2018-04-04 00:23:33 +07:00
|
|
|
// SPDX-License-Identifier: GPL-2.0
|
2008-06-12 08:53:53 +07:00
|
|
|
/*
|
|
|
|
|
* Copyright (C) 2007 Oracle. All rights reserved.
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
#include <linux/kernel.h>
|
|
|
|
|
#include <linux/bio.h>
|
|
|
|
|
#include <linux/buffer_head.h>
|
|
|
|
|
#include <linux/file.h>
|
|
|
|
|
#include <linux/fs.h>
|
2008-10-10 00:39:39 +07:00
|
|
|
#include <linux/fsnotify.h>
|
2008-06-12 08:53:53 +07:00
|
|
|
#include <linux/pagemap.h>
|
|
|
|
|
#include <linux/highmem.h>
|
|
|
|
|
#include <linux/time.h>
|
|
|
|
|
#include <linux/init.h>
|
|
|
|
|
#include <linux/string.h>
|
|
|
|
|
#include <linux/backing-dev.h>
|
2008-10-10 00:39:39 +07:00
|
|
|
#include <linux/mount.h>
|
2008-06-12 08:53:53 +07:00
|
|
|
#include <linux/mpage.h>
|
2008-10-10 00:39:39 +07:00
|
|
|
#include <linux/namei.h>
|
2008-06-12 08:53:53 +07:00
|
|
|
#include <linux/swap.h>
|
|
|
|
|
#include <linux/writeback.h>
|
|
|
|
|
#include <linux/compat.h>
|
|
|
|
|
#include <linux/bit_spinlock.h>
|
2008-10-10 00:39:39 +07:00
|
|
|
#include <linux/security.h>
|
2008-06-12 08:53:53 +07:00
|
|
|
#include <linux/xattr.h>
|
2017-06-01 00:32:09 +07:00
|
|
|
#include <linux/mm.h>
|
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h
percpu.h is included by sched.h and module.h and thus ends up being
included when building most .c files. percpu.h includes slab.h which
in turn includes gfp.h making everything defined by the two files
universally available and complicating inclusion dependencies.
percpu.h -> slab.h dependency is about to be removed. Prepare for
this change by updating users of gfp and slab facilities include those
headers directly instead of assuming availability. As this conversion
needs to touch large number of source files, the following script is
used as the basis of conversion.
http://userweb.kernel.org/~tj/misc/slabh-sweep.py
The script does the followings.
* Scan files for gfp and slab usages and update includes such that
only the necessary includes are there. ie. if only gfp is used,
gfp.h, if slab is used, slab.h.
* When the script inserts a new include, it looks at the include
blocks and try to put the new include such that its order conforms
to its surrounding. It's put in the include block which contains
core kernel includes, in the same order that the rest are ordered -
alphabetical, Christmas tree, rev-Xmas-tree or at the end if there
doesn't seem to be any matching order.
* If the script can't find a place to put a new include (mostly
because the file doesn't have fitting include block), it prints out
an error message indicating which .h file needs to be added to the
file.
The conversion was done in the following steps.
1. The initial automatic conversion of all .c files updated slightly
over 4000 files, deleting around 700 includes and adding ~480 gfp.h
and ~3000 slab.h inclusions. The script emitted errors for ~400
files.
2. Each error was manually checked. Some didn't need the inclusion,
some needed manual addition while adding it to implementation .h or
embedding .c file was more appropriate for others. This step added
inclusions to around 150 files.
3. The script was run again and the output was compared to the edits
from #2 to make sure no file was left behind.
4. Several build tests were done and a couple of problems were fixed.
e.g. lib/decompress_*.c used malloc/free() wrappers around slab
APIs requiring slab.h to be added manually.
5. The script was run on all .h files but without automatically
editing them as sprinkling gfp.h and slab.h inclusions around .h
files could easily lead to inclusion dependency hell. Most gfp.h
inclusion directives were ignored as stuff from gfp.h was usually
wildly available and often used in preprocessor macros. Each
slab.h inclusion directive was examined and added manually as
necessary.
6. percpu.h was updated not to include slab.h.
7. Build test were done on the following configurations and failures
were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my
distributed build env didn't work with gcov compiles) and a few
more options had to be turned off depending on archs to make things
build (like ipr on powerpc/64 which failed due to missing writeq).
* x86 and x86_64 UP and SMP allmodconfig and a custom test config.
* powerpc and powerpc64 SMP allmodconfig
* sparc and sparc64 SMP allmodconfig
* ia64 SMP allmodconfig
* s390 SMP allmodconfig
* alpha SMP allmodconfig
* um on x86_64 SMP allmodconfig
8. percpu.h modifications were reverted so that it could be applied as
a separate patch and serve as bisection point.
Given the fact that I had only a couple of failures from tests on step
6, I'm fairly confident about the coverage of this conversion patch.
If there is a breakage, it's likely to be something in one of the arch
headers which should be easily discoverable easily on most builds of
the specific arch.
Signed-off-by: Tejun Heo <tj@kernel.org>
Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 15:04:11 +07:00
|
|
|
#include <linux/slab.h>
|
2011-03-24 17:24:28 +07:00
|
|
|
#include <linux/blkdev.h>
|
2012-07-25 22:35:53 +07:00
|
|
|
#include <linux/uuid.h>
|
2013-01-29 13:04:50 +07:00
|
|
|
#include <linux/btrfs.h>
|
2013-08-07 01:42:51 +07:00
|
|
|
#include <linux/uaccess.h>
|
2018-01-29 18:41:30 +07:00
|
|
|
#include <linux/iversion.h>
|
2008-06-12 08:53:53 +07:00
|
|
|
#include "ctree.h"
|
|
|
|
|
#include "disk-io.h"
|
|
|
|
|
#include "transaction.h"
|
|
|
|
|
#include "btrfs_inode.h"
|
|
|
|
|
#include "print-tree.h"
|
|
|
|
|
#include "volumes.h"
|
2008-06-26 03:01:30 +07:00
|
|
|
#include "locking.h"
|
Btrfs: Cache free inode numbers in memory
Currently btrfs stores the highest objectid of the fs tree, and it always
returns (highest+1) inode number when we create a file, so inode numbers
won't be reclaimed when we delete files, so we'll run out of inode numbers
as we keep create/delete files in 32bits machines.
This fixes it, and it works similarly to how we cache free space in block
cgroups.
We start a kernel thread to read the file tree. By scanning inode items,
we know which chunks of inode numbers are free, and we cache them in
an rb-tree.
Because we are searching the commit root, we have to carefully handle the
cross-transaction case.
The rb-tree is a hybrid extent+bitmap tree, so if we have too many small
chunks of inode numbers, we'll use bitmaps. Initially we allow 16K ram
of extents, and a bitmap will be used if we exceed this threshold. The
extents threshold is adjusted in runtime.
Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
2011-04-20 09:06:11 +07:00
|
|
|
#include "inode-map.h"
|
2011-07-07 21:48:38 +07:00
|
|
|
#include "backref.h"
|
2012-06-05 01:03:51 +07:00
|
|
|
#include "rcu-string.h"
|
2012-07-26 04:19:24 +07:00
|
|
|
#include "send.h"
|
2012-11-06 21:08:53 +07:00
|
|
|
#include "dev-replace.h"
|
Btrfs: add support for inode properties
This change adds infrastructure to allow for generic properties for
inodes. Properties are name/value pairs that can be associated with
inodes for different purposes. They are stored as xattrs with the
prefix "btrfs."
Properties can be inherited - this means when a directory inode has
inheritable properties set, these are added to new inodes created
under that directory. Further, subvolumes can also have properties
associated with them, and they can be inherited from their parent
subvolume. Naturally, directory properties have priority over subvolume
properties (in practice a subvolume property is just a regular
property associated with the root inode, objectid 256, of the
subvolume's fs tree).
This change also adds one specific property implementation, named
"compression", whose values can be "lzo" or "zlib" and it's an
inheritable property.
The corresponding changes to btrfs-progs were also implemented.
A patch with xfstests for this feature will follow once there's
agreement on this change/feature.
Further, the script at the bottom of this commit message was used to
do some benchmarks to measure any performance penalties of this feature.
Basically the tests correspond to:
Test 1 - create a filesystem and mount it with compress-force=lzo,
then sequentially create N files of 64Kb each, measure how long it took
to create the files, unmount the filesystem, mount the filesystem and
perform an 'ls -lha' against the test directory holding the N files, and
report the time the command took.
Test 2 - create a filesystem and don't use any compression option when
mounting it - instead set the compression property of the subvolume's
root to 'lzo'. Then create N files of 64Kb, and report the time it took.
The unmount the filesystem, mount it again and perform an 'ls -lha' like
in the former test. This means every single file ends up with a property
(xattr) associated to it.
Test 3 - same as test 2, but uses 4 properties - 3 are duplicates of the
compression property, have no real effect other than adding more work
when inheriting properties and taking more btree leaf space.
Test 4 - same as test 3 but with 10 properties per file.
Results (in seconds, and averages of 5 runs each), for different N
numbers of files follow.
* Without properties (test 1)
file creation time ls -lha time
10 000 files 3.49 0.76
100 000 files 47.19 8.37
1 000 000 files 518.51 107.06
* With 1 property (compression property set to lzo - test 2)
file creation time ls -lha time
10 000 files 3.63 0.93
100 000 files 48.56 9.74
1 000 000 files 537.72 125.11
* With 4 properties (test 3)
file creation time ls -lha time
10 000 files 3.94 1.20
100 000 files 52.14 11.48
1 000 000 files 572.70 142.13
* With 10 properties (test 4)
file creation time ls -lha time
10 000 files 4.61 1.35
100 000 files 58.86 13.83
1 000 000 files 656.01 177.61
The increased latencies with properties are essencialy because of:
*) When creating an inode, we now synchronously write 1 more item
(an xattr item) for each property inherited from the parent dir
(or subvolume). This could be done in an asynchronous way such
as we do for dir intex items (delayed-inode.c), which could help
reduce the file creation latency;
*) With properties, we now have larger fs trees. For this particular
test each xattr item uses 75 bytes of leaf space in the fs tree.
This could be less by using a new item for xattr items, instead of
the current btrfs_dir_item, since we could cut the 'location' and
'type' fields (saving 18 bytes) and maybe 'transid' too (saving a
total of 26 bytes per xattr item) from the btrfs_dir_item type.
Also tried batching the xattr insertions (ignoring proper hash
collision handling, since it didn't exist) when creating files that
inherit properties from their parent inode/subvolume, but the end
results were (surprisingly) essentially the same.
Test script:
$ cat test.pl
#!/usr/bin/perl -w
use strict;
use Time::HiRes qw(time);
use constant NUM_FILES => 10_000;
use constant FILE_SIZES => (64 * 1024);
use constant DEV => '/dev/sdb4';
use constant MNT_POINT => '/home/fdmanana/btrfs-tests/dev';
use constant TEST_DIR => (MNT_POINT . '/testdir');
system("mkfs.btrfs", "-l", "16384", "-f", DEV) == 0 or die "mkfs.btrfs failed!";
# following line for testing without properties
#system("mount", "-o", "compress-force=lzo", DEV, MNT_POINT) == 0 or die "mount failed!";
# following 2 lines for testing with properties
system("mount", DEV, MNT_POINT) == 0 or die "mount failed!";
system("btrfs", "prop", "set", MNT_POINT, "compression", "lzo") == 0 or die "set prop failed!";
system("mkdir", TEST_DIR) == 0 or die "mkdir failed!";
my ($t1, $t2);
$t1 = time();
for (my $i = 1; $i <= NUM_FILES; $i++) {
my $p = TEST_DIR . '/file_' . $i;
open(my $f, '>', $p) or die "Error opening file!";
$f->autoflush(1);
for (my $j = 0; $j < FILE_SIZES; $j += 4096) {
print $f ('A' x 4096) or die "Error writing to file!";
}
close($f);
}
$t2 = time();
print "Time to create " . NUM_FILES . ": " . ($t2 - $t1) . " seconds.\n";
system("umount", DEV) == 0 or die "umount failed!";
system("mount", DEV, MNT_POINT) == 0 or die "mount failed!";
$t1 = time();
system("bash -c 'ls -lha " . TEST_DIR . " > /dev/null'") == 0 or die "ls failed!";
$t2 = time();
print "Time to ls -lha all files: " . ($t2 - $t1) . " seconds.\n";
system("umount", DEV) == 0 or die "umount failed!";
Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com>
Signed-off-by: Josef Bacik <jbacik@fb.com>
Signed-off-by: Chris Mason <clm@fb.com>
2014-01-07 18:47:46 +07:00
|
|
|
#include "props.h"
|
2013-11-02 00:07:02 +07:00
|
|
|
#include "sysfs.h"
|
2014-05-14 07:30:47 +07:00
|
|
|
#include "qgroup.h"
|
Btrfs: fix unreplayable log after snapshot delete + parent dir fsync
If we delete a snapshot, fsync its parent directory and crash/power fail
before the next transaction commit, on the next mount when we attempt to
replay the log tree of the root containing the parent directory we will
fail and prevent the filesystem from mounting, which is solvable by wiping
out the log trees with the btrfs-zero-log tool but very inconvenient as
we will lose any data and metadata fsynced before the parent directory
was fsynced.
For example:
$ mkfs.btrfs -f /dev/sdc
$ mount /dev/sdc /mnt
$ mkdir /mnt/testdir
$ btrfs subvolume snapshot /mnt /mnt/testdir/snap
$ btrfs subvolume delete /mnt/testdir/snap
$ xfs_io -c "fsync" /mnt/testdir
< crash / power failure and reboot >
$ mount /dev/sdc /mnt
mount: mount(2) failed: No such file or directory
And in dmesg/syslog we get the following message and trace:
[192066.361162] BTRFS info (device dm-0): failed to delete reference to snap, inode 257 parent 257
[192066.363010] ------------[ cut here ]------------
[192066.365268] WARNING: CPU: 4 PID: 5130 at fs/btrfs/inode.c:3986 __btrfs_unlink_inode+0x17a/0x354 [btrfs]()
[192066.367250] BTRFS: Transaction aborted (error -2)
[192066.368401] Modules linked in: btrfs dm_flakey dm_mod ppdev sha256_generic xor raid6_pq hmac drbg ansi_cprng aesni_intel acpi_cpufreq tpm_tis aes_x86_64 tpm ablk_helper evdev cryptd sg parport_pc i2c_piix4 psmouse lrw parport i2c_core pcspkr gf128mul processor serio_raw glue_helper button loop autofs4 ext4 crc16 mbcache jbd2 sd_mod sr_mod cdrom ata_generic virtio_scsi ata_piix libata virtio_pci virtio_ring crc32c_intel scsi_mod e1000 virtio floppy [last unloaded: btrfs]
[192066.377154] CPU: 4 PID: 5130 Comm: mount Tainted: G W 4.4.0-rc6-btrfs-next-20+ #1
[192066.378875] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS by qemu-project.org 04/01/2014
[192066.380889] 0000000000000000 ffff880143923670 ffffffff81257570 ffff8801439236b8
[192066.382561] ffff8801439236a8 ffffffff8104ec07 ffffffffa039dc2c 00000000fffffffe
[192066.384191] ffff8801ed31d000 ffff8801b9fc9c88 ffff8801086875e0 ffff880143923710
[192066.385827] Call Trace:
[192066.386373] [<ffffffff81257570>] dump_stack+0x4e/0x79
[192066.387387] [<ffffffff8104ec07>] warn_slowpath_common+0x99/0xb2
[192066.388429] [<ffffffffa039dc2c>] ? __btrfs_unlink_inode+0x17a/0x354 [btrfs]
[192066.389236] [<ffffffff8104ec68>] warn_slowpath_fmt+0x48/0x50
[192066.389884] [<ffffffffa039dc2c>] __btrfs_unlink_inode+0x17a/0x354 [btrfs]
[192066.390621] [<ffffffff81184b55>] ? iput+0xb0/0x266
[192066.391200] [<ffffffffa039ea25>] btrfs_unlink_inode+0x1c/0x3d [btrfs]
[192066.391930] [<ffffffffa03ca623>] check_item_in_log+0x1fe/0x29b [btrfs]
[192066.392715] [<ffffffffa03ca827>] replay_dir_deletes+0x167/0x1cf [btrfs]
[192066.393510] [<ffffffffa03cccc7>] replay_one_buffer+0x417/0x570 [btrfs]
[192066.394241] [<ffffffffa03ca164>] walk_up_log_tree+0x10e/0x1dc [btrfs]
[192066.394958] [<ffffffffa03cac72>] walk_log_tree+0xa5/0x190 [btrfs]
[192066.395628] [<ffffffffa03ce8b8>] btrfs_recover_log_trees+0x239/0x32c [btrfs]
[192066.396790] [<ffffffffa03cc8b0>] ? replay_one_extent+0x50a/0x50a [btrfs]
[192066.397891] [<ffffffffa0394041>] open_ctree+0x1d8b/0x2167 [btrfs]
[192066.398897] [<ffffffffa03706e1>] btrfs_mount+0x5ef/0x729 [btrfs]
[192066.399823] [<ffffffff8108ad98>] ? trace_hardirqs_on+0xd/0xf
[192066.400739] [<ffffffff8108959b>] ? lockdep_init_map+0xb9/0x1b3
[192066.401700] [<ffffffff811714b9>] mount_fs+0x67/0x131
[192066.402482] [<ffffffff81188560>] vfs_kern_mount+0x6c/0xde
[192066.403930] [<ffffffffa03702bd>] btrfs_mount+0x1cb/0x729 [btrfs]
[192066.404831] [<ffffffff8108ad98>] ? trace_hardirqs_on+0xd/0xf
[192066.405726] [<ffffffff8108959b>] ? lockdep_init_map+0xb9/0x1b3
[192066.406621] [<ffffffff811714b9>] mount_fs+0x67/0x131
[192066.407401] [<ffffffff81188560>] vfs_kern_mount+0x6c/0xde
[192066.408247] [<ffffffff8118ae36>] do_mount+0x893/0x9d2
[192066.409047] [<ffffffff8113009b>] ? strndup_user+0x3f/0x8c
[192066.409842] [<ffffffff8118b187>] SyS_mount+0x75/0xa1
[192066.410621] [<ffffffff8147e517>] entry_SYSCALL_64_fastpath+0x12/0x6b
[192066.411572] ---[ end trace 2de42126c1e0a0f0 ]---
[192066.412344] BTRFS: error (device dm-0) in __btrfs_unlink_inode:3986: errno=-2 No such entry
[192066.413748] BTRFS: error (device dm-0) in btrfs_replay_log:2464: errno=-2 No such entry (Failed to recover log tree)
[192066.415458] BTRFS error (device dm-0): cleaner transaction attach returned -30
[192066.444613] BTRFS: open_ctree failed
This happens because when we are replaying the log and processing the
directory entry pointing to the snapshot in the subvolume tree, we treat
its btrfs_dir_item item as having a location with a key type matching
BTRFS_INODE_ITEM_KEY, which is wrong because the type matches
BTRFS_ROOT_ITEM_KEY and therefore must be processed differently, as the
object id refers to a root number and not to an inode in the root
containing the parent directory.
So fix this by triggering a transaction commit if an fsync against the
parent directory is requested after deleting a snapshot. This is the
simplest approach for a rare use case. Some alternative that avoids the
transaction commit would require more code to explicitly delete the
snapshot at log replay time (factoring out common code from ioctl.c:
btrfs_ioctl_snap_destroy()), special care at fsync time to remove the
log tree of the snapshot's root from the log root of the root of tree
roots, amongst other steps.
A test case for xfstests that triggers the issue follows.
seq=`basename $0`
seqres=$RESULT_DIR/$seq
echo "QA output created by $seq"
tmp=/tmp/$$
status=1 # failure is the default!
trap "_cleanup; exit \$status" 0 1 2 3 15
_cleanup()
{
_cleanup_flakey
cd /
rm -f $tmp.*
}
# get standard environment, filters and checks
. ./common/rc
. ./common/filter
. ./common/dmflakey
# real QA test starts here
_need_to_be_root
_supported_fs btrfs
_supported_os Linux
_require_scratch
_require_dm_target flakey
_require_metadata_journaling $SCRATCH_DEV
rm -f $seqres.full
_scratch_mkfs >>$seqres.full 2>&1
_init_flakey
_mount_flakey
# Create a snapshot at the root of our filesystem (mount point path), delete it,
# fsync the mount point path, crash and mount to replay the log. This should
# succeed and after the filesystem is mounted the snapshot should not be visible
# anymore.
_run_btrfs_util_prog subvolume snapshot $SCRATCH_MNT $SCRATCH_MNT/snap1
_run_btrfs_util_prog subvolume delete $SCRATCH_MNT/snap1
$XFS_IO_PROG -c "fsync" $SCRATCH_MNT
_flakey_drop_and_remount
[ -e $SCRATCH_MNT/snap1 ] && \
echo "Snapshot snap1 still exists after log replay"
# Similar scenario as above, but this time the snapshot is created inside a
# directory and not directly under the root (mount point path).
mkdir $SCRATCH_MNT/testdir
_run_btrfs_util_prog subvolume snapshot $SCRATCH_MNT $SCRATCH_MNT/testdir/snap2
_run_btrfs_util_prog subvolume delete $SCRATCH_MNT/testdir/snap2
$XFS_IO_PROG -c "fsync" $SCRATCH_MNT/testdir
_flakey_drop_and_remount
[ -e $SCRATCH_MNT/testdir/snap2 ] && \
echo "Snapshot snap2 still exists after log replay"
_unmount_flakey
echo "Silence is golden"
status=0
exit
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Tested-by: Liu Bo <bo.li.liu@oracle.com>
Reviewed-by: Liu Bo <bo.li.liu@oracle.com>
Signed-off-by: Chris Mason <clm@fb.com>
2016-02-10 17:42:25 +07:00
|
|
|
#include "tree-log.h"
|
2016-03-10 16:26:59 +07:00
|
|
|
#include "compression.h"
|
2008-06-12 08:53:53 +07:00
|
|
|
|
2014-01-31 03:17:00 +07:00
|
|
|
#ifdef CONFIG_64BIT
|
|
|
|
|
/* If we have a 32-bit userspace and 64-bit kernel, then the UAPI
|
|
|
|
|
* structures are incorrect, as the timespec structure from userspace
|
|
|
|
|
* is 4 bytes too small. We define these alternatives here to teach
|
|
|
|
|
* the kernel about the 32-bit struct packing.
|
|
|
|
|
*/
|
|
|
|
|
struct btrfs_ioctl_timespec_32 {
|
|
|
|
|
__u64 sec;
|
|
|
|
|
__u32 nsec;
|
|
|
|
|
} __attribute__ ((__packed__));
|
|
|
|
|
|
|
|
|
|
struct btrfs_ioctl_received_subvol_args_32 {
|
|
|
|
|
char uuid[BTRFS_UUID_SIZE]; /* in */
|
|
|
|
|
__u64 stransid; /* in */
|
|
|
|
|
__u64 rtransid; /* out */
|
|
|
|
|
struct btrfs_ioctl_timespec_32 stime; /* in */
|
|
|
|
|
struct btrfs_ioctl_timespec_32 rtime; /* out */
|
|
|
|
|
__u64 flags; /* in */
|
|
|
|
|
__u64 reserved[16]; /* in */
|
|
|
|
|
} __attribute__ ((__packed__));
|
|
|
|
|
|
|
|
|
|
#define BTRFS_IOC_SET_RECEIVED_SUBVOL_32 _IOWR(BTRFS_IOCTL_MAGIC, 37, \
|
|
|
|
|
struct btrfs_ioctl_received_subvol_args_32)
|
|
|
|
|
#endif
|
|
|
|
|
|
2017-09-27 21:43:13 +07:00
|
|
|
#if defined(CONFIG_64BIT) && defined(CONFIG_COMPAT)
|
|
|
|
|
struct btrfs_ioctl_send_args_32 {
|
|
|
|
|
__s64 send_fd; /* in */
|
|
|
|
|
__u64 clone_sources_count; /* in */
|
|
|
|
|
compat_uptr_t clone_sources; /* in */
|
|
|
|
|
__u64 parent_root; /* in */
|
|
|
|
|
__u64 flags; /* in */
|
|
|
|
|
__u64 reserved[4]; /* in */
|
|
|
|
|
} __attribute__ ((__packed__));
|
|
|
|
|
|
|
|
|
|
#define BTRFS_IOC_SEND_32 _IOW(BTRFS_IOCTL_MAGIC, 38, \
|
|
|
|
|
struct btrfs_ioctl_send_args_32)
|
|
|
|
|
#endif
|
2014-01-31 03:17:00 +07:00
|
|
|
|
2013-08-07 01:42:51 +07:00
|
|
|
static int btrfs_clone(struct inode *src, struct inode *inode,
|
2015-07-01 04:42:08 +07:00
|
|
|
u64 off, u64 olen, u64 olen_aligned, u64 destoff,
|
|
|
|
|
int no_time_update);
|
2013-08-07 01:42:51 +07:00
|
|
|
|
2009-04-17 15:37:41 +07:00
|
|
|
/* Mask out flags that are inappropriate for the given type of inode. */
|
2018-03-26 23:29:41 +07:00
|
|
|
static unsigned int btrfs_mask_flags(umode_t mode, unsigned int flags)
|
2009-04-17 15:37:41 +07:00
|
|
|
{
|
|
|
|
|
if (S_ISDIR(mode))
|
|
|
|
|
return flags;
|
|
|
|
|
else if (S_ISREG(mode))
|
|
|
|
|
return flags & ~FS_DIRSYNC_FL;
|
|
|
|
|
else
|
|
|
|
|
return flags & (FS_NODUMP_FL | FS_NOATIME_FL);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Export inode flags to the format expected by the FS_IOC_GETFLAGS ioctl.
|
|
|
|
|
*/
|
|
|
|
|
static unsigned int btrfs_flags_to_ioctl(unsigned int flags)
|
|
|
|
|
{
|
|
|
|
|
unsigned int iflags = 0;
|
|
|
|
|
|
|
|
|
|
if (flags & BTRFS_INODE_SYNC)
|
|
|
|
|
iflags |= FS_SYNC_FL;
|
|
|
|
|
if (flags & BTRFS_INODE_IMMUTABLE)
|
|
|
|
|
iflags |= FS_IMMUTABLE_FL;
|
|
|
|
|
if (flags & BTRFS_INODE_APPEND)
|
|
|
|
|
iflags |= FS_APPEND_FL;
|
|
|
|
|
if (flags & BTRFS_INODE_NODUMP)
|
|
|
|
|
iflags |= FS_NODUMP_FL;
|
|
|
|
|
if (flags & BTRFS_INODE_NOATIME)
|
|
|
|
|
iflags |= FS_NOATIME_FL;
|
|
|
|
|
if (flags & BTRFS_INODE_DIRSYNC)
|
|
|
|
|
iflags |= FS_DIRSYNC_FL;
|
2011-04-15 10:03:06 +07:00
|
|
|
if (flags & BTRFS_INODE_NODATACOW)
|
|
|
|
|
iflags |= FS_NOCOW_FL;
|
|
|
|
|
|
2016-03-15 07:09:59 +07:00
|
|
|
if (flags & BTRFS_INODE_NOCOMPRESS)
|
2011-04-15 10:03:06 +07:00
|
|
|
iflags |= FS_NOCOMP_FL;
|
2016-03-15 07:09:59 +07:00
|
|
|
else if (flags & BTRFS_INODE_COMPRESS)
|
|
|
|
|
iflags |= FS_COMPR_FL;
|
2009-04-17 15:37:41 +07:00
|
|
|
|
|
|
|
|
return iflags;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Update inode->i_flags based on the btrfs internal flags.
|
|
|
|
|
*/
|
|
|
|
|
void btrfs_update_iflags(struct inode *inode)
|
|
|
|
|
{
|
|
|
|
|
struct btrfs_inode *ip = BTRFS_I(inode);
|
2014-06-26 04:36:02 +07:00
|
|
|
unsigned int new_fl = 0;
|
2009-04-17 15:37:41 +07:00
|
|
|
|
|
|
|
|
if (ip->flags & BTRFS_INODE_SYNC)
|
2014-06-26 04:36:02 +07:00
|
|
|
new_fl |= S_SYNC;
|
2009-04-17 15:37:41 +07:00
|
|
|
if (ip->flags & BTRFS_INODE_IMMUTABLE)
|
2014-06-26 04:36:02 +07:00
|
|
|
new_fl |= S_IMMUTABLE;
|
2009-04-17 15:37:41 +07:00
|
|
|
if (ip->flags & BTRFS_INODE_APPEND)
|
2014-06-26 04:36:02 +07:00
|
|
|
new_fl |= S_APPEND;
|
2009-04-17 15:37:41 +07:00
|
|
|
if (ip->flags & BTRFS_INODE_NOATIME)
|
2014-06-26 04:36:02 +07:00
|
|
|
new_fl |= S_NOATIME;
|
2009-04-17 15:37:41 +07:00
|
|
|
if (ip->flags & BTRFS_INODE_DIRSYNC)
|
2014-06-26 04:36:02 +07:00
|
|
|
new_fl |= S_DIRSYNC;
|
|
|
|
|
|
|
|
|
|
set_mask_bits(&inode->i_flags,
|
|
|
|
|
S_SYNC | S_APPEND | S_IMMUTABLE | S_NOATIME | S_DIRSYNC,
|
|
|
|
|
new_fl);
|
2009-04-17 15:37:41 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static int btrfs_ioctl_getflags(struct file *file, void __user *arg)
|
|
|
|
|
{
|
2013-01-24 05:07:38 +07:00
|
|
|
struct btrfs_inode *ip = BTRFS_I(file_inode(file));
|
2009-04-17 15:37:41 +07:00
|
|
|
unsigned int flags = btrfs_flags_to_ioctl(ip->flags);
|
|
|
|
|
|
|
|
|
|
if (copy_to_user(arg, &flags, sizeof(flags)))
|
|
|
|
|
return -EFAULT;
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
Btrfs: Per file/directory controls for COW and compression
Data compression and data cow are controlled across the entire FS by mount
options right now. ioctls are needed to set this on a per file or per
directory basis. This has been proposed previously, but VFS developers
wanted us to use generic ioctls rather than btrfs-specific ones.
According to Chris's comment, there should be just one true compression
method(probably LZO) stored in the super. However, before this, we would
wait for that one method is stable enough to be adopted into the super.
So I list it as a long term goal, and just store it in ram today.
After applying this patch, we can use the generic "FS_IOC_SETFLAGS" ioctl to
control file and directory's datacow and compression attribute.
NOTE:
- The compression type is selected by such rules:
If we mount btrfs with compress options, ie, zlib/lzo, the type is it.
Otherwise, we'll use the default compress type (zlib today).
v1->v2:
- rebase to the latest btrfs.
v2->v3:
- fix a problem, i.e. when a file is set NOCOW via mount option, then this NOCOW
will be screwed by inheritance from parent directory.
Signed-off-by: Liu Bo <liubo2009@cn.fujitsu.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
2011-03-22 17:12:20 +07:00
|
|
|
static int check_flags(unsigned int flags)
|
|
|
|
|
{
|
|
|
|
|
if (flags & ~(FS_IMMUTABLE_FL | FS_APPEND_FL | \
|
|
|
|
|
FS_NOATIME_FL | FS_NODUMP_FL | \
|
|
|
|
|
FS_SYNC_FL | FS_DIRSYNC_FL | \
|
2011-04-15 10:02:49 +07:00
|
|
|
FS_NOCOMP_FL | FS_COMPR_FL |
|
|
|
|
|
FS_NOCOW_FL))
|
Btrfs: Per file/directory controls for COW and compression
Data compression and data cow are controlled across the entire FS by mount
options right now. ioctls are needed to set this on a per file or per
directory basis. This has been proposed previously, but VFS developers
wanted us to use generic ioctls rather than btrfs-specific ones.
According to Chris's comment, there should be just one true compression
method(probably LZO) stored in the super. However, before this, we would
wait for that one method is stable enough to be adopted into the super.
So I list it as a long term goal, and just store it in ram today.
After applying this patch, we can use the generic "FS_IOC_SETFLAGS" ioctl to
control file and directory's datacow and compression attribute.
NOTE:
- The compression type is selected by such rules:
If we mount btrfs with compress options, ie, zlib/lzo, the type is it.
Otherwise, we'll use the default compress type (zlib today).
v1->v2:
- rebase to the latest btrfs.
v2->v3:
- fix a problem, i.e. when a file is set NOCOW via mount option, then this NOCOW
will be screwed by inheritance from parent directory.
Signed-off-by: Liu Bo <liubo2009@cn.fujitsu.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
2011-03-22 17:12:20 +07:00
|
|
|
return -EOPNOTSUPP;
|
|
|
|
|
|
|
|
|
|
if ((flags & FS_NOCOMP_FL) && (flags & FS_COMPR_FL))
|
|
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
2009-04-17 15:37:41 +07:00
|
|
|
static int btrfs_ioctl_setflags(struct file *file, void __user *arg)
|
|
|
|
|
{
|
2013-01-24 05:07:38 +07:00
|
|
|
struct inode *inode = file_inode(file);
|
2016-06-23 05:54:23 +07:00
|
|
|
struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
|
2009-04-17 15:37:41 +07:00
|
|
|
struct btrfs_inode *ip = BTRFS_I(inode);
|
|
|
|
|
struct btrfs_root *root = ip->root;
|
|
|
|
|
struct btrfs_trans_handle *trans;
|
|
|
|
|
unsigned int flags, oldflags;
|
|
|
|
|
int ret;
|
2011-12-29 12:36:45 +07:00
|
|
|
u64 ip_oldflags;
|
|
|
|
|
unsigned int i_oldflags;
|
btrfs: allow setting NOCOW for a zero sized file via ioctl
Hi,
the patch si simple, but it has user visible impact and I'm not quite sure how
to resolve it.
In short, $subj says it, chattr -C supports it and we want to use it.
The conditions that acutally allow to change the NOCOW flag are clear. What if
I try to set the flag on a file that is not empty? Options:
1) whole ioctl will fail, EINVAL
2.1) ioctl will succeed, the NOCOW flag will be silently removed, but the file
will stay COW-ed and checksummed
2.2) ioctl will succeed, flag will not be removed and a syslog message will
warn that the COW flag has not been changed
2.2.1) dtto, no syslog message
Man page of chattr states that
"If it is set on a file which already has data blocks, it is undefined when
the blocks assigned to the file will be fully stable."
Yes, it's undefined and with current implementation it'll never happen. So from
this end, the user cannot expect anything. I'm trying to find a reasonable
behaviour, so that a command like 'chattr -R -aijS +C' to tweak a broad set of
flags in a deep directory does not fail unnecessarily and does not pollute the
log.
My personal preference is 2.2.1, but my dev's oppinion is skewed, not counting
the fact that I know the code and otherwise would look there before consulting
the documentation.
The patch implements 2.2.1.
david
-------------8<-------------------
From: David Sterba <dsterba@suse.cz>
It's safe to turn off checksums for a zero sized file.
http://thread.gmane.org/gmane.comp.file-systems.btrfs/18030
"We cannot switch on NODATASUM for a file that already has extents that
are checksummed. The invariant here is that either all the extents or
none are checksummed.
Theoretically it's possible to add/remove all checksums from a given
file, but it's a potentially longtime operation, the file has to be in
some intermediate state where the checksums partially exist but have to
be ignored (for the csum->nocsum) until the file is fully converted,
this brings more special cases to extent handling, it has to survive
power failure and remain consistent, and probably needs to be restarted
after next mount."
Signed-off-by: David Sterba <dsterba@suse.cz>
2012-09-07 18:56:55 +07:00
|
|
|
umode_t mode;
|
2009-04-17 15:37:41 +07:00
|
|
|
|
2014-01-16 21:50:22 +07:00
|
|
|
if (!inode_owner_or_capable(inode))
|
|
|
|
|
return -EPERM;
|
|
|
|
|
|
2010-12-20 15:04:08 +07:00
|
|
|
if (btrfs_root_readonly(root))
|
|
|
|
|
return -EROFS;
|
|
|
|
|
|
2009-04-17 15:37:41 +07:00
|
|
|
if (copy_from_user(&flags, arg, sizeof(flags)))
|
|
|
|
|
return -EFAULT;
|
|
|
|
|
|
Btrfs: Per file/directory controls for COW and compression
Data compression and data cow are controlled across the entire FS by mount
options right now. ioctls are needed to set this on a per file or per
directory basis. This has been proposed previously, but VFS developers
wanted us to use generic ioctls rather than btrfs-specific ones.
According to Chris's comment, there should be just one true compression
method(probably LZO) stored in the super. However, before this, we would
wait for that one method is stable enough to be adopted into the super.
So I list it as a long term goal, and just store it in ram today.
After applying this patch, we can use the generic "FS_IOC_SETFLAGS" ioctl to
control file and directory's datacow and compression attribute.
NOTE:
- The compression type is selected by such rules:
If we mount btrfs with compress options, ie, zlib/lzo, the type is it.
Otherwise, we'll use the default compress type (zlib today).
v1->v2:
- rebase to the latest btrfs.
v2->v3:
- fix a problem, i.e. when a file is set NOCOW via mount option, then this NOCOW
will be screwed by inheritance from parent directory.
Signed-off-by: Liu Bo <liubo2009@cn.fujitsu.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
2011-03-22 17:12:20 +07:00
|
|
|
ret = check_flags(flags);
|
|
|
|
|
if (ret)
|
|
|
|
|
return ret;
|
2008-06-12 08:53:53 +07:00
|
|
|
|
2012-06-12 21:20:32 +07:00
|
|
|
ret = mnt_want_write_file(file);
|
|
|
|
|
if (ret)
|
|
|
|
|
return ret;
|
|
|
|
|
|
2016-01-23 03:40:57 +07:00
|
|
|
inode_lock(inode);
|
2009-04-17 15:37:41 +07:00
|
|
|
|
2011-12-29 12:36:45 +07:00
|
|
|
ip_oldflags = ip->flags;
|
|
|
|
|
i_oldflags = inode->i_flags;
|
btrfs: allow setting NOCOW for a zero sized file via ioctl
Hi,
the patch si simple, but it has user visible impact and I'm not quite sure how
to resolve it.
In short, $subj says it, chattr -C supports it and we want to use it.
The conditions that acutally allow to change the NOCOW flag are clear. What if
I try to set the flag on a file that is not empty? Options:
1) whole ioctl will fail, EINVAL
2.1) ioctl will succeed, the NOCOW flag will be silently removed, but the file
will stay COW-ed and checksummed
2.2) ioctl will succeed, flag will not be removed and a syslog message will
warn that the COW flag has not been changed
2.2.1) dtto, no syslog message
Man page of chattr states that
"If it is set on a file which already has data blocks, it is undefined when
the blocks assigned to the file will be fully stable."
Yes, it's undefined and with current implementation it'll never happen. So from
this end, the user cannot expect anything. I'm trying to find a reasonable
behaviour, so that a command like 'chattr -R -aijS +C' to tweak a broad set of
flags in a deep directory does not fail unnecessarily and does not pollute the
log.
My personal preference is 2.2.1, but my dev's oppinion is skewed, not counting
the fact that I know the code and otherwise would look there before consulting
the documentation.
The patch implements 2.2.1.
david
-------------8<-------------------
From: David Sterba <dsterba@suse.cz>
It's safe to turn off checksums for a zero sized file.
http://thread.gmane.org/gmane.comp.file-systems.btrfs/18030
"We cannot switch on NODATASUM for a file that already has extents that
are checksummed. The invariant here is that either all the extents or
none are checksummed.
Theoretically it's possible to add/remove all checksums from a given
file, but it's a potentially longtime operation, the file has to be in
some intermediate state where the checksums partially exist but have to
be ignored (for the csum->nocsum) until the file is fully converted,
this brings more special cases to extent handling, it has to survive
power failure and remain consistent, and probably needs to be restarted
after next mount."
Signed-off-by: David Sterba <dsterba@suse.cz>
2012-09-07 18:56:55 +07:00
|
|
|
mode = inode->i_mode;
|
2011-12-29 12:36:45 +07:00
|
|
|
|
2009-04-17 15:37:41 +07:00
|
|
|
flags = btrfs_mask_flags(inode->i_mode, flags);
|
|
|
|
|
oldflags = btrfs_flags_to_ioctl(ip->flags);
|
|
|
|
|
if ((flags ^ oldflags) & (FS_APPEND_FL | FS_IMMUTABLE_FL)) {
|
|
|
|
|
if (!capable(CAP_LINUX_IMMUTABLE)) {
|
|
|
|
|
ret = -EPERM;
|
|
|
|
|
goto out_unlock;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (flags & FS_SYNC_FL)
|
|
|
|
|
ip->flags |= BTRFS_INODE_SYNC;
|
|
|
|
|
else
|
|
|
|
|
ip->flags &= ~BTRFS_INODE_SYNC;
|
|
|
|
|
if (flags & FS_IMMUTABLE_FL)
|
|
|
|
|
ip->flags |= BTRFS_INODE_IMMUTABLE;
|
|
|
|
|
else
|
|
|
|
|
ip->flags &= ~BTRFS_INODE_IMMUTABLE;
|
|
|
|
|
if (flags & FS_APPEND_FL)
|
|
|
|
|
ip->flags |= BTRFS_INODE_APPEND;
|
|
|
|
|
else
|
|
|
|
|
ip->flags &= ~BTRFS_INODE_APPEND;
|
|
|
|
|
if (flags & FS_NODUMP_FL)
|
|
|
|
|
ip->flags |= BTRFS_INODE_NODUMP;
|
|
|
|
|
else
|
|
|
|
|
ip->flags &= ~BTRFS_INODE_NODUMP;
|
|
|
|
|
if (flags & FS_NOATIME_FL)
|
|
|
|
|
ip->flags |= BTRFS_INODE_NOATIME;
|
|
|
|
|
else
|
|
|
|
|
ip->flags &= ~BTRFS_INODE_NOATIME;
|
|
|
|
|
if (flags & FS_DIRSYNC_FL)
|
|
|
|
|
ip->flags |= BTRFS_INODE_DIRSYNC;
|
|
|
|
|
else
|
|
|
|
|
ip->flags &= ~BTRFS_INODE_DIRSYNC;
|
btrfs: allow setting NOCOW for a zero sized file via ioctl
Hi,
the patch si simple, but it has user visible impact and I'm not quite sure how
to resolve it.
In short, $subj says it, chattr -C supports it and we want to use it.
The conditions that acutally allow to change the NOCOW flag are clear. What if
I try to set the flag on a file that is not empty? Options:
1) whole ioctl will fail, EINVAL
2.1) ioctl will succeed, the NOCOW flag will be silently removed, but the file
will stay COW-ed and checksummed
2.2) ioctl will succeed, flag will not be removed and a syslog message will
warn that the COW flag has not been changed
2.2.1) dtto, no syslog message
Man page of chattr states that
"If it is set on a file which already has data blocks, it is undefined when
the blocks assigned to the file will be fully stable."
Yes, it's undefined and with current implementation it'll never happen. So from
this end, the user cannot expect anything. I'm trying to find a reasonable
behaviour, so that a command like 'chattr -R -aijS +C' to tweak a broad set of
flags in a deep directory does not fail unnecessarily and does not pollute the
log.
My personal preference is 2.2.1, but my dev's oppinion is skewed, not counting
the fact that I know the code and otherwise would look there before consulting
the documentation.
The patch implements 2.2.1.
david
-------------8<-------------------
From: David Sterba <dsterba@suse.cz>
It's safe to turn off checksums for a zero sized file.
http://thread.gmane.org/gmane.comp.file-systems.btrfs/18030
"We cannot switch on NODATASUM for a file that already has extents that
are checksummed. The invariant here is that either all the extents or
none are checksummed.
Theoretically it's possible to add/remove all checksums from a given
file, but it's a potentially longtime operation, the file has to be in
some intermediate state where the checksums partially exist but have to
be ignored (for the csum->nocsum) until the file is fully converted,
this brings more special cases to extent handling, it has to survive
power failure and remain consistent, and probably needs to be restarted
after next mount."
Signed-off-by: David Sterba <dsterba@suse.cz>
2012-09-07 18:56:55 +07:00
|
|
|
if (flags & FS_NOCOW_FL) {
|
|
|
|
|
if (S_ISREG(mode)) {
|
|
|
|
|
/*
|
|
|
|
|
* It's safe to turn csums off here, no extents exist.
|
|
|
|
|
* Otherwise we want the flag to reflect the real COW
|
|
|
|
|
* status of the file and will not set it.
|
|
|
|
|
*/
|
|
|
|
|
if (inode->i_size == 0)
|
|
|
|
|
ip->flags |= BTRFS_INODE_NODATACOW
|
|
|
|
|
| BTRFS_INODE_NODATASUM;
|
|
|
|
|
} else {
|
|
|
|
|
ip->flags |= BTRFS_INODE_NODATACOW;
|
|
|
|
|
}
|
|
|
|
|
} else {
|
|
|
|
|
/*
|
2016-05-20 08:18:45 +07:00
|
|
|
* Revert back under same assumptions as above
|
btrfs: allow setting NOCOW for a zero sized file via ioctl
Hi,
the patch si simple, but it has user visible impact and I'm not quite sure how
to resolve it.
In short, $subj says it, chattr -C supports it and we want to use it.
The conditions that acutally allow to change the NOCOW flag are clear. What if
I try to set the flag on a file that is not empty? Options:
1) whole ioctl will fail, EINVAL
2.1) ioctl will succeed, the NOCOW flag will be silently removed, but the file
will stay COW-ed and checksummed
2.2) ioctl will succeed, flag will not be removed and a syslog message will
warn that the COW flag has not been changed
2.2.1) dtto, no syslog message
Man page of chattr states that
"If it is set on a file which already has data blocks, it is undefined when
the blocks assigned to the file will be fully stable."
Yes, it's undefined and with current implementation it'll never happen. So from
this end, the user cannot expect anything. I'm trying to find a reasonable
behaviour, so that a command like 'chattr -R -aijS +C' to tweak a broad set of
flags in a deep directory does not fail unnecessarily and does not pollute the
log.
My personal preference is 2.2.1, but my dev's oppinion is skewed, not counting
the fact that I know the code and otherwise would look there before consulting
the documentation.
The patch implements 2.2.1.
david
-------------8<-------------------
From: David Sterba <dsterba@suse.cz>
It's safe to turn off checksums for a zero sized file.
http://thread.gmane.org/gmane.comp.file-systems.btrfs/18030
"We cannot switch on NODATASUM for a file that already has extents that
are checksummed. The invariant here is that either all the extents or
none are checksummed.
Theoretically it's possible to add/remove all checksums from a given
file, but it's a potentially longtime operation, the file has to be in
some intermediate state where the checksums partially exist but have to
be ignored (for the csum->nocsum) until the file is fully converted,
this brings more special cases to extent handling, it has to survive
power failure and remain consistent, and probably needs to be restarted
after next mount."
Signed-off-by: David Sterba <dsterba@suse.cz>
2012-09-07 18:56:55 +07:00
|
|
|
*/
|
|
|
|
|
if (S_ISREG(mode)) {
|
|
|
|
|
if (inode->i_size == 0)
|
|
|
|
|
ip->flags &= ~(BTRFS_INODE_NODATACOW
|
|
|
|
|
| BTRFS_INODE_NODATASUM);
|
|
|
|
|
} else {
|
|
|
|
|
ip->flags &= ~BTRFS_INODE_NODATACOW;
|
|
|
|
|
}
|
|
|
|
|
}
|
2009-04-17 15:37:41 +07:00
|
|
|
|
Btrfs: Per file/directory controls for COW and compression
Data compression and data cow are controlled across the entire FS by mount
options right now. ioctls are needed to set this on a per file or per
directory basis. This has been proposed previously, but VFS developers
wanted us to use generic ioctls rather than btrfs-specific ones.
According to Chris's comment, there should be just one true compression
method(probably LZO) stored in the super. However, before this, we would
wait for that one method is stable enough to be adopted into the super.
So I list it as a long term goal, and just store it in ram today.
After applying this patch, we can use the generic "FS_IOC_SETFLAGS" ioctl to
control file and directory's datacow and compression attribute.
NOTE:
- The compression type is selected by such rules:
If we mount btrfs with compress options, ie, zlib/lzo, the type is it.
Otherwise, we'll use the default compress type (zlib today).
v1->v2:
- rebase to the latest btrfs.
v2->v3:
- fix a problem, i.e. when a file is set NOCOW via mount option, then this NOCOW
will be screwed by inheritance from parent directory.
Signed-off-by: Liu Bo <liubo2009@cn.fujitsu.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
2011-03-22 17:12:20 +07:00
|
|
|
/*
|
|
|
|
|
* The COMPRESS flag can only be changed by users, while the NOCOMPRESS
|
|
|
|
|
* flag may be changed automatically if compression code won't make
|
|
|
|
|
* things smaller.
|
|
|
|
|
*/
|
|
|
|
|
if (flags & FS_NOCOMP_FL) {
|
|
|
|
|
ip->flags &= ~BTRFS_INODE_COMPRESS;
|
|
|
|
|
ip->flags |= BTRFS_INODE_NOCOMPRESS;
|
Btrfs: add support for inode properties
This change adds infrastructure to allow for generic properties for
inodes. Properties are name/value pairs that can be associated with
inodes for different purposes. They are stored as xattrs with the
prefix "btrfs."
Properties can be inherited - this means when a directory inode has
inheritable properties set, these are added to new inodes created
under that directory. Further, subvolumes can also have properties
associated with them, and they can be inherited from their parent
subvolume. Naturally, directory properties have priority over subvolume
properties (in practice a subvolume property is just a regular
property associated with the root inode, objectid 256, of the
subvolume's fs tree).
This change also adds one specific property implementation, named
"compression", whose values can be "lzo" or "zlib" and it's an
inheritable property.
The corresponding changes to btrfs-progs were also implemented.
A patch with xfstests for this feature will follow once there's
agreement on this change/feature.
Further, the script at the bottom of this commit message was used to
do some benchmarks to measure any performance penalties of this feature.
Basically the tests correspond to:
Test 1 - create a filesystem and mount it with compress-force=lzo,
then sequentially create N files of 64Kb each, measure how long it took
to create the files, unmount the filesystem, mount the filesystem and
perform an 'ls -lha' against the test directory holding the N files, and
report the time the command took.
Test 2 - create a filesystem and don't use any compression option when
mounting it - instead set the compression property of the subvolume's
root to 'lzo'. Then create N files of 64Kb, and report the time it took.
The unmount the filesystem, mount it again and perform an 'ls -lha' like
in the former test. This means every single file ends up with a property
(xattr) associated to it.
Test 3 - same as test 2, but uses 4 properties - 3 are duplicates of the
compression property, have no real effect other than adding more work
when inheriting properties and taking more btree leaf space.
Test 4 - same as test 3 but with 10 properties per file.
Results (in seconds, and averages of 5 runs each), for different N
numbers of files follow.
* Without properties (test 1)
file creation time ls -lha time
10 000 files 3.49 0.76
100 000 files 47.19 8.37
1 000 000 files 518.51 107.06
* With 1 property (compression property set to lzo - test 2)
file creation time ls -lha time
10 000 files 3.63 0.93
100 000 files 48.56 9.74
1 000 000 files 537.72 125.11
* With 4 properties (test 3)
file creation time ls -lha time
10 000 files 3.94 1.20
100 000 files 52.14 11.48
1 000 000 files 572.70 142.13
* With 10 properties (test 4)
file creation time ls -lha time
10 000 files 4.61 1.35
100 000 files 58.86 13.83
1 000 000 files 656.01 177.61
The increased latencies with properties are essencialy because of:
*) When creating an inode, we now synchronously write 1 more item
(an xattr item) for each property inherited from the parent dir
(or subvolume). This could be done in an asynchronous way such
as we do for dir intex items (delayed-inode.c), which could help
reduce the file creation latency;
*) With properties, we now have larger fs trees. For this particular
test each xattr item uses 75 bytes of leaf space in the fs tree.
This could be less by using a new item for xattr items, instead of
the current btrfs_dir_item, since we could cut the 'location' and
'type' fields (saving 18 bytes) and maybe 'transid' too (saving a
total of 26 bytes per xattr item) from the btrfs_dir_item type.
Also tried batching the xattr insertions (ignoring proper hash
collision handling, since it didn't exist) when creating files that
inherit properties from their parent inode/subvolume, but the end
results were (surprisingly) essentially the same.
Test script:
$ cat test.pl
#!/usr/bin/perl -w
use strict;
use Time::HiRes qw(time);
use constant NUM_FILES => 10_000;
use constant FILE_SIZES => (64 * 1024);
use constant DEV => '/dev/sdb4';
use constant MNT_POINT => '/home/fdmanana/btrfs-tests/dev';
use constant TEST_DIR => (MNT_POINT . '/testdir');
system("mkfs.btrfs", "-l", "16384", "-f", DEV) == 0 or die "mkfs.btrfs failed!";
# following line for testing without properties
#system("mount", "-o", "compress-force=lzo", DEV, MNT_POINT) == 0 or die "mount failed!";
# following 2 lines for testing with properties
system("mount", DEV, MNT_POINT) == 0 or die "mount failed!";
system("btrfs", "prop", "set", MNT_POINT, "compression", "lzo") == 0 or die "set prop failed!";
system("mkdir", TEST_DIR) == 0 or die "mkdir failed!";
my ($t1, $t2);
$t1 = time();
for (my $i = 1; $i <= NUM_FILES; $i++) {
my $p = TEST_DIR . '/file_' . $i;
open(my $f, '>', $p) or die "Error opening file!";
$f->autoflush(1);
for (my $j = 0; $j < FILE_SIZES; $j += 4096) {
print $f ('A' x 4096) or die "Error writing to file!";
}
close($f);
}
$t2 = time();
print "Time to create " . NUM_FILES . ": " . ($t2 - $t1) . " seconds.\n";
system("umount", DEV) == 0 or die "umount failed!";
system("mount", DEV, MNT_POINT) == 0 or die "mount failed!";
$t1 = time();
system("bash -c 'ls -lha " . TEST_DIR . " > /dev/null'") == 0 or die "ls failed!";
$t2 = time();
print "Time to ls -lha all files: " . ($t2 - $t1) . " seconds.\n";
system("umount", DEV) == 0 or die "umount failed!";
Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com>
Signed-off-by: Josef Bacik <jbacik@fb.com>
Signed-off-by: Chris Mason <clm@fb.com>
2014-01-07 18:47:46 +07:00
|
|
|
|
|
|
|
|
ret = btrfs_set_prop(inode, "btrfs.compression", NULL, 0, 0);
|
|
|
|
|
if (ret && ret != -ENODATA)
|
|
|
|
|
goto out_drop;
|
Btrfs: Per file/directory controls for COW and compression
Data compression and data cow are controlled across the entire FS by mount
options right now. ioctls are needed to set this on a per file or per
directory basis. This has been proposed previously, but VFS developers
wanted us to use generic ioctls rather than btrfs-specific ones.
According to Chris's comment, there should be just one true compression
method(probably LZO) stored in the super. However, before this, we would
wait for that one method is stable enough to be adopted into the super.
So I list it as a long term goal, and just store it in ram today.
After applying this patch, we can use the generic "FS_IOC_SETFLAGS" ioctl to
control file and directory's datacow and compression attribute.
NOTE:
- The compression type is selected by such rules:
If we mount btrfs with compress options, ie, zlib/lzo, the type is it.
Otherwise, we'll use the default compress type (zlib today).
v1->v2:
- rebase to the latest btrfs.
v2->v3:
- fix a problem, i.e. when a file is set NOCOW via mount option, then this NOCOW
will be screwed by inheritance from parent directory.
Signed-off-by: Liu Bo <liubo2009@cn.fujitsu.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
2011-03-22 17:12:20 +07:00
|
|
|
} else if (flags & FS_COMPR_FL) {
|
Btrfs: add support for inode properties
This change adds infrastructure to allow for generic properties for
inodes. Properties are name/value pairs that can be associated with
inodes for different purposes. They are stored as xattrs with the
prefix "btrfs."
Properties can be inherited - this means when a directory inode has
inheritable properties set, these are added to new inodes created
under that directory. Further, subvolumes can also have properties
associated with them, and they can be inherited from their parent
subvolume. Naturally, directory properties have priority over subvolume
properties (in practice a subvolume property is just a regular
property associated with the root inode, objectid 256, of the
subvolume's fs tree).
This change also adds one specific property implementation, named
"compression", whose values can be "lzo" or "zlib" and it's an
inheritable property.
The corresponding changes to btrfs-progs were also implemented.
A patch with xfstests for this feature will follow once there's
agreement on this change/feature.
Further, the script at the bottom of this commit message was used to
do some benchmarks to measure any performance penalties of this feature.
Basically the tests correspond to:
Test 1 - create a filesystem and mount it with compress-force=lzo,
then sequentially create N files of 64Kb each, measure how long it took
to create the files, unmount the filesystem, mount the filesystem and
perform an 'ls -lha' against the test directory holding the N files, and
report the time the command took.
Test 2 - create a filesystem and don't use any compression option when
mounting it - instead set the compression property of the subvolume's
root to 'lzo'. Then create N files of 64Kb, and report the time it took.
The unmount the filesystem, mount it again and perform an 'ls -lha' like
in the former test. This means every single file ends up with a property
(xattr) associated to it.
Test 3 - same as test 2, but uses 4 properties - 3 are duplicates of the
compression property, have no real effect other than adding more work
when inheriting properties and taking more btree leaf space.
Test 4 - same as test 3 but with 10 properties per file.
Results (in seconds, and averages of 5 runs each), for different N
numbers of files follow.
* Without properties (test 1)
file creation time ls -lha time
10 000 files 3.49 0.76
100 000 files 47.19 8.37
1 000 000 files 518.51 107.06
* With 1 property (compression property set to lzo - test 2)
file creation time ls -lha time
10 000 files 3.63 0.93
100 000 files 48.56 9.74
1 000 000 files 537.72 125.11
* With 4 properties (test 3)
file creation time ls -lha time
10 000 files 3.94 1.20
100 000 files 52.14 11.48
1 000 000 files 572.70 142.13
* With 10 properties (test 4)
file creation time ls -lha time
10 000 files 4.61 1.35
100 000 files 58.86 13.83
1 000 000 files 656.01 177.61
The increased latencies with properties are essencialy because of:
*) When creating an inode, we now synchronously write 1 more item
(an xattr item) for each property inherited from the parent dir
(or subvolume). This could be done in an asynchronous way such
as we do for dir intex items (delayed-inode.c), which could help
reduce the file creation latency;
*) With properties, we now have larger fs trees. For this particular
test each xattr item uses 75 bytes of leaf space in the fs tree.
This could be less by using a new item for xattr items, instead of
the current btrfs_dir_item, since we could cut the 'location' and
'type' fields (saving 18 bytes) and maybe 'transid' too (saving a
total of 26 bytes per xattr item) from the btrfs_dir_item type.
Also tried batching the xattr insertions (ignoring proper hash
collision handling, since it didn't exist) when creating files that
inherit properties from their parent inode/subvolume, but the end
results were (surprisingly) essentially the same.
Test script:
$ cat test.pl
#!/usr/bin/perl -w
use strict;
use Time::HiRes qw(time);
use constant NUM_FILES => 10_000;
use constant FILE_SIZES => (64 * 1024);
use constant DEV => '/dev/sdb4';
use constant MNT_POINT => '/home/fdmanana/btrfs-tests/dev';
use constant TEST_DIR => (MNT_POINT . '/testdir');
system("mkfs.btrfs", "-l", "16384", "-f", DEV) == 0 or die "mkfs.btrfs failed!";
# following line for testing without properties
#system("mount", "-o", "compress-force=lzo", DEV, MNT_POINT) == 0 or die "mount failed!";
# following 2 lines for testing with properties
system("mount", DEV, MNT_POINT) == 0 or die "mount failed!";
system("btrfs", "prop", "set", MNT_POINT, "compression", "lzo") == 0 or die "set prop failed!";
system("mkdir", TEST_DIR) == 0 or die "mkdir failed!";
my ($t1, $t2);
$t1 = time();
for (my $i = 1; $i <= NUM_FILES; $i++) {
my $p = TEST_DIR . '/file_' . $i;
open(my $f, '>', $p) or die "Error opening file!";
$f->autoflush(1);
for (my $j = 0; $j < FILE_SIZES; $j += 4096) {
print $f ('A' x 4096) or die "Error writing to file!";
}
close($f);
}
$t2 = time();
print "Time to create " . NUM_FILES . ": " . ($t2 - $t1) . " seconds.\n";
system("umount", DEV) == 0 or die "umount failed!";
system("mount", DEV, MNT_POINT) == 0 or die "mount failed!";
$t1 = time();
system("bash -c 'ls -lha " . TEST_DIR . " > /dev/null'") == 0 or die "ls failed!";
$t2 = time();
print "Time to ls -lha all files: " . ($t2 - $t1) . " seconds.\n";
system("umount", DEV) == 0 or die "umount failed!";
Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com>
Signed-off-by: Josef Bacik <jbacik@fb.com>
Signed-off-by: Chris Mason <clm@fb.com>
2014-01-07 18:47:46 +07:00
|
|
|
const char *comp;
|
|
|
|
|
|
Btrfs: Per file/directory controls for COW and compression
Data compression and data cow are controlled across the entire FS by mount
options right now. ioctls are needed to set this on a per file or per
directory basis. This has been proposed previously, but VFS developers
wanted us to use generic ioctls rather than btrfs-specific ones.
According to Chris's comment, there should be just one true compression
method(probably LZO) stored in the super. However, before this, we would
wait for that one method is stable enough to be adopted into the super.
So I list it as a long term goal, and just store it in ram today.
After applying this patch, we can use the generic "FS_IOC_SETFLAGS" ioctl to
control file and directory's datacow and compression attribute.
NOTE:
- The compression type is selected by such rules:
If we mount btrfs with compress options, ie, zlib/lzo, the type is it.
Otherwise, we'll use the default compress type (zlib today).
v1->v2:
- rebase to the latest btrfs.
v2->v3:
- fix a problem, i.e. when a file is set NOCOW via mount option, then this NOCOW
will be screwed by inheritance from parent directory.
Signed-off-by: Liu Bo <liubo2009@cn.fujitsu.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
2011-03-22 17:12:20 +07:00
|
|
|
ip->flags |= BTRFS_INODE_COMPRESS;
|
|
|
|
|
ip->flags &= ~BTRFS_INODE_NOCOMPRESS;
|
Btrfs: add support for inode properties
This change adds infrastructure to allow for generic properties for
inodes. Properties are name/value pairs that can be associated with
inodes for different purposes. They are stored as xattrs with the
prefix "btrfs."
Properties can be inherited - this means when a directory inode has
inheritable properties set, these are added to new inodes created
under that directory. Further, subvolumes can also have properties
associated with them, and they can be inherited from their parent
subvolume. Naturally, directory properties have priority over subvolume
properties (in practice a subvolume property is just a regular
property associated with the root inode, objectid 256, of the
subvolume's fs tree).
This change also adds one specific property implementation, named
"compression", whose values can be "lzo" or "zlib" and it's an
inheritable property.
The corresponding changes to btrfs-progs were also implemented.
A patch with xfstests for this feature will follow once there's
agreement on this change/feature.
Further, the script at the bottom of this commit message was used to
do some benchmarks to measure any performance penalties of this feature.
Basically the tests correspond to:
Test 1 - create a filesystem and mount it with compress-force=lzo,
then sequentially create N files of 64Kb each, measure how long it took
to create the files, unmount the filesystem, mount the filesystem and
perform an 'ls -lha' against the test directory holding the N files, and
report the time the command took.
Test 2 - create a filesystem and don't use any compression option when
mounting it - instead set the compression property of the subvolume's
root to 'lzo'. Then create N files of 64Kb, and report the time it took.
The unmount the filesystem, mount it again and perform an 'ls -lha' like
in the former test. This means every single file ends up with a property
(xattr) associated to it.
Test 3 - same as test 2, but uses 4 properties - 3 are duplicates of the
compression property, have no real effect other than adding more work
when inheriting properties and taking more btree leaf space.
Test 4 - same as test 3 but with 10 properties per file.
Results (in seconds, and averages of 5 runs each), for different N
numbers of files follow.
* Without properties (test 1)
file creation time ls -lha time
10 000 files 3.49 0.76
100 000 files 47.19 8.37
1 000 000 files 518.51 107.06
* With 1 property (compression property set to lzo - test 2)
file creation time ls -lha time
10 000 files 3.63 0.93
100 000 files 48.56 9.74
1 000 000 files 537.72 125.11
* With 4 properties (test 3)
file creation time ls -lha time
10 000 files 3.94 1.20
100 000 files 52.14 11.48
1 000 000 files 572.70 142.13
* With 10 properties (test 4)
file creation time ls -lha time
10 000 files 4.61 1.35
100 000 files 58.86 13.83
1 000 000 files 656.01 177.61
The increased latencies with properties are essencialy because of:
*) When creating an inode, we now synchronously write 1 more item
(an xattr item) for each property inherited from the parent dir
(or subvolume). This could be done in an asynchronous way such
as we do for dir intex items (delayed-inode.c), which could help
reduce the file creation latency;
*) With properties, we now have larger fs trees. For this particular
test each xattr item uses 75 bytes of leaf space in the fs tree.
This could be less by using a new item for xattr items, instead of
the current btrfs_dir_item, since we could cut the 'location' and
'type' fields (saving 18 bytes) and maybe 'transid' too (saving a
total of 26 bytes per xattr item) from the btrfs_dir_item type.
Also tried batching the xattr insertions (ignoring proper hash
collision handling, since it didn't exist) when creating files that
inherit properties from their parent inode/subvolume, but the end
results were (surprisingly) essentially the same.
Test script:
$ cat test.pl
#!/usr/bin/perl -w
use strict;
use Time::HiRes qw(time);
use constant NUM_FILES => 10_000;
use constant FILE_SIZES => (64 * 1024);
use constant DEV => '/dev/sdb4';
use constant MNT_POINT => '/home/fdmanana/btrfs-tests/dev';
use constant TEST_DIR => (MNT_POINT . '/testdir');
system("mkfs.btrfs", "-l", "16384", "-f", DEV) == 0 or die "mkfs.btrfs failed!";
# following line for testing without properties
#system("mount", "-o", "compress-force=lzo", DEV, MNT_POINT) == 0 or die "mount failed!";
# following 2 lines for testing with properties
system("mount", DEV, MNT_POINT) == 0 or die "mount failed!";
system("btrfs", "prop", "set", MNT_POINT, "compression", "lzo") == 0 or die "set prop failed!";
system("mkdir", TEST_DIR) == 0 or die "mkdir failed!";
my ($t1, $t2);
$t1 = time();
for (my $i = 1; $i <= NUM_FILES; $i++) {
my $p = TEST_DIR . '/file_' . $i;
open(my $f, '>', $p) or die "Error opening file!";
$f->autoflush(1);
for (my $j = 0; $j < FILE_SIZES; $j += 4096) {
print $f ('A' x 4096) or die "Error writing to file!";
}
close($f);
}
$t2 = time();
print "Time to create " . NUM_FILES . ": " . ($t2 - $t1) . " seconds.\n";
system("umount", DEV) == 0 or die "umount failed!";
system("mount", DEV, MNT_POINT) == 0 or die "mount failed!";
$t1 = time();
system("bash -c 'ls -lha " . TEST_DIR . " > /dev/null'") == 0 or die "ls failed!";
$t2 = time();
print "Time to ls -lha all files: " . ($t2 - $t1) . " seconds.\n";
system("umount", DEV) == 0 or die "umount failed!";
Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com>
Signed-off-by: Josef Bacik <jbacik@fb.com>
Signed-off-by: Chris Mason <clm@fb.com>
2014-01-07 18:47:46 +07:00
|
|
|
|
2017-10-31 23:32:41 +07:00
|
|
|
comp = btrfs_compress_type2str(fs_info->compress_type);
|
|
|
|
|
if (!comp || comp[0] == 0)
|
|
|
|
|
comp = btrfs_compress_type2str(BTRFS_COMPRESS_ZLIB);
|
|
|
|
|
|
Btrfs: add support for inode properties
This change adds infrastructure to allow for generic properties for
inodes. Properties are name/value pairs that can be associated with
inodes for different purposes. They are stored as xattrs with the
prefix "btrfs."
Properties can be inherited - this means when a directory inode has
inheritable properties set, these are added to new inodes created
under that directory. Further, subvolumes can also have properties
associated with them, and they can be inherited from their parent
subvolume. Naturally, directory properties have priority over subvolume
properties (in practice a subvolume property is just a regular
property associated with the root inode, objectid 256, of the
subvolume's fs tree).
This change also adds one specific property implementation, named
"compression", whose values can be "lzo" or "zlib" and it's an
inheritable property.
The corresponding changes to btrfs-progs were also implemented.
A patch with xfstests for this feature will follow once there's
agreement on this change/feature.
Further, the script at the bottom of this commit message was used to
do some benchmarks to measure any performance penalties of this feature.
Basically the tests correspond to:
Test 1 - create a filesystem and mount it with compress-force=lzo,
then sequentially create N files of 64Kb each, measure how long it took
to create the files, unmount the filesystem, mount the filesystem and
perform an 'ls -lha' against the test directory holding the N files, and
report the time the command took.
Test 2 - create a filesystem and don't use any compression option when
mounting it - instead set the compression property of the subvolume's
root to 'lzo'. Then create N files of 64Kb, and report the time it took.
The unmount the filesystem, mount it again and perform an 'ls -lha' like
in the former test. This means every single file ends up with a property
(xattr) associated to it.
Test 3 - same as test 2, but uses 4 properties - 3 are duplicates of the
compression property, have no real effect other than adding more work
when inheriting properties and taking more btree leaf space.
Test 4 - same as test 3 but with 10 properties per file.
Results (in seconds, and averages of 5 runs each), for different N
numbers of files follow.
* Without properties (test 1)
file creation time ls -lha time
10 000 files 3.49 0.76
100 000 files 47.19 8.37
1 000 000 files 518.51 107.06
* With 1 property (compression property set to lzo - test 2)
file creation time ls -lha time
10 000 files 3.63 0.93
100 000 files 48.56 9.74
1 000 000 files 537.72 125.11
* With 4 properties (test 3)
file creation time ls -lha time
10 000 files 3.94 1.20
100 000 files 52.14 11.48
1 000 000 files 572.70 142.13
* With 10 properties (test 4)
file creation time ls -lha time
10 000 files 4.61 1.35
100 000 files 58.86 13.83
1 000 000 files 656.01 177.61
The increased latencies with properties are essencialy because of:
*) When creating an inode, we now synchronously write 1 more item
(an xattr item) for each property inherited from the parent dir
(or subvolume). This could be done in an asynchronous way such
as we do for dir intex items (delayed-inode.c), which could help
reduce the file creation latency;
*) With properties, we now have larger fs trees. For this particular
test each xattr item uses 75 bytes of leaf space in the fs tree.
This could be less by using a new item for xattr items, instead of
the current btrfs_dir_item, since we could cut the 'location' and
'type' fields (saving 18 bytes) and maybe 'transid' too (saving a
total of 26 bytes per xattr item) from the btrfs_dir_item type.
Also tried batching the xattr insertions (ignoring proper hash
collision handling, since it didn't exist) when creating files that
inherit properties from their parent inode/subvolume, but the end
results were (surprisingly) essentially the same.
Test script:
$ cat test.pl
#!/usr/bin/perl -w
use strict;
use Time::HiRes qw(time);
use constant NUM_FILES => 10_000;
use constant FILE_SIZES => (64 * 1024);
use constant DEV => '/dev/sdb4';
use constant MNT_POINT => '/home/fdmanana/btrfs-tests/dev';
use constant TEST_DIR => (MNT_POINT . '/testdir');
system("mkfs.btrfs", "-l", "16384", "-f", DEV) == 0 or die "mkfs.btrfs failed!";
# following line for testing without properties
#system("mount", "-o", "compress-force=lzo", DEV, MNT_POINT) == 0 or die "mount failed!";
# following 2 lines for testing with properties
system("mount", DEV, MNT_POINT) == 0 or die "mount failed!";
system("btrfs", "prop", "set", MNT_POINT, "compression", "lzo") == 0 or die "set prop failed!";
system("mkdir", TEST_DIR) == 0 or die "mkdir failed!";
my ($t1, $t2);
$t1 = time();
for (my $i = 1; $i <= NUM_FILES; $i++) {
my $p = TEST_DIR . '/file_' . $i;
open(my $f, '>', $p) or die "Error opening file!";
$f->autoflush(1);
for (my $j = 0; $j < FILE_SIZES; $j += 4096) {
print $f ('A' x 4096) or die "Error writing to file!";
}
close($f);
}
$t2 = time();
print "Time to create " . NUM_FILES . ": " . ($t2 - $t1) . " seconds.\n";
system("umount", DEV) == 0 or die "umount failed!";
system("mount", DEV, MNT_POINT) == 0 or die "mount failed!";
$t1 = time();
system("bash -c 'ls -lha " . TEST_DIR . " > /dev/null'") == 0 or die "ls failed!";
$t2 = time();
print "Time to ls -lha all files: " . ($t2 - $t1) . " seconds.\n";
system("umount", DEV) == 0 or die "umount failed!";
Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com>
Signed-off-by: Josef Bacik <jbacik@fb.com>
Signed-off-by: Chris Mason <clm@fb.com>
2014-01-07 18:47:46 +07:00
|
|
|
ret = btrfs_set_prop(inode, "btrfs.compression",
|
|
|
|
|
comp, strlen(comp), 0);
|
|
|
|
|
if (ret)
|
|
|
|
|
goto out_drop;
|
|
|
|
|
|
2011-04-15 10:03:17 +07:00
|
|
|
} else {
|
2014-09-11 17:44:49 +07:00
|
|
|
ret = btrfs_set_prop(inode, "btrfs.compression", NULL, 0, 0);
|
|
|
|
|
if (ret && ret != -ENODATA)
|
|
|
|
|
goto out_drop;
|
2011-04-15 10:03:17 +07:00
|
|
|
ip->flags &= ~(BTRFS_INODE_COMPRESS | BTRFS_INODE_NOCOMPRESS);
|
Btrfs: Per file/directory controls for COW and compression
Data compression and data cow are controlled across the entire FS by mount
options right now. ioctls are needed to set this on a per file or per
directory basis. This has been proposed previously, but VFS developers
wanted us to use generic ioctls rather than btrfs-specific ones.
According to Chris's comment, there should be just one true compression
method(probably LZO) stored in the super. However, before this, we would
wait for that one method is stable enough to be adopted into the super.
So I list it as a long term goal, and just store it in ram today.
After applying this patch, we can use the generic "FS_IOC_SETFLAGS" ioctl to
control file and directory's datacow and compression attribute.
NOTE:
- The compression type is selected by such rules:
If we mount btrfs with compress options, ie, zlib/lzo, the type is it.
Otherwise, we'll use the default compress type (zlib today).
v1->v2:
- rebase to the latest btrfs.
v2->v3:
- fix a problem, i.e. when a file is set NOCOW via mount option, then this NOCOW
will be screwed by inheritance from parent directory.
