mirror of
https://github.com/AuxXxilium/linux_dsm_epyc7002.git
synced 2024-12-28 05:25:20 +07:00
2105c2820d
AFS directories are retained locally as a structured file, with lookup
being effected by a local search of the file contents. When a modification
(such as mkdir) happens, the dir file content is modified locally rather
than redownloading the directory.
The directory contents are accessed in a number of ways, with a number of
different locks schemes:
(1) Download of contents - dvnode->validate_lock/write in afs_read_dir().
(2) Lookup and readdir - dvnode->validate_lock/read in afs_dir_iterate(),
downgrading from (1) if necessary.
(3) d_revalidate of child dentry - dvnode->validate_lock/read in
afs_do_lookup_one() downgrading from (1) if necessary.
(4) Edit of dir after modification - page locks on individual dir pages.
Unfortunately, because (4) uses different locking scheme to (1) - (3),
nothing protects against the page being scanned whilst the edit is
underway. Even download is not safe as it doesn't lock the pages - relying
instead on the validate_lock to serialise as a whole (the theory being that
directory contents are treated as a block and always downloaded as a
block).
Fix this by write-locking dvnode->validate_lock around the edits. Care
must be taken in the rename case as there may be two different dirs - but
they need not be locked at the same time. In any case, once the lock is
taken, the directory version must be rechecked, and the edit skipped if a
later version has been downloaded by revalidation (there can't have been
any local changes because the VFS holds the inode lock, but there can have
been remote changes).
Fixes: 63a4681ff3
("afs: Locally edit directory data for mkdir/create/unlink/...")
Signed-off-by: David Howells <dhowells@redhat.com>
254 lines
6.6 KiB
C
254 lines
6.6 KiB
C
// SPDX-License-Identifier: GPL-2.0-or-later
|
|
/* AFS silly rename handling
|
|
*
|
|
* Copyright (C) 2019 Red Hat, Inc. All Rights Reserved.
|
|
* Written by David Howells (dhowells@redhat.com)
|
|
* - Derived from NFS's sillyrename.
|
|
*/
|
|
|
|
#include <linux/kernel.h>
|
|
#include <linux/fs.h>
|
|
#include <linux/namei.h>
|
|
#include <linux/fsnotify.h>
|
|
#include "internal.h"
|
|
|
|
/*
|
|
* Actually perform the silly rename step.
|
|
*/
|
|
static int afs_do_silly_rename(struct afs_vnode *dvnode, struct afs_vnode *vnode,
|
|
struct dentry *old, struct dentry *new,
|
|
struct key *key)
|
|
{
|
|
struct afs_fs_cursor fc;
|
|
struct afs_status_cb *scb;
|
|
afs_dataversion_t dir_data_version;
|
|
int ret = -ERESTARTSYS;
|
|
|
|
_enter("%pd,%pd", old, new);
|
|
|
|
scb = kzalloc(sizeof(struct afs_status_cb), GFP_KERNEL);
|
|
if (!scb)
|
|
return -ENOMEM;
|
|
|
|
trace_afs_silly_rename(vnode, false);
|
|
if (afs_begin_vnode_operation(&fc, dvnode, key, true)) {
|
|
dir_data_version = dvnode->status.data_version + 1;
|
|
|
|
while (afs_select_fileserver(&fc)) {
|
|
fc.cb_break = afs_calc_vnode_cb_break(dvnode);
|
|
afs_fs_rename(&fc, old->d_name.name,
|
|
dvnode, new->d_name.name,
|
|
scb, scb);
|
|
}
|
|
|
|
afs_vnode_commit_status(&fc, dvnode, fc.cb_break,
|
|
&dir_data_version, scb);
|
|
ret = afs_end_vnode_operation(&fc);
|
|
}
|
|
|
|
if (ret == 0) {
|
|
spin_lock(&old->d_lock);
|
|
old->d_flags |= DCACHE_NFSFS_RENAMED;
|
|
spin_unlock(&old->d_lock);
|
|
if (dvnode->silly_key != key) {
|
|
key_put(dvnode->silly_key);
|
|
dvnode->silly_key = key_get(key);
|
|
}
|
|
|
|
down_write(&dvnode->validate_lock);
|
|
if (test_bit(AFS_VNODE_DIR_VALID, &dvnode->flags) &&
|
|
dvnode->status.data_version == dir_data_version) {
|
|
afs_edit_dir_remove(dvnode, &old->d_name,
|
|
afs_edit_dir_for_silly_0);
|
|
afs_edit_dir_add(dvnode, &new->d_name,
|
|
&vnode->fid, afs_edit_dir_for_silly_1);
|
|
}
|
|
up_write(&dvnode->validate_lock);
|
|
}
|
|
|
|
kfree(scb);
|
|
_leave(" = %d", ret);
|
|
return ret;
|
|
}
|
|
|
|
/**
|
|
* afs_sillyrename - Perform a silly-rename of a dentry
|
|
*
|
|
* AFS is stateless and the server doesn't know when the client is holding a
|
|
* file open. To prevent application problems when a file is unlinked while
|
|
* it's still open, the client performs a "silly-rename". That is, it renames
|
|
* the file to a hidden file in the same directory, and only performs the
|
|
* unlink once the last reference to it is put.
|
|
*
|
|
* The final cleanup is done during dentry_iput.