Signed-off-by: Liu Bo <liubo2009@cn.fujitsu.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
2011-03-22 17:12:20 +07:00
|
|
|
}
|
2009-04-17 15:37:41 +07:00
|
|
|
|
2011-12-29 12:39:50 +07:00
|
|
|
trans = btrfs_start_transaction(root, 1);
|
2011-12-29 12:36:45 +07:00
|
|
|
if (IS_ERR(trans)) {
|
|
|
|
|
ret = PTR_ERR(trans);
|
|
|
|
|
goto out_drop;
|
|
|
|
|
}
|
2009-04-17 15:37:41 +07:00
|
|
|
|
2011-12-15 08:12:02 +07:00
|
|
|
btrfs_update_iflags(inode);
|
2012-04-06 02:03:02 +07:00
|
|
|
inode_inc_iversion(inode);
|
2016-09-14 21:48:06 +07:00
|
|
|
inode->i_ctime = current_time(inode);
|
2009-04-17 15:37:41 +07:00
|
|
|
ret = btrfs_update_inode(trans, root, inode);
|
|
|
|
|
|
2016-09-10 08:39:03 +07:00
|
|
|
btrfs_end_transaction(trans);
|
2011-12-29 12:36:45 +07:00
|
|
|
out_drop:
|
|
|
|
|
if (ret) {
|
|
|
|
|
ip->flags = ip_oldflags;
|
|
|
|
|
inode->i_flags = i_oldflags;
|
|
|
|
|
}
|
2009-04-17 15:37:41 +07:00
|
|
|
|
|
|
|
|
out_unlock:
|
2016-01-23 03:40:57 +07:00
|
|
|
inode_unlock(inode);
|
2012-06-12 21:20:32 +07:00
|
|
|
mnt_drop_write_file(file);
|
2011-02-24 16:38:16 +07:00
|
|
|
return ret;
|
2009-04-17 15:37:41 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static int btrfs_ioctl_getversion(struct file *file, int __user *arg)
|
|
|
|
|
{
|
2013-01-24 05:07:38 +07:00
|
|
|
struct inode *inode = file_inode(file);
|
2009-04-17 15:37:41 +07:00
|
|
|
|
|
|
|
|
return put_user(inode->i_generation, arg);
|
|
|
|
|
}
|
2008-06-12 08:53:53 +07:00
|
|
|
|
2011-03-24 17:24:28 +07:00
|
|
|
static noinline int btrfs_ioctl_fitrim(struct file *file, void __user *arg)
|
|
|
|
|
{
|
2016-06-23 05:54:23 +07:00
|
|
|
struct inode *inode = file_inode(file);
|
|
|
|
|
struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
|
2011-03-24 17:24:28 +07:00
|
|
|
struct btrfs_device *device;
|
|
|
|
|
struct request_queue *q;
|
|
|
|
|
struct fstrim_range range;
|
|
|
|
|
u64 minlen = ULLONG_MAX;
|
|
|
|
|
u64 num_devices = 0;
|
2011-11-18 03:40:49 +07:00
|
|
|
u64 total_bytes = btrfs_super_total_bytes(fs_info->super_copy);
|
2011-03-24 17:24:28 +07:00
|
|
|
int ret;
|
|
|
|
|
|
|
|
|
|
if (!capable(CAP_SYS_ADMIN))
|
|
|
|
|
return -EPERM;
|
|
|
|
|
|
2011-04-20 17:09:16 +07:00
|
|
|
rcu_read_lock();
|
|
|
|
|
list_for_each_entry_rcu(device, &fs_info->fs_devices->devices,
|
|
|
|
|
dev_list) {
|
2011-03-24 17:24:28 +07:00
|
|
|
if (!device->bdev)
|
|
|
|
|
continue;
|
|
|
|
|
q = bdev_get_queue(device->bdev);
|
|
|
|
|
if (blk_queue_discard(q)) {
|
|
|
|
|
num_devices++;
|
2016-12-15 20:38:28 +07:00
|
|
|
minlen = min_t(u64, q->limits.discard_granularity,
|
2011-03-24 17:24:28 +07:00
|
|
|
minlen);
|
|
|
|
|
}
|
|
|
|
|
}
|
2011-04-20 17:09:16 +07:00
|
|
|
rcu_read_unlock();
|
2011-09-05 21:34:54 +07:00
|
|
|
|
2011-03-24 17:24:28 +07:00
|
|
|
if (!num_devices)
|
|
|
|
|
return -EOPNOTSUPP;
|
|
|
|
|
if (copy_from_user(&range, arg, sizeof(range)))
|
|
|
|
|
return -EFAULT;
|
2012-10-16 16:34:36 +07:00
|
|
|
if (range.start > total_bytes ||
|
|
|
|
|
range.len < fs_info->sb->s_blocksize)
|
2011-09-05 21:34:54 +07:00
|
|
|
return -EINVAL;
|
2011-03-24 17:24:28 +07:00
|
|
|
|
2011-09-05 21:34:54 +07:00
|
|
|
range.len = min(range.len, total_bytes - range.start);
|
2011-03-24 17:24:28 +07:00
|
|
|
range.minlen = max(range.minlen, minlen);
|
2016-06-23 05:54:24 +07:00
|
|
|
ret = btrfs_trim_fs(fs_info, &range);
|
2011-03-24 17:24:28 +07:00
|
|
|
if (ret < 0)
|
|
|
|
|
return ret;
|
|
|
|
|
|
|
|
|
|
if (copy_to_user(arg, &range, sizeof(range)))
|
|
|
|
|
return -EFAULT;
|
|
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
2013-08-15 22:11:20 +07:00
|
|
|
int btrfs_is_empty_uuid(u8 *uuid)
|
|
|
|
|
{
|
2013-11-15 18:14:55 +07:00
|
|
|
int i;
|
|
|
|
|
|
|
|
|
|
for (i = 0; i < BTRFS_UUID_SIZE; i++) {
|
|
|
|
|
if (uuid[i])
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
return 1;
|
2013-08-15 22:11:20 +07:00
|
|
|
}
|
|
|
|
|
|
2013-02-28 17:04:33 +07:00
|
|
|
static noinline int create_subvol(struct inode *dir,
|
2008-10-10 00:39:39 +07:00
|
|
|
struct dentry *dentry,
|
2017-02-15 00:33:53 +07:00
|
|
|
const char *name, int namelen,
|
2011-09-14 20:58:21 +07:00
|
|
|
u64 *async_transid,
|
2013-02-07 13:02:44 +07:00
|
|
|
struct btrfs_qgroup_inherit *inherit)
|
2008-06-12 08:53:53 +07:00
|
|
|
{
|
2016-06-23 05:54:23 +07:00
|
|
|
struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb);
|
2008-06-12 08:53:53 +07:00
|
|
|
struct btrfs_trans_handle *trans;
|
|
|
|
|
struct btrfs_key key;
|
2016-03-24 23:49:22 +07:00
|
|
|
struct btrfs_root_item *root_item;
|
2008-06-12 08:53:53 +07:00
|
|
|
struct btrfs_inode_item *inode_item;
|
|
|
|
|
struct extent_buffer *leaf;
|
2013-02-28 17:04:33 +07:00
|
|
|
struct btrfs_root *root = BTRFS_I(dir)->root;
|
2009-09-22 03:00:26 +07:00
|
|
|
struct btrfs_root *new_root;
|
2013-02-28 17:04:33 +07:00
|
|
|
struct btrfs_block_rsv block_rsv;
|
2016-09-14 21:48:06 +07:00
|
|
|
struct timespec cur_time = current_time(dir);
|
2013-12-13 07:51:42 +07:00
|
|
|
struct inode *inode;
|
2008-06-12 08:53:53 +07:00
|
|
|
int ret;
|
|
|
|
|
int err;
|
|
|
|
|
u64 objectid;
|
|
|
|
|
u64 new_dirid = BTRFS_FIRST_FREE_OBJECTID;
|
2008-11-18 09:02:50 +07:00
|
|
|
u64 index = 0;
|
2013-02-28 17:04:33 +07:00
|
|
|
u64 qgroup_reserved;
|
2012-07-25 22:35:53 +07:00
|
|
|
uuid_le new_uuid;
|
2008-06-12 08:53:53 +07:00
|
|
|
|
2016-03-24 23:49:22 +07:00
|
|
|
root_item = kzalloc(sizeof(*root_item), GFP_KERNEL);
|
|
|
|
|
if (!root_item)
|
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
2016-06-23 05:54:23 +07:00
|
|
|
ret = btrfs_find_free_objectid(fs_info->tree_root, &objectid);
|
2011-07-17 08:38:06 +07:00
|
|
|
if (ret)
|
2016-03-24 23:49:22 +07:00
|
|
|
goto fail_free;
|
2010-11-20 16:48:00 +07:00
|
|
|
|
2015-02-27 15:24:23 +07:00
|
|
|
/*
|
|
|
|
|
* Don't create subvolume whose level is not zero. Or qgroup will be
|
2016-05-20 08:18:45 +07:00
|
|
|
* screwed up since it assumes subvolume qgroup's level to be 0.
|
2015-02-27 15:24:23 +07:00
|
|
|
*/
|
2016-03-24 23:49:22 +07:00
|
|
|
if (btrfs_qgroup_level(objectid)) {
|
|
|
|
|
ret = -ENOSPC;
|
|
|
|
|
goto fail_free;
|
|
|
|
|
}
|
2015-02-27 15:24:23 +07:00
|
|
|
|
2013-02-28 17:04:33 +07:00
|
|
|
btrfs_init_block_rsv(&block_rsv, BTRFS_BLOCK_RSV_TEMP);
|
2009-09-12 03:12:44 +07:00
|
|
|
/*
|
2013-02-28 17:04:33 +07:00
|
|
|
* The same as the snapshot creation, please see the comment
|
|
|
|
|
* of create_snapshot().
|
2009-09-12 03:12:44 +07:00
|
|
|
*/
|
2013-02-28 17:04:33 +07:00
|
|
|
ret = btrfs_subvolume_reserve_metadata(root, &block_rsv,
|
2013-08-15 22:11:20 +07:00
|
|
|
8, &qgroup_reserved, false);
|
2013-02-28 17:04:33 +07:00
|
|
|
if (ret)
|
2016-03-24 23:49:22 +07:00
|
|
|
goto fail_free;
|
2013-02-28 17:04:33 +07:00
|
|
|
|
|
|
|
|
trans = btrfs_start_transaction(root, 0);
|
|
|
|
|
if (IS_ERR(trans)) {
|
|
|
|
|
ret = PTR_ERR(trans);
|
2017-02-11 01:18:18 +07:00
|
|
|
btrfs_subvolume_release_metadata(fs_info, &block_rsv);
|
2016-03-24 23:49:22 +07:00
|
|
|
goto fail_free;
|
2013-02-28 17:04:33 +07:00
|
|
|
}
|
|
|
|
|
trans->block_rsv = &block_rsv;
|
|
|
|
|
trans->bytes_reserved = block_rsv.size;
|
2008-06-12 08:53:53 +07:00
|
|
|
|
2016-06-23 05:54:23 +07:00
|
|
|
ret = btrfs_qgroup_inherit(trans, fs_info, 0, objectid, inherit);
|
2011-09-14 20:58:21 +07:00
|
|
|
if (ret)
|
|
|
|
|
goto fail;
|
|
|
|
|
|
2014-06-15 06:54:12 +07:00
|
|
|
leaf = btrfs_alloc_tree_block(trans, root, 0, objectid, NULL, 0, 0, 0);
|
2008-07-24 23:17:14 +07:00
|
|
|
if (IS_ERR(leaf)) {
|
|
|
|
|
ret = PTR_ERR(leaf);
|
|
|
|
|
goto fail;
|
|
|
|
|
}
|
2008-06-12 08:53:53 +07:00
|
|
|
|
2016-11-09 00:09:03 +07:00
|
|
|
memzero_extent_buffer(leaf, 0, sizeof(struct btrfs_header));
|
2008-06-12 08:53:53 +07:00
|
|
|
btrfs_set_header_bytenr(leaf, leaf->start);
|
|
|
|
|
btrfs_set_header_generation(leaf, trans->transid);
|
Btrfs: Mixed back reference (FORWARD ROLLING FORMAT CHANGE)
This commit introduces a new kind of back reference for btrfs metadata.
Once a filesystem has been mounted with this commit, IT WILL NO LONGER
BE MOUNTABLE BY OLDER KERNELS.
When a tree block in subvolume tree is cow'd, the reference counts of all
extents it points to are increased by one. At transaction commit time,
the old root of the subvolume is recorded in a "dead root" data structure,
and the btree it points to is later walked, dropping reference counts
and freeing any blocks where the reference count goes to 0.
The increments done during cow and decrements done after commit cancel out,
and the walk is a very expensive way to go about freeing the blocks that
are no longer referenced by the new btree root. This commit reduces the
transaction overhead by avoiding the need for dead root records.
When a non-shared tree block is cow'd, we free the old block at once, and the
new block inherits old block's references. When a tree block with reference
count > 1 is cow'd, we increase the reference counts of all extents
the new block points to by one, and decrease the old block's reference count by
one.
This dead tree avoidance code removes the need to modify the reference
counts of lower level extents when a non-shared tree block is cow'd.
But we still need to update back ref for all pointers in the block.
This is because the location of the block is recorded in the back ref
item.
We can solve this by introducing a new type of back ref. The new
back ref provides information about pointer's key, level and in which
tree the pointer lives. This information allow us to find the pointer
by searching the tree. The shortcoming of the new back ref is that it
only works for pointers in tree blocks referenced by their owner trees.
This is mostly a problem for snapshots, where resolving one of these
fuzzy back references would be O(number_of_snapshots) and quite slow.
The solution used here is to use the fuzzy back references in the common
case where a given tree block is only referenced by one root,
and use the full back references when multiple roots have a reference
on a given block.
This commit adds per subvolume red-black tree to keep trace of cached
inodes. The red-black tree helps the balancing code to find cached
inodes whose inode numbers within a given range.
This commit improves the balancing code by introducing several data
structures to keep the state of balancing. The most important one
is the back ref cache. It caches how the upper level tree blocks are
referenced. This greatly reduce the overhead of checking back ref.
The improved balancing code scales significantly better with a large
number of snapshots.
This is a very large commit and was written in a number of
pieces. But, they depend heavily on the disk format change and were
squashed together to make sure git bisect didn't end up in a
bad state wrt space balancing or the format change.
Signed-off-by: Yan Zheng <zheng.yan@oracle.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
2009-06-10 21:45:14 +07:00
|
|
|
btrfs_set_header_backref_rev(leaf, BTRFS_MIXED_BACKREF_REV);
|
2008-06-12 08:53:53 +07:00
|
|
|
btrfs_set_header_owner(leaf, objectid);
|
|
|
|
|
|
2016-06-23 05:54:23 +07:00
|
|
|
write_extent_buffer_fsid(leaf, fs_info->fsid);
|
|
|
|
|
write_extent_buffer_chunk_tree_uuid(leaf, fs_info->chunk_tree_uuid);
|
2008-06-12 08:53:53 +07:00
|
|
|
btrfs_mark_buffer_dirty(leaf);
|
|
|
|
|
|
2016-03-24 23:49:22 +07:00
|
|
|
inode_item = &root_item->inode;
|
2013-07-16 10:19:18 +07:00
|
|
|
btrfs_set_stack_inode_generation(inode_item, 1);
|
|
|
|
|
btrfs_set_stack_inode_size(inode_item, 3);
|
|
|
|
|
btrfs_set_stack_inode_nlink(inode_item, 1);
|
2016-06-15 20:22:56 +07:00
|
|
|
btrfs_set_stack_inode_nbytes(inode_item,
|
2016-06-23 05:54:23 +07:00
|
|
|
fs_info->nodesize);
|
2013-07-16 10:19:18 +07:00
|
|
|
btrfs_set_stack_inode_mode(inode_item, S_IFDIR | 0755);
|
2008-06-12 08:53:53 +07:00
|
|
|
|
2016-03-24 23:49:22 +07:00
|
|
|
btrfs_set_root_flags(root_item, 0);
|
|
|
|
|
btrfs_set_root_limit(root_item, 0);
|
2013-07-16 10:19:18 +07:00
|
|
|
btrfs_set_stack_inode_flags(inode_item, BTRFS_INODE_ROOT_ITEM_INIT);
|
2011-03-28 09:01:25 +07:00
|
|
|
|
2016-03-24 23:49:22 +07:00
|
|
|
btrfs_set_root_bytenr(root_item, leaf->start);
|
|
|
|
|
btrfs_set_root_generation(root_item, trans->transid);
|
|
|
|
|
btrfs_set_root_level(root_item, 0);
|
|
|
|
|
btrfs_set_root_refs(root_item, 1);
|
|
|
|
|
btrfs_set_root_used(root_item, leaf->len);
|
|
|
|
|
btrfs_set_root_last_snapshot(root_item, 0);
|
2008-06-12 08:53:53 +07:00
|
|
|
|
2016-03-24 23:49:22 +07:00
|
|
|
btrfs_set_root_generation_v2(root_item,
|
|
|
|
|
btrfs_root_generation(root_item));
|
2012-07-25 22:35:53 +07:00
|
|
|
uuid_le_gen(&new_uuid);
|
2016-03-24 23:49:22 +07:00
|
|
|
memcpy(root_item->uuid, new_uuid.b, BTRFS_UUID_SIZE);
|
|
|
|
|
btrfs_set_stack_timespec_sec(&root_item->otime, cur_time.tv_sec);
|
|
|
|
|
btrfs_set_stack_timespec_nsec(&root_item->otime, cur_time.tv_nsec);
|
|
|
|
|
root_item->ctime = root_item->otime;
|
|
|
|
|
btrfs_set_root_ctransid(root_item, trans->transid);
|
|
|
|
|
btrfs_set_root_otransid(root_item, trans->transid);
|
2008-06-12 08:53:53 +07:00
|
|
|
|
2008-06-26 03:01:30 +07:00
|
|
|
btrfs_tree_unlock(leaf);
|
2008-06-12 08:53:53 +07:00
|
|
|
free_extent_buffer(leaf);
|
|
|
|
|
leaf = NULL;
|
|
|
|
|
|
2016-03-24 23:49:22 +07:00
|
|
|
btrfs_set_root_dirid(root_item, new_dirid);
|
2008-06-12 08:53:53 +07:00
|
|
|
|
|
|
|
|
key.objectid = objectid;
|
Btrfs: Mixed back reference (FORWARD ROLLING FORMAT CHANGE)
This commit introduces a new kind of back reference for btrfs metadata.
Once a filesystem has been mounted with this commit, IT WILL NO LONGER
BE MOUNTABLE BY OLDER KERNELS.
When a tree block in subvolume tree is cow'd, the reference counts of all
extents it points to are increased by one. At transaction commit time,
the old root of the subvolume is recorded in a "dead root" data structure,
and the btree it points to is later walked, dropping reference counts
and freeing any blocks where the reference count goes to 0.
The increments done during cow and decrements done after commit cancel out,
and the walk is a very expensive way to go about freeing the blocks that
are no longer referenced by the new btree root. This commit reduces the
transaction overhead by avoiding the need for dead root records.
When a non-shared tree block is cow'd, we free the old block at once, and the
new block inherits old block's references. When a tree block with reference
count > 1 is cow'd, we increase the reference counts of all extents
the new block points to by one, and decrease the old block's reference count by
one.
This dead tree avoidance code removes the need to modify the reference
counts of lower level extents when a non-shared tree block is cow'd.
But we still need to update back ref for all pointers in the block.
This is because the location of the block is recorded in the back ref
item.
We can solve this by introducing a new type of back ref. The new
back ref provides information about pointer's key, level and in which
tree the pointer lives. This information allow us to find the pointer
by searching the tree. The shortcoming of the new back ref is that it
only works for pointers in tree blocks referenced by their owner trees.
This is mostly a problem for snapshots, where resolving one of these
fuzzy back references would be O(number_of_snapshots) and quite slow.
The solution used here is to use the fuzzy back references in the common
case where a given tree block is only referenced by one root,
and use the full back references when multiple roots have a reference
on a given block.
This commit adds per subvolume red-black tree to keep trace of cached
inodes. The red-black tree helps the balancing code to find cached
inodes whose inode numbers within a given range.
This commit improves the balancing code by introducing several data
structures to keep the state of balancing. The most important one
is the back ref cache. It caches how the upper level tree blocks are
referenced. This greatly reduce the overhead of checking back ref.
The improved balancing code scales significantly better with a large
number of snapshots.
This is a very large commit and was written in a number of
pieces. But, they depend heavily on the disk format change and were
squashed together to make sure git bisect didn't end up in a
bad state wrt space balancing or the format change.
Signed-off-by: Yan Zheng <zheng.yan@oracle.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
2009-06-10 21:45:14 +07:00
|
|
|
key.offset = 0;
|
2014-06-04 23:41:45 +07:00
|
|
|
key.type = BTRFS_ROOT_ITEM_KEY;
|
2016-06-23 05:54:23 +07:00
|
|
|
ret = btrfs_insert_root(trans, fs_info->tree_root, &key,
|
2016-03-24 23:49:22 +07:00
|
|
|
root_item);
|
2008-06-12 08:53:53 +07:00
|
|
|
if (ret)
|
|
|
|
|
goto fail;
|
|
|
|
|
|
2009-09-22 03:00:26 +07:00
|
|
|
key.offset = (u64)-1;
|
2016-06-23 05:54:23 +07:00
|
|
|
new_root = btrfs_read_fs_root_no_name(fs_info, &key);
|
2012-03-12 22:03:00 +07:00
|
|
|
if (IS_ERR(new_root)) {
|
|
|
|
|
ret = PTR_ERR(new_root);
|
2016-06-11 05:19:25 +07:00
|
|
|
btrfs_abort_transaction(trans, ret);
|
2012-03-12 22:03:00 +07:00
|
|
|
goto fail;
|
|
|
|
|
}
|
2009-09-22 03:00:26 +07:00
|
|
|
|
|
|
|
|
btrfs_record_root_in_trans(trans, new_root);
|
|
|
|
|
|
Btrfs: add support for inode properties
This change adds infrastructure to allow for generic properties for
inodes. Properties are name/value pairs that can be associated with
inodes for different purposes. They are stored as xattrs with the
prefix "btrfs."
Properties can be inherited - this means when a directory inode has
inheritable properties set, these are added to new inodes created
under that directory. Further, subvolumes can also have properties
associated with them, and they can be inherited from their parent
subvolume. Naturally, directory properties have priority over subvolume
properties (in practice a subvolume property is just a regular
property associated with the root inode, objectid 256, of the
subvolume's fs tree).
This change also adds one specific property implementation, named
"compression", whose values can be "lzo" or "zlib" and it's an
inheritable property.
The corresponding changes to btrfs-progs were also implemented.
A patch with xfstests for this feature will follow once there's
agreement on this change/feature.
Further, the script at the bottom of this commit message was used to
do some benchmarks to measure any performance penalties of this feature.
Basically the tests correspond to:
Test 1 - create a filesystem and mount it with compress-force=lzo,
then sequentially create N files of 64Kb each, measure how long it took
to create the files, unmount the filesystem, mount the filesystem and
perform an 'ls -lha' against the test directory holding the N files, and
report the time the command took.
Test 2 - create a filesystem and don't use any compression option when
mounting it - instead set the compression property of the subvolume's
root to 'lzo'. Then create N files of 64Kb, and report the time it took.
The unmount the filesystem, mount it again and perform an 'ls -lha' like
in the former test. This means every single file ends up with a property
(xattr) associated to it.
Test 3 - same as test 2, but uses 4 properties - 3 are duplicates of the
compression property, have no real effect other than adding more work
when inheriting properties and taking more btree leaf space.
Test 4 - same as test 3 but with 10 properties per file.
Results (in seconds, and averages of 5 runs each), for different N
numbers of files follow.
* Without properties (test 1)
file creation time ls -lha time
10 000 files 3.49 0.76
100 000 files 47.19 8.37
1 000 000 files 518.51 107.06
* With 1 property (compression property set to lzo - test 2)
file creation time ls -lha time
10 000 files 3.63 0.93
100 000 files 48.56 9.74
1 000 000 files 537.72 125.11
* With 4 properties (test 3)
file creation time ls -lha time
10 000 files 3.94 1.20
100 000 files 52.14 11.48
1 000 000 files 572.70 142.13
* With 10 properties (test 4)
file creation time ls -lha time
10 000 files 4.61 1.35
100 000 files 58.86 13.83
1 000 000 files 656.01 177.61
The increased latencies with properties are essencialy because of:
*) When creating an inode, we now synchronously write 1 more item
(an xattr item) for each property inherited from the parent dir
(or subvolume). This could be done in an asynchronous way such
as we do for dir intex items (delayed-inode.c), which could help
reduce the file creation latency;
*) With properties, we now have larger fs trees. For this particular
test each xattr item uses 75 bytes of leaf space in the fs tree.
This could be less by using a new item for xattr items, instead of
the current btrfs_dir_item, since we could cut the 'location' and
'type' fields (saving 18 bytes) and maybe 'transid' too (saving a
total of 26 bytes per xattr item) from the btrfs_dir_item type.
Also tried batching the xattr insertions (ignoring proper hash
collision handling, since it didn't exist) when creating files that
inherit properties from their parent inode/subvolume, but the end
results were (surprisingly) essentially the same.
Test script:
$ cat test.pl
#!/usr/bin/perl -w
use strict;
use Time::HiRes qw(time);
use constant NUM_FILES => 10_000;
use constant FILE_SIZES => (64 * 1024);
use constant DEV => '/dev/sdb4';
use constant MNT_POINT => '/home/fdmanana/btrfs-tests/dev';
use constant TEST_DIR => (MNT_POINT . '/testdir');
system("mkfs.btrfs", "-l", "16384", "-f", DEV) == 0 or die "mkfs.btrfs failed!";
# following line for testing without properties
#system("mount", "-o", "compress-force=lzo", DEV, MNT_POINT) == 0 or die "mount failed!";
# following 2 lines for testing with properties
system("mount", DEV, MNT_POINT) == 0 or die "mount failed!";
system("btrfs", "prop", "set", MNT_POINT, "compression", "lzo") == 0 or die "set prop failed!";
system("mkdir", TEST_DIR) == 0 or die "mkdir failed!";
my ($t1, $t2);
$t1 = time();
for (my $i = 1; $i <= NUM_FILES; $i++) {
my $p = TEST_DIR . '/file_' . $i;
open(my $f, '>', $p) or die "Error opening file!";
$f->autoflush(1);
for (my $j = 0; $j < FILE_SIZES; $j += 4096) {
print $f ('A' x 4096) or die "Error writing to file!";
}
close($f);
}
$t2 = time();
print "Time to create " . NUM_FILES . ": " . ($t2 - $t1) . " seconds.\n";
system("umount", DEV) == 0 or die "umount failed!";
system("mount", DEV, MNT_POINT) == 0 or die "mount failed!";
$t1 = time();
system("bash -c 'ls -lha " . TEST_DIR . " > /dev/null'") == 0 or die "ls failed!";
$t2 = time();
print "Time to ls -lha all files: " . ($t2 - $t1) . " seconds.\n";
system("umount", DEV) == 0 or die "umount failed!";
Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com>
Signed-off-by: Josef Bacik <jbacik@fb.com>
Signed-off-by: Chris Mason <clm@fb.com>
2014-01-07 18:47:46 +07:00
|
|
|
ret = btrfs_create_subvol_root(trans, new_root, root, new_dirid);
|
2011-07-27 01:32:23 +07:00
|
|
|
if (ret) {
|
|
|
|
|
/* We potentially lose an unused inode item here */
|
2016-06-11 05:19:25 +07:00
|
|
|
btrfs_abort_transaction(trans, ret);
|
2011-07-27 01:32:23 +07:00
|
|
|
goto fail;
|
|
|
|
|
}
|
|
|
|
|
|
2016-01-07 20:26:59 +07:00
|
|
|
mutex_lock(&new_root->objectid_mutex);
|
|
|
|
|
new_root->highest_objectid = new_dirid;
|
|
|
|
|
mutex_unlock(&new_root->objectid_mutex);
|
|
|
|
|
|
2008-06-12 08:53:53 +07:00
|
|
|
/*
|
|
|
|
|
* insert the directory item
|
|
|
|
|
*/
|
2017-02-20 18:50:33 +07:00
|
|
|
ret = btrfs_set_inode_index(BTRFS_I(dir), &index);
|
2012-03-12 22:03:00 +07:00
|
|
|
if (ret) {
|
2016-06-11 05:19:25 +07:00
|
|
|
btrfs_abort_transaction(trans, ret);
|
2012-03-12 22:03:00 +07:00
|
|
|
goto fail;
|
|
|
|
|
}
|
2008-11-18 09:02:50 +07:00
|
|
|
|
|
|
|
|
ret = btrfs_insert_dir_item(trans, root,
|
2017-02-20 18:50:31 +07:00
|
|
|
name, namelen, BTRFS_I(dir), &key,
|
2008-11-18 09:02:50 +07:00
|
|
|
BTRFS_FT_DIR, index);
|
2012-03-12 22:03:00 +07:00
|
|
|
if (ret) {
|
2016-06-11 05:19:25 +07:00
|
|
|
btrfs_abort_transaction(trans, ret);
|
2008-06-12 08:53:53 +07:00
|
|
|
goto fail;
|
2012-03-12 22:03:00 +07:00
|
|
|
}
|
2008-11-18 08:37:39 +07:00
|
|
|
|
2017-02-20 18:50:34 +07:00
|
|
|
btrfs_i_size_write(BTRFS_I(dir), dir->i_size + namelen * 2);
|
2009-01-06 03:43:43 +07:00
|
|
|
ret = btrfs_update_inode(trans, root, dir);
|
|
|
|
|
BUG_ON(ret);
|
|
|
|
|
|
2016-06-23 05:54:23 +07:00
|
|
|
ret = btrfs_add_root_ref(trans, fs_info,
|
2009-09-22 02:56:00 +07:00
|
|
|
objectid, root->root_key.objectid,
|
2017-01-11 01:35:31 +07:00
|
|
|
btrfs_ino(BTRFS_I(dir)), index, name, namelen);
|
2009-09-22 03:00:26 +07:00
|
|
|
BUG_ON(ret);
|
2008-06-12 08:53:53 +07:00
|
|
|
|
2016-06-23 05:54:23 +07:00
|
|
|
ret = btrfs_uuid_tree_add(trans, fs_info, root_item->uuid,
|
2016-06-22 08:16:51 +07:00
|
|
|
BTRFS_UUID_KEY_SUBVOL, objectid);
|
2013-08-15 22:11:20 +07:00
|
|
|
if (ret)
|
2016-06-11 05:19:25 +07:00
|
|
|
btrfs_abort_transaction(trans, ret);
|
2013-08-15 22:11:20 +07:00
|
|
|
|
2008-06-12 08:53:53 +07:00
|
|
|
fail:
|
2016-03-24 23:49:22 +07:00
|
|
|
kfree(root_item);
|
2013-02-28 17:04:33 +07:00
|
|
|
trans->block_rsv = NULL;
|
|
|
|
|
trans->bytes_reserved = 0;
|
2017-02-11 01:18:18 +07:00
|
|
|
btrfs_subvolume_release_metadata(fs_info, &block_rsv);
|
2014-01-09 13:57:06 +07:00
|
|
|
|
2010-10-30 02:41:32 +07:00
|
|
|
if (async_transid) {
|
|
|
|
|
*async_transid = trans->transid;
|
2016-09-10 08:39:03 +07:00
|
|
|
err = btrfs_commit_transaction_async(trans, 1);
|
2013-03-04 16:45:06 +07:00
|
|
|
if (err)
|
2016-09-10 08:39:03 +07:00
|
|
|
err = btrfs_commit_transaction(trans);
|
2010-10-30 02:41:32 +07:00
|
|
|
} else {
|
2016-09-10 08:39:03 +07:00
|
|
|
err = btrfs_commit_transaction(trans);
|
2010-10-30 02:41:32 +07:00
|
|
|
}
|
2008-06-12 08:53:53 +07:00
|
|
|
if (err && !ret)
|
|
|
|
|
ret = err;
|
2013-02-07 00:06:02 +07:00
|
|
|
|
2013-12-13 07:51:42 +07:00
|
|
|
if (!ret) {
|
|
|
|
|
inode = btrfs_lookup_dentry(dir, dentry);
|
2014-01-09 13:57:06 +07:00
|
|
|
if (IS_ERR(inode))
|
|
|
|
|
return PTR_ERR(inode);
|
2013-12-13 07:51:42 +07:00
|
|
|
d_instantiate(dentry, inode);
|
|
|
|
|
}
|
2008-06-12 08:53:53 +07:00
|
|
|
return ret;
|
2016-03-24 23:49:22 +07:00
|
|
|
|
|
|
|
|
fail_free:
|
|
|
|
|
kfree(root_item);
|
|
|
|
|
return ret;
|
2008-06-12 08:53:53 +07:00
|
|
|
}
|
|
|
|
|
|
2013-02-28 17:01:15 +07:00
|
|
|
static int create_snapshot(struct btrfs_root *root, struct inode *dir,
|
2017-02-11 01:54:06 +07:00
|
|
|
struct dentry *dentry,
|
2013-02-28 17:01:15 +07:00
|
|
|
u64 *async_transid, bool readonly,
|
|
|
|
|
struct btrfs_qgroup_inherit *inherit)
|
2008-06-12 08:53:53 +07:00
|
|
|
{
|
2016-06-23 05:54:23 +07:00
|
|
|
struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb);
|
2009-11-12 16:37:02 +07:00
|
|
|
struct inode *inode;
|
2008-06-12 08:53:53 +07:00
|
|
|
struct btrfs_pending_snapshot *pending_snapshot;
|
|
|
|
|
struct btrfs_trans_handle *trans;
|
2009-11-12 16:37:02 +07:00
|
|
|
int ret;
|
2008-06-12 08:53:53 +07:00
|
|
|
|
2014-04-02 18:51:05 +07:00
|
|
|
if (!test_bit(BTRFS_ROOT_REF_COWS, &root->state))
|
2008-06-12 08:53:53 +07:00
|
|
|
return -EINVAL;
|
|
|
|
|
|
2017-02-13 17:03:44 +07:00
|
|
|
pending_snapshot = kzalloc(sizeof(*pending_snapshot), GFP_KERNEL);
|
2015-11-11 00:53:56 +07:00
|
|
|
if (!pending_snapshot)
|
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
2015-11-11 00:54:00 +07:00
|
|
|
pending_snapshot->root_item = kzalloc(sizeof(struct btrfs_root_item),
|
2017-02-13 17:03:44 +07:00
|
|
|
GFP_KERNEL);
|
2015-11-11 00:54:03 +07:00
|
|
|
pending_snapshot->path = btrfs_alloc_path();
|
|
|
|
|
if (!pending_snapshot->root_item || !pending_snapshot->path) {
|
2015-11-11 00:54:00 +07:00
|
|
|
ret = -ENOMEM;
|
|
|
|
|
goto free_pending;
|
|
|
|
|
}
|
|
|
|
|
|
2017-06-22 07:19:11 +07:00
|
|
|
atomic_inc(&root->will_be_snapshotted);
|
2014-03-18 00:06:10 +07:00
|
|
|
smp_mb__after_atomic();
|
2017-09-02 05:14:29 +07:00
|
|
|
/* wait for no snapshot writes */
|
|
|
|
|
wait_event(root->subv_writers->wait,
|
|
|
|
|
percpu_counter_sum(&root->subv_writers->counter) == 0);
|
2014-03-06 12:38:19 +07:00
|
|
|
|
2013-05-15 14:48:24 +07:00
|
|
|
ret = btrfs_start_delalloc_inodes(root, 0);
|
|
|
|
|
if (ret)
|
2015-11-11 00:53:56 +07:00
|
|
|
goto dec_and_free;
|
2013-05-15 14:48:24 +07:00
|
|
|
|
2017-06-23 23:48:21 +07:00
|
|
|
btrfs_wait_ordered_extents(root, U64_MAX, 0, (u64)-1);
|
2013-05-15 14:48:24 +07:00
|
|
|
|
2012-09-06 17:02:28 +07:00
|
|
|
btrfs_init_block_rsv(&pending_snapshot->block_rsv,
|
|
|
|
|
BTRFS_BLOCK_RSV_TEMP);
|
2013-02-28 17:04:33 +07:00
|
|
|
/*
|
|
|
|
|
* 1 - parent dir inode
|
|
|
|
|
* 2 - dir entries
|
|
|
|
|
* 1 - root item
|
|
|
|
|
* 2 - root ref/backref
|
|
|
|
|
* 1 - root of snapshot
|
2013-08-15 22:11:20 +07:00
|
|
|
* 1 - UUID item
|
2013-02-28 17:04:33 +07:00
|
|
|
*/
|
|
|
|
|
ret = btrfs_subvolume_reserve_metadata(BTRFS_I(dir)->root,
|
2013-08-15 22:11:20 +07:00
|
|
|
&pending_snapshot->block_rsv, 8,
|
2013-07-10 03:37:21 +07:00
|
|
|
&pending_snapshot->qgroup_reserved,
|
|
|
|
|
false);
|
2013-02-28 17:04:33 +07:00
|
|
|
if (ret)
|
2015-11-11 00:53:56 +07:00
|
|
|
goto dec_and_free;
|
2013-02-28 17:04:33 +07:00
|
|
|
|
2008-11-18 09:02:50 +07:00
|
|
|
pending_snapshot->dentry = dentry;
|
2008-06-12 08:53:53 +07:00
|
|
|
pending_snapshot->root = root;
|
2010-12-20 15:04:08 +07:00
|
|
|
pending_snapshot->readonly = readonly;
|
2013-02-28 17:01:15 +07:00
|
|
|
pending_snapshot->dir = dir;
|
2013-02-07 13:02:44 +07:00
|
|
|
pending_snapshot->inherit = inherit;
|
2010-05-16 21:48:46 +07:00
|
|
|
|
2013-02-28 17:04:33 +07:00
|
|
|
trans = btrfs_start_transaction(root, 0);
|
2010-05-16 21:48:46 +07:00
|
|
|
if (IS_ERR(trans)) {
|
|
|
|
|
ret = PTR_ERR(trans);
|
|
|
|
|
goto fail;
|
|
|
|
|
}
|
|
|
|
|
|
2016-06-23 05:54:23 +07:00
|
|
|
spin_lock(&fs_info->trans_lock);
|
2008-06-12 08:53:53 +07:00
|
|
|
list_add(&pending_snapshot->list,
|
|
|
|
|
&trans->transaction->pending_snapshots);
|
2016-06-23 05:54:23 +07:00
|
|
|
spin_unlock(&fs_info->trans_lock);
|
2010-10-30 02:41:32 +07:00
|
|
|
if (async_transid) {
|
|
|
|
|
*async_transid = trans->transid;
|
2016-09-10 08:39:03 +07:00
|
|
|
ret = btrfs_commit_transaction_async(trans, 1);
|
2013-03-04 16:45:06 +07:00
|
|
|
if (ret)
|
2016-09-10 08:39:03 +07:00
|
|
|
ret = btrfs_commit_transaction(trans);
|
2010-10-30 02:41:32 +07:00
|
|
|
} else {
|
2016-09-10 08:39:03 +07:00
|
|
|
ret = btrfs_commit_transaction(trans);
|
2010-10-30 02:41:32 +07:00
|
|
|
}
|
2013-03-04 16:44:29 +07:00
|
|
|
if (ret)
|
2012-10-23 02:51:44 +07:00
|
|
|
goto fail;
|
2010-05-16 21:48:46 +07:00
|
|
|
|
|
|
|
|
ret = pending_snapshot->error;
|
|
|
|
|
if (ret)
|
|
|
|
|
goto fail;
|
|
|
|
|
|
2014-10-16 03:50:56 +07:00
|
|
|
ret = btrfs_orphan_cleanup(pending_snapshot->snap);
|
|
|
|
|
if (ret)
|
|
|
|
|
goto fail;
|
|
|
|
|
|
2015-03-18 05:25:59 +07:00
|
|
|
inode = btrfs_lookup_dentry(d_inode(dentry->d_parent), dentry);
|
2009-11-12 16:37:02 +07:00
|
|
|
if (IS_ERR(inode)) {
|
|
|
|
|
ret = PTR_ERR(inode);
|
|
|
|
|
goto fail;
|
|
|
|
|
}
|
2013-12-13 07:51:42 +07:00
|
|
|
|
2009-11-12 16:37:02 +07:00
|
|
|
d_instantiate(dentry, inode);
|
|
|
|
|
ret = 0;
|
|
|
|
|
fail:
|
2017-02-11 01:18:18 +07:00
|
|
|
btrfs_subvolume_release_metadata(fs_info, &pending_snapshot->block_rsv);
|
2015-11-11 00:53:56 +07:00
|
|
|
dec_and_free:
|
2017-06-22 07:19:11 +07:00
|
|
|
if (atomic_dec_and_test(&root->will_be_snapshotted))
|
2018-03-15 17:43:08 +07:00
|
|
|
wake_up_var(&root->will_be_snapshotted);
|
2015-11-11 00:54:00 +07:00
|
|
|
free_pending:
|
|
|
|
|
kfree(pending_snapshot->root_item);
|
2015-11-11 00:54:03 +07:00
|
|
|
btrfs_free_path(pending_snapshot->path);
|
2015-11-11 00:53:56 +07:00
|
|
|
kfree(pending_snapshot);
|
|
|
|
|
|
2008-06-12 08:53:53 +07:00
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
2010-10-30 02:46:43 +07:00
|
|
|
/* copy of may_delete in fs/namei.c()
|
|
|
|
|
* Check whether we can remove a link victim from directory dir, check
|
|
|
|
|
* whether the type of victim is right.
|
|
|
|
|
* 1. We can't do it if dir is read-only (done in permission())
|
|
|
|
|
* 2. We should have write and exec permissions on dir
|
|
|
|
|
* 3. We can't remove anything from append-only dir
|
|
|
|
|
* 4. We can't do anything with immutable dir (done in permission())
|
|
|
|
|
* 5. If the sticky bit on dir is set we should either
|
|
|
|
|
* a. be owner of dir, or
|
|
|
|
|
* b. be owner of victim, or
|
|
|
|
|
* c. have CAP_FOWNER capability
|
2016-05-20 08:18:45 +07:00
|
|
|
* 6. If the victim is append-only or immutable we can't do anything with
|
2010-10-30 02:46:43 +07:00
|
|
|
* links pointing to it.
|
|
|
|
|
* 7. If we were asked to remove a directory and victim isn't one - ENOTDIR.
|
|
|
|
|
* 8. If we were asked to remove a non-directory and victim isn't one - EISDIR.
|
|
|
|
|
* 9. We can't remove a root or mountpoint.
|
|
|
|
|
* 10. We don't allow removal of NFS sillyrenamed files; it's handled by
|
|
|
|
|
* nfs_async_unlink().
|
|
|
|
|
*/
|
|
|
|
|
|
2013-10-31 12:03:04 +07:00
|
|
|
static int btrfs_may_delete(struct inode *dir, struct dentry *victim, int isdir)
|
2010-10-30 02:46:43 +07:00
|
|
|
{
|
|
|
|
|
int error;
|
|
|
|
|
|
2015-03-18 05:25:59 +07:00
|
|
|
if (d_really_is_negative(victim))
|
2010-10-30 02:46:43 +07:00
|
|
|
return -ENOENT;
|
|
|
|
|
|
2015-03-18 05:25:59 +07:00
|
|
|
BUG_ON(d_inode(victim->d_parent) != dir);
|
2012-10-11 02:25:25 +07:00
|
|
|
audit_inode_child(dir, victim, AUDIT_TYPE_CHILD_DELETE);
|
2010-10-30 02:46:43 +07:00
|
|
|
|
|
|
|
|
error = inode_permission(dir, MAY_WRITE | MAY_EXEC);
|
|
|
|
|
if (error)
|
|
|
|
|
return error;
|
|
|
|
|
if (IS_APPEND(dir))
|
|
|
|
|
return -EPERM;
|
2015-03-18 05:25:59 +07:00
|
|
|
if (check_sticky(dir, d_inode(victim)) || IS_APPEND(d_inode(victim)) ||
|
|
|
|
|
IS_IMMUTABLE(d_inode(victim)) || IS_SWAPFILE(d_inode(victim)))
|
2010-10-30 02:46:43 +07:00
|
|
|
return -EPERM;
|
|
|
|
|
if (isdir) {
|
VFS: (Scripted) Convert S_ISLNK/DIR/REG(dentry->d_inode) to d_is_*(dentry)
Convert the following where appropriate:
(1) S_ISLNK(dentry->d_inode) to d_is_symlink(dentry).
(2) S_ISREG(dentry->d_inode) to d_is_reg(dentry).
(3) S_ISDIR(dentry->d_inode) to d_is_dir(dentry). This is actually more
complicated than it appears as some calls should be converted to
d_can_lookup() instead. The difference is whether the directory in
question is a real dir with a ->lookup op or whether it's a fake dir with
a ->d_automount op.
In some circumstances, we can subsume checks for dentry->d_inode not being
NULL into this, provided we the code isn't in a filesystem that expects
d_inode to be NULL if the dirent really *is* negative (ie. if we're going to
use d_inode() rather than d_backing_inode() to get the inode pointer).
Note that the dentry type field may be set to something other than
DCACHE_MISS_TYPE when d_inode is NULL in the case of unionmount, where the VFS
manages the fall-through from a negative dentry to a lower layer. In such a
case, the dentry type of the negative union dentry is set to the same as the
type of the lower dentry.
However, if you know d_inode is not NULL at the call site, then you can use
the d_is_xxx() functions even in a filesystem.
There is one further complication: a 0,0 chardev dentry may be labelled
DCACHE_WHITEOUT_TYPE rather than DCACHE_SPECIAL_TYPE. Strictly, this was
intended for special directory entry types that don't have attached inodes.
The following perl+coccinelle script was used:
use strict;
my @callers;
open($fd, 'git grep -l \'S_IS[A-Z].*->d_inode\' |') ||
die "Can't grep for S_ISDIR and co. callers";
@callers = <$fd>;
close($fd);
unless (@callers) {
print "No matches\n";
exit(0);
}
my @cocci = (
'@@',
'expression E;',
'@@',
'',
'- S_ISLNK(E->d_inode->i_mode)',
'+ d_is_symlink(E)',
'',
'@@',
'expression E;',
'@@',
'',
'- S_ISDIR(E->d_inode->i_mode)',
'+ d_is_dir(E)',
'',
'@@',
'expression E;',
'@@',
'',
'- S_ISREG(E->d_inode->i_mode)',
'+ d_is_reg(E)' );
my $coccifile = "tmp.sp.cocci";
open($fd, ">$coccifile") || die $coccifile;
print($fd "$_\n") || die $coccifile foreach (@cocci);
close($fd);
foreach my $file (@callers) {
chomp $file;
print "Processing ", $file, "\n";
system("spatch", "--sp-file", $coccifile, $file, "--in-place", "--no-show-diff") == 0 ||
die "spatch failed";
}
[AV: overlayfs parts skipped]
Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-01-29 19:02:35 +07:00
|
|
|
if (!d_is_dir(victim))
|
2010-10-30 02:46:43 +07:00
|
|
|
return -ENOTDIR;
|
|
|
|
|
if (IS_ROOT(victim))
|
|
|
|
|
return -EBUSY;
|
VFS: (Scripted) Convert S_ISLNK/DIR/REG(dentry->d_inode) to d_is_*(dentry)
Convert the following where appropriate:
(1) S_ISLNK(dentry->d_inode) to d_is_symlink(dentry).
(2) S_ISREG(dentry->d_inode) to d_is_reg(dentry).
(3) S_ISDIR(dentry->d_inode) to d_is_dir(dentry). This is actually more
complicated than it appears as some calls should be converted to
d_can_lookup() instead. The difference is whether the directory in
question is a real dir with a ->lookup op or whether it's a fake dir with
a ->d_automount op.
In some circumstances, we can subsume checks for dentry->d_inode not being
NULL into this, provided we the code isn't in a filesystem that expects
d_inode to be NULL if the dirent really *is* negative (ie. if we're going to
use d_inode() rather than d_backing_inode() to get the inode pointer).
Note that the dentry type field may be set to something other than
DCACHE_MISS_TYPE when d_inode is NULL in the case of unionmount, where the VFS
manages the fall-through from a negative dentry to a lower layer. In such a
case, the dentry type of the negative union dentry is set to the same as the
type of the lower dentry.
However, if you know d_inode is not NULL at the call site, then you can use
the d_is_xxx() functions even in a filesystem.
There is one further complication: a 0,0 chardev dentry may be labelled
DCACHE_WHITEOUT_TYPE rather than DCACHE_SPECIAL_TYPE. Strictly, this was
intended for special directory entry types that don't have attached inodes.
The following perl+coccinelle script was used:
use strict;
my @callers;
open($fd, 'git grep -l \'S_IS[A-Z].*->d_inode\' |') ||
die "Can't grep for S_ISDIR and co. callers";
@callers = <$fd>;
close($fd);
unless (@callers) {
print "No matches\n";
exit(0);
}
my @cocci = (
'@@',
'expression E;',
'@@',
'',
'- S_ISLNK(E->d_inode->i_mode)',
'+ d_is_symlink(E)',
'',
'@@',
'expression E;',
'@@',
'',
'- S_ISDIR(E->d_inode->i_mode)',
'+ d_is_dir(E)',
'',
'@@',
'expression E;',
'@@',
'',
'- S_ISREG(E->d_inode->i_mode)',
'+ d_is_reg(E)' );
my $coccifile = "tmp.sp.cocci";
open($fd, ">$coccifile") || die $coccifile;
print($fd "$_\n") || die $coccifile foreach (@cocci);
close($fd);
foreach my $file (@callers) {
chomp $file;
print "Processing ", $file, "\n";
system("spatch", "--sp-file", $coccifile, $file, "--in-place", "--no-show-diff") == 0 ||
die "spatch failed";
}
[AV: overlayfs parts skipped]
Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2015-01-29 19:02:35 +07:00
|
|
|
} else if (d_is_dir(victim))
|
2010-10-30 02:46:43 +07:00
|
|
|
return -EISDIR;
|
|
|
|
|
if (IS_DEADDIR(dir))
|
|
|
|
|
return -ENOENT;
|
|
|
|
|
if (victim->d_flags & DCACHE_NFSFS_RENAMED)
|
|
|
|
|
return -EBUSY;
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
2008-10-10 00:39:39 +07:00
|
|
|
/* copy of may_create in fs/namei.c() */
|
|
|
|
|
static inline int btrfs_may_create(struct inode *dir, struct dentry *child)
|
|
|
|
|
{
|
2015-03-18 05:25:59 +07:00
|
|
|
if (d_really_is_positive(child))
|
2008-10-10 00:39:39 +07:00
|
|
|
return -EEXIST;
|
|
|
|
|
if (IS_DEADDIR(dir))
|
|
|
|
|
return -ENOENT;
|
|
|
|
|
return inode_permission(dir, MAY_WRITE | MAY_EXEC);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Create a new subvolume below @parent. This is largely modeled after
|
|
|
|
|
* sys_mkdirat and vfs_mkdir, but we only do a single component lookup
|
|
|
|
|
* inside this filesystem so it's quite a bit simpler.
|
|
|
|
|
*/
|
2016-11-21 07:34:31 +07:00
|
|
|
static noinline int btrfs_mksubvol(const struct path *parent,
|
2017-02-15 00:33:53 +07:00
|
|
|
const char *name, int namelen,
|
2010-10-30 02:41:32 +07:00
|
|
|
struct btrfs_root *snap_src,
|
2011-09-14 20:58:21 +07:00
|
|
|
u64 *async_transid, bool readonly,
|
2013-02-07 13:02:44 +07:00
|
|
|
struct btrfs_qgroup_inherit *inherit)
|
2008-10-10 00:39:39 +07:00
|
|
|
{
|
2016-06-23 05:54:23 +07:00
|
|
|
struct inode *dir = d_inode(parent->dentry);
|
|
|
|
|
struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb);
|
2008-10-10 00:39:39 +07:00
|
|
|
struct dentry *dentry;
|
|
|
|
|
int error;
|
|
|
|
|
|
2016-05-26 11:05:12 +07:00
|
|
|
error = down_write_killable_nested(&dir->i_rwsem, I_MUTEX_PARENT);
|
|
|
|
|
if (error == -EINTR)
|
|
|
|
|
return error;
|
2008-10-10 00:39:39 +07:00
|
|
|
|
|
|
|
|
dentry = lookup_one_len(name, parent->dentry, namelen);
|
|
|
|
|
error = PTR_ERR(dentry);
|
|
|
|
|
if (IS_ERR(dentry))
|
|
|
|
|
goto out_unlock;
|
|
|
|
|
|
2009-09-22 03:00:26 +07:00
|
|
|
error = btrfs_may_create(dir, dentry);
|
2008-10-10 00:39:39 +07:00
|
|
|
if (error)
|
2012-06-29 16:58:46 +07:00
|
|
|
goto out_dput;
|
2008-10-10 00:39:39 +07:00
|
|
|
|
2012-12-18 02:26:57 +07:00
|
|
|
/*
|
|
|
|
|
* even if this name doesn't exist, we may get hash collisions.
|
|
|
|
|
* check for them now when we can safely fail
|
|
|
|
|
*/
|
|
|
|
|
error = btrfs_check_dir_item_collision(BTRFS_I(dir)->root,
|
|
|
|
|
dir->i_ino, name,
|
|
|
|
|
namelen);
|
|
|
|
|
if (error)
|
|
|
|
|
goto out_dput;
|
|
|
|
|
|
2016-06-23 05:54:23 +07:00
|
|
|
down_read(&fs_info->subvol_sem);
|
2009-09-22 03:00:26 +07:00
|
|
|
|
|
|
|
|
if (btrfs_root_refs(&BTRFS_I(dir)->root->root_item) == 0)
|
|
|
|
|
goto out_up_read;
|
|
|
|
|
|
2008-11-18 09:02:50 +07:00
|
|
|
if (snap_src) {
|
2017-02-11 01:54:06 +07:00
|
|
|
error = create_snapshot(snap_src, dir, dentry,
|
2011-09-14 20:58:21 +07:00
|
|
|
async_transid, readonly, inherit);
|
2008-11-18 09:02:50 +07:00
|
|
|
} else {
|
2013-02-28 17:04:33 +07:00
|
|
|
error = create_subvol(dir, dentry, name, namelen,
|
|
|
|
|
async_transid, inherit);
|
2008-11-18 09:02:50 +07:00
|
|
|
}
|
2009-09-22 03:00:26 +07:00
|
|
|
if (!error)
|
|
|
|
|
fsnotify_mkdir(dir, dentry);
|
|
|
|
|
out_up_read:
|
2016-06-23 05:54:23 +07:00
|
|
|
up_read(&fs_info->subvol_sem);
|
2008-10-10 00:39:39 +07:00
|
|
|
out_dput:
|
|
|
|
|
dput(dentry);
|
|
|
|
|
out_unlock:
|
2016-01-23 03:40:57 +07:00
|
|
|
inode_unlock(dir);
|
2008-10-10 00:39:39 +07:00
|
|
|
return error;
|
|
|
|
|
}
|
|
|
|
|
|
2011-05-25 02:35:30 +07:00
|
|
|
/*
|
|
|
|
|
* When we're defragging a range, we don't want to kick it off again
|
|
|
|
|
* if it is really just waiting for delalloc to send it down.
|
|
|
|
|
* If we find a nice big extent or delalloc range for the bytes in the
|
|
|
|
|
* file you want to defrag, we return 0 to let you know to skip this
|
|
|
|
|
* part of the file
|
|
|
|
|
*/
|
2014-07-29 22:32:10 +07:00
|
|
|
static int check_defrag_in_cache(struct inode *inode, u64 offset, u32 thresh)
|
2011-05-25 02:35:30 +07:00
|
|
|
{
|
|
|
|
|
struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
|
|
|
|
|
struct extent_map *em = NULL;
|
|
|
|
|
struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
|
|
|
|
|
u64 end;
|
|
|
|
|
|
|
|
|
|
read_lock(&em_tree->lock);
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
em = lookup_extent_mapping(em_tree, offset, PAGE_SIZE);
|
2011-05-25 02:35:30 +07:00
|
|
|
read_unlock(&em_tree->lock);
|
|
|
|
|
|
|
|
|
|
if (em) {
|
|
|
|
|
end = extent_map_end(em);
|
|
|
|
|
free_extent_map(em);
|
|
|
|
|
if (end - offset > thresh)
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
/* if we already have a nice delalloc here, just stop */
|
|
|
|
|
thresh /= 2;
|
|
|
|
|
end = count_range_bits(io_tree, &offset, offset + thresh,
|
|
|
|
|
thresh, EXTENT_DELALLOC, 1);
|
|
|
|
|
if (end >= thresh)
|
|
|
|
|
return 0;
|
|
|
|
|
return 1;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* helper function to walk through a file and find extents
|
|
|
|
|
* newer than a specific transid, and smaller than thresh.
|
|
|
|
|
*
|
|
|
|
|
* This is used by the defragging code to find new and small
|
|
|
|
|
* extents
|
|
|
|
|
*/
|
|
|
|
|
static int find_new_extents(struct btrfs_root *root,
|
|
|
|
|
struct inode *inode, u64 newer_than,
|
2014-07-29 22:32:10 +07:00
|
|
|
u64 *off, u32 thresh)
|
2011-05-25 02:35:30 +07:00
|
|
|
{
|
|
|
|
|
struct btrfs_path *path;
|
|
|
|
|
struct btrfs_key min_key;
|
|
|
|
|
struct extent_buffer *leaf;
|
|
|
|
|
struct btrfs_file_extent_item *extent;
|
|
|
|
|
int type;
|
|
|
|
|
int ret;
|
2017-01-11 01:35:31 +07:00
|
|
|
u64 ino = btrfs_ino(BTRFS_I(inode));
|
2011-05-25 02:35:30 +07:00
|
|
|
|
|
|
|
|
path = btrfs_alloc_path();
|
|
|
|
|
if (!path)
|
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
2011-06-01 00:08:14 +07:00
|
|
|
min_key.objectid = ino;
|
2011-05-25 02:35:30 +07:00
|
|
|
min_key.type = BTRFS_EXTENT_DATA_KEY;
|
|
|
|
|
min_key.offset = *off;
|
|
|
|
|
|
2013-10-31 12:03:04 +07:00
|
|
|
while (1) {
|
2013-10-01 22:13:42 +07:00
|
|
|
ret = btrfs_search_forward(root, &min_key, path, newer_than);
|
2011-05-25 02:35:30 +07:00
|
|
|
if (ret != 0)
|
|
|
|
|
goto none;
|
Btrfs: less fs tree lock contention when using autodefrag
When finding new extents during an autodefrag, don't do so many fs tree
lookups to find an extent with a size smaller then the target treshold.
Instead, after each fs tree forward search immediately unlock upper
levels and process the entire leaf while holding a read lock on the leaf,
since our leaf processing is very fast.
This reduces lock contention, allowing for higher concurrency when other
tasks want to write/update items related to other inodes in the fs tree,
as we're not holding read locks on upper tree levels while processing the
leaf and we do less tree searches.
Test:
sysbench --test=fileio --file-num=512 --file-total-size=16G \
--file-test-mode=rndrw --num-threads=32 --file-block-size=32768 \
--file-rw-ratio=3 --file-io-mode=sync --max-time=1800 \
--max-requests=10000000000 [prepare|run]
(fileystem mounted with -o autodefrag, averages of 5 runs)
Before this change: 58.852Mb/sec throughtput, read 77.589Gb, written 25.863Gb
After this change: 63.034Mb/sec throughtput, read 83.102Gb, written 27.701Gb
Test machine: quad core intel i5-3570K, 32Gb of RAM, SSD.
Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com>
Signed-off-by: Chris Mason <clm@fb.com>
2014-03-12 08:28:24 +07:00
|
|
|
process_slot:
|
2011-06-01 00:08:14 +07:00
|
|
|
if (min_key.objectid != ino)
|
2011-05-25 02:35:30 +07:00
|
|
|
goto none;
|
|
|
|
|
if (min_key.type != BTRFS_EXTENT_DATA_KEY)
|
|
|
|
|
goto none;
|
|
|
|
|
|
|
|
|
|
leaf = path->nodes[0];
|
|
|
|
|
extent = btrfs_item_ptr(leaf, path->slots[0],
|
|
|
|
|
struct btrfs_file_extent_item);
|
|
|
|
|
|
|
|
|
|
type = btrfs_file_extent_type(leaf, extent);
|
|
|
|
|
if (type == BTRFS_FILE_EXTENT_REG &&
|
|
|
|
|
btrfs_file_extent_num_bytes(leaf, extent) < thresh &&
|
|
|
|
|
check_defrag_in_cache(inode, min_key.offset, thresh)) {
|
|
|
|
|
*off = min_key.offset;
|
|
|
|
|
btrfs_free_path(path);
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
Btrfs: less fs tree lock contention when using autodefrag
When finding new extents during an autodefrag, don't do so many fs tree
lookups to find an extent with a size smaller then the target treshold.
Instead, after each fs tree forward search immediately unlock upper
levels and process the entire leaf while holding a read lock on the leaf,
since our leaf processing is very fast.
This reduces lock contention, allowing for higher concurrency when other
tasks want to write/update items related to other inodes in the fs tree,
as we're not holding read locks on upper tree levels while processing the
leaf and we do less tree searches.
Test:
sysbench --test=fileio --file-num=512 --file-total-size=16G \
--file-test-mode=rndrw --num-threads=32 --file-block-size=32768 \
--file-rw-ratio=3 --file-io-mode=sync --max-time=1800 \
--max-requests=10000000000 [prepare|run]
(fileystem mounted with -o autodefrag, averages of 5 runs)
Before this change: 58.852Mb/sec throughtput, read 77.589Gb, written 25.863Gb
After this change: 63.034Mb/sec throughtput, read 83.102Gb, written 27.701Gb
Test machine: quad core intel i5-3570K, 32Gb of RAM, SSD.
Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com>
Signed-off-by: Chris Mason <clm@fb.com>
2014-03-12 08:28:24 +07:00
|
|
|
path->slots[0]++;
|
|
|
|
|
if (path->slots[0] < btrfs_header_nritems(leaf)) {
|
|
|
|
|
btrfs_item_key_to_cpu(leaf, &min_key, path->slots[0]);
|
|
|
|
|
goto process_slot;
|
|
|
|
|
}
|
|
|
|
|
|
2011-05-25 02:35:30 +07:00
|
|
|
if (min_key.offset == (u64)-1)
|
|
|
|
|
goto none;
|
|
|
|
|
|
|
|
|
|
min_key.offset++;
|
|
|
|
|
btrfs_release_path(path);
|
|
|
|
|
}
|
|
|
|
|
none:
|
|
|
|
|
btrfs_free_path(path);
|
|
|
|
|
return -ENOENT;
|
|
|
|
|
}
|
|
|
|
|
|
2012-06-11 15:03:35 +07:00
|
|
|
static struct extent_map *defrag_lookup_extent(struct inode *inode, u64 start)
|
2012-03-29 20:57:45 +07:00
|
|
|
{
|
|
|
|
|
struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
|
2012-06-11 15:03:35 +07:00
|
|
|
struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
|
|
|
|
|
struct extent_map *em;
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
u64 len = PAGE_SIZE;
|
2012-03-29 20:57:45 +07:00
|
|
|
|
2012-06-11 15:03:35 +07:00
|
|
|
/*
|
|
|
|
|
* hopefully we have this extent in the tree already, try without
|
|
|
|
|
* the full extent lock
|
|
|
|
|
*/
|
2012-03-29 20:57:45 +07:00
|
|
|
read_lock(&em_tree->lock);
|
2012-06-11 15:03:35 +07:00
|
|
|
em = lookup_extent_mapping(em_tree, start, len);
|
2012-03-29 20:57:45 +07:00
|
|
|
read_unlock(&em_tree->lock);
|
|
|
|
|
|
2012-06-11 15:03:35 +07:00
|
|
|
if (!em) {
|
2014-03-11 20:56:15 +07:00
|
|
|
struct extent_state *cached = NULL;
|
|
|
|
|
u64 end = start + len - 1;
|
|
|
|
|
|
2012-06-11 15:03:35 +07:00
|
|
|
/* get the big lock and read metadata off disk */
|
2015-12-03 20:30:40 +07:00
|
|
|
lock_extent_bits(io_tree, start, end, &cached);
|
2017-02-20 18:51:06 +07:00
|
|
|
em = btrfs_get_extent(BTRFS_I(inode), NULL, 0, start, len, 0);
|
2017-12-13 03:43:52 +07:00
|
|
|
unlock_extent_cached(io_tree, start, end, &cached);
|
2012-06-11 15:03:35 +07:00
|
|
|
|
|
|
|
|
if (IS_ERR(em))
|
|
|
|
|
return NULL;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return em;
|
|
|
|
|
}
|
2012-03-29 20:57:45 +07:00
|
|
|
|
2012-06-11 15:03:35 +07:00
|
|
|
static bool defrag_check_next_extent(struct inode *inode, struct extent_map *em)
|
|
|
|
|
{
|
|
|
|
|
struct extent_map *next;
|
|
|
|
|
bool ret = true;
|
|
|
|
|
|
|
|
|
|
/* this is the last extent */
|
|
|
|
|
if (em->start + em->len >= i_size_read(inode))
|
|
|
|
|
return false;
|
|
|
|
|
|
|
|
|
|
next = defrag_lookup_extent(inode, em->start + em->len);
|
2014-08-27 03:55:54 +07:00
|
|
|
if (!next || next->block_start >= EXTENT_MAP_LAST_BYTE)
|
|
|
|
|
ret = false;
|
|
|
|
|
else if ((em->block_start + em->block_len == next->block_start) &&
|
2015-12-14 23:42:10 +07:00
|
|
|
(em->block_len > SZ_128K && next->block_len > SZ_128K))
|
2012-06-11 15:03:35 +07:00
|
|
|
ret = false;
|
|
|
|
|
|
|
|
|
|
free_extent_map(next);
|
2012-03-29 20:57:45 +07:00
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
2014-07-29 22:32:10 +07:00
|
|
|
static int should_defrag_range(struct inode *inode, u64 start, u32 thresh,
|
2012-06-20 08:08:32 +07:00
|
|
|
u64 *last_len, u64 *skip, u64 *defrag_end,
|
|
|
|
|
int compress)
|
2010-03-10 22:52:59 +07:00
|
|
|
{
|
2012-06-11 15:03:35 +07:00
|
|
|
struct extent_map *em;
|
2010-03-10 22:52:59 +07:00
|
|
|
int ret = 1;
|
2012-06-11 15:03:35 +07:00
|
|
|
bool next_mergeable = true;
|
Btrfs: fix defrag to merge tail file extent
The file layout is
[extent 1]...[extent n][4k extent][HOLE][extent x]
extent 1~n and 4k extent can be merged during defrag, and the whole
defrag bytes is larger than our defrag thresh(256k), 4k extent as a
tail is left unmerged since we check if its next extent can be merged
(the next one is a hole, so the check will fail), the layout thus can
be
[new extent][4k extent][HOLE][extent x]
(1~n)
To fix it, beside looking at the next one, this also looks at the
previous one by checking @defrag_end, which is set to 0 when we
decide to stop merging contiguous extents, otherwise, we can merge
the previous one with our extent.
Also, this makes btrfs behave consistent with how xfs and ext4 do.
Signed-off-by: Liu Bo <bo.li.liu@oracle.com>
Signed-off-by: Chris Mason <clm@fb.com>
2015-08-07 15:48:41 +07:00
|
|
|
bool prev_mergeable = true;
|
2010-03-10 22:52:59 +07:00
|
|
|
|
|
|
|
|
/*
|
2011-09-02 14:57:07 +07:00
|
|
|
* make sure that once we start defragging an extent, we keep on
|
2010-03-10 22:52:59 +07:00
|
|
|
* defragging it
|
|
|
|
|
*/
|
|
|
|
|
if (start < *defrag_end)
|
|
|
|
|
return 1;
|
|
|
|
|
|
|
|
|
|
*skip = 0;
|
|
|
|
|
|
2012-06-11 15:03:35 +07:00
|
|
|
em = defrag_lookup_extent(inode, start);
|
|
|
|
|
if (!em)
|
|
|
|
|
return 0;
|
2010-03-10 22:52:59 +07:00
|
|
|
|
|
|
|
|
/* this will cover holes, and inline extents */
|
2012-03-29 20:57:45 +07:00
|
|
|
if (em->block_start >= EXTENT_MAP_LAST_BYTE) {
|
2010-03-10 22:52:59 +07:00
|
|
|
ret = 0;
|
2012-03-29 20:57:45 +07:00
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
|
Btrfs: fix defrag to merge tail file extent
The file layout is
[extent 1]...[extent n][4k extent][HOLE][extent x]
extent 1~n and 4k extent can be merged during defrag, and the whole
defrag bytes is larger than our defrag thresh(256k), 4k extent as a
tail is left unmerged since we check if its next extent can be merged
(the next one is a hole, so the check will fail), the layout thus can
be
[new extent][4k extent][HOLE][extent x]
(1~n)
To fix it, beside looking at the next one, this also looks at the
previous one by checking @defrag_end, which is set to 0 when we
decide to stop merging contiguous extents, otherwise, we can merge
the previous one with our extent.
Also, this makes btrfs behave consistent with how xfs and ext4 do.
Signed-off-by: Liu Bo <bo.li.liu@oracle.com>
Signed-off-by: Chris Mason <clm@fb.com>
2015-08-07 15:48:41 +07:00
|
|
|
if (!*defrag_end)
|
|
|
|
|
prev_mergeable = false;
|
|
|
|
|
|
2012-06-11 15:03:35 +07:00
|
|
|
next_mergeable = defrag_check_next_extent(inode, em);
|
2010-03-10 22:52:59 +07:00
|
|
|
/*
|
2012-06-11 15:03:35 +07:00
|
|
|
* we hit a real extent, if it is big or the next extent is not a
|
|
|
|
|
* real extent, don't bother defragging it
|
2010-03-10 22:52:59 +07:00
|
|
|
*/
|
2012-06-20 08:08:32 +07:00
|
|
|
if (!compress && (*last_len == 0 || *last_len >= thresh) &&
|
Btrfs: fix defrag to merge tail file extent
The file layout is
[extent 1]...[extent n][4k extent][HOLE][extent x]
extent 1~n and 4k extent can be merged during defrag, and the whole
defrag bytes is larger than our defrag thresh(256k), 4k extent as a
tail is left unmerged since we check if its next extent can be merged
(the next one is a hole, so the check will fail), the layout thus can
be
[new extent][4k extent][HOLE][extent x]
(1~n)
To fix it, beside looking at the next one, this also looks at the
previous one by checking @defrag_end, which is set to 0 when we
decide to stop merging contiguous extents, otherwise, we can merge
the previous one with our extent.
Also, this makes btrfs behave consistent with how xfs and ext4 do.
Signed-off-by: Liu Bo <bo.li.liu@oracle.com>
Signed-off-by: Chris Mason <clm@fb.com>
2015-08-07 15:48:41 +07:00
|
|
|
(em->len >= thresh || (!next_mergeable && !prev_mergeable)))
|
2010-03-10 22:52:59 +07:00
|
|
|
ret = 0;
|
2012-03-29 20:57:45 +07:00
|
|
|
out:
|
2010-03-10 22:52:59 +07:00
|
|
|
/*
|
|
|
|
|
* last_len ends up being a counter of how many bytes we've defragged.
|
|
|
|
|
* every time we choose not to defrag an extent, we reset *last_len
|
|
|
|
|
* so that the next tiny extent will force a defrag.
|
|
|
|
|
*
|
|
|
|
|
* The end result of this is that tiny extents before a single big
|
|
|
|
|
* extent will force at least part of that big extent to be defragged.
|
|
|
|
|
*/
|
|
|
|
|
if (ret) {
|
|
|
|
|
*defrag_end = extent_map_end(em);
|
|
|
|
|
} else {
|
|
|
|
|
*last_len = 0;
|
|
|
|
|
*skip = extent_map_end(em);
|
|
|
|
|
*defrag_end = 0;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
free_extent_map(em);
|
|
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
2011-05-25 02:35:30 +07:00
|
|
|
/*
|
|
|
|
|
* it doesn't do much good to defrag one or two pages
|
|
|
|
|
* at a time. This pulls in a nice chunk of pages
|
|
|
|
|
* to COW and defrag.
|
|
|
|
|
*
|
|
|
|
|
* It also makes sure the delalloc code has enough
|
|
|
|
|
* dirty data to avoid making new small extents as part
|
|
|
|
|
* of the defrag
|
|
|
|
|
*
|
|
|
|
|
* It's a good idea to start RA on this range
|
|
|
|
|
* before calling this.
|
|
|
|
|
*/
|
|
|
|
|
static int cluster_pages_for_defrag(struct inode *inode,
|
|
|
|
|
struct page **pages,
|
|
|
|
|
unsigned long start_index,
|
2014-01-22 02:18:29 +07:00
|
|
|
unsigned long num_pages)
|
2008-06-12 08:53:53 +07:00
|
|
|
{
|
2011-05-25 02:35:30 +07:00
|
|
|
unsigned long file_end;
|
|
|
|
|
u64 isize = i_size_read(inode);
|
|
|
|
|
u64 page_start;
|
|
|
|
|
u64 page_end;
|
2012-03-29 20:57:44 +07:00
|
|
|
u64 page_cnt;
|
2011-05-25 02:35:30 +07:00
|
|
|
int ret;
|
|
|
|
|
int i;
|
|
|
|
|
int i_done;
|
2008-07-24 22:57:52 +07:00
|
|
|
struct btrfs_ordered_extent *ordered;
|
2011-05-25 02:35:30 +07:00
|
|
|
struct extent_state *cached_state = NULL;
|
2012-02-16 14:01:24 +07:00
|
|
|
struct extent_io_tree *tree;
|
2017-02-27 14:10:38 +07:00
|
|
|
struct extent_changeset *data_reserved = NULL;
|
2011-09-22 02:05:58 +07:00
|
|
|
gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping);
|
2011-05-25 02:35:30 +07:00
|
|
|
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
file_end = (isize - 1) >> PAGE_SHIFT;
|
2012-03-29 20:57:44 +07:00
|
|
|
if (!isize || start_index > file_end)
|
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
|
|
page_cnt = min_t(u64, (u64)num_pages, (u64)file_end - start_index + 1);
|
2011-05-25 02:35:30 +07:00
|
|
|
|
2017-02-27 14:10:38 +07:00
|
|
|
ret = btrfs_delalloc_reserve_space(inode, &data_reserved,
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
start_index << PAGE_SHIFT,
|
|
|
|
|
page_cnt << PAGE_SHIFT);
|
2011-05-25 02:35:30 +07:00
|
|
|
if (ret)
|
|
|
|
|
return ret;
|
|
|
|
|
i_done = 0;
|
2012-02-16 14:01:24 +07:00
|
|
|
tree = &BTRFS_I(inode)->io_tree;
|
2011-05-25 02:35:30 +07:00
|
|
|
|
|
|
|
|
/* step one, lock all the pages */
|
2012-03-29 20:57:44 +07:00
|
|
|
for (i = 0; i < page_cnt; i++) {
|
2011-05-25 02:35:30 +07:00
|
|
|
struct page *page;
|
2012-02-16 14:01:24 +07:00
|
|
|
again:
|
2011-07-11 21:47:06 +07:00
|
|
|
page = find_or_create_page(inode->i_mapping,
|
2012-02-16 14:01:24 +07:00
|
|
|
start_index + i, mask);
|
2011-05-25 02:35:30 +07:00
|
|
|
if (!page)
|
|
|
|
|
break;
|
|
|
|
|
|
2012-02-16 14:01:24 +07:00
|
|
|
page_start = page_offset(page);
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
page_end = page_start + PAGE_SIZE - 1;
|
2012-02-16 14:01:24 +07:00
|
|
|
while (1) {
|
2014-03-11 20:56:15 +07:00
|
|
|
lock_extent_bits(tree, page_start, page_end,
|
2015-12-03 20:30:40 +07:00
|
|
|
&cached_state);
|
2012-02-16 14:01:24 +07:00
|
|
|
ordered = btrfs_lookup_ordered_extent(inode,
|
|
|
|
|
page_start);
|
2014-03-11 20:56:15 +07:00
|
|
|
unlock_extent_cached(tree, page_start, page_end,
|
2017-12-13 03:43:52 +07:00
|
|
|
&cached_state);
|
2012-02-16 14:01:24 +07:00
|
|
|
if (!ordered)
|
|
|
|
|
break;
|
|
|
|
|
|
|
|
|
|
unlock_page(page);
|
|
|
|
|
btrfs_start_ordered_extent(inode, ordered, 1);
|
|
|
|
|
btrfs_put_ordered_extent(ordered);
|
|
|
|
|
lock_page(page);
|
2012-03-29 20:57:44 +07:00
|
|
|
/*
|
|
|
|
|
* we unlocked the page above, so we need check if
|
|
|
|
|
* it was released or not.
|
|
|
|
|
*/
|
|
|
|
|
if (page->mapping != inode->i_mapping) {
|
|
|
|
|
unlock_page(page);
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
put_page(page);
|
2012-03-29 20:57:44 +07:00
|
|
|
goto again;
|
|
|
|
|
}
|
2012-02-16 14:01:24 +07:00
|
|
|
}
|
|
|
|
|
|
2011-05-25 02:35:30 +07:00
|
|
|
if (!PageUptodate(page)) {
|
|
|
|
|
btrfs_readpage(NULL, page);
|
|
|
|
|
lock_page(page);
|
|
|
|
|
if (!PageUptodate(page)) {
|
|
|
|
|
unlock_page(page);
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
put_page(page);
|
2011-05-25 02:35:30 +07:00
|
|
|
ret = -EIO;
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
}
|
2012-02-16 14:01:24 +07:00
|
|
|
|
|
|
|
|
if (page->mapping != inode->i_mapping) {
|
|
|
|
|
unlock_page(page);
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
put_page(page);
|
2012-02-16 14:01:24 +07:00
|
|
|
goto again;
|
|
|
|
|
}
|
|
|
|
|
|
2011-05-25 02:35:30 +07:00
|
|
|
pages[i] = page;
|
|
|
|
|
i_done++;
|
|
|
|
|
}
|
|
|
|
|
if (!i_done || ret)
|
|
|
|
|
goto out;
|
|
|
|
|
|
2017-11-28 04:05:09 +07:00
|
|
|
if (!(inode->i_sb->s_flags & SB_ACTIVE))
|
2011-05-25 02:35:30 +07:00
|
|
|
goto out;
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* so now we have a nice long stream of locked
|
|
|
|
|
* and up to date pages, lets wait on them
|
|
|
|
|
*/
|
|
|
|
|
for (i = 0; i < i_done; i++)
|
|
|
|
|
wait_on_page_writeback(pages[i]);
|
|
|
|
|
|
|
|
|
|
page_start = page_offset(pages[0]);
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
page_end = page_offset(pages[i_done - 1]) + PAGE_SIZE;
|
2011-05-25 02:35:30 +07:00
|
|
|
|
|
|
|
|
lock_extent_bits(&BTRFS_I(inode)->io_tree,
|
2015-12-03 20:30:40 +07:00
|
|
|
page_start, page_end - 1, &cached_state);
|
2011-05-25 02:35:30 +07:00
|
|
|
clear_extent_bit(&BTRFS_I(inode)->io_tree, page_start,
|
|
|
|
|
page_end - 1, EXTENT_DIRTY | EXTENT_DELALLOC |
|
2012-09-06 08:10:51 +07:00
|
|
|
EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG, 0, 0,
|
2017-10-31 22:37:52 +07:00
|
|
|
&cached_state);
|
2011-05-25 02:35:30 +07:00
|
|
|
|
2012-03-29 20:57:44 +07:00
|
|
|
if (i_done != page_cnt) {
|
2011-07-15 22:16:44 +07:00
|
|
|
spin_lock(&BTRFS_I(inode)->lock);
|
|
|
|
|
BTRFS_I(inode)->outstanding_extents++;
|
|
|
|
|
spin_unlock(&BTRFS_I(inode)->lock);
|
btrfs: qgroup: Fix qgroup reserved space underflow by only freeing reserved ranges
[BUG]
For the following case, btrfs can underflow qgroup reserved space
at an error path:
(Page size 4K, function name without "btrfs_" prefix)
Task A | Task B
----------------------------------------------------------------------
Buffered_write [0, 2K) |
|- check_data_free_space() |
| |- qgroup_reserve_data() |
| Range aligned to page |
| range [0, 4K) <<< |
| 4K bytes reserved <<< |
|- copy pages to page cache |
| Buffered_write [2K, 4K)
| |- check_data_free_space()
| | |- qgroup_reserved_data()
| | Range alinged to page
| | range [0, 4K)
| | Already reserved by A <<<
| | 0 bytes reserved <<<
| |- delalloc_reserve_metadata()
| | And it *FAILED* (Maybe EQUOTA)
| |- free_reserved_data_space()
|- qgroup_free_data()
Range aligned to page range
[0, 4K)
Freeing 4K
(Special thanks to Chandan for the detailed report and analyse)
[CAUSE]
Above Task B is freeing reserved data range [0, 4K) which is actually
reserved by Task A.
And at writeback time, page dirty by Task A will go through writeback
routine, which will free 4K reserved data space at file extent insert
time, causing the qgroup underflow.
[FIX]
For btrfs_qgroup_free_data(), add @reserved parameter to only free
data ranges reserved by previous btrfs_qgroup_reserve_data().
So in above case, Task B will try to free 0 byte, so no underflow.
Reported-by: Chandan Rajendra <chandan@linux.vnet.ibm.com>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Reviewed-by: Chandan Rajendra <chandan@linux.vnet.ibm.com>
Tested-by: Chandan Rajendra <chandan@linux.vnet.ibm.com>
Signed-off-by: David Sterba <dsterba@suse.com>
2017-02-27 14:10:39 +07:00
|
|
|
btrfs_delalloc_release_space(inode, data_reserved,
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
start_index << PAGE_SHIFT,
|
btrfs: qgroup: Use separate meta reservation type for delalloc
Before this patch, btrfs qgroup is mixing per-transcation meta rsv with
preallocated meta rsv, making it quite easy to underflow qgroup meta
reservation.
Since we have the new qgroup meta rsv types, apply it to delalloc
reservation.
Now for delalloc, most of its reserved space will use META_PREALLOC qgroup
rsv type.
And for callers reducing outstanding extent like btrfs_finish_ordered_io(),
they will convert corresponding META_PREALLOC reservation to
META_PERTRANS.
This is mainly due to the fact that current qgroup numbers will only be
updated in btrfs_commit_transaction(), that's to say if we don't keep
such placeholder reservation, we can exceed qgroup limitation.
And for callers freeing outstanding extent in error handler, we will
just free META_PREALLOC bytes.
This behavior makes callers of btrfs_qgroup_release_meta() or
btrfs_qgroup_convert_meta() to be aware of which type they are.
So in this patch, btrfs_delalloc_release_metadata() and its callers get
an extra parameter to info qgroup to do correct meta convert/release.
The good news is, even we use the wrong type (convert or free), it won't
cause obvious bug, as prealloc type is always in good shape, and the
type only affects how per-trans meta is increased or not.
So the worst case will be at most metadata limitation can be sometimes
exceeded (no convert at all) or metadata limitation is reached too soon
(no free at all).
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
2017-12-12 14:34:32 +07:00
|
|
|
(page_cnt - i_done) << PAGE_SHIFT, true);
|
2011-05-25 02:35:30 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
2012-09-06 08:10:51 +07:00
|
|
|
set_extent_defrag(&BTRFS_I(inode)->io_tree, page_start, page_end - 1,
|
2016-04-27 04:54:39 +07:00
|
|
|
&cached_state);
|
2011-05-25 02:35:30 +07:00
|
|
|
|
|
|
|
|
unlock_extent_cached(&BTRFS_I(inode)->io_tree,
|
2017-12-13 03:43:52 +07:00
|
|
|
page_start, page_end - 1, &cached_state);
|
2011-05-25 02:35:30 +07:00
|
|
|
|
|
|
|
|
for (i = 0; i < i_done; i++) {
|
|
|
|
|
clear_page_dirty_for_io(pages[i]);
|
|
|
|
|
ClearPageChecked(pages[i]);
|
|
|
|
|
set_page_extent_mapped(pages[i]);
|
|
|
|
|
set_page_dirty(pages[i]);
|
|
|
|
|
unlock_page(pages[i]);
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
put_page(pages[i]);
|
2011-05-25 02:35:30 +07:00
|
|
|
}
|
btrfs: qgroup: Use separate meta reservation type for delalloc
Before this patch, btrfs qgroup is mixing per-transcation meta rsv with
preallocated meta rsv, making it quite easy to underflow qgroup meta
reservation.
Since we have the new qgroup meta rsv types, apply it to delalloc
reservation.
Now for delalloc, most of its reserved space will use META_PREALLOC qgroup
rsv type.
And for callers reducing outstanding extent like btrfs_finish_ordered_io(),
they will convert corresponding META_PREALLOC reservation to
META_PERTRANS.
This is mainly due to the fact that current qgroup numbers will only be
updated in btrfs_commit_transaction(), that's to say if we don't keep
such placeholder reservation, we can exceed qgroup limitation.
And for callers freeing outstanding extent in error handler, we will
just free META_PREALLOC bytes.
This behavior makes callers of btrfs_qgroup_release_meta() or
btrfs_qgroup_convert_meta() to be aware of which type they are.
So in this patch, btrfs_delalloc_release_metadata() and its callers get
an extra parameter to info qgroup to do correct meta convert/release.
The good news is, even we use the wrong type (convert or free), it won't
cause obvious bug, as prealloc type is always in good shape, and the
type only affects how per-trans meta is increased or not.
So the worst case will be at most metadata limitation can be sometimes
exceeded (no convert at all) or metadata limitation is reached too soon
(no free at all).
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
2017-12-12 14:34:32 +07:00
|
|
|
btrfs_delalloc_release_extents(BTRFS_I(inode), page_cnt << PAGE_SHIFT,
|
|
|
|
|
false);
|
2017-02-27 14:10:38 +07:00
|
|
|
extent_changeset_free(data_reserved);
|
2011-05-25 02:35:30 +07:00
|
|
|
return i_done;
|
|
|
|
|
out:
|
|
|
|
|
for (i = 0; i < i_done; i++) {
|
|
|
|
|
unlock_page(pages[i]);
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
put_page(pages[i]);
|
2011-05-25 02:35:30 +07:00
|
|
|
}
|
btrfs: qgroup: Fix qgroup reserved space underflow by only freeing reserved ranges
[BUG]
For the following case, btrfs can underflow qgroup reserved space
at an error path:
(Page size 4K, function name without "btrfs_" prefix)
Task A | Task B
----------------------------------------------------------------------
Buffered_write [0, 2K) |
|- check_data_free_space() |
| |- qgroup_reserve_data() |
| Range aligned to page |
| range [0, 4K) <<< |
| 4K bytes reserved <<< |
|- copy pages to page cache |
| Buffered_write [2K, 4K)
| |- check_data_free_space()
| | |- qgroup_reserved_data()
| | Range alinged to page
| | range [0, 4K)
| | Already reserved by A <<<
| | 0 bytes reserved <<<
| |- delalloc_reserve_metadata()
| | And it *FAILED* (Maybe EQUOTA)
| |- free_reserved_data_space()
|- qgroup_free_data()
Range aligned to page range
[0, 4K)
Freeing 4K
(Special thanks to Chandan for the detailed report and analyse)
[CAUSE]
Above Task B is freeing reserved data range [0, 4K) which is actually
reserved by Task A.
And at writeback time, page dirty by Task A will go through writeback
routine, which will free 4K reserved data space at file extent insert
time, causing the qgroup underflow.
[FIX]
For btrfs_qgroup_free_data(), add @reserved parameter to only free
data ranges reserved by previous btrfs_qgroup_reserve_data().
So in above case, Task B will try to free 0 byte, so no underflow.
Reported-by: Chandan Rajendra <chandan@linux.vnet.ibm.com>
Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Reviewed-by: Chandan Rajendra <chandan@linux.vnet.ibm.com>
Tested-by: Chandan Rajendra <chandan@linux.vnet.ibm.com>
Signed-off-by: David Sterba <dsterba@suse.com>
2017-02-27 14:10:39 +07:00
|
|
|
btrfs_delalloc_release_space(inode, data_reserved,
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
start_index << PAGE_SHIFT,
|
btrfs: qgroup: Use separate meta reservation type for delalloc
Before this patch, btrfs qgroup is mixing per-transcation meta rsv with
preallocated meta rsv, making it quite easy to underflow qgroup meta
reservation.
Since we have the new qgroup meta rsv types, apply it to delalloc
reservation.
Now for delalloc, most of its reserved space will use META_PREALLOC qgroup
rsv type.
And for callers reducing outstanding extent like btrfs_finish_ordered_io(),
they will convert corresponding META_PREALLOC reservation to
META_PERTRANS.
This is mainly due to the fact that current qgroup numbers will only be
updated in btrfs_commit_transaction(), that's to say if we don't keep
such placeholder reservation, we can exceed qgroup limitation.
And for callers freeing outstanding extent in error handler, we will
just free META_PREALLOC bytes.
This behavior makes callers of btrfs_qgroup_release_meta() or
btrfs_qgroup_convert_meta() to be aware of which type they are.
So in this patch, btrfs_delalloc_release_metadata() and its callers get
an extra parameter to info qgroup to do correct meta convert/release.
The good news is, even we use the wrong type (convert or free), it won't
cause obvious bug, as prealloc type is always in good shape, and the
type only affects how per-trans meta is increased or not.
So the worst case will be at most metadata limitation can be sometimes
exceeded (no convert at all) or metadata limitation is reached too soon
(no free at all).
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
2017-12-12 14:34:32 +07:00
|
|
|
page_cnt << PAGE_SHIFT, true);
|
|
|
|
|
btrfs_delalloc_release_extents(BTRFS_I(inode), page_cnt << PAGE_SHIFT,
|
|
|
|
|
true);
|
2017-02-27 14:10:38 +07:00
|
|
|
extent_changeset_free(data_reserved);
|
2011-05-25 02:35:30 +07:00
|
|
|
return ret;
|
|
|
|
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
int btrfs_defrag_file(struct inode *inode, struct file *file,
|
|
|
|
|
struct btrfs_ioctl_defrag_range_args *range,
|
|
|
|
|
u64 newer_than, unsigned long max_to_defrag)
|
|
|
|
|
{
|
2016-06-23 05:54:23 +07:00
|
|
|
struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
|
2011-05-25 02:35:30 +07:00
|
|
|
struct btrfs_root *root = BTRFS_I(inode)->root;
|
|
|
|
|
struct file_ra_state *ra = NULL;
|
2008-06-12 08:53:53 +07:00
|
|
|
unsigned long last_index;
|
2011-09-02 14:56:39 +07:00
|
|
|
u64 isize = i_size_read(inode);
|
2010-03-10 22:52:59 +07:00
|
|
|
u64 last_len = 0;
|
|
|
|
|
u64 skip = 0;
|
|
|
|
|
u64 defrag_end = 0;
|
2011-05-25 02:35:30 +07:00
|
|
|
u64 newer_off = range->start;
|
2008-06-12 08:53:53 +07:00
|
|
|
unsigned long i;
|
2011-09-02 14:57:07 +07:00
|
|
|
unsigned long ra_index = 0;
|
2008-06-12 08:53:53 +07:00
|
|
|
int ret;
|
2011-05-25 02:35:30 +07:00
|
|
|
int defrag_count = 0;
|
2010-10-25 14:12:50 +07:00
|
|
|
int compress_type = BTRFS_COMPRESS_ZLIB;
|
2014-07-29 22:32:10 +07:00
|
|
|
u32 extent_thresh = range->extent_thresh;
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
unsigned long max_cluster = SZ_256K >> PAGE_SHIFT;
|
2014-01-22 02:18:29 +07:00
|
|
|
unsigned long cluster = max_cluster;
|
2015-12-14 23:42:10 +07:00
|
|
|
u64 new_align = ~((u64)SZ_128K - 1);
|
2011-05-25 02:35:30 +07:00
|
|
|
struct page **pages = NULL;
|
2017-07-18 01:01:59 +07:00
|
|
|
bool do_compress = range->flags & BTRFS_DEFRAG_RANGE_COMPRESS;
|
2011-05-25 02:35:30 +07:00
|
|
|
|
2013-04-16 16:20:28 +07:00
|
|
|
if (isize == 0)
|
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
|
|
if (range->start >= isize)
|
|
|
|
|
return -EINVAL;
|
2010-10-25 14:12:50 +07:00
|
|
|
|
2017-07-18 01:01:59 +07:00
|
|
|
if (do_compress) {
|
2010-10-25 14:12:50 +07:00
|
|
|
if (range->compress_type > BTRFS_COMPRESS_TYPES)
|
|
|
|
|
return -EINVAL;
|
|
|
|
|
if (range->compress_type)
|
|
|
|
|
compress_type = range->compress_type;
|
|
|
|
|
}
|
2008-06-12 08:53:53 +07:00
|
|
|
|
2013-04-16 16:20:28 +07:00
|
|
|
if (extent_thresh == 0)
|
2015-12-14 23:42:10 +07:00
|
|
|
extent_thresh = SZ_256K;
|
2010-03-10 22:52:59 +07:00
|
|
|
|
2011-05-25 02:35:30 +07:00
|
|
|
/*
|
2017-06-22 08:22:58 +07:00
|
|
|
* If we were not given a file, allocate a readahead context. As
|
|
|
|
|
* readahead is just an optimization, defrag will work without it so
|
|
|
|
|
* we don't error out.
|
2011-05-25 02:35:30 +07:00
|
|
|
*/
|
|
|
|
|
if (!file) {
|
2017-06-22 08:13:02 +07:00
|
|
|
ra = kzalloc(sizeof(*ra), GFP_KERNEL);
|
2017-06-22 08:22:58 +07:00
|
|
|
if (ra)
|
|
|
|
|
file_ra_state_init(ra, inode->i_mapping);
|
2011-05-25 02:35:30 +07:00
|
|
|
} else {
|
|
|
|
|
ra = &file->f_ra;
|
|
|
|
|
}
|
|
|
|
|
|
2017-06-22 08:13:02 +07:00
|
|
|
pages = kmalloc_array(max_cluster, sizeof(struct page *), GFP_KERNEL);
|
2011-05-25 02:35:30 +07:00
|
|
|
if (!pages) {
|
|
|
|
|
ret = -ENOMEM;
|
|
|
|
|
goto out_ra;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* find the last page to defrag */
|
2010-03-11 21:42:04 +07:00
|
|
|
if (range->start + range->len > range->start) {
|
2011-09-02 14:56:39 +07:00
|
|
|
last_index = min_t(u64, isize - 1,
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
range->start + range->len - 1) >> PAGE_SHIFT;
|
2010-03-11 21:42:04 +07:00
|
|
|
} else {
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
last_index = (isize - 1) >> PAGE_SHIFT;
|
2010-03-11 21:42:04 +07:00
|
|
|
}
|
|
|
|
|
|
2011-05-25 02:35:30 +07:00
|
|
|
if (newer_than) {
|
|
|
|
|
ret = find_new_extents(root, inode, newer_than,
|
2015-12-14 23:42:10 +07:00
|
|
|
&newer_off, SZ_64K);
|
2011-05-25 02:35:30 +07:00
|
|
|
if (!ret) {
|
|
|
|
|
range->start = newer_off;
|
|
|
|
|
/*
|
|
|
|
|
* we always align our defrag to help keep
|
|
|
|
|
* the extents in the file evenly spaced
|
|
|
|
|
*/
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
i = (newer_off & new_align) >> PAGE_SHIFT;
|
2011-05-25 02:35:30 +07:00
|
|
|
} else
|
|
|
|
|
goto out_ra;
|
|
|
|
|
} else {
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
i = range->start >> PAGE_SHIFT;
|
2011-05-25 02:35:30 +07:00
|
|
|
}
|
|
|
|
|
if (!max_to_defrag)
|
2015-06-09 12:05:11 +07:00
|
|
|
max_to_defrag = last_index - i + 1;
|
2011-05-25 02:35:30 +07:00
|
|
|
|
2011-10-11 02:43:34 +07:00
|
|
|
/*
|
|
|
|
|
* make writeback starts from i, so the defrag range can be
|
|
|
|
|
* written sequentially.
|
|
|
|
|
*/
|
|
|
|
|
if (i < inode->i_mapping->writeback_index)
|
|
|
|
|
inode->i_mapping->writeback_index = i;
|
|
|
|
|
|
2011-10-11 22:41:40 +07:00
|
|
|
while (i <= last_index && defrag_count < max_to_defrag &&
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
(i < DIV_ROUND_UP(i_size_read(inode), PAGE_SIZE))) {
|
2011-05-25 02:35:30 +07:00
|
|
|
/*
|
|
|
|
|
* make sure we stop running if someone unmounts
|
|
|
|
|
* the FS
|
|
|
|
|
*/
|
2017-11-28 04:05:09 +07:00
|
|
|
if (!(inode->i_sb->s_flags & SB_ACTIVE))
|
2011-05-25 02:35:30 +07:00
|
|
|
break;
|
|
|
|
|
|
2016-06-23 05:54:23 +07:00
|
|
|
if (btrfs_defrag_cancelled(fs_info)) {
|
|
|
|
|
btrfs_debug(fs_info, "defrag_file cancelled");
|
2013-02-10 06:38:06 +07:00
|
|
|
ret = -EAGAIN;
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
if (!should_defrag_range(inode, (u64)i << PAGE_SHIFT,
|
2012-06-11 15:03:35 +07:00
|
|
|
extent_thresh, &last_len, &skip,
|
2017-07-18 01:01:59 +07:00
|
|
|
&defrag_end, do_compress)){
|
2010-03-10 22:52:59 +07:00
|
|
|
unsigned long next;
|
|
|
|
|
/*
|
|
|
|
|
* the should_defrag function tells us how much to skip
|
|
|
|
|
* bump our counter by the suggested amount
|
|
|
|
|
*/
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
next = DIV_ROUND_UP(skip, PAGE_SIZE);
|
2010-03-10 22:52:59 +07:00
|
|
|
i = max(i + 1, next);
|
|
|
|
|
continue;
|
|
|
|
|
}
|
2011-09-02 14:57:07 +07:00
|
|
|
|
|
|
|
|
if (!newer_than) {
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
cluster = (PAGE_ALIGN(defrag_end) >>
|
|
|
|
|
PAGE_SHIFT) - i;
|
2011-09-02 14:57:07 +07:00
|
|
|
cluster = min(cluster, max_cluster);
|
|
|
|
|
} else {
|
|
|
|
|
cluster = max_cluster;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (i + cluster > ra_index) {
|
|
|
|
|
ra_index = max(i, ra_index);
|
2017-06-22 08:22:58 +07:00
|
|
|
if (ra)
|
2017-06-22 08:35:28 +07:00
|
|
|
page_cache_sync_readahead(inode->i_mapping, ra,
|
|
|
|
|
file, ra_index, cluster);
|
2015-06-09 19:08:32 +07:00
|
|
|
ra_index += cluster;
|
2011-09-02 14:57:07 +07:00
|
|
|
}
|
2010-03-10 22:52:59 +07:00
|
|
|
|
2016-01-23 03:40:57 +07:00
|
|
|
inode_lock(inode);
|
2017-07-18 01:01:59 +07:00
|
|
|
if (do_compress)
|
2017-07-18 00:41:31 +07:00
|
|
|
BTRFS_I(inode)->defrag_compress = compress_type;
|
2011-09-02 14:57:07 +07:00
|
|
|
ret = cluster_pages_for_defrag(inode, pages, i, cluster);
|
2012-03-29 20:57:44 +07:00
|
|
|
if (ret < 0) {
|
2016-01-23 03:40:57 +07:00
|
|
|
inode_unlock(inode);
|
2011-05-25 02:35:30 +07:00
|
|
|
goto out_ra;
|
2012-03-29 20:57:44 +07:00
|
|
|
}
|
2011-05-25 02:35:30 +07:00
|
|
|
|
|
|
|
|
defrag_count += ret;
|
2012-12-12 07:00:21 +07:00
|
|
|
balance_dirty_pages_ratelimited(inode->i_mapping);
|
2016-01-23 03:40:57 +07:00
|
|
|
inode_unlock(inode);
|
2011-05-25 02:35:30 +07:00
|
|
|
|
|
|
|
|
if (newer_than) {
|
|
|
|
|
if (newer_off == (u64)-1)
|
|
|
|
|
break;
|
|
|
|
|
|
2012-03-29 20:57:45 +07:00
|
|
|
if (ret > 0)
|
|
|
|
|
i += ret;
|
|
|
|
|
|
2011-05-25 02:35:30 +07:00
|
|
|
newer_off = max(newer_off + 1,
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
(u64)i << PAGE_SHIFT);
|
2011-05-25 02:35:30 +07:00
|
|
|
|
2015-12-14 23:42:10 +07:00
|
|
|
ret = find_new_extents(root, inode, newer_than,
|
|
|
|
|
&newer_off, SZ_64K);
|
2011-05-25 02:35:30 +07:00
|
|
|
if (!ret) {
|
|
|
|
|
range->start = newer_off;
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
i = (newer_off & new_align) >> PAGE_SHIFT;
|
2011-05-25 02:35:30 +07:00
|
|
|
} else {
|
|
|
|
|
break;
|
2008-06-12 08:53:53 +07:00
|
|
|
}
|
2011-05-25 02:35:30 +07:00
|
|
|
} else {
|
2011-09-02 14:57:07 +07:00
|
|
|
if (ret > 0) {
|
2011-09-02 14:56:25 +07:00
|
|
|
i += ret;
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
last_len += ret << PAGE_SHIFT;
|
2011-09-02 14:57:07 +07:00
|
|
|
} else {
|
2011-09-02 14:56:25 +07:00
|
|
|
i++;
|
2011-09-02 14:57:07 +07:00
|
|
|
last_len = 0;
|
|
|
|
|
}
|
2008-06-12 08:53:53 +07:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2014-03-01 17:55:54 +07:00
|
|
|
if ((range->flags & BTRFS_DEFRAG_RANGE_START_IO)) {
|
2010-03-11 21:42:04 +07:00
|
|
|
filemap_flush(inode->i_mapping);
|
2014-03-01 17:55:54 +07:00
|
|
|
if (test_bit(BTRFS_INODE_HAS_ASYNC_EXTENT,
|
|
|
|
|
&BTRFS_I(inode)->runtime_flags))
|
|
|
|
|
filemap_flush(inode->i_mapping);
|
|
|
|
|
}
|
2010-03-11 21:42:04 +07:00
|
|
|
|
2010-10-25 14:12:50 +07:00
|
|
|
if (range->compress_type == BTRFS_COMPRESS_LZO) {
|
2016-06-23 05:54:23 +07:00
|
|
|
btrfs_set_fs_incompat(fs_info, COMPRESS_LZO);
|
btrfs: Add zstd support
Add zstd compression and decompression support to BtrFS. zstd at its
fastest level compresses almost as well as zlib, while offering much
faster compression and decompression, approaching lzo speeds.
I benchmarked btrfs with zstd compression against no compression, lzo
compression, and zlib compression. I benchmarked two scenarios. Copying
a set of files to btrfs, and then reading the files. Copying a tarball
to btrfs, extracting it to btrfs, and then reading the extracted files.
After every operation, I call `sync` and include the sync time.
Between every pair of operations I unmount and remount the filesystem
to avoid caching. The benchmark files can be found in the upstream
zstd source repository under
`contrib/linux-kernel/{btrfs-benchmark.sh,btrfs-extract-benchmark.sh}`
[1] [2].
I ran the benchmarks on a Ubuntu 14.04 VM with 2 cores and 4 GiB of RAM.
The VM is running on a MacBook Pro with a 3.1 GHz Intel Core i7 processor,
16 GB of RAM, and a SSD.
The first compression benchmark is copying 10 copies of the unzipped
Silesia corpus [3] into a BtrFS filesystem mounted with
`-o compress-force=Method`. The decompression benchmark times how long
it takes to `tar` all 10 copies into `/dev/null`. The compression ratio is
measured by comparing the output of `df` and `du`. See the benchmark file
[1] for details. I benchmarked multiple zstd compression levels, although
the patch uses zstd level 1.
| Method | Ratio | Compression MB/s | Decompression speed |
|---------|-------|------------------|---------------------|
| None | 0.99 | 504 | 686 |
| lzo | 1.66 | 398 | 442 |
| zlib | 2.58 | 65 | 241 |
| zstd 1 | 2.57 | 260 | 383 |
| zstd 3 | 2.71 | 174 | 408 |
| zstd 6 | 2.87 | 70 | 398 |
| zstd 9 | 2.92 | 43 | 406 |
| zstd 12 | 2.93 | 21 | 408 |
| zstd 15 | 3.01 | 11 | 354 |
The next benchmark first copies `linux-4.11.6.tar` [4] to btrfs. Then it
measures the compression ratio, extracts the tar, and deletes the tar.
Then it measures the compression ratio again, and `tar`s the extracted
files into `/dev/null`. See the benchmark file [2] for details.
| Method | Tar Ratio | Extract Ratio | Copy (s) | Extract (s)| Read (s) |
|--------|-----------|---------------|----------|------------|----------|
| None | 0.97 | 0.78 | 0.981 | 5.501 | 8.807 |
| lzo | 2.06 | 1.38 | 1.631 | 8.458 | 8.585 |
| zlib | 3.40 | 1.86 | 7.750 | 21.544 | 11.744 |
| zstd 1 | 3.57 | 1.85 | 2.579 | 11.479 | 9.389 |
[1] https://github.com/facebook/zstd/blob/dev/contrib/linux-kernel/btrfs-benchmark.sh
[2] https://github.com/facebook/zstd/blob/dev/contrib/linux-kernel/btrfs-extract-benchmark.sh
[3] http://sun.aei.polsl.pl/~sdeor/index.php?page=silesia
[4] https://cdn.kernel.org/pub/linux/kernel/v4.x/linux-4.11.6.tar.xz
zstd source repository: https://github.com/facebook/zstd
Signed-off-by: Nick Terrell <terrelln@fb.com>
Signed-off-by: Chris Mason <clm@fb.com>
2017-08-10 09:39:02 +07:00
|
|
|
} else if (range->compress_type == BTRFS_COMPRESS_ZSTD) {
|
|
|
|
|
btrfs_set_fs_incompat(fs_info, COMPRESS_ZSTD);
|
2010-10-25 14:12:50 +07:00
|
|
|
}
|
|
|
|
|
|
2011-09-01 21:33:57 +07:00
|
|
|
ret = defrag_count;
|
2010-03-10 22:52:59 +07:00
|
|
|
|
2011-05-25 02:35:30 +07:00
|
|
|
out_ra:
|
2017-07-18 01:01:59 +07:00
|
|
|
if (do_compress) {
|
2016-01-23 03:40:57 +07:00
|
|
|
inode_lock(inode);
|
2017-07-18 00:41:31 +07:00
|
|
|
BTRFS_I(inode)->defrag_compress = BTRFS_COMPRESS_NONE;
|
2016-01-23 03:40:57 +07:00
|
|
|
inode_unlock(inode);
|
2013-08-16 21:23:33 +07:00
|
|
|
}
|
2011-05-25 02:35:30 +07:00
|
|
|
if (!file)
|
|
|
|
|
kfree(ra);
|
|
|
|
|
kfree(pages);
|
2010-03-10 22:52:59 +07:00
|
|
|
return ret;
|
2008-06-12 08:53:53 +07:00
|
|
|
}
|
|
|
|
|
|
2012-11-26 15:43:45 +07:00
|
|
|
static noinline int btrfs_ioctl_resize(struct file *file,
|
2009-09-22 03:00:26 +07:00
|
|
|
void __user *arg)
|
2008-06-12 08:53:53 +07:00
|
|
|
{
|
2016-06-23 05:54:23 +07:00
|
|
|
struct inode *inode = file_inode(file);
|
|
|
|
|
struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
|
2008-06-12 08:53:53 +07:00
|
|
|
u64 new_size;
|
|
|
|
|
u64 old_size;
|
|
|
|
|
u64 devid = 1;
|
2016-06-23 05:54:23 +07:00
|
|
|
struct btrfs_root *root = BTRFS_I(inode)->root;
|
2008-06-12 08:53:53 +07:00
|
|
|
struct btrfs_ioctl_vol_args *vol_args;
|
|
|
|
|
struct btrfs_trans_handle *trans;
|
|
|
|
|
struct btrfs_device *device = NULL;
|
|
|
|
|
char *sizestr;
|
2014-03-31 17:03:25 +07:00
|
|
|
char *retptr;
|
2008-06-12 08:53:53 +07:00
|
|
|
char *devstr = NULL;
|
|
|
|
|
int ret = 0;
|
|
|
|
|
int mod = 0;
|
|
|
|
|
|
2009-01-06 04:57:23 +07:00
|
|
|
if (!capable(CAP_SYS_ADMIN))
|
|
|
|
|
return -EPERM;
|
|
|
|
|
|
2012-11-26 15:43:45 +07:00
|
|
|
ret = mnt_want_write_file(file);
|
|
|
|
|
if (ret)
|
|
|
|
|
return ret;
|
|
|
|
|
|
2017-03-28 19:44:21 +07:00
|
|
|
if (test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags)) {
|
2012-12-21 17:38:50 +07:00
|
|
|
mnt_drop_write_file(file);
|
2013-08-21 10:44:48 +07:00
|
|
|
return BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS;
|
2012-01-17 03:04:47 +07:00
|
|
|
}
|
|
|
|
|
|
2016-06-23 05:54:23 +07:00
|
|
|
mutex_lock(&fs_info->volume_mutex);
|
2009-04-08 14:06:54 +07:00
|
|
|
vol_args = memdup_user(arg, sizeof(*vol_args));
|
2012-01-17 03:04:47 +07:00
|
|
|
if (IS_ERR(vol_args)) {
|
|
|
|
|
ret = PTR_ERR(vol_args);
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
2008-07-24 23:20:14 +07:00
|
|
|
|
|
|
|
|
vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
|
2008-06-12 08:53:53 +07:00
|
|
|
|
|
|
|
|
sizestr = vol_args->name;
|
|
|
|
|
devstr = strchr(sizestr, ':');
|
|
|
|
|
if (devstr) {
|
|
|
|
|
sizestr = devstr + 1;
|
|
|
|
|
*devstr = '\0';
|
|
|
|
|
devstr = vol_args->name;
|
2014-05-13 15:36:08 +07:00
|
|
|
ret = kstrtoull(devstr, 10, &devid);
|
|
|
|
|
if (ret)
|
|
|
|
|
goto out_free;
|
2012-12-21 16:21:30 +07:00
|
|
|
if (!devid) {
|
|
|
|
|
ret = -EINVAL;
|
|
|
|
|
goto out_free;
|
|
|
|
|
}
|
2016-06-23 05:54:23 +07:00
|
|
|
btrfs_info(fs_info, "resizing devid %llu", devid);
|
2008-06-12 08:53:53 +07:00
|
|
|
}
|
2012-12-21 16:19:51 +07:00
|
|
|
|
2016-06-23 05:54:23 +07:00
|
|
|
device = btrfs_find_device(fs_info, devid, NULL, NULL);
|
2008-06-12 08:53:53 +07:00
|
|
|
if (!device) {
|
2016-06-23 05:54:23 +07:00
|
|
|
btrfs_info(fs_info, "resizer unable to find device %llu",
|
|
|
|
|
devid);
|
2012-12-21 16:21:30 +07:00
|
|
|
ret = -ENODEV;
|
2012-01-17 03:04:47 +07:00
|
|
|
goto out_free;
|
2008-06-12 08:53:53 +07:00
|
|
|
}
|
2012-12-21 16:19:51 +07:00
|
|
|
|
2017-12-04 11:54:52 +07:00
|
|
|
if (!test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state)) {
|
2016-06-23 05:54:23 +07:00
|
|
|
btrfs_info(fs_info,
|
2013-12-20 23:37:06 +07:00
|
|
|
"resizer unable to apply on readonly device %llu",
|
2013-08-20 18:20:07 +07:00
|
|
|
devid);
|
2012-12-21 16:21:30 +07:00
|
|
|
ret = -EPERM;
|
2012-06-14 15:23:19 +07:00
|
|
|
goto out_free;
|
|
|
|
|
}
|
|
|
|
|
|
2008-06-12 08:53:53 +07:00
|
|
|
if (!strcmp(sizestr, "max"))
|
|
|
|
|
new_size = device->bdev->bd_inode->i_size;
|
|
|
|
|
else {
|
|
|
|
|
if (sizestr[0] == '-') {
|
|
|
|
|
mod = -1;
|
|
|
|
|
sizestr++;
|
|
|
|
|
} else if (sizestr[0] == '+') {
|
|
|
|
|
mod = 1;
|
|
|
|
|
sizestr++;
|
|
|
|
|
}
|
2014-03-31 17:03:25 +07:00
|
|
|
new_size = memparse(sizestr, &retptr);
|
|
|
|
|
if (*retptr != '\0' || new_size == 0) {
|
2008-06-12 08:53:53 +07:00
|
|
|
ret = -EINVAL;
|
2012-01-17 03:04:47 +07:00
|
|
|
goto out_free;
|
2008-06-12 08:53:53 +07:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2017-12-04 11:54:55 +07:00
|
|
|
if (test_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state)) {
|
2012-12-21 16:21:30 +07:00
|
|
|
ret = -EPERM;
|
2012-11-06 00:29:28 +07:00
|
|
|
goto out_free;
|
|
|
|
|
}
|
|
|
|
|
|
2014-09-03 20:35:38 +07:00
|
|
|
old_size = btrfs_device_get_total_bytes(device);
|
2008-06-12 08:53:53 +07:00
|
|
|
|
|
|
|
|
if (mod < 0) {
|
|
|
|
|
if (new_size > old_size) {
|
|
|
|
|
ret = -EINVAL;
|
2012-01-17 03:04:47 +07:00
|
|
|
goto out_free;
|
2008-06-12 08:53:53 +07:00
|
|
|
}
|
|
|
|
|
new_size = old_size - new_size;
|
|
|
|
|
} else if (mod > 0) {
|
2013-12-20 14:28:56 +07:00
|
|
|
if (new_size > ULLONG_MAX - old_size) {
|
2014-05-29 08:19:58 +07:00
|
|
|
ret = -ERANGE;
|
2013-12-20 14:28:56 +07:00
|
|
|
goto out_free;
|
|
|
|
|
}
|
2008-06-12 08:53:53 +07:00
|
|
|
new_size = old_size + new_size;
|
|
|
|
|
}
|
|
|
|
|
|
2015-12-14 23:42:10 +07:00
|
|
|
if (new_size < SZ_256M) {
|
2008-06-12 08:53:53 +07:00
|
|
|
ret = -EINVAL;
|
2012-01-17 03:04:47 +07:00
|
|
|
goto out_free;
|
2008-06-12 08:53:53 +07:00
|
|
|
}
|
|
|
|
|
if (new_size > device->bdev->bd_inode->i_size) {
|
|
|
|
|
ret = -EFBIG;
|
2012-01-17 03:04:47 +07:00
|
|
|
goto out_free;
|
2008-06-12 08:53:53 +07:00
|
|
|
}
|
|
|
|
|
|
2017-07-18 19:39:08 +07:00
|
|
|
new_size = round_down(new_size, fs_info->sectorsize);
|
2008-06-12 08:53:53 +07:00
|
|
|
|
2016-06-23 05:54:23 +07:00
|
|
|
btrfs_info_in_rcu(fs_info, "new size for %s is %llu",
|
|
|
|
|
rcu_str_deref(device->name), new_size);
|
2008-06-12 08:53:53 +07:00
|
|
|
|
|
|
|
|
if (new_size > old_size) {
|
2010-05-16 21:48:46 +07:00
|
|
|
trans = btrfs_start_transaction(root, 0);
|
2011-01-20 13:19:37 +07:00
|
|
|
if (IS_ERR(trans)) {
|
|
|
|
|
ret = PTR_ERR(trans);
|
2012-01-17 03:04:47 +07:00
|
|
|
goto out_free;
|
2011-01-20 13:19:37 +07:00
|
|
|
}
|
2008-06-12 08:53:53 +07:00
|
|
|
ret = btrfs_grow_device(trans, device, new_size);
|
2016-09-10 08:39:03 +07:00
|
|
|
btrfs_commit_transaction(trans);
|
2011-11-19 01:55:01 +07:00
|
|
|
} else if (new_size < old_size) {
|
2008-06-12 08:53:53 +07:00
|
|
|
ret = btrfs_shrink_device(device, new_size);
|
2012-10-27 19:06:39 +07:00
|
|
|
} /* equal, nothing need to do */
|
2008-06-12 08:53:53 +07:00
|
|
|
|
2012-01-17 03:04:47 +07:00
|
|
|
out_free:
|
2008-06-12 08:53:53 +07:00
|
|
|
kfree(vol_args);
|
2012-01-17 03:04:47 +07:00
|
|
|
out:
|
2016-06-23 05:54:23 +07:00
|
|
|
mutex_unlock(&fs_info->volume_mutex);
|
2017-03-28 19:44:21 +07:00
|
|
|
clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
|
2013-01-20 20:57:57 +07:00
|
|
|
mnt_drop_write_file(file);
|
2008-06-12 08:53:53 +07:00
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
2010-10-30 02:41:32 +07:00
|
|
|
static noinline int btrfs_ioctl_snap_create_transid(struct file *file,
|
2017-02-15 00:33:53 +07:00
|
|
|
const char *name, unsigned long fd, int subvol,
|
2011-09-14 20:58:21 +07:00
|
|
|
u64 *transid, bool readonly,
|
2013-02-07 13:02:44 +07:00
|
|
|
struct btrfs_qgroup_inherit *inherit)
|
2008-06-12 08:53:53 +07:00
|
|
|
{
|
|
|
|
|
int namelen;
|
2008-11-18 09:02:50 +07:00
|
|
|
int ret = 0;
|
2008-06-12 08:53:53 +07:00
|
|
|
|
2016-09-21 19:31:29 +07:00
|
|
|
if (!S_ISDIR(file_inode(file)->i_mode))
|
|
|
|
|
return -ENOTDIR;
|
|
|
|
|
|
2012-06-29 16:58:46 +07:00
|
|
|
ret = mnt_want_write_file(file);
|
|
|
|
|
if (ret)
|
|
|
|
|
goto out;
|
|
|
|
|
|
2010-10-30 02:41:32 +07:00
|
|
|
namelen = strlen(name);
|
|
|
|
|
if (strchr(name, '/')) {
|
2008-06-12 08:53:53 +07:00
|
|
|
ret = -EINVAL;
|
2012-06-29 16:58:46 +07:00
|
|
|
goto out_drop_write;
|
2008-06-12 08:53:53 +07:00
|
|
|
}
|
|
|
|
|
|
2012-02-21 10:14:55 +07:00
|
|
|
if (name[0] == '.' &&
|
|
|
|
|
(namelen == 1 || (name[1] == '.' && namelen == 2))) {
|
|
|
|
|
ret = -EEXIST;
|
2012-06-29 16:58:46 +07:00
|
|
|
goto out_drop_write;
|
2012-02-21 10:14:55 +07:00
|
|
|
}
|
|
|
|
|
|
2008-11-18 09:02:50 +07:00
|
|
|
if (subvol) {
|
2010-10-30 02:41:32 +07:00
|
|
|
ret = btrfs_mksubvol(&file->f_path, name, namelen,
|
2011-09-14 20:58:21 +07:00
|
|
|
NULL, transid, readonly, inherit);
|
2008-10-10 00:39:39 +07:00
|
|
|
} else {
|
2012-08-28 23:52:22 +07:00
|
|
|
struct fd src = fdget(fd);
|
2008-11-18 09:02:50 +07:00
|
|
|
struct inode *src_inode;
|
2012-08-28 23:52:22 +07:00
|
|
|
if (!src.file) {
|
2008-11-18 09:02:50 +07:00
|
|
|
ret = -EINVAL;
|
2012-06-29 16:58:46 +07:00
|
|
|
goto out_drop_write;
|
2008-11-18 09:02:50 +07:00
|
|
|
}
|
|
|
|
|
|
2013-01-24 05:07:38 +07:00
|
|
|
src_inode = file_inode(src.file);
|
|
|
|
|
if (src_inode->i_sb != file_inode(file)->i_sb) {
|
2016-03-25 21:02:41 +07:00
|
|
|
btrfs_info(BTRFS_I(file_inode(file))->root->fs_info,
|
2013-12-20 23:37:06 +07:00
|
|
|
"Snapshot src from another FS");
|
2014-01-30 14:32:02 +07:00
|
|
|
ret = -EXDEV;
|
2014-01-16 00:15:52 +07:00
|
|
|
} else if (!inode_owner_or_capable(src_inode)) {
|
|
|
|
|
/*
|
|
|
|
|
* Subvolume creation is not restricted, but snapshots
|
|
|
|
|
* are limited to own subvolumes only
|
|
|
|
|
*/
|
|
|
|
|
ret = -EPERM;
|
2012-08-27 08:20:24 +07:00
|
|
|
} else {
|
|
|
|
|
ret = btrfs_mksubvol(&file->f_path, name, namelen,
|
|
|
|
|
BTRFS_I(src_inode)->root,
|
|
|
|
|
transid, readonly, inherit);
|
2008-11-18 09:02:50 +07:00
|
|
|
}
|
2012-08-28 23:52:22 +07:00
|
|
|
fdput(src);
|
2008-10-10 00:39:39 +07:00
|
|
|
}
|
2012-06-29 16:58:46 +07:00
|
|
|
out_drop_write:
|
|
|
|
|
mnt_drop_write_file(file);
|
2008-06-12 08:53:53 +07:00
|
|
|
out:
|
2010-10-30 02:41:32 +07:00
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static noinline int btrfs_ioctl_snap_create(struct file *file,
|
2010-12-20 14:53:28 +07:00
|
|
|
void __user *arg, int subvol)
|
2010-10-30 02:41:32 +07:00
|
|
|
{
|
2010-12-20 14:53:28 +07:00
|
|
|
struct btrfs_ioctl_vol_args *vol_args;
|
2010-10-30 02:41:32 +07:00
|
|
|
int ret;
|
|
|
|
|
|
2016-09-21 19:31:29 +07:00
|
|
|
if (!S_ISDIR(file_inode(file)->i_mode))
|
|
|
|
|
return -ENOTDIR;
|
|
|
|
|
|
2010-12-20 14:53:28 +07:00
|
|
|
vol_args = memdup_user(arg, sizeof(*vol_args));
|
|
|
|
|
if (IS_ERR(vol_args))
|
|
|
|
|
return PTR_ERR(vol_args);
|
|
|
|
|
vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
|
2010-10-30 02:41:32 +07:00
|
|
|
|
2010-12-20 14:53:28 +07:00
|
|
|
ret = btrfs_ioctl_snap_create_transid(file, vol_args->name,
|
2010-12-20 15:04:08 +07:00
|
|
|
vol_args->fd, subvol,
|
2011-09-14 20:58:21 +07:00
|
|
|
NULL, false, NULL);
|
2010-12-10 13:41:56 +07:00
|
|
|
|
2010-12-20 14:53:28 +07:00
|
|
|
kfree(vol_args);
|
|
|
|
|
return ret;
|
|
|
|
|
}
|
2010-12-10 13:41:56 +07:00
|
|
|
|
2010-12-20 14:53:28 +07:00
|
|
|
static noinline int btrfs_ioctl_snap_create_v2(struct file *file,
|
|
|
|
|
void __user *arg, int subvol)
|
|
|
|
|
{
|
|
|
|
|
struct btrfs_ioctl_vol_args_v2 *vol_args;
|
|
|
|
|
int ret;
|
|
|
|
|
u64 transid = 0;
|
|
|
|
|
u64 *ptr = NULL;
|
2010-12-20 15:04:08 +07:00
|
|
|
bool readonly = false;
|
2011-09-14 20:58:21 +07:00
|
|
|
struct btrfs_qgroup_inherit *inherit = NULL;
|
2010-12-10 07:36:28 +07:00
|
|
|
|
2016-09-21 19:31:29 +07:00
|
|
|
if (!S_ISDIR(file_inode(file)->i_mode))
|
|
|
|
|
return -ENOTDIR;
|
|
|
|
|
|
2010-12-20 14:53:28 +07:00
|
|
|
vol_args = memdup_user(arg, sizeof(*vol_args));
|
|
|
|
|
if (IS_ERR(vol_args))
|
|
|
|
|
return PTR_ERR(vol_args);
|
|
|
|
|
vol_args->name[BTRFS_SUBVOL_NAME_MAX] = '\0';
|
2010-12-10 07:36:28 +07:00
|
|
|
|
2010-12-20 15:04:08 +07:00
|
|
|
if (vol_args->flags &
|
2011-09-14 20:58:21 +07:00
|
|
|
~(BTRFS_SUBVOL_CREATE_ASYNC | BTRFS_SUBVOL_RDONLY |
|
|
|
|
|
BTRFS_SUBVOL_QGROUP_INHERIT)) {
|
2010-12-20 15:04:08 +07:00
|
|
|
ret = -EOPNOTSUPP;
|
2014-09-04 18:09:15 +07:00
|
|
|
goto free_args;
|
2010-10-30 02:41:32 +07:00
|
|
|
}
|
2010-12-20 14:53:28 +07:00
|
|
|
|
|
|
|
|
if (vol_args->flags & BTRFS_SUBVOL_CREATE_ASYNC)
|
|
|
|
|
ptr = &transid;
|
2010-12-20 15:04:08 +07:00
|
|
|
if (vol_args->flags & BTRFS_SUBVOL_RDONLY)
|
|
|
|
|
readonly = true;
|
2011-09-14 20:58:21 +07:00
|
|
|
if (vol_args->flags & BTRFS_SUBVOL_QGROUP_INHERIT) {
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
if (vol_args->size > PAGE_SIZE) {
|
2011-09-14 20:58:21 +07:00
|
|
|
ret = -EINVAL;
|
2014-09-04 18:09:15 +07:00
|
|
|
goto free_args;
|
2011-09-14 20:58:21 +07:00
|
|
|
}
|
|
|
|
|
inherit = memdup_user(vol_args->qgroup_inherit, vol_args->size);
|
|
|
|
|
if (IS_ERR(inherit)) {
|
|
|
|
|
ret = PTR_ERR(inherit);
|
2014-09-04 18:09:15 +07:00
|
|
|
goto free_args;
|
2011-09-14 20:58:21 +07:00
|
|
|
}
|
|
|
|
|
}
|
2010-12-20 14:53:28 +07:00
|
|
|
|
|
|
|
|
ret = btrfs_ioctl_snap_create_transid(file, vol_args->name,
|
2011-09-14 20:58:21 +07:00
|
|
|
vol_args->fd, subvol, ptr,
|
2013-02-07 13:02:44 +07:00
|
|
|
readonly, inherit);
|
2014-09-04 18:09:15 +07:00
|
|
|
if (ret)
|
|
|
|
|
goto free_inherit;
|
2010-12-20 14:53:28 +07:00
|
|
|
|
2014-09-04 18:09:15 +07:00
|
|
|
if (ptr && copy_to_user(arg +
|
|
|
|
|
offsetof(struct btrfs_ioctl_vol_args_v2,
|
|
|
|
|
transid),
|
|
|
|
|
ptr, sizeof(*ptr)))
|
2010-12-20 14:53:28 +07:00
|
|
|
ret = -EFAULT;
|
2014-09-04 18:09:15 +07:00
|
|
|
|
|
|
|
|
free_inherit:
|
2011-09-14 20:58:21 +07:00
|
|
|
kfree(inherit);
|
2014-09-04 18:09:15 +07:00
|
|
|
free_args:
|
|
|
|
|
kfree(vol_args);
|
2008-06-12 08:53:53 +07:00
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
2010-12-20 15:30:25 +07:00
|
|
|
static noinline int btrfs_ioctl_subvol_getflags(struct file *file,
|
|
|
|
|
void __user *arg)
|
|
|
|
|
{
|
2013-01-24 05:07:38 +07:00
|
|
|
struct inode *inode = file_inode(file);
|
2016-06-23 05:54:23 +07:00
|
|
|
struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
|
2010-12-20 15:30:25 +07:00
|
|
|
struct btrfs_root *root = BTRFS_I(inode)->root;
|
|
|
|
|
int ret = 0;
|
|
|
|
|
u64 flags = 0;
|
|
|
|
|
|
2017-01-11 01:35:31 +07:00
|
|
|
if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FIRST_FREE_OBJECTID)
|
2010-12-20 15:30:25 +07:00
|
|
|
return -EINVAL;
|
|
|
|
|
|
2016-06-23 05:54:23 +07:00
|
|
|
down_read(&fs_info->subvol_sem);
|
2010-12-20 15:30:25 +07:00
|
|
|
if (btrfs_root_readonly(root))
|
|
|
|
|
flags |= BTRFS_SUBVOL_RDONLY;
|
2016-06-23 05:54:23 +07:00
|
|
|
up_read(&fs_info->subvol_sem);
|
2010-12-20 15:30:25 +07:00
|
|
|
|
|
|
|
|
if (copy_to_user(arg, &flags, sizeof(flags)))
|
|
|
|
|
ret = -EFAULT;
|
|
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static noinline int btrfs_ioctl_subvol_setflags(struct file *file,
|
|
|
|
|
void __user *arg)
|
|
|
|
|
{
|
2013-01-24 05:07:38 +07:00
|
|
|
struct inode *inode = file_inode(file);
|
2016-06-23 05:54:23 +07:00
|
|
|
struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
|
2010-12-20 15:30:25 +07:00
|
|
|
struct btrfs_root *root = BTRFS_I(inode)->root;
|
|
|
|
|
struct btrfs_trans_handle *trans;
|
|
|
|
|
u64 root_flags;
|
|
|
|
|
u64 flags;
|
|
|
|
|
int ret = 0;
|
|
|
|
|
|
2014-01-16 21:50:22 +07:00
|
|
|
if (!inode_owner_or_capable(inode))
|
|
|
|
|
return -EPERM;
|
|
|
|
|
|
2012-06-29 16:58:49 +07:00
|
|
|
ret = mnt_want_write_file(file);
|
|
|
|
|
if (ret)
|
|
|
|
|
goto out;
|
2010-12-20 15:30:25 +07:00
|
|
|
|
2017-01-11 01:35:31 +07:00
|
|
|
if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FIRST_FREE_OBJECTID) {
|
2012-06-29 16:58:49 +07:00
|
|
|
ret = -EINVAL;
|
|
|
|
|
goto out_drop_write;
|
|
|
|
|
}
|
2010-12-20 15:30:25 +07:00
|
|
|
|
2012-06-29 16:58:49 +07:00
|
|
|
if (copy_from_user(&flags, arg, sizeof(flags))) {
|
|
|
|
|
ret = -EFAULT;
|
|
|
|
|
goto out_drop_write;
|
|
|
|
|
}
|
2010-12-20 15:30:25 +07:00
|
|
|
|
2012-06-29 16:58:49 +07:00
|
|
|
if (flags & BTRFS_SUBVOL_CREATE_ASYNC) {
|
|
|
|
|
ret = -EINVAL;
|
|
|
|
|
goto out_drop_write;
|
|
|
|
|
}
|
2010-12-20 15:30:25 +07:00
|
|
|
|
2012-06-29 16:58:49 +07:00
|
|
|
if (flags & ~BTRFS_SUBVOL_RDONLY) {
|
|
|
|
|
ret = -EOPNOTSUPP;
|
|
|
|
|
goto out_drop_write;
|
|
|
|
|
}
|
2010-12-20 15:30:25 +07:00
|
|
|
|
2016-06-23 05:54:23 +07:00
|
|
|
down_write(&fs_info->subvol_sem);
|
2010-12-20 15:30:25 +07:00
|
|
|
|
|
|
|
|
/* nothing to do */
|
|
|
|
|
if (!!(flags & BTRFS_SUBVOL_RDONLY) == btrfs_root_readonly(root))
|
2012-06-29 16:58:49 +07:00
|
|
|
goto out_drop_sem;
|
2010-12-20 15:30:25 +07:00
|
|
|
|
|
|
|
|
root_flags = btrfs_root_flags(&root->root_item);
|
2013-12-16 23:34:17 +07:00
|
|
|
if (flags & BTRFS_SUBVOL_RDONLY) {
|
2010-12-20 15:30:25 +07:00
|
|
|
btrfs_set_root_flags(&root->root_item,
|
|
|
|
|
root_flags | BTRFS_ROOT_SUBVOL_RDONLY);
|
2013-12-16 23:34:17 +07:00
|
|
|
} else {
|
|
|
|
|
/*
|
|
|
|
|
* Block RO -> RW transition if this subvolume is involved in
|
|
|
|
|
* send
|
|
|
|
|
*/
|
|
|
|
|
spin_lock(&root->root_item_lock);
|
|
|
|
|
if (root->send_in_progress == 0) {
|
|
|
|
|
btrfs_set_root_flags(&root->root_item,
|
2010-12-20 15:30:25 +07:00
|
|
|
root_flags & ~BTRFS_ROOT_SUBVOL_RDONLY);
|
2013-12-16 23:34:17 +07:00
|
|
|
spin_unlock(&root->root_item_lock);
|
|
|
|
|
} else {
|
|
|
|
|
spin_unlock(&root->root_item_lock);
|
2016-06-23 05:54:23 +07:00
|
|
|
btrfs_warn(fs_info,
|
|
|
|
|
"Attempt to set subvolume %llu read-write during send",
|
|
|
|
|
root->root_key.objectid);
|
2013-12-16 23:34:17 +07:00
|
|
|
ret = -EPERM;
|
|
|
|
|
goto out_drop_sem;
|
|
|
|
|
}
|
|
|
|
|
}
|
2010-12-20 15:30:25 +07:00
|
|
|
|
|
|
|
|
trans = btrfs_start_transaction(root, 1);
|
|
|
|
|
if (IS_ERR(trans)) {
|
|
|
|
|
ret = PTR_ERR(trans);
|
|
|
|
|
goto out_reset;
|
|
|
|
|
}
|
|
|
|
|
|
2016-06-23 05:54:23 +07:00
|
|
|
ret = btrfs_update_root(trans, fs_info->tree_root,
|
2010-12-20 15:30:25 +07:00
|
|
|
&root->root_key, &root->root_item);
|
2017-09-28 14:53:17 +07:00
|
|
|
if (ret < 0) {
|
|
|
|
|
btrfs_end_transaction(trans);
|
|
|
|
|
goto out_reset;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
ret = btrfs_commit_transaction(trans);
|
2010-12-20 15:30:25 +07:00
|
|
|
|
|
|
|
|
out_reset:
|
|
|
|
|
if (ret)
|
|
|
|
|
btrfs_set_root_flags(&root->root_item, root_flags);
|
2012-06-29 16:58:49 +07:00
|
|
|
out_drop_sem:
|
2016-06-23 05:54:23 +07:00
|
|
|
up_write(&fs_info->subvol_sem);
|
2012-06-29 16:58:49 +07:00
|
|
|
out_drop_write:
|
|
|
|
|
mnt_drop_write_file(file);
|
|
|
|
|
out:
|
2010-12-20 15:30:25 +07:00
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
2010-03-01 03:39:26 +07:00
|
|
|
static noinline int key_in_sk(struct btrfs_key *key,
|
|
|
|
|
struct btrfs_ioctl_search_key *sk)
|
|
|
|
|
{
|
2010-03-18 23:10:08 +07:00
|
|
|
struct btrfs_key test;
|
|
|
|
|
int ret;
|
|
|
|
|
|
|
|
|
|
test.objectid = sk->min_objectid;
|
|
|
|
|
test.type = sk->min_type;
|
|
|
|
|
test.offset = sk->min_offset;
|
|
|
|
|
|
|
|
|
|
ret = btrfs_comp_cpu_keys(key, &test);
|
|
|
|
|
if (ret < 0)
|
2010-03-01 03:39:26 +07:00
|
|
|
return 0;
|
2010-03-18 23:10:08 +07:00
|
|
|
|
|
|
|
|
test.objectid = sk->max_objectid;
|
|
|
|
|
test.type = sk->max_type;
|
|
|
|
|
test.offset = sk->max_offset;
|
|
|
|
|
|
|
|
|
|
ret = btrfs_comp_cpu_keys(key, &test);
|
|
|
|
|
if (ret > 0)
|
2010-03-01 03:39:26 +07:00
|
|
|
return 0;
|
|
|
|
|
return 1;
|
|
|
|
|
}
|
|
|
|
|
|
2016-06-22 07:18:21 +07:00
|
|
|
static noinline int copy_to_sk(struct btrfs_path *path,
|
2010-03-01 03:39:26 +07:00
|
|
|
struct btrfs_key *key,
|
|
|
|
|
struct btrfs_ioctl_search_key *sk,
|
2014-01-30 22:24:00 +07:00
|
|
|
size_t *buf_size,
|
2014-01-30 22:24:02 +07:00
|
|
|
char __user *ubuf,
|
2010-03-01 03:39:26 +07:00
|
|
|
unsigned long *sk_offset,
|
|
|
|
|
int *num_found)
|
|
|
|
|
{
|
|
|
|
|
u64 found_transid;
|
|
|
|
|
struct extent_buffer *leaf;
|
|
|
|
|
struct btrfs_ioctl_search_header sh;
|
2015-06-30 09:25:43 +07:00
|
|
|
struct btrfs_key test;
|
2010-03-01 03:39:26 +07:00
|
|
|
unsigned long item_off;
|
|
|
|
|
unsigned long item_len;
|
|
|
|
|
int nritems;
|
|
|
|
|
int i;
|
|
|
|
|
int slot;
|
|
|
|
|
int ret = 0;
|
|
|
|
|
|
|
|
|
|
leaf = path->nodes[0];
|
|
|
|
|
slot = path->slots[0];
|
|
|
|
|
nritems = btrfs_header_nritems(leaf);
|
|
|
|
|
|
|
|
|
|
if (btrfs_header_generation(leaf) > sk->max_transid) {
|
|
|
|
|
i = nritems;
|
|
|
|
|
goto advance_key;
|
|
|
|
|
}
|
|
|
|
|
found_transid = btrfs_header_generation(leaf);
|
|
|
|
|
|
|
|
|
|
for (i = slot; i < nritems; i++) {
|
|
|
|
|
item_off = btrfs_item_ptr_offset(leaf, i);
|
|
|
|
|
item_len = btrfs_item_size_nr(leaf, i);
|
|
|
|
|
|
2013-05-07 00:40:18 +07:00
|
|
|
btrfs_item_key_to_cpu(leaf, key, i);
|
|
|
|
|
if (!key_in_sk(key, sk))
|
|
|
|
|
continue;
|
|
|
|
|
|
2014-01-30 22:24:00 +07:00
|
|
|
if (sizeof(sh) + item_len > *buf_size) {
|
2014-01-30 22:23:59 +07:00
|
|
|
if (*num_found) {
|
|
|
|
|
ret = 1;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* return one empty item back for v1, which does not
|
|
|
|
|
* handle -EOVERFLOW
|
|
|
|
|
*/
|
|
|
|
|
|
2014-01-30 22:24:00 +07:00
|
|
|
*buf_size = sizeof(sh) + item_len;
|
2010-03-01 03:39:26 +07:00
|
|
|
item_len = 0;
|
2014-01-30 22:23:59 +07:00
|
|
|
ret = -EOVERFLOW;
|
|
|
|
|
}
|
2010-03-01 03:39:26 +07:00
|
|
|
|
2014-01-30 22:24:00 +07:00
|
|
|
if (sizeof(sh) + item_len + *sk_offset > *buf_size) {
|
2010-03-01 03:39:26 +07:00
|
|
|
ret = 1;
|
2014-01-30 22:23:57 +07:00
|
|
|
goto out;
|
2010-03-01 03:39:26 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
sh.objectid = key->objectid;
|
|
|
|
|
sh.offset = key->offset;
|
|
|
|
|
sh.type = key->type;
|
|
|
|
|
sh.len = item_len;
|
|
|
|
|
sh.transid = found_transid;
|
|
|
|
|
|
|
|
|
|
/* copy search result header */
|
2014-01-30 22:24:02 +07:00
|
|
|
if (copy_to_user(ubuf + *sk_offset, &sh, sizeof(sh))) {
|
|
|
|
|
ret = -EFAULT;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
|
2010-03-01 03:39:26 +07:00
|
|
|
*sk_offset += sizeof(sh);
|
|
|
|
|
|
|
|
|
|
if (item_len) {
|
2014-01-30 22:24:02 +07:00
|
|
|
char __user *up = ubuf + *sk_offset;
|
2010-03-01 03:39:26 +07:00
|
|
|
/* copy the item */
|
2014-01-30 22:24:02 +07:00
|
|
|
if (read_extent_buffer_to_user(leaf, up,
|
|
|
|
|
item_off, item_len)) {
|
|
|
|
|
ret = -EFAULT;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
|
2010-03-01 03:39:26 +07:00
|
|
|
*sk_offset += item_len;
|
|
|
|
|
}
|
2011-05-15 00:43:41 +07:00
|
|
|
(*num_found)++;
|
2010-03-01 03:39:26 +07:00
|
|
|
|
2014-01-30 22:23:59 +07:00
|
|
|
if (ret) /* -EOVERFLOW from above */
|
|
|
|
|
goto out;
|
|
|
|
|
|
2014-01-30 22:23:57 +07:00
|
|
|
if (*num_found >= sk->nr_items) {
|
|
|
|
|
ret = 1;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
2010-03-01 03:39:26 +07:00
|
|
|
}
|
|
|
|
|
advance_key:
|
2010-03-18 23:10:08 +07:00
|
|
|
ret = 0;
|
2015-06-30 09:25:43 +07:00
|
|
|
test.objectid = sk->max_objectid;
|
|
|
|
|
test.type = sk->max_type;
|
|
|
|
|
test.offset = sk->max_offset;
|
|
|
|
|
if (btrfs_comp_cpu_keys(key, &test) >= 0)
|
|
|
|
|
ret = 1;
|
|
|
|
|
else if (key->offset < (u64)-1)
|
2010-03-01 03:39:26 +07:00
|
|
|
key->offset++;
|
2015-06-30 09:25:43 +07:00
|
|
|
else if (key->type < (u8)-1) {
|
2010-03-18 23:10:08 +07:00
|
|
|
key->offset = 0;
|
2010-03-01 03:39:26 +07:00
|
|
|
key->type++;
|
2015-06-30 09:25:43 +07:00
|
|
|
} else if (key->objectid < (u64)-1) {
|
2010-03-18 23:10:08 +07:00
|
|
|
key->offset = 0;
|
|
|
|
|
key->type = 0;
|
2010-03-01 03:39:26 +07:00
|
|
|
key->objectid++;
|
2010-03-18 23:10:08 +07:00
|
|
|
} else
|
|
|
|
|
ret = 1;
|
2014-01-30 22:23:57 +07:00
|
|
|
out:
|
2014-01-30 22:24:02 +07:00
|
|
|
/*
|
|
|
|
|
* 0: all items from this leaf copied, continue with next
|
|
|
|
|
* 1: * more items can be copied, but unused buffer is too small
|
|
|
|
|
* * all items were found
|
|
|
|
|
* Either way, it will stops the loop which iterates to the next
|
|
|
|
|
* leaf
|
|
|
|
|
* -EOVERFLOW: item was to large for buffer
|
|
|
|
|
* -EFAULT: could not copy extent buffer back to userspace
|
|
|
|
|
*/
|
2010-03-01 03:39:26 +07:00
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static noinline int search_ioctl(struct inode *inode,
|
2014-01-30 22:23:58 +07:00
|
|
|
struct btrfs_ioctl_search_key *sk,
|
2014-01-30 22:24:00 +07:00
|
|
|
size_t *buf_size,
|
2014-01-30 22:24:02 +07:00
|
|
|
char __user *ubuf)
|
2010-03-01 03:39:26 +07:00
|
|
|
{
|
2016-06-23 05:54:23 +07:00
|
|
|
struct btrfs_fs_info *info = btrfs_sb(inode->i_sb);
|
2010-03-01 03:39:26 +07:00
|
|
|
struct btrfs_root *root;
|
|
|
|
|
struct btrfs_key key;
|
|
|
|
|
struct btrfs_path *path;
|
|
|
|
|
int ret;
|
|
|
|
|
int num_found = 0;
|
|
|
|
|
unsigned long sk_offset = 0;
|
|
|
|
|
|
2014-01-30 22:24:00 +07:00
|
|
|
if (*buf_size < sizeof(struct btrfs_ioctl_search_header)) {
|
|
|
|
|
*buf_size = sizeof(struct btrfs_ioctl_search_header);
|
2014-01-30 22:23:58 +07:00
|
|
|
return -EOVERFLOW;
|
2014-01-30 22:24:00 +07:00
|
|
|
}
|
2014-01-30 22:23:58 +07:00
|
|
|
|
2010-03-01 03:39:26 +07:00
|
|
|
path = btrfs_alloc_path();
|
|
|
|
|
if (!path)
|
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
|
|
if (sk->tree_id == 0) {
|
|
|
|
|
/* search the root of the inode that was passed */
|
|
|
|
|
root = BTRFS_I(inode)->root;
|
|
|
|
|
} else {
|
|
|
|
|
key.objectid = sk->tree_id;
|
|
|
|
|
key.type = BTRFS_ROOT_ITEM_KEY;
|
|
|
|
|
key.offset = (u64)-1;
|
|
|
|
|
root = btrfs_read_fs_root_no_name(info, &key);
|
|
|
|
|
if (IS_ERR(root)) {
|
|
|
|
|
btrfs_free_path(path);
|
|
|
|
|
return -ENOENT;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
key.objectid = sk->min_objectid;
|
|
|
|
|
key.type = sk->min_type;
|
|
|
|
|
key.offset = sk->min_offset;
|
|
|
|
|
|
2013-10-31 12:03:04 +07:00
|
|
|
while (1) {
|
2013-10-01 22:13:42 +07:00
|
|
|
ret = btrfs_search_forward(root, &key, path, sk->min_transid);
|
2010-03-01 03:39:26 +07:00
|
|
|
if (ret != 0) {
|
|
|
|
|
if (ret > 0)
|
|
|
|
|
ret = 0;
|
|
|
|
|
goto err;
|
|
|
|
|
}
|
2016-06-22 07:18:21 +07:00
|
|
|
ret = copy_to_sk(path, &key, sk, buf_size, ubuf,
|
2010-03-01 03:39:26 +07:00
|
|
|
&sk_offset, &num_found);
|
2011-04-21 06:20:15 +07:00
|
|
|
btrfs_release_path(path);
|
2014-01-30 22:23:57 +07:00
|
|
|
if (ret)
|
2010-03-01 03:39:26 +07:00
|
|
|
break;
|
|
|
|
|
|
|
|
|
|
}
|
2014-01-30 22:23:59 +07:00
|
|
|
if (ret > 0)
|
|
|
|
|
ret = 0;
|
2010-03-01 03:39:26 +07:00
|
|
|
err:
|
|
|
|
|
sk->nr_items = num_found;
|
|
|
|
|
btrfs_free_path(path);
|
|
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static noinline int btrfs_ioctl_tree_search(struct file *file,
|
|
|
|
|
void __user *argp)
|
|
|
|
|
{
|
2014-01-30 22:24:02 +07:00
|
|
|
struct btrfs_ioctl_search_args __user *uargs;
|
|
|
|
|
struct btrfs_ioctl_search_key sk;
|
2014-01-30 22:24:00 +07:00
|
|
|
struct inode *inode;
|
|
|
|
|
int ret;
|
|
|
|
|
size_t buf_size;
|
2010-03-01 03:39:26 +07:00
|
|
|
|
|
|
|
|
if (!capable(CAP_SYS_ADMIN))
|
|
|
|
|
return -EPERM;
|
|
|
|
|
|
2014-01-30 22:24:02 +07:00
|
|
|
uargs = (struct btrfs_ioctl_search_args __user *)argp;
|
|
|
|
|
|
|
|
|
|
if (copy_from_user(&sk, &uargs->key, sizeof(sk)))
|
|
|
|
|
return -EFAULT;
|
2010-03-01 03:39:26 +07:00
|
|
|
|
2014-01-30 22:24:02 +07:00
|
|
|
buf_size = sizeof(uargs->buf);
|
2010-03-01 03:39:26 +07:00
|
|
|
|
2013-01-24 05:07:38 +07:00
|
|
|
inode = file_inode(file);
|
2014-01-30 22:24:02 +07:00
|
|
|
ret = search_ioctl(inode, &sk, &buf_size, uargs->buf);
|
2014-01-30 22:23:59 +07:00
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* In the origin implementation an overflow is handled by returning a
|
|
|
|
|
* search header with a len of zero, so reset ret.
|
|
|
|
|
*/
|
|
|
|
|
if (ret == -EOVERFLOW)
|
|
|
|
|
ret = 0;
|
|
|
|
|
|
2014-01-30 22:24:02 +07:00
|
|
|
if (ret == 0 && copy_to_user(&uargs->key, &sk, sizeof(sk)))
|
2010-03-01 03:39:26 +07:00
|
|
|
ret = -EFAULT;
|
|
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
2014-01-30 22:24:03 +07:00
|
|
|
static noinline int btrfs_ioctl_tree_search_v2(struct file *file,
|
|
|
|
|
void __user *argp)
|
|
|
|
|
{
|
|
|
|
|
struct btrfs_ioctl_search_args_v2 __user *uarg;
|
|
|
|
|
struct btrfs_ioctl_search_args_v2 args;
|
|
|
|
|
struct inode *inode;
|
|
|
|
|
int ret;
|
|
|
|
|
size_t buf_size;
|
2015-12-14 23:42:10 +07:00
|
|
|
const size_t buf_limit = SZ_16M;
|
2014-01-30 22:24:03 +07:00
|
|
|
|
|
|
|
|
if (!capable(CAP_SYS_ADMIN))
|
|
|
|
|
return -EPERM;
|
|
|
|
|
|
|
|
|
|
/* copy search header and buffer size */
|
|
|
|
|
uarg = (struct btrfs_ioctl_search_args_v2 __user *)argp;
|
|
|
|
|
if (copy_from_user(&args, uarg, sizeof(args)))
|
|
|
|
|
return -EFAULT;
|
|
|
|
|
|
|
|
|
|
buf_size = args.buf_size;
|
|
|
|
|
|
|
|
|
|
/* limit result size to 16MB */
|
|
|
|
|
if (buf_size > buf_limit)
|
|
|
|
|
buf_size = buf_limit;
|
|
|
|
|
|
|
|
|
|
inode = file_inode(file);
|
|
|
|
|
ret = search_ioctl(inode, &args.key, &buf_size,
|
2017-08-23 13:46:05 +07:00
|
|
|
(char __user *)(&uarg->buf[0]));
|
2014-01-30 22:24:03 +07:00
|
|
|
if (ret == 0 && copy_to_user(&uarg->key, &args.key, sizeof(args.key)))
|
|
|
|
|
ret = -EFAULT;
|
|
|
|
|
else if (ret == -EOVERFLOW &&
|
|
|
|
|
copy_to_user(&uarg->buf_size, &buf_size, sizeof(buf_size)))
|
|
|
|
|
ret = -EFAULT;
|
|
|
|
|
|
2010-03-01 03:39:26 +07:00
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
2009-11-18 12:42:14 +07:00
|
|
|
/*
|
2010-03-01 03:39:26 +07:00
|
|
|
* Search INODE_REFs to identify path name of 'dirid' directory
|
|
|
|
|
* in a 'tree_id' tree. and sets path name to 'name'.
|
|
|
|
|
*/
|
2009-11-18 12:42:14 +07:00
|
|
|
static noinline int btrfs_search_path_in_tree(struct btrfs_fs_info *info,
|
|
|
|
|
u64 tree_id, u64 dirid, char *name)
|
|
|
|
|
{
|
|
|
|
|
struct btrfs_root *root;
|
|
|
|
|
struct btrfs_key key;
|
2010-03-01 03:39:26 +07:00
|
|
|
char *ptr;
|
2009-11-18 12:42:14 +07:00
|
|
|
int ret = -1;
|
|
|
|
|
int slot;
|
|
|
|
|
int len;
|
|
|
|
|
int total_len = 0;
|
|
|
|
|
struct btrfs_inode_ref *iref;
|
|
|
|
|
struct extent_buffer *l;
|
|
|
|
|
struct btrfs_path *path;
|
|
|
|
|
|
|
|
|
|
if (dirid == BTRFS_FIRST_FREE_OBJECTID) {
|
|
|
|
|
name[0]='\0';
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
path = btrfs_alloc_path();
|
|
|
|
|
if (!path)
|
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
2017-12-01 16:19:42 +07:00
|
|
|
ptr = &name[BTRFS_INO_LOOKUP_PATH_MAX - 1];
|
2009-11-18 12:42:14 +07:00
|
|
|
|
|
|
|
|
key.objectid = tree_id;
|
|
|
|
|
key.type = BTRFS_ROOT_ITEM_KEY;
|
|
|
|
|
key.offset = (u64)-1;
|
|
|
|
|
root = btrfs_read_fs_root_no_name(info, &key);
|
|
|
|
|
if (IS_ERR(root)) {
|
2015-10-08 16:37:06 +07:00
|
|
|
btrfs_err(info, "could not find root %llu", tree_id);
|
2010-03-18 23:23:10 +07:00
|
|
|
ret = -ENOENT;
|
|
|
|
|
goto out;
|
2009-11-18 12:42:14 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
key.objectid = dirid;
|
|
|
|
|
key.type = BTRFS_INODE_REF_KEY;
|
2010-03-18 23:23:10 +07:00
|
|
|
key.offset = (u64)-1;
|
2009-11-18 12:42:14 +07:00
|
|
|
|
2013-10-31 12:03:04 +07:00
|
|
|
while (1) {
|
2009-11-18 12:42:14 +07:00
|
|
|
ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
|
|
|
|
|
if (ret < 0)
|
|
|
|
|
goto out;
|
2013-08-14 09:00:21 +07:00
|
|
|
else if (ret > 0) {
|
|
|
|
|
ret = btrfs_previous_item(root, path, dirid,
|
|
|
|
|
BTRFS_INODE_REF_KEY);
|
|
|
|
|
if (ret < 0)
|
|
|
|
|
goto out;
|
|
|
|
|
else if (ret > 0) {
|
|
|
|
|
ret = -ENOENT;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
}
|
2009-11-18 12:42:14 +07:00
|
|
|
|
|
|
|
|
l = path->nodes[0];
|
|
|
|
|
slot = path->slots[0];
|
|
|
|
|
btrfs_item_key_to_cpu(l, &key, slot);
|
|
|
|
|
|
|
|
|
|
iref = btrfs_item_ptr(l, slot, struct btrfs_inode_ref);
|
|
|
|
|
len = btrfs_inode_ref_name_len(l, iref);
|
|
|
|
|
ptr -= len + 1;
|
|
|
|
|
total_len += len + 1;
|
2013-08-14 09:00:20 +07:00
|
|
|
if (ptr < name) {
|
|
|
|
|
ret = -ENAMETOOLONG;
|
2009-11-18 12:42:14 +07:00
|
|
|
goto out;
|
2013-08-14 09:00:20 +07:00
|
|
|
}
|
2009-11-18 12:42:14 +07:00
|
|
|
|
|
|
|
|
*(ptr + len) = '/';
|
2013-10-31 12:03:04 +07:00
|
|
|
read_extent_buffer(l, ptr, (unsigned long)(iref + 1), len);
|
2009-11-18 12:42:14 +07:00
|
|
|
|
|
|
|
|
if (key.offset == BTRFS_FIRST_FREE_OBJECTID)
|
|
|
|
|
break;
|
|
|
|
|
|
2011-04-21 06:20:15 +07:00
|
|
|
btrfs_release_path(path);
|
2009-11-18 12:42:14 +07:00
|
|
|
key.objectid = key.offset;
|
2010-03-18 23:23:10 +07:00
|
|
|
key.offset = (u64)-1;
|
2009-11-18 12:42:14 +07:00
|
|
|
dirid = key.objectid;
|
|
|
|
|
}
|
2011-07-14 10:16:00 +07:00
|
|
|
memmove(name, ptr, total_len);
|
2013-10-31 12:03:04 +07:00
|
|
|
name[total_len] = '\0';
|
2009-11-18 12:42:14 +07:00
|
|
|
ret = 0;
|
|
|
|
|
out:
|
|
|
|
|
btrfs_free_path(path);
|
2010-03-01 03:39:26 +07:00
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static noinline int btrfs_ioctl_ino_lookup(struct file *file,
|
|
|
|
|
void __user *argp)
|
|
|
|
|
{
|
|
|
|
|
struct btrfs_ioctl_ino_lookup_args *args;
|
|
|
|
|
struct inode *inode;
|
2015-05-13 00:14:49 +07:00
|
|
|
int ret = 0;
|
2010-03-01 03:39:26 +07:00
|
|
|
|
2010-10-30 02:14:18 +07:00
|
|
|
args = memdup_user(argp, sizeof(*args));
|
|
|
|
|
if (IS_ERR(args))
|
|
|
|
|
return PTR_ERR(args);
|
2010-03-20 18:24:15 +07:00
|
|
|
|
2013-01-24 05:07:38 +07:00
|
|
|
inode = file_inode(file);
|
2010-03-01 03:39:26 +07:00
|
|
|
|
2015-05-13 00:14:49 +07:00
|
|
|
/*
|
|
|
|
|
* Unprivileged query to obtain the containing subvolume root id. The
|
|
|
|
|
* path is reset so it's consistent with btrfs_search_path_in_tree.
|
|
|
|
|
*/
|
2010-03-18 23:17:05 +07:00
|
|
|
if (args->treeid == 0)
|
|
|
|
|
args->treeid = BTRFS_I(inode)->root->root_key.objectid;
|
|
|
|
|
|
2015-05-13 00:14:49 +07:00
|
|
|
if (args->objectid == BTRFS_FIRST_FREE_OBJECTID) {
|
|
|
|
|
args->name[0] = 0;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (!capable(CAP_SYS_ADMIN)) {
|
|
|
|
|
ret = -EPERM;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
|
2010-03-01 03:39:26 +07:00
|
|
|
ret = btrfs_search_path_in_tree(BTRFS_I(inode)->root->fs_info,
|
|
|
|
|
args->treeid, args->objectid,
|
|
|
|
|
args->name);
|
|
|
|
|
|
2015-05-13 00:14:49 +07:00
|
|
|
out:
|
2010-03-01 03:39:26 +07:00
|
|
|
if (ret == 0 && copy_to_user(argp, args, sizeof(*args)))
|
|
|
|
|
ret = -EFAULT;
|
|
|
|
|
|
|
|
|
|
kfree(args);
|
2009-11-18 12:42:14 +07:00
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
2009-09-22 03:00:26 +07:00
|
|
|
static noinline int btrfs_ioctl_snap_destroy(struct file *file,
|
|
|
|
|
void __user *arg)
|
|
|
|
|
{
|
2013-09-02 02:57:51 +07:00
|
|
|
struct dentry *parent = file->f_path.dentry;
|
2016-06-23 05:54:23 +07:00
|
|
|
struct btrfs_fs_info *fs_info = btrfs_sb(parent->d_sb);
|
2009-09-22 03:00:26 +07:00
|
|
|
struct dentry *dentry;
|
2015-03-18 05:25:59 +07:00
|
|
|
struct inode *dir = d_inode(parent);
|
2009-09-22 03:00:26 +07:00
|
|
|
struct inode *inode;
|
|
|
|
|
struct btrfs_root *root = BTRFS_I(dir)->root;
|
|
|
|
|
struct btrfs_root *dest = NULL;
|
|
|
|
|
struct btrfs_ioctl_vol_args *vol_args;
|
|
|
|
|
int namelen;
|
|
|
|
|
int err = 0;
|
|
|
|
|
|
2016-09-21 19:31:29 +07:00
|
|
|
if (!S_ISDIR(dir->i_mode))
|
|
|
|
|
return -ENOTDIR;
|
|
|
|
|
|
2009-09-22 03:00:26 +07:00
|
|
|
vol_args = memdup_user(arg, sizeof(*vol_args));
|
|
|
|
|
if (IS_ERR(vol_args))
|
|
|
|
|
return PTR_ERR(vol_args);
|
|
|
|
|
|
|
|
|
|
vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
|
|
|
|
|
namelen = strlen(vol_args->name);
|
|
|
|
|
if (strchr(vol_args->name, '/') ||
|
|
|
|
|
strncmp(vol_args->name, "..", namelen) == 0) {
|
|
|
|
|
err = -EINVAL;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
|
2011-11-23 23:57:51 +07:00
|
|
|
err = mnt_want_write_file(file);
|
2009-09-22 03:00:26 +07:00
|
|
|
if (err)
|
|
|
|
|
goto out;
|
|
|
|
|
|
2014-04-15 21:41:44 +07:00
|
|
|
|
2016-05-26 11:05:12 +07:00
|
|
|
err = down_write_killable_nested(&dir->i_rwsem, I_MUTEX_PARENT);
|
|
|
|
|
if (err == -EINTR)
|
|
|
|
|
goto out_drop_write;
|
2009-09-22 03:00:26 +07:00
|
|
|
dentry = lookup_one_len(vol_args->name, parent, namelen);
|
|
|
|
|
if (IS_ERR(dentry)) {
|
|
|
|
|
err = PTR_ERR(dentry);
|
|
|
|
|
goto out_unlock_dir;
|
|
|
|
|
}
|
|
|
|
|
|
2015-03-18 05:25:59 +07:00
|
|
|
if (d_really_is_negative(dentry)) {
|
2009-09-22 03:00:26 +07:00
|
|
|
err = -ENOENT;
|
|
|
|
|
goto out_dput;
|
|
|
|
|
}
|
|
|
|
|
|
2015-03-18 05:25:59 +07:00
|
|
|
inode = d_inode(dentry);
|
2010-10-30 02:46:43 +07:00
|
|
|
dest = BTRFS_I(inode)->root;
|
2013-10-31 12:03:04 +07:00
|
|
|
if (!capable(CAP_SYS_ADMIN)) {
|
2010-10-30 02:46:43 +07:00
|
|
|
/*
|
|
|
|
|
* Regular user. Only allow this with a special mount
|
|
|
|
|
* option, when the user has write+exec access to the
|
|
|
|
|
* subvol root, and when rmdir(2) would have been
|
|
|
|
|
* allowed.
|
|
|
|
|
*
|
|
|
|
|
* Note that this is _not_ check that the subvol is
|
|
|
|
|
* empty or doesn't contain data that we wouldn't
|
|
|
|
|
* otherwise be able to delete.
|
|
|
|
|
*
|
|
|
|
|
* Users who want to delete empty subvols should try
|
|
|
|
|
* rmdir(2).
|
|
|
|
|
*/
|
|
|
|
|
err = -EPERM;
|
2016-06-23 05:54:23 +07:00
|
|
|
if (!btrfs_test_opt(fs_info, USER_SUBVOL_RM_ALLOWED))
|
2010-10-30 02:46:43 +07:00
|
|
|
goto out_dput;
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Do not allow deletion if the parent dir is the same
|
|
|
|
|
* as the dir to be deleted. That means the ioctl
|
|
|
|
|
* must be called on the dentry referencing the root
|
|
|
|
|
* of the subvol, not a random directory contained
|
|
|
|
|
* within it.
|
|
|
|
|
*/
|
|
|
|
|
err = -EINVAL;
|
|
|
|
|
if (root == dest)
|
|
|
|
|
goto out_dput;
|
|
|
|
|
|
|
|
|
|
err = inode_permission(inode, MAY_WRITE | MAY_EXEC);
|
|
|
|
|
if (err)
|
|
|
|
|
goto out_dput;
|
|
|
|
|
}
|
|
|
|
|
|
2012-10-22 18:39:53 +07:00
|
|
|
/* check if subvolume may be deleted by a user */
|
|
|
|
|
err = btrfs_may_delete(dir, dentry, 1);
|
|
|
|
|
if (err)
|
|
|
|
|
goto out_dput;
|
|
|
|
|
|
2017-01-11 01:35:31 +07:00
|
|
|
if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FIRST_FREE_OBJECTID) {
|
2009-09-22 03:00:26 +07:00
|
|
|
err = -EINVAL;
|
|
|
|
|
goto out_dput;
|
|
|
|
|
}
|
|
|
|
|
|
2016-01-23 03:40:57 +07:00
|
|
|
inode_lock(inode);
|
2018-04-18 09:34:52 +07:00
|
|
|
err = btrfs_delete_subvolume(dir, dentry);
|
2016-01-23 03:40:57 +07:00
|
|
|
inode_unlock(inode);
|
2018-04-18 09:34:52 +07:00
|
|
|
if (!err)
|
2009-09-22 03:00:26 +07:00
|
|
|
d_delete(dentry);
|
Btrfs: fix cleaner thread not working with inode cache option
Right now inode cache inode is treated as the same as space cache
inode, ie. keep inode in memory till putting super.
But this leads to an awkward situation.
If we're going to delete a snapshot/subvolume, btrfs will not
actually delete it and return free space, but will add it to dead
roots list until the last inode on this snap/subvol being destroyed.
Then we'll fetch deleted roots and cleanup them via cleaner thread.
So here is the problem, if we enable inode cache option, each
snap/subvol has a cached inode which is used to store inode allcation
information. And this cache inode will be kept in memory, as the above
said. So with inode cache, snap/subvol can only be added into
dead roots list during freeing roots stage in umount, so that we can
ONLY get space back after another remount(we cleanup dead roots on mount).
But the real thing is we'll no more use the snap/subvol if we mark it
deleted, so we can safely iput its cache inode when we delete snap/subvol.
Another thing is that we need to change the rules of droping inode, we
don't keep snap/subvol's cache inode in memory till end so that we can
add snap/subvol into dead roots list in time.
Reported-by: Mitch Harder <mitch.harder@sabayonlinux.org>
Signed-off-by: Liu Bo <bo.li.liu@oracle.com>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-02-20 21:10:23 +07:00
|
|
|
|
2009-09-22 03:00:26 +07:00
|
|
|
out_dput:
|
|
|
|
|
dput(dentry);
|
|
|
|
|
out_unlock_dir:
|
2016-01-23 03:40:57 +07:00
|
|
|
inode_unlock(dir);
|
2016-05-26 11:05:12 +07:00
|
|
|
out_drop_write:
|
2011-12-09 20:06:57 +07:00
|
|
|
mnt_drop_write_file(file);
|
2009-09-22 03:00:26 +07:00
|
|
|
out:
|
|
|
|
|
kfree(vol_args);
|
|
|
|
|
return err;
|
|
|
|
|
}
|
|
|
|
|
|
2010-03-11 21:42:04 +07:00
|
|
|
static int btrfs_ioctl_defrag(struct file *file, void __user *argp)
|
2008-06-12 08:53:53 +07:00
|
|
|
{
|
2013-01-24 05:07:38 +07:00
|
|
|
struct inode *inode = file_inode(file);
|
2008-06-12 08:53:53 +07:00
|
|
|
struct btrfs_root *root = BTRFS_I(inode)->root;
|
2010-03-11 21:42:04 +07:00
|
|
|
struct btrfs_ioctl_defrag_range_args *range;
|
2008-11-13 02:34:12 +07:00
|
|
|
int ret;
|
|
|
|
|
|
2013-01-20 20:57:57 +07:00
|
|
|
ret = mnt_want_write_file(file);
|
|
|
|
|
if (ret)
|
|
|
|
|
return ret;
|
2010-12-20 15:04:08 +07:00
|
|
|
|
2013-01-20 20:57:57 +07:00
|
|
|
if (btrfs_root_readonly(root)) {
|
|
|
|
|
ret = -EROFS;
|
|
|
|
|
goto out;
|
2012-11-05 23:54:08 +07:00
|
|
|
}
|
2008-06-12 08:53:53 +07:00
|
|
|
|
|
|
|
|
switch (inode->i_mode & S_IFMT) {
|
|
|
|
|
case S_IFDIR:
|
2009-01-06 04:57:23 +07:00
|
|
|
if (!capable(CAP_SYS_ADMIN)) {
|
|
|
|
|
ret = -EPERM;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
2013-02-01 01:21:12 +07:00
|
|
|
ret = btrfs_defrag_root(root);
|
2008-06-12 08:53:53 +07:00
|
|
|
break;
|
|
|
|
|
case S_IFREG:
|
2009-01-06 04:57:23 +07:00
|
|
|
if (!(file->f_mode & FMODE_WRITE)) {
|
|
|
|
|
ret = -EINVAL;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
2010-03-11 21:42:04 +07:00
|
|
|
|
|
|
|
|
range = kzalloc(sizeof(*range), GFP_KERNEL);
|
|
|
|
|
if (!range) {
|
|
|
|
|
ret = -ENOMEM;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (argp) {
|
|
|
|
|
if (copy_from_user(range, argp,
|
|
|
|
|
sizeof(*range))) {
|
|
|
|
|
ret = -EFAULT;
|
|
|
|
|
kfree(range);
|
2010-03-20 18:24:48 +07:00
|
|
|
goto out;
|
2010-03-11 21:42:04 +07:00
|
|
|
}
|
|
|
|
|
/* compression requires us to start the IO */
|
|
|
|
|
if ((range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)) {
|
|
|
|
|
range->flags |= BTRFS_DEFRAG_RANGE_START_IO;
|
|
|
|
|
range->extent_thresh = (u32)-1;
|
|
|
|
|
}
|
|
|
|
|
} else {
|
|
|
|
|
/* the rest are all set to zero by kzalloc */
|
|
|
|
|
range->len = (u64)-1;
|
|
|
|
|
}
|
2013-01-24 05:07:38 +07:00
|
|
|
ret = btrfs_defrag_file(file_inode(file), file,
|
2018-03-07 16:29:18 +07:00
|
|
|
range, BTRFS_OLDEST_GENERATION, 0);
|
2011-05-25 02:35:30 +07:00
|
|
|
if (ret > 0)
|
|
|
|
|
ret = 0;
|
2010-03-11 21:42:04 +07:00
|
|
|
kfree(range);
|
2008-06-12 08:53:53 +07:00
|
|
|
break;
|
2010-05-16 21:49:58 +07:00
|
|
|
default:
|
|
|
|
|
ret = -EINVAL;
|
2008-06-12 08:53:53 +07:00
|
|
|
}
|
2009-01-06 04:57:23 +07:00
|
|
|
out:
|
2013-01-20 20:57:57 +07:00
|
|
|
mnt_drop_write_file(file);
|
2009-01-06 04:57:23 +07:00
|
|
|
return ret;
|
2008-06-12 08:53:53 +07:00
|
|
|
}
|
|
|
|
|
|
2016-06-23 05:54:24 +07:00
|
|
|
static long btrfs_ioctl_add_dev(struct btrfs_fs_info *fs_info, void __user *arg)
|
2008-06-12 08:53:53 +07:00
|
|
|
{
|
|
|
|
|
struct btrfs_ioctl_vol_args *vol_args;
|
|
|
|
|
int ret;
|
|
|
|
|
|
2009-01-06 04:57:23 +07:00
|
|
|
if (!capable(CAP_SYS_ADMIN))
|
|
|
|
|
return -EPERM;
|
|
|
|
|
|
2017-03-28 19:44:21 +07:00
|
|
|
if (test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags))
|
2013-08-21 10:44:48 +07:00
|
|
|
return BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS;
|
2012-01-17 03:04:47 +07:00
|
|
|
|
2016-06-23 05:54:23 +07:00
|
|
|
mutex_lock(&fs_info->volume_mutex);
|
2009-04-08 14:06:54 +07:00
|
|
|
vol_args = memdup_user(arg, sizeof(*vol_args));
|
2012-01-17 03:04:47 +07:00
|
|
|
if (IS_ERR(vol_args)) {
|
|
|
|
|
ret = PTR_ERR(vol_args);
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
2008-06-12 08:53:53 +07:00
|
|
|
|
2008-07-24 23:20:14 +07:00
|
|
|
vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
|
2016-06-23 05:54:24 +07:00
|
|
|
ret = btrfs_init_new_device(fs_info, vol_args->name);
|
2008-06-12 08:53:53 +07:00
|
|
|
|
2014-06-30 23:58:56 +07:00
|
|
|
if (!ret)
|
2016-06-23 05:54:23 +07:00
|
|
|
btrfs_info(fs_info, "disk added %s", vol_args->name);
|
2014-06-30 23:58:56 +07:00
|
|
|
|
2008-06-12 08:53:53 +07:00
|
|
|
kfree(vol_args);
|
2012-01-17 03:04:47 +07:00
|
|
|
out:
|
2016-06-23 05:54:23 +07:00
|
|
|
mutex_unlock(&fs_info->volume_mutex);
|
2017-03-28 19:44:21 +07:00
|
|
|
clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
|
2008-06-12 08:53:53 +07:00
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
2016-02-13 09:01:39 +07:00
|
|
|
static long btrfs_ioctl_rm_dev_v2(struct file *file, void __user *arg)
|
2008-06-12 08:53:53 +07:00
|
|
|
{
|
2016-06-23 05:54:23 +07:00
|
|
|
struct inode *inode = file_inode(file);
|
|
|
|
|
struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
|
2016-02-13 09:01:39 +07:00
|
|
|
struct btrfs_ioctl_vol_args_v2 *vol_args;
|
2008-06-12 08:53:53 +07:00
|
|
|
int ret;
|
|
|
|
|
|
2009-01-06 04:57:23 +07:00
|
|
|
if (!capable(CAP_SYS_ADMIN))
|
|
|
|
|
return -EPERM;
|
|
|
|
|
|
2012-11-26 15:44:50 +07:00
|
|
|
ret = mnt_want_write_file(file);
|
|
|
|
|
if (ret)
|
|
|
|
|
return ret;
|
2008-11-13 02:34:12 +07:00
|
|
|
|
2009-04-08 14:06:54 +07:00
|
|
|
vol_args = memdup_user(arg, sizeof(*vol_args));
|
2012-01-17 03:04:47 +07:00
|
|
|
if (IS_ERR(vol_args)) {
|
|
|
|
|
ret = PTR_ERR(vol_args);
|
2014-09-04 18:09:15 +07:00
|
|
|
goto err_drop;
|
2012-01-17 03:04:47 +07:00
|
|
|
}
|
2008-06-12 08:53:53 +07:00
|
|
|
|
2016-02-13 09:01:39 +07:00
|
|
|
/* Check for compatibility reject unknown flags */
|
2016-02-16 00:15:21 +07:00
|
|
|
if (vol_args->flags & ~BTRFS_VOL_ARG_V2_FLAGS_SUPPORTED)
|
|
|
|
|
return -EOPNOTSUPP;
|
2008-06-12 08:53:53 +07:00
|
|
|
|
2017-03-28 19:44:21 +07:00
|
|
|
if (test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags)) {
|
2013-05-17 17:52:45 +07:00
|
|
|
ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
2018-03-20 23:20:45 +07:00
|
|
|
mutex_lock(&fs_info->volume_mutex);
|
2013-05-17 17:52:45 +07:00
|
|
|
|
2016-02-16 00:15:21 +07:00
|
|
|
if (vol_args->flags & BTRFS_DEVICE_SPEC_BY_ID) {
|
2016-06-23 05:54:24 +07:00
|
|
|
ret = btrfs_rm_device(fs_info, NULL, vol_args->devid);
|
2016-02-13 09:01:39 +07:00
|
|
|
} else {
|
|
|
|
|
vol_args->name[BTRFS_SUBVOL_NAME_MAX] = '\0';
|
2016-06-23 05:54:24 +07:00
|
|
|
ret = btrfs_rm_device(fs_info, vol_args->name, 0);
|
2016-02-13 09:01:39 +07:00
|
|
|
}
|
2018-03-20 23:20:45 +07:00
|
|
|
mutex_unlock(&fs_info->volume_mutex);
|
2017-03-28 19:44:21 +07:00
|
|
|
clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
|
2013-05-17 17:52:45 +07:00
|
|
|
|
2016-02-13 09:01:39 +07:00
|
|
|
if (!ret) {
|
2016-02-16 00:15:21 +07:00
|
|
|
if (vol_args->flags & BTRFS_DEVICE_SPEC_BY_ID)
|
2016-06-23 05:54:23 +07:00
|
|
|
btrfs_info(fs_info, "device deleted: id %llu",
|
2016-02-13 09:01:39 +07:00
|
|
|
vol_args->devid);
|
|
|
|
|
else
|
2016-06-23 05:54:23 +07:00
|
|
|
btrfs_info(fs_info, "device deleted: %s",
|
2016-02-13 09:01:39 +07:00
|
|
|
vol_args->name);
|
|
|
|
|
}
|
2013-05-17 17:52:45 +07:00
|
|
|
out:
|
|
|
|
|
kfree(vol_args);
|
2014-09-04 18:09:15 +07:00
|
|
|
err_drop:
|
2013-01-20 20:57:57 +07:00
|
|
|
mnt_drop_write_file(file);
|
2008-06-12 08:53:53 +07:00
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
2012-11-26 15:44:50 +07:00
|
|
|
static long btrfs_ioctl_rm_dev(struct file *file, void __user *arg)
|
2008-06-12 08:53:53 +07:00
|
|
|
{
|
2016-06-23 05:54:23 +07:00
|
|
|
struct inode *inode = file_inode(file);
|
|
|
|
|
struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
|
2008-06-12 08:53:53 +07:00
|
|
|
struct btrfs_ioctl_vol_args *vol_args;
|
|
|
|
|
int ret;
|
|
|
|
|
|
2009-01-06 04:57:23 +07:00
|
|
|
if (!capable(CAP_SYS_ADMIN))
|
|
|
|
|
return -EPERM;
|
|
|
|
|
|
2012-11-26 15:44:50 +07:00
|
|
|
ret = mnt_want_write_file(file);
|
|
|
|
|
if (ret)
|
|
|
|
|
return ret;
|
2008-11-13 02:34:12 +07:00
|
|
|
|
2017-03-28 19:44:21 +07:00
|
|
|
if (test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags)) {
|
2013-05-17 17:52:45 +07:00
|
|
|
ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS;
|
2016-05-04 19:10:47 +07:00
|
|
|
goto out_drop_write;
|
|
|
|
|
}
|
2018-03-20 23:20:45 +07:00
|
|
|
mutex_lock(&fs_info->volume_mutex);
|
2016-05-04 19:10:47 +07:00
|
|
|
|
|
|
|
|
vol_args = memdup_user(arg, sizeof(*vol_args));
|
|
|
|
|
if (IS_ERR(vol_args)) {
|
|
|
|
|
ret = PTR_ERR(vol_args);
|
2013-05-17 17:52:45 +07:00
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
|
2016-05-04 19:10:47 +07:00
|
|
|
vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
|
2016-06-23 05:54:24 +07:00
|
|
|
ret = btrfs_rm_device(fs_info, vol_args->name, 0);
|
2013-05-17 17:52:45 +07:00
|
|
|
|
2014-06-30 23:58:57 +07:00
|
|
|
if (!ret)
|
2016-06-23 05:54:23 +07:00
|
|
|
btrfs_info(fs_info, "disk deleted %s", vol_args->name);
|
2013-05-17 17:52:45 +07:00
|
|
|
kfree(vol_args);
|
2016-05-04 19:10:47 +07:00
|
|
|
out:
|
2018-03-20 23:20:45 +07:00
|
|
|
mutex_unlock(&fs_info->volume_mutex);
|
2017-03-28 19:44:21 +07:00
|
|
|
clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
|
2016-05-04 19:10:47 +07:00
|
|
|
out_drop_write:
|
2013-01-20 20:57:57 +07:00
|
|
|
mnt_drop_write_file(file);
|
2016-05-04 19:10:47 +07:00
|
|
|
|
2008-06-12 08:53:53 +07:00
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
2016-06-23 05:54:24 +07:00
|
|
|
static long btrfs_ioctl_fs_info(struct btrfs_fs_info *fs_info,
|
|
|
|
|
void __user *arg)
|
2011-03-11 21:41:01 +07:00
|
|
|
{
|
2011-06-08 15:27:56 +07:00
|
|
|
struct btrfs_ioctl_fs_info_args *fi_args;
|
2011-03-11 21:41:01 +07:00
|
|
|
struct btrfs_device *device;
|
2016-06-23 05:54:23 +07:00
|
|
|
struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
|
2011-06-08 15:27:56 +07:00
|
|
|
int ret = 0;
|
2011-03-11 21:41:01 +07:00
|
|
|
|
2011-06-08 15:27:56 +07:00
|
|
|
fi_args = kzalloc(sizeof(*fi_args), GFP_KERNEL);
|
|
|
|
|
if (!fi_args)
|
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
2017-06-16 05:09:21 +07:00
|
|
|
rcu_read_lock();
|
2011-06-08 15:27:56 +07:00
|
|
|
fi_args->num_devices = fs_devices->num_devices;
|
2011-03-11 21:41:01 +07:00
|
|
|
|
2017-06-16 05:09:21 +07:00
|
|
|
list_for_each_entry_rcu(device, &fs_devices->devices, dev_list) {
|
2011-06-08 15:27:56 +07:00
|
|
|
if (device->devid > fi_args->max_id)
|
|
|
|
|
fi_args->max_id = device->devid;
|
2011-03-11 21:41:01 +07:00
|
|
|
}
|
2017-06-16 05:09:21 +07:00
|
|
|
rcu_read_unlock();
|
2011-03-11 21:41:01 +07:00
|
|
|
|
2017-06-16 05:09:21 +07:00
|
|
|
memcpy(&fi_args->fsid, fs_info->fsid, sizeof(fi_args->fsid));
|
2017-08-23 13:46:00 +07:00
|
|
|
fi_args->nodesize = fs_info->nodesize;
|
|
|
|
|
fi_args->sectorsize = fs_info->sectorsize;
|
|
|
|
|
fi_args->clone_alignment = fs_info->sectorsize;
|
2014-05-07 23:17:06 +07:00
|
|
|
|
2011-06-08 15:27:56 +07:00
|
|
|
if (copy_to_user(arg, fi_args, sizeof(*fi_args)))
|
|
|
|
|
ret = -EFAULT;
|
2011-03-11 21:41:01 +07:00
|
|
|
|
2011-06-08 15:27:56 +07:00
|
|
|
kfree(fi_args);
|
|
|
|
|
return ret;
|
2011-03-11 21:41:01 +07:00
|
|
|
}
|
|
|
|
|
|
2016-06-23 05:54:24 +07:00
|
|
|
static long btrfs_ioctl_dev_info(struct btrfs_fs_info *fs_info,
|
|
|
|
|
void __user *arg)
|
2011-03-11 21:41:01 +07:00
|
|
|
{
|
|
|
|
|
struct btrfs_ioctl_dev_info_args *di_args;
|
|
|
|
|
struct btrfs_device *dev;
|
|
|
|
|
int ret = 0;
|
|
|
|
|
char *s_uuid = NULL;
|
|
|
|
|
|
|
|
|
|
di_args = memdup_user(arg, sizeof(*di_args));
|
|
|
|
|
if (IS_ERR(di_args))
|
|
|
|
|
return PTR_ERR(di_args);
|
|
|
|
|
|
2013-08-15 22:11:20 +07:00
|
|
|
if (!btrfs_is_empty_uuid(di_args->uuid))
|
2011-03-11 21:41:01 +07:00
|
|
|
s_uuid = di_args->uuid;
|
|
|
|
|
|
2017-06-16 05:09:21 +07:00
|
|
|
rcu_read_lock();
|
2016-06-23 05:54:23 +07:00
|
|
|
dev = btrfs_find_device(fs_info, di_args->devid, s_uuid, NULL);
|
2011-03-11 21:41:01 +07:00
|
|
|
|
|
|
|
|
if (!dev) {
|
|
|
|
|
ret = -ENODEV;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
di_args->devid = dev->devid;
|
2014-09-03 20:35:38 +07:00
|
|
|
di_args->bytes_used = btrfs_device_get_bytes_used(dev);
|
|
|
|
|
di_args->total_bytes = btrfs_device_get_total_bytes(dev);
|
2011-03-11 21:41:01 +07:00
|
|
|
memcpy(di_args->uuid, dev->uuid, sizeof(di_args->uuid));
|
2012-04-26 23:36:56 +07:00
|
|
|
if (dev->name) {
|
2012-06-05 01:03:51 +07:00
|
|
|
struct rcu_string *name;
|
|
|
|
|
|
|
|
|
|
name = rcu_dereference(dev->name);
|
2018-01-08 18:51:22 +07:00
|
|
|
strncpy(di_args->path, name->str, sizeof(di_args->path) - 1);
|
2012-04-26 23:36:56 +07:00
|
|
|
di_args->path[sizeof(di_args->path) - 1] = 0;
|
|
|
|
|
} else {
|
2012-03-19 22:17:22 +07:00
|
|
|
di_args->path[0] = '\0';
|
2012-04-26 23:36:56 +07:00
|
|
|
}
|
2011-03-11 21:41:01 +07:00
|
|
|
|
|
|
|
|
out:
|
2017-06-16 05:09:21 +07:00
|
|
|
rcu_read_unlock();
|
2011-03-11 21:41:01 +07:00
|
|
|
if (ret == 0 && copy_to_user(arg, di_args, sizeof(*di_args)))
|
|
|
|
|
ret = -EFAULT;
|
|
|
|
|
|
|
|
|
|
kfree(di_args);
|
|
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
2015-07-01 04:42:05 +07:00
|
|
|
static struct page *extent_same_get_page(struct inode *inode, pgoff_t index)
|
2013-08-07 01:42:51 +07:00
|
|
|
{
|
|
|
|
|
struct page *page;
|
|
|
|
|
|
|
|
|
|
page = grab_cache_page(inode->i_mapping, index);
|
|
|
|
|
if (!page)
|
2016-01-28 01:37:47 +07:00
|
|
|
return ERR_PTR(-ENOMEM);
|
2013-08-07 01:42:51 +07:00
|
|
|
|
|
|
|
|
if (!PageUptodate(page)) {
|
2016-01-28 01:37:47 +07:00
|
|
|
int ret;
|
|
|
|
|
|
|
|
|
|
ret = btrfs_readpage(NULL, page);
|
|
|
|
|
if (ret)
|
|
|
|
|
return ERR_PTR(ret);
|
2013-08-07 01:42:51 +07:00
|
|
|
lock_page(page);
|
|
|
|
|
if (!PageUptodate(page)) {
|
|
|
|
|
unlock_page(page);
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
put_page(page);
|
2016-01-28 01:37:47 +07:00
|
|
|
return ERR_PTR(-EIO);
|
|
|
|
|
}
|
|
|
|
|
if (page->mapping != inode->i_mapping) {
|
|
|
|
|
unlock_page(page);
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
put_page(page);
|
2016-01-28 01:37:47 +07:00
|
|
|
return ERR_PTR(-EAGAIN);
|
2013-08-07 01:42:51 +07:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return page;
|
|
|
|
|
}
|
|
|
|
|
|
2015-07-01 04:42:05 +07:00
|
|
|
static int gather_extent_pages(struct inode *inode, struct page **pages,
|
|
|
|
|
int num_pages, u64 off)
|
|
|
|
|
{
|
|
|
|
|
int i;
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
pgoff_t index = off >> PAGE_SHIFT;
|
2015-07-01 04:42:05 +07:00
|
|
|
|
|
|
|
|
for (i = 0; i < num_pages; i++) {
|
2016-01-28 01:37:47 +07:00
|
|
|
again:
|
2015-07-01 04:42:05 +07:00
|
|
|
pages[i] = extent_same_get_page(inode, index + i);
|
2016-01-28 01:37:47 +07:00
|
|
|
if (IS_ERR(pages[i])) {
|
|
|
|
|
int err = PTR_ERR(pages[i]);
|
|
|
|
|
|
|
|
|
|
if (err == -EAGAIN)
|
|
|
|
|
goto again;
|
|
|
|
|
pages[i] = NULL;
|
|
|
|
|
return err;
|
|
|
|
|
}
|
2015-07-01 04:42:05 +07:00
|
|
|
}
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
2016-01-27 17:20:58 +07:00
|
|
|
static int lock_extent_range(struct inode *inode, u64 off, u64 len,
|
|
|
|
|
bool retry_range_locking)
|
2013-08-07 01:42:48 +07:00
|
|
|
{
|
2016-01-27 17:20:58 +07:00
|
|
|
/*
|
|
|
|
|
* Do any pending delalloc/csum calculations on inode, one way or
|
|
|
|
|
* another, and lock file content.
|
|
|
|
|
* The locking order is:
|
|
|
|
|
*
|
|
|
|
|
* 1) pages
|
|
|
|
|
* 2) range in the inode's io tree
|
|
|
|
|
*/
|
2013-08-07 01:42:48 +07:00
|
|
|
while (1) {
|
|
|
|
|
struct btrfs_ordered_extent *ordered;
|
|
|
|
|
lock_extent(&BTRFS_I(inode)->io_tree, off, off + len - 1);
|
|
|
|
|
ordered = btrfs_lookup_first_ordered_extent(inode,
|
|
|
|
|
off + len - 1);
|
2014-05-30 23:56:24 +07:00
|
|
|
if ((!ordered ||
|
|
|
|
|
ordered->file_offset + ordered->len <= off ||
|
|
|
|
|
ordered->file_offset >= off + len) &&
|
2013-08-07 01:42:48 +07:00
|
|
|
!test_range_bit(&BTRFS_I(inode)->io_tree, off,
|
2014-05-30 23:56:24 +07:00
|
|
|
off + len - 1, EXTENT_DELALLOC, 0, NULL)) {
|
|
|
|
|
if (ordered)
|
|
|
|
|
btrfs_put_ordered_extent(ordered);
|
2013-08-07 01:42:48 +07:00
|
|
|
break;
|
2014-05-30 23:56:24 +07:00
|
|
|
}
|
2013-08-07 01:42:48 +07:00
|
|
|
unlock_extent(&BTRFS_I(inode)->io_tree, off, off + len - 1);
|
|
|
|
|
if (ordered)
|
|
|
|
|
btrfs_put_ordered_extent(ordered);
|
2016-01-27 17:20:58 +07:00
|
|
|
if (!retry_range_locking)
|
|
|
|
|
return -EAGAIN;
|
2013-08-07 01:42:48 +07:00
|
|
|
btrfs_wait_ordered_range(inode, off, len);
|
|
|
|
|
}
|
2016-01-27 17:20:58 +07:00
|
|
|
return 0;
|
2013-08-07 01:42:48 +07:00
|
|
|
}
|
|
|
|
|
|
2015-07-01 04:42:05 +07:00
|
|
|
static void btrfs_double_inode_unlock(struct inode *inode1, struct inode *inode2)
|
2013-08-07 01:42:51 +07:00
|
|
|
{
|
2016-01-23 03:40:57 +07:00
|
|
|
inode_unlock(inode1);
|
|
|
|
|
inode_unlock(inode2);
|
2013-08-07 01:42:51 +07:00
|
|
|
}
|
|
|
|
|
|
2015-07-01 04:42:05 +07:00
|
|
|
static void btrfs_double_inode_lock(struct inode *inode1, struct inode *inode2)
|
|
|
|
|
{
|
|
|
|
|
if (inode1 < inode2)
|
|
|
|
|
swap(inode1, inode2);
|
|
|
|
|
|
2016-01-23 03:40:57 +07:00
|
|
|
inode_lock_nested(inode1, I_MUTEX_PARENT);
|
|
|
|
|
inode_lock_nested(inode2, I_MUTEX_CHILD);
|
2015-07-01 04:42:05 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void btrfs_double_extent_unlock(struct inode *inode1, u64 loff1,
|
|
|
|
|
struct inode *inode2, u64 loff2, u64 len)
|
|
|
|
|
{
|
|
|
|
|
unlock_extent(&BTRFS_I(inode1)->io_tree, loff1, loff1 + len - 1);
|
|
|
|
|
unlock_extent(&BTRFS_I(inode2)->io_tree, loff2, loff2 + len - 1);
|
|
|
|
|
}
|
|
|
|
|
|
2016-01-27 17:20:58 +07:00
|
|
|
static int btrfs_double_extent_lock(struct inode *inode1, u64 loff1,
|
|
|
|
|
struct inode *inode2, u64 loff2, u64 len,
|
|
|
|
|
bool retry_range_locking)
|
2013-08-07 01:42:51 +07:00
|
|
|
{
|
2016-01-27 17:20:58 +07:00
|
|
|
int ret;
|
|
|
|
|
|
2013-08-07 01:42:51 +07:00
|
|
|
if (inode1 < inode2) {
|
|
|
|
|
swap(inode1, inode2);
|
|
|
|
|
swap(loff1, loff2);
|
|
|
|
|
}
|
2016-01-27 17:20:58 +07:00
|
|
|
ret = lock_extent_range(inode1, loff1, len, retry_range_locking);
|
|
|
|
|
if (ret)
|
|
|
|
|
return ret;
|
|
|
|
|
ret = lock_extent_range(inode2, loff2, len, retry_range_locking);
|
|
|
|
|
if (ret)
|
|
|
|
|
unlock_extent(&BTRFS_I(inode1)->io_tree, loff1,
|
|
|
|
|
loff1 + len - 1);
|
|
|
|
|
return ret;
|
2015-07-01 04:42:05 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
struct cmp_pages {
|
|
|
|
|
int num_pages;
|
|
|
|
|
struct page **src_pages;
|
|
|
|
|
struct page **dst_pages;
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
static void btrfs_cmp_data_free(struct cmp_pages *cmp)
|
|
|
|
|
{
|
|
|
|
|
int i;
|
|
|
|
|
struct page *pg;
|
|
|
|
|
|
|
|
|
|
for (i = 0; i < cmp->num_pages; i++) {
|
|
|
|
|
pg = cmp->src_pages[i];
|
2016-01-27 17:20:58 +07:00
|
|
|
if (pg) {
|
|
|
|
|
unlock_page(pg);
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
put_page(pg);
|
2016-01-27 17:20:58 +07:00
|
|
|
}
|
2015-07-01 04:42:05 +07:00
|
|
|
pg = cmp->dst_pages[i];
|
2016-01-27 17:20:58 +07:00
|
|
|
if (pg) {
|
|
|
|
|
unlock_page(pg);
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
put_page(pg);
|
2016-01-27 17:20:58 +07:00
|
|
|
}
|
2015-07-01 04:42:05 +07:00
|
|
|
}
|
|
|
|
|
kfree(cmp->src_pages);
|
|
|
|
|
kfree(cmp->dst_pages);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static int btrfs_cmp_data_prepare(struct inode *src, u64 loff,
|
|
|
|
|
struct inode *dst, u64 dst_loff,
|
|
|
|
|
u64 len, struct cmp_pages *cmp)
|
|
|
|
|
{
|
|
|
|
|
int ret;
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
int num_pages = PAGE_ALIGN(len) >> PAGE_SHIFT;
|
2015-07-01 04:42:05 +07:00
|
|
|
struct page **src_pgarr, **dst_pgarr;
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* We must gather up all the pages before we initiate our
|
|
|
|
|
* extent locking. We use an array for the page pointers. Size
|
|
|
|
|
* of the array is bounded by len, which is in turn bounded by
|
|
|
|
|
* BTRFS_MAX_DEDUPE_LEN.
|
|
|
|
|
*/
|
2016-02-11 21:01:38 +07:00
|
|
|
src_pgarr = kcalloc(num_pages, sizeof(struct page *), GFP_KERNEL);
|
|
|
|
|
dst_pgarr = kcalloc(num_pages, sizeof(struct page *), GFP_KERNEL);
|
2015-07-01 04:42:05 +07:00
|
|
|
if (!src_pgarr || !dst_pgarr) {
|
|
|
|
|
kfree(src_pgarr);
|
|
|
|
|
kfree(dst_pgarr);
|
|
|
|
|
return -ENOMEM;
|
2013-08-07 01:42:51 +07:00
|
|
|
}
|
2015-07-01 04:42:05 +07:00
|
|
|
cmp->num_pages = num_pages;
|
|
|
|
|
cmp->src_pages = src_pgarr;
|
|
|
|
|
cmp->dst_pages = dst_pgarr;
|
|
|
|
|
|
Btrfs: fix deadlock between dedup on same file and starting writeback
If we are deduping two ranges of the same file we need to make sure that
we lock all pages in ascending order, that is, lock first the pages from
the range with lower offset and then the pages from the other range, as
otherwise we can deadlock with a concurrent task that is starting delalloc
(writeback). Example trace:
[74073.052218] INFO: task kworker/u32:10:17997 blocked for more than 120 seconds.
[74073.053889] Tainted: G W 4.9.0-rc7-btrfs-next-36+ #1
[74073.055071] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
[74073.056696] kworker/u32:10 D 0 17997 2 0x00000000
[74073.058606] Workqueue: writeback wb_workfn (flush-btrfs-53176)
[74073.061370] ffff880031e79858 ffff8802159d2580 ffff880237004580 ffff880031e79240
[74073.064784] ffff88023f4978c0 ffffc9000817b638 ffffffff814c15e1 0000000000000000
[74073.068386] ffff88023f4978d8 ffff88023f4978c0 000000000017b620 ffff880031e79240
[74073.071712] Call Trace:
[74073.072884] [<ffffffff814c15e1>] ? __schedule+0x48f/0x6f4
[74073.075395] [<ffffffff814c1c8b>] ? bit_wait+0x2f/0x2f
[74073.077511] [<ffffffff814c18d2>] schedule+0x8c/0xa0
[74073.079440] [<ffffffff814c4b36>] schedule_timeout+0x43/0xff
[74073.081637] [<ffffffff8110953e>] ? time_hardirqs_on+0x9/0x14
[74073.083809] [<ffffffff81095c67>] ? trace_hardirqs_on_caller+0x16/0x197
[74073.086314] [<ffffffff810bde98>] ? timekeeping_get_ns+0x1e/0x32
[74073.100654] [<ffffffff810be048>] ? ktime_get+0x41/0x52
[74073.102619] [<ffffffff814c10f0>] io_schedule_timeout+0xa0/0x102
[74073.104771] [<ffffffff814c10f0>] ? io_schedule_timeout+0xa0/0x102
[74073.106969] [<ffffffff814c1ca6>] bit_wait_io+0x1b/0x39
[74073.108954] [<ffffffff814c1fb8>] __wait_on_bit_lock+0x4f/0x99
[74073.110981] [<ffffffff8112b692>] __lock_page+0x6b/0x6d
[74073.112833] [<ffffffff8108ceb4>] ? autoremove_wake_function+0x3a/0x3a
[74073.115010] [<ffffffffa031178b>] lock_page+0x2f/0x32 [btrfs]
[74073.116999] [<ffffffffa0311d9f>] lock_delalloc_pages+0xc7/0x1a0 [btrfs]
[74073.119243] [<ffffffffa0313d15>] find_lock_delalloc_range+0xc3/0x1a4 [btrfs]
[74073.121636] [<ffffffffa0313e81>] writepage_delalloc.isra.31+0x8b/0x134 [btrfs]
[74073.124229] [<ffffffffa0315d69>] __extent_writepage+0x1c1/0x2bf [btrfs]
[74073.126372] [<ffffffffa03160f2>] extent_write_cache_pages.isra.30.constprop.49+0x28b/0x36c [btrfs]
[74073.129371] [<ffffffffa03165b9>] extent_writepages+0x4b/0x5c [btrfs]
[74073.131440] [<ffffffffa02fcb59>] ? insert_reserved_file_extent.constprop.42+0x261/0x261 [btrfs]
[74073.134303] [<ffffffff811b4ce4>] ? writeback_sb_inodes+0xe0/0x4a1
[74073.136298] [<ffffffffa02fab7f>] btrfs_writepages+0x28/0x2a [btrfs]
[74073.138248] [<ffffffff81138200>] do_writepages+0x23/0x2c
[74073.139910] [<ffffffff811b3cab>] __writeback_single_inode+0x105/0x6d2
[74073.142003] [<ffffffff811b4e96>] writeback_sb_inodes+0x292/0x4a1
[74073.136298] [<ffffffffa02fab7f>] btrfs_writepages+0x28/0x2a [btrfs]
[74073.138248] [<ffffffff81138200>] do_writepages+0x23/0x2c
[74073.139910] [<ffffffff811b3cab>] __writeback_single_inode+0x105/0x6d2
[74073.142003] [<ffffffff811b4e96>] writeback_sb_inodes+0x292/0x4a1
[74073.143911] [<ffffffff811b511b>] __writeback_inodes_wb+0x76/0xae
[74073.145787] [<ffffffff811b53ca>] wb_writeback+0x1cc/0x4d7
[74073.147452] [<ffffffff811b60cd>] wb_workfn+0x194/0x37d
[74073.149084] [<ffffffff811b60cd>] ? wb_workfn+0x194/0x37d
[74073.150726] [<ffffffff8106ce77>] ? process_one_work+0x154/0x4e4
[74073.152694] [<ffffffff8106cf96>] process_one_work+0x273/0x4e4
[74073.154452] [<ffffffff8106d6db>] worker_thread+0x1eb/0x2ca
[74073.156138] [<ffffffff8106d4f0>] ? rescuer_thread+0x2b6/0x2b6
[74073.157837] [<ffffffff81072a81>] kthread+0xd5/0xdd
[74073.159339] [<ffffffff810729ac>] ? __kthread_unpark+0x5a/0x5a
[74073.161088] [<ffffffff814c6257>] ret_from_fork+0x27/0x40
[74073.162680] INFO: lockdep is turned off.
[74073.163855] INFO: task do-dedup:30264 blocked for more than 120 seconds.
[74073.181180] Tainted: G W 4.9.0-rc7-btrfs-next-36+ #1
[74073.181180] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
[74073.185296] fdm-stress D 0 30264 29974 0x00000000
[74073.186810] ffff880089595118 ffff880211b8eac0 ffff880237030380 ffff880089594b00
[74073.188998] ffff88023f2978c0 ffffc900063abb68 ffffffff814c15e1 0000000000000000
[74073.191070] ffff88023f2978d8 ffff88023f2978c0 00000000003abb50 ffff880089594b00
[74073.193286] Call Trace:
[74073.193990] [<ffffffff814c15e1>] ? __schedule+0x48f/0x6f4
[74073.195418] [<ffffffff814c1c8b>] ? bit_wait+0x2f/0x2f
[74073.196796] [<ffffffff814c18d2>] schedule+0x8c/0xa0
[74073.198163] [<ffffffff814c4b36>] schedule_timeout+0x43/0xff
[74073.199621] [<ffffffff81095df5>] ? trace_hardirqs_on+0xd/0xf
[74073.201100] [<ffffffff810bde98>] ? timekeeping_get_ns+0x1e/0x32
[74073.202686] [<ffffffff810be048>] ? ktime_get+0x41/0x52
[74073.204051] [<ffffffff814c10f0>] io_schedule_timeout+0xa0/0x102
[74073.205585] [<ffffffff814c10f0>] ? io_schedule_timeout+0xa0/0x102
[74073.207123] [<ffffffff814c1ca6>] bit_wait_io+0x1b/0x39
[74073.208238] [<ffffffff814c1fb8>] __wait_on_bit_lock+0x4f/0x99
[74073.208871] [<ffffffff8112b692>] __lock_page+0x6b/0x6d
[74073.209430] [<ffffffff8108ceb4>] ? autoremove_wake_function+0x3a/0x3a
[74073.210101] [<ffffffff8112b800>] lock_page+0x2f/0x32
[74073.210636] [<ffffffff8112c502>] pagecache_get_page+0x5e/0x153
[74073.211270] [<ffffffffa03257eb>] gather_extent_pages+0x4e/0x109 [btrfs]
[74073.212166] [<ffffffffa032a04c>] btrfs_dedupe_file_range+0x1e1/0x4dd [btrfs]
[74073.213257] [<ffffffff8118d9b5>] vfs_dedupe_file_range+0x1c1/0x221
[74073.214086] [<ffffffff8119e0c4>] do_vfs_ioctl+0x442/0x600
[74073.214767] [<ffffffff811a7874>] ? rcu_read_unlock+0x5b/0x5d
[74073.215619] [<ffffffff811a7953>] ? __fget+0x6b/0x77
[74073.216338] [<ffffffff8119e2d9>] SyS_ioctl+0x57/0x79
[74073.217149] [<ffffffff814c5fea>] entry_SYSCALL_64_fastpath+0x18/0xad
[74073.218102] [<ffffffff81109552>] ? time_hardirqs_off+0x9/0x14
[74073.218968] [<ffffffff810938ce>] ? trace_hardirqs_off_caller+0x1f/0xaa
[74073.219938] INFO: lockdep is turned off.
What happened was the following:
CPU 1 CPU 2
btrfs_dedupe_file_range()
--> using same inode as source
and target
--> src range is [768K, 1Mb[
--> dst range is [0, 256K[
btrfs_cmp_data_prepare()
--> calls gather_extent_pages()
for range [768K, 1Mb[ and
locks all pages in that range
do_writepages()
btrfs_writepages()
extent_writepages()
extent_write_cache_pages()
__extent_writepage()
writepage_delalloc()
find_lock_delalloc_range()
--> finds range [0, 1Mb[
lock_delalloc_pages()
--> locks all pages in the
range [0, 768K[
--> tries to lock page at
offset 768K
--> deadlock
--> calls gather_extent_pages()
to lock pages in the range
[0, 256K[
--> deadlock, task at CPU 1
already locked that
range and it's trying
to lock the range we
locked previously
So fix this by making sure that during a dedup we always lock first the
pages from the range with lower offset.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Chris Mason <clm@fb.com>
2017-02-22 00:14:52 +07:00
|
|
|
/*
|
|
|
|
|
* If deduping ranges in the same inode, locking rules make it mandatory
|
|
|
|
|
* to always lock pages in ascending order to avoid deadlocks with
|
|
|
|
|
* concurrent tasks (such as starting writeback/delalloc).
|
|
|
|
|
*/
|
|
|
|
|
if (src == dst && dst_loff < loff) {
|
|
|
|
|
swap(src_pgarr, dst_pgarr);
|
|
|
|
|
swap(loff, dst_loff);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
ret = gather_extent_pages(src, src_pgarr, cmp->num_pages, loff);
|
2015-07-01 04:42:05 +07:00
|
|
|
if (ret)
|
|
|
|
|
goto out;
|
|
|
|
|
|
Btrfs: fix deadlock between dedup on same file and starting writeback
If we are deduping two ranges of the same file we need to make sure that
we lock all pages in ascending order, that is, lock first the pages from
the range with lower offset and then the pages from the other range, as
otherwise we can deadlock with a concurrent task that is starting delalloc
(writeback). Example trace:
[74073.052218] INFO: task kworker/u32:10:17997 blocked for more than 120 seconds.
[74073.053889] Tainted: G W 4.9.0-rc7-btrfs-next-36+ #1
[74073.055071] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
[74073.056696] kworker/u32:10 D 0 17997 2 0x00000000
[74073.058606] Workqueue: writeback wb_workfn (flush-btrfs-53176)
[74073.061370] ffff880031e79858 ffff8802159d2580 ffff880237004580 ffff880031e79240
[74073.064784] ffff88023f4978c0 ffffc9000817b638 ffffffff814c15e1 0000000000000000
[74073.068386] ffff88023f4978d8 ffff88023f4978c0 000000000017b620 ffff880031e79240
[74073.071712] Call Trace:
[74073.072884] [<ffffffff814c15e1>] ? __schedule+0x48f/0x6f4
[74073.075395] [<ffffffff814c1c8b>] ? bit_wait+0x2f/0x2f
[74073.077511] [<ffffffff814c18d2>] schedule+0x8c/0xa0
[74073.079440] [<ffffffff814c4b36>] schedule_timeout+0x43/0xff
[74073.081637] [<ffffffff8110953e>] ? time_hardirqs_on+0x9/0x14
[74073.083809] [<ffffffff81095c67>] ? trace_hardirqs_on_caller+0x16/0x197
[74073.086314] [<ffffffff810bde98>] ? timekeeping_get_ns+0x1e/0x32
[74073.100654] [<ffffffff810be048>] ? ktime_get+0x41/0x52
[74073.102619] [<ffffffff814c10f0>] io_schedule_timeout+0xa0/0x102
[74073.104771] [<ffffffff814c10f0>] ? io_schedule_timeout+0xa0/0x102
[74073.106969] [<ffffffff814c1ca6>] bit_wait_io+0x1b/0x39
[74073.108954] [<ffffffff814c1fb8>] __wait_on_bit_lock+0x4f/0x99
[74073.110981] [<ffffffff8112b692>] __lock_page+0x6b/0x6d
[74073.112833] [<ffffffff8108ceb4>] ? autoremove_wake_function+0x3a/0x3a
[74073.115010] [<ffffffffa031178b>] lock_page+0x2f/0x32 [btrfs]
[74073.116999] [<ffffffffa0311d9f>] lock_delalloc_pages+0xc7/0x1a0 [btrfs]
[74073.119243] [<ffffffffa0313d15>] find_lock_delalloc_range+0xc3/0x1a4 [btrfs]
[74073.121636] [<ffffffffa0313e81>] writepage_delalloc.isra.31+0x8b/0x134 [btrfs]
[74073.124229] [<ffffffffa0315d69>] __extent_writepage+0x1c1/0x2bf [btrfs]
[74073.126372] [<ffffffffa03160f2>] extent_write_cache_pages.isra.30.constprop.49+0x28b/0x36c [btrfs]
[74073.129371] [<ffffffffa03165b9>] extent_writepages+0x4b/0x5c [btrfs]
[74073.131440] [<ffffffffa02fcb59>] ? insert_reserved_file_extent.constprop.42+0x261/0x261 [btrfs]
[74073.134303] [<ffffffff811b4ce4>] ? writeback_sb_inodes+0xe0/0x4a1
[74073.136298] [<ffffffffa02fab7f>] btrfs_writepages+0x28/0x2a [btrfs]
[74073.138248] [<ffffffff81138200>] do_writepages+0x23/0x2c
[74073.139910] [<ffffffff811b3cab>] __writeback_single_inode+0x105/0x6d2
[74073.142003] [<ffffffff811b4e96>] writeback_sb_inodes+0x292/0x4a1
[74073.136298] [<ffffffffa02fab7f>] btrfs_writepages+0x28/0x2a [btrfs]
[74073.138248] [<ffffffff81138200>] do_writepages+0x23/0x2c
[74073.139910] [<ffffffff811b3cab>] __writeback_single_inode+0x105/0x6d2
[74073.142003] [<ffffffff811b4e96>] writeback_sb_inodes+0x292/0x4a1
[74073.143911] [<ffffffff811b511b>] __writeback_inodes_wb+0x76/0xae
[74073.145787] [<ffffffff811b53ca>] wb_writeback+0x1cc/0x4d7
[74073.147452] [<ffffffff811b60cd>] wb_workfn+0x194/0x37d
[74073.149084] [<ffffffff811b60cd>] ? wb_workfn+0x194/0x37d
[74073.150726] [<ffffffff8106ce77>] ? process_one_work+0x154/0x4e4
[74073.152694] [<ffffffff8106cf96>] process_one_work+0x273/0x4e4
[74073.154452] [<ffffffff8106d6db>] worker_thread+0x1eb/0x2ca
[74073.156138] [<ffffffff8106d4f0>] ? rescuer_thread+0x2b6/0x2b6
[74073.157837] [<ffffffff81072a81>] kthread+0xd5/0xdd
[74073.159339] [<ffffffff810729ac>] ? __kthread_unpark+0x5a/0x5a
[74073.161088] [<ffffffff814c6257>] ret_from_fork+0x27/0x40
[74073.162680] INFO: lockdep is turned off.
[74073.163855] INFO: task do-dedup:30264 blocked for more than 120 seconds.
[74073.181180] Tainted: G W 4.9.0-rc7-btrfs-next-36+ #1
[74073.181180] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
[74073.185296] fdm-stress D 0 30264 29974 0x00000000
[74073.186810] ffff880089595118 ffff880211b8eac0 ffff880237030380 ffff880089594b00
[74073.188998] ffff88023f2978c0 ffffc900063abb68 ffffffff814c15e1 0000000000000000
[74073.191070] ffff88023f2978d8 ffff88023f2978c0 00000000003abb50 ffff880089594b00
[74073.193286] Call Trace:
[74073.193990] [<ffffffff814c15e1>] ? __schedule+0x48f/0x6f4
[74073.195418] [<ffffffff814c1c8b>] ? bit_wait+0x2f/0x2f
[74073.196796] [<ffffffff814c18d2>] schedule+0x8c/0xa0
[74073.198163] [<ffffffff814c4b36>] schedule_timeout+0x43/0xff
[74073.199621] [<ffffffff81095df5>] ? trace_hardirqs_on+0xd/0xf
[74073.201100] [<ffffffff810bde98>] ? timekeeping_get_ns+0x1e/0x32
[74073.202686] [<ffffffff810be048>] ? ktime_get+0x41/0x52
[74073.204051] [<ffffffff814c10f0>] io_schedule_timeout+0xa0/0x102
[74073.205585] [<ffffffff814c10f0>] ? io_schedule_timeout+0xa0/0x102
[74073.207123] [<ffffffff814c1ca6>] bit_wait_io+0x1b/0x39
[74073.208238] [<ffffffff814c1fb8>] __wait_on_bit_lock+0x4f/0x99
[74073.208871] [<ffffffff8112b692>] __lock_page+0x6b/0x6d
[74073.209430] [<ffffffff8108ceb4>] ? autoremove_wake_function+0x3a/0x3a
[74073.210101] [<ffffffff8112b800>] lock_page+0x2f/0x32
[74073.210636] [<ffffffff8112c502>] pagecache_get_page+0x5e/0x153
[74073.211270] [<ffffffffa03257eb>] gather_extent_pages+0x4e/0x109 [btrfs]
[74073.212166] [<ffffffffa032a04c>] btrfs_dedupe_file_range+0x1e1/0x4dd [btrfs]
[74073.213257] [<ffffffff8118d9b5>] vfs_dedupe_file_range+0x1c1/0x221
[74073.214086] [<ffffffff8119e0c4>] do_vfs_ioctl+0x442/0x600
[74073.214767] [<ffffffff811a7874>] ? rcu_read_unlock+0x5b/0x5d
[74073.215619] [<ffffffff811a7953>] ? __fget+0x6b/0x77
[74073.216338] [<ffffffff8119e2d9>] SyS_ioctl+0x57/0x79
[74073.217149] [<ffffffff814c5fea>] entry_SYSCALL_64_fastpath+0x18/0xad
[74073.218102] [<ffffffff81109552>] ? time_hardirqs_off+0x9/0x14
[74073.218968] [<ffffffff810938ce>] ? trace_hardirqs_off_caller+0x1f/0xaa
[74073.219938] INFO: lockdep is turned off.
What happened was the following:
CPU 1 CPU 2
btrfs_dedupe_file_range()
--> using same inode as source
and target
--> src range is [768K, 1Mb[
--> dst range is [0, 256K[
btrfs_cmp_data_prepare()
--> calls gather_extent_pages()
for range [768K, 1Mb[ and
locks all pages in that range
do_writepages()
btrfs_writepages()
extent_writepages()
extent_write_cache_pages()
__extent_writepage()
writepage_delalloc()
find_lock_delalloc_range()
--> finds range [0, 1Mb[
lock_delalloc_pages()
--> locks all pages in the
range [0, 768K[
--> tries to lock page at
offset 768K
--> deadlock
--> calls gather_extent_pages()
to lock pages in the range
[0, 256K[
--> deadlock, task at CPU 1
already locked that
range and it's trying
to lock the range we
locked previously
So fix this by making sure that during a dedup we always lock first the
pages from the range with lower offset.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Chris Mason <clm@fb.com>
2017-02-22 00:14:52 +07:00
|
|
|
ret = gather_extent_pages(dst, dst_pgarr, cmp->num_pages, dst_loff);
|
2015-07-01 04:42:05 +07:00
|
|
|
|
|
|
|
|
out:
|
|
|
|
|
if (ret)
|
|
|
|
|
btrfs_cmp_data_free(cmp);
|
2017-09-08 15:48:55 +07:00
|
|
|
return ret;
|
2013-08-07 01:42:51 +07:00
|
|
|
}
|
|
|
|
|
|
2017-02-11 02:15:10 +07:00
|
|
|
static int btrfs_cmp_data(u64 len, struct cmp_pages *cmp)
|
2013-08-07 01:42:51 +07:00
|
|
|
{
|
|
|
|
|
int ret = 0;
|
2015-07-01 04:42:05 +07:00
|
|
|
int i;
|
2013-08-07 01:42:51 +07:00
|
|
|
struct page *src_page, *dst_page;
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
unsigned int cmp_len = PAGE_SIZE;
|
2013-08-07 01:42:51 +07:00
|
|
|
void *addr, *dst_addr;
|
|
|
|
|
|
2015-07-01 04:42:05 +07:00
|
|
|
i = 0;
|
2013-08-07 01:42:51 +07:00
|
|
|
while (len) {
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
if (len < PAGE_SIZE)
|
2013-08-07 01:42:51 +07:00
|
|
|
cmp_len = len;
|
|
|
|
|
|
2015-07-01 04:42:05 +07:00
|
|
|
BUG_ON(i >= cmp->num_pages);
|
|
|
|
|
|
|
|
|
|
src_page = cmp->src_pages[i];
|
|
|
|
|
dst_page = cmp->dst_pages[i];
|
2016-01-27 17:20:58 +07:00
|
|
|
ASSERT(PageLocked(src_page));
|
|
|
|
|
ASSERT(PageLocked(dst_page));
|
2015-07-01 04:42:05 +07:00
|
|
|
|
2013-08-07 01:42:51 +07:00
|
|
|
addr = kmap_atomic(src_page);
|
|
|
|
|
dst_addr = kmap_atomic(dst_page);
|
|
|
|
|
|
|
|
|
|
flush_dcache_page(src_page);
|
|
|
|
|
flush_dcache_page(dst_page);
|
|
|
|
|
|
|
|
|
|
if (memcmp(addr, dst_addr, cmp_len))
|
2015-12-19 15:56:05 +07:00
|
|
|
ret = -EBADE;
|
2013-08-07 01:42:51 +07:00
|
|
|
|
|
|
|
|
kunmap_atomic(addr);
|
|
|
|
|
kunmap_atomic(dst_addr);
|
|
|
|
|
|
|
|
|
|
if (ret)
|
|
|
|
|
break;
|
|
|
|
|
|
|
|
|
|
len -= cmp_len;
|
2015-07-01 04:42:05 +07:00
|
|
|
i++;
|
2013-08-07 01:42:51 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
2015-06-09 05:05:25 +07:00
|
|
|
static int extent_same_check_offsets(struct inode *inode, u64 off, u64 *plen,
|
|
|
|
|
u64 olen)
|
2013-08-07 01:42:51 +07:00
|
|
|
{
|
2015-06-09 05:05:25 +07:00
|
|
|
u64 len = *plen;
|
2013-08-07 01:42:51 +07:00
|
|
|
u64 bs = BTRFS_I(inode)->root->fs_info->sb->s_blocksize;
|
|
|
|
|
|
2015-06-09 05:05:25 +07:00
|
|
|
if (off + olen > inode->i_size || off + olen < off)
|
2013-08-07 01:42:51 +07:00
|
|
|
return -EINVAL;
|
2015-06-09 05:05:25 +07:00
|
|
|
|
|
|
|
|
/* if we extend to eof, continue to block boundary */
|
|
|
|
|
if (off + len == inode->i_size)
|
|
|
|
|
*plen = len = ALIGN(inode->i_size, bs) - off;
|
|
|
|
|
|
2013-08-07 01:42:51 +07:00
|
|
|
/* Check that we are block aligned - btrfs_clone() requires this */
|
|
|
|
|
if (!IS_ALIGNED(off, bs) || !IS_ALIGNED(off + len, bs))
|
|
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
2015-06-09 05:05:25 +07:00
|
|
|
static int btrfs_extent_same(struct inode *src, u64 loff, u64 olen,
|
2013-08-07 01:42:51 +07:00
|
|
|
struct inode *dst, u64 dst_loff)
|
|
|
|
|
{
|
|
|
|
|
int ret;
|
2015-06-09 05:05:25 +07:00
|
|
|
u64 len = olen;
|
2015-07-01 04:42:05 +07:00
|
|
|
struct cmp_pages cmp;
|
2017-01-18 14:37:38 +07:00
|
|
|
bool same_inode = (src == dst);
|
2015-07-01 04:42:07 +07:00
|
|
|
u64 same_lock_start = 0;
|
|
|
|
|
u64 same_lock_len = 0;
|
2013-08-07 01:42:51 +07:00
|
|
|
|
2015-03-31 00:26:47 +07:00
|
|
|
if (len == 0)
|
|
|
|
|
return 0;
|
|
|
|
|
|
2017-01-18 14:37:38 +07:00
|
|
|
if (same_inode)
|
2016-01-23 03:40:57 +07:00
|
|
|
inode_lock(src);
|
2017-01-18 14:37:38 +07:00
|
|
|
else
|
|
|
|
|
btrfs_double_inode_lock(src, dst);
|
2013-08-07 01:42:51 +07:00
|
|
|
|
2017-01-18 14:37:38 +07:00
|
|
|
ret = extent_same_check_offsets(src, loff, &len, olen);
|
|
|
|
|
if (ret)
|
|
|
|
|
goto out_unlock;
|
2013-08-07 01:42:51 +07:00
|
|
|
|
2017-01-18 14:37:38 +07:00
|
|
|
ret = extent_same_check_offsets(dst, dst_loff, &len, olen);
|
|
|
|
|
if (ret)
|
|
|
|
|
goto out_unlock;
|
|
|
|
|
|
|
|
|
|
if (same_inode) {
|
2015-07-01 04:42:07 +07:00
|
|
|
/*
|
|
|
|
|
* Single inode case wants the same checks, except we
|
|
|
|
|
* don't want our length pushed out past i_size as
|
|
|
|
|
* comparing that data range makes no sense.
|
|
|
|
|
*
|
|
|
|
|
* extent_same_check_offsets() will do this for an
|
|
|
|
|
* unaligned length at i_size, so catch it here and
|
|
|
|
|
* reject the request.
|
|
|
|
|
*
|
|
|
|
|
* This effectively means we require aligned extents
|
|
|
|
|
* for the single-inode case, whereas the other cases
|
|
|
|
|
* allow an unaligned length so long as it ends at
|
|
|
|
|
* i_size.
|
|
|
|
|
*/
|
|
|
|
|
if (len != olen) {
|
|
|
|
|
ret = -EINVAL;
|
|
|
|
|
goto out_unlock;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* Check for overlapping ranges */
|
|
|
|
|
if (dst_loff + len > loff && dst_loff < loff + len) {
|
|
|
|
|
ret = -EINVAL;
|
|
|
|
|
goto out_unlock;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
same_lock_start = min_t(u64, loff, dst_loff);
|
|
|
|
|
same_lock_len = max_t(u64, loff, dst_loff) + len - same_lock_start;
|
|
|
|
|
}
|
2013-08-07 01:42:51 +07:00
|
|
|
|
|
|
|
|
/* don't make the dst file partly checksummed */
|
|
|
|
|
if ((BTRFS_I(src)->flags & BTRFS_INODE_NODATASUM) !=
|
|
|
|
|
(BTRFS_I(dst)->flags & BTRFS_INODE_NODATASUM)) {
|
|
|
|
|
ret = -EINVAL;
|
|
|
|
|
goto out_unlock;
|
|
|
|
|
}
|
|
|
|
|
|
2016-01-27 17:20:58 +07:00
|
|
|
again:
|
2015-07-01 04:42:05 +07:00
|
|
|
ret = btrfs_cmp_data_prepare(src, loff, dst, dst_loff, olen, &cmp);
|
|
|
|
|
if (ret)
|
|
|
|
|
goto out_unlock;
|
|
|
|
|
|
2015-07-01 04:42:07 +07:00
|
|
|
if (same_inode)
|
2016-01-27 17:20:58 +07:00
|
|
|
ret = lock_extent_range(src, same_lock_start, same_lock_len,
|
|
|
|
|
false);
|
2015-07-01 04:42:07 +07:00
|
|
|
else
|
2016-01-27 17:20:58 +07:00
|
|
|
ret = btrfs_double_extent_lock(src, loff, dst, dst_loff, len,
|
|
|
|
|
false);
|
|
|
|
|
/*
|
|
|
|
|
* If one of the inodes has dirty pages in the respective range or
|
|
|
|
|
* ordered extents, we need to flush dellaloc and wait for all ordered
|
|
|
|
|
* extents in the range. We must unlock the pages and the ranges in the
|
|
|
|
|
* io trees to avoid deadlocks when flushing delalloc (requires locking
|
|
|
|
|
* pages) and when waiting for ordered extents to complete (they require
|
|
|
|
|
* range locking).
|
|
|
|
|
*/
|
|
|
|
|
if (ret == -EAGAIN) {
|
|
|
|
|
/*
|
|
|
|
|
* Ranges in the io trees already unlocked. Now unlock all
|
|
|
|
|
* pages before waiting for all IO to complete.
|
|
|
|
|
*/
|
|
|
|
|
btrfs_cmp_data_free(&cmp);
|
|
|
|
|
if (same_inode) {
|
|
|
|
|
btrfs_wait_ordered_range(src, same_lock_start,
|
|
|
|
|
same_lock_len);
|
|
|
|
|
} else {
|
|
|
|
|
btrfs_wait_ordered_range(src, loff, len);
|
|
|
|
|
btrfs_wait_ordered_range(dst, dst_loff, len);
|
|
|
|
|
}
|
|
|
|
|
goto again;
|
|
|
|
|
}
|
|
|
|
|
ASSERT(ret == 0);
|
|
|
|
|
if (WARN_ON(ret)) {
|
|
|
|
|
/* ranges in the io trees already unlocked */
|
|
|
|
|
btrfs_cmp_data_free(&cmp);
|
|
|
|
|
return ret;
|
|
|
|
|
}
|
2015-07-01 04:42:05 +07:00
|
|
|
|
2015-07-01 04:42:04 +07:00
|
|
|
/* pass original length for comparison so we stay within i_size */
|
2017-02-11 02:15:10 +07:00
|
|
|
ret = btrfs_cmp_data(olen, &cmp);
|
2013-08-07 01:42:51 +07:00
|
|
|
if (ret == 0)
|
2015-07-01 04:42:08 +07:00
|
|
|
ret = btrfs_clone(src, dst, loff, olen, len, dst_loff, 1);
|
2013-08-07 01:42:51 +07:00
|
|
|
|
2015-07-01 04:42:07 +07:00
|
|
|
if (same_inode)
|
|
|
|
|
unlock_extent(&BTRFS_I(src)->io_tree, same_lock_start,
|
|
|
|
|
same_lock_start + same_lock_len - 1);
|
|
|
|
|
else
|
|
|
|
|
btrfs_double_extent_unlock(src, loff, dst, dst_loff, len);
|
2015-07-01 04:42:05 +07:00
|
|
|
|
|
|
|
|
btrfs_cmp_data_free(&cmp);
|
2013-08-07 01:42:51 +07:00
|
|
|
out_unlock:
|
2015-07-01 04:42:07 +07:00
|
|
|
if (same_inode)
|
2016-01-23 03:40:57 +07:00
|
|
|
inode_unlock(src);
|
2015-07-01 04:42:07 +07:00
|
|
|
else
|
|
|
|
|
btrfs_double_inode_unlock(src, dst);
|
2013-08-07 01:42:51 +07:00
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
2015-12-14 23:42:10 +07:00
|
|
|
#define BTRFS_MAX_DEDUPE_LEN SZ_16M
|
2013-08-07 01:42:51 +07:00
|
|
|
|
2015-12-19 15:56:05 +07:00
|
|
|
ssize_t btrfs_dedupe_file_range(struct file *src_file, u64 loff, u64 olen,
|
|
|
|
|
struct file *dst_file, u64 dst_loff)
|
2013-08-07 01:42:51 +07:00
|
|
|
{
|
2015-12-19 15:56:05 +07:00
|
|
|
struct inode *src = file_inode(src_file);
|
|
|
|
|
struct inode *dst = file_inode(dst_file);
|
2013-08-07 01:42:51 +07:00
|
|
|
u64 bs = BTRFS_I(src)->root->fs_info->sb->s_blocksize;
|
2015-12-19 15:56:05 +07:00
|
|
|
ssize_t res;
|
2013-08-07 01:42:51 +07:00
|
|
|
|
2015-12-19 15:56:05 +07:00
|
|
|
if (olen > BTRFS_MAX_DEDUPE_LEN)
|
|
|
|
|
olen = BTRFS_MAX_DEDUPE_LEN;
|
2013-08-07 01:42:51 +07:00
|
|
|
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
if (WARN_ON_ONCE(bs < PAGE_SIZE)) {
|
2013-08-07 01:42:51 +07:00
|
|
|
/*
|
|
|
|
|
* Btrfs does not support blocksize < page_size. As a
|
|
|
|
|
* result, btrfs_cmp_data() won't correctly handle
|
|
|
|
|
* this situation without an update.
|
|
|
|
|
*/
|
2015-12-19 15:56:05 +07:00
|
|
|
return -EINVAL;
|
2013-08-07 01:42:51 +07:00
|
|
|
}
|
|
|
|
|
|
2015-12-19 15:56:05 +07:00
|
|
|
res = btrfs_extent_same(src, loff, olen, dst, dst_loff);
|
|
|
|
|
if (res)
|
|
|
|
|
return res;
|
|
|
|
|
return olen;
|
2013-08-07 01:42:51 +07:00
|
|
|
}
|
|
|
|
|
|
2014-06-01 07:50:28 +07:00
|
|
|
static int clone_finish_inode_update(struct btrfs_trans_handle *trans,
|
|
|
|
|
struct inode *inode,
|
|
|
|
|
u64 endoff,
|
|
|
|
|
const u64 destoff,
|
2015-07-01 04:42:08 +07:00
|
|
|
const u64 olen,
|
|
|
|
|
int no_time_update)
|
2014-06-01 07:50:28 +07:00
|
|
|
{
|
|
|
|
|
struct btrfs_root *root = BTRFS_I(inode)->root;
|
|
|
|
|
int ret;
|
|
|
|
|
|
|
|
|
|
inode_inc_iversion(inode);
|
2015-07-01 04:42:08 +07:00
|
|
|
if (!no_time_update)
|
2016-09-14 21:48:06 +07:00
|
|
|
inode->i_mtime = inode->i_ctime = current_time(inode);
|
2014-06-01 07:50:28 +07:00
|
|
|
/*
|
|
|
|
|
* We round up to the block size at eof when determining which
|
|
|
|
|
* extents to clone above, but shouldn't round up the file size.
|
|
|
|
|
*/
|
|
|
|
|
if (endoff > destoff + olen)
|
|
|
|
|
endoff = destoff + olen;
|
|
|
|
|
if (endoff > inode->i_size)
|
2017-02-20 18:50:34 +07:00
|
|
|
btrfs_i_size_write(BTRFS_I(inode), endoff);
|
2014-06-01 07:50:28 +07:00
|
|
|
|
|
|
|
|
ret = btrfs_update_inode(trans, root, inode);
|
|
|
|
|
if (ret) {
|
2016-06-11 05:19:25 +07:00
|
|
|
btrfs_abort_transaction(trans, ret);
|
2016-09-10 08:39:03 +07:00
|
|
|
btrfs_end_transaction(trans);
|
2014-06-01 07:50:28 +07:00
|
|
|
goto out;
|
|
|
|
|
}
|
2016-09-10 08:39:03 +07:00
|
|
|
ret = btrfs_end_transaction(trans);
|
2014-06-01 07:50:28 +07:00
|
|
|
out:
|
|
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
2017-02-20 18:51:04 +07:00
|
|
|
static void clone_update_extent_map(struct btrfs_inode *inode,
|
2014-06-09 09:48:05 +07:00
|
|
|
const struct btrfs_trans_handle *trans,
|
|
|
|
|
const struct btrfs_path *path,
|
|
|
|
|
const u64 hole_offset,
|
|
|
|
|
const u64 hole_len)
|
|
|
|
|
{
|
2017-02-20 18:51:04 +07:00
|
|
|
struct extent_map_tree *em_tree = &inode->extent_tree;
|
2014-06-09 09:48:05 +07:00
|
|
|
struct extent_map *em;
|
|
|
|
|
int ret;
|
|
|
|
|
|
|
|
|
|
em = alloc_extent_map();
|
|
|
|
|
if (!em) {
|
2017-02-20 18:51:04 +07:00
|
|
|
set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, &inode->runtime_flags);
|
2014-06-09 09:48:05 +07:00
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
|
2014-06-30 03:45:40 +07:00
|
|
|
if (path) {
|
|
|
|
|
struct btrfs_file_extent_item *fi;
|
|
|
|
|
|
|
|
|
|
fi = btrfs_item_ptr(path->nodes[0], path->slots[0],
|
|
|
|
|
struct btrfs_file_extent_item);
|
2017-02-20 18:51:04 +07:00
|
|
|
btrfs_extent_item_to_extent_map(inode, path, fi, false, em);
|
2014-06-09 09:48:05 +07:00
|
|
|
em->generation = -1;
|
|
|
|
|
if (btrfs_file_extent_type(path->nodes[0], fi) ==
|
|
|
|
|
BTRFS_FILE_EXTENT_INLINE)
|
|
|
|
|
set_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
|
2017-02-20 18:51:04 +07:00
|
|
|
&inode->runtime_flags);
|
2014-06-09 09:48:05 +07:00
|
|
|
} else {
|
|
|
|
|
em->start = hole_offset;
|
|
|
|
|
em->len = hole_len;
|
|
|
|
|
em->ram_bytes = em->len;
|
|
|
|
|
em->orig_start = hole_offset;
|
|
|
|
|
em->block_start = EXTENT_MAP_HOLE;
|
|
|
|
|
em->block_len = 0;
|
|
|
|
|
em->orig_block_len = 0;
|
|
|
|
|
em->compress_type = BTRFS_COMPRESS_NONE;
|
|
|
|
|
em->generation = trans->transid;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
while (1) {
|
|
|
|
|
write_lock(&em_tree->lock);
|
|
|
|
|
ret = add_extent_mapping(em_tree, em, 1);
|
|
|
|
|
write_unlock(&em_tree->lock);
|
|
|
|
|
if (ret != -EEXIST) {
|
|
|
|
|
free_extent_map(em);
|
|
|
|
|
break;
|
|
|
|
|
}
|
2017-02-20 18:51:04 +07:00
|
|
|
btrfs_drop_extent_cache(inode, em->start,
|
2014-06-09 09:48:05 +07:00
|
|
|
em->start + em->len - 1, 0);
|
|
|
|
|
}
|
|
|
|
|
|
2014-09-30 06:33:33 +07:00
|
|
|
if (ret)
|
2017-02-20 18:51:04 +07:00
|
|
|
set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, &inode->runtime_flags);
|
2014-06-09 09:48:05 +07:00
|
|
|
}
|
|
|
|
|
|
Btrfs: fix file corruption and data loss after cloning inline extents
Currently the clone ioctl allows to clone an inline extent from one file
to another that already has other (non-inlined) extents. This is a problem
because btrfs is not designed to deal with files having inline and regular
extents, if a file has an inline extent then it must be the only extent
in the file and must start at file offset 0. Having a file with an inline
extent followed by regular extents results in EIO errors when doing reads
or writes against the first 4K of the file.
Also, the clone ioctl allows one to lose data if the source file consists
of a single inline extent, with a size of N bytes, and the destination
file consists of a single inline extent with a size of M bytes, where we
have M > N. In this case the clone operation removes the inline extent
from the destination file and then copies the inline extent from the
source file into the destination file - we lose the M - N bytes from the
destination file, a read operation will get the value 0x00 for any bytes
in the the range [N, M] (the destination inode's i_size remained as M,
that's why we can read past N bytes).
So fix this by not allowing such destructive operations to happen and
return errno EOPNOTSUPP to user space.
Currently the fstest btrfs/035 tests the data loss case but it totally
ignores this - i.e. expects the operation to succeed and does not check
the we got data loss.
The following test case for fstests exercises all these cases that result
in file corruption and data loss:
seq=`basename $0`
seqres=$RESULT_DIR/$seq
echo "QA output created by $seq"
tmp=/tmp/$$
status=1 # failure is the default!
trap "_cleanup; exit \$status" 0 1 2 3 15
_cleanup()
{
rm -f $tmp.*
}
# get standard environment, filters and checks
. ./common/rc
. ./common/filter
# real QA test starts here
_need_to_be_root
_supported_fs btrfs
_supported_os Linux
_require_scratch
_require_cloner
_require_btrfs_fs_feature "no_holes"
_require_btrfs_mkfs_feature "no-holes"
rm -f $seqres.full
test_cloning_inline_extents()
{
local mkfs_opts=$1
local mount_opts=$2
_scratch_mkfs $mkfs_opts >>$seqres.full 2>&1
_scratch_mount $mount_opts
# File bar, the source for all the following clone operations, consists
# of a single inline extent (50 bytes).
$XFS_IO_PROG -f -c "pwrite -S 0xbb 0 50" $SCRATCH_MNT/bar \
| _filter_xfs_io
# Test cloning into a file with an extent (non-inlined) where the
# destination offset overlaps that extent. It should not be possible to
# clone the inline extent from file bar into this file.
$XFS_IO_PROG -f -c "pwrite -S 0xaa 0K 16K" $SCRATCH_MNT/foo \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo
# Doing IO against any range in the first 4K of the file should work.
# Due to a past clone ioctl bug which allowed cloning the inline extent,
# these operations resulted in EIO errors.
echo "File foo data after clone operation:"
# All bytes should have the value 0xaa (clone operation failed and did
# not modify our file).
od -t x1 $SCRATCH_MNT/foo
$XFS_IO_PROG -c "pwrite -S 0xcc 0 100" $SCRATCH_MNT/foo | _filter_xfs_io
# Test cloning the inline extent against a file which has a hole in its
# first 4K followed by a non-inlined extent. It should not be possible
# as well to clone the inline extent from file bar into this file.
$XFS_IO_PROG -f -c "pwrite -S 0xdd 4K 12K" $SCRATCH_MNT/foo2 \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo2
# Doing IO against any range in the first 4K of the file should work.
# Due to a past clone ioctl bug which allowed cloning the inline extent,
# these operations resulted in EIO errors.
echo "File foo2 data after clone operation:"
# All bytes should have the value 0x00 (clone operation failed and did
# not modify our file).
od -t x1 $SCRATCH_MNT/foo2
$XFS_IO_PROG -c "pwrite -S 0xee 0 90" $SCRATCH_MNT/foo2 | _filter_xfs_io
# Test cloning the inline extent against a file which has a size of zero
# but has a prealloc extent. It should not be possible as well to clone
# the inline extent from file bar into this file.
$XFS_IO_PROG -f -c "falloc -k 0 1M" $SCRATCH_MNT/foo3 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo3
# Doing IO against any range in the first 4K of the file should work.
# Due to a past clone ioctl bug which allowed cloning the inline extent,
# these operations resulted in EIO errors.
echo "First 50 bytes of foo3 after clone operation:"
# Should not be able to read any bytes, file has 0 bytes i_size (the
# clone operation failed and did not modify our file).
od -t x1 $SCRATCH_MNT/foo3
$XFS_IO_PROG -c "pwrite -S 0xff 0 90" $SCRATCH_MNT/foo3 | _filter_xfs_io
# Test cloning the inline extent against a file which consists of a
# single inline extent that has a size not greater than the size of
# bar's inline extent (40 < 50).
# It should be possible to do the extent cloning from bar to this file.
$XFS_IO_PROG -f -c "pwrite -S 0x01 0 40" $SCRATCH_MNT/foo4 \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo4
# Doing IO against any range in the first 4K of the file should work.
echo "File foo4 data after clone operation:"
# Must match file bar's content.
od -t x1 $SCRATCH_MNT/foo4
$XFS_IO_PROG -c "pwrite -S 0x02 0 90" $SCRATCH_MNT/foo4 | _filter_xfs_io
# Test cloning the inline extent against a file which consists of a
# single inline extent that has a size greater than the size of bar's
# inline extent (60 > 50).
# It should not be possible to clone the inline extent from file bar
# into this file.
$XFS_IO_PROG -f -c "pwrite -S 0x03 0 60" $SCRATCH_MNT/foo5 \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo5
# Reading the file should not fail.
echo "File foo5 data after clone operation:"
# Must have a size of 60 bytes, with all bytes having a value of 0x03
# (the clone operation failed and did not modify our file).
od -t x1 $SCRATCH_MNT/foo5
# Test cloning the inline extent against a file which has no extents but
# has a size greater than bar's inline extent (16K > 50).
# It should not be possible to clone the inline extent from file bar
# into this file.
$XFS_IO_PROG -f -c "truncate 16K" $SCRATCH_MNT/foo6 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo6
# Reading the file should not fail.
echo "File foo6 data after clone operation:"
# Must have a size of 16K, with all bytes having a value of 0x00 (the
# clone operation failed and did not modify our file).
od -t x1 $SCRATCH_MNT/foo6
# Test cloning the inline extent against a file which has no extents but
# has a size not greater than bar's inline extent (30 < 50).
# It should be possible to clone the inline extent from file bar into
# this file.
$XFS_IO_PROG -f -c "truncate 30" $SCRATCH_MNT/foo7 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo7
# Reading the file should not fail.
echo "File foo7 data after clone operation:"
# Must have a size of 50 bytes, with all bytes having a value of 0xbb.
od -t x1 $SCRATCH_MNT/foo7
# Test cloning the inline extent against a file which has a size not
# greater than the size of bar's inline extent (20 < 50) but has
# a prealloc extent that goes beyond the file's size. It should not be
# possible to clone the inline extent from bar into this file.
$XFS_IO_PROG -f -c "falloc -k 0 1M" \
-c "pwrite -S 0x88 0 20" \
$SCRATCH_MNT/foo8 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo8
echo "File foo8 data after clone operation:"
# Must have a size of 20 bytes, with all bytes having a value of 0x88
# (the clone operation did not modify our file).
od -t x1 $SCRATCH_MNT/foo8
_scratch_unmount
}
echo -e "\nTesting without compression and without the no-holes feature...\n"
test_cloning_inline_extents
echo -e "\nTesting with compression and without the no-holes feature...\n"
test_cloning_inline_extents "" "-o compress"
echo -e "\nTesting without compression and with the no-holes feature...\n"
test_cloning_inline_extents "-O no-holes" ""
echo -e "\nTesting with compression and with the no-holes feature...\n"
test_cloning_inline_extents "-O no-holes" "-o compress"
status=0
exit
Cc: stable@vger.kernel.org
Signed-off-by: Filipe Manana <fdmanana@suse.com>
2015-10-13 21:15:00 +07:00
|
|
|
/*
|
|
|
|
|
* Make sure we do not end up inserting an inline extent into a file that has
|
|
|
|
|
* already other (non-inline) extents. If a file has an inline extent it can
|
|
|
|
|
* not have any other extents and the (single) inline extent must start at the
|
|
|
|
|
* file offset 0. Failing to respect these rules will lead to file corruption,
|
|
|
|
|
* resulting in EIO errors on read/write operations, hitting BUG_ON's in mm, etc
|
|
|
|
|
*
|
|
|
|
|
* We can have extents that have been already written to disk or we can have
|
|
|
|
|
* dirty ranges still in delalloc, in which case the extent maps and items are
|
|
|
|
|
* created only when we run delalloc, and the delalloc ranges might fall outside
|
|
|
|
|
* the range we are currently locking in the inode's io tree. So we check the
|
|
|
|
|
* inode's i_size because of that (i_size updates are done while holding the
|
|
|
|
|
* i_mutex, which we are holding here).
|
|
|
|
|
* We also check to see if the inode has a size not greater than "datal" but has
|
|
|
|
|
* extents beyond it, due to an fallocate with FALLOC_FL_KEEP_SIZE (and we are
|
|
|
|
|
* protected against such concurrent fallocate calls by the i_mutex).
|
|
|
|
|
*
|
|
|
|
|
* If the file has no extents but a size greater than datal, do not allow the
|
|
|
|
|
* copy because we would need turn the inline extent into a non-inline one (even
|
|
|
|
|
* with NO_HOLES enabled). If we find our destination inode only has one inline
|
|
|
|
|
* extent, just overwrite it with the source inline extent if its size is less
|
|
|
|
|
* than the source extent's size, or we could copy the source inline extent's
|
|
|
|
|
* data into the destination inode's inline extent if the later is greater then
|
|
|
|
|
* the former.
|
|
|
|
|
*/
|
2017-02-11 02:18:49 +07:00
|
|
|
static int clone_copy_inline_extent(struct inode *dst,
|
Btrfs: fix file corruption and data loss after cloning inline extents
Currently the clone ioctl allows to clone an inline extent from one file
to another that already has other (non-inlined) extents. This is a problem
because btrfs is not designed to deal with files having inline and regular
extents, if a file has an inline extent then it must be the only extent
in the file and must start at file offset 0. Having a file with an inline
extent followed by regular extents results in EIO errors when doing reads
or writes against the first 4K of the file.
Also, the clone ioctl allows one to lose data if the source file consists
of a single inline extent, with a size of N bytes, and the destination
file consists of a single inline extent with a size of M bytes, where we
have M > N. In this case the clone operation removes the inline extent
from the destination file and then copies the inline extent from the
source file into the destination file - we lose the M - N bytes from the
destination file, a read operation will get the value 0x00 for any bytes
in the the range [N, M] (the destination inode's i_size remained as M,
that's why we can read past N bytes).
So fix this by not allowing such destructive operations to happen and
return errno EOPNOTSUPP to user space.
Currently the fstest btrfs/035 tests the data loss case but it totally
ignores this - i.e. expects the operation to succeed and does not check
the we got data loss.
The following test case for fstests exercises all these cases that result
in file corruption and data loss:
seq=`basename $0`
seqres=$RESULT_DIR/$seq
echo "QA output created by $seq"
tmp=/tmp/$$
status=1 # failure is the default!
trap "_cleanup; exit \$status" 0 1 2 3 15
_cleanup()
{
rm -f $tmp.*
}
# get standard environment, filters and checks
. ./common/rc
. ./common/filter
# real QA test starts here
_need_to_be_root
_supported_fs btrfs
_supported_os Linux
_require_scratch
_require_cloner
_require_btrfs_fs_feature "no_holes"
_require_btrfs_mkfs_feature "no-holes"
rm -f $seqres.full
test_cloning_inline_extents()
{
local mkfs_opts=$1
local mount_opts=$2
_scratch_mkfs $mkfs_opts >>$seqres.full 2>&1
_scratch_mount $mount_opts
# File bar, the source for all the following clone operations, consists
# of a single inline extent (50 bytes).
$XFS_IO_PROG -f -c "pwrite -S 0xbb 0 50" $SCRATCH_MNT/bar \
| _filter_xfs_io
# Test cloning into a file with an extent (non-inlined) where the
# destination offset overlaps that extent. It should not be possible to
# clone the inline extent from file bar into this file.
$XFS_IO_PROG -f -c "pwrite -S 0xaa 0K 16K" $SCRATCH_MNT/foo \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo
# Doing IO against any range in the first 4K of the file should work.
# Due to a past clone ioctl bug which allowed cloning the inline extent,
# these operations resulted in EIO errors.
echo "File foo data after clone operation:"
# All bytes should have the value 0xaa (clone operation failed and did
# not modify our file).
od -t x1 $SCRATCH_MNT/foo
$XFS_IO_PROG -c "pwrite -S 0xcc 0 100" $SCRATCH_MNT/foo | _filter_xfs_io
# Test cloning the inline extent against a file which has a hole in its
# first 4K followed by a non-inlined extent. It should not be possible
# as well to clone the inline extent from file bar into this file.
$XFS_IO_PROG -f -c "pwrite -S 0xdd 4K 12K" $SCRATCH_MNT/foo2 \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo2
# Doing IO against any range in the first 4K of the file should work.
# Due to a past clone ioctl bug which allowed cloning the inline extent,
# these operations resulted in EIO errors.
echo "File foo2 data after clone operation:"
# All bytes should have the value 0x00 (clone operation failed and did
# not modify our file).
od -t x1 $SCRATCH_MNT/foo2
$XFS_IO_PROG -c "pwrite -S 0xee 0 90" $SCRATCH_MNT/foo2 | _filter_xfs_io
# Test cloning the inline extent against a file which has a size of zero
# but has a prealloc extent. It should not be possible as well to clone
# the inline extent from file bar into this file.
$XFS_IO_PROG -f -c "falloc -k 0 1M" $SCRATCH_MNT/foo3 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo3
# Doing IO against any range in the first 4K of the file should work.
# Due to a past clone ioctl bug which allowed cloning the inline extent,
# these operations resulted in EIO errors.
echo "First 50 bytes of foo3 after clone operation:"
# Should not be able to read any bytes, file has 0 bytes i_size (the
# clone operation failed and did not modify our file).
od -t x1 $SCRATCH_MNT/foo3
$XFS_IO_PROG -c "pwrite -S 0xff 0 90" $SCRATCH_MNT/foo3 | _filter_xfs_io
# Test cloning the inline extent against a file which consists of a
# single inline extent that has a size not greater than the size of
# bar's inline extent (40 < 50).
# It should be possible to do the extent cloning from bar to this file.
$XFS_IO_PROG -f -c "pwrite -S 0x01 0 40" $SCRATCH_MNT/foo4 \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo4
# Doing IO against any range in the first 4K of the file should work.
echo "File foo4 data after clone operation:"
# Must match file bar's content.
od -t x1 $SCRATCH_MNT/foo4
$XFS_IO_PROG -c "pwrite -S 0x02 0 90" $SCRATCH_MNT/foo4 | _filter_xfs_io
# Test cloning the inline extent against a file which consists of a
# single inline extent that has a size greater than the size of bar's
# inline extent (60 > 50).
# It should not be possible to clone the inline extent from file bar
# into this file.
$XFS_IO_PROG -f -c "pwrite -S 0x03 0 60" $SCRATCH_MNT/foo5 \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo5
# Reading the file should not fail.
echo "File foo5 data after clone operation:"
# Must have a size of 60 bytes, with all bytes having a value of 0x03
# (the clone operation failed and did not modify our file).
od -t x1 $SCRATCH_MNT/foo5
# Test cloning the inline extent against a file which has no extents but
# has a size greater than bar's inline extent (16K > 50).
# It should not be possible to clone the inline extent from file bar
# into this file.
$XFS_IO_PROG -f -c "truncate 16K" $SCRATCH_MNT/foo6 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo6
# Reading the file should not fail.
echo "File foo6 data after clone operation:"
# Must have a size of 16K, with all bytes having a value of 0x00 (the
# clone operation failed and did not modify our file).
od -t x1 $SCRATCH_MNT/foo6
# Test cloning the inline extent against a file which has no extents but
# has a size not greater than bar's inline extent (30 < 50).
# It should be possible to clone the inline extent from file bar into
# this file.
$XFS_IO_PROG -f -c "truncate 30" $SCRATCH_MNT/foo7 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo7
# Reading the file should not fail.
echo "File foo7 data after clone operation:"
# Must have a size of 50 bytes, with all bytes having a value of 0xbb.
od -t x1 $SCRATCH_MNT/foo7
# Test cloning the inline extent against a file which has a size not
# greater than the size of bar's inline extent (20 < 50) but has
# a prealloc extent that goes beyond the file's size. It should not be
# possible to clone the inline extent from bar into this file.
$XFS_IO_PROG -f -c "falloc -k 0 1M" \
-c "pwrite -S 0x88 0 20" \
$SCRATCH_MNT/foo8 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo8
echo "File foo8 data after clone operation:"
# Must have a size of 20 bytes, with all bytes having a value of 0x88
# (the clone operation did not modify our file).
od -t x1 $SCRATCH_MNT/foo8
_scratch_unmount
}
echo -e "\nTesting without compression and without the no-holes feature...\n"
test_cloning_inline_extents
echo -e "\nTesting with compression and without the no-holes feature...\n"
test_cloning_inline_extents "" "-o compress"
echo -e "\nTesting without compression and with the no-holes feature...\n"
test_cloning_inline_extents "-O no-holes" ""
echo -e "\nTesting with compression and with the no-holes feature...\n"
test_cloning_inline_extents "-O no-holes" "-o compress"
status=0
exit
Cc: stable@vger.kernel.org
Signed-off-by: Filipe Manana <fdmanana@suse.com>
2015-10-13 21:15:00 +07:00
|
|
|
struct btrfs_trans_handle *trans,
|
|
|
|
|
struct btrfs_path *path,
|
|
|
|
|
struct btrfs_key *new_key,
|
|
|
|
|
const u64 drop_start,
|
|
|
|
|
const u64 datal,
|
|
|
|
|
const u64 skip,
|
|
|
|
|
const u64 size,
|
|
|
|
|
char *inline_data)
|
|
|
|
|
{
|
2016-06-23 05:54:23 +07:00
|
|
|
struct btrfs_fs_info *fs_info = btrfs_sb(dst->i_sb);
|
Btrfs: fix file corruption and data loss after cloning inline extents
Currently the clone ioctl allows to clone an inline extent from one file
to another that already has other (non-inlined) extents. This is a problem
because btrfs is not designed to deal with files having inline and regular
extents, if a file has an inline extent then it must be the only extent
in the file and must start at file offset 0. Having a file with an inline
extent followed by regular extents results in EIO errors when doing reads
or writes against the first 4K of the file.
Also, the clone ioctl allows one to lose data if the source file consists
of a single inline extent, with a size of N bytes, and the destination
file consists of a single inline extent with a size of M bytes, where we
have M > N. In this case the clone operation removes the inline extent
from the destination file and then copies the inline extent from the
source file into the destination file - we lose the M - N bytes from the
destination file, a read operation will get the value 0x00 for any bytes
in the the range [N, M] (the destination inode's i_size remained as M,
that's why we can read past N bytes).
So fix this by not allowing such destructive operations to happen and
return errno EOPNOTSUPP to user space.
Currently the fstest btrfs/035 tests the data loss case but it totally
ignores this - i.e. expects the operation to succeed and does not check
the we got data loss.
The following test case for fstests exercises all these cases that result
in file corruption and data loss:
seq=`basename $0`
seqres=$RESULT_DIR/$seq
echo "QA output created by $seq"
tmp=/tmp/$$
status=1 # failure is the default!
trap "_cleanup; exit \$status" 0 1 2 3 15
_cleanup()
{
rm -f $tmp.*
}
# get standard environment, filters and checks
. ./common/rc
. ./common/filter
# real QA test starts here
_need_to_be_root
_supported_fs btrfs
_supported_os Linux
_require_scratch
_require_cloner
_require_btrfs_fs_feature "no_holes"
_require_btrfs_mkfs_feature "no-holes"
rm -f $seqres.full
test_cloning_inline_extents()
{
local mkfs_opts=$1
local mount_opts=$2
_scratch_mkfs $mkfs_opts >>$seqres.full 2>&1
_scratch_mount $mount_opts
# File bar, the source for all the following clone operations, consists
# of a single inline extent (50 bytes).
$XFS_IO_PROG -f -c "pwrite -S 0xbb 0 50" $SCRATCH_MNT/bar \
| _filter_xfs_io
# Test cloning into a file with an extent (non-inlined) where the
# destination offset overlaps that extent. It should not be possible to
# clone the inline extent from file bar into this file.
$XFS_IO_PROG -f -c "pwrite -S 0xaa 0K 16K" $SCRATCH_MNT/foo \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo
# Doing IO against any range in the first 4K of the file should work.
# Due to a past clone ioctl bug which allowed cloning the inline extent,
# these operations resulted in EIO errors.
echo "File foo data after clone operation:"
# All bytes should have the value 0xaa (clone operation failed and did
# not modify our file).
od -t x1 $SCRATCH_MNT/foo
$XFS_IO_PROG -c "pwrite -S 0xcc 0 100" $SCRATCH_MNT/foo | _filter_xfs_io
# Test cloning the inline extent against a file which has a hole in its
# first 4K followed by a non-inlined extent. It should not be possible
# as well to clone the inline extent from file bar into this file.
$XFS_IO_PROG -f -c "pwrite -S 0xdd 4K 12K" $SCRATCH_MNT/foo2 \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo2
# Doing IO against any range in the first 4K of the file should work.
# Due to a past clone ioctl bug which allowed cloning the inline extent,
# these operations resulted in EIO errors.
echo "File foo2 data after clone operation:"
# All bytes should have the value 0x00 (clone operation failed and did
# not modify our file).
od -t x1 $SCRATCH_MNT/foo2
$XFS_IO_PROG -c "pwrite -S 0xee 0 90" $SCRATCH_MNT/foo2 | _filter_xfs_io
# Test cloning the inline extent against a file which has a size of zero
# but has a prealloc extent. It should not be possible as well to clone
# the inline extent from file bar into this file.
$XFS_IO_PROG -f -c "falloc -k 0 1M" $SCRATCH_MNT/foo3 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo3
# Doing IO against any range in the first 4K of the file should work.
# Due to a past clone ioctl bug which allowed cloning the inline extent,
# these operations resulted in EIO errors.
echo "First 50 bytes of foo3 after clone operation:"
# Should not be able to read any bytes, file has 0 bytes i_size (the
# clone operation failed and did not modify our file).
od -t x1 $SCRATCH_MNT/foo3
$XFS_IO_PROG -c "pwrite -S 0xff 0 90" $SCRATCH_MNT/foo3 | _filter_xfs_io
# Test cloning the inline extent against a file which consists of a
# single inline extent that has a size not greater than the size of
# bar's inline extent (40 < 50).
# It should be possible to do the extent cloning from bar to this file.
$XFS_IO_PROG -f -c "pwrite -S 0x01 0 40" $SCRATCH_MNT/foo4 \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo4
# Doing IO against any range in the first 4K of the file should work.
echo "File foo4 data after clone operation:"
# Must match file bar's content.
od -t x1 $SCRATCH_MNT/foo4
$XFS_IO_PROG -c "pwrite -S 0x02 0 90" $SCRATCH_MNT/foo4 | _filter_xfs_io
# Test cloning the inline extent against a file which consists of a
# single inline extent that has a size greater than the size of bar's
# inline extent (60 > 50).
# It should not be possible to clone the inline extent from file bar
# into this file.
$XFS_IO_PROG -f -c "pwrite -S 0x03 0 60" $SCRATCH_MNT/foo5 \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo5
# Reading the file should not fail.
echo "File foo5 data after clone operation:"
# Must have a size of 60 bytes, with all bytes having a value of 0x03
# (the clone operation failed and did not modify our file).
od -t x1 $SCRATCH_MNT/foo5
# Test cloning the inline extent against a file which has no extents but
# has a size greater than bar's inline extent (16K > 50).
# It should not be possible to clone the inline extent from file bar
# into this file.
$XFS_IO_PROG -f -c "truncate 16K" $SCRATCH_MNT/foo6 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo6
# Reading the file should not fail.
echo "File foo6 data after clone operation:"
# Must have a size of 16K, with all bytes having a value of 0x00 (the
# clone operation failed and did not modify our file).
od -t x1 $SCRATCH_MNT/foo6
# Test cloning the inline extent against a file which has no extents but
# has a size not greater than bar's inline extent (30 < 50).
# It should be possible to clone the inline extent from file bar into
# this file.
$XFS_IO_PROG -f -c "truncate 30" $SCRATCH_MNT/foo7 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo7
# Reading the file should not fail.
echo "File foo7 data after clone operation:"
# Must have a size of 50 bytes, with all bytes having a value of 0xbb.
od -t x1 $SCRATCH_MNT/foo7
# Test cloning the inline extent against a file which has a size not
# greater than the size of bar's inline extent (20 < 50) but has
# a prealloc extent that goes beyond the file's size. It should not be
# possible to clone the inline extent from bar into this file.
$XFS_IO_PROG -f -c "falloc -k 0 1M" \
-c "pwrite -S 0x88 0 20" \
$SCRATCH_MNT/foo8 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo8
echo "File foo8 data after clone operation:"
# Must have a size of 20 bytes, with all bytes having a value of 0x88
# (the clone operation did not modify our file).
od -t x1 $SCRATCH_MNT/foo8
_scratch_unmount
}
echo -e "\nTesting without compression and without the no-holes feature...\n"
test_cloning_inline_extents
echo -e "\nTesting with compression and without the no-holes feature...\n"
test_cloning_inline_extents "" "-o compress"
echo -e "\nTesting without compression and with the no-holes feature...\n"
test_cloning_inline_extents "-O no-holes" ""
echo -e "\nTesting with compression and with the no-holes feature...\n"
test_cloning_inline_extents "-O no-holes" "-o compress"
status=0
exit
Cc: stable@vger.kernel.org
Signed-off-by: Filipe Manana <fdmanana@suse.com>
2015-10-13 21:15:00 +07:00
|
|
|
struct btrfs_root *root = BTRFS_I(dst)->root;
|
|
|
|
|
const u64 aligned_end = ALIGN(new_key->offset + datal,
|
2016-06-23 05:54:23 +07:00
|
|
|
fs_info->sectorsize);
|
Btrfs: fix file corruption and data loss after cloning inline extents
Currently the clone ioctl allows to clone an inline extent from one file
to another that already has other (non-inlined) extents. This is a problem
because btrfs is not designed to deal with files having inline and regular
extents, if a file has an inline extent then it must be the only extent
in the file and must start at file offset 0. Having a file with an inline
extent followed by regular extents results in EIO errors when doing reads
or writes against the first 4K of the file.
Also, the clone ioctl allows one to lose data if the source file consists
of a single inline extent, with a size of N bytes, and the destination
file consists of a single inline extent with a size of M bytes, where we
have M > N. In this case the clone operation removes the inline extent
from the destination file and then copies the inline extent from the
source file into the destination file - we lose the M - N bytes from the
destination file, a read operation will get the value 0x00 for any bytes
in the the range [N, M] (the destination inode's i_size remained as M,
that's why we can read past N bytes).
So fix this by not allowing such destructive operations to happen and
return errno EOPNOTSUPP to user space.
Currently the fstest btrfs/035 tests the data loss case but it totally
ignores this - i.e. expects the operation to succeed and does not check
the we got data loss.
The following test case for fstests exercises all these cases that result
in file corruption and data loss:
seq=`basename $0`
seqres=$RESULT_DIR/$seq
echo "QA output created by $seq"
tmp=/tmp/$$
status=1 # failure is the default!
trap "_cleanup; exit \$status" 0 1 2 3 15
_cleanup()
{
rm -f $tmp.*
}
# get standard environment, filters and checks
. ./common/rc
. ./common/filter
# real QA test starts here
_need_to_be_root
_supported_fs btrfs
_supported_os Linux
_require_scratch
_require_cloner
_require_btrfs_fs_feature "no_holes"
_require_btrfs_mkfs_feature "no-holes"
rm -f $seqres.full
test_cloning_inline_extents()
{
local mkfs_opts=$1
local mount_opts=$2
_scratch_mkfs $mkfs_opts >>$seqres.full 2>&1
_scratch_mount $mount_opts
# File bar, the source for all the following clone operations, consists
# of a single inline extent (50 bytes).
$XFS_IO_PROG -f -c "pwrite -S 0xbb 0 50" $SCRATCH_MNT/bar \
| _filter_xfs_io
# Test cloning into a file with an extent (non-inlined) where the
# destination offset overlaps that extent. It should not be possible to
# clone the inline extent from file bar into this file.
$XFS_IO_PROG -f -c "pwrite -S 0xaa 0K 16K" $SCRATCH_MNT/foo \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo
# Doing IO against any range in the first 4K of the file should work.
# Due to a past clone ioctl bug which allowed cloning the inline extent,
# these operations resulted in EIO errors.
echo "File foo data after clone operation:"
# All bytes should have the value 0xaa (clone operation failed and did
# not modify our file).
od -t x1 $SCRATCH_MNT/foo
$XFS_IO_PROG -c "pwrite -S 0xcc 0 100" $SCRATCH_MNT/foo | _filter_xfs_io
# Test cloning the inline extent against a file which has a hole in its
# first 4K followed by a non-inlined extent. It should not be possible
# as well to clone the inline extent from file bar into this file.
$XFS_IO_PROG -f -c "pwrite -S 0xdd 4K 12K" $SCRATCH_MNT/foo2 \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo2
# Doing IO against any range in the first 4K of the file should work.
# Due to a past clone ioctl bug which allowed cloning the inline extent,
# these operations resulted in EIO errors.
echo "File foo2 data after clone operation:"
# All bytes should have the value 0x00 (clone operation failed and did
# not modify our file).
od -t x1 $SCRATCH_MNT/foo2
$XFS_IO_PROG -c "pwrite -S 0xee 0 90" $SCRATCH_MNT/foo2 | _filter_xfs_io
# Test cloning the inline extent against a file which has a size of zero
# but has a prealloc extent. It should not be possible as well to clone
# the inline extent from file bar into this file.
$XFS_IO_PROG -f -c "falloc -k 0 1M" $SCRATCH_MNT/foo3 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo3
# Doing IO against any range in the first 4K of the file should work.
# Due to a past clone ioctl bug which allowed cloning the inline extent,
# these operations resulted in EIO errors.
echo "First 50 bytes of foo3 after clone operation:"
# Should not be able to read any bytes, file has 0 bytes i_size (the
# clone operation failed and did not modify our file).
od -t x1 $SCRATCH_MNT/foo3
$XFS_IO_PROG -c "pwrite -S 0xff 0 90" $SCRATCH_MNT/foo3 | _filter_xfs_io
# Test cloning the inline extent against a file which consists of a
# single inline extent that has a size not greater than the size of
# bar's inline extent (40 < 50).
# It should be possible to do the extent cloning from bar to this file.
$XFS_IO_PROG -f -c "pwrite -S 0x01 0 40" $SCRATCH_MNT/foo4 \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo4
# Doing IO against any range in the first 4K of the file should work.
echo "File foo4 data after clone operation:"
# Must match file bar's content.
od -t x1 $SCRATCH_MNT/foo4
$XFS_IO_PROG -c "pwrite -S 0x02 0 90" $SCRATCH_MNT/foo4 | _filter_xfs_io
# Test cloning the inline extent against a file which consists of a
# single inline extent that has a size greater than the size of bar's
# inline extent (60 > 50).
# It should not be possible to clone the inline extent from file bar
# into this file.
$XFS_IO_PROG -f -c "pwrite -S 0x03 0 60" $SCRATCH_MNT/foo5 \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo5
# Reading the file should not fail.
echo "File foo5 data after clone operation:"
# Must have a size of 60 bytes, with all bytes having a value of 0x03
# (the clone operation failed and did not modify our file).
od -t x1 $SCRATCH_MNT/foo5
# Test cloning the inline extent against a file which has no extents but
# has a size greater than bar's inline extent (16K > 50).
# It should not be possible to clone the inline extent from file bar
# into this file.
$XFS_IO_PROG -f -c "truncate 16K" $SCRATCH_MNT/foo6 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo6
# Reading the file should not fail.
echo "File foo6 data after clone operation:"
# Must have a size of 16K, with all bytes having a value of 0x00 (the
# clone operation failed and did not modify our file).
od -t x1 $SCRATCH_MNT/foo6
# Test cloning the inline extent against a file which has no extents but
# has a size not greater than bar's inline extent (30 < 50).
# It should be possible to clone the inline extent from file bar into
# this file.
$XFS_IO_PROG -f -c "truncate 30" $SCRATCH_MNT/foo7 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo7
# Reading the file should not fail.
echo "File foo7 data after clone operation:"
# Must have a size of 50 bytes, with all bytes having a value of 0xbb.
od -t x1 $SCRATCH_MNT/foo7
# Test cloning the inline extent against a file which has a size not
# greater than the size of bar's inline extent (20 < 50) but has
# a prealloc extent that goes beyond the file's size. It should not be
# possible to clone the inline extent from bar into this file.
$XFS_IO_PROG -f -c "falloc -k 0 1M" \
-c "pwrite -S 0x88 0 20" \
$SCRATCH_MNT/foo8 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo8
echo "File foo8 data after clone operation:"
# Must have a size of 20 bytes, with all bytes having a value of 0x88
# (the clone operation did not modify our file).
od -t x1 $SCRATCH_MNT/foo8
_scratch_unmount
}
echo -e "\nTesting without compression and without the no-holes feature...\n"
test_cloning_inline_extents
echo -e "\nTesting with compression and without the no-holes feature...\n"
test_cloning_inline_extents "" "-o compress"
echo -e "\nTesting without compression and with the no-holes feature...\n"
test_cloning_inline_extents "-O no-holes" ""
echo -e "\nTesting with compression and with the no-holes feature...\n"
test_cloning_inline_extents "-O no-holes" "-o compress"
status=0
exit
Cc: stable@vger.kernel.org
Signed-off-by: Filipe Manana <fdmanana@suse.com>
2015-10-13 21:15:00 +07:00
|
|
|
int ret;
|
|
|
|
|
struct btrfs_key key;
|
|
|
|
|
|
|
|
|
|
if (new_key->offset > 0)
|
|
|
|
|
return -EOPNOTSUPP;
|
|
|
|
|
|
2017-01-11 01:35:31 +07:00
|
|
|
key.objectid = btrfs_ino(BTRFS_I(dst));
|
Btrfs: fix file corruption and data loss after cloning inline extents
Currently the clone ioctl allows to clone an inline extent from one file
to another that already has other (non-inlined) extents. This is a problem
because btrfs is not designed to deal with files having inline and regular
extents, if a file has an inline extent then it must be the only extent
in the file and must start at file offset 0. Having a file with an inline
extent followed by regular extents results in EIO errors when doing reads
or writes against the first 4K of the file.
Also, the clone ioctl allows one to lose data if the source file consists
of a single inline extent, with a size of N bytes, and the destination
file consists of a single inline extent with a size of M bytes, where we
have M > N. In this case the clone operation removes the inline extent
from the destination file and then copies the inline extent from the
source file into the destination file - we lose the M - N bytes from the
destination file, a read operation will get the value 0x00 for any bytes
in the the range [N, M] (the destination inode's i_size remained as M,
that's why we can read past N bytes).
So fix this by not allowing such destructive operations to happen and
return errno EOPNOTSUPP to user space.
Currently the fstest btrfs/035 tests the data loss case but it totally
ignores this - i.e. expects the operation to succeed and does not check
the we got data loss.
The following test case for fstests exercises all these cases that result
in file corruption and data loss:
seq=`basename $0`
seqres=$RESULT_DIR/$seq
echo "QA output created by $seq"
tmp=/tmp/$$
status=1 # failure is the default!
trap "_cleanup; exit \$status" 0 1 2 3 15
_cleanup()
{
rm -f $tmp.*
}
# get standard environment, filters and checks
. ./common/rc
. ./common/filter
# real QA test starts here
_need_to_be_root
_supported_fs btrfs
_supported_os Linux
_require_scratch
_require_cloner
_require_btrfs_fs_feature "no_holes"
_require_btrfs_mkfs_feature "no-holes"
rm -f $seqres.full
test_cloning_inline_extents()
{
local mkfs_opts=$1
local mount_opts=$2
_scratch_mkfs $mkfs_opts >>$seqres.full 2>&1
_scratch_mount $mount_opts
# File bar, the source for all the following clone operations, consists
# of a single inline extent (50 bytes).
$XFS_IO_PROG -f -c "pwrite -S 0xbb 0 50" $SCRATCH_MNT/bar \
| _filter_xfs_io
# Test cloning into a file with an extent (non-inlined) where the
# destination offset overlaps that extent. It should not be possible to
# clone the inline extent from file bar into this file.
$XFS_IO_PROG -f -c "pwrite -S 0xaa 0K 16K" $SCRATCH_MNT/foo \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo
# Doing IO against any range in the first 4K of the file should work.
# Due to a past clone ioctl bug which allowed cloning the inline extent,
# these operations resulted in EIO errors.
echo "File foo data after clone operation:"
# All bytes should have the value 0xaa (clone operation failed and did
# not modify our file).
od -t x1 $SCRATCH_MNT/foo
$XFS_IO_PROG -c "pwrite -S 0xcc 0 100" $SCRATCH_MNT/foo | _filter_xfs_io
# Test cloning the inline extent against a file which has a hole in its
# first 4K followed by a non-inlined extent. It should not be possible
# as well to clone the inline extent from file bar into this file.
$XFS_IO_PROG -f -c "pwrite -S 0xdd 4K 12K" $SCRATCH_MNT/foo2 \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo2
# Doing IO against any range in the first 4K of the file should work.
# Due to a past clone ioctl bug which allowed cloning the inline extent,
# these operations resulted in EIO errors.
echo "File foo2 data after clone operation:"
# All bytes should have the value 0x00 (clone operation failed and did
# not modify our file).
od -t x1 $SCRATCH_MNT/foo2
$XFS_IO_PROG -c "pwrite -S 0xee 0 90" $SCRATCH_MNT/foo2 | _filter_xfs_io
# Test cloning the inline extent against a file which has a size of zero
# but has a prealloc extent. It should not be possible as well to clone
# the inline extent from file bar into this file.
$XFS_IO_PROG -f -c "falloc -k 0 1M" $SCRATCH_MNT/foo3 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo3
# Doing IO against any range in the first 4K of the file should work.
# Due to a past clone ioctl bug which allowed cloning the inline extent,
# these operations resulted in EIO errors.
echo "First 50 bytes of foo3 after clone operation:"
# Should not be able to read any bytes, file has 0 bytes i_size (the
# clone operation failed and did not modify our file).
od -t x1 $SCRATCH_MNT/foo3
$XFS_IO_PROG -c "pwrite -S 0xff 0 90" $SCRATCH_MNT/foo3 | _filter_xfs_io
# Test cloning the inline extent against a file which consists of a
# single inline extent that has a size not greater than the size of
# bar's inline extent (40 < 50).
# It should be possible to do the extent cloning from bar to this file.
$XFS_IO_PROG -f -c "pwrite -S 0x01 0 40" $SCRATCH_MNT/foo4 \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo4
# Doing IO against any range in the first 4K of the file should work.
echo "File foo4 data after clone operation:"
# Must match file bar's content.
od -t x1 $SCRATCH_MNT/foo4
$XFS_IO_PROG -c "pwrite -S 0x02 0 90" $SCRATCH_MNT/foo4 | _filter_xfs_io
# Test cloning the inline extent against a file which consists of a
# single inline extent that has a size greater than the size of bar's
# inline extent (60 > 50).
# It should not be possible to clone the inline extent from file bar
# into this file.
$XFS_IO_PROG -f -c "pwrite -S 0x03 0 60" $SCRATCH_MNT/foo5 \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo5
# Reading the file should not fail.
echo "File foo5 data after clone operation:"
# Must have a size of 60 bytes, with all bytes having a value of 0x03
# (the clone operation failed and did not modify our file).
od -t x1 $SCRATCH_MNT/foo5
# Test cloning the inline extent against a file which has no extents but
# has a size greater than bar's inline extent (16K > 50).
# It should not be possible to clone the inline extent from file bar
# into this file.
$XFS_IO_PROG -f -c "truncate 16K" $SCRATCH_MNT/foo6 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo6
# Reading the file should not fail.
echo "File foo6 data after clone operation:"
# Must have a size of 16K, with all bytes having a value of 0x00 (the
# clone operation failed and did not modify our file).
od -t x1 $SCRATCH_MNT/foo6
# Test cloning the inline extent against a file which has no extents but
# has a size not greater than bar's inline extent (30 < 50).
# It should be possible to clone the inline extent from file bar into
# this file.
$XFS_IO_PROG -f -c "truncate 30" $SCRATCH_MNT/foo7 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo7
# Reading the file should not fail.
echo "File foo7 data after clone operation:"
# Must have a size of 50 bytes, with all bytes having a value of 0xbb.
od -t x1 $SCRATCH_MNT/foo7
# Test cloning the inline extent against a file which has a size not
# greater than the size of bar's inline extent (20 < 50) but has
# a prealloc extent that goes beyond the file's size. It should not be
# possible to clone the inline extent from bar into this file.
$XFS_IO_PROG -f -c "falloc -k 0 1M" \
-c "pwrite -S 0x88 0 20" \
$SCRATCH_MNT/foo8 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo8
echo "File foo8 data after clone operation:"
# Must have a size of 20 bytes, with all bytes having a value of 0x88
# (the clone operation did not modify our file).
od -t x1 $SCRATCH_MNT/foo8
_scratch_unmount
}
echo -e "\nTesting without compression and without the no-holes feature...\n"
test_cloning_inline_extents
echo -e "\nTesting with compression and without the no-holes feature...\n"
test_cloning_inline_extents "" "-o compress"
echo -e "\nTesting without compression and with the no-holes feature...\n"
test_cloning_inline_extents "-O no-holes" ""
echo -e "\nTesting with compression and with the no-holes feature...\n"
test_cloning_inline_extents "-O no-holes" "-o compress"
status=0
exit
Cc: stable@vger.kernel.org
Signed-off-by: Filipe Manana <fdmanana@suse.com>
2015-10-13 21:15:00 +07:00
|
|
|
key.type = BTRFS_EXTENT_DATA_KEY;
|
|
|
|
|
key.offset = 0;
|
|
|
|
|
ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
|
|
|
|
|
if (ret < 0) {
|
|
|
|
|
return ret;
|
|
|
|
|
} else if (ret > 0) {
|
|
|
|
|
if (path->slots[0] >= btrfs_header_nritems(path->nodes[0])) {
|
|
|
|
|
ret = btrfs_next_leaf(root, path);
|
|
|
|
|
if (ret < 0)
|
|
|
|
|
return ret;
|
|
|
|
|
else if (ret > 0)
|
|
|
|
|
goto copy_inline_extent;
|
|
|
|
|
}
|
|
|
|
|
btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
|
2017-01-11 01:35:31 +07:00
|
|
|
if (key.objectid == btrfs_ino(BTRFS_I(dst)) &&
|
Btrfs: fix file corruption and data loss after cloning inline extents
Currently the clone ioctl allows to clone an inline extent from one file
to another that already has other (non-inlined) extents. This is a problem
because btrfs is not designed to deal with files having inline and regular
extents, if a file has an inline extent then it must be the only extent
in the file and must start at file offset 0. Having a file with an inline
extent followed by regular extents results in EIO errors when doing reads
or writes against the first 4K of the file.
Also, the clone ioctl allows one to lose data if the source file consists
of a single inline extent, with a size of N bytes, and the destination
file consists of a single inline extent with a size of M bytes, where we
have M > N. In this case the clone operation removes the inline extent
from the destination file and then copies the inline extent from the
source file into the destination file - we lose the M - N bytes from the
destination file, a read operation will get the value 0x00 for any bytes
in the the range [N, M] (the destination inode's i_size remained as M,
that's why we can read past N bytes).
So fix this by not allowing such destructive operations to happen and
return errno EOPNOTSUPP to user space.
Currently the fstest btrfs/035 tests the data loss case but it totally
ignores this - i.e. expects the operation to succeed and does not check
the we got data loss.
The following test case for fstests exercises all these cases that result
in file corruption and data loss:
seq=`basename $0`
seqres=$RESULT_DIR/$seq
echo "QA output created by $seq"
tmp=/tmp/$$
status=1 # failure is the default!
trap "_cleanup; exit \$status" 0 1 2 3 15
_cleanup()
{
rm -f $tmp.*
}
# get standard environment, filters and checks
. ./common/rc
. ./common/filter
# real QA test starts here
_need_to_be_root
_supported_fs btrfs
_supported_os Linux
_require_scratch
_require_cloner
_require_btrfs_fs_feature "no_holes"
_require_btrfs_mkfs_feature "no-holes"
rm -f $seqres.full
test_cloning_inline_extents()
{
local mkfs_opts=$1
local mount_opts=$2
_scratch_mkfs $mkfs_opts >>$seqres.full 2>&1
_scratch_mount $mount_opts
# File bar, the source for all the following clone operations, consists
# of a single inline extent (50 bytes).
$XFS_IO_PROG -f -c "pwrite -S 0xbb 0 50" $SCRATCH_MNT/bar \
| _filter_xfs_io
# Test cloning into a file with an extent (non-inlined) where the
# destination offset overlaps that extent. It should not be possible to
# clone the inline extent from file bar into this file.
$XFS_IO_PROG -f -c "pwrite -S 0xaa 0K 16K" $SCRATCH_MNT/foo \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo
# Doing IO against any range in the first 4K of the file should work.
# Due to a past clone ioctl bug which allowed cloning the inline extent,
# these operations resulted in EIO errors.
echo "File foo data after clone operation:"
# All bytes should have the value 0xaa (clone operation failed and did
# not modify our file).
od -t x1 $SCRATCH_MNT/foo
$XFS_IO_PROG -c "pwrite -S 0xcc 0 100" $SCRATCH_MNT/foo | _filter_xfs_io
# Test cloning the inline extent against a file which has a hole in its
# first 4K followed by a non-inlined extent. It should not be possible
# as well to clone the inline extent from file bar into this file.
$XFS_IO_PROG -f -c "pwrite -S 0xdd 4K 12K" $SCRATCH_MNT/foo2 \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo2
# Doing IO against any range in the first 4K of the file should work.
# Due to a past clone ioctl bug which allowed cloning the inline extent,
# these operations resulted in EIO errors.
echo "File foo2 data after clone operation:"
# All bytes should have the value 0x00 (clone operation failed and did
# not modify our file).
od -t x1 $SCRATCH_MNT/foo2
$XFS_IO_PROG -c "pwrite -S 0xee 0 90" $SCRATCH_MNT/foo2 | _filter_xfs_io
# Test cloning the inline extent against a file which has a size of zero
# but has a prealloc extent. It should not be possible as well to clone
# the inline extent from file bar into this file.
$XFS_IO_PROG -f -c "falloc -k 0 1M" $SCRATCH_MNT/foo3 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo3
# Doing IO against any range in the first 4K of the file should work.
# Due to a past clone ioctl bug which allowed cloning the inline extent,
# these operations resulted in EIO errors.
echo "First 50 bytes of foo3 after clone operation:"
# Should not be able to read any bytes, file has 0 bytes i_size (the
# clone operation failed and did not modify our file).
od -t x1 $SCRATCH_MNT/foo3
$XFS_IO_PROG -c "pwrite -S 0xff 0 90" $SCRATCH_MNT/foo3 | _filter_xfs_io
# Test cloning the inline extent against a file which consists of a
# single inline extent that has a size not greater than the size of
# bar's inline extent (40 < 50).
# It should be possible to do the extent cloning from bar to this file.
$XFS_IO_PROG -f -c "pwrite -S 0x01 0 40" $SCRATCH_MNT/foo4 \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo4
# Doing IO against any range in the first 4K of the file should work.
echo "File foo4 data after clone operation:"
# Must match file bar's content.
od -t x1 $SCRATCH_MNT/foo4
$XFS_IO_PROG -c "pwrite -S 0x02 0 90" $SCRATCH_MNT/foo4 | _filter_xfs_io
# Test cloning the inline extent against a file which consists of a
# single inline extent that has a size greater than the size of bar's
# inline extent (60 > 50).
# It should not be possible to clone the inline extent from file bar
# into this file.
$XFS_IO_PROG -f -c "pwrite -S 0x03 0 60" $SCRATCH_MNT/foo5 \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo5
# Reading the file should not fail.
echo "File foo5 data after clone operation:"
# Must have a size of 60 bytes, with all bytes having a value of 0x03
# (the clone operation failed and did not modify our file).
od -t x1 $SCRATCH_MNT/foo5
# Test cloning the inline extent against a file which has no extents but
# has a size greater than bar's inline extent (16K > 50).
# It should not be possible to clone the inline extent from file bar
# into this file.
$XFS_IO_PROG -f -c "truncate 16K" $SCRATCH_MNT/foo6 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo6
# Reading the file should not fail.
echo "File foo6 data after clone operation:"
# Must have a size of 16K, with all bytes having a value of 0x00 (the
# clone operation failed and did not modify our file).
od -t x1 $SCRATCH_MNT/foo6
# Test cloning the inline extent against a file which has no extents but
# has a size not greater than bar's inline extent (30 < 50).
# It should be possible to clone the inline extent from file bar into
# this file.
$XFS_IO_PROG -f -c "truncate 30" $SCRATCH_MNT/foo7 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo7
# Reading the file should not fail.
echo "File foo7 data after clone operation:"
# Must have a size of 50 bytes, with all bytes having a value of 0xbb.
od -t x1 $SCRATCH_MNT/foo7
# Test cloning the inline extent against a file which has a size not
# greater than the size of bar's inline extent (20 < 50) but has
# a prealloc extent that goes beyond the file's size. It should not be
# possible to clone the inline extent from bar into this file.
$XFS_IO_PROG -f -c "falloc -k 0 1M" \
-c "pwrite -S 0x88 0 20" \
$SCRATCH_MNT/foo8 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo8
echo "File foo8 data after clone operation:"
# Must have a size of 20 bytes, with all bytes having a value of 0x88
# (the clone operation did not modify our file).
od -t x1 $SCRATCH_MNT/foo8
_scratch_unmount
}
echo -e "\nTesting without compression and without the no-holes feature...\n"
test_cloning_inline_extents
echo -e "\nTesting with compression and without the no-holes feature...\n"
test_cloning_inline_extents "" "-o compress"
echo -e "\nTesting without compression and with the no-holes feature...\n"
test_cloning_inline_extents "-O no-holes" ""
echo -e "\nTesting with compression and with the no-holes feature...\n"
test_cloning_inline_extents "-O no-holes" "-o compress"
status=0
exit
Cc: stable@vger.kernel.org
Signed-off-by: Filipe Manana <fdmanana@suse.com>
2015-10-13 21:15:00 +07:00
|
|
|
key.type == BTRFS_EXTENT_DATA_KEY) {
|
|
|
|
|
ASSERT(key.offset > 0);
|
|
|
|
|
return -EOPNOTSUPP;
|
|
|
|
|
}
|
|
|
|
|
} else if (i_size_read(dst) <= datal) {
|
|
|
|
|
struct btrfs_file_extent_item *ei;
|
|
|
|
|
u64 ext_len;
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* If the file size is <= datal, make sure there are no other
|
|
|
|
|
* extents following (can happen do to an fallocate call with
|
|
|
|
|
* the flag FALLOC_FL_KEEP_SIZE).
|
|
|
|
|
*/
|
|
|
|
|
ei = btrfs_item_ptr(path->nodes[0], path->slots[0],
|
|
|
|
|
struct btrfs_file_extent_item);
|
|
|
|
|
/*
|
|
|
|
|
* If it's an inline extent, it can not have other extents
|
|
|
|
|
* following it.
|
|
|
|
|
*/
|
|
|
|
|
if (btrfs_file_extent_type(path->nodes[0], ei) ==
|
|
|
|
|
BTRFS_FILE_EXTENT_INLINE)
|
|
|
|
|
goto copy_inline_extent;
|
|
|
|
|
|
|
|
|
|
ext_len = btrfs_file_extent_num_bytes(path->nodes[0], ei);
|
|
|
|
|
if (ext_len > aligned_end)
|
|
|
|
|
return -EOPNOTSUPP;
|
|
|
|
|
|
|
|
|
|
ret = btrfs_next_item(root, path);
|
|
|
|
|
if (ret < 0) {
|
|
|
|
|
return ret;
|
|
|
|
|
} else if (ret == 0) {
|
|
|
|
|
btrfs_item_key_to_cpu(path->nodes[0], &key,
|
|
|
|
|
path->slots[0]);
|
2017-01-11 01:35:31 +07:00
|
|
|
if (key.objectid == btrfs_ino(BTRFS_I(dst)) &&
|
Btrfs: fix file corruption and data loss after cloning inline extents
Currently the clone ioctl allows to clone an inline extent from one file
to another that already has other (non-inlined) extents. This is a problem
because btrfs is not designed to deal with files having inline and regular
extents, if a file has an inline extent then it must be the only extent
in the file and must start at file offset 0. Having a file with an inline
extent followed by regular extents results in EIO errors when doing reads
or writes against the first 4K of the file.
Also, the clone ioctl allows one to lose data if the source file consists
of a single inline extent, with a size of N bytes, and the destination
file consists of a single inline extent with a size of M bytes, where we
have M > N. In this case the clone operation removes the inline extent
from the destination file and then copies the inline extent from the
source file into the destination file - we lose the M - N bytes from the
destination file, a read operation will get the value 0x00 for any bytes
in the the range [N, M] (the destination inode's i_size remained as M,
that's why we can read past N bytes).
So fix this by not allowing such destructive operations to happen and
return errno EOPNOTSUPP to user space.
Currently the fstest btrfs/035 tests the data loss case but it totally
ignores this - i.e. expects the operation to succeed and does not check
the we got data loss.
The following test case for fstests exercises all these cases that result
in file corruption and data loss:
seq=`basename $0`
seqres=$RESULT_DIR/$seq
echo "QA output created by $seq"
tmp=/tmp/$$
status=1 # failure is the default!
trap "_cleanup; exit \$status" 0 1 2 3 15
_cleanup()
{
rm -f $tmp.*
}
# get standard environment, filters and checks
. ./common/rc
. ./common/filter
# real QA test starts here
_need_to_be_root
_supported_fs btrfs
_supported_os Linux
_require_scratch
_require_cloner
_require_btrfs_fs_feature "no_holes"
_require_btrfs_mkfs_feature "no-holes"
rm -f $seqres.full
test_cloning_inline_extents()
{
local mkfs_opts=$1
local mount_opts=$2
_scratch_mkfs $mkfs_opts >>$seqres.full 2>&1
_scratch_mount $mount_opts
# File bar, the source for all the following clone operations, consists
# of a single inline extent (50 bytes).
$XFS_IO_PROG -f -c "pwrite -S 0xbb 0 50" $SCRATCH_MNT/bar \
| _filter_xfs_io
# Test cloning into a file with an extent (non-inlined) where the
# destination offset overlaps that extent. It should not be possible to
# clone the inline extent from file bar into this file.
$XFS_IO_PROG -f -c "pwrite -S 0xaa 0K 16K" $SCRATCH_MNT/foo \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo
# Doing IO against any range in the first 4K of the file should work.
# Due to a past clone ioctl bug which allowed cloning the inline extent,
# these operations resulted in EIO errors.
echo "File foo data after clone operation:"
# All bytes should have the value 0xaa (clone operation failed and did
# not modify our file).
od -t x1 $SCRATCH_MNT/foo
$XFS_IO_PROG -c "pwrite -S 0xcc 0 100" $SCRATCH_MNT/foo | _filter_xfs_io
# Test cloning the inline extent against a file which has a hole in its
# first 4K followed by a non-inlined extent. It should not be possible
# as well to clone the inline extent from file bar into this file.
$XFS_IO_PROG -f -c "pwrite -S 0xdd 4K 12K" $SCRATCH_MNT/foo2 \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo2
# Doing IO against any range in the first 4K of the file should work.
# Due to a past clone ioctl bug which allowed cloning the inline extent,
# these operations resulted in EIO errors.
echo "File foo2 data after clone operation:"
# All bytes should have the value 0x00 (clone operation failed and did
# not modify our file).
od -t x1 $SCRATCH_MNT/foo2
$XFS_IO_PROG -c "pwrite -S 0xee 0 90" $SCRATCH_MNT/foo2 | _filter_xfs_io
# Test cloning the inline extent against a file which has a size of zero
# but has a prealloc extent. It should not be possible as well to clone
# the inline extent from file bar into this file.
$XFS_IO_PROG -f -c "falloc -k 0 1M" $SCRATCH_MNT/foo3 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo3
# Doing IO against any range in the first 4K of the file should work.
# Due to a past clone ioctl bug which allowed cloning the inline extent,
# these operations resulted in EIO errors.
echo "First 50 bytes of foo3 after clone operation:"
# Should not be able to read any bytes, file has 0 bytes i_size (the
# clone operation failed and did not modify our file).
od -t x1 $SCRATCH_MNT/foo3
$XFS_IO_PROG -c "pwrite -S 0xff 0 90" $SCRATCH_MNT/foo3 | _filter_xfs_io
# Test cloning the inline extent against a file which consists of a
# single inline extent that has a size not greater than the size of
# bar's inline extent (40 < 50).
# It should be possible to do the extent cloning from bar to this file.
$XFS_IO_PROG -f -c "pwrite -S 0x01 0 40" $SCRATCH_MNT/foo4 \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo4
# Doing IO against any range in the first 4K of the file should work.
echo "File foo4 data after clone operation:"
# Must match file bar's content.
od -t x1 $SCRATCH_MNT/foo4
$XFS_IO_PROG -c "pwrite -S 0x02 0 90" $SCRATCH_MNT/foo4 | _filter_xfs_io
# Test cloning the inline extent against a file which consists of a
# single inline extent that has a size greater than the size of bar's
# inline extent (60 > 50).
# It should not be possible to clone the inline extent from file bar
# into this file.
$XFS_IO_PROG -f -c "pwrite -S 0x03 0 60" $SCRATCH_MNT/foo5 \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo5
# Reading the file should not fail.
echo "File foo5 data after clone operation:"
# Must have a size of 60 bytes, with all bytes having a value of 0x03
# (the clone operation failed and did not modify our file).
od -t x1 $SCRATCH_MNT/foo5
# Test cloning the inline extent against a file which has no extents but
# has a size greater than bar's inline extent (16K > 50).
# It should not be possible to clone the inline extent from file bar
# into this file.
$XFS_IO_PROG -f -c "truncate 16K" $SCRATCH_MNT/foo6 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo6
# Reading the file should not fail.
echo "File foo6 data after clone operation:"
# Must have a size of 16K, with all bytes having a value of 0x00 (the
# clone operation failed and did not modify our file).
od -t x1 $SCRATCH_MNT/foo6
# Test cloning the inline extent against a file which has no extents but
# has a size not greater than bar's inline extent (30 < 50).
# It should be possible to clone the inline extent from file bar into
# this file.
$XFS_IO_PROG -f -c "truncate 30" $SCRATCH_MNT/foo7 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo7
# Reading the file should not fail.
echo "File foo7 data after clone operation:"
# Must have a size of 50 bytes, with all bytes having a value of 0xbb.
od -t x1 $SCRATCH_MNT/foo7
# Test cloning the inline extent against a file which has a size not
# greater than the size of bar's inline extent (20 < 50) but has
# a prealloc extent that goes beyond the file's size. It should not be
# possible to clone the inline extent from bar into this file.
$XFS_IO_PROG -f -c "falloc -k 0 1M" \
-c "pwrite -S 0x88 0 20" \
$SCRATCH_MNT/foo8 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo8
echo "File foo8 data after clone operation:"
# Must have a size of 20 bytes, with all bytes having a value of 0x88
# (the clone operation did not modify our file).
od -t x1 $SCRATCH_MNT/foo8
_scratch_unmount
}
echo -e "\nTesting without compression and without the no-holes feature...\n"
test_cloning_inline_extents
echo -e "\nTesting with compression and without the no-holes feature...\n"
test_cloning_inline_extents "" "-o compress"
echo -e "\nTesting without compression and with the no-holes feature...\n"
test_cloning_inline_extents "-O no-holes" ""
echo -e "\nTesting with compression and with the no-holes feature...\n"
test_cloning_inline_extents "-O no-holes" "-o compress"
status=0
exit
Cc: stable@vger.kernel.org
Signed-off-by: Filipe Manana <fdmanana@suse.com>
2015-10-13 21:15:00 +07:00
|
|
|
key.type == BTRFS_EXTENT_DATA_KEY)
|
|
|
|
|
return -EOPNOTSUPP;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
copy_inline_extent:
|
|
|
|
|
/*
|
|
|
|
|
* We have no extent items, or we have an extent at offset 0 which may
|
|
|
|
|
* or may not be inlined. All these cases are dealt the same way.
|
|
|
|
|
*/
|
|
|
|
|
if (i_size_read(dst) > datal) {
|
|
|
|
|
/*
|
|
|
|
|
* If the destination inode has an inline extent...
|
|
|
|
|
* This would require copying the data from the source inline
|
|
|
|
|
* extent into the beginning of the destination's inline extent.
|
|
|
|
|
* But this is really complex, both extents can be compressed
|
|
|
|
|
* or just one of them, which would require decompressing and
|
|
|
|
|
* re-compressing data (which could increase the new compressed
|
|
|
|
|
* size, not allowing the compressed data to fit anymore in an
|
|
|
|
|
* inline extent).
|
|
|
|
|
* So just don't support this case for now (it should be rare,
|
|
|
|
|
* we are not really saving space when cloning inline extents).
|
|
|
|
|
*/
|
|
|
|
|
return -EOPNOTSUPP;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
btrfs_release_path(path);
|
|
|
|
|
ret = btrfs_drop_extents(trans, root, dst, drop_start, aligned_end, 1);
|
|
|
|
|
if (ret)
|
|
|
|
|
return ret;
|
|
|
|
|
ret = btrfs_insert_empty_item(trans, root, path, new_key, size);
|
|
|
|
|
if (ret)
|
|
|
|
|
return ret;
|
|
|
|
|
|
|
|
|
|
if (skip) {
|
|
|
|
|
const u32 start = btrfs_file_extent_calc_inline_size(0);
|
|
|
|
|
|
|
|
|
|
memmove(inline_data + start, inline_data + start + skip, datal);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
write_extent_buffer(path->nodes[0], inline_data,
|
|
|
|
|
btrfs_item_ptr_offset(path->nodes[0],
|
|
|
|
|
path->slots[0]),
|
|
|
|
|
size);
|
|
|
|
|
inode_add_bytes(dst, datal);
|
|
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
2013-08-07 01:42:49 +07:00
|
|
|
/**
|
|
|
|
|
* btrfs_clone() - clone a range from inode file to another
|
|
|
|
|
*
|
|
|
|
|
* @src: Inode to clone from
|
|
|
|
|
* @inode: Inode to clone to
|
|
|
|
|
* @off: Offset within source to start clone from
|
|
|
|
|
* @olen: Original length, passed by user, of range to clone
|
2015-07-01 04:42:08 +07:00
|
|
|
* @olen_aligned: Block-aligned value of olen
|
2013-08-07 01:42:49 +07:00
|
|
|
* @destoff: Offset within @inode to start clone
|
2015-07-01 04:42:08 +07:00
|
|
|
* @no_time_update: Whether to update mtime/ctime on the target inode
|
2013-08-07 01:42:49 +07:00
|
|
|
*/
|
|
|
|
|
static int btrfs_clone(struct inode *src, struct inode *inode,
|
2014-06-01 07:50:28 +07:00
|
|
|
const u64 off, const u64 olen, const u64 olen_aligned,
|
2015-07-01 04:42:08 +07:00
|
|
|
const u64 destoff, int no_time_update)
|
2008-06-12 08:53:53 +07:00
|
|
|
{
|
2016-06-23 05:54:23 +07:00
|
|
|
struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
|
2008-06-12 08:53:53 +07:00
|
|
|
struct btrfs_root *root = BTRFS_I(inode)->root;
|
2013-08-07 01:42:49 +07:00
|
|
|
struct btrfs_path *path = NULL;
|
2008-06-12 08:53:53 +07:00
|
|
|
struct extent_buffer *leaf;
|
2013-08-07 01:42:49 +07:00
|
|
|
struct btrfs_trans_handle *trans;
|
|
|
|
|
char *buf = NULL;
|
2008-08-05 10:23:47 +07:00
|
|
|
struct btrfs_key key;
|
2008-06-12 08:53:53 +07:00
|
|
|
u32 nritems;
|
|
|
|
|
int slot;
|
2008-08-05 10:23:47 +07:00
|
|
|
int ret;
|
2014-06-01 07:50:28 +07:00
|
|
|
const u64 len = olen_aligned;
|
|
|
|
|
u64 last_dest_end = destoff;
|
2008-08-05 10:23:47 +07:00
|
|
|
|
|
|
|
|
ret = -ENOMEM;
|
2017-05-09 05:57:27 +07:00
|
|
|
buf = kvmalloc(fs_info->nodesize, GFP_KERNEL);
|
|
|
|
|
if (!buf)
|
|
|
|
|
return ret;
|
2008-08-05 10:23:47 +07:00
|
|
|
|
|
|
|
|
path = btrfs_alloc_path();
|
|
|
|
|
if (!path) {
|
2016-04-11 23:40:08 +07:00
|
|
|
kvfree(buf);
|
2013-08-07 01:42:49 +07:00
|
|
|
return ret;
|
2011-09-11 21:52:25 +07:00
|
|
|
}
|
|
|
|
|
|
2015-11-27 22:31:35 +07:00
|
|
|
path->reada = READA_FORWARD;
|
2008-11-13 02:32:25 +07:00
|
|
|
/* clone data */
|
2017-01-11 01:35:31 +07:00
|
|
|
key.objectid = btrfs_ino(BTRFS_I(src));
|
2008-08-05 10:23:47 +07:00
|
|
|
key.type = BTRFS_EXTENT_DATA_KEY;
|
2014-05-31 08:31:05 +07:00
|
|
|
key.offset = off;
|
2008-06-12 08:53:53 +07:00
|
|
|
|
|
|
|
|
while (1) {
|
2015-04-11 19:09:06 +07:00
|
|
|
u64 next_key_min_offset = key.offset + 1;
|
2015-03-31 20:56:46 +07:00
|
|
|
|
2008-06-12 08:53:53 +07:00
|
|
|
/*
|
|
|
|
|
* note the key will change type as we walk through the
|
|
|
|
|
* tree.
|
|
|
|
|
*/
|
2014-01-14 02:35:01 +07:00
|
|
|
path->leave_spinning = 1;
|
2011-08-01 23:11:57 +07:00
|
|
|
ret = btrfs_search_slot(NULL, BTRFS_I(src)->root, &key, path,
|
|
|
|
|
0, 0);
|
2008-06-12 08:53:53 +07:00
|
|
|
if (ret < 0)
|
|
|
|
|
goto out;
|
2014-05-31 08:31:05 +07:00
|
|
|
/*
|
|
|
|
|
* First search, if no extent item that starts at offset off was
|
|
|
|
|
* found but the previous item is an extent item, it's possible
|
|
|
|
|
* it might overlap our target range, therefore process it.
|
|
|
|
|
*/
|
|
|
|
|
if (key.offset == off && ret > 0 && path->slots[0] > 0) {
|
|
|
|
|
btrfs_item_key_to_cpu(path->nodes[0], &key,
|
|
|
|
|
path->slots[0] - 1);
|
|
|
|
|
if (key.type == BTRFS_EXTENT_DATA_KEY)
|
|
|
|
|
path->slots[0]--;
|
|
|
|
|
}
|
2008-06-12 08:53:53 +07:00
|
|
|
|
2008-08-05 10:23:47 +07:00
|
|
|
nritems = btrfs_header_nritems(path->nodes[0]);
|
2014-01-14 02:35:01 +07:00
|
|
|
process_slot:
|
2008-08-05 10:23:47 +07:00
|
|
|
if (path->slots[0] >= nritems) {
|
2011-08-01 23:11:57 +07:00
|
|
|
ret = btrfs_next_leaf(BTRFS_I(src)->root, path);
|
2008-06-12 08:53:53 +07:00
|
|
|
if (ret < 0)
|
|
|
|
|
goto out;
|
|
|
|
|
if (ret > 0)
|
|
|
|
|
break;
|
2008-08-05 10:23:47 +07:00
|
|
|
nritems = btrfs_header_nritems(path->nodes[0]);
|
2008-06-12 08:53:53 +07:00
|
|
|
}
|
|
|
|
|
leaf = path->nodes[0];
|
|
|
|
|
slot = path->slots[0];
|
|
|
|
|
|
2008-08-05 10:23:47 +07:00
|
|
|
btrfs_item_key_to_cpu(leaf, &key, slot);
|
2014-06-04 23:41:45 +07:00
|
|
|
if (key.type > BTRFS_EXTENT_DATA_KEY ||
|
2017-01-11 01:35:31 +07:00
|
|
|
key.objectid != btrfs_ino(BTRFS_I(src)))
|
2008-06-12 08:53:53 +07:00
|
|
|
break;
|
|
|
|
|
|
2014-06-04 23:41:45 +07:00
|
|
|
if (key.type == BTRFS_EXTENT_DATA_KEY) {
|
2008-11-13 02:32:25 +07:00
|
|
|
struct btrfs_file_extent_item *extent;
|
|
|
|
|
int type;
|
2008-09-24 00:14:14 +07:00
|
|
|
u32 size;
|
|
|
|
|
struct btrfs_key new_key;
|
2008-11-13 02:32:25 +07:00
|
|
|
u64 disko = 0, diskl = 0;
|
|
|
|
|
u64 datao = 0, datal = 0;
|
|
|
|
|
u8 comp;
|
2014-06-01 07:50:28 +07:00
|
|
|
u64 drop_start;
|
2008-09-24 00:14:14 +07:00
|
|
|
|
2008-11-13 02:32:25 +07:00
|
|
|
extent = btrfs_item_ptr(leaf, slot,
|
|
|
|
|
struct btrfs_file_extent_item);
|
|
|
|
|
comp = btrfs_file_extent_compression(leaf, extent);
|
|
|
|
|
type = btrfs_file_extent_type(leaf, extent);
|
2009-06-28 08:07:03 +07:00
|
|
|
if (type == BTRFS_FILE_EXTENT_REG ||
|
|
|
|
|
type == BTRFS_FILE_EXTENT_PREALLOC) {
|
2009-01-06 09:25:51 +07:00
|
|
|
disko = btrfs_file_extent_disk_bytenr(leaf,
|
|
|
|
|
extent);
|
|
|
|
|
diskl = btrfs_file_extent_disk_num_bytes(leaf,
|
|
|
|
|
extent);
|
2008-11-13 02:32:25 +07:00
|
|
|
datao = btrfs_file_extent_offset(leaf, extent);
|
2009-01-06 09:25:51 +07:00
|
|
|
datal = btrfs_file_extent_num_bytes(leaf,
|
|
|
|
|
extent);
|
2008-11-13 02:32:25 +07:00
|
|
|
} else if (type == BTRFS_FILE_EXTENT_INLINE) {
|
|
|
|
|
/* take upper bound, may be compressed */
|
|
|
|
|
datal = btrfs_file_extent_ram_bytes(leaf,
|
|
|
|
|
extent);
|
|
|
|
|
}
|
2008-09-24 00:14:14 +07:00
|
|
|
|
2014-05-31 08:31:05 +07:00
|
|
|
/*
|
|
|
|
|
* The first search might have left us at an extent
|
|
|
|
|
* item that ends before our target range's start, can
|
|
|
|
|
* happen if we have holes and NO_HOLES feature enabled.
|
|
|
|
|
*/
|
|
|
|
|
if (key.offset + datal <= off) {
|
2014-01-14 02:35:01 +07:00
|
|
|
path->slots[0]++;
|
|
|
|
|
goto process_slot;
|
2014-05-31 08:31:05 +07:00
|
|
|
} else if (key.offset >= off + len) {
|
|
|
|
|
break;
|
2014-01-14 02:35:01 +07:00
|
|
|
}
|
2015-03-31 20:56:46 +07:00
|
|
|
next_key_min_offset = key.offset + datal;
|
2014-01-14 02:35:01 +07:00
|
|
|
size = btrfs_item_size_nr(leaf, slot);
|
|
|
|
|
read_extent_buffer(leaf, buf,
|
|
|
|
|
btrfs_item_ptr_offset(leaf, slot),
|
|
|
|
|
size);
|
|
|
|
|
|
|
|
|
|
btrfs_release_path(path);
|
|
|
|
|
path->leave_spinning = 0;
|
2008-11-13 02:32:25 +07:00
|
|
|
|
2008-09-24 00:14:14 +07:00
|
|
|
memcpy(&new_key, &key, sizeof(new_key));
|
2017-01-11 01:35:31 +07:00
|
|
|
new_key.objectid = btrfs_ino(BTRFS_I(inode));
|
2011-01-26 13:10:43 +07:00
|
|
|
if (off <= key.offset)
|
|
|
|
|
new_key.offset = key.offset + destoff - off;
|
|
|
|
|
else
|
|
|
|
|
new_key.offset = destoff;
|
2008-09-24 00:14:14 +07:00
|
|
|
|
2014-06-01 07:50:28 +07:00
|
|
|
/*
|
|
|
|
|
* Deal with a hole that doesn't have an extent item
|
|
|
|
|
* that represents it (NO_HOLES feature enabled).
|
|
|
|
|
* This hole is either in the middle of the cloning
|
|
|
|
|
* range or at the beginning (fully overlaps it or
|
|
|
|
|
* partially overlaps it).
|
|
|
|
|
*/
|
|
|
|
|
if (new_key.offset != last_dest_end)
|
|
|
|
|
drop_start = last_dest_end;
|
|
|
|
|
else
|
|
|
|
|
drop_start = new_key.offset;
|
|
|
|
|
|
2011-09-21 01:48:51 +07:00
|
|
|
/*
|
|
|
|
|
* 1 - adjusting old extent (we may have to split it)
|
|
|
|
|
* 1 - add new extent
|
|
|
|
|
* 1 - inode update
|
|
|
|
|
*/
|
|
|
|
|
trans = btrfs_start_transaction(root, 3);
|
2010-05-16 21:48:46 +07:00
|
|
|
if (IS_ERR(trans)) {
|
|
|
|
|
ret = PTR_ERR(trans);
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
|
2009-06-28 08:07:03 +07:00
|
|
|
if (type == BTRFS_FILE_EXTENT_REG ||
|
|
|
|
|
type == BTRFS_FILE_EXTENT_PREALLOC) {
|
2011-09-11 21:52:25 +07:00
|
|
|
/*
|
|
|
|
|
* a | --- range to clone ---| b
|
|
|
|
|
* | ------------- extent ------------- |
|
|
|
|
|
*/
|
|
|
|
|
|
2014-06-04 19:03:48 +07:00
|
|
|
/* subtract range b */
|
2011-09-11 21:52:25 +07:00
|
|
|
if (key.offset + datal > off + len)
|
|
|
|
|
datal = off + len - key.offset;
|
|
|
|
|
|
2014-06-04 19:03:48 +07:00
|
|
|
/* subtract range a */
|
2010-05-16 21:48:46 +07:00
|
|
|
if (off > key.offset) {
|
|
|
|
|
datao += off - key.offset;
|
|
|
|
|
datal -= off - key.offset;
|
|
|
|
|
}
|
|
|
|
|
|
Btrfs: turbo charge fsync
At least for the vm workload. Currently on fsync we will
1) Truncate all items in the log tree for the given inode if they exist
and
2) Copy all items for a given inode into the log
The problem with this is that for things like VMs you can have lots of
extents from the fragmented writing behavior, and worst yet you may have
only modified a few extents, not the entire thing. This patch fixes this
problem by tracking which transid modified our extent, and then when we do
the tree logging we find all of the extents we've modified in our current
transaction, sort them and commit them. We also only truncate up to the
xattrs of the inode and copy that stuff in normally, and then just drop any
extents in the range we have that exist in the log already. Here are some
numbers of a 50 meg fio job that does random writes and fsync()s after every
write
Original Patched
SATA drive 82KB/s 140KB/s
Fusion drive 431KB/s 2532KB/s
So around 2-6 times faster depending on your hardware. There are a few
corner cases, for example if you truncate at all we have to do it the old
way since there is no way to be sure what is in the log is ok. This
probably could be done smarter, but if you write-fsync-truncate-write-fsync
you deserve what you get. All this work is in RAM of course so if your
inode gets evicted from cache and you read it in and fsync it we'll do it
the slow way if we are still in the same transaction that we last modified
the inode in.
The biggest cool part of this is that it requires no changes to the recovery
code, so if you fsync with this patch and crash and load an old kernel, it
will run the recovery and be a-ok. I have tested this pretty thoroughly
with an fsync tester and everything comes back fine, as well as xfstests.
Thanks,
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2012-08-18 00:14:17 +07:00
|
|
|
ret = btrfs_drop_extents(trans, root, inode,
|
2014-06-01 07:50:28 +07:00
|
|
|
drop_start,
|
2010-05-16 21:48:46 +07:00
|
|
|
new_key.offset + datal,
|
2012-08-29 23:24:27 +07:00
|
|
|
1);
|
2012-03-12 22:03:00 +07:00
|
|
|
if (ret) {
|
2014-04-15 23:50:17 +07:00
|
|
|
if (ret != -EOPNOTSUPP)
|
2014-03-10 17:56:07 +07:00
|
|
|
btrfs_abort_transaction(trans,
|
2016-06-11 05:19:25 +07:00
|
|
|
ret);
|
2016-09-10 08:39:03 +07:00
|
|
|
btrfs_end_transaction(trans);
|
2012-03-12 22:03:00 +07:00
|
|
|
goto out;
|
|
|
|
|
}
|
2010-05-16 21:48:46 +07:00
|
|
|
|
2008-11-13 02:32:25 +07:00
|
|
|
ret = btrfs_insert_empty_item(trans, root, path,
|
|
|
|
|
&new_key, size);
|
2012-03-12 22:03:00 +07:00
|
|
|
if (ret) {
|
2016-06-11 05:19:25 +07:00
|
|
|
btrfs_abort_transaction(trans, ret);
|
2016-09-10 08:39:03 +07:00
|
|
|
btrfs_end_transaction(trans);
|
2012-03-12 22:03:00 +07:00
|
|
|
goto out;
|
|
|
|
|
}
|
2008-11-13 02:32:25 +07:00
|
|
|
|
|
|
|
|
leaf = path->nodes[0];
|
|
|
|
|
slot = path->slots[0];
|
|
|
|
|
write_extent_buffer(leaf, buf,
|
2008-09-24 00:14:14 +07:00
|
|
|
btrfs_item_ptr_offset(leaf, slot),
|
|
|
|
|
size);
|
2008-08-05 10:23:47 +07:00
|
|
|
|
2008-11-13 02:32:25 +07:00
|
|
|
extent = btrfs_item_ptr(leaf, slot,
|
2008-06-12 08:53:53 +07:00
|
|
|
struct btrfs_file_extent_item);
|
2008-11-13 02:32:25 +07:00
|
|
|
|
|
|
|
|
/* disko == 0 means it's a hole */
|
|
|
|
|
if (!disko)
|
|
|
|
|
datao = 0;
|
|
|
|
|
|
|
|
|
|
btrfs_set_file_extent_offset(leaf, extent,
|
|
|
|
|
datao);
|
|
|
|
|
btrfs_set_file_extent_num_bytes(leaf, extent,
|
|
|
|
|
datal);
|
2014-05-14 07:30:47 +07:00
|
|
|
|
2008-11-13 02:32:25 +07:00
|
|
|
if (disko) {
|
|
|
|
|
inode_add_bytes(inode, datal);
|
2016-06-23 05:54:24 +07:00
|
|
|
ret = btrfs_inc_extent_ref(trans,
|
2017-09-30 02:43:49 +07:00
|
|
|
root,
|
Btrfs: Mixed back reference (FORWARD ROLLING FORMAT CHANGE)
This commit introduces a new kind of back reference for btrfs metadata.
Once a filesystem has been mounted with this commit, IT WILL NO LONGER
BE MOUNTABLE BY OLDER KERNELS.
When a tree block in subvolume tree is cow'd, the reference counts of all
extents it points to are increased by one. At transaction commit time,
the old root of the subvolume is recorded in a "dead root" data structure,
and the btree it points to is later walked, dropping reference counts
and freeing any blocks where the reference count goes to 0.
The increments done during cow and decrements done after commit cancel out,
and the walk is a very expensive way to go about freeing the blocks that
are no longer referenced by the new btree root. This commit reduces the
transaction overhead by avoiding the need for dead root records.
When a non-shared tree block is cow'd, we free the old block at once, and the
new block inherits old block's references. When a tree block with reference
count > 1 is cow'd, we increase the reference counts of all extents
the new block points to by one, and decrease the old block's reference count by
one.
This dead tree avoidance code removes the need to modify the reference
counts of lower level extents when a non-shared tree block is cow'd.
But we still need to update back ref for all pointers in the block.
This is because the location of the block is recorded in the back ref
item.
We can solve this by introducing a new type of back ref. The new
back ref provides information about pointer's key, level and in which
tree the pointer lives. This information allow us to find the pointer
by searching the tree. The shortcoming of the new back ref is that it
only works for pointers in tree blocks referenced by their owner trees.
This is mostly a problem for snapshots, where resolving one of these
fuzzy back references would be O(number_of_snapshots) and quite slow.
The solution used here is to use the fuzzy back references in the common
case where a given tree block is only referenced by one root,
and use the full back references when multiple roots have a reference
on a given block.
This commit adds per subvolume red-black tree to keep trace of cached
inodes. The red-black tree helps the balancing code to find cached
inodes whose inode numbers within a given range.
This commit improves the balancing code by introducing several data
structures to keep the state of balancing. The most important one
is the back ref cache. It caches how the upper level tree blocks are
referenced. This greatly reduce the overhead of checking back ref.
The improved balancing code scales significantly better with a large
number of snapshots.
This is a very large commit and was written in a number of
pieces. But, they depend heavily on the disk format change and were
squashed together to make sure git bisect didn't end up in a
bad state wrt space balancing or the format change.
Signed-off-by: Yan Zheng <zheng.yan@oracle.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
2009-06-10 21:45:14 +07:00
|
|
|
disko, diskl, 0,
|
|
|
|
|
root->root_key.objectid,
|
2017-01-11 01:35:31 +07:00
|
|
|
btrfs_ino(BTRFS_I(inode)),
|
Btrfs: fix regression running delayed references when using qgroups
In the kernel 4.2 merge window we had a big changes to the implementation
of delayed references and qgroups which made the no_quota field of delayed
references not used anymore. More specifically the no_quota field is not
used anymore as of:
commit 0ed4792af0e8 ("btrfs: qgroup: Switch to new extent-oriented qgroup mechanism.")
Leaving the no_quota field actually prevents delayed references from
getting merged, which in turn cause the following BUG_ON(), at
fs/btrfs/extent-tree.c, to be hit when qgroups are enabled:
static int run_delayed_tree_ref(...)
{
(...)
BUG_ON(node->ref_mod != 1);
(...)
}
This happens on a scenario like the following:
1) Ref1 bytenr X, action = BTRFS_ADD_DELAYED_REF, no_quota = 1, added.
2) Ref2 bytenr X, action = BTRFS_DROP_DELAYED_REF, no_quota = 0, added.
It's not merged with Ref1 because Ref1->no_quota != Ref2->no_quota.
3) Ref3 bytenr X, action = BTRFS_ADD_DELAYED_REF, no_quota = 1, added.
It's not merged with the reference at the tail of the list of refs
for bytenr X because the reference at the tail, Ref2 is incompatible
due to Ref2->no_quota != Ref3->no_quota.
4) Ref4 bytenr X, action = BTRFS_DROP_DELAYED_REF, no_quota = 0, added.
It's not merged with the reference at the tail of the list of refs
for bytenr X because the reference at the tail, Ref3 is incompatible
due to Ref3->no_quota != Ref4->no_quota.
5) We run delayed references, trigger merging of delayed references,
through __btrfs_run_delayed_refs() -> btrfs_merge_delayed_refs().
6) Ref1 and Ref3 are merged as Ref1->no_quota = Ref3->no_quota and
all other conditions are satisfied too. So Ref1 gets a ref_mod
value of 2.
7) Ref2 and Ref4 are merged as Ref2->no_quota = Ref4->no_quota and
all other conditions are satisfied too. So Ref2 gets a ref_mod
value of 2.
8) Ref1 and Ref2 aren't merged, because they have different values
for their no_quota field.
9) Delayed reference Ref1 is picked for running (select_delayed_ref()
always prefers references with an action == BTRFS_ADD_DELAYED_REF).
So run_delayed_tree_ref() is called for Ref1 which triggers the
BUG_ON because Ref1->red_mod != 1 (equals 2).
So fix this by removing the no_quota field, as it's not used anymore as
of commit 0ed4792af0e8 ("btrfs: qgroup: Switch to new extent-oriented
qgroup mechanism.").
The use of no_quota was also buggy in at least two places:
1) At delayed-refs.c:btrfs_add_delayed_tree_ref() - we were setting
no_quota to 0 instead of 1 when the following condition was true:
is_fstree(ref_root) || !fs_info->quota_enabled
2) At extent-tree.c:__btrfs_inc_extent_ref() - we were attempting to
reset a node's no_quota when the condition "!is_fstree(root_objectid)
|| !root->fs_info->quota_enabled" was true but we did it only in
an unused local stack variable, that is, we never reset the no_quota
value in the node itself.
This fixes the remainder of problems several people have been having when
running delayed references, mostly while a balance is running in parallel,
on a 4.2+ kernel.
Very special thanks to Stéphane Lesimple for helping debugging this issue
and testing this fix on his multi terabyte filesystem (which took more
than one day to balance alone, plus fsck, etc).
Also, this fixes deadlock issue when using the clone ioctl with qgroups
enabled, as reported by Elias Probst in the mailing list. The deadlock
happens because after calling btrfs_insert_empty_item we have our path
holding a write lock on a leaf of the fs/subvol tree and then before
releasing the path we called check_ref() which did backref walking, when
qgroups are enabled, and tried to read lock the same leaf. The trace for
this case is the following:
INFO: task systemd-nspawn:6095 blocked for more than 120 seconds.
(...)
Call Trace:
[<ffffffff86999201>] schedule+0x74/0x83
[<ffffffff863ef64c>] btrfs_tree_read_lock+0xc0/0xea
[<ffffffff86137ed7>] ? wait_woken+0x74/0x74
[<ffffffff8639f0a7>] btrfs_search_old_slot+0x51a/0x810
[<ffffffff863a129b>] btrfs_next_old_leaf+0xdf/0x3ce
[<ffffffff86413a00>] ? ulist_add_merge+0x1b/0x127
[<ffffffff86411688>] __resolve_indirect_refs+0x62a/0x667
[<ffffffff863ef546>] ? btrfs_clear_lock_blocking_rw+0x78/0xbe
[<ffffffff864122d3>] find_parent_nodes+0xaf3/0xfc6
[<ffffffff86412838>] __btrfs_find_all_roots+0x92/0xf0
[<ffffffff864128f2>] btrfs_find_all_roots+0x45/0x65
[<ffffffff8639a75b>] ? btrfs_get_tree_mod_seq+0x2b/0x88
[<ffffffff863e852e>] check_ref+0x64/0xc4
[<ffffffff863e9e01>] btrfs_clone+0x66e/0xb5d
[<ffffffff863ea77f>] btrfs_ioctl_clone+0x48f/0x5bb
[<ffffffff86048a68>] ? native_sched_clock+0x28/0x77
[<ffffffff863ed9b0>] btrfs_ioctl+0xabc/0x25cb
(...)
The problem goes away by eleminating check_ref(), which no longer is
needed as its purpose was to get a value for the no_quota field of
a delayed reference (this patch removes the no_quota field as mentioned
earlier).
Reported-by: Stéphane Lesimple <stephane_btrfs@lesimple.fr>
Tested-by: Stéphane Lesimple <stephane_btrfs@lesimple.fr>
Reported-by: Elias Probst <mail@eliasprobst.eu>
Reported-by: Peter Becker <floyd.net@gmail.com>
Reported-by: Malte Schröder <malte@tnxip.de>
Reported-by: Derek Dongray <derek@valedon.co.uk>
Reported-by: Erkki Seppala <flux-btrfs@inside.org>
Cc: stable@vger.kernel.org # 4.2+
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
2015-10-23 13:52:54 +07:00
|
|
|
new_key.offset - datao);
|
2012-03-12 22:03:00 +07:00
|
|
|
if (ret) {
|
|
|
|
|
btrfs_abort_transaction(trans,
|
|
|
|
|
ret);
|
2016-09-10 08:39:03 +07:00
|
|
|
btrfs_end_transaction(trans);
|
2012-03-12 22:03:00 +07:00
|
|
|
goto out;
|
|
|
|
|
|
|
|
|
|
}
|
2008-06-12 08:53:53 +07:00
|
|
|
}
|
2008-11-13 02:32:25 +07:00
|
|
|
} else if (type == BTRFS_FILE_EXTENT_INLINE) {
|
|
|
|
|
u64 skip = 0;
|
|
|
|
|
u64 trim = 0;
|
Btrfs: fix file corruption after cloning inline extents
Using the clone ioctl (or extent_same ioctl, which calls the same extent
cloning function as well) we end up allowing copy an inline extent from
the source file into a non-zero offset of the destination file. This is
something not expected and that the btrfs code is not prepared to deal
with - all inline extents must be at a file offset equals to 0.
For example, the following excerpt of a test case for fstests triggers
a crash/BUG_ON() on a write operation after an inline extent is cloned
into a non-zero offset:
_scratch_mkfs >>$seqres.full 2>&1
_scratch_mount
# Create our test files. File foo has the same 2K of data at offset 4K
# as file bar has at its offset 0.
$XFS_IO_PROG -f -s -c "pwrite -S 0xaa 0 4K" \
-c "pwrite -S 0xbb 4k 2K" \
-c "pwrite -S 0xcc 8K 4K" \
$SCRATCH_MNT/foo | _filter_xfs_io
# File bar consists of a single inline extent (2K size).
$XFS_IO_PROG -f -s -c "pwrite -S 0xbb 0 2K" \
$SCRATCH_MNT/bar | _filter_xfs_io
# Now call the clone ioctl to clone the extent of file bar into file
# foo at its offset 4K. This made file foo have an inline extent at
# offset 4K, something which the btrfs code can not deal with in future
# IO operations because all inline extents are supposed to start at an
# offset of 0, resulting in all sorts of chaos.
# So here we validate that clone ioctl returns an EOPNOTSUPP, which is
# what it returns for other cases dealing with inlined extents.
$CLONER_PROG -s 0 -d $((4 * 1024)) -l $((2 * 1024)) \
$SCRATCH_MNT/bar $SCRATCH_MNT/foo
# Because of the inline extent at offset 4K, the following write made
# the kernel crash with a BUG_ON().
$XFS_IO_PROG -c "pwrite -S 0xdd 6K 2K" $SCRATCH_MNT/foo | _filter_xfs_io
status=0
exit
The stack trace of the BUG_ON() triggered by the last write is:
[152154.035903] ------------[ cut here ]------------
[152154.036424] kernel BUG at mm/page-writeback.c:2286!
[152154.036424] invalid opcode: 0000 [#1] PREEMPT SMP DEBUG_PAGEALLOC
[152154.036424] Modules linked in: btrfs dm_flakey dm_mod crc32c_generic xor raid6_pq nfsd auth_rpcgss oid_registry nfs_acl nfs lockd grace fscache sunrpc loop fuse parport_pc acpi_cpu$
[152154.036424] CPU: 2 PID: 17873 Comm: xfs_io Tainted: G W 4.1.0-rc6-btrfs-next-11+ #2
[152154.036424] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.8.1-0-g4adadbd-20150316_085822-nilsson.home.kraxel.org 04/01/2014
[152154.036424] task: ffff880429f70990 ti: ffff880429efc000 task.ti: ffff880429efc000
[152154.036424] RIP: 0010:[<ffffffff8111a9d5>] [<ffffffff8111a9d5>] clear_page_dirty_for_io+0x1e/0x90
[152154.036424] RSP: 0018:ffff880429effc68 EFLAGS: 00010246
[152154.036424] RAX: 0200000000000806 RBX: ffffea0006a6d8f0 RCX: 0000000000000001
[152154.036424] RDX: 0000000000000000 RSI: ffffffff81155d1b RDI: ffffea0006a6d8f0
[152154.036424] RBP: ffff880429effc78 R08: ffff8801ce389fe0 R09: 0000000000000001
[152154.036424] R10: 0000000000002000 R11: ffffffffffffffff R12: ffff8800200dce68
[152154.036424] R13: 0000000000000000 R14: ffff8800200dcc88 R15: ffff8803d5736d80
[152154.036424] FS: 00007fbf119f6700(0000) GS:ffff88043d280000(0000) knlGS:0000000000000000
[152154.036424] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[152154.036424] CR2: 0000000001bdc000 CR3: 00000003aa555000 CR4: 00000000000006e0
[152154.036424] Stack:
[152154.036424] ffff8803d5736d80 0000000000000001 ffff880429effcd8 ffffffffa04e97c1
[152154.036424] ffff880429effd68 ffff880429effd60 0000000000000001 ffff8800200dc9c8
[152154.036424] 0000000000000001 ffff8800200dcc88 0000000000000000 0000000000001000
[152154.036424] Call Trace:
[152154.036424] [<ffffffffa04e97c1>] lock_and_cleanup_extent_if_need+0x147/0x18d [btrfs]
[152154.036424] [<ffffffffa04ea82c>] __btrfs_buffered_write+0x245/0x4c8 [btrfs]
[152154.036424] [<ffffffffa04ed14b>] ? btrfs_file_write_iter+0x150/0x3e0 [btrfs]
[152154.036424] [<ffffffffa04ed15a>] ? btrfs_file_write_iter+0x15f/0x3e0 [btrfs]
[152154.036424] [<ffffffffa04ed2c7>] btrfs_file_write_iter+0x2cc/0x3e0 [btrfs]
[152154.036424] [<ffffffff81165a4a>] __vfs_write+0x7c/0xa5
[152154.036424] [<ffffffff81165f89>] vfs_write+0xa0/0xe4
[152154.036424] [<ffffffff81166855>] SyS_pwrite64+0x64/0x82
[152154.036424] [<ffffffff81465197>] system_call_fastpath+0x12/0x6f
[152154.036424] Code: 48 89 c7 e8 0f ff ff ff 5b 41 5c 5d c3 0f 1f 44 00 00 55 48 89 e5 41 54 53 48 89 fb e8 ae ef 00 00 49 89 c4 48 8b 03 a8 01 75 02 <0f> 0b 4d 85 e4 74 59 49 8b 3c 2$
[152154.036424] RIP [<ffffffff8111a9d5>] clear_page_dirty_for_io+0x1e/0x90
[152154.036424] RSP <ffff880429effc68>
[152154.242621] ---[ end trace e3d3376b23a57041 ]---
Fix this by returning the error EOPNOTSUPP if an attempt to copy an
inline extent into a non-zero offset happens, just like what is done for
other scenarios that would require copying/splitting inline extents,
which were introduced by the following commits:
00fdf13a2e9f ("Btrfs: fix a crash of clone with inline extents's split")
3f9e3df8da3c ("btrfs: replace error code from btrfs_drop_extents")
Cc: stable@vger.kernel.org
Signed-off-by: Filipe Manana <fdmanana@suse.com>
2015-07-14 22:09:39 +07:00
|
|
|
|
2008-11-13 02:32:25 +07:00
|
|
|
if (off > key.offset) {
|
|
|
|
|
skip = off - key.offset;
|
|
|
|
|
new_key.offset += skip;
|
|
|
|
|
}
|
2009-01-06 09:25:51 +07:00
|
|
|
|
2012-09-18 16:52:23 +07:00
|
|
|
if (key.offset + datal > off + len)
|
|
|
|
|
trim = key.offset + datal - (off + len);
|
2009-01-06 09:25:51 +07:00
|
|
|
|
2008-11-13 02:32:25 +07:00
|
|
|
if (comp && (skip || trim)) {
|
|
|
|
|
ret = -EINVAL;
|
2016-09-10 08:39:03 +07:00
|
|
|
btrfs_end_transaction(trans);
|
2008-11-13 02:32:25 +07:00
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
size -= skip + trim;
|
|
|
|
|
datal -= skip + trim;
|
2010-05-16 21:48:46 +07:00
|
|
|
|
2017-02-11 02:18:49 +07:00
|
|
|
ret = clone_copy_inline_extent(inode,
|
Btrfs: fix file corruption and data loss after cloning inline extents
Currently the clone ioctl allows to clone an inline extent from one file
to another that already has other (non-inlined) extents. This is a problem
because btrfs is not designed to deal with files having inline and regular
extents, if a file has an inline extent then it must be the only extent
in the file and must start at file offset 0. Having a file with an inline
extent followed by regular extents results in EIO errors when doing reads
or writes against the first 4K of the file.
Also, the clone ioctl allows one to lose data if the source file consists
of a single inline extent, with a size of N bytes, and the destination
file consists of a single inline extent with a size of M bytes, where we
have M > N. In this case the clone operation removes the inline extent
from the destination file and then copies the inline extent from the
source file into the destination file - we lose the M - N bytes from the
destination file, a read operation will get the value 0x00 for any bytes
in the the range [N, M] (the destination inode's i_size remained as M,
that's why we can read past N bytes).
So fix this by not allowing such destructive operations to happen and
return errno EOPNOTSUPP to user space.
Currently the fstest btrfs/035 tests the data loss case but it totally
ignores this - i.e. expects the operation to succeed and does not check
the we got data loss.
The following test case for fstests exercises all these cases that result
in file corruption and data loss:
seq=`basename $0`
seqres=$RESULT_DIR/$seq
echo "QA output created by $seq"
tmp=/tmp/$$
status=1 # failure is the default!
trap "_cleanup; exit \$status" 0 1 2 3 15
_cleanup()
{
rm -f $tmp.*
}
# get standard environment, filters and checks
. ./common/rc
. ./common/filter
# real QA test starts here
_need_to_be_root
_supported_fs btrfs
_supported_os Linux
_require_scratch
_require_cloner
_require_btrfs_fs_feature "no_holes"
_require_btrfs_mkfs_feature "no-holes"
rm -f $seqres.full
test_cloning_inline_extents()
{
local mkfs_opts=$1
local mount_opts=$2
_scratch_mkfs $mkfs_opts >>$seqres.full 2>&1
_scratch_mount $mount_opts
# File bar, the source for all the following clone operations, consists
# of a single inline extent (50 bytes).
$XFS_IO_PROG -f -c "pwrite -S 0xbb 0 50" $SCRATCH_MNT/bar \
| _filter_xfs_io
# Test cloning into a file with an extent (non-inlined) where the
# destination offset overlaps that extent. It should not be possible to
# clone the inline extent from file bar into this file.
$XFS_IO_PROG -f -c "pwrite -S 0xaa 0K 16K" $SCRATCH_MNT/foo \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo
# Doing IO against any range in the first 4K of the file should work.
# Due to a past clone ioctl bug which allowed cloning the inline extent,
# these operations resulted in EIO errors.
echo "File foo data after clone operation:"
# All bytes should have the value 0xaa (clone operation failed and did
# not modify our file).
od -t x1 $SCRATCH_MNT/foo
$XFS_IO_PROG -c "pwrite -S 0xcc 0 100" $SCRATCH_MNT/foo | _filter_xfs_io
# Test cloning the inline extent against a file which has a hole in its
# first 4K followed by a non-inlined extent. It should not be possible
# as well to clone the inline extent from file bar into this file.
$XFS_IO_PROG -f -c "pwrite -S 0xdd 4K 12K" $SCRATCH_MNT/foo2 \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo2
# Doing IO against any range in the first 4K of the file should work.
# Due to a past clone ioctl bug which allowed cloning the inline extent,
# these operations resulted in EIO errors.
echo "File foo2 data after clone operation:"
# All bytes should have the value 0x00 (clone operation failed and did
# not modify our file).
od -t x1 $SCRATCH_MNT/foo2
$XFS_IO_PROG -c "pwrite -S 0xee 0 90" $SCRATCH_MNT/foo2 | _filter_xfs_io
# Test cloning the inline extent against a file which has a size of zero
# but has a prealloc extent. It should not be possible as well to clone
# the inline extent from file bar into this file.
$XFS_IO_PROG -f -c "falloc -k 0 1M" $SCRATCH_MNT/foo3 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo3
# Doing IO against any range in the first 4K of the file should work.
# Due to a past clone ioctl bug which allowed cloning the inline extent,
# these operations resulted in EIO errors.
echo "First 50 bytes of foo3 after clone operation:"
# Should not be able to read any bytes, file has 0 bytes i_size (the
# clone operation failed and did not modify our file).
od -t x1 $SCRATCH_MNT/foo3
$XFS_IO_PROG -c "pwrite -S 0xff 0 90" $SCRATCH_MNT/foo3 | _filter_xfs_io
# Test cloning the inline extent against a file which consists of a
# single inline extent that has a size not greater than the size of
# bar's inline extent (40 < 50).
# It should be possible to do the extent cloning from bar to this file.
$XFS_IO_PROG -f -c "pwrite -S 0x01 0 40" $SCRATCH_MNT/foo4 \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo4
# Doing IO against any range in the first 4K of the file should work.
echo "File foo4 data after clone operation:"
# Must match file bar's content.
od -t x1 $SCRATCH_MNT/foo4
$XFS_IO_PROG -c "pwrite -S 0x02 0 90" $SCRATCH_MNT/foo4 | _filter_xfs_io
# Test cloning the inline extent against a file which consists of a
# single inline extent that has a size greater than the size of bar's
# inline extent (60 > 50).
# It should not be possible to clone the inline extent from file bar
# into this file.
$XFS_IO_PROG -f -c "pwrite -S 0x03 0 60" $SCRATCH_MNT/foo5 \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo5
# Reading the file should not fail.
echo "File foo5 data after clone operation:"
# Must have a size of 60 bytes, with all bytes having a value of 0x03
# (the clone operation failed and did not modify our file).
od -t x1 $SCRATCH_MNT/foo5
# Test cloning the inline extent against a file which has no extents but
# has a size greater than bar's inline extent (16K > 50).
# It should not be possible to clone the inline extent from file bar
# into this file.
$XFS_IO_PROG -f -c "truncate 16K" $SCRATCH_MNT/foo6 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo6
# Reading the file should not fail.
echo "File foo6 data after clone operation:"
# Must have a size of 16K, with all bytes having a value of 0x00 (the
# clone operation failed and did not modify our file).
od -t x1 $SCRATCH_MNT/foo6
# Test cloning the inline extent against a file which has no extents but
# has a size not greater than bar's inline extent (30 < 50).
# It should be possible to clone the inline extent from file bar into
# this file.
$XFS_IO_PROG -f -c "truncate 30" $SCRATCH_MNT/foo7 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo7
# Reading the file should not fail.
echo "File foo7 data after clone operation:"
# Must have a size of 50 bytes, with all bytes having a value of 0xbb.
od -t x1 $SCRATCH_MNT/foo7
# Test cloning the inline extent against a file which has a size not
# greater than the size of bar's inline extent (20 < 50) but has
# a prealloc extent that goes beyond the file's size. It should not be
# possible to clone the inline extent from bar into this file.
$XFS_IO_PROG -f -c "falloc -k 0 1M" \
-c "pwrite -S 0x88 0 20" \
$SCRATCH_MNT/foo8 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo8
echo "File foo8 data after clone operation:"
# Must have a size of 20 bytes, with all bytes having a value of 0x88
# (the clone operation did not modify our file).
od -t x1 $SCRATCH_MNT/foo8
_scratch_unmount
}
echo -e "\nTesting without compression and without the no-holes feature...\n"
test_cloning_inline_extents
echo -e "\nTesting with compression and without the no-holes feature...\n"
test_cloning_inline_extents "" "-o compress"
echo -e "\nTesting without compression and with the no-holes feature...\n"
test_cloning_inline_extents "-O no-holes" ""
echo -e "\nTesting with compression and with the no-holes feature...\n"
test_cloning_inline_extents "-O no-holes" "-o compress"
status=0
exit
Cc: stable@vger.kernel.org
Signed-off-by: Filipe Manana <fdmanana@suse.com>
2015-10-13 21:15:00 +07:00
|
|
|
trans, path,
|
|
|
|
|
&new_key,
|
|
|
|
|
drop_start,
|
|
|
|
|
datal,
|
|
|
|
|
skip, size, buf);
|
2012-03-12 22:03:00 +07:00
|
|
|
if (ret) {
|
2014-04-15 23:50:17 +07:00
|
|
|
if (ret != -EOPNOTSUPP)
|
2014-04-07 21:10:40 +07:00
|
|
|
btrfs_abort_transaction(trans,
|
Btrfs: fix file corruption and data loss after cloning inline extents
Currently the clone ioctl allows to clone an inline extent from one file
to another that already has other (non-inlined) extents. This is a problem
because btrfs is not designed to deal with files having inline and regular
extents, if a file has an inline extent then it must be the only extent
in the file and must start at file offset 0. Having a file with an inline
extent followed by regular extents results in EIO errors when doing reads
or writes against the first 4K of the file.
Also, the clone ioctl allows one to lose data if the source file consists
of a single inline extent, with a size of N bytes, and the destination
file consists of a single inline extent with a size of M bytes, where we
have M > N. In this case the clone operation removes the inline extent
from the destination file and then copies the inline extent from the
source file into the destination file - we lose the M - N bytes from the
destination file, a read operation will get the value 0x00 for any bytes
in the the range [N, M] (the destination inode's i_size remained as M,
that's why we can read past N bytes).
So fix this by not allowing such destructive operations to happen and
return errno EOPNOTSUPP to user space.
Currently the fstest btrfs/035 tests the data loss case but it totally
ignores this - i.e. expects the operation to succeed and does not check
the we got data loss.
The following test case for fstests exercises all these cases that result
in file corruption and data loss:
seq=`basename $0`
seqres=$RESULT_DIR/$seq
echo "QA output created by $seq"
tmp=/tmp/$$
status=1 # failure is the default!
trap "_cleanup; exit \$status" 0 1 2 3 15
_cleanup()
{
rm -f $tmp.*
}
# get standard environment, filters and checks
. ./common/rc
. ./common/filter
# real QA test starts here
_need_to_be_root
_supported_fs btrfs
_supported_os Linux
_require_scratch
_require_cloner
_require_btrfs_fs_feature "no_holes"
_require_btrfs_mkfs_feature "no-holes"
rm -f $seqres.full
test_cloning_inline_extents()
{
local mkfs_opts=$1
local mount_opts=$2
_scratch_mkfs $mkfs_opts >>$seqres.full 2>&1
_scratch_mount $mount_opts
# File bar, the source for all the following clone operations, consists
# of a single inline extent (50 bytes).
$XFS_IO_PROG -f -c "pwrite -S 0xbb 0 50" $SCRATCH_MNT/bar \
| _filter_xfs_io
# Test cloning into a file with an extent (non-inlined) where the
# destination offset overlaps that extent. It should not be possible to
# clone the inline extent from file bar into this file.
$XFS_IO_PROG -f -c "pwrite -S 0xaa 0K 16K" $SCRATCH_MNT/foo \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo
# Doing IO against any range in the first 4K of the file should work.
# Due to a past clone ioctl bug which allowed cloning the inline extent,
# these operations resulted in EIO errors.
echo "File foo data after clone operation:"
# All bytes should have the value 0xaa (clone operation failed and did
# not modify our file).
od -t x1 $SCRATCH_MNT/foo
$XFS_IO_PROG -c "pwrite -S 0xcc 0 100" $SCRATCH_MNT/foo | _filter_xfs_io
# Test cloning the inline extent against a file which has a hole in its
# first 4K followed by a non-inlined extent. It should not be possible
# as well to clone the inline extent from file bar into this file.
$XFS_IO_PROG -f -c "pwrite -S 0xdd 4K 12K" $SCRATCH_MNT/foo2 \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo2
# Doing IO against any range in the first 4K of the file should work.
# Due to a past clone ioctl bug which allowed cloning the inline extent,
# these operations resulted in EIO errors.
echo "File foo2 data after clone operation:"
# All bytes should have the value 0x00 (clone operation failed and did
# not modify our file).
od -t x1 $SCRATCH_MNT/foo2
$XFS_IO_PROG -c "pwrite -S 0xee 0 90" $SCRATCH_MNT/foo2 | _filter_xfs_io
# Test cloning the inline extent against a file which has a size of zero
# but has a prealloc extent. It should not be possible as well to clone
# the inline extent from file bar into this file.
$XFS_IO_PROG -f -c "falloc -k 0 1M" $SCRATCH_MNT/foo3 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo3
# Doing IO against any range in the first 4K of the file should work.
# Due to a past clone ioctl bug which allowed cloning the inline extent,
# these operations resulted in EIO errors.
echo "First 50 bytes of foo3 after clone operation:"
# Should not be able to read any bytes, file has 0 bytes i_size (the
# clone operation failed and did not modify our file).
od -t x1 $SCRATCH_MNT/foo3
$XFS_IO_PROG -c "pwrite -S 0xff 0 90" $SCRATCH_MNT/foo3 | _filter_xfs_io
# Test cloning the inline extent against a file which consists of a
# single inline extent that has a size not greater than the size of
# bar's inline extent (40 < 50).
# It should be possible to do the extent cloning from bar to this file.
$XFS_IO_PROG -f -c "pwrite -S 0x01 0 40" $SCRATCH_MNT/foo4 \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo4
# Doing IO against any range in the first 4K of the file should work.
echo "File foo4 data after clone operation:"
# Must match file bar's content.
od -t x1 $SCRATCH_MNT/foo4
$XFS_IO_PROG -c "pwrite -S 0x02 0 90" $SCRATCH_MNT/foo4 | _filter_xfs_io
# Test cloning the inline extent against a file which consists of a
# single inline extent that has a size greater than the size of bar's
# inline extent (60 > 50).
# It should not be possible to clone the inline extent from file bar
# into this file.
$XFS_IO_PROG -f -c "pwrite -S 0x03 0 60" $SCRATCH_MNT/foo5 \
| _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo5
# Reading the file should not fail.
echo "File foo5 data after clone operation:"
# Must have a size of 60 bytes, with all bytes having a value of 0x03
# (the clone operation failed and did not modify our file).
od -t x1 $SCRATCH_MNT/foo5
# Test cloning the inline extent against a file which has no extents but
# has a size greater than bar's inline extent (16K > 50).
# It should not be possible to clone the inline extent from file bar
# into this file.
$XFS_IO_PROG -f -c "truncate 16K" $SCRATCH_MNT/foo6 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo6
# Reading the file should not fail.
echo "File foo6 data after clone operation:"
# Must have a size of 16K, with all bytes having a value of 0x00 (the
# clone operation failed and did not modify our file).
od -t x1 $SCRATCH_MNT/foo6
# Test cloning the inline extent against a file which has no extents but
# has a size not greater than bar's inline extent (30 < 50).
# It should be possible to clone the inline extent from file bar into
# this file.
$XFS_IO_PROG -f -c "truncate 30" $SCRATCH_MNT/foo7 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo7
# Reading the file should not fail.
echo "File foo7 data after clone operation:"
# Must have a size of 50 bytes, with all bytes having a value of 0xbb.
od -t x1 $SCRATCH_MNT/foo7
# Test cloning the inline extent against a file which has a size not
# greater than the size of bar's inline extent (20 < 50) but has
# a prealloc extent that goes beyond the file's size. It should not be
# possible to clone the inline extent from bar into this file.
$XFS_IO_PROG -f -c "falloc -k 0 1M" \
-c "pwrite -S 0x88 0 20" \
$SCRATCH_MNT/foo8 | _filter_xfs_io
$CLONER_PROG -s 0 -d 0 -l 0 $SCRATCH_MNT/bar $SCRATCH_MNT/foo8
echo "File foo8 data after clone operation:"
# Must have a size of 20 bytes, with all bytes having a value of 0x88
# (the clone operation did not modify our file).
od -t x1 $SCRATCH_MNT/foo8
_scratch_unmount
}
echo -e "\nTesting without compression and without the no-holes feature...\n"
test_cloning_inline_extents
echo -e "\nTesting with compression and without the no-holes feature...\n"
test_cloning_inline_extents "" "-o compress"
echo -e "\nTesting without compression and with the no-holes feature...\n"
test_cloning_inline_extents "-O no-holes" ""
echo -e "\nTesting with compression and with the no-holes feature...\n"
test_cloning_inline_extents "-O no-holes" "-o compress"
status=0
exit
Cc: stable@vger.kernel.org
Signed-off-by: Filipe Manana <fdmanana@suse.com>
2015-10-13 21:15:00 +07:00
|
|
|
ret);
|
2016-09-10 08:39:03 +07:00
|
|
|
btrfs_end_transaction(trans);
|
2012-03-12 22:03:00 +07:00
|
|
|
goto out;
|
|
|
|
|
}
|
2008-11-13 02:32:25 +07:00
|
|
|
leaf = path->nodes[0];
|
|
|
|
|
slot = path->slots[0];
|
2008-06-12 08:53:53 +07:00
|
|
|
}
|
2008-11-13 02:32:25 +07:00
|
|
|
|
2014-06-09 09:48:05 +07:00
|
|
|
/* If we have an implicit hole (NO_HOLES feature). */
|
|
|
|
|
if (drop_start < new_key.offset)
|
2017-02-20 18:51:04 +07:00
|
|
|
clone_update_extent_map(BTRFS_I(inode), trans,
|
2014-06-30 03:45:40 +07:00
|
|
|
NULL, drop_start,
|
2014-06-09 09:48:05 +07:00
|
|
|
new_key.offset - drop_start);
|
|
|
|
|
|
2017-02-20 18:51:04 +07:00
|
|
|
clone_update_extent_map(BTRFS_I(inode), trans,
|
|
|
|
|
path, 0, 0);
|
2014-06-09 09:48:05 +07:00
|
|
|
|
2008-11-13 02:32:25 +07:00
|
|
|
btrfs_mark_buffer_dirty(leaf);
|
2011-04-21 06:20:15 +07:00
|
|
|
btrfs_release_path(path);
|
2008-11-13 02:32:25 +07:00
|
|
|
|
2014-08-08 08:47:06 +07:00
|
|
|
last_dest_end = ALIGN(new_key.offset + datal,
|
2016-06-23 05:54:23 +07:00
|
|
|
fs_info->sectorsize);
|
2014-06-01 07:50:28 +07:00
|
|
|
ret = clone_finish_inode_update(trans, inode,
|
|
|
|
|
last_dest_end,
|
2015-07-01 04:42:08 +07:00
|
|
|
destoff, olen,
|
|
|
|
|
no_time_update);
|
2014-06-01 07:50:28 +07:00
|
|
|
if (ret)
|
2012-03-12 22:03:00 +07:00
|
|
|
goto out;
|
2014-05-31 08:31:05 +07:00
|
|
|
if (new_key.offset + datal >= destoff + len)
|
|
|
|
|
break;
|
2010-05-16 21:48:46 +07:00
|
|
|
}
|
2011-04-21 06:20:15 +07:00
|
|
|
btrfs_release_path(path);
|
2015-03-31 20:56:46 +07:00
|
|
|
key.offset = next_key_min_offset;
|
2016-10-13 08:23:39 +07:00
|
|
|
|
|
|
|
|
if (fatal_signal_pending(current)) {
|
|
|
|
|
ret = -EINTR;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
2008-06-12 08:53:53 +07:00
|
|
|
}
|
|
|
|
|
ret = 0;
|
2013-08-07 01:42:49 +07:00
|
|
|
|
2014-06-01 07:50:28 +07:00
|
|
|
if (last_dest_end < destoff + len) {
|
|
|
|
|
/*
|
|
|
|
|
* We have an implicit hole (NO_HOLES feature is enabled) that
|
|
|
|
|
* fully or partially overlaps our cloning range at its end.
|
|
|
|
|
*/
|
|
|
|
|
btrfs_release_path(path);
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* 1 - remove extent(s)
|
|
|
|
|
* 1 - inode update
|
|
|
|
|
*/
|
|
|
|
|
trans = btrfs_start_transaction(root, 2);
|
|
|
|
|
if (IS_ERR(trans)) {
|
|
|
|
|
ret = PTR_ERR(trans);
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
ret = btrfs_drop_extents(trans, root, inode,
|
|
|
|
|
last_dest_end, destoff + len, 1);
|
|
|
|
|
if (ret) {
|
|
|
|
|
if (ret != -EOPNOTSUPP)
|
2016-06-11 05:19:25 +07:00
|
|
|
btrfs_abort_transaction(trans, ret);
|
2016-09-10 08:39:03 +07:00
|
|
|
btrfs_end_transaction(trans);
|
2014-06-01 07:50:28 +07:00
|
|
|
goto out;
|
|
|
|
|
}
|
2017-02-20 18:51:04 +07:00
|
|
|
clone_update_extent_map(BTRFS_I(inode), trans, NULL,
|
|
|
|
|
last_dest_end,
|
|
|
|
|
destoff + len - last_dest_end);
|
2014-06-01 07:50:28 +07:00
|
|
|
ret = clone_finish_inode_update(trans, inode, destoff + len,
|
2015-07-01 04:42:08 +07:00
|
|
|
destoff, olen, no_time_update);
|
2014-06-01 07:50:28 +07:00
|
|
|
}
|
|
|
|
|
|
2008-06-12 08:53:53 +07:00
|
|
|
out:
|
2013-08-07 01:42:49 +07:00
|
|
|
btrfs_free_path(path);
|
2016-04-11 23:40:08 +07:00
|
|
|
kvfree(buf);
|
2013-08-07 01:42:49 +07:00
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
2015-11-11 04:53:32 +07:00
|
|
|
static noinline int btrfs_clone_files(struct file *file, struct file *file_src,
|
|
|
|
|
u64 off, u64 olen, u64 destoff)
|
2013-08-07 01:42:49 +07:00
|
|
|
{
|
2013-09-02 02:57:51 +07:00
|
|
|
struct inode *inode = file_inode(file);
|
2015-11-11 04:53:32 +07:00
|
|
|
struct inode *src = file_inode(file_src);
|
2016-06-23 05:54:23 +07:00
|
|
|
struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
|
2013-08-07 01:42:49 +07:00
|
|
|
struct btrfs_root *root = BTRFS_I(inode)->root;
|
|
|
|
|
int ret;
|
|
|
|
|
u64 len = olen;
|
2016-06-23 05:54:23 +07:00
|
|
|
u64 bs = fs_info->sb->s_blocksize;
|
2015-11-11 04:53:32 +07:00
|
|
|
int same_inode = src == inode;
|
2013-08-07 01:42:49 +07:00
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* TODO:
|
|
|
|
|
* - split compressed inline extents. annoying: we need to
|
|
|
|
|
* decompress into destination's address_space (the file offset
|
|
|
|
|
* may change, so source mapping won't do), then recompress (or
|
|
|
|
|
* otherwise reinsert) a subrange.
|
2014-03-10 17:56:07 +07:00
|
|
|
*
|
|
|
|
|
* - split destination inode's inline extents. The inline extents can
|
|
|
|
|
* be either compressed or non-compressed.
|
2013-08-07 01:42:49 +07:00
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
if (btrfs_root_readonly(root))
|
|
|
|
|
return -EROFS;
|
|
|
|
|
|
2015-11-11 04:53:32 +07:00
|
|
|
if (file_src->f_path.mnt != file->f_path.mnt ||
|
|
|
|
|
src->i_sb != inode->i_sb)
|
|
|
|
|
return -EXDEV;
|
2013-08-07 01:42:49 +07:00
|
|
|
|
|
|
|
|
/* don't make the dst file partly checksummed */
|
|
|
|
|
if ((BTRFS_I(src)->flags & BTRFS_INODE_NODATASUM) !=
|
|
|
|
|
(BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM))
|
2015-11-11 04:53:32 +07:00
|
|
|
return -EINVAL;
|
2013-08-07 01:42:49 +07:00
|
|
|
|
|
|
|
|
if (S_ISDIR(src->i_mode) || S_ISDIR(inode->i_mode))
|
2015-11-11 04:53:32 +07:00
|
|
|
return -EISDIR;
|
2013-08-07 01:42:49 +07:00
|
|
|
|
|
|
|
|
if (!same_inode) {
|
2015-07-01 04:42:06 +07:00
|
|
|
btrfs_double_inode_lock(src, inode);
|
2013-08-07 01:42:49 +07:00
|
|
|
} else {
|
2016-01-23 03:40:57 +07:00
|
|
|
inode_lock(src);
|
2013-08-07 01:42:49 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* determine range to clone */
|
|
|
|
|
ret = -EINVAL;
|
|
|
|
|
if (off + len > src->i_size || off + len < off)
|
|
|
|
|
goto out_unlock;
|
|
|
|
|
if (len == 0)
|
|
|
|
|
olen = len = src->i_size - off;
|
|
|
|
|
/* if we extend to eof, continue to block boundary */
|
|
|
|
|
if (off + len == src->i_size)
|
|
|
|
|
len = ALIGN(src->i_size, bs) - off;
|
|
|
|
|
|
2015-03-31 00:23:59 +07:00
|
|
|
if (len == 0) {
|
|
|
|
|
ret = 0;
|
|
|
|
|
goto out_unlock;
|
|
|
|
|
}
|
|
|
|
|
|
2013-08-07 01:42:49 +07:00
|
|
|
/* verify the end result is block aligned */
|
|
|
|
|
if (!IS_ALIGNED(off, bs) || !IS_ALIGNED(off + len, bs) ||
|
|
|
|
|
!IS_ALIGNED(destoff, bs))
|
|
|
|
|
goto out_unlock;
|
|
|
|
|
|
|
|
|
|
/* verify if ranges are overlapped within the same file */
|
|
|
|
|
if (same_inode) {
|
|
|
|
|
if (destoff + len > off && destoff < off + len)
|
|
|
|
|
goto out_unlock;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (destoff > inode->i_size) {
|
|
|
|
|
ret = btrfs_cont_expand(inode, inode->i_size, destoff);
|
|
|
|
|
if (ret)
|
|
|
|
|
goto out_unlock;
|
|
|
|
|
}
|
|
|
|
|
|
Btrfs: ensure readers see new data after a clone operation
We were cleaning the clone target file range from the page cache before
we did replace the file extent items in the fs tree. This was racy,
as right after cleaning the relevant range from the page cache and before
replacing the file extent items, a read against that range could be
performed by another task and populate again the page cache with stale
data (stale after the cloning finishes). This would result in reads after
the clone operation successfully finishes to get old data (and potentially
for a very long time). Therefore evict the pages after replacing the file
extent items, so that subsequent reads will always get the new data.
Similarly, we were prone to races while cloning the file extent items
because we weren't locking the target range and wait for any existing
ordered extents against that range to complete. It was possible that
after cloning the extent items, a write operation that was performed
before the clone operation and overlaps the same range, would end up
undoing all or part of the work the clone operation did (a worker task
running inode.c:btrfs_finish_ordered_io). Therefore lock the target
range in the io tree, wait for all pending ordered extents against that
range to finish and then safely perform the cloning.
The issue of reading stale data after the clone operation is easy to
reproduce by running the following C program in a loop until it exits
with return value 1.
#include <unistd.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <errno.h>
#include <pthread.h>
#include <fcntl.h>
#include <assert.h>
#include <asm/types.h>
#include <linux/ioctl.h>
#include <sys/stat.h>
#include <sys/types.h>
#include <sys/ioctl.h>
#define SRC_FILE "/mnt/sdd/foo"
#define DST_FILE "/mnt/sdd/bar"
#define FILE_SIZE (16 * 1024)
#define PATTERN_SRC 'X'
#define PATTERN_DST 'Y'
struct btrfs_ioctl_clone_range_args {
__s64 src_fd;
__u64 src_offset, src_length;
__u64 dest_offset;
};
#define BTRFS_IOCTL_MAGIC 0x94
#define BTRFS_IOC_CLONE_RANGE _IOW(BTRFS_IOCTL_MAGIC, 13, \
struct btrfs_ioctl_clone_range_args)
static pthread_mutex_t mutex = PTHREAD_MUTEX_INITIALIZER;
static int clone_done = 0;
static int reader_ready = 0;
static int stale_data = 0;
static void *reader_loop(void *arg)
{
char buf[4096], want_buf[4096];
memset(want_buf, PATTERN_SRC, 4096);
pthread_mutex_lock(&mutex);
reader_ready = 1;
pthread_mutex_unlock(&mutex);
while (1) {
int done, fd, ret;
fd = open(DST_FILE, O_RDONLY);
assert(fd != -1);
pthread_mutex_lock(&mutex);
done = clone_done;
pthread_mutex_unlock(&mutex);
ret = read(fd, buf, 4096);
assert(ret == 4096);
close(fd);
if (done) {
ret = memcmp(buf, want_buf, 4096);
if (ret == 0) {
printf("Found new content\n");
} else {
printf("Found old content\n");
pthread_mutex_lock(&mutex);
stale_data = 1;
pthread_mutex_unlock(&mutex);
}
break;
}
}
return NULL;
}
int main(int argc, char *argv[])
{
pthread_t reader;
int ret, i, fd;
struct btrfs_ioctl_clone_range_args clone_args;
int fd1, fd2;
ret = remove(SRC_FILE);
if (ret == -1 && errno != ENOENT) {
fprintf(stderr, "Error deleting src file: %s\n", strerror(errno));
return 1;
}
ret = remove(DST_FILE);
if (ret == -1 && errno != ENOENT) {
fprintf(stderr, "Error deleting dst file: %s\n", strerror(errno));
return 1;
}
fd = open(SRC_FILE, O_CREAT | O_WRONLY | O_TRUNC, S_IRWXU);
assert(fd != -1);
for (i = 0; i < FILE_SIZE; i++) {
char c = PATTERN_SRC;
ret = write(fd, &c, 1);
assert(ret == 1);
}
close(fd);
fd = open(DST_FILE, O_CREAT | O_WRONLY | O_TRUNC, S_IRWXU);
assert(fd != -1);
for (i = 0; i < FILE_SIZE; i++) {
char c = PATTERN_DST;
ret = write(fd, &c, 1);
assert(ret == 1);
}
close(fd);
sync();
ret = pthread_create(&reader, NULL, reader_loop, NULL);
assert(ret == 0);
while (1) {
int r;
pthread_mutex_lock(&mutex);
r = reader_ready;
pthread_mutex_unlock(&mutex);
if (r) break;
}
fd1 = open(SRC_FILE, O_RDONLY);
if (fd1 < 0) {
fprintf(stderr, "Error open src file: %s\n", strerror(errno));
return 1;
}
fd2 = open(DST_FILE, O_RDWR);
if (fd2 < 0) {
fprintf(stderr, "Error open dst file: %s\n", strerror(errno));
return 1;
}
clone_args.src_fd = fd1;
clone_args.src_offset = 0;
clone_args.src_length = 4096;
clone_args.dest_offset = 0;
ret = ioctl(fd2, BTRFS_IOC_CLONE_RANGE, &clone_args);
assert(ret == 0);
close(fd1);
close(fd2);
pthread_mutex_lock(&mutex);
clone_done = 1;
pthread_mutex_unlock(&mutex);
ret = pthread_join(reader, NULL);
assert(ret == 0);
pthread_mutex_lock(&mutex);
ret = stale_data ? 1 : 0;
pthread_mutex_unlock(&mutex);
return ret;
}
Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com>
Signed-off-by: Chris Mason <clm@fb.com>
2014-05-23 11:03:34 +07:00
|
|
|
/*
|
|
|
|
|
* Lock the target range too. Right after we replace the file extent
|
|
|
|
|
* items in the fs tree (which now point to the cloned data), we might
|
|
|
|
|
* have a worker replace them with extent items relative to a write
|
|
|
|
|
* operation that was issued before this clone operation (i.e. confront
|
|
|
|
|
* with inode.c:btrfs_finish_ordered_io).
|
|
|
|
|
*/
|
|
|
|
|
if (same_inode) {
|
|
|
|
|
u64 lock_start = min_t(u64, off, destoff);
|
|
|
|
|
u64 lock_len = max_t(u64, off, destoff) + len - lock_start;
|
2013-08-07 01:42:49 +07:00
|
|
|
|
2016-01-27 17:20:58 +07:00
|
|
|
ret = lock_extent_range(src, lock_start, lock_len, true);
|
Btrfs: ensure readers see new data after a clone operation
We were cleaning the clone target file range from the page cache before
we did replace the file extent items in the fs tree. This was racy,
as right after cleaning the relevant range from the page cache and before
replacing the file extent items, a read against that range could be
performed by another task and populate again the page cache with stale
data (stale after the cloning finishes). This would result in reads after
the clone operation successfully finishes to get old data (and potentially
for a very long time). Therefore evict the pages after replacing the file
extent items, so that subsequent reads will always get the new data.
Similarly, we were prone to races while cloning the file extent items
because we weren't locking the target range and wait for any existing
ordered extents against that range to complete. It was possible that
after cloning the extent items, a write operation that was performed
before the clone operation and overlaps the same range, would end up
undoing all or part of the work the clone operation did (a worker task
running inode.c:btrfs_finish_ordered_io). Therefore lock the target
range in the io tree, wait for all pending ordered extents against that
range to finish and then safely perform the cloning.
The issue of reading stale data after the clone operation is easy to
reproduce by running the following C program in a loop until it exits
with return value 1.
#include <unistd.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <errno.h>
#include <pthread.h>
#include <fcntl.h>
#include <assert.h>
#include <asm/types.h>
#include <linux/ioctl.h>
#include <sys/stat.h>
#include <sys/types.h>
#include <sys/ioctl.h>
#define SRC_FILE "/mnt/sdd/foo"
#define DST_FILE "/mnt/sdd/bar"
#define FILE_SIZE (16 * 1024)
#define PATTERN_SRC 'X'
#define PATTERN_DST 'Y'
struct btrfs_ioctl_clone_range_args {
__s64 src_fd;
__u64 src_offset, src_length;
__u64 dest_offset;
};
#define BTRFS_IOCTL_MAGIC 0x94
#define BTRFS_IOC_CLONE_RANGE _IOW(BTRFS_IOCTL_MAGIC, 13, \
struct btrfs_ioctl_clone_range_args)
static pthread_mutex_t mutex = PTHREAD_MUTEX_INITIALIZER;
static int clone_done = 0;
static int reader_ready = 0;
static int stale_data = 0;
static void *reader_loop(void *arg)
{
char buf[4096], want_buf[4096];
memset(want_buf, PATTERN_SRC, 4096);
pthread_mutex_lock(&mutex);
reader_ready = 1;
pthread_mutex_unlock(&mutex);
while (1) {
int done, fd, ret;
fd = open(DST_FILE, O_RDONLY);
assert(fd != -1);
pthread_mutex_lock(&mutex);
done = clone_done;
pthread_mutex_unlock(&mutex);
ret = read(fd, buf, 4096);
assert(ret == 4096);
close(fd);
if (done) {
ret = memcmp(buf, want_buf, 4096);
if (ret == 0) {
printf("Found new content\n");
} else {
printf("Found old content\n");
pthread_mutex_lock(&mutex);
stale_data = 1;
pthread_mutex_unlock(&mutex);
}
break;
}
}
return NULL;
}
int main(int argc, char *argv[])
{
pthread_t reader;
int ret, i, fd;
struct btrfs_ioctl_clone_range_args clone_args;
int fd1, fd2;
ret = remove(SRC_FILE);
if (ret == -1 && errno != ENOENT) {
fprintf(stderr, "Error deleting src file: %s\n", strerror(errno));
return 1;
}
ret = remove(DST_FILE);
if (ret == -1 && errno != ENOENT) {
fprintf(stderr, "Error deleting dst file: %s\n", strerror(errno));
return 1;
}
fd = open(SRC_FILE, O_CREAT | O_WRONLY | O_TRUNC, S_IRWXU);
assert(fd != -1);
for (i = 0; i < FILE_SIZE; i++) {
char c = PATTERN_SRC;
ret = write(fd, &c, 1);
assert(ret == 1);
}
close(fd);
fd = open(DST_FILE, O_CREAT | O_WRONLY | O_TRUNC, S_IRWXU);
assert(fd != -1);
for (i = 0; i < FILE_SIZE; i++) {
char c = PATTERN_DST;
ret = write(fd, &c, 1);
assert(ret == 1);
}
close(fd);
sync();
ret = pthread_create(&reader, NULL, reader_loop, NULL);
assert(ret == 0);
while (1) {
int r;
pthread_mutex_lock(&mutex);
r = reader_ready;
pthread_mutex_unlock(&mutex);
if (r) break;
}
fd1 = open(SRC_FILE, O_RDONLY);
if (fd1 < 0) {
fprintf(stderr, "Error open src file: %s\n", strerror(errno));
return 1;
}
fd2 = open(DST_FILE, O_RDWR);
if (fd2 < 0) {
fprintf(stderr, "Error open dst file: %s\n", strerror(errno));
return 1;
}
clone_args.src_fd = fd1;
clone_args.src_offset = 0;
clone_args.src_length = 4096;
clone_args.dest_offset = 0;
ret = ioctl(fd2, BTRFS_IOC_CLONE_RANGE, &clone_args);
assert(ret == 0);
close(fd1);
close(fd2);
pthread_mutex_lock(&mutex);
clone_done = 1;
pthread_mutex_unlock(&mutex);
ret = pthread_join(reader, NULL);
assert(ret == 0);
pthread_mutex_lock(&mutex);
ret = stale_data ? 1 : 0;
pthread_mutex_unlock(&mutex);
return ret;
}
Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com>
Signed-off-by: Chris Mason <clm@fb.com>
2014-05-23 11:03:34 +07:00
|
|
|
} else {
|
2016-01-27 17:20:58 +07:00
|
|
|
ret = btrfs_double_extent_lock(src, off, inode, destoff, len,
|
|
|
|
|
true);
|
|
|
|
|
}
|
|
|
|
|
ASSERT(ret == 0);
|
|
|
|
|
if (WARN_ON(ret)) {
|
|
|
|
|
/* ranges in the io trees already unlocked */
|
|
|
|
|
goto out_unlock;
|
Btrfs: ensure readers see new data after a clone operation
We were cleaning the clone target file range from the page cache before
we did replace the file extent items in the fs tree. This was racy,
as right after cleaning the relevant range from the page cache and before
replacing the file extent items, a read against that range could be
performed by another task and populate again the page cache with stale
data (stale after the cloning finishes). This would result in reads after
the clone operation successfully finishes to get old data (and potentially
for a very long time). Therefore evict the pages after replacing the file
extent items, so that subsequent reads will always get the new data.
Similarly, we were prone to races while cloning the file extent items
because we weren't locking the target range and wait for any existing
ordered extents against that range to complete. It was possible that
after cloning the extent items, a write operation that was performed
before the clone operation and overlaps the same range, would end up
undoing all or part of the work the clone operation did (a worker task
running inode.c:btrfs_finish_ordered_io). Therefore lock the target
range in the io tree, wait for all pending ordered extents against that
range to finish and then safely perform the cloning.
The issue of reading stale data after the clone operation is easy to
reproduce by running the following C program in a loop until it exits
with return value 1.
#include <unistd.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <errno.h>
#include <pthread.h>
#include <fcntl.h>
#include <assert.h>
#include <asm/types.h>
#include <linux/ioctl.h>
#include <sys/stat.h>
#include <sys/types.h>
#include <sys/ioctl.h>
#define SRC_FILE "/mnt/sdd/foo"
#define DST_FILE "/mnt/sdd/bar"
#define FILE_SIZE (16 * 1024)
#define PATTERN_SRC 'X'
#define PATTERN_DST 'Y'
struct btrfs_ioctl_clone_range_args {
__s64 src_fd;
__u64 src_offset, src_length;
__u64 dest_offset;
};
#define BTRFS_IOCTL_MAGIC 0x94
#define BTRFS_IOC_CLONE_RANGE _IOW(BTRFS_IOCTL_MAGIC, 13, \
struct btrfs_ioctl_clone_range_args)
static pthread_mutex_t mutex = PTHREAD_MUTEX_INITIALIZER;
static int clone_done = 0;
static int reader_ready = 0;
static int stale_data = 0;
static void *reader_loop(void *arg)
{
char buf[4096], want_buf[4096];
memset(want_buf, PATTERN_SRC, 4096);
pthread_mutex_lock(&mutex);
reader_ready = 1;
pthread_mutex_unlock(&mutex);
while (1) {
int done, fd, ret;
fd = open(DST_FILE, O_RDONLY);
assert(fd != -1);
pthread_mutex_lock(&mutex);
done = clone_done;
pthread_mutex_unlock(&mutex);
ret = read(fd, buf, 4096);
assert(ret == 4096);
close(fd);
if (done) {
ret = memcmp(buf, want_buf, 4096);
if (ret == 0) {
printf("Found new content\n");
} else {
printf("Found old content\n");
pthread_mutex_lock(&mutex);
stale_data = 1;
pthread_mutex_unlock(&mutex);
}
break;
}
}
return NULL;
}
int main(int argc, char *argv[])
{
pthread_t reader;
int ret, i, fd;
struct btrfs_ioctl_clone_range_args clone_args;
int fd1, fd2;
ret = remove(SRC_FILE);
if (ret == -1 && errno != ENOENT) {
fprintf(stderr, "Error deleting src file: %s\n", strerror(errno));
return 1;
}
ret = remove(DST_FILE);
if (ret == -1 && errno != ENOENT) {
fprintf(stderr, "Error deleting dst file: %s\n", strerror(errno));
return 1;
}
fd = open(SRC_FILE, O_CREAT | O_WRONLY | O_TRUNC, S_IRWXU);
assert(fd != -1);
for (i = 0; i < FILE_SIZE; i++) {
char c = PATTERN_SRC;
ret = write(fd, &c, 1);
assert(ret == 1);
}
close(fd);
fd = open(DST_FILE, O_CREAT | O_WRONLY | O_TRUNC, S_IRWXU);
assert(fd != -1);
for (i = 0; i < FILE_SIZE; i++) {
char c = PATTERN_DST;
ret = write(fd, &c, 1);
assert(ret == 1);
}
close(fd);
sync();
ret = pthread_create(&reader, NULL, reader_loop, NULL);
assert(ret == 0);
while (1) {
int r;
pthread_mutex_lock(&mutex);
r = reader_ready;
pthread_mutex_unlock(&mutex);
if (r) break;
}
fd1 = open(SRC_FILE, O_RDONLY);
if (fd1 < 0) {
fprintf(stderr, "Error open src file: %s\n", strerror(errno));
return 1;
}
fd2 = open(DST_FILE, O_RDWR);
if (fd2 < 0) {
fprintf(stderr, "Error open dst file: %s\n", strerror(errno));
return 1;
}
clone_args.src_fd = fd1;
clone_args.src_offset = 0;
clone_args.src_length = 4096;
clone_args.dest_offset = 0;
ret = ioctl(fd2, BTRFS_IOC_CLONE_RANGE, &clone_args);
assert(ret == 0);
close(fd1);
close(fd2);
pthread_mutex_lock(&mutex);
clone_done = 1;
pthread_mutex_unlock(&mutex);
ret = pthread_join(reader, NULL);
assert(ret == 0);
pthread_mutex_lock(&mutex);
ret = stale_data ? 1 : 0;
pthread_mutex_unlock(&mutex);
return ret;
}
Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com>
Signed-off-by: Chris Mason <clm@fb.com>
2014-05-23 11:03:34 +07:00
|
|
|
}
|
2013-08-07 01:42:49 +07:00
|
|
|
|
2015-07-01 04:42:08 +07:00
|
|
|
ret = btrfs_clone(src, inode, off, olen, len, destoff, 0);
|
2013-08-07 01:42:49 +07:00
|
|
|
|
Btrfs: ensure readers see new data after a clone operation
We were cleaning the clone target file range from the page cache before
we did replace the file extent items in the fs tree. This was racy,
as right after cleaning the relevant range from the page cache and before
replacing the file extent items, a read against that range could be
performed by another task and populate again the page cache with stale
data (stale after the cloning finishes). This would result in reads after
the clone operation successfully finishes to get old data (and potentially
for a very long time). Therefore evict the pages after replacing the file
extent items, so that subsequent reads will always get the new data.
Similarly, we were prone to races while cloning the file extent items
because we weren't locking the target range and wait for any existing
ordered extents against that range to complete. It was possible that
after cloning the extent items, a write operation that was performed
before the clone operation and overlaps the same range, would end up
undoing all or part of the work the clone operation did (a worker task
running inode.c:btrfs_finish_ordered_io). Therefore lock the target
range in the io tree, wait for all pending ordered extents against that
range to finish and then safely perform the cloning.
The issue of reading stale data after the clone operation is easy to
reproduce by running the following C program in a loop until it exits
with return value 1.
#include <unistd.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <errno.h>
#include <pthread.h>
#include <fcntl.h>
#include <assert.h>
#include <asm/types.h>
#include <linux/ioctl.h>
#include <sys/stat.h>
#include <sys/types.h>
#include <sys/ioctl.h>
#define SRC_FILE "/mnt/sdd/foo"
#define DST_FILE "/mnt/sdd/bar"
#define FILE_SIZE (16 * 1024)
#define PATTERN_SRC 'X'
#define PATTERN_DST 'Y'
struct btrfs_ioctl_clone_range_args {
__s64 src_fd;
__u64 src_offset, src_length;
__u64 dest_offset;
};
#define BTRFS_IOCTL_MAGIC 0x94
#define BTRFS_IOC_CLONE_RANGE _IOW(BTRFS_IOCTL_MAGIC, 13, \
struct btrfs_ioctl_clone_range_args)
static pthread_mutex_t mutex = PTHREAD_MUTEX_INITIALIZER;
static int clone_done = 0;
static int reader_ready = 0;
static int stale_data = 0;
static void *reader_loop(void *arg)
{
char buf[4096], want_buf[4096];
memset(want_buf, PATTERN_SRC, 4096);
pthread_mutex_lock(&mutex);
reader_ready = 1;
pthread_mutex_unlock(&mutex);
while (1) {
int done, fd, ret;
fd = open(DST_FILE, O_RDONLY);
assert(fd != -1);
pthread_mutex_lock(&mutex);
done = clone_done;
pthread_mutex_unlock(&mutex);
ret = read(fd, buf, 4096);
assert(ret == 4096);
close(fd);
if (done) {
ret = memcmp(buf, want_buf, 4096);
if (ret == 0) {
printf("Found new content\n");
} else {
printf("Found old content\n");
pthread_mutex_lock(&mutex);
stale_data = 1;
pthread_mutex_unlock(&mutex);
}
break;
}
}
return NULL;
}
int main(int argc, char *argv[])
{
pthread_t reader;
int ret, i, fd;
struct btrfs_ioctl_clone_range_args clone_args;
int fd1, fd2;
ret = remove(SRC_FILE);
if (ret == -1 && errno != ENOENT) {
fprintf(stderr, "Error deleting src file: %s\n", strerror(errno));
return 1;
}
ret = remove(DST_FILE);
if (ret == -1 && errno != ENOENT) {
fprintf(stderr, "Error deleting dst file: %s\n", strerror(errno));
return 1;
}
fd = open(SRC_FILE, O_CREAT | O_WRONLY | O_TRUNC, S_IRWXU);
assert(fd != -1);
for (i = 0; i < FILE_SIZE; i++) {
char c = PATTERN_SRC;
ret = write(fd, &c, 1);
assert(ret == 1);
}
close(fd);
fd = open(DST_FILE, O_CREAT | O_WRONLY | O_TRUNC, S_IRWXU);
assert(fd != -1);
for (i = 0; i < FILE_SIZE; i++) {
char c = PATTERN_DST;
ret = write(fd, &c, 1);
assert(ret == 1);
}
close(fd);
sync();
ret = pthread_create(&reader, NULL, reader_loop, NULL);
assert(ret == 0);
while (1) {
int r;
pthread_mutex_lock(&mutex);
r = reader_ready;
pthread_mutex_unlock(&mutex);
if (r) break;
}
fd1 = open(SRC_FILE, O_RDONLY);
if (fd1 < 0) {
fprintf(stderr, "Error open src file: %s\n", strerror(errno));
return 1;
}
fd2 = open(DST_FILE, O_RDWR);
if (fd2 < 0) {
fprintf(stderr, "Error open dst file: %s\n", strerror(errno));
return 1;
}
clone_args.src_fd = fd1;
clone_args.src_offset = 0;
clone_args.src_length = 4096;
clone_args.dest_offset = 0;
ret = ioctl(fd2, BTRFS_IOC_CLONE_RANGE, &clone_args);
assert(ret == 0);
close(fd1);
close(fd2);
pthread_mutex_lock(&mutex);
clone_done = 1;
pthread_mutex_unlock(&mutex);
ret = pthread_join(reader, NULL);
assert(ret == 0);
pthread_mutex_lock(&mutex);
ret = stale_data ? 1 : 0;
pthread_mutex_unlock(&mutex);
return ret;
}
Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com>
Signed-off-by: Chris Mason <clm@fb.com>
2014-05-23 11:03:34 +07:00
|
|
|
if (same_inode) {
|
|
|
|
|
u64 lock_start = min_t(u64, off, destoff);
|
|
|
|
|
u64 lock_end = max_t(u64, off, destoff) + len - 1;
|
|
|
|
|
|
|
|
|
|
unlock_extent(&BTRFS_I(src)->io_tree, lock_start, lock_end);
|
|
|
|
|
} else {
|
2015-07-01 04:42:06 +07:00
|
|
|
btrfs_double_extent_unlock(src, off, inode, destoff, len);
|
Btrfs: ensure readers see new data after a clone operation
We were cleaning the clone target file range from the page cache before
we did replace the file extent items in the fs tree. This was racy,
as right after cleaning the relevant range from the page cache and before
replacing the file extent items, a read against that range could be
performed by another task and populate again the page cache with stale
data (stale after the cloning finishes). This would result in reads after
the clone operation successfully finishes to get old data (and potentially
for a very long time). Therefore evict the pages after replacing the file
extent items, so that subsequent reads will always get the new data.
Similarly, we were prone to races while cloning the file extent items
because we weren't locking the target range and wait for any existing
ordered extents against that range to complete. It was possible that
after cloning the extent items, a write operation that was performed
before the clone operation and overlaps the same range, would end up
undoing all or part of the work the clone operation did (a worker task
running inode.c:btrfs_finish_ordered_io). Therefore lock the target
range in the io tree, wait for all pending ordered extents against that
range to finish and then safely perform the cloning.
The issue of reading stale data after the clone operation is easy to
reproduce by running the following C program in a loop until it exits
with return value 1.
#include <unistd.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <errno.h>
#include <pthread.h>
#include <fcntl.h>
#include <assert.h>
#include <asm/types.h>
#include <linux/ioctl.h>
#include <sys/stat.h>
#include <sys/types.h>
#include <sys/ioctl.h>
#define SRC_FILE "/mnt/sdd/foo"
#define DST_FILE "/mnt/sdd/bar"
#define FILE_SIZE (16 * 1024)
#define PATTERN_SRC 'X'
#define PATTERN_DST 'Y'
struct btrfs_ioctl_clone_range_args {
__s64 src_fd;
__u64 src_offset, src_length;
__u64 dest_offset;
};
#define BTRFS_IOCTL_MAGIC 0x94
#define BTRFS_IOC_CLONE_RANGE _IOW(BTRFS_IOCTL_MAGIC, 13, \
struct btrfs_ioctl_clone_range_args)
static pthread_mutex_t mutex = PTHREAD_MUTEX_INITIALIZER;
static int clone_done = 0;
static int reader_ready = 0;
static int stale_data = 0;
static void *reader_loop(void *arg)
{
char buf[4096], want_buf[4096];
memset(want_buf, PATTERN_SRC, 4096);
pthread_mutex_lock(&mutex);
reader_ready = 1;
pthread_mutex_unlock(&mutex);
while (1) {
int done, fd, ret;
fd = open(DST_FILE, O_RDONLY);
assert(fd != -1);
pthread_mutex_lock(&mutex);
done = clone_done;
pthread_mutex_unlock(&mutex);
ret = read(fd, buf, 4096);
assert(ret == 4096);
close(fd);
if (done) {
ret = memcmp(buf, want_buf, 4096);
if (ret == 0) {
printf("Found new content\n");
} else {
printf("Found old content\n");
pthread_mutex_lock(&mutex);
stale_data = 1;
pthread_mutex_unlock(&mutex);
}
break;
}
}
return NULL;
}
int main(int argc, char *argv[])
{
pthread_t reader;
int ret, i, fd;
struct btrfs_ioctl_clone_range_args clone_args;
int fd1, fd2;
ret = remove(SRC_FILE);
if (ret == -1 && errno != ENOENT) {
fprintf(stderr, "Error deleting src file: %s\n", strerror(errno));
return 1;
}
ret = remove(DST_FILE);
if (ret == -1 && errno != ENOENT) {
fprintf(stderr, "Error deleting dst file: %s\n", strerror(errno));
return 1;
}
fd = open(SRC_FILE, O_CREAT | O_WRONLY | O_TRUNC, S_IRWXU);
assert(fd != -1);
for (i = 0; i < FILE_SIZE; i++) {
char c = PATTERN_SRC;
ret = write(fd, &c, 1);
assert(ret == 1);
}
close(fd);
fd = open(DST_FILE, O_CREAT | O_WRONLY | O_TRUNC, S_IRWXU);
assert(fd != -1);
for (i = 0; i < FILE_SIZE; i++) {
char c = PATTERN_DST;
ret = write(fd, &c, 1);
assert(ret == 1);
}
close(fd);
sync();
ret = pthread_create(&reader, NULL, reader_loop, NULL);
assert(ret == 0);
while (1) {
int r;
pthread_mutex_lock(&mutex);
r = reader_ready;
pthread_mutex_unlock(&mutex);
if (r) break;
}
fd1 = open(SRC_FILE, O_RDONLY);
if (fd1 < 0) {
fprintf(stderr, "Error open src file: %s\n", strerror(errno));
return 1;
}
fd2 = open(DST_FILE, O_RDWR);
if (fd2 < 0) {
fprintf(stderr, "Error open dst file: %s\n", strerror(errno));
return 1;
}
clone_args.src_fd = fd1;
clone_args.src_offset = 0;
clone_args.src_length = 4096;
clone_args.dest_offset = 0;
ret = ioctl(fd2, BTRFS_IOC_CLONE_RANGE, &clone_args);
assert(ret == 0);
close(fd1);
close(fd2);
pthread_mutex_lock(&mutex);
clone_done = 1;
pthread_mutex_unlock(&mutex);
ret = pthread_join(reader, NULL);
assert(ret == 0);
pthread_mutex_lock(&mutex);
ret = stale_data ? 1 : 0;
pthread_mutex_unlock(&mutex);
return ret;
}
Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com>
Signed-off-by: Chris Mason <clm@fb.com>
2014-05-23 11:03:34 +07:00
|
|
|
}
|
|
|
|
|
/*
|
|
|
|
|
* Truncate page cache pages so that future reads will see the cloned
|
|
|
|
|
* data immediately and not the previous data.
|
|
|
|
|
*/
|
2016-01-21 17:26:04 +07:00
|
|
|
truncate_inode_pages_range(&inode->i_data,
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
round_down(destoff, PAGE_SIZE),
|
|
|
|
|
round_up(destoff + len, PAGE_SIZE) - 1);
|
2008-06-12 08:53:53 +07:00
|
|
|
out_unlock:
|
2015-07-01 04:42:06 +07:00
|
|
|
if (!same_inode)
|
|
|
|
|
btrfs_double_inode_unlock(src, inode);
|
|
|
|
|
else
|
2016-01-23 03:40:57 +07:00
|
|
|
inode_unlock(src);
|
2015-11-11 04:53:32 +07:00
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
2015-12-03 18:59:50 +07:00
|
|
|
int btrfs_clone_file_range(struct file *src_file, loff_t off,
|
|
|
|
|
struct file *dst_file, loff_t destoff, u64 len)
|
2015-11-11 04:53:32 +07:00
|
|
|
{
|
2015-12-03 18:59:50 +07:00
|
|
|
return btrfs_clone_files(dst_file, src_file, off, len, destoff);
|
2008-11-13 02:32:25 +07:00
|
|
|
}
|
|
|
|
|
|
2009-12-12 04:11:29 +07:00
|
|
|
static long btrfs_ioctl_default_subvol(struct file *file, void __user *argp)
|
|
|
|
|
{
|
2013-01-24 05:07:38 +07:00
|
|
|
struct inode *inode = file_inode(file);
|
2016-06-23 05:54:23 +07:00
|
|
|
struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
|
2009-12-12 04:11:29 +07:00
|
|
|
struct btrfs_root *root = BTRFS_I(inode)->root;
|
|
|
|
|
struct btrfs_root *new_root;
|
|
|
|
|
struct btrfs_dir_item *di;
|
|
|
|
|
struct btrfs_trans_handle *trans;
|
|
|
|
|
struct btrfs_path *path;
|
|
|
|
|
struct btrfs_key location;
|
|
|
|
|
struct btrfs_disk_key disk_key;
|
|
|
|
|
u64 objectid = 0;
|
|
|
|
|
u64 dir_id;
|
2012-11-26 15:43:07 +07:00
|
|
|
int ret;
|
2009-12-12 04:11:29 +07:00
|
|
|
|
|
|
|
|
if (!capable(CAP_SYS_ADMIN))
|
|
|
|
|
return -EPERM;
|
|
|
|
|
|
2012-11-26 15:43:07 +07:00
|
|
|
ret = mnt_want_write_file(file);
|
|
|
|
|
if (ret)
|
|
|
|
|
return ret;
|
|
|
|
|
|
|
|
|
|
if (copy_from_user(&objectid, argp, sizeof(objectid))) {
|
|
|
|
|
ret = -EFAULT;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
2009-12-12 04:11:29 +07:00
|
|
|
|
|
|
|
|
if (!objectid)
|
2013-09-13 21:04:10 +07:00
|
|
|
objectid = BTRFS_FS_TREE_OBJECTID;
|
2009-12-12 04:11:29 +07:00
|
|
|
|
|
|
|
|
location.objectid = objectid;
|
|
|
|
|
location.type = BTRFS_ROOT_ITEM_KEY;
|
|
|
|
|
location.offset = (u64)-1;
|
|
|
|
|
|
2016-06-23 05:54:23 +07:00
|
|
|
new_root = btrfs_read_fs_root_no_name(fs_info, &location);
|
2012-11-26 15:43:07 +07:00
|
|
|
if (IS_ERR(new_root)) {
|
|
|
|
|
ret = PTR_ERR(new_root);
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
2017-09-12 20:42:52 +07:00
|
|
|
if (!is_fstree(new_root->objectid)) {
|
|
|
|
|
ret = -ENOENT;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
2009-12-12 04:11:29 +07:00
|
|
|
|
|
|
|
|
path = btrfs_alloc_path();
|
2012-11-26 15:43:07 +07:00
|
|
|
if (!path) {
|
|
|
|
|
ret = -ENOMEM;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
2009-12-12 04:11:29 +07:00
|
|
|
path->leave_spinning = 1;
|
|
|
|
|
|
|
|
|
|
trans = btrfs_start_transaction(root, 1);
|
2011-01-20 13:19:37 +07:00
|
|
|
if (IS_ERR(trans)) {
|
2009-12-12 04:11:29 +07:00
|
|
|
btrfs_free_path(path);
|
2012-11-26 15:43:07 +07:00
|
|
|
ret = PTR_ERR(trans);
|
|
|
|
|
goto out;
|
2009-12-12 04:11:29 +07:00
|
|
|
}
|
|
|
|
|
|
2016-06-23 05:54:23 +07:00
|
|
|
dir_id = btrfs_super_root_dir(fs_info->super_copy);
|
|
|
|
|
di = btrfs_lookup_dir_item(trans, fs_info->tree_root, path,
|
2009-12-12 04:11:29 +07:00
|
|
|
dir_id, "default", 7, 1);
|
2010-05-29 16:47:24 +07:00
|
|
|
if (IS_ERR_OR_NULL(di)) {
|
2009-12-12 04:11:29 +07:00
|
|
|
btrfs_free_path(path);
|
2016-09-10 08:39:03 +07:00
|
|
|
btrfs_end_transaction(trans);
|
2016-06-23 05:54:23 +07:00
|
|
|
btrfs_err(fs_info,
|
2016-09-20 21:05:00 +07:00
|
|
|
"Umm, you don't have the default diritem, this isn't going to work");
|
2012-11-26 15:43:07 +07:00
|
|
|
ret = -ENOENT;
|
|
|
|
|
goto out;
|
2009-12-12 04:11:29 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
btrfs_cpu_key_to_disk(&disk_key, &new_root->root_key);
|
|
|
|
|
btrfs_set_dir_item_key(path->nodes[0], di, &disk_key);
|
|
|
|
|
btrfs_mark_buffer_dirty(path->nodes[0]);
|
|
|
|
|
btrfs_free_path(path);
|
|
|
|
|
|
2016-06-23 05:54:23 +07:00
|
|
|
btrfs_set_fs_incompat(fs_info, DEFAULT_SUBVOL);
|
2016-09-10 08:39:03 +07:00
|
|
|
btrfs_end_transaction(trans);
|
2012-11-26 15:43:07 +07:00
|
|
|
out:
|
|
|
|
|
mnt_drop_write_file(file);
|
|
|
|
|
return ret;
|
2009-12-12 04:11:29 +07:00
|
|
|
}
|
|
|
|
|
|
2018-04-02 16:24:11 +07:00
|
|
|
static void get_block_group_info(struct list_head *groups_list,
|
|
|
|
|
struct btrfs_ioctl_space_info *space)
|
2010-09-29 22:22:36 +07:00
|
|
|
{
|
|
|
|
|
struct btrfs_block_group_cache *block_group;
|
|
|
|
|
|
|
|
|
|
space->total_bytes = 0;
|
|
|
|
|
space->used_bytes = 0;
|
|
|
|
|
space->flags = 0;
|
|
|
|
|
list_for_each_entry(block_group, groups_list, list) {
|
|
|
|
|
space->flags = block_group->flags;
|
|
|
|
|
space->total_bytes += block_group->key.offset;
|
|
|
|
|
space->used_bytes +=
|
|
|
|
|
btrfs_block_group_used(&block_group->item);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2016-06-23 05:54:24 +07:00
|
|
|
static long btrfs_ioctl_space_info(struct btrfs_fs_info *fs_info,
|
|
|
|
|
void __user *arg)
|
2010-01-14 01:19:06 +07:00
|
|
|
{
|
|
|
|
|
struct btrfs_ioctl_space_args space_args;
|
|
|
|
|
struct btrfs_ioctl_space_info space;
|
|
|
|
|
struct btrfs_ioctl_space_info *dest;
|
2010-03-17 02:40:10 +07:00
|
|
|
struct btrfs_ioctl_space_info *dest_orig;
|
2011-04-11 22:56:31 +07:00
|
|
|
struct btrfs_ioctl_space_info __user *user_dest;
|
2010-01-14 01:19:06 +07:00
|
|
|
struct btrfs_space_info *info;
|
2017-09-19 22:01:23 +07:00
|
|
|
static const u64 types[] = {
|
|
|
|
|
BTRFS_BLOCK_GROUP_DATA,
|
|
|
|
|
BTRFS_BLOCK_GROUP_SYSTEM,
|
|
|
|
|
BTRFS_BLOCK_GROUP_METADATA,
|
|
|
|
|
BTRFS_BLOCK_GROUP_DATA | BTRFS_BLOCK_GROUP_METADATA
|
|
|
|
|
};
|
2010-09-29 22:22:36 +07:00
|
|
|
int num_types = 4;
|
2010-03-17 02:40:10 +07:00
|
|
|
int alloc_size;
|
2010-01-14 01:19:06 +07:00
|
|
|
int ret = 0;
|
2011-02-15 04:04:23 +07:00
|
|
|
u64 slot_count = 0;
|
2010-09-29 22:22:36 +07:00
|
|
|
int i, c;
|
2010-01-14 01:19:06 +07:00
|
|
|
|
|
|
|
|
if (copy_from_user(&space_args,
|
|
|
|
|
(struct btrfs_ioctl_space_args __user *)arg,
|
|
|
|
|
sizeof(space_args)))
|
|
|
|
|
return -EFAULT;
|
|
|
|
|
|
2010-09-29 22:22:36 +07:00
|
|
|
for (i = 0; i < num_types; i++) {
|
|
|
|
|
struct btrfs_space_info *tmp;
|
|
|
|
|
|
|
|
|
|
info = NULL;
|
|
|
|
|
rcu_read_lock();
|
2016-06-23 05:54:23 +07:00
|
|
|
list_for_each_entry_rcu(tmp, &fs_info->space_info,
|
2010-09-29 22:22:36 +07:00
|
|
|
list) {
|
|
|
|
|
if (tmp->flags == types[i]) {
|
|
|
|
|
info = tmp;
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
rcu_read_unlock();
|
|
|
|
|
|
|
|
|
|
if (!info)
|
|
|
|
|
continue;
|
|
|
|
|
|
|
|
|
|
down_read(&info->groups_sem);
|
|
|
|
|
for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) {
|
|
|
|
|
if (!list_empty(&info->block_groups[c]))
|
|
|
|
|
slot_count++;
|
|
|
|
|
}
|
|
|
|
|
up_read(&info->groups_sem);
|
|
|
|
|
}
|
2010-03-17 02:40:10 +07:00
|
|
|
|
2014-02-07 20:34:12 +07:00
|
|
|
/*
|
|
|
|
|
* Global block reserve, exported as a space_info
|
|
|
|
|
*/
|
|
|
|
|
slot_count++;
|
|
|
|
|
|
2010-03-17 02:40:10 +07:00
|
|
|
/* space_slots == 0 means they are asking for a count */
|
|
|
|
|
if (space_args.space_slots == 0) {
|
|
|
|
|
space_args.total_spaces = slot_count;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
2010-09-29 22:22:36 +07:00
|
|
|
|
2011-02-15 04:04:23 +07:00
|
|
|
slot_count = min_t(u64, space_args.space_slots, slot_count);
|
2010-09-29 22:22:36 +07:00
|
|
|
|
2010-03-17 02:40:10 +07:00
|
|
|
alloc_size = sizeof(*dest) * slot_count;
|
2010-09-29 22:22:36 +07:00
|
|
|
|
2010-03-17 02:40:10 +07:00
|
|
|
/* we generally have at most 6 or so space infos, one for each raid
|
|
|
|
|
* level. So, a whole page should be more than enough for everyone
|
|
|
|
|
*/
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 19:29:47 +07:00
|
|
|
if (alloc_size > PAGE_SIZE)
|
2010-03-17 02:40:10 +07:00
|
|
|
return -ENOMEM;
|
|
|
|
|
|
2010-01-14 01:19:06 +07:00
|
|
|
space_args.total_spaces = 0;
|
2015-11-04 21:38:29 +07:00
|
|
|
dest = kmalloc(alloc_size, GFP_KERNEL);
|
2010-03-17 02:40:10 +07:00
|
|
|
if (!dest)
|
|
|
|
|
return -ENOMEM;
|
|
|
|
|
dest_orig = dest;
|
2010-01-14 01:19:06 +07:00
|
|
|
|
2010-03-17 02:40:10 +07:00
|
|
|
/* now we have a buffer to copy into */
|
2010-09-29 22:22:36 +07:00
|
|
|
for (i = 0; i < num_types; i++) {
|
|
|
|
|
struct btrfs_space_info *tmp;
|
|
|
|
|
|
2011-02-15 04:04:23 +07:00
|
|
|
if (!slot_count)
|
|
|
|
|
break;
|
|
|
|
|
|
2010-09-29 22:22:36 +07:00
|
|
|
info = NULL;
|
|
|
|
|
rcu_read_lock();
|
2016-06-23 05:54:23 +07:00
|
|
|
list_for_each_entry_rcu(tmp, &fs_info->space_info,
|
2010-09-29 22:22:36 +07:00
|
|
|
list) {
|
|
|
|
|
if (tmp->flags == types[i]) {
|
|
|
|
|
info = tmp;
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
rcu_read_unlock();
|
2010-03-17 02:40:10 +07:00
|
|
|
|
2010-09-29 22:22:36 +07:00
|
|
|
if (!info)
|
|
|
|
|
continue;
|
|
|
|
|
down_read(&info->groups_sem);
|
|
|
|
|
for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) {
|
|
|
|
|
if (!list_empty(&info->block_groups[c])) {
|
2018-04-02 16:24:11 +07:00
|
|
|
get_block_group_info(&info->block_groups[c],
|
|
|
|
|
&space);
|
2010-09-29 22:22:36 +07:00
|
|
|
memcpy(dest, &space, sizeof(space));
|
|
|
|
|
dest++;
|
|
|
|
|
space_args.total_spaces++;
|
2011-02-15 04:04:23 +07:00
|
|
|
slot_count--;
|
2010-09-29 22:22:36 +07:00
|
|
|
}
|
2011-02-15 04:04:23 +07:00
|
|
|
if (!slot_count)
|
|
|
|
|
break;
|
2010-09-29 22:22:36 +07:00
|
|
|
}
|
|
|
|
|
up_read(&info->groups_sem);
|
2010-01-14 01:19:06 +07:00
|
|
|
}
|
|
|
|
|
|
2014-02-07 20:34:12 +07:00
|
|
|
/*
|
|
|
|
|
* Add global block reserve
|
|
|
|
|
*/
|
|
|
|
|
if (slot_count) {
|
2016-06-23 05:54:23 +07:00
|
|
|
struct btrfs_block_rsv *block_rsv = &fs_info->global_block_rsv;
|
2014-02-07 20:34:12 +07:00
|
|
|
|
|
|
|
|
spin_lock(&block_rsv->lock);
|
|
|
|
|
space.total_bytes = block_rsv->size;
|
|
|
|
|
space.used_bytes = block_rsv->size - block_rsv->reserved;
|
|
|
|
|
spin_unlock(&block_rsv->lock);
|
|
|
|
|
space.flags = BTRFS_SPACE_INFO_GLOBAL_RSV;
|
|
|
|
|
memcpy(dest, &space, sizeof(space));
|
|
|
|
|
space_args.total_spaces++;
|
|
|
|
|
}
|
|
|
|
|
|
2012-04-25 23:37:14 +07:00
|
|
|
user_dest = (struct btrfs_ioctl_space_info __user *)
|
2010-03-17 02:40:10 +07:00
|
|
|
(arg + sizeof(struct btrfs_ioctl_space_args));
|
|
|
|
|
|
|
|
|
|
if (copy_to_user(user_dest, dest_orig, alloc_size))
|
|
|
|
|
ret = -EFAULT;
|
|
|
|
|
|
|
|
|
|
kfree(dest_orig);
|
|
|
|
|
out:
|
|
|
|
|
if (ret == 0 && copy_to_user(arg, &space_args, sizeof(space_args)))
|
2010-01-14 01:19:06 +07:00
|
|
|
ret = -EFAULT;
|
|
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
2012-11-26 15:40:43 +07:00
|
|
|
static noinline long btrfs_ioctl_start_sync(struct btrfs_root *root,
|
|
|
|
|
void __user *argp)
|
Btrfs: add START_SYNC, WAIT_SYNC ioctls
START_SYNC will start a sync/commit, but not wait for it to
complete. Any modification started after the ioctl returns is
guaranteed not to be included in the commit. If a non-NULL
pointer is passed, the transaction id will be returned to
userspace.
WAIT_SYNC will wait for any in-progress commit to complete. If a
transaction id is specified, the ioctl will block and then
return (success) when the specified transaction has committed.
If it has already committed when we call the ioctl, it returns
immediately. If the specified transaction doesn't exist, it
returns EINVAL.
If no transaction id is specified, WAIT_SYNC will wait for the
currently committing transaction to finish it's commit to disk.
If there is no currently committing transaction, it returns
success.
These ioctls are useful for applications which want to impose an
ordering on when fs modifications reach disk, but do not want to
wait for the full (slow) commit process to do so.
Picky callers can take the transid returned by START_SYNC and
feed it to WAIT_SYNC, and be certain to wait only as long as
necessary for the transaction _they_ started to reach disk.
Sloppy callers can START_SYNC and WAIT_SYNC without a transid,
and provided they didn't wait too long between the calls, they
will get the same result. However, if a second commit starts
before they call WAIT_SYNC, they may end up waiting longer for
it to commit as well. Even so, a START_SYNC+WAIT_SYNC still
guarantees that any operation completed before the START_SYNC
reaches disk.
Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
2010-10-30 02:41:32 +07:00
|
|
|
{
|
|
|
|
|
struct btrfs_trans_handle *trans;
|
|
|
|
|
u64 transid;
|
2011-03-23 15:14:16 +07:00
|
|
|
int ret;
|
Btrfs: add START_SYNC, WAIT_SYNC ioctls
START_SYNC will start a sync/commit, but not wait for it to
complete. Any modification started after the ioctl returns is
guaranteed not to be included in the commit. If a non-NULL
pointer is passed, the transaction id will be returned to
userspace.
WAIT_SYNC will wait for any in-progress commit to complete. If a
transaction id is specified, the ioctl will block and then
return (success) when the specified transaction has committed.
If it has already committed when we call the ioctl, it returns
immediately. If the specified transaction doesn't exist, it
returns EINVAL.
If no transaction id is specified, WAIT_SYNC will wait for the
currently committing transaction to finish it's commit to disk.
If there is no currently committing transaction, it returns
success.
These ioctls are useful for applications which want to impose an
ordering on when fs modifications reach disk, but do not want to
wait for the full (slow) commit process to do so.
Picky callers can take the transid returned by START_SYNC and
feed it to WAIT_SYNC, and be certain to wait only as long as
necessary for the transaction _they_ started to reach disk.
Sloppy callers can START_SYNC and WAIT_SYNC without a transid,
and provided they didn't wait too long between the calls, they
will get the same result. However, if a second commit starts
before they call WAIT_SYNC, they may end up waiting longer for
it to commit as well. Even so, a START_SYNC+WAIT_SYNC still
guarantees that any operation completed before the START_SYNC
reaches disk.
Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
2010-10-30 02:41:32 +07:00
|
|
|
|
Btrfs: fix uncompleted transaction
In some cases, we need commit the current transaction, but don't want
to start a new one if there is no running transaction, so we introduce
the function - btrfs_attach_transaction(), which can catch the current
transaction, and return -ENOENT if there is no running transaction.
But no running transaction doesn't mean the current transction completely,
because we removed the running transaction before it completes. In some
cases, it doesn't matter. But in some special cases, such as freeze fs, we
hope the transaction is fully on disk, it will introduce some bugs, for
example, we may feeze the fs and dump the data in the disk, if the transction
doesn't complete, we would dump inconsistent data. So we need fix the above
problem for those cases.
We fixes this problem by introducing a function:
btrfs_attach_transaction_barrier()
if we hope all the transaction is fully on the disk, even they are not
running, we can use this function.
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-02-20 16:17:06 +07:00
|
|
|
trans = btrfs_attach_transaction_barrier(root);
|
2012-11-26 15:41:29 +07:00
|
|
|
if (IS_ERR(trans)) {
|
|
|
|
|
if (PTR_ERR(trans) != -ENOENT)
|
|
|
|
|
return PTR_ERR(trans);
|
|
|
|
|
|
|
|
|
|
/* No running transaction, don't bother */
|
|
|
|
|
transid = root->fs_info->last_trans_committed;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
Btrfs: add START_SYNC, WAIT_SYNC ioctls
START_SYNC will start a sync/commit, but not wait for it to
complete. Any modification started after the ioctl returns is
guaranteed not to be included in the commit. If a non-NULL
pointer is passed, the transaction id will be returned to
userspace.
WAIT_SYNC will wait for any in-progress commit to complete. If a
transaction id is specified, the ioctl will block and then
return (success) when the specified transaction has committed.
If it has already committed when we call the ioctl, it returns
immediately. If the specified transaction doesn't exist, it
returns EINVAL.
If no transaction id is specified, WAIT_SYNC will wait for the
currently committing transaction to finish it's commit to disk.
If there is no currently committing transaction, it returns
success.
These ioctls are useful for applications which want to impose an
ordering on when fs modifications reach disk, but do not want to
wait for the full (slow) commit process to do so.
Picky callers can take the transid returned by START_SYNC and
feed it to WAIT_SYNC, and be certain to wait only as long as
necessary for the transaction _they_ started to reach disk.
Sloppy callers can START_SYNC and WAIT_SYNC without a transid,
and provided they didn't wait too long between the calls, they
will get the same result. However, if a second commit starts
before they call WAIT_SYNC, they may end up waiting longer for
it to commit as well. Even so, a START_SYNC+WAIT_SYNC still
guarantees that any operation completed before the START_SYNC
reaches disk.
Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
2010-10-30 02:41:32 +07:00
|
|
|
transid = trans->transid;
|
2016-09-10 08:39:03 +07:00
|
|
|
ret = btrfs_commit_transaction_async(trans, 0);
|
2011-04-04 08:52:13 +07:00
|
|
|
if (ret) {
|
2016-09-10 08:39:03 +07:00
|
|
|
btrfs_end_transaction(trans);
|
2011-03-23 15:14:16 +07:00
|
|
|
return ret;
|
2011-04-04 08:52:13 +07:00
|
|
|
}
|
2012-11-26 15:41:29 +07:00
|
|
|
out:
|
Btrfs: add START_SYNC, WAIT_SYNC ioctls
START_SYNC will start a sync/commit, but not wait for it to
complete. Any modification started after the ioctl returns is
guaranteed not to be included in the commit. If a non-NULL
pointer is passed, the transaction id will be returned to
userspace.
WAIT_SYNC will wait for any in-progress commit to complete. If a
transaction id is specified, the ioctl will block and then
return (success) when the specified transaction has committed.
If it has already committed when we call the ioctl, it returns
immediately. If the specified transaction doesn't exist, it
returns EINVAL.
If no transaction id is specified, WAIT_SYNC will wait for the
currently committing transaction to finish it's commit to disk.
If there is no currently committing transaction, it returns
success.
These ioctls are useful for applications which want to impose an
ordering on when fs modifications reach disk, but do not want to
wait for the full (slow) commit process to do so.
Picky callers can take the transid returned by START_SYNC and
feed it to WAIT_SYNC, and be certain to wait only as long as
necessary for the transaction _they_ started to reach disk.
Sloppy callers can START_SYNC and WAIT_SYNC without a transid,
and provided they didn't wait too long between the calls, they
will get the same result. However, if a second commit starts
before they call WAIT_SYNC, they may end up waiting longer for
it to commit as well. Even so, a START_SYNC+WAIT_SYNC still
guarantees that any operation completed before the START_SYNC
reaches disk.
Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
2010-10-30 02:41:32 +07:00
|
|
|
if (argp)
|
|
|
|
|
if (copy_to_user(argp, &transid, sizeof(transid)))
|
|
|
|
|
return -EFAULT;
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
2016-06-23 05:54:24 +07:00
|
|
|
static noinline long btrfs_ioctl_wait_sync(struct btrfs_fs_info *fs_info,
|
2012-11-26 15:40:43 +07:00
|
|
|
void __user *argp)
|
Btrfs: add START_SYNC, WAIT_SYNC ioctls
START_SYNC will start a sync/commit, but not wait for it to
complete. Any modification started after the ioctl returns is
guaranteed not to be included in the commit. If a non-NULL
pointer is passed, the transaction id will be returned to
userspace.
WAIT_SYNC will wait for any in-progress commit to complete. If a
transaction id is specified, the ioctl will block and then
return (success) when the specified transaction has committed.
If it has already committed when we call the ioctl, it returns
immediately. If the specified transaction doesn't exist, it
returns EINVAL.
If no transaction id is specified, WAIT_SYNC will wait for the
currently committing transaction to finish it's commit to disk.
If there is no currently committing transaction, it returns
success.
These ioctls are useful for applications which want to impose an
ordering on when fs modifications reach disk, but do not want to
wait for the full (slow) commit process to do so.
Picky callers can take the transid returned by START_SYNC and
feed it to WAIT_SYNC, and be certain to wait only as long as
necessary for the transaction _they_ started to reach disk.
Sloppy callers can START_SYNC and WAIT_SYNC without a transid,
and provided they didn't wait too long between the calls, they
will get the same result. However, if a second commit starts
before they call WAIT_SYNC, they may end up waiting longer for
it to commit as well. Even so, a START_SYNC+WAIT_SYNC still
guarantees that any operation completed before the START_SYNC
reaches disk.
Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
2010-10-30 02:41:32 +07:00
|
|
|
{
|
|
|
|
|
u64 transid;
|
|
|
|
|
|
|
|
|
|
if (argp) {
|
|
|
|
|
if (copy_from_user(&transid, argp, sizeof(transid)))
|
|
|
|
|
return -EFAULT;
|
|
|
|
|
} else {
|
|
|
|
|
transid = 0; /* current trans */
|
|
|
|
|
}
|
2016-06-23 05:54:24 +07:00
|
|
|
return btrfs_wait_for_commit(fs_info, transid);
|
Btrfs: add START_SYNC, WAIT_SYNC ioctls
START_SYNC will start a sync/commit, but not wait for it to
complete. Any modification started after the ioctl returns is
guaranteed not to be included in the commit. If a non-NULL
pointer is passed, the transaction id will be returned to
userspace.
WAIT_SYNC will wait for any in-progress commit to complete. If a
transaction id is specified, the ioctl will block and then
return (success) when the specified transaction has committed.
If it has already committed when we call the ioctl, it returns
immediately. If the specified transaction doesn't exist, it
returns EINVAL.
If no transaction id is specified, WAIT_SYNC will wait for the
currently committing transaction to finish it's commit to disk.
If there is no currently committing transaction, it returns
success.
These ioctls are useful for applications which want to impose an
ordering on when fs modifications reach disk, but do not want to
wait for the full (slow) commit process to do so.
Picky callers can take the transid returned by START_SYNC and
feed it to WAIT_SYNC, and be certain to wait only as long as
necessary for the transaction _they_ started to reach disk.
Sloppy callers can START_SYNC and WAIT_SYNC without a transid,
and provided they didn't wait too long between the calls, they
will get the same result. However, if a second commit starts
before they call WAIT_SYNC, they may end up waiting longer for
it to commit as well. Even so, a START_SYNC+WAIT_SYNC still
guarantees that any operation completed before the START_SYNC
reaches disk.
Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
2010-10-30 02:41:32 +07:00
|
|
|
}
|
|
|
|
|
|
2012-11-26 15:48:01 +07:00
|
|
|
static long btrfs_ioctl_scrub(struct file *file, void __user *arg)
|
2011-03-11 21:41:01 +07:00
|
|
|
{
|
2016-06-23 05:54:23 +07:00
|
|
|
struct btrfs_fs_info *fs_info = btrfs_sb(file_inode(file)->i_sb);
|
2011-03-11 21:41:01 +07:00
|
|
|
struct btrfs_ioctl_scrub_args *sa;
|
2012-11-26 15:48:01 +07:00
|
|
|
int ret;
|
2011-03-11 21:41:01 +07:00
|
|
|
|
|
|
|
|
if (!capable(CAP_SYS_ADMIN))
|
|
|
|
|
return -EPERM;
|
|
|
|
|
|
|
|
|
|
sa = memdup_user(arg, sizeof(*sa));
|
|
|
|
|
if (IS_ERR(sa))
|
|
|
|
|
return PTR_ERR(sa);
|
|
|
|
|
|
2012-11-26 15:48:01 +07:00
|
|
|
if (!(sa->flags & BTRFS_SCRUB_READONLY)) {
|
|
|
|
|
ret = mnt_want_write_file(file);
|
|
|
|
|
if (ret)
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
|
2016-06-23 05:54:23 +07:00
|
|
|
ret = btrfs_scrub_dev(fs_info, sa->devid, sa->start, sa->end,
|
2012-11-06 00:29:28 +07:00
|
|
|
&sa->progress, sa->flags & BTRFS_SCRUB_READONLY,
|
|
|
|
|
0);
|
2011-03-11 21:41:01 +07:00
|
|
|
|
|
|
|
|
if (copy_to_user(arg, sa, sizeof(*sa)))
|
|
|
|
|
ret = -EFAULT;
|
|
|
|
|
|
2012-11-26 15:48:01 +07:00
|
|
|
if (!(sa->flags & BTRFS_SCRUB_READONLY))
|
|
|
|
|
mnt_drop_write_file(file);
|
|
|
|
|
out:
|
2011-03-11 21:41:01 +07:00
|
|
|
kfree(sa);
|
|
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
2016-06-23 05:54:24 +07:00
|
|
|
static long btrfs_ioctl_scrub_cancel(struct btrfs_fs_info *fs_info)
|
2011-03-11 21:41:01 +07:00
|
|
|
{
|
|
|
|
|
if (!capable(CAP_SYS_ADMIN))
|
|
|
|
|
return -EPERM;
|
|
|
|
|
|
2016-06-23 05:54:24 +07:00
|
|
|
return btrfs_scrub_cancel(fs_info);
|
2011-03-11 21:41:01 +07:00
|
|
|
}
|
|
|
|
|
|
2016-06-23 05:54:24 +07:00
|
|
|
static long btrfs_ioctl_scrub_progress(struct btrfs_fs_info *fs_info,
|
2011-03-11 21:41:01 +07:00
|
|
|
void __user *arg)
|
|
|
|
|
{
|
|
|
|
|
struct btrfs_ioctl_scrub_args *sa;
|
|
|
|
|
int ret;
|
|
|
|
|
|
|
|
|
|
if (!capable(CAP_SYS_ADMIN))
|
|
|
|
|
return -EPERM;
|
|
|
|
|
|
|
|
|
|
sa = memdup_user(arg, sizeof(*sa));
|
|
|
|
|
if (IS_ERR(sa))
|
|
|
|
|
return PTR_ERR(sa);
|
|
|
|
|
|
2016-06-23 05:54:24 +07:00
|
|
|
ret = btrfs_scrub_progress(fs_info, sa->devid, &sa->progress);
|
2011-03-11 21:41:01 +07:00
|
|
|
|
|
|
|
|
if (copy_to_user(arg, sa, sizeof(*sa)))
|
|
|
|
|
ret = -EFAULT;
|
|
|
|
|
|
|
|
|
|
kfree(sa);
|
|
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
2016-06-23 05:54:24 +07:00
|
|
|
static long btrfs_ioctl_get_dev_stats(struct btrfs_fs_info *fs_info,
|
2012-06-22 19:30:39 +07:00
|
|
|
void __user *arg)
|
2012-05-25 21:06:09 +07:00
|
|
|
{
|
|
|
|
|
struct btrfs_ioctl_get_dev_stats *sa;
|
|
|
|
|
int ret;
|
|
|
|
|
|
|
|
|
|
sa = memdup_user(arg, sizeof(*sa));
|
|
|
|
|
if (IS_ERR(sa))
|
|
|
|
|
return PTR_ERR(sa);
|
|
|
|
|
|
2012-06-22 19:30:39 +07:00
|
|
|
if ((sa->flags & BTRFS_DEV_STATS_RESET) && !capable(CAP_SYS_ADMIN)) {
|
|
|
|
|
kfree(sa);
|
|
|
|
|
return -EPERM;
|
|
|
|
|
}
|
|
|
|
|
|
2016-06-23 05:54:24 +07:00
|
|
|
ret = btrfs_get_dev_stats(fs_info, sa);
|
2012-05-25 21:06:09 +07:00
|
|
|
|
|
|
|
|
if (copy_to_user(arg, sa, sizeof(*sa)))
|
|
|
|
|
ret = -EFAULT;
|
|
|
|
|
|
|
|
|
|
kfree(sa);
|
|
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
2016-06-23 05:54:24 +07:00
|
|
|
static long btrfs_ioctl_dev_replace(struct btrfs_fs_info *fs_info,
|
|
|
|
|
void __user *arg)
|
2012-11-06 21:08:53 +07:00
|
|
|
{
|
|
|
|
|
struct btrfs_ioctl_dev_replace_args *p;
|
|
|
|
|
int ret;
|
|
|
|
|
|
|
|
|
|
if (!capable(CAP_SYS_ADMIN))
|
|
|
|
|
return -EPERM;
|
|
|
|
|
|
|
|
|
|
p = memdup_user(arg, sizeof(*p));
|
|
|
|
|
if (IS_ERR(p))
|
|
|
|
|
return PTR_ERR(p);
|
|
|
|
|
|
|
|
|
|
switch (p->cmd) {
|
|
|
|
|
case BTRFS_IOCTL_DEV_REPLACE_CMD_START:
|
2017-07-17 14:45:34 +07:00
|
|
|
if (sb_rdonly(fs_info->sb)) {
|
2013-10-11 00:39:28 +07:00
|
|
|
ret = -EROFS;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
2017-03-28 19:44:21 +07:00
|
|
|
if (test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags)) {
|
2013-08-21 10:44:48 +07:00
|
|
|
ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS;
|
2012-11-06 21:08:53 +07:00
|
|
|
} else {
|
2016-06-23 05:54:24 +07:00
|
|
|
ret = btrfs_dev_replace_by_ioctl(fs_info, p);
|
2017-03-28 19:44:21 +07:00
|
|
|
clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
|
2012-11-06 21:08:53 +07:00
|
|
|
}
|
|
|
|
|
break;
|
|
|
|
|
case BTRFS_IOCTL_DEV_REPLACE_CMD_STATUS:
|
2016-06-23 05:54:23 +07:00
|
|
|
btrfs_dev_replace_status(fs_info, p);
|
2012-11-06 21:08:53 +07:00
|
|
|
ret = 0;
|
|
|
|
|
break;
|
|
|
|
|
case BTRFS_IOCTL_DEV_REPLACE_CMD_CANCEL:
|
2018-02-12 22:33:30 +07:00
|
|
|
p->result = btrfs_dev_replace_cancel(fs_info);
|
2018-02-12 22:33:29 +07:00
|
|
|
ret = 0;
|
2012-11-06 21:08:53 +07:00
|
|
|
break;
|
|
|
|
|
default:
|
|
|
|
|
ret = -EINVAL;
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (copy_to_user(arg, p, sizeof(*p)))
|
|
|
|
|
ret = -EFAULT;
|
2013-10-11 00:39:28 +07:00
|
|
|
out:
|
2012-11-06 21:08:53 +07:00
|
|
|
kfree(p);
|
|
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
2011-07-07 21:48:38 +07:00
|
|
|
static long btrfs_ioctl_ino_to_path(struct btrfs_root *root, void __user *arg)
|
|
|
|
|
{
|
|
|
|
|
int ret = 0;
|
|
|
|
|
int i;
|
2011-11-03 02:48:34 +07:00
|
|
|
u64 rel_ptr;
|
2011-07-07 21:48:38 +07:00
|
|
|
int size;
|
2011-11-06 15:07:10 +07:00
|
|
|
struct btrfs_ioctl_ino_path_args *ipa = NULL;
|
2011-07-07 21:48:38 +07:00
|
|
|
struct inode_fs_paths *ipath = NULL;
|
|
|
|
|
struct btrfs_path *path;
|
|
|
|
|
|
2013-01-28 18:33:31 +07:00
|
|
|
if (!capable(CAP_DAC_READ_SEARCH))
|
2011-07-07 21:48:38 +07:00
|
|
|
return -EPERM;
|
|
|
|
|
|
|
|
|
|
path = btrfs_alloc_path();
|
|
|
|
|
if (!path) {
|
|
|
|
|
ret = -ENOMEM;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
ipa = memdup_user(arg, sizeof(*ipa));
|
|
|
|
|
if (IS_ERR(ipa)) {
|
|
|
|
|
ret = PTR_ERR(ipa);
|
|
|
|
|
ipa = NULL;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
size = min_t(u32, ipa->size, 4096);
|
|
|
|
|
ipath = init_ipath(size, root, path);
|
|
|
|
|
if (IS_ERR(ipath)) {
|
|
|
|
|
ret = PTR_ERR(ipath);
|
|
|
|
|
ipath = NULL;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
ret = paths_from_inode(ipa->inum, ipath);
|
|
|
|
|
if (ret < 0)
|
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
|
|
for (i = 0; i < ipath->fspath->elem_cnt; ++i) {
|
2011-11-20 19:31:57 +07:00
|
|
|
rel_ptr = ipath->fspath->val[i] -
|
|
|
|
|
(u64)(unsigned long)ipath->fspath->val;
|
2011-11-03 02:48:34 +07:00
|
|
|
ipath->fspath->val[i] = rel_ptr;
|
2011-07-07 21:48:38 +07:00
|
|
|
}
|
|
|
|
|
|
2017-08-23 13:46:05 +07:00
|
|
|
ret = copy_to_user((void __user *)(unsigned long)ipa->fspath,
|
|
|
|
|
ipath->fspath, size);
|
2011-07-07 21:48:38 +07:00
|
|
|
if (ret) {
|
|
|
|
|
ret = -EFAULT;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
out:
|
|
|
|
|
btrfs_free_path(path);
|
|
|
|
|
free_ipath(ipath);
|
|
|
|
|
kfree(ipa);
|
|
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static int build_ino_list(u64 inum, u64 offset, u64 root, void *ctx)
|
|
|
|
|
{
|
|
|
|
|
struct btrfs_data_container *inodes = ctx;
|
|
|
|
|
const size_t c = 3 * sizeof(u64);
|
|
|
|
|
|
|
|
|
|
if (inodes->bytes_left >= c) {
|
|
|
|
|
inodes->bytes_left -= c;
|
|
|
|
|
inodes->val[inodes->elem_cnt] = inum;
|
|
|
|
|
inodes->val[inodes->elem_cnt + 1] = offset;
|
|
|
|
|
inodes->val[inodes->elem_cnt + 2] = root;
|
|
|
|
|
inodes->elem_cnt += 3;
|
|
|
|
|
} else {
|
|
|
|
|
inodes->bytes_missing += c - inodes->bytes_left;
|
|
|
|
|
inodes->bytes_left = 0;
|
|
|
|
|
inodes->elem_missed += 3;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
2016-06-23 05:54:24 +07:00
|
|
|
static long btrfs_ioctl_logical_to_ino(struct btrfs_fs_info *fs_info,
|
btrfs: add a flags argument to LOGICAL_INO and call it LOGICAL_INO_V2
Now that check_extent_in_eb()'s extent offset filter can be turned off,
we need a way to do it from userspace.
Add a 'flags' field to the btrfs_logical_ino_args structure to disable
extent offset filtering, taking the place of one of the existing
reserved[] fields.
Previous versions of LOGICAL_INO neglected to check whether any of the
reserved fields have non-zero values. Assigning meaning to those fields
now may change the behavior of existing programs that left these fields
uninitialized. The lack of a zero check also means that new programs
have no way to know whether the kernel is honoring the flags field.
To avoid these problems, define a new ioctl LOGICAL_INO_V2. We can
use the same argument layout as LOGICAL_INO, but shorten the reserved[]
array by one element and turn it into the 'flags' field. The V2 ioctl
explicitly checks that reserved fields and unsupported flag bits are zero
so that userspace can negotiate future feature bits as they are defined.
Since the memory layouts of the two ioctls' arguments are compatible,
there is no need for a separate function for logical_to_ino_v2 (contrast
with tree_search_v2 vs tree_search where the layout and code are quite
different). A version parameter and an 'if' statement will suffice.
Now that we have a flags field in logical_ino_args, add a flag
BTRFS_LOGICAL_INO_ARGS_IGNORE_OFFSET to get the behavior we want,
and pass it down the stack to iterate_inodes_from_logical.
Motivation and background, copied from the patchset cover letter:
Suppose we have a file with one extent:
root@tester:~# zcat /usr/share/doc/cpio/changelog.gz > /test/a
root@tester:~# sync
Split the extent by overwriting it in the middle:
root@tester:~# cat /dev/urandom | dd bs=4k seek=2 skip=2 count=1 conv=notrunc of=/test/a
We should now have 3 extent refs to 2 extents, with one block unreachable.
The extent tree looks like:
root@tester:~# btrfs-debug-tree /dev/vdc -t 2
[...]
item 9 key (1103101952 EXTENT_ITEM 73728) itemoff 15942 itemsize 53
extent refs 2 gen 29 flags DATA
extent data backref root 5 objectid 261 offset 0 count 2
[...]
item 11 key (1103175680 EXTENT_ITEM 4096) itemoff 15865 itemsize 53
extent refs 1 gen 30 flags DATA
extent data backref root 5 objectid 261 offset 8192 count 1
[...]
and the ref tree looks like:
root@tester:~# btrfs-debug-tree /dev/vdc -t 5
[...]
item 6 key (261 EXTENT_DATA 0) itemoff 15825 itemsize 53
extent data disk byte 1103101952 nr 73728
extent data offset 0 nr 8192 ram 73728
extent compression(none)
item 7 key (261 EXTENT_DATA 8192) itemoff 15772 itemsize 53
extent data disk byte 1103175680 nr 4096
extent data offset 0 nr 4096 ram 4096
extent compression(none)
item 8 key (261 EXTENT_DATA 12288) itemoff 15719 itemsize 53
extent data disk byte 1103101952 nr 73728
extent data offset 12288 nr 61440 ram 73728
extent compression(none)
[...]
There are two references to the same extent with different, non-overlapping
byte offsets:
[------------------72K extent at 1103101952----------------------]
[--8K----------------|--4K unreachable----|--60K-----------------]
^ ^
| |
[--8K ref offset 0--][--4K ref offset 0--][--60K ref offset 12K--]
|
v
[-----4K extent-----] at 1103175680
We want to find all of the references to extent bytenr 1103101952.
Without the patch (and without running btrfs-debug-tree), we have to
do it with 18 LOGICAL_INO calls:
root@tester:~# btrfs ins log 1103101952 -P /test/
Using LOGICAL_INO
inode 261 offset 0 root 5
root@tester:~# for x in $(seq 0 17); do btrfs ins log $((1103101952 + x * 4096)) -P /test/; done 2>&1 | grep inode
inode 261 offset 0 root 5
inode 261 offset 4096 root 5 <- same extent ref as offset 0
(offset 8192 returns empty set, not reachable)
inode 261 offset 12288 root 5
inode 261 offset 16384 root 5 \
inode 261 offset 20480 root 5 |
inode 261 offset 24576 root 5 |
inode 261 offset 28672 root 5 |
inode 261 offset 32768 root 5 |
inode 261 offset 36864 root 5 \
inode 261 offset 40960 root 5 > all the same extent ref as offset 12288.
inode 261 offset 45056 root 5 / More processing required in userspace
inode 261 offset 49152 root 5 | to figure out these are all duplicates.
inode 261 offset 53248 root 5 |
inode 261 offset 57344 root 5 |
inode 261 offset 61440 root 5 |
inode 261 offset 65536 root 5 |
inode 261 offset 69632 root 5 /
In the worst case the extents are 128MB long, and we have to do 32768
iterations of the loop to find one 4K extent ref.
With the patch, we just use one call to map all refs to the extent at once:
root@tester:~# btrfs ins log 1103101952 -P /test/
Using LOGICAL_INO_V2
inode 261 offset 0 root 5
inode 261 offset 12288 root 5
The TREE_SEARCH ioctl allows userspace to retrieve the offset and
extent bytenr fields easily once the root, inode and offset are known.
This is sufficient information to build a complete map of the extent
and all of its references. Userspace can use this information to make
better choices to dedup or defrag.
Signed-off-by: Zygo Blaxell <ce3g8jdj@umail.furryterror.org>
Reviewed-by: Hans van Kranenburg <hans.van.kranenburg@mendix.com>
Tested-by: Hans van Kranenburg <hans.van.kranenburg@mendix.com>
[ copy background and motivation from cover letter ]
Signed-off-by: David Sterba <dsterba@suse.com>
2017-09-23 00:58:46 +07:00
|
|
|
void __user *arg, int version)
|
2011-07-07 21:48:38 +07:00
|
|
|
{
|
|
|
|
|
int ret = 0;
|
|
|
|
|
int size;
|
|
|
|
|
struct btrfs_ioctl_logical_ino_args *loi;
|
|
|
|
|
struct btrfs_data_container *inodes = NULL;
|
|
|
|
|
struct btrfs_path *path = NULL;
|
btrfs: add a flags argument to LOGICAL_INO and call it LOGICAL_INO_V2
Now that check_extent_in_eb()'s extent offset filter can be turned off,
we need a way to do it from userspace.
Add a 'flags' field to the btrfs_logical_ino_args structure to disable
extent offset filtering, taking the place of one of the existing
reserved[] fields.
Previous versions of LOGICAL_INO neglected to check whether any of the
reserved fields have non-zero values. Assigning meaning to those fields
now may change the behavior of existing programs that left these fields
uninitialized. The lack of a zero check also means that new programs
have no way to know whether the kernel is honoring the flags field.
To avoid these problems, define a new ioctl LOGICAL_INO_V2. We can
use the same argument layout as LOGICAL_INO, but shorten the reserved[]
array by one element and turn it into the 'flags' field. The V2 ioctl
explicitly checks that reserved fields and unsupported flag bits are zero
so that userspace can negotiate future feature bits as they are defined.
Since the memory layouts of the two ioctls' arguments are compatible,
there is no need for a separate function for logical_to_ino_v2 (contrast
with tree_search_v2 vs tree_search where the layout and code are quite
different). A version parameter and an 'if' statement will suffice.
Now that we have a flags field in logical_ino_args, add a flag
BTRFS_LOGICAL_INO_ARGS_IGNORE_OFFSET to get the behavior we want,
and pass it down the stack to iterate_inodes_from_logical.
Motivation and background, copied from the patchset cover letter:
Suppose we have a file with one extent:
root@tester:~# zcat /usr/share/doc/cpio/changelog.gz > /test/a
root@tester:~# sync
Split the extent by overwriting it in the middle:
root@tester:~# cat /dev/urandom | dd bs=4k seek=2 skip=2 count=1 conv=notrunc of=/test/a
We should now have 3 extent refs to 2 extents, with one block unreachable.
The extent tree looks like:
root@tester:~# btrfs-debug-tree /dev/vdc -t 2
[...]
item 9 key (1103101952 EXTENT_ITEM 73728) itemoff 15942 itemsize 53
extent refs 2 gen 29 flags DATA
extent data backref root 5 objectid 261 offset 0 count 2
[...]
item 11 key (1103175680 EXTENT_ITEM 4096) itemoff 15865 itemsize 53
extent refs 1 gen 30 flags DATA
extent data backref root 5 objectid 261 offset 8192 count 1
[...]
and the ref tree looks like:
root@tester:~# btrfs-debug-tree /dev/vdc -t 5
[...]
item 6 key (261 EXTENT_DATA 0) itemoff 15825 itemsize 53
extent data disk byte 1103101952 nr 73728
extent data offset 0 nr 8192 ram 73728
extent compression(none)
item 7 key (261 EXTENT_DATA 8192) itemoff 15772 itemsize 53
extent data disk byte 1103175680 nr 4096
extent data offset 0 nr 4096 ram 4096
extent compression(none)
item 8 key (261 EXTENT_DATA 12288) itemoff 15719 itemsize 53
extent data disk byte 1103101952 nr 73728
extent data offset 12288 nr 61440 ram 73728
extent compression(none)
[...]
There are two references to the same extent with different, non-overlapping
byte offsets:
[------------------72K extent at 1103101952----------------------]
[--8K----------------|--4K unreachable----|--60K-----------------]
^ ^
| |
[--8K ref offset 0--][--4K ref offset 0--][--60K ref offset 12K--]
|
v
[-----4K extent-----] at 1103175680
We want to find all of the references to extent bytenr 1103101952.
Without the patch (and without running btrfs-debug-tree), we have to
do it with 18 LOGICAL_INO calls:
root@tester:~# btrfs ins log 1103101952 -P /test/
Using LOGICAL_INO
inode 261 offset 0 root 5
root@tester:~# for x in $(seq 0 17); do btrfs ins log $((1103101952 + x * 4096)) -P /test/; done 2>&1 | grep inode
inode 261 offset 0 root 5
inode 261 offset 4096 root 5 <- same extent ref as offset 0
(offset 8192 returns empty set, not reachable)
inode 261 offset 12288 root 5
inode 261 offset 16384 root 5 \
inode 261 offset 20480 root 5 |
inode 261 offset 24576 root 5 |
inode 261 offset 28672 root 5 |
inode 261 offset 32768 root 5 |
inode 261 offset 36864 root 5 \
inode 261 offset 40960 root 5 > all the same extent ref as offset 12288.
inode 261 offset 45056 root 5 / More processing required in userspace
inode 261 offset 49152 root 5 | to figure out these are all duplicates.
inode 261 offset 53248 root 5 |
inode 261 offset 57344 root 5 |
inode 261 offset 61440 root 5 |
inode 261 offset 65536 root 5 |
inode 261 offset 69632 root 5 /
In the worst case the extents are 128MB long, and we have to do 32768
iterations of the loop to find one 4K extent ref.
With the patch, we just use one call to map all refs to the extent at once:
root@tester:~# btrfs ins log 1103101952 -P /test/
Using LOGICAL_INO_V2
inode 261 offset 0 root 5
inode 261 offset 12288 root 5
The TREE_SEARCH ioctl allows userspace to retrieve the offset and
extent bytenr fields easily once the root, inode and offset are known.
This is sufficient information to build a complete map of the extent
and all of its references. Userspace can use this information to make
better choices to dedup or defrag.
Signed-off-by: Zygo Blaxell <ce3g8jdj@umail.furryterror.org>
Reviewed-by: Hans van Kranenburg <hans.van.kranenburg@mendix.com>
Tested-by: Hans van Kranenburg <hans.van.kranenburg@mendix.com>
[ copy background and motivation from cover letter ]
Signed-off-by: David Sterba <dsterba@suse.com>
2017-09-23 00:58:46 +07:00
|
|
|
bool ignore_offset;
|
2011-07-07 21:48:38 +07:00
|
|
|
|
|
|
|
|
if (!capable(CAP_SYS_ADMIN))
|
|
|
|
|
return -EPERM;
|
|
|
|
|
|
|
|
|
|
loi = memdup_user(arg, sizeof(*loi));
|
2016-11-10 16:47:41 +07:00
|
|
|
if (IS_ERR(loi))
|
|
|
|
|
return PTR_ERR(loi);
|
2011-07-07 21:48:38 +07:00
|
|
|
|
btrfs: add a flags argument to LOGICAL_INO and call it LOGICAL_INO_V2
Now that check_extent_in_eb()'s extent offset filter can be turned off,
we need a way to do it from userspace.
Add a 'flags' field to the btrfs_logical_ino_args structure to disable
extent offset filtering, taking the place of one of the existing
reserved[] fields.
Previous versions of LOGICAL_INO neglected to check whether any of the
reserved fields have non-zero values. Assigning meaning to those fields
now may change the behavior of existing programs that left these fields
uninitialized. The lack of a zero check also means that new programs
have no way to know whether the kernel is honoring the flags field.
To avoid these problems, define a new ioctl LOGICAL_INO_V2. We can
use the same argument layout as LOGICAL_INO, but shorten the reserved[]
array by one element and turn it into the 'flags' field. The V2 ioctl
explicitly checks that reserved fields and unsupported flag bits are zero
so that userspace can negotiate future feature bits as they are defined.
Since the memory layouts of the two ioctls' arguments are compatible,
there is no need for a separate function for logical_to_ino_v2 (contrast
with tree_search_v2 vs tree_search where the layout and code are quite
different). A version parameter and an 'if' statement will suffice.
Now that we have a flags field in logical_ino_args, add a flag
BTRFS_LOGICAL_INO_ARGS_IGNORE_OFFSET to get the behavior we want,
and pass it down the stack to iterate_inodes_from_logical.
Motivation and background, copied from the patchset cover letter:
Suppose we have a file with one extent:
root@tester:~# zcat /usr/share/doc/cpio/changelog.gz > /test/a
root@tester:~# sync
Split the extent by overwriting it in the middle:
root@tester:~# cat /dev/urandom | dd bs=4k seek=2 skip=2 count=1 conv=notrunc of=/test/a
We should now have 3 extent refs to 2 extents, with one block unreachable.
The extent tree looks like:
root@tester:~# btrfs-debug-tree /dev/vdc -t 2
[...]
item 9 key (1103101952 EXTENT_ITEM 73728) itemoff 15942 itemsize 53
extent refs 2 gen 29 flags DATA
extent data backref root 5 objectid 261 offset 0 count 2
[...]
item 11 key (1103175680 EXTENT_ITEM 4096) itemoff 15865 itemsize 53
extent refs 1 gen 30 flags DATA
extent data backref root 5 objectid 261 offset 8192 count 1
[...]
and the ref tree looks like:
root@tester:~# btrfs-debug-tree /dev/vdc -t 5
[...]
item 6 key (261 EXTENT_DATA 0) itemoff 15825 itemsize 53
extent data disk byte 1103101952 nr 73728
extent data offset 0 nr 8192 ram 73728
extent compression(none)
item 7 key (261 EXTENT_DATA 8192) itemoff 15772 itemsize 53
extent data disk byte 1103175680 nr 4096
extent data offset 0 nr 4096 ram 4096
extent compression(none)
item 8 key (261 EXTENT_DATA 12288) itemoff 15719 itemsize 53
extent data disk byte 1103101952 nr 73728
extent data offset 12288 nr 61440 ram 73728
extent compression(none)
[...]
There are two references to the same extent with different, non-overlapping
byte offsets:
[------------------72K extent at 1103101952----------------------]
[--8K----------------|--4K unreachable----|--60K-----------------]
^ ^
| |
[--8K ref offset 0--][--4K ref offset 0--][--60K ref offset 12K--]
|
v
[-----4K extent-----] at 1103175680
We want to find all of the references to extent bytenr 1103101952.
Without the patch (and without running btrfs-debug-tree), we have to
do it with 18 LOGICAL_INO calls:
root@tester:~# btrfs ins log 1103101952 -P /test/
Using LOGICAL_INO
inode 261 offset 0 root 5
root@tester:~# for x in $(seq 0 17); do btrfs ins log $((1103101952 + x * 4096)) -P /test/; done 2>&1 | grep inode
inode 261 offset 0 root 5
inode 261 offset 4096 root 5 <- same extent ref as offset 0
(offset 8192 returns empty set, not reachable)
inode 261 offset 12288 root 5
inode 261 offset 16384 root 5 \
inode 261 offset 20480 root 5 |
inode 261 offset 24576 root 5 |
inode 261 offset 28672 root 5 |
inode 261 offset 32768 root 5 |
inode 261 offset 36864 root 5 \
inode 261 offset 40960 root 5 > all the same extent ref as offset 12288.
inode 261 offset 45056 root 5 / More processing required in userspace
inode 261 offset 49152 root 5 | to figure out these are all duplicates.
inode 261 offset 53248 root 5 |
inode 261 offset 57344 root 5 |
inode 261 offset 61440 root 5 |
inode 261 offset 65536 root 5 |
inode 261 offset 69632 root 5 /
In the worst case the extents are 128MB long, and we have to do 32768
iterations of the loop to find one 4K extent ref.
With the patch, we just use one call to map all refs to the extent at once:
root@tester:~# btrfs ins log 1103101952 -P /test/
Using LOGICAL_INO_V2
inode 261 offset 0 root 5
inode 261 offset 12288 root 5
The TREE_SEARCH ioctl allows userspace to retrieve the offset and
extent bytenr fields easily once the root, inode and offset are known.
This is sufficient information to build a complete map of the extent
and all of its references. Userspace can use this information to make
better choices to dedup or defrag.
Signed-off-by: Zygo Blaxell <ce3g8jdj@umail.furryterror.org>
Reviewed-by: Hans van Kranenburg <hans.van.kranenburg@mendix.com>
Tested-by: Hans van Kranenburg <hans.van.kranenburg@mendix.com>
[ copy background and motivation from cover letter ]
Signed-off-by: David Sterba <dsterba@suse.com>
2017-09-23 00:58:46 +07:00
|
|
|
if (version == 1) {
|
|
|
|
|
ignore_offset = false;
|
2017-09-23 00:58:47 +07:00
|
|
|
size = min_t(u32, loi->size, SZ_64K);
|
btrfs: add a flags argument to LOGICAL_INO and call it LOGICAL_INO_V2
Now that check_extent_in_eb()'s extent offset filter can be turned off,
we need a way to do it from userspace.
Add a 'flags' field to the btrfs_logical_ino_args structure to disable
extent offset filtering, taking the place of one of the existing
reserved[] fields.
Previous versions of LOGICAL_INO neglected to check whether any of the
reserved fields have non-zero values. Assigning meaning to those fields
now may change the behavior of existing programs that left these fields
uninitialized. The lack of a zero check also means that new programs
have no way to know whether the kernel is honoring the flags field.
To avoid these problems, define a new ioctl LOGICAL_INO_V2. We can
use the same argument layout as LOGICAL_INO, but shorten the reserved[]
array by one element and turn it into the 'flags' field. The V2 ioctl
explicitly checks that reserved fields and unsupported flag bits are zero
so that userspace can negotiate future feature bits as they are defined.
Since the memory layouts of the two ioctls' arguments are compatible,
there is no need for a separate function for logical_to_ino_v2 (contrast
with tree_search_v2 vs tree_search where the layout and code are quite
different). A version parameter and an 'if' statement will suffice.
Now that we have a flags field in logical_ino_args, add a flag
BTRFS_LOGICAL_INO_ARGS_IGNORE_OFFSET to get the behavior we want,
and pass it down the stack to iterate_inodes_from_logical.
Motivation and background, copied from the patchset cover letter:
Suppose we have a file with one extent:
root@tester:~# zcat /usr/share/doc/cpio/changelog.gz > /test/a
root@tester:~# sync
Split the extent by overwriting it in the middle:
root@tester:~# cat /dev/urandom | dd bs=4k seek=2 skip=2 count=1 conv=notrunc of=/test/a
We should now have 3 extent refs to 2 extents, with one block unreachable.
The extent tree looks like:
root@tester:~# btrfs-debug-tree /dev/vdc -t 2
[...]
item 9 key (1103101952 EXTENT_ITEM 73728) itemoff 15942 itemsize 53
extent refs 2 gen 29 flags DATA
extent data backref root 5 objectid 261 offset 0 count 2
[...]
item 11 key (1103175680 EXTENT_ITEM 4096) itemoff 15865 itemsize 53
extent refs 1 gen 30 flags DATA
extent data backref root 5 objectid 261 offset 8192 count 1
[...]
and the ref tree looks like:
root@tester:~# btrfs-debug-tree /dev/vdc -t 5
[...]
item 6 key (261 EXTENT_DATA 0) itemoff 15825 itemsize 53
extent data disk byte 1103101952 nr 73728
extent data offset 0 nr 8192 ram 73728
extent compression(none)
item 7 key (261 EXTENT_DATA 8192) itemoff 15772 itemsize 53
extent data disk byte 1103175680 nr 4096
extent data offset 0 nr 4096 ram 4096
extent compression(none)
item 8 key (261 EXTENT_DATA 12288) itemoff 15719 itemsize 53
extent data disk byte 1103101952 nr 73728
extent data offset 12288 nr 61440 ram 73728
extent compression(none)
[...]
There are two references to the same extent with different, non-overlapping
byte offsets:
[------------------72K extent at 1103101952----------------------]
[--8K----------------|--4K unreachable----|--60K-----------------]
^ ^
| |
[--8K ref offset 0--][--4K ref offset 0--][--60K ref offset 12K--]
|
v
[-----4K extent-----] at 1103175680
We want to find all of the references to extent bytenr 1103101952.
Without the patch (and without running btrfs-debug-tree), we have to
do it with 18 LOGICAL_INO calls:
root@tester:~# btrfs ins log 1103101952 -P /test/
Using LOGICAL_INO
inode 261 offset 0 root 5
root@tester:~# for x in $(seq 0 17); do btrfs ins log $((1103101952 + x * 4096)) -P /test/; done 2>&1 | grep inode
inode 261 offset 0 root 5
inode 261 offset 4096 root 5 <- same extent ref as offset 0
(offset 8192 returns empty set, not reachable)
inode 261 offset 12288 root 5
inode 261 offset 16384 root 5 \
inode 261 offset 20480 root 5 |
inode 261 offset 24576 root 5 |
inode 261 offset 28672 root 5 |
inode 261 offset 32768 root 5 |
inode 261 offset 36864 root 5 \
inode 261 offset 40960 root 5 > all the same extent ref as offset 12288.
inode 261 offset 45056 root 5 / More processing required in userspace
inode 261 offset 49152 root 5 | to figure out these are all duplicates.
inode 261 offset 53248 root 5 |
inode 261 offset 57344 root 5 |
inode 261 offset 61440 root 5 |
inode 261 offset 65536 root 5 |
inode 261 offset 69632 root 5 /
In the worst case the extents are 128MB long, and we have to do 32768
iterations of the loop to find one 4K extent ref.
With the patch, we just use one call to map all refs to the extent at once:
root@tester:~# btrfs ins log 1103101952 -P /test/
Using LOGICAL_INO_V2
inode 261 offset 0 root 5
inode 261 offset 12288 root 5
The TREE_SEARCH ioctl allows userspace to retrieve the offset and
extent bytenr fields easily once the root, inode and offset are known.
This is sufficient information to build a complete map of the extent
and all of its references. Userspace can use this information to make
better choices to dedup or defrag.
Signed-off-by: Zygo Blaxell <ce3g8jdj@umail.furryterror.org>
Reviewed-by: Hans van Kranenburg <hans.van.kranenburg@mendix.com>
Tested-by: Hans van Kranenburg <hans.van.kranenburg@mendix.com>
[ copy background and motivation from cover letter ]
Signed-off-by: David Sterba <dsterba@suse.com>
2017-09-23 00:58:46 +07:00
|
|
|
} else {
|
|
|
|
|
/* All reserved bits must be 0 for now */
|
|
|
|
|
if (memchr_inv(loi->reserved, 0, sizeof(loi->reserved))) {
|
|
|
|
|
ret = -EINVAL;
|
|
|
|
|
goto out_loi;
|
|
|
|
|
}
|
|
|
|
|
/* Only accept flags we have defined so far */
|
|
|
|
|
if (loi->flags & ~(BTRFS_LOGICAL_INO_ARGS_IGNORE_OFFSET)) {
|
|
|
|
|
ret = -EINVAL;
|
|
|
|
|
goto out_loi;
|
|
|
|
|
}
|
|
|
|
|
ignore_offset = loi->flags & BTRFS_LOGICAL_INO_ARGS_IGNORE_OFFSET;
|
2017-09-23 00:58:47 +07:00
|
|
|
size = min_t(u32, loi->size, SZ_16M);
|
btrfs: add a flags argument to LOGICAL_INO and call it LOGICAL_INO_V2
Now that check_extent_in_eb()'s extent offset filter can be turned off,
we need a way to do it from userspace.
Add a 'flags' field to the btrfs_logical_ino_args structure to disable
extent offset filtering, taking the place of one of the existing
reserved[] fields.
Previous versions of LOGICAL_INO neglected to check whether any of the
reserved fields have non-zero values. Assigning meaning to those fields
now may change the behavior of existing programs that left these fields
uninitialized. The lack of a zero check also means that new programs
have no way to know whether the kernel is honoring the flags field.
To avoid these problems, define a new ioctl LOGICAL_INO_V2. We can
use the same argument layout as LOGICAL_INO, but shorten the reserved[]
array by one element and turn it into the 'flags' field. The V2 ioctl
explicitly checks that reserved fields and unsupported flag bits are zero
so that userspace can negotiate future feature bits as they are defined.
Since the memory layouts of the two ioctls' arguments are compatible,
there is no need for a separate function for logical_to_ino_v2 (contrast
with tree_search_v2 vs tree_search where the layout and code are quite
different). A version parameter and an 'if' statement will suffice.
Now that we have a flags field in logical_ino_args, add a flag
BTRFS_LOGICAL_INO_ARGS_IGNORE_OFFSET to get the behavior we want,
and pass it down the stack to iterate_inodes_from_logical.
Motivation and background, copied from the patchset cover letter:
Suppose we have a file with one extent:
root@tester:~# zcat /usr/share/doc/cpio/changelog.gz > /test/a
root@tester:~# sync
Split the extent by overwriting it in the middle:
root@tester:~# cat /dev/urandom | dd bs=4k seek=2 skip=2 count=1 conv=notrunc of=/test/a
We should now have 3 extent refs to 2 extents, with one block unreachable.
The extent tree looks like:
root@tester:~# btrfs-debug-tree /dev/vdc -t 2
[...]
item 9 key (1103101952 EXTENT_ITEM 73728) itemoff 15942 itemsize 53
extent refs 2 gen 29 flags DATA
extent data backref root 5 objectid 261 offset 0 count 2
[...]
item 11 key (1103175680 EXTENT_ITEM 4096) itemoff 15865 itemsize 53
extent refs 1 gen 30 flags DATA
extent data backref root 5 objectid 261 offset 8192 count 1
[...]
and the ref tree looks like:
root@tester:~# btrfs-debug-tree /dev/vdc -t 5
[...]
item 6 key (261 EXTENT_DATA 0) itemoff 15825 itemsize 53
extent data disk byte 1103101952 nr 73728
extent data offset 0 nr 8192 ram 73728
extent compression(none)
item 7 key (261 EXTENT_DATA 8192) itemoff 15772 itemsize 53
extent data disk byte 1103175680 nr 4096
extent data offset 0 nr 4096 ram 4096
extent compression(none)
item 8 key (261 EXTENT_DATA 12288) itemoff 15719 itemsize 53
extent data disk byte 1103101952 nr 73728
extent data offset 12288 nr 61440 ram 73728
extent compression(none)
[...]
There are two references to the same extent with different, non-overlapping
byte offsets:
[------------------72K extent at 1103101952----------------------]
[--8K----------------|--4K unreachable----|--60K-----------------]
^ ^
| |
[--8K ref offset 0--][--4K ref offset 0--][--60K ref offset 12K--]
|
v
[-----4K extent-----] at 1103175680
We want to find all of the references to extent bytenr 1103101952.
Without the patch (and without running btrfs-debug-tree), we have to
do it with 18 LOGICAL_INO calls:
root@tester:~# btrfs ins log 1103101952 -P /test/
Using LOGICAL_INO
inode 261 offset 0 root 5
root@tester:~# for x in $(seq 0 17); do btrfs ins log $((1103101952 + x * 4096)) -P /test/; done 2>&1 | grep inode
inode 261 offset 0 root 5
inode 261 offset 4096 root 5 <- same extent ref as offset 0
(offset 8192 returns empty set, not reachable)
inode 261 offset 12288 root 5
inode 261 offset 16384 root 5 \
inode 261 offset 20480 root 5 |
inode 261 offset 24576 root 5 |
inode 261 offset 28672 root 5 |
inode 261 offset 32768 root 5 |
inode 261 offset 36864 root 5 \
inode 261 offset 40960 root 5 > all the same extent ref as offset 12288.
inode 261 offset 45056 root 5 / More processing required in userspace
inode 261 offset 49152 root 5 | to figure out these are all duplicates.
inode 261 offset 53248 root 5 |
inode 261 offset 57344 root 5 |
inode 261 offset 61440 root 5 |
inode 261 offset 65536 root 5 |
inode 261 offset 69632 root 5 /
In the worst case the extents are 128MB long, and we have to do 32768
iterations of the loop to find one 4K extent ref.
With the patch, we just use one call to map all refs to the extent at once:
root@tester:~# btrfs ins log 1103101952 -P /test/
Using LOGICAL_INO_V2
inode 261 offset 0 root 5
inode 261 offset 12288 root 5
The TREE_SEARCH ioctl allows userspace to retrieve the offset and
extent bytenr fields easily once the root, inode and offset are known.
This is sufficient information to build a complete map of the extent
and all of its references. Userspace can use this information to make
better choices to dedup or defrag.
Signed-off-by: Zygo Blaxell <ce3g8jdj@umail.furryterror.org>
Reviewed-by: Hans van Kranenburg <hans.van.kranenburg@mendix.com>
Tested-by: Hans van Kranenburg <hans.van.kranenburg@mendix.com>
[ copy background and motivation from cover letter ]
Signed-off-by: David Sterba <dsterba@suse.com>
2017-09-23 00:58:46 +07:00
|
|
|
}
|
|
|
|
|
|
2011-07-07 21:48:38 +07:00
|
|
|
path = btrfs_alloc_path();
|
|
|
|
|
if (!path) {
|
|
|
|
|
ret = -ENOMEM;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
inodes = init_data_container(size);
|
|
|
|
|
if (IS_ERR(inodes)) {
|
|
|
|
|
ret = PTR_ERR(inodes);
|
|
|
|
|
inodes = NULL;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
|
2016-06-23 05:54:24 +07:00
|
|
|
ret = iterate_inodes_from_logical(loi->logical, fs_info, path,
|
btrfs: add a flags argument to LOGICAL_INO and call it LOGICAL_INO_V2
Now that check_extent_in_eb()'s extent offset filter can be turned off,
we need a way to do it from userspace.
Add a 'flags' field to the btrfs_logical_ino_args structure to disable
extent offset filtering, taking the place of one of the existing
reserved[] fields.
Previous versions of LOGICAL_INO neglected to check whether any of the
reserved fields have non-zero values. Assigning meaning to those fields
now may change the behavior of existing programs that left these fields
uninitialized. The lack of a zero check also means that new programs
have no way to know whether the kernel is honoring the flags field.
To avoid these problems, define a new ioctl LOGICAL_INO_V2. We can
use the same argument layout as LOGICAL_INO, but shorten the reserved[]
array by one element and turn it into the 'flags' field. The V2 ioctl
explicitly checks that reserved fields and unsupported flag bits are zero
so that userspace can negotiate future feature bits as they are defined.
Since the memory layouts of the two ioctls' arguments are compatible,
there is no need for a separate function for logical_to_ino_v2 (contrast
with tree_search_v2 vs tree_search where the layout and code are quite
different). A version parameter and an 'if' statement will suffice.
Now that we have a flags field in logical_ino_args, add a flag
BTRFS_LOGICAL_INO_ARGS_IGNORE_OFFSET to get the behavior we want,
and pass it down the stack to iterate_inodes_from_logical.
Motivation and background, copied from the patchset cover letter:
Suppose we have a file with one extent:
root@tester:~# zcat /usr/share/doc/cpio/changelog.gz > /test/a
root@tester:~# sync
Split the extent by overwriting it in the middle:
root@tester:~# cat /dev/urandom | dd bs=4k seek=2 skip=2 count=1 conv=notrunc of=/test/a
We should now have 3 extent refs to 2 extents, with one block unreachable.
The extent tree looks like:
root@tester:~# btrfs-debug-tree /dev/vdc -t 2
[...]
item 9 key (1103101952 EXTENT_ITEM 73728) itemoff 15942 itemsize 53
extent refs 2 gen 29 flags DATA
extent data backref root 5 objectid 261 offset 0 count 2
[...]
item 11 key (1103175680 EXTENT_ITEM 4096) itemoff 15865 itemsize 53
extent refs 1 gen 30 flags DATA
extent data backref root 5 objectid 261 offset 8192 count 1
[...]
and the ref tree looks like:
root@tester:~# btrfs-debug-tree /dev/vdc -t 5
[...]
item 6 key (261 EXTENT_DATA 0) itemoff 15825 itemsize 53
extent data disk byte 1103101952 nr 73728
extent data offset 0 nr 8192 ram 73728
extent compression(none)
item 7 key (261 EXTENT_DATA 8192) itemoff 15772 itemsize 53
extent data disk byte 1103175680 nr 4096
extent data offset 0 nr 4096 ram 4096
extent compression(none)
item 8 key (261 EXTENT_DATA 12288) itemoff 15719 itemsize 53
extent data disk byte 1103101952 nr 73728
extent data offset 12288 nr 61440 ram 73728
extent compression(none)
[...]
There are two references to the same extent with different, non-overlapping
byte offsets:
[------------------72K extent at 1103101952----------------------]
[--8K----------------|--4K unreachable----|--60K-----------------]
^ ^
| |
[--8K ref offset 0--][--4K ref offset 0--][--60K ref offset 12K--]
|
v
[-----4K extent-----] at 1103175680
We want to find all of the references to extent bytenr 1103101952.
Without the patch (and without running btrfs-debug-tree), we have to
do it with 18 LOGICAL_INO calls:
root@tester:~# btrfs ins log 1103101952 -P /test/
Using LOGICAL_INO
inode 261 offset 0 root 5
root@tester:~# for x in $(seq 0 17); do btrfs ins log $((1103101952 + x * 4096)) -P /test/; done 2>&1 | grep inode
inode 261 offset 0 root 5
inode 261 offset 4096 root 5 <- same extent ref as offset 0
(offset 8192 returns empty set, not reachable)
inode 261 offset 12288 root 5
inode 261 offset 16384 root 5 \
inode 261 offset 20480 root 5 |
inode 261 offset 24576 root 5 |
inode 261 offset 28672 root 5 |
inode 261 offset 32768 root 5 |
inode 261 offset 36864 root 5 \
inode 261 offset 40960 root 5 > all the same extent ref as offset 12288.
inode 261 offset 45056 root 5 / More processing required in userspace
inode 261 offset 49152 root 5 | to figure out these are all duplicates.
inode 261 offset 53248 root 5 |
inode 261 offset 57344 root 5 |
inode 261 offset 61440 root 5 |
inode 261 offset 65536 root 5 |
inode 261 offset 69632 root 5 /
In the worst case the extents are 128MB long, and we have to do 32768
iterations of the loop to find one 4K extent ref.
With the patch, we just use one call to map all refs to the extent at once:
root@tester:~# btrfs ins log 1103101952 -P /test/
Using LOGICAL_INO_V2
inode 261 offset 0 root 5
inode 261 offset 12288 root 5
The TREE_SEARCH ioctl allows userspace to retrieve the offset and
extent bytenr fields easily once the root, inode and offset are known.
This is sufficient information to build a complete map of the extent
and all of its references. Userspace can use this information to make
better choices to dedup or defrag.
Signed-off-by: Zygo Blaxell <ce3g8jdj@umail.furryterror.org>
Reviewed-by: Hans van Kranenburg <hans.van.kranenburg@mendix.com>
Tested-by: Hans van Kranenburg <hans.van.kranenburg@mendix.com>
[ copy background and motivation from cover letter ]
Signed-off-by: David Sterba <dsterba@suse.com>
2017-09-23 00:58:46 +07:00
|
|
|
build_ino_list, inodes, ignore_offset);
|
2012-09-08 09:01:29 +07:00
|
|
|
if (ret == -EINVAL)
|
2011-07-07 21:48:38 +07:00
|
|
|
ret = -ENOENT;
|
|
|
|
|
if (ret < 0)
|
|
|
|
|
goto out;
|
|
|
|
|
|
2017-08-23 13:46:05 +07:00
|
|
|
ret = copy_to_user((void __user *)(unsigned long)loi->inodes, inodes,
|
|
|
|
|
size);
|
2011-07-07 21:48:38 +07:00
|
|
|
if (ret)
|
|
|
|
|
ret = -EFAULT;
|
|
|
|
|
|
|
|
|
|
out:
|
|
|
|
|
btrfs_free_path(path);
|
2017-06-01 00:32:09 +07:00
|
|
|
kvfree(inodes);
|
btrfs: add a flags argument to LOGICAL_INO and call it LOGICAL_INO_V2
Now that check_extent_in_eb()'s extent offset filter can be turned off,
we need a way to do it from userspace.
Add a 'flags' field to the btrfs_logical_ino_args structure to disable
extent offset filtering, taking the place of one of the existing
reserved[] fields.
Previous versions of LOGICAL_INO neglected to check whether any of the
reserved fields have non-zero values. Assigning meaning to those fields
now may change the behavior of existing programs that left these fields
uninitialized. The lack of a zero check also means that new programs
have no way to know whether the kernel is honoring the flags field.
To avoid these problems, define a new ioctl LOGICAL_INO_V2. We can
use the same argument layout as LOGICAL_INO, but shorten the reserved[]
array by one element and turn it into the 'flags' field. The V2 ioctl
explicitly checks that reserved fields and unsupported flag bits are zero
so that userspace can negotiate future feature bits as they are defined.
Since the memory layouts of the two ioctls' arguments are compatible,
there is no need for a separate function for logical_to_ino_v2 (contrast
with tree_search_v2 vs tree_search where the layout and code are quite
different). A version parameter and an 'if' statement will suffice.
Now that we have a flags field in logical_ino_args, add a flag
BTRFS_LOGICAL_INO_ARGS_IGNORE_OFFSET to get the behavior we want,
and pass it down the stack to iterate_inodes_from_logical.
Motivation and background, copied from the patchset cover letter:
Suppose we have a file with one extent:
root@tester:~# zcat /usr/share/doc/cpio/changelog.gz > /test/a
root@tester:~# sync
Split the extent by overwriting it in the middle:
root@tester:~# cat /dev/urandom | dd bs=4k seek=2 skip=2 count=1 conv=notrunc of=/test/a
We should now have 3 extent refs to 2 extents, with one block unreachable.
The extent tree looks like:
root@tester:~# btrfs-debug-tree /dev/vdc -t 2
[...]
item 9 key (1103101952 EXTENT_ITEM 73728) itemoff 15942 itemsize 53
extent refs 2 gen 29 flags DATA
extent data backref root 5 objectid 261 offset 0 count 2
[...]
item 11 key (1103175680 EXTENT_ITEM 4096) itemoff 15865 itemsize 53
extent refs 1 gen 30 flags DATA
extent data backref root 5 objectid 261 offset 8192 count 1
[...]
and the ref tree looks like:
root@tester:~# btrfs-debug-tree /dev/vdc -t 5
[...]
item 6 key (261 EXTENT_DATA 0) itemoff 15825 itemsize 53
extent data disk byte 1103101952 nr 73728
extent data offset 0 nr 8192 ram 73728
extent compression(none)
item 7 key (261 EXTENT_DATA 8192) itemoff 15772 itemsize 53
extent data disk byte 1103175680 nr 4096
extent data offset 0 nr 4096 ram 4096
extent compression(none)
item 8 key (261 EXTENT_DATA 12288) itemoff 15719 itemsize 53
extent data disk byte 1103101952 nr 73728
extent data offset 12288 nr 61440 ram 73728
extent compression(none)
[...]
There are two references to the same extent with different, non-overlapping
byte offsets:
[------------------72K extent at 1103101952----------------------]
[--8K----------------|--4K unreachable----|--60K-----------------]
^ ^
| |
[--8K ref offset 0--][--4K ref offset 0--][--60K ref offset 12K--]
|
v
[-----4K extent-----] at 1103175680
We want to find all of the references to extent bytenr 1103101952.
Without the patch (and without running btrfs-debug-tree), we have to
do it with 18 LOGICAL_INO calls:
root@tester:~# btrfs ins log 1103101952 -P /test/
Using LOGICAL_INO
inode 261 offset 0 root 5
root@tester:~# for x in $(seq 0 17); do btrfs ins log $((1103101952 + x * 4096)) -P /test/; done 2>&1 | grep inode
inode 261 offset 0 root 5
inode 261 offset 4096 root 5 <- same extent ref as offset 0
(offset 8192 returns empty set, not reachable)
inode 261 offset 12288 root 5
inode 261 offset 16384 root 5 \
inode 261 offset 20480 root 5 |
inode 261 offset 24576 root 5 |
inode 261 offset 28672 root 5 |
inode 261 offset 32768 root 5 |
inode 261 offset 36864 root 5 \
inode 261 offset 40960 root 5 > all the same extent ref as offset 12288.
inode 261 offset 45056 root 5 / More processing required in userspace
inode 261 offset 49152 root 5 | to figure out these are all duplicates.
inode 261 offset 53248 root 5 |
inode 261 offset 57344 root 5 |
inode 261 offset 61440 root 5 |
inode 261 offset 65536 root 5 |
inode 261 offset 69632 root 5 /
In the worst case the extents are 128MB long, and we have to do 32768
iterations of the loop to find one 4K extent ref.
With the patch, we just use one call to map all refs to the extent at once:
root@tester:~# btrfs ins log 1103101952 -P /test/
Using LOGICAL_INO_V2
inode 261 offset 0 root 5
inode 261 offset 12288 root 5
The TREE_SEARCH ioctl allows userspace to retrieve the offset and
extent bytenr fields easily once the root, inode and offset are known.
This is sufficient information to build a complete map of the extent
and all of its references. Userspace can use this information to make
better choices to dedup or defrag.
Signed-off-by: Zygo Blaxell <ce3g8jdj@umail.furryterror.org>
Reviewed-by: Hans van Kranenburg <hans.van.kranenburg@mendix.com>
Tested-by: Hans van Kranenburg <hans.van.kranenburg@mendix.com>
[ copy background and motivation from cover letter ]
Signed-off-by: David Sterba <dsterba@suse.com>
2017-09-23 00:58:46 +07:00
|
|
|
out_loi:
|
2011-07-07 21:48:38 +07:00
|
|
|
kfree(loi);
|
|
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
2012-01-17 03:04:49 +07:00
|
|
|
void update_ioctl_balance_args(struct btrfs_fs_info *fs_info, int lock,
|
2012-01-17 03:04:47 +07:00
|
|
|
struct btrfs_ioctl_balance_args *bargs)
|
|
|
|
|
{
|
|
|
|
|
struct btrfs_balance_control *bctl = fs_info->balance_ctl;
|
|
|
|
|
|
|
|
|
|
bargs->flags = bctl->flags;
|
|
|
|
|
|
2012-01-17 03:04:49 +07:00
|
|
|
if (atomic_read(&fs_info->balance_running))
|
|
|
|
|
bargs->state |= BTRFS_BALANCE_STATE_RUNNING;
|
|
|
|
|
if (atomic_read(&fs_info->balance_pause_req))
|
|
|
|
|
bargs->state |= BTRFS_BALANCE_STATE_PAUSE_REQ;
|
2012-01-17 03:04:49 +07:00
|
|
|
if (atomic_read(&fs_info->balance_cancel_req))
|
|
|
|
|
bargs->state |= BTRFS_BALANCE_STATE_CANCEL_REQ;
|
2012-01-17 03:04:49 +07:00
|
|
|
|
2012-01-17 03:04:47 +07:00
|
|
|
memcpy(&bargs->data, &bctl->data, sizeof(bargs->data));
|
|
|
|
|
memcpy(&bargs->meta, &bctl->meta, sizeof(bargs->meta));
|
|
|
|
|
memcpy(&bargs->sys, &bctl->sys, sizeof(bargs->sys));
|
2012-01-17 03:04:49 +07:00
|
|
|
|
|
|
|
|
if (lock) {
|
|
|
|
|
spin_lock(&fs_info->balance_lock);
|
|
|
|
|
memcpy(&bargs->stat, &bctl->stat, sizeof(bargs->stat));
|
|
|
|
|
spin_unlock(&fs_info->balance_lock);
|
|
|
|
|
} else {
|
|
|
|
|
memcpy(&bargs->stat, &bctl->stat, sizeof(bargs->stat));
|
|
|
|
|
}
|
2012-01-17 03:04:47 +07:00
|
|
|
}
|
|
|
|
|
|
2012-05-11 17:11:26 +07:00
|
|
|
static long btrfs_ioctl_balance(struct file *file, void __user *arg)
|
2012-01-17 03:04:47 +07:00
|
|
|
{
|
2013-01-24 05:07:38 +07:00
|
|
|
struct btrfs_root *root = BTRFS_I(file_inode(file))->root;
|
2012-01-17 03:04:47 +07:00
|
|
|
struct btrfs_fs_info *fs_info = root->fs_info;
|
|
|
|
|
struct btrfs_ioctl_balance_args *bargs;
|
|
|
|
|
struct btrfs_balance_control *bctl;
|
2013-01-20 20:57:57 +07:00
|
|
|
bool need_unlock; /* for mut. excl. ops lock */
|
2012-01-17 03:04:47 +07:00
|
|
|
int ret;
|
|
|
|
|
|
|
|
|
|
if (!capable(CAP_SYS_ADMIN))
|
|
|
|
|
return -EPERM;
|
|
|
|
|
|
2012-06-29 16:58:48 +07:00
|
|
|
ret = mnt_want_write_file(file);
|
2012-05-11 17:11:26 +07:00
|
|
|
if (ret)
|
|
|
|
|
return ret;
|
|
|
|
|
|
2013-01-20 20:57:57 +07:00
|
|
|
again:
|
2017-03-28 19:44:21 +07:00
|
|
|
if (!test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags)) {
|
2013-01-20 20:57:57 +07:00
|
|
|
mutex_lock(&fs_info->volume_mutex);
|
|
|
|
|
mutex_lock(&fs_info->balance_mutex);
|
|
|
|
|
need_unlock = true;
|
|
|
|
|
goto locked;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
2016-05-20 08:18:45 +07:00
|
|
|
* mut. excl. ops lock is locked. Three possibilities:
|
2013-01-20 20:57:57 +07:00
|
|
|
* (1) some other op is running
|
|
|
|
|
* (2) balance is running
|
|
|
|
|
* (3) balance is paused -- special case (think resume)
|
|
|
|
|
*/
|
2012-01-17 03:04:47 +07:00
|
|
|
mutex_lock(&fs_info->balance_mutex);
|
2013-01-20 20:57:57 +07:00
|
|
|
if (fs_info->balance_ctl) {
|
|
|
|
|
/* this is either (2) or (3) */
|
|
|
|
|
if (!atomic_read(&fs_info->balance_running)) {
|
|
|
|
|
mutex_unlock(&fs_info->balance_mutex);
|
|
|
|
|
if (!mutex_trylock(&fs_info->volume_mutex))
|
|
|
|
|
goto again;
|
|
|
|
|
mutex_lock(&fs_info->balance_mutex);
|
|
|
|
|
|
|
|
|
|
if (fs_info->balance_ctl &&
|
|
|
|
|
!atomic_read(&fs_info->balance_running)) {
|
|
|
|
|
/* this is (3) */
|
|
|
|
|
need_unlock = false;
|
|
|
|
|
goto locked;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
mutex_unlock(&fs_info->balance_mutex);
|
|
|
|
|
mutex_unlock(&fs_info->volume_mutex);
|
|
|
|
|
goto again;
|
|
|
|
|
} else {
|
|
|
|
|
/* this is (2) */
|
|
|
|
|
mutex_unlock(&fs_info->balance_mutex);
|
|
|
|
|
ret = -EINPROGRESS;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
} else {
|
|
|
|
|
/* this is (1) */
|
|
|
|
|
mutex_unlock(&fs_info->balance_mutex);
|
2013-08-21 10:44:48 +07:00
|
|
|
ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS;
|
2013-01-20 20:57:57 +07:00
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
locked:
|
2017-03-28 19:44:21 +07:00
|
|
|
BUG_ON(!test_bit(BTRFS_FS_EXCL_OP, &fs_info->flags));
|
2012-01-17 03:04:47 +07:00
|
|
|
|
|
|
|
|
if (arg) {
|
|
|
|
|
bargs = memdup_user(arg, sizeof(*bargs));
|
|
|
|
|
if (IS_ERR(bargs)) {
|
|
|
|
|
ret = PTR_ERR(bargs);
|
2013-01-20 20:57:57 +07:00
|
|
|
goto out_unlock;
|
2012-01-17 03:04:47 +07:00
|
|
|
}
|
2012-01-17 03:04:49 +07:00
|
|
|
|
|
|
|
|
if (bargs->flags & BTRFS_BALANCE_RESUME) {
|
|
|
|
|
if (!fs_info->balance_ctl) {
|
|
|
|
|
ret = -ENOTCONN;
|
|
|
|
|
goto out_bargs;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
bctl = fs_info->balance_ctl;
|
|
|
|
|
spin_lock(&fs_info->balance_lock);
|
|
|
|
|
bctl->flags |= BTRFS_BALANCE_RESUME;
|
|
|
|
|
spin_unlock(&fs_info->balance_lock);
|
|
|
|
|
|
|
|
|
|
goto do_balance;
|
|
|
|
|
}
|
2012-01-17 03:04:47 +07:00
|
|
|
} else {
|
|
|
|
|
bargs = NULL;
|
|
|
|
|
}
|
|
|
|
|
|
2013-01-20 20:57:57 +07:00
|
|
|
if (fs_info->balance_ctl) {
|
2012-01-17 03:04:49 +07:00
|
|
|
ret = -EINPROGRESS;
|
|
|
|
|
goto out_bargs;
|
|
|
|
|
}
|
|
|
|
|
|
2015-11-04 21:38:29 +07:00
|
|
|
bctl = kzalloc(sizeof(*bctl), GFP_KERNEL);
|
2012-01-17 03:04:47 +07:00
|
|
|
if (!bctl) {
|
|
|
|
|
ret = -ENOMEM;
|
|
|
|
|
goto out_bargs;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
bctl->fs_info = fs_info;
|
|
|
|
|
if (arg) {
|
|
|
|
|
memcpy(&bctl->data, &bargs->data, sizeof(bctl->data));
|
|
|
|
|
memcpy(&bctl->meta, &bargs->meta, sizeof(bctl->meta));
|
|
|
|
|
memcpy(&bctl->sys, &bargs->sys, sizeof(bctl->sys));
|
|
|
|
|
|
|
|
|
|
bctl->flags = bargs->flags;
|
2012-01-17 03:04:47 +07:00
|
|
|
} else {
|
|
|
|
|
/* balance everything - no filters */
|
|
|
|
|
bctl->flags |= BTRFS_BALANCE_TYPE_MASK;
|
2012-01-17 03:04:47 +07:00
|
|
|
}
|
|
|
|
|
|
2015-10-12 21:55:54 +07:00
|
|
|
if (bctl->flags & ~(BTRFS_BALANCE_ARGS_MASK | BTRFS_BALANCE_TYPE_MASK)) {
|
|
|
|
|
ret = -EINVAL;
|
2015-10-21 05:50:06 +07:00
|
|
|
goto out_bctl;
|
2015-10-12 21:55:54 +07:00
|
|
|
}
|
|
|
|
|
|
2012-01-17 03:04:49 +07:00
|
|
|
do_balance:
|
2012-01-17 03:04:47 +07:00
|
|
|
/*
|
2017-03-28 19:44:21 +07:00
|
|
|
* Ownership of bctl and filesystem flag BTRFS_FS_EXCL_OP
|
2013-01-20 20:57:57 +07:00
|
|
|
* goes to to btrfs_balance. bctl is freed in __cancel_balance,
|
|
|
|
|
* or, if restriper was paused all the way until unmount, in
|
2018-03-20 23:28:05 +07:00
|
|
|
* free_fs_info. The flag should be cleared after __cancel_balance.
|
2012-01-17 03:04:47 +07:00
|
|
|
*/
|
2013-01-20 20:57:57 +07:00
|
|
|
need_unlock = false;
|
|
|
|
|
|
|
|
|
|
ret = btrfs_balance(bctl, bargs);
|
2015-10-21 05:50:06 +07:00
|
|
|
bctl = NULL;
|
2013-01-20 20:57:57 +07:00
|
|
|
|
2012-01-17 03:04:47 +07:00
|
|
|
if (arg) {
|
|
|
|
|
if (copy_to_user(arg, bargs, sizeof(*bargs)))
|
|
|
|
|
ret = -EFAULT;
|
|
|
|
|
}
|
|
|
|
|
|
2015-10-21 05:50:06 +07:00
|
|
|
out_bctl:
|
|
|
|
|
kfree(bctl);
|
2012-01-17 03:04:47 +07:00
|
|
|
out_bargs:
|
|
|
|
|
kfree(bargs);
|
2013-01-20 20:57:57 +07:00
|
|
|
out_unlock:
|
2012-01-17 03:04:47 +07:00
|
|
|
mutex_unlock(&fs_info->balance_mutex);
|
|
|
|
|
mutex_unlock(&fs_info->volume_mutex);
|
2013-01-20 20:57:57 +07:00
|
|
|
if (need_unlock)
|
2017-03-28 19:44:21 +07:00
|
|
|
clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
|
2013-01-20 20:57:57 +07:00
|
|
|
out:
|
2012-06-29 16:58:48 +07:00
|
|
|
mnt_drop_write_file(file);
|
2012-01-17 03:04:47 +07:00
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
2016-06-23 05:54:24 +07:00
|
|
|
static long btrfs_ioctl_balance_ctl(struct btrfs_fs_info *fs_info, int cmd)
|
2012-01-17 03:04:49 +07:00
|
|
|
{
|
|
|
|
|
if (!capable(CAP_SYS_ADMIN))
|
|
|
|
|
return -EPERM;
|
|
|
|
|
|
|
|
|
|
switch (cmd) {
|
|
|
|
|
case BTRFS_BALANCE_CTL_PAUSE:
|
2016-06-23 05:54:23 +07:00
|
|
|
return btrfs_pause_balance(fs_info);
|
2012-01-17 03:04:49 +07:00
|
|
|
case BTRFS_BALANCE_CTL_CANCEL:
|
2016-06-23 05:54:23 +07:00
|
|
|
return btrfs_cancel_balance(fs_info);
|
2012-01-17 03:04:49 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return -EINVAL;
|
|
|
|
|
}
|
|
|
|
|
|
2016-06-23 05:54:24 +07:00
|
|
|
static long btrfs_ioctl_balance_progress(struct btrfs_fs_info *fs_info,
|
2012-01-17 03:04:49 +07:00
|
|
|
void __user *arg)
|
|
|
|
|
{
|
|
|
|
|
struct btrfs_ioctl_balance_args *bargs;
|
|
|
|
|
int ret = 0;
|
|
|
|
|
|
|
|
|
|
if (!capable(CAP_SYS_ADMIN))
|
|
|
|
|
return -EPERM;
|
|
|
|
|
|
|
|
|
|
mutex_lock(&fs_info->balance_mutex);
|
|
|
|
|
if (!fs_info->balance_ctl) {
|
|
|
|
|
ret = -ENOTCONN;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
|
2015-11-04 21:38:29 +07:00
|
|
|
bargs = kzalloc(sizeof(*bargs), GFP_KERNEL);
|
2012-01-17 03:04:49 +07:00
|
|
|
if (!bargs) {
|
|
|
|
|
ret = -ENOMEM;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
update_ioctl_balance_args(fs_info, 1, bargs);
|
|
|
|
|
|
|
|
|
|
if (copy_to_user(arg, bargs, sizeof(*bargs)))
|
|
|
|
|
ret = -EFAULT;
|
|
|
|
|
|
|
|
|
|
kfree(bargs);
|
|
|
|
|
out:
|
|
|
|
|
mutex_unlock(&fs_info->balance_mutex);
|
|
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
2012-11-26 15:50:11 +07:00
|
|
|
static long btrfs_ioctl_quota_ctl(struct file *file, void __user *arg)
|
2011-09-14 20:53:51 +07:00
|
|
|
{
|
2016-06-23 05:54:23 +07:00
|
|
|
struct inode *inode = file_inode(file);
|
|
|
|
|
struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
|
2011-09-14 20:53:51 +07:00
|
|
|
struct btrfs_ioctl_quota_ctl_args *sa;
|
|
|
|
|
struct btrfs_trans_handle *trans = NULL;
|
|
|
|
|
int ret;
|
|
|
|
|
int err;
|
|
|
|
|
|
|
|
|
|
if (!capable(CAP_SYS_ADMIN))
|
|
|
|
|
return -EPERM;
|
|
|
|
|
|
2012-11-26 15:50:11 +07:00
|
|
|
ret = mnt_want_write_file(file);
|
|
|
|
|
if (ret)
|
|
|
|
|
return ret;
|
2011-09-14 20:53:51 +07:00
|
|
|
|
|
|
|
|
sa = memdup_user(arg, sizeof(*sa));
|
2012-11-26 15:50:11 +07:00
|
|
|
if (IS_ERR(sa)) {
|
|
|
|
|
ret = PTR_ERR(sa);
|
|
|
|
|
goto drop_write;
|
|
|
|
|
}
|
2011-09-14 20:53:51 +07:00
|
|
|
|
2016-06-23 05:54:23 +07:00
|
|
|
down_write(&fs_info->subvol_sem);
|
|
|
|
|
trans = btrfs_start_transaction(fs_info->tree_root, 2);
|
2013-04-25 23:04:51 +07:00
|
|
|
if (IS_ERR(trans)) {
|
|
|
|
|
ret = PTR_ERR(trans);
|
|
|
|
|
goto out;
|
2011-09-14 20:53:51 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
switch (sa->cmd) {
|
|
|
|
|
case BTRFS_QUOTA_CTL_ENABLE:
|
2016-06-23 05:54:23 +07:00
|
|
|
ret = btrfs_quota_enable(trans, fs_info);
|
2011-09-14 20:53:51 +07:00
|
|
|
break;
|
|
|
|
|
case BTRFS_QUOTA_CTL_DISABLE:
|
2016-06-23 05:54:23 +07:00
|
|
|
ret = btrfs_quota_disable(trans, fs_info);
|
2011-09-14 20:53:51 +07:00
|
|
|
break;
|
|
|
|
|
default:
|
|
|
|
|
ret = -EINVAL;
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
|
2016-09-10 08:39:03 +07:00
|
|
|
err = btrfs_commit_transaction(trans);
|
2013-04-25 23:04:51 +07:00
|
|
|
if (err && !ret)
|
|
|
|
|
ret = err;
|
2011-09-14 20:53:51 +07:00
|
|
|
out:
|
|
|
|
|
kfree(sa);
|
2016-06-23 05:54:23 +07:00
|
|
|
up_write(&fs_info->subvol_sem);
|
2012-11-26 15:50:11 +07:00
|
|
|
drop_write:
|
|
|
|
|
mnt_drop_write_file(file);
|
2011-09-14 20:53:51 +07:00
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
2012-11-26 15:50:11 +07:00
|
|
|
static long btrfs_ioctl_qgroup_assign(struct file *file, void __user *arg)
|
2011-09-14 20:53:51 +07:00
|
|
|
{
|
2016-06-23 05:54:23 +07:00
|
|
|
struct inode *inode = file_inode(file);
|
|
|
|
|
struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
|
|
|
|
|
struct btrfs_root *root = BTRFS_I(inode)->root;
|
2011-09-14 20:53:51 +07:00
|
|
|
struct btrfs_ioctl_qgroup_assign_args *sa;
|
|
|
|
|
struct btrfs_trans_handle *trans;
|
|
|
|
|
int ret;
|
|
|
|
|
int err;
|
|
|
|
|
|
|
|
|
|
if (!capable(CAP_SYS_ADMIN))
|
|
|
|
|
return -EPERM;
|
|
|
|
|
|
2012-11-26 15:50:11 +07:00
|
|
|
ret = mnt_want_write_file(file);
|
|
|
|
|
if (ret)
|
|
|
|
|
return ret;
|
2011-09-14 20:53:51 +07:00
|
|
|
|
|
|
|
|
sa = memdup_user(arg, sizeof(*sa));
|
2012-11-26 15:50:11 +07:00
|
|
|
if (IS_ERR(sa)) {
|
|
|
|
|
ret = PTR_ERR(sa);
|
|
|
|
|
goto drop_write;
|
|
|
|
|
}
|
2011-09-14 20:53:51 +07:00
|
|
|
|
|
|
|
|
trans = btrfs_join_transaction(root);
|
|
|
|
|
if (IS_ERR(trans)) {
|
|
|
|
|
ret = PTR_ERR(trans);
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (sa->assign) {
|
2016-06-23 05:54:23 +07:00
|
|
|
ret = btrfs_add_qgroup_relation(trans, fs_info,
|
2011-09-14 20:53:51 +07:00
|
|
|
sa->src, sa->dst);
|
|
|
|
|
} else {
|
2016-06-23 05:54:23 +07:00
|
|
|
ret = btrfs_del_qgroup_relation(trans, fs_info,
|
2011-09-14 20:53:51 +07:00
|
|
|
sa->src, sa->dst);
|
|
|
|
|
}
|
|
|
|
|
|
2015-02-27 15:24:28 +07:00
|
|
|
/* update qgroup status and info */
|
2016-06-23 05:54:23 +07:00
|
|
|
err = btrfs_run_qgroups(trans, fs_info);
|
2015-02-27 15:24:28 +07:00
|
|
|
if (err < 0)
|
2016-06-23 05:54:23 +07:00
|
|
|
btrfs_handle_fs_error(fs_info, err,
|
|
|
|
|
"failed to update qgroup status and info");
|
2016-09-10 08:39:03 +07:00
|
|
|
err = btrfs_end_transaction(trans);
|
2011-09-14 20:53:51 +07:00
|
|
|
if (err && !ret)
|
|
|
|
|
ret = err;
|
|
|
|
|
|
|
|
|
|
out:
|
|
|
|
|
kfree(sa);
|
2012-11-26 15:50:11 +07:00
|
|
|
drop_write:
|
|
|
|
|
mnt_drop_write_file(file);
|
2011-09-14 20:53:51 +07:00
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
2012-11-26 15:50:11 +07:00
|
|
|
static long btrfs_ioctl_qgroup_create(struct file *file, void __user *arg)
|
2011-09-14 20:53:51 +07:00
|
|
|
{
|
2016-06-23 05:54:23 +07:00
|
|
|
struct inode *inode = file_inode(file);
|
|
|
|
|
struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
|
|
|
|
|
struct btrfs_root *root = BTRFS_I(inode)->root;
|
2011-09-14 20:53:51 +07:00
|
|
|
struct btrfs_ioctl_qgroup_create_args *sa;
|
|
|
|
|
struct btrfs_trans_handle *trans;
|
|
|
|
|
int ret;
|
|
|
|
|
int err;
|
|
|
|
|
|
|
|
|
|
if (!capable(CAP_SYS_ADMIN))
|
|
|
|
|
return -EPERM;
|
|
|
|
|
|
2012-11-26 15:50:11 +07:00
|
|
|
ret = mnt_want_write_file(file);
|
|
|
|
|
if (ret)
|
|
|
|
|
return ret;
|
2011-09-14 20:53:51 +07:00
|
|
|
|
|
|
|
|
sa = memdup_user(arg, sizeof(*sa));
|
2012-11-26 15:50:11 +07:00
|
|
|
if (IS_ERR(sa)) {
|
|
|
|
|
ret = PTR_ERR(sa);
|
|
|
|
|
goto drop_write;
|
|
|
|
|
}
|
2011-09-14 20:53:51 +07:00
|
|
|
|
2012-11-15 18:35:41 +07:00
|
|
|
if (!sa->qgroupid) {
|
|
|
|
|
ret = -EINVAL;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
|
2011-09-14 20:53:51 +07:00
|
|
|
trans = btrfs_join_transaction(root);
|
|
|
|
|
if (IS_ERR(trans)) {
|
|
|
|
|
ret = PTR_ERR(trans);
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (sa->create) {
|
2016-06-23 05:54:23 +07:00
|
|
|
ret = btrfs_create_qgroup(trans, fs_info, sa->qgroupid);
|
2011-09-14 20:53:51 +07:00
|
|
|
} else {
|
2016-06-23 05:54:23 +07:00
|
|
|
ret = btrfs_remove_qgroup(trans, fs_info, sa->qgroupid);
|
2011-09-14 20:53:51 +07:00
|
|
|
}
|
|
|
|
|
|
2016-09-10 08:39:03 +07:00
|
|
|
err = btrfs_end_transaction(trans);
|
2011-09-14 20:53:51 +07:00
|
|
|
if (err && !ret)
|
|
|
|
|
ret = err;
|
|
|
|
|
|
|
|
|
|
out:
|
|
|
|
|
kfree(sa);
|
2012-11-26 15:50:11 +07:00
|
|
|
drop_write:
|
|
|
|
|
mnt_drop_write_file(file);
|
2011-09-14 20:53:51 +07:00
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
2012-11-26 15:50:11 +07:00
|
|
|
static long btrfs_ioctl_qgroup_limit(struct file *file, void __user *arg)
|
2011-09-14 20:53:51 +07:00
|
|
|
{
|
2016-06-23 05:54:23 +07:00
|
|
|
struct inode *inode = file_inode(file);
|
|
|
|
|
struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
|
|
|
|
|
struct btrfs_root *root = BTRFS_I(inode)->root;
|
2011-09-14 20:53:51 +07:00
|
|
|
struct btrfs_ioctl_qgroup_limit_args *sa;
|
|
|
|
|
struct btrfs_trans_handle *trans;
|
|
|
|
|
int ret;
|
|
|
|
|
int err;
|
|
|
|
|
u64 qgroupid;
|
|
|
|
|
|
|
|
|
|
if (!capable(CAP_SYS_ADMIN))
|
|
|
|
|
return -EPERM;
|
|
|
|
|
|
2012-11-26 15:50:11 +07:00
|
|
|
ret = mnt_want_write_file(file);
|
|
|
|
|
if (ret)
|
|
|
|
|
return ret;
|
2011-09-14 20:53:51 +07:00
|
|
|
|
|
|
|
|
sa = memdup_user(arg, sizeof(*sa));
|
2012-11-26 15:50:11 +07:00
|
|
|
if (IS_ERR(sa)) {
|
|
|
|
|
ret = PTR_ERR(sa);
|
|
|
|
|
goto drop_write;
|
|
|
|
|
}
|
2011-09-14 20:53:51 +07:00
|
|
|
|
|
|
|
|
trans = btrfs_join_transaction(root);
|
|
|
|
|
if (IS_ERR(trans)) {
|
|
|
|
|
ret = PTR_ERR(trans);
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
qgroupid = sa->qgroupid;
|
|
|
|
|
if (!qgroupid) {
|
|
|
|
|
/* take the current subvol as qgroup */
|
|
|
|
|
qgroupid = root->root_key.objectid;
|
|
|
|
|
}
|
|
|
|
|
|
2016-06-23 05:54:23 +07:00
|
|
|
ret = btrfs_limit_qgroup(trans, fs_info, qgroupid, &sa->lim);
|
2011-09-14 20:53:51 +07:00
|
|
|
|
2016-09-10 08:39:03 +07:00
|
|
|
err = btrfs_end_transaction(trans);
|
2011-09-14 20:53:51 +07:00
|
|
|
if (err && !ret)
|
|
|
|
|
ret = err;
|
|
|
|
|
|
|
|
|
|
out:
|
|
|
|
|
kfree(sa);
|
2012-11-26 15:50:11 +07:00
|
|
|
drop_write:
|
|
|
|
|
mnt_drop_write_file(file);
|
2011-09-14 20:53:51 +07:00
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
2013-04-25 23:04:51 +07:00
|
|
|
static long btrfs_ioctl_quota_rescan(struct file *file, void __user *arg)
|
|
|
|
|
{
|
2016-06-23 05:54:23 +07:00
|
|
|
struct inode *inode = file_inode(file);
|
|
|
|
|
struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
|
2013-04-25 23:04:51 +07:00
|
|
|
struct btrfs_ioctl_quota_rescan_args *qsa;
|
|
|
|
|
int ret;
|
|
|
|
|
|
|
|
|
|
if (!capable(CAP_SYS_ADMIN))
|
|
|
|
|
return -EPERM;
|
|
|
|
|
|
|
|
|
|
ret = mnt_want_write_file(file);
|
|
|
|
|
if (ret)
|
|
|
|
|
return ret;
|
|
|
|
|
|
|
|
|
|
qsa = memdup_user(arg, sizeof(*qsa));
|
|
|
|
|
if (IS_ERR(qsa)) {
|
|
|
|
|
ret = PTR_ERR(qsa);
|
|
|
|
|
goto drop_write;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (qsa->flags) {
|
|
|
|
|
ret = -EINVAL;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
|
2016-06-23 05:54:23 +07:00
|
|
|
ret = btrfs_qgroup_rescan(fs_info);
|
2013-04-25 23:04:51 +07:00
|
|
|
|
|
|
|
|
out:
|
|
|
|
|
kfree(qsa);
|
|
|
|
|
drop_write:
|
|
|
|
|
mnt_drop_write_file(file);
|
|
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static long btrfs_ioctl_quota_rescan_status(struct file *file, void __user *arg)
|
|
|
|
|
{
|
2016-06-23 05:54:23 +07:00
|
|
|
struct inode *inode = file_inode(file);
|
|
|
|
|
struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
|
2013-04-25 23:04:51 +07:00
|
|
|
struct btrfs_ioctl_quota_rescan_args *qsa;
|
|
|
|
|
int ret = 0;
|
|
|
|
|
|
|
|
|
|
if (!capable(CAP_SYS_ADMIN))
|
|
|
|
|
return -EPERM;
|
|
|
|
|
|
2015-11-04 21:38:29 +07:00
|
|
|
qsa = kzalloc(sizeof(*qsa), GFP_KERNEL);
|
2013-04-25 23:04:51 +07:00
|
|
|
if (!qsa)
|
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
2016-06-23 05:54:23 +07:00
|
|
|
if (fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_RESCAN) {
|
2013-04-25 23:04:51 +07:00
|
|
|
qsa->flags = 1;
|
2016-06-23 05:54:23 +07:00
|
|
|
qsa->progress = fs_info->qgroup_rescan_progress.objectid;
|
2013-04-25 23:04:51 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (copy_to_user(arg, qsa, sizeof(*qsa)))
|
|
|
|
|
ret = -EFAULT;
|
|
|
|
|
|
|
|
|
|
kfree(qsa);
|
|
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
2013-05-07 02:14:17 +07:00
|
|
|
static long btrfs_ioctl_quota_rescan_wait(struct file *file, void __user *arg)
|
|
|
|
|
{
|
2016-06-23 05:54:23 +07:00
|
|
|
struct inode *inode = file_inode(file);
|
|
|
|
|
struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
|
2013-05-07 02:14:17 +07:00
|
|
|
|
|
|
|
|
if (!capable(CAP_SYS_ADMIN))
|
|
|
|
|
return -EPERM;
|
|
|
|
|
|
2016-06-23 05:54:23 +07:00
|
|
|
return btrfs_qgroup_wait_for_completion(fs_info, true);
|
2013-05-07 02:14:17 +07:00
|
|
|
}
|
|
|
|
|
|
2014-01-31 03:17:00 +07:00
|
|
|
static long _btrfs_ioctl_set_received_subvol(struct file *file,
|
|
|
|
|
struct btrfs_ioctl_received_subvol_args *sa)
|
2012-07-25 22:35:53 +07:00
|
|
|
{
|
2013-01-24 05:07:38 +07:00
|
|
|
struct inode *inode = file_inode(file);
|
2016-06-23 05:54:23 +07:00
|
|
|
struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
|
2012-07-25 22:35:53 +07:00
|
|
|
struct btrfs_root *root = BTRFS_I(inode)->root;
|
|
|
|
|
struct btrfs_root_item *root_item = &root->root_item;
|
|
|
|
|
struct btrfs_trans_handle *trans;
|
2016-09-14 21:48:06 +07:00
|
|
|
struct timespec ct = current_time(inode);
|
2012-07-25 22:35:53 +07:00
|
|
|
int ret = 0;
|
2013-08-15 22:11:20 +07:00
|
|
|
int received_uuid_changed;
|
2012-07-25 22:35:53 +07:00
|
|
|
|
2014-01-16 21:50:22 +07:00
|
|
|
if (!inode_owner_or_capable(inode))
|
|
|
|
|
return -EPERM;
|
|
|
|
|
|
2012-07-25 22:35:53 +07:00
|
|
|
ret = mnt_want_write_file(file);
|
|
|
|
|
if (ret < 0)
|
|
|
|
|
return ret;
|
|
|
|
|
|
2016-06-23 05:54:23 +07:00
|
|
|
down_write(&fs_info->subvol_sem);
|
2012-07-25 22:35:53 +07:00
|
|
|
|
2017-01-11 01:35:31 +07:00
|
|
|
if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FIRST_FREE_OBJECTID) {
|
2012-07-25 22:35:53 +07:00
|
|
|
ret = -EINVAL;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (btrfs_root_readonly(root)) {
|
|
|
|
|
ret = -EROFS;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
|
2013-08-15 22:11:20 +07:00
|
|
|
/*
|
|
|
|
|
* 1 - root item
|
|
|
|
|
* 2 - uuid items (received uuid + subvol uuid)
|
|
|
|
|
*/
|
|
|
|
|
trans = btrfs_start_transaction(root, 3);
|
2012-07-25 22:35:53 +07:00
|
|
|
if (IS_ERR(trans)) {
|
|
|
|
|
ret = PTR_ERR(trans);
|
|
|
|
|
trans = NULL;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
sa->rtransid = trans->transid;
|
|
|
|
|
sa->rtime.sec = ct.tv_sec;
|
|
|
|
|
sa->rtime.nsec = ct.tv_nsec;
|
|
|
|
|
|
2013-08-15 22:11:20 +07:00
|
|
|
received_uuid_changed = memcmp(root_item->received_uuid, sa->uuid,
|
|
|
|
|
BTRFS_UUID_SIZE);
|
|
|
|
|
if (received_uuid_changed &&
|
2018-03-12 19:48:09 +07:00
|
|
|
!btrfs_is_empty_uuid(root_item->received_uuid)) {
|
|
|
|
|
ret = btrfs_uuid_tree_rem(trans, fs_info,
|
|
|
|
|
root_item->received_uuid,
|
|
|
|
|
BTRFS_UUID_KEY_RECEIVED_SUBVOL,
|
|
|
|
|
root->root_key.objectid);
|
|
|
|
|
if (ret && ret != -ENOENT) {
|
|
|
|
|
btrfs_abort_transaction(trans, ret);
|
|
|
|
|
btrfs_end_transaction(trans);
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
}
|
2012-07-25 22:35:53 +07:00
|
|
|
memcpy(root_item->received_uuid, sa->uuid, BTRFS_UUID_SIZE);
|
|
|
|
|
btrfs_set_root_stransid(root_item, sa->stransid);
|
|
|
|
|
btrfs_set_root_rtransid(root_item, sa->rtransid);
|
2013-07-16 10:19:18 +07:00
|
|
|
btrfs_set_stack_timespec_sec(&root_item->stime, sa->stime.sec);
|
|
|
|
|
btrfs_set_stack_timespec_nsec(&root_item->stime, sa->stime.nsec);
|
|
|
|
|
btrfs_set_stack_timespec_sec(&root_item->rtime, sa->rtime.sec);
|
|
|
|
|
btrfs_set_stack_timespec_nsec(&root_item->rtime, sa->rtime.nsec);
|
2012-07-25 22:35:53 +07:00
|
|
|
|
2016-06-23 05:54:23 +07:00
|
|
|
ret = btrfs_update_root(trans, fs_info->tree_root,
|
2012-07-25 22:35:53 +07:00
|
|
|
&root->root_key, &root->root_item);
|
|
|
|
|
if (ret < 0) {
|
2016-09-10 08:39:03 +07:00
|
|
|
btrfs_end_transaction(trans);
|
2012-07-25 22:35:53 +07:00
|
|
|
goto out;
|
2013-08-15 22:11:20 +07:00
|
|
|
}
|
|
|
|
|
if (received_uuid_changed && !btrfs_is_empty_uuid(sa->uuid)) {
|
2016-06-23 05:54:23 +07:00
|
|
|
ret = btrfs_uuid_tree_add(trans, fs_info, sa->uuid,
|
2013-08-15 22:11:20 +07:00
|
|
|
BTRFS_UUID_KEY_RECEIVED_SUBVOL,
|
|
|
|
|
root->root_key.objectid);
|
|
|
|
|
if (ret < 0 && ret != -EEXIST) {
|
2016-06-11 05:19:25 +07:00
|
|
|
btrfs_abort_transaction(trans, ret);
|
2017-09-28 15:45:26 +07:00
|
|
|
btrfs_end_transaction(trans);
|
2012-07-25 22:35:53 +07:00
|
|
|
goto out;
|
2013-08-15 22:11:20 +07:00
|
|
|
}
|
|
|
|
|
}
|
2016-09-10 08:39:03 +07:00
|
|
|
ret = btrfs_commit_transaction(trans);
|
2014-01-31 03:17:00 +07:00
|
|
|
out:
|
2016-06-23 05:54:23 +07:00
|
|
|
up_write(&fs_info->subvol_sem);
|
2014-01-31 03:17:00 +07:00
|
|
|
mnt_drop_write_file(file);
|
|
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
#ifdef CONFIG_64BIT
|
|
|
|
|
static long btrfs_ioctl_set_received_subvol_32(struct file *file,
|
|
|
|
|
void __user *arg)
|
|
|
|
|
{
|
|
|
|
|
struct btrfs_ioctl_received_subvol_args_32 *args32 = NULL;
|
|
|
|
|
struct btrfs_ioctl_received_subvol_args *args64 = NULL;
|
|
|
|
|
int ret = 0;
|
|
|
|
|
|
|
|
|
|
args32 = memdup_user(arg, sizeof(*args32));
|
2016-11-10 16:47:41 +07:00
|
|
|
if (IS_ERR(args32))
|
|
|
|
|
return PTR_ERR(args32);
|
2014-01-31 03:17:00 +07:00
|
|
|
|
2015-11-04 21:38:29 +07:00
|
|
|
args64 = kmalloc(sizeof(*args64), GFP_KERNEL);
|
2014-03-28 15:06:00 +07:00
|
|
|
if (!args64) {
|
|
|
|
|
ret = -ENOMEM;
|
2014-01-31 03:17:00 +07:00
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
memcpy(args64->uuid, args32->uuid, BTRFS_UUID_SIZE);
|
|
|
|
|
args64->stransid = args32->stransid;
|
|
|
|
|
args64->rtransid = args32->rtransid;
|
|
|
|
|
args64->stime.sec = args32->stime.sec;
|
|
|
|
|
args64->stime.nsec = args32->stime.nsec;
|
|
|
|
|
args64->rtime.sec = args32->rtime.sec;
|
|
|
|
|
args64->rtime.nsec = args32->rtime.nsec;
|
|
|
|
|
args64->flags = args32->flags;
|
|
|
|
|
|
|
|
|
|
ret = _btrfs_ioctl_set_received_subvol(file, args64);
|
|
|
|
|
if (ret)
|
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
|
|
memcpy(args32->uuid, args64->uuid, BTRFS_UUID_SIZE);
|
|
|
|
|
args32->stransid = args64->stransid;
|
|
|
|
|
args32->rtransid = args64->rtransid;
|
|
|
|
|
args32->stime.sec = args64->stime.sec;
|
|
|
|
|
args32->stime.nsec = args64->stime.nsec;
|
|
|
|
|
args32->rtime.sec = args64->rtime.sec;
|
|
|
|
|
args32->rtime.nsec = args64->rtime.nsec;
|
|
|
|
|
args32->flags = args64->flags;
|
|
|
|
|
|
|
|
|
|
ret = copy_to_user(arg, args32, sizeof(*args32));
|
|
|
|
|
if (ret)
|
|
|
|
|
ret = -EFAULT;
|
|
|
|
|
|
|
|
|
|
out:
|
|
|
|
|
kfree(args32);
|
|
|
|
|
kfree(args64);
|
|
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
static long btrfs_ioctl_set_received_subvol(struct file *file,
|
|
|
|
|
void __user *arg)
|
|
|
|
|
{
|
|
|
|
|
struct btrfs_ioctl_received_subvol_args *sa = NULL;
|
|
|
|
|
int ret = 0;
|
|
|
|
|
|
|
|
|
|
sa = memdup_user(arg, sizeof(*sa));
|
2016-11-10 16:47:41 +07:00
|
|
|
if (IS_ERR(sa))
|
|
|
|
|
return PTR_ERR(sa);
|
2014-01-31 03:17:00 +07:00
|
|
|
|
|
|
|
|
ret = _btrfs_ioctl_set_received_subvol(file, sa);
|
|
|
|
|
|
|
|
|
|
if (ret)
|
|
|
|
|
goto out;
|
|
|
|
|
|
2012-07-25 22:35:53 +07:00
|
|
|
ret = copy_to_user(arg, sa, sizeof(*sa));
|
|
|
|
|
if (ret)
|
|
|
|
|
ret = -EFAULT;
|
|
|
|
|
|
|
|
|
|
out:
|
|
|
|
|
kfree(sa);
|
|
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
2013-01-05 09:48:01 +07:00
|
|
|
static int btrfs_ioctl_get_fslabel(struct file *file, void __user *arg)
|
|
|
|
|
{
|
2016-06-23 05:54:23 +07:00
|
|
|
struct inode *inode = file_inode(file);
|
|
|
|
|
struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
|
2013-07-19 16:39:32 +07:00
|
|
|
size_t len;
|
2013-01-05 09:48:01 +07:00
|
|
|
int ret;
|
2013-07-19 16:39:32 +07:00
|
|
|
char label[BTRFS_LABEL_SIZE];
|
|
|
|
|
|
2016-06-23 05:54:23 +07:00
|
|
|
spin_lock(&fs_info->super_lock);
|
|
|
|
|
memcpy(label, fs_info->super_copy->label, BTRFS_LABEL_SIZE);
|
|
|
|
|
spin_unlock(&fs_info->super_lock);
|
2013-07-19 16:39:32 +07:00
|
|
|
|
|
|
|
|
len = strnlen(label, BTRFS_LABEL_SIZE);
|
2013-01-05 09:48:01 +07:00
|
|
|
|
|
|
|
|
if (len == BTRFS_LABEL_SIZE) {
|
2016-06-23 05:54:23 +07:00
|
|
|
btrfs_warn(fs_info,
|
|
|
|
|
"label is too long, return the first %zu bytes",
|
|
|
|
|
--len);
|
2013-01-05 09:48:01 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
ret = copy_to_user(arg, label, len);
|
|
|
|
|
|
|
|
|
|
return ret ? -EFAULT : 0;
|
|
|
|
|
}
|
|
|
|
|
|
2013-01-05 09:48:08 +07:00
|
|
|
static int btrfs_ioctl_set_fslabel(struct file *file, void __user *arg)
|
|
|
|
|
{
|
2016-06-23 05:54:23 +07:00
|
|
|
struct inode *inode = file_inode(file);
|
|
|
|
|
struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
|
|
|
|
|
struct btrfs_root *root = BTRFS_I(inode)->root;
|
|
|
|
|
struct btrfs_super_block *super_block = fs_info->super_copy;
|
2013-01-05 09:48:08 +07:00
|
|
|
struct btrfs_trans_handle *trans;
|
|
|
|
|
char label[BTRFS_LABEL_SIZE];
|
|
|
|
|
int ret;
|
|
|
|
|
|
|
|
|
|
if (!capable(CAP_SYS_ADMIN))
|
|
|
|
|
return -EPERM;
|
|
|
|
|
|
|
|
|
|
if (copy_from_user(label, arg, sizeof(label)))
|
|
|
|
|
return -EFAULT;
|
|
|
|
|
|
|
|
|
|
if (strnlen(label, BTRFS_LABEL_SIZE) == BTRFS_LABEL_SIZE) {
|
2016-06-23 05:54:23 +07:00
|
|
|
btrfs_err(fs_info,
|
2016-09-20 21:05:00 +07:00
|
|
|
"unable to set label with more than %d bytes",
|
|
|
|
|
BTRFS_LABEL_SIZE - 1);
|
2013-01-05 09:48:08 +07:00
|
|
|
return -EINVAL;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
ret = mnt_want_write_file(file);
|
|
|
|
|
if (ret)
|
|
|
|
|
return ret;
|
|
|
|
|
|
|
|
|
|
trans = btrfs_start_transaction(root, 0);
|
|
|
|
|
if (IS_ERR(trans)) {
|
|
|
|
|
ret = PTR_ERR(trans);
|
|
|
|
|
goto out_unlock;
|
|
|
|
|
}
|
|
|
|
|
|
2016-06-23 05:54:23 +07:00
|
|
|
spin_lock(&fs_info->super_lock);
|
2013-01-05 09:48:08 +07:00
|
|
|
strcpy(super_block->label, label);
|
2016-06-23 05:54:23 +07:00
|
|
|
spin_unlock(&fs_info->super_lock);
|
2016-09-10 08:39:03 +07:00
|
|
|
ret = btrfs_commit_transaction(trans);
|
2013-01-05 09:48:08 +07:00
|
|
|
|
|
|
|
|
out_unlock:
|
|
|
|
|
mnt_drop_write_file(file);
|
|
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
2013-11-16 03:33:55 +07:00
|
|
|
#define INIT_FEATURE_FLAGS(suffix) \
|
|
|
|
|
{ .compat_flags = BTRFS_FEATURE_COMPAT_##suffix, \
|
|
|
|
|
.compat_ro_flags = BTRFS_FEATURE_COMPAT_RO_##suffix, \
|
|
|
|
|
.incompat_flags = BTRFS_FEATURE_INCOMPAT_##suffix }
|
|
|
|
|
|
2016-02-17 21:26:27 +07:00
|
|
|
int btrfs_ioctl_get_supported_features(void __user *arg)
|
2013-11-16 03:33:55 +07:00
|
|
|
{
|
2015-11-19 17:42:31 +07:00
|
|
|
static const struct btrfs_ioctl_feature_flags features[3] = {
|
2013-11-16 03:33:55 +07:00
|
|
|
INIT_FEATURE_FLAGS(SUPP),
|
|
|
|
|
INIT_FEATURE_FLAGS(SAFE_SET),
|
|
|
|
|
INIT_FEATURE_FLAGS(SAFE_CLEAR)
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
if (copy_to_user(arg, &features, sizeof(features)))
|
|
|
|
|
return -EFAULT;
|
|
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static int btrfs_ioctl_get_features(struct file *file, void __user *arg)
|
|
|
|
|
{
|
2016-06-23 05:54:23 +07:00
|
|
|
struct inode *inode = file_inode(file);
|
|
|
|
|
struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
|
|
|
|
|
struct btrfs_super_block *super_block = fs_info->super_copy;
|
2013-11-16 03:33:55 +07:00
|
|
|
struct btrfs_ioctl_feature_flags features;
|
|
|
|
|
|
|
|
|
|
features.compat_flags = btrfs_super_compat_flags(super_block);
|
|
|
|
|
features.compat_ro_flags = btrfs_super_compat_ro_flags(super_block);
|
|
|
|
|
features.incompat_flags = btrfs_super_incompat_flags(super_block);
|
|
|
|
|
|
|
|
|
|
if (copy_to_user(arg, &features, sizeof(features)))
|
|
|
|
|
return -EFAULT;
|
|
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
2016-06-23 05:54:24 +07:00
|
|
|
static int check_feature_bits(struct btrfs_fs_info *fs_info,
|
2013-11-02 00:07:02 +07:00
|
|
|
enum btrfs_feature_set set,
|
2013-11-16 03:33:55 +07:00
|
|
|
u64 change_mask, u64 flags, u64 supported_flags,
|
|
|
|
|
u64 safe_set, u64 safe_clear)
|
|
|
|
|
{
|
2013-11-02 00:07:02 +07:00
|
|
|
const char *type = btrfs_feature_set_names[set];
|
|
|
|
|
char *names;
|
2013-11-16 03:33:55 +07:00
|
|
|
u64 disallowed, unsupported;
|
|
|
|
|
u64 set_mask = flags & change_mask;
|
|
|
|
|
u64 clear_mask = ~flags & change_mask;
|
|
|
|
|
|
|
|
|
|
unsupported = set_mask & ~supported_flags;
|
|
|
|
|
if (unsupported) {
|
2013-11-02 00:07:02 +07:00
|
|
|
names = btrfs_printable_features(set, unsupported);
|
|
|
|
|
if (names) {
|
2016-06-23 05:54:23 +07:00
|
|
|
btrfs_warn(fs_info,
|
|
|
|
|
"this kernel does not support the %s feature bit%s",
|
|
|
|
|
names, strchr(names, ',') ? "s" : "");
|
2013-11-02 00:07:02 +07:00
|
|
|
kfree(names);
|
|
|
|
|
} else
|
2016-06-23 05:54:23 +07:00
|
|
|
btrfs_warn(fs_info,
|
|
|
|
|
"this kernel does not support %s bits 0x%llx",
|
|
|
|
|
type, unsupported);
|
2013-11-16 03:33:55 +07:00
|
|
|
return -EOPNOTSUPP;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
disallowed = set_mask & ~safe_set;
|
|
|
|
|
if (disallowed) {
|
2013-11-02 00:07:02 +07:00
|
|
|
names = btrfs_printable_features(set, disallowed);
|
|
|
|
|
if (names) {
|
2016-06-23 05:54:23 +07:00
|
|
|
btrfs_warn(fs_info,
|
|
|
|
|
"can't set the %s feature bit%s while mounted",
|
|
|
|
|
names, strchr(names, ',') ? "s" : "");
|
2013-11-02 00:07:02 +07:00
|
|
|
kfree(names);
|
|
|
|
|
} else
|
2016-06-23 05:54:23 +07:00
|
|
|
btrfs_warn(fs_info,
|
|
|
|
|
"can't set %s bits 0x%llx while mounted",
|
|
|
|
|
type, disallowed);
|
2013-11-16 03:33:55 +07:00
|
|
|
return -EPERM;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
disallowed = clear_mask & ~safe_clear;
|
|
|
|
|
if (disallowed) {
|
2013-11-02 00:07:02 +07:00
|
|
|
names = btrfs_printable_features(set, disallowed);
|
|
|
|
|
if (names) {
|
2016-06-23 05:54:23 +07:00
|
|
|
btrfs_warn(fs_info,
|
|
|
|
|
"can't clear the %s feature bit%s while mounted",
|
|
|
|
|
names, strchr(names, ',') ? "s" : "");
|
2013-11-02 00:07:02 +07:00
|
|
|
kfree(names);
|
|
|
|
|
} else
|
2016-06-23 05:54:23 +07:00
|
|
|
btrfs_warn(fs_info,
|
|
|
|
|
"can't clear %s bits 0x%llx while mounted",
|
|
|
|
|
type, disallowed);
|
2013-11-16 03:33:55 +07:00
|
|
|
return -EPERM;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
2016-06-23 05:54:24 +07:00
|
|
|
#define check_feature(fs_info, change_mask, flags, mask_base) \
|
|
|
|
|
check_feature_bits(fs_info, FEAT_##mask_base, change_mask, flags, \
|
2013-11-16 03:33:55 +07:00
|
|
|
BTRFS_FEATURE_ ## mask_base ## _SUPP, \
|
|
|
|
|
BTRFS_FEATURE_ ## mask_base ## _SAFE_SET, \
|
|
|
|
|
BTRFS_FEATURE_ ## mask_base ## _SAFE_CLEAR)
|
|
|
|
|
|
|
|
|
|
static int btrfs_ioctl_set_features(struct file *file, void __user *arg)
|
|
|
|
|
{
|
2016-06-23 05:54:23 +07:00
|
|
|
struct inode *inode = file_inode(file);
|
|
|
|
|
struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
|
|
|
|
|
struct btrfs_root *root = BTRFS_I(inode)->root;
|
|
|
|
|
struct btrfs_super_block *super_block = fs_info->super_copy;
|
2013-11-16 03:33:55 +07:00
|
|
|
struct btrfs_ioctl_feature_flags flags[2];
|
|
|
|
|
struct btrfs_trans_handle *trans;
|
|
|
|
|
u64 newflags;
|
|
|
|
|
int ret;
|
|
|
|
|
|
|
|
|
|
if (!capable(CAP_SYS_ADMIN))
|
|
|
|
|
return -EPERM;
|
|
|
|
|
|
|
|
|
|
if (copy_from_user(flags, arg, sizeof(flags)))
|
|
|
|
|
return -EFAULT;
|
|
|
|
|
|
|
|
|
|
/* Nothing to do */
|
|
|
|
|
if (!flags[0].compat_flags && !flags[0].compat_ro_flags &&
|
|
|
|
|
!flags[0].incompat_flags)
|
|
|
|
|
return 0;
|
|
|
|
|
|
2016-06-23 05:54:24 +07:00
|
|
|
ret = check_feature(fs_info, flags[0].compat_flags,
|
2013-11-16 03:33:55 +07:00
|
|
|
flags[1].compat_flags, COMPAT);
|
|
|
|
|
if (ret)
|
|
|
|
|
return ret;
|
|
|
|
|
|
2016-06-23 05:54:24 +07:00
|
|
|
ret = check_feature(fs_info, flags[0].compat_ro_flags,
|
2013-11-16 03:33:55 +07:00
|
|
|
flags[1].compat_ro_flags, COMPAT_RO);
|
|
|
|
|
if (ret)
|
|
|
|
|
return ret;
|
|
|
|
|
|
2016-06-23 05:54:24 +07:00
|
|
|
ret = check_feature(fs_info, flags[0].incompat_flags,
|
2013-11-16 03:33:55 +07:00
|
|
|
flags[1].incompat_flags, INCOMPAT);
|
|
|
|
|
if (ret)
|
|
|
|
|
return ret;
|
|
|
|
|
|
2016-05-04 16:32:00 +07:00
|
|
|
ret = mnt_want_write_file(file);
|
|
|
|
|
if (ret)
|
|
|
|
|
return ret;
|
|
|
|
|
|
2014-02-07 20:34:04 +07:00
|
|
|
trans = btrfs_start_transaction(root, 0);
|
2016-05-04 16:32:00 +07:00
|
|
|
if (IS_ERR(trans)) {
|
|
|
|
|
ret = PTR_ERR(trans);
|
|
|
|
|
goto out_drop_write;
|
|
|
|
|
}
|
2013-11-16 03:33:55 +07:00
|
|
|
|
2016-06-23 05:54:23 +07:00
|
|
|
spin_lock(&fs_info->super_lock);
|
2013-11-16 03:33:55 +07:00
|
|
|
newflags = btrfs_super_compat_flags(super_block);
|
|
|
|
|
newflags |= flags[0].compat_flags & flags[1].compat_flags;
|
|
|
|
|
newflags &= ~(flags[0].compat_flags & ~flags[1].compat_flags);
|
|
|
|
|
btrfs_set_super_compat_flags(super_block, newflags);
|
|
|
|
|
|
|
|
|
|
newflags = btrfs_super_compat_ro_flags(super_block);
|
|
|
|
|
newflags |= flags[0].compat_ro_flags & flags[1].compat_ro_flags;
|
|
|
|
|
newflags &= ~(flags[0].compat_ro_flags & ~flags[1].compat_ro_flags);
|
|
|
|
|
btrfs_set_super_compat_ro_flags(super_block, newflags);
|
|
|
|
|
|
|
|
|
|
newflags = btrfs_super_incompat_flags(super_block);
|
|
|
|
|
newflags |= flags[0].incompat_flags & flags[1].incompat_flags;
|
|
|
|
|
newflags &= ~(flags[0].incompat_flags & ~flags[1].incompat_flags);
|
|
|
|
|
btrfs_set_super_incompat_flags(super_block, newflags);
|
2016-06-23 05:54:23 +07:00
|
|
|
spin_unlock(&fs_info->super_lock);
|
2013-11-16 03:33:55 +07:00
|
|
|
|
2016-09-10 08:39:03 +07:00
|
|
|
ret = btrfs_commit_transaction(trans);
|
2016-05-04 16:32:00 +07:00
|
|
|
out_drop_write:
|
|
|
|
|
mnt_drop_write_file(file);
|
|
|
|
|
|
|
|
|
|
return ret;
|
2013-11-16 03:33:55 +07:00
|
|
|
}
|
|
|
|
|
|
2017-09-27 21:43:13 +07:00
|
|
|
static int _btrfs_ioctl_send(struct file *file, void __user *argp, bool compat)
|
|
|
|
|
{
|
|
|
|
|
struct btrfs_ioctl_send_args *arg;
|
|
|
|
|
int ret;
|
|
|
|
|
|
|
|
|
|
if (compat) {
|
|
|
|
|
#if defined(CONFIG_64BIT) && defined(CONFIG_COMPAT)
|
|
|
|
|
struct btrfs_ioctl_send_args_32 args32;
|
|
|
|
|
|
|
|
|
|
ret = copy_from_user(&args32, argp, sizeof(args32));
|
|
|
|
|
if (ret)
|
|
|
|
|
return -EFAULT;
|
|
|
|
|
arg = kzalloc(sizeof(*arg), GFP_KERNEL);
|
|
|
|
|
if (!arg)
|
|
|
|
|
return -ENOMEM;
|
|
|
|
|
arg->send_fd = args32.send_fd;
|
|
|
|
|
arg->clone_sources_count = args32.clone_sources_count;
|
|
|
|
|
arg->clone_sources = compat_ptr(args32.clone_sources);
|
|
|
|
|
arg->parent_root = args32.parent_root;
|
|
|
|
|
arg->flags = args32.flags;
|
|
|
|
|
memcpy(arg->reserved, args32.reserved,
|
|
|
|
|
sizeof(args32.reserved));
|
|
|
|
|
#else
|
|
|
|
|
return -ENOTTY;
|
|
|
|
|
#endif
|
|
|
|
|
} else {
|
|
|
|
|
arg = memdup_user(argp, sizeof(*arg));
|
|
|
|
|
if (IS_ERR(arg))
|
|
|
|
|
return PTR_ERR(arg);
|
|
|
|
|
}
|
|
|
|
|
ret = btrfs_ioctl_send(file, arg);
|
|
|
|
|
kfree(arg);
|
|
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
|
2008-06-12 08:53:53 +07:00
|
|
|
long btrfs_ioctl(struct file *file, unsigned int
|
|
|
|
|
cmd, unsigned long arg)
|
|
|
|
|
{
|
2016-06-23 05:54:23 +07:00
|
|
|
struct inode *inode = file_inode(file);
|
|
|
|
|
struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
|
|
|
|
|
struct btrfs_root *root = BTRFS_I(inode)->root;
|
2008-12-02 18:36:08 +07:00
|
|
|
void __user *argp = (void __user *)arg;
|
2008-06-12 08:53:53 +07:00
|
|
|
|
|
|
|
|
switch (cmd) {
|
2009-04-17 15:37:41 +07:00
|
|
|
case FS_IOC_GETFLAGS:
|
|
|
|
|
return btrfs_ioctl_getflags(file, argp);
|
|
|
|
|
case FS_IOC_SETFLAGS:
|
|
|
|
|
return btrfs_ioctl_setflags(file, argp);
|
|
|
|
|
case FS_IOC_GETVERSION:
|
|
|
|
|
return btrfs_ioctl_getversion(file, argp);
|
2011-03-24 17:24:28 +07:00
|
|
|
case FITRIM:
|
|
|
|
|
return btrfs_ioctl_fitrim(file, argp);
|
2008-06-12 08:53:53 +07:00
|
|
|
case BTRFS_IOC_SNAP_CREATE:
|
2010-12-20 14:53:28 +07:00
|
|
|
return btrfs_ioctl_snap_create(file, argp, 0);
|
2010-12-10 13:41:56 +07:00
|
|
|
case BTRFS_IOC_SNAP_CREATE_V2:
|
2010-12-20 14:53:28 +07:00
|
|
|
return btrfs_ioctl_snap_create_v2(file, argp, 0);
|
2008-11-18 09:02:50 +07:00
|
|
|
case BTRFS_IOC_SUBVOL_CREATE:
|
2010-12-20 14:53:28 +07:00
|
|
|
return btrfs_ioctl_snap_create(file, argp, 1);
|
2011-09-14 20:58:21 +07:00
|
|
|
case BTRFS_IOC_SUBVOL_CREATE_V2:
|
|
|
|
|
return btrfs_ioctl_snap_create_v2(file, argp, 1);
|
2009-09-22 03:00:26 +07:00
|
|
|
case BTRFS_IOC_SNAP_DESTROY:
|
|
|
|
|
return btrfs_ioctl_snap_destroy(file, argp);
|
2010-12-20 15:30:25 +07:00
|
|
|
case BTRFS_IOC_SUBVOL_GETFLAGS:
|
|
|
|
|
return btrfs_ioctl_subvol_getflags(file, argp);
|
|
|
|
|
case BTRFS_IOC_SUBVOL_SETFLAGS:
|
|
|
|
|
return btrfs_ioctl_subvol_setflags(file, argp);
|
2009-12-12 04:11:29 +07:00
|
|
|
case BTRFS_IOC_DEFAULT_SUBVOL:
|
|
|
|
|
return btrfs_ioctl_default_subvol(file, argp);
|
2008-06-12 08:53:53 +07:00
|
|
|
case BTRFS_IOC_DEFRAG:
|
2010-03-11 21:42:04 +07:00
|
|
|
return btrfs_ioctl_defrag(file, NULL);
|
|
|
|
|
case BTRFS_IOC_DEFRAG_RANGE:
|
|
|
|
|
return btrfs_ioctl_defrag(file, argp);
|
2008-06-12 08:53:53 +07:00
|
|
|
case BTRFS_IOC_RESIZE:
|
2012-11-26 15:43:45 +07:00
|
|
|
return btrfs_ioctl_resize(file, argp);
|
2008-06-12 08:53:53 +07:00
|
|
|
case BTRFS_IOC_ADD_DEV:
|
2016-06-23 05:54:24 +07:00
|
|
|
return btrfs_ioctl_add_dev(fs_info, argp);
|
2008-06-12 08:53:53 +07:00
|
|
|
case BTRFS_IOC_RM_DEV:
|
2012-11-26 15:44:50 +07:00
|
|
|
return btrfs_ioctl_rm_dev(file, argp);
|
2016-02-13 09:01:39 +07:00
|
|
|
case BTRFS_IOC_RM_DEV_V2:
|
|
|
|
|
return btrfs_ioctl_rm_dev_v2(file, argp);
|
2011-03-11 21:41:01 +07:00
|
|
|
case BTRFS_IOC_FS_INFO:
|
2016-06-23 05:54:24 +07:00
|
|
|
return btrfs_ioctl_fs_info(fs_info, argp);
|
2011-03-11 21:41:01 +07:00
|
|
|
case BTRFS_IOC_DEV_INFO:
|
2016-06-23 05:54:24 +07:00
|
|
|
return btrfs_ioctl_dev_info(fs_info, argp);
|
2008-06-12 08:53:53 +07:00
|
|
|
case BTRFS_IOC_BALANCE:
|
2012-05-11 17:11:26 +07:00
|
|
|
return btrfs_ioctl_balance(file, NULL);
|
2010-03-01 03:39:26 +07:00
|
|
|
case BTRFS_IOC_TREE_SEARCH:
|
|
|
|
|
return btrfs_ioctl_tree_search(file, argp);
|
2014-01-30 22:24:03 +07:00
|
|
|
case BTRFS_IOC_TREE_SEARCH_V2:
|
|
|
|
|
return btrfs_ioctl_tree_search_v2(file, argp);
|
2010-03-01 03:39:26 +07:00
|
|
|
case BTRFS_IOC_INO_LOOKUP:
|
|
|
|
|
return btrfs_ioctl_ino_lookup(file, argp);
|
2011-07-07 21:48:38 +07:00
|
|
|
case BTRFS_IOC_INO_PATHS:
|
|
|
|
|
return btrfs_ioctl_ino_to_path(root, argp);
|
|
|
|
|
case BTRFS_IOC_LOGICAL_INO:
|
btrfs: add a flags argument to LOGICAL_INO and call it LOGICAL_INO_V2
Now that check_extent_in_eb()'s extent offset filter can be turned off,
we need a way to do it from userspace.
Add a 'flags' field to the btrfs_logical_ino_args structure to disable
extent offset filtering, taking the place of one of the existing
reserved[] fields.
Previous versions of LOGICAL_INO neglected to check whether any of the
reserved fields have non-zero values. Assigning meaning to those fields
now may change the behavior of existing programs that left these fields
uninitialized. The lack of a zero check also means that new programs
have no way to know whether the kernel is honoring the flags field.
To avoid these problems, define a new ioctl LOGICAL_INO_V2. We can
use the same argument layout as LOGICAL_INO, but shorten the reserved[]
array by one element and turn it into the 'flags' field. The V2 ioctl
explicitly checks that reserved fields and unsupported flag bits are zero
so that userspace can negotiate future feature bits as they are defined.
Since the memory layouts of the two ioctls' arguments are compatible,
there is no need for a separate function for logical_to_ino_v2 (contrast
with tree_search_v2 vs tree_search where the layout and code are quite
different). A version parameter and an 'if' statement will suffice.
Now that we have a flags field in logical_ino_args, add a flag
BTRFS_LOGICAL_INO_ARGS_IGNORE_OFFSET to get the behavior we want,
and pass it down the stack to iterate_inodes_from_logical.
Motivation and background, copied from the patchset cover letter:
Suppose we have a file with one extent:
root@tester:~# zcat /usr/share/doc/cpio/changelog.gz > /test/a
root@tester:~# sync
Split the extent by overwriting it in the middle:
root@tester:~# cat /dev/urandom | dd bs=4k seek=2 skip=2 count=1 conv=notrunc of=/test/a
We should now have 3 extent refs to 2 extents, with one block unreachable.
The extent tree looks like:
root@tester:~# btrfs-debug-tree /dev/vdc -t 2
[...]
item 9 key (1103101952 EXTENT_ITEM 73728) itemoff 15942 itemsize 53
extent refs 2 gen 29 flags DATA
extent data backref root 5 objectid 261 offset 0 count 2
[...]
item 11 key (1103175680 EXTENT_ITEM 4096) itemoff 15865 itemsize 53
extent refs 1 gen 30 flags DATA
extent data backref root 5 objectid 261 offset 8192 count 1
[...]
and the ref tree looks like:
root@tester:~# btrfs-debug-tree /dev/vdc -t 5
[...]
item 6 key (261 EXTENT_DATA 0) itemoff 15825 itemsize 53
extent data disk byte 1103101952 nr 73728
extent data offset 0 nr 8192 ram 73728
extent compression(none)
item 7 key (261 EXTENT_DATA 8192) itemoff 15772 itemsize 53
extent data disk byte 1103175680 nr 4096
extent data offset 0 nr 4096 ram 4096
extent compression(none)
item 8 key (261 EXTENT_DATA 12288) itemoff 15719 itemsize 53
extent data disk byte 1103101952 nr 73728
extent data offset 12288 nr 61440 ram 73728
extent compression(none)
[...]
There are two references to the same extent with different, non-overlapping
byte offsets:
[------------------72K extent at 1103101952----------------------]
[--8K----------------|--4K unreachable----|--60K-----------------]
^ ^
| |
[--8K ref offset 0--][--4K ref offset 0--][--60K ref offset 12K--]
|
v
[-----4K extent-----] at 1103175680
We want to find all of the references to extent bytenr 1103101952.
Without the patch (and without running btrfs-debug-tree), we have to
do it with 18 LOGICAL_INO calls:
root@tester:~# btrfs ins log 1103101952 -P /test/
Using LOGICAL_INO
inode 261 offset 0 root 5
root@tester:~# for x in $(seq 0 17); do btrfs ins log $((1103101952 + x * 4096)) -P /test/; done 2>&1 | grep inode
inode 261 offset 0 root 5
inode 261 offset 4096 root 5 <- same extent ref as offset 0
(offset 8192 returns empty set, not reachable)
inode 261 offset 12288 root 5
inode 261 offset 16384 root 5 \
inode 261 offset 20480 root 5 |
inode 261 offset 24576 root 5 |
inode 261 offset 28672 root 5 |
inode 261 offset 32768 root 5 |
inode 261 offset 36864 root 5 \
inode 261 offset 40960 root 5 > all the same extent ref as offset 12288.
inode 261 offset 45056 root 5 / More processing required in userspace
inode 261 offset 49152 root 5 | to figure out these are all duplicates.
inode 261 offset 53248 root 5 |
inode 261 offset 57344 root 5 |
inode 261 offset 61440 root 5 |
inode 261 offset 65536 root 5 |
inode 261 offset 69632 root 5 /
In the worst case the extents are 128MB long, and we have to do 32768
iterations of the loop to find one 4K extent ref.
With the patch, we just use one call to map all refs to the extent at once:
root@tester:~# btrfs ins log 1103101952 -P /test/
Using LOGICAL_INO_V2
inode 261 offset 0 root 5
inode 261 offset 12288 root 5
The TREE_SEARCH ioctl allows userspace to retrieve the offset and
extent bytenr fields easily once the root, inode and offset are known.
This is sufficient information to build a complete map of the extent
and all of its references. Userspace can use this information to make
better choices to dedup or defrag.
Signed-off-by: Zygo Blaxell <ce3g8jdj@umail.furryterror.org>
Reviewed-by: Hans van Kranenburg <hans.van.kranenburg@mendix.com>
Tested-by: Hans van Kranenburg <hans.van.kranenburg@mendix.com>
[ copy background and motivation from cover letter ]
Signed-off-by: David Sterba <dsterba@suse.com>
2017-09-23 00:58:46 +07:00
|
|
|
return btrfs_ioctl_logical_to_ino(fs_info, argp, 1);
|
|
|
|
|
case BTRFS_IOC_LOGICAL_INO_V2:
|
|
|
|
|
return btrfs_ioctl_logical_to_ino(fs_info, argp, 2);
|
2010-01-14 01:19:06 +07:00
|
|
|
case BTRFS_IOC_SPACE_INFO:
|
2016-06-23 05:54:24 +07:00
|
|
|
return btrfs_ioctl_space_info(fs_info, argp);
|
2013-09-23 17:35:11 +07:00
|
|
|
case BTRFS_IOC_SYNC: {
|
|
|
|
|
int ret;
|
|
|
|
|
|
2016-06-23 05:54:23 +07:00
|
|
|
ret = btrfs_start_delalloc_roots(fs_info, 0, -1);
|
2013-09-23 17:35:11 +07:00
|
|
|
if (ret)
|
|
|
|
|
return ret;
|
2016-06-23 05:54:23 +07:00
|
|
|
ret = btrfs_sync_fs(inode->i_sb, 1);
|
2014-07-23 19:39:35 +07:00
|
|
|
/*
|
|
|
|
|
* The transaction thread may want to do more work,
|
2016-05-20 08:18:45 +07:00
|
|
|
* namely it pokes the cleaner kthread that will start
|
2014-07-23 19:39:35 +07:00
|
|
|
* processing uncleaned subvols.
|
|
|
|
|
*/
|
2016-06-23 05:54:23 +07:00
|
|
|
wake_up_process(fs_info->transaction_kthread);
|
2013-09-23 17:35:11 +07:00
|
|
|
return ret;
|
|
|
|
|
}
|
Btrfs: add START_SYNC, WAIT_SYNC ioctls
START_SYNC will start a sync/commit, but not wait for it to
complete. Any modification started after the ioctl returns is
guaranteed not to be included in the commit. If a non-NULL
pointer is passed, the transaction id will be returned to
userspace.
WAIT_SYNC will wait for any in-progress commit to complete. If a
transaction id is specified, the ioctl will block and then
return (success) when the specified transaction has committed.
If it has already committed when we call the ioctl, it returns
immediately. If the specified transaction doesn't exist, it
returns EINVAL.
If no transaction id is specified, WAIT_SYNC will wait for the
currently committing transaction to finish it's commit to disk.
If there is no currently committing transaction, it returns
success.
These ioctls are useful for applications which want to impose an
ordering on when fs modifications reach disk, but do not want to
wait for the full (slow) commit process to do so.
Picky callers can take the transid returned by START_SYNC and
feed it to WAIT_SYNC, and be certain to wait only as long as
necessary for the transaction _they_ started to reach disk.
Sloppy callers can START_SYNC and WAIT_SYNC without a transid,
and provided they didn't wait too long between the calls, they
will get the same result. However, if a second commit starts
before they call WAIT_SYNC, they may end up waiting longer for
it to commit as well. Even so, a START_SYNC+WAIT_SYNC still
guarantees that any operation completed before the START_SYNC
reaches disk.
Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
2010-10-30 02:41:32 +07:00
|
|
|
case BTRFS_IOC_START_SYNC:
|
2012-11-26 15:40:43 +07:00
|
|
|
return btrfs_ioctl_start_sync(root, argp);
|
Btrfs: add START_SYNC, WAIT_SYNC ioctls
START_SYNC will start a sync/commit, but not wait for it to
complete. Any modification started after the ioctl returns is
guaranteed not to be included in the commit. If a non-NULL
pointer is passed, the transaction id will be returned to
userspace.
WAIT_SYNC will wait for any in-progress commit to complete. If a
transaction id is specified, the ioctl will block and then
return (success) when the specified transaction has committed.
If it has already committed when we call the ioctl, it returns
immediately. If the specified transaction doesn't exist, it
returns EINVAL.
If no transaction id is specified, WAIT_SYNC will wait for the
currently committing transaction to finish it's commit to disk.
If there is no currently committing transaction, it returns
success.
These ioctls are useful for applications which want to impose an
ordering on when fs modifications reach disk, but do not want to
wait for the full (slow) commit process to do so.
Picky callers can take the transid returned by START_SYNC and
feed it to WAIT_SYNC, and be certain to wait only as long as
necessary for the transaction _they_ started to reach disk.
Sloppy callers can START_SYNC and WAIT_SYNC without a transid,
and provided they didn't wait too long between the calls, they
will get the same result. However, if a second commit starts
before they call WAIT_SYNC, they may end up waiting longer for
it to commit as well. Even so, a START_SYNC+WAIT_SYNC still
guarantees that any operation completed before the START_SYNC
reaches disk.
Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
2010-10-30 02:41:32 +07:00
|
|
|
case BTRFS_IOC_WAIT_SYNC:
|
2016-06-23 05:54:24 +07:00
|
|
|
return btrfs_ioctl_wait_sync(fs_info, argp);
|
2011-03-11 21:41:01 +07:00
|
|
|
case BTRFS_IOC_SCRUB:
|
2012-11-26 15:48:01 +07:00
|
|
|
return btrfs_ioctl_scrub(file, argp);
|
2011-03-11 21:41:01 +07:00
|
|
|
case BTRFS_IOC_SCRUB_CANCEL:
|
2016-06-23 05:54:24 +07:00
|
|
|
return btrfs_ioctl_scrub_cancel(fs_info);
|
2011-03-11 21:41:01 +07:00
|
|
|
case BTRFS_IOC_SCRUB_PROGRESS:
|
2016-06-23 05:54:24 +07:00
|
|
|
return btrfs_ioctl_scrub_progress(fs_info, argp);
|
2012-01-17 03:04:47 +07:00
|
|
|
case BTRFS_IOC_BALANCE_V2:
|
2012-05-11 17:11:26 +07:00
|
|
|
return btrfs_ioctl_balance(file, argp);
|
2012-01-17 03:04:49 +07:00
|
|
|
case BTRFS_IOC_BALANCE_CTL:
|
2016-06-23 05:54:24 +07:00
|
|
|
return btrfs_ioctl_balance_ctl(fs_info, arg);
|
2012-01-17 03:04:49 +07:00
|
|
|
case BTRFS_IOC_BALANCE_PROGRESS:
|
2016-06-23 05:54:24 +07:00
|
|
|
return btrfs_ioctl_balance_progress(fs_info, argp);
|
2012-07-25 22:35:53 +07:00
|
|
|
case BTRFS_IOC_SET_RECEIVED_SUBVOL:
|
|
|
|
|
return btrfs_ioctl_set_received_subvol(file, argp);
|
2014-01-31 03:17:00 +07:00
|
|
|
#ifdef CONFIG_64BIT
|
|
|
|
|
case BTRFS_IOC_SET_RECEIVED_SUBVOL_32:
|
|
|
|
|
return btrfs_ioctl_set_received_subvol_32(file, argp);
|
|
|
|
|
#endif
|
2012-07-26 04:19:24 +07:00
|
|
|
case BTRFS_IOC_SEND:
|
2017-09-27 21:43:13 +07:00
|
|
|
return _btrfs_ioctl_send(file, argp, false);
|
|
|
|
|
#if defined(CONFIG_64BIT) && defined(CONFIG_COMPAT)
|
|
|
|
|
case BTRFS_IOC_SEND_32:
|
|
|
|
|
return _btrfs_ioctl_send(file, argp, true);
|
|
|
|
|
#endif
|
2012-05-25 21:06:09 +07:00
|
|
|
case BTRFS_IOC_GET_DEV_STATS:
|
2016-06-23 05:54:24 +07:00
|
|
|
return btrfs_ioctl_get_dev_stats(fs_info, argp);
|
2011-09-14 20:53:51 +07:00
|
|
|
case BTRFS_IOC_QUOTA_CTL:
|
2012-11-26 15:50:11 +07:00
|
|
|
return btrfs_ioctl_quota_ctl(file, argp);
|
2011-09-14 20:53:51 +07:00
|
|
|
case BTRFS_IOC_QGROUP_ASSIGN:
|
2012-11-26 15:50:11 +07:00
|
|
|
return btrfs_ioctl_qgroup_assign(file, argp);
|
2011-09-14 20:53:51 +07:00
|
|
|
case BTRFS_IOC_QGROUP_CREATE:
|
2012-11-26 15:50:11 +07:00
|
|
|
return btrfs_ioctl_qgroup_create(file, argp);
|
2011-09-14 20:53:51 +07:00
|
|
|
case BTRFS_IOC_QGROUP_LIMIT:
|
2012-11-26 15:50:11 +07:00
|
|
|
return btrfs_ioctl_qgroup_limit(file, argp);
|
2013-04-25 23:04:51 +07:00
|
|
|
case BTRFS_IOC_QUOTA_RESCAN:
|
|
|
|
|
return btrfs_ioctl_quota_rescan(file, argp);
|
|
|
|
|
case BTRFS_IOC_QUOTA_RESCAN_STATUS:
|
|
|
|
|
return btrfs_ioctl_quota_rescan_status(file, argp);
|
2013-05-07 02:14:17 +07:00
|
|
|
case BTRFS_IOC_QUOTA_RESCAN_WAIT:
|
|
|
|
|
return btrfs_ioctl_quota_rescan_wait(file, argp);
|
2012-11-06 21:08:53 +07:00
|
|
|
case BTRFS_IOC_DEV_REPLACE:
|
2016-06-23 05:54:24 +07:00
|
|
|
return btrfs_ioctl_dev_replace(fs_info, argp);
|
2013-01-05 09:48:01 +07:00
|
|
|
case BTRFS_IOC_GET_FSLABEL:
|
|
|
|
|
return btrfs_ioctl_get_fslabel(file, argp);
|
2013-01-05 09:48:08 +07:00
|
|
|
case BTRFS_IOC_SET_FSLABEL:
|
|
|
|
|
return btrfs_ioctl_set_fslabel(file, argp);
|
2013-11-16 03:33:55 +07:00
|
|
|
case BTRFS_IOC_GET_SUPPORTED_FEATURES:
|
2016-02-17 21:26:27 +07:00
|
|
|
return btrfs_ioctl_get_supported_features(argp);
|
2013-11-16 03:33:55 +07:00
|
|
|
case BTRFS_IOC_GET_FEATURES:
|
|
|
|
|
return btrfs_ioctl_get_features(file, argp);
|
|
|
|
|
case BTRFS_IOC_SET_FEATURES:
|
|
|
|
|
return btrfs_ioctl_set_features(file, argp);
|
2008-06-12 08:53:53 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return -ENOTTY;
|
|
|
|
|
}
|
2015-10-29 15:22:21 +07:00
|
|
|
|
|
|
|
|
#ifdef CONFIG_COMPAT
|
|
|
|
|
long btrfs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
|
|
|
|
|
{
|
2017-02-07 07:39:09 +07:00
|
|
|
/*
|
|
|
|
|
* These all access 32-bit values anyway so no further
|
|
|
|
|
* handling is necessary.
|
|
|
|
|
*/
|
2015-10-29 15:22:21 +07:00
|
|
|
switch (cmd) {
|
|
|
|
|
case FS_IOC32_GETFLAGS:
|
|
|
|
|
cmd = FS_IOC_GETFLAGS;
|
|
|
|
|
break;
|
|
|
|
|
case FS_IOC32_SETFLAGS:
|
|
|
|
|
cmd = FS_IOC_SETFLAGS;
|
|
|
|
|
break;
|
|
|
|
|
case FS_IOC32_GETVERSION:
|
|
|
|
|
cmd = FS_IOC_GETVERSION;
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return btrfs_ioctl(file, cmd, (unsigned long) compat_ptr(arg));
|
|
|
|
|
}
|
|
|
|
|
#endif
|