|
|
*/
|
|
int afs_sillyrename(struct afs_vnode *dvnode, struct afs_vnode *vnode,
|
|
struct dentry *dentry, struct key *key)
|
|
{
|
|
static unsigned int sillycounter;
|
|
struct dentry *sdentry = NULL;
|
|
unsigned char silly[16];
|
|
int ret = -EBUSY;
|
|
|
|
_enter("");
|
|
|
|
/* We don't allow a dentry to be silly-renamed twice. */
|
|
if (dentry->d_flags & DCACHE_NFSFS_RENAMED)
|
|
return -EBUSY;
|
|
|
|
sdentry = NULL;
|
|
do {
|
|
int slen;
|
|
|
|
dput(sdentry);
|
|
sillycounter++;
|
|
|
|
/* Create a silly name. Note that the ".__afs" prefix is
|
|
* understood by the salvager and must not be changed.
|
|
*/
|
|
slen = scnprintf(silly, sizeof(silly), ".__afs%04X", sillycounter);
|
|
sdentry = lookup_one_len(silly, dentry->d_parent, slen);
|
|
|
|
/* N.B. Better to return EBUSY here ... it could be dangerous
|
|
* to delete the file while it's in use.
|
|
*/
|
|
if (IS_ERR(sdentry))
|
|
goto out;
|
|
} while (!d_is_negative(sdentry));
|
|
|
|
ihold(&vnode->vfs_inode);
|
|
|
|
ret = afs_do_silly_rename(dvnode, vnode, dentry, sdentry, key);
|
|
switch (ret) {
|
|
case 0:
|
|
/* The rename succeeded. */
|
|
d_move(dentry, sdentry);
|
|
break;
|
|
case -ERESTARTSYS:
|
|
/* The result of the rename is unknown. Play it safe by forcing
|
|
* a new lookup.
|
|
*/
|
|
d_drop(dentry);
|
|
d_drop(sdentry);
|
|
}
|
|
|
|
iput(&vnode->vfs_inode);
|
|
dput(sdentry);
|
|
out:
|
|
_leave(" = %d", ret);
|
|
return ret;
|
|
}
|
|
|
|
/*
|
|
* Tell the server to remove a sillyrename file.
|
|
*/
|
|
static int afs_do_silly_unlink(struct afs_vnode *dvnode, struct afs_vnode *vnode,
|
|
struct dentry *dentry, struct key *key)
|
|
{
|
|
struct afs_fs_cursor fc;
|
|
struct afs_status_cb *scb;
|
|
int ret = -ERESTARTSYS;
|
|
|
|
_enter("");
|
|
|
|
scb = kcalloc(2, sizeof(struct afs_status_cb), GFP_KERNEL);
|
|
if (!scb)
|
|
return -ENOMEM;
|
|
|
|
trace_afs_silly_rename(vnode, true);
|
|
if (afs_begin_vnode_operation(&fc, dvnode, key, false)) {
|
|
afs_dataversion_t dir_data_version = dvnode->status.data_version + 1;
|
|
|
|
while (afs_select_fileserver(&fc)) {
|
|
fc.cb_break = afs_calc_vnode_cb_break(dvnode);
|
|
|
|
if (test_bit(AFS_SERVER_FL_IS_YFS, &fc.cbi->server->flags) &&
|
|
!test_bit(AFS_SERVER_FL_NO_RM2, &fc.cbi->server->flags)) {
|
|
yfs_fs_remove_file2(&fc, vnode, dentry->d_name.name,
|
|
&scb[0], &scb[1]);
|
|
if (fc.ac.error != -ECONNABORTED ||
|
|
fc.ac.abort_code != RXGEN_OPCODE)
|
|
continue;
|
|
set_bit(AFS_SERVER_FL_NO_RM2, &fc.cbi->server->flags);
|
|
}
|
|
|
|
afs_fs_remove(&fc, vnode, dentry->d_name.name, false, &scb[0]);
|
|
}
|
|
|
|
afs_vnode_commit_status(&fc, dvnode, fc.cb_break,
|
|
&dir_data_version, &scb[0]);
|
|
ret = afs_end_vnode_operation(&fc);
|
|
if (ret == 0) {
|
|
drop_nlink(&vnode->vfs_inode);
|
|
if (vnode->vfs_inode.i_nlink == 0) {
|
|
set_bit(AFS_VNODE_DELETED, &vnode->flags);
|
|
clear_bit(AFS_VNODE_CB_PROMISED, &vnode->flags);
|
|
}
|
|
}
|
|
if (ret == 0) {
|
|
down_write(&dvnode->validate_lock);
|
|
if (test_bit(AFS_VNODE_DIR_VALID, &dvnode->flags) &&
|
|
dvnode->status.data_version == dir_data_version)
|
|
afs_edit_dir_remove(dvnode, &dentry->d_name,
|
|
afs_edit_dir_for_unlink);
|
|
up_write(&dvnode->validate_lock);
|
|
}
|
|
}
|
|
|
|
kfree(scb);
|
|
_leave(" = %d", ret);
|
|
return ret;
|
|
}
|
|
|
|
/*
|
|
* Remove sillyrename file on iput.
|
|
*/
|
|
int afs_silly_iput(struct dentry *dentry, struct inode *inode)
|
|
{
|
|
struct afs_vnode *dvnode = AFS_FS_I(d_inode(dentry->d_parent));
|
|
struct afs_vnode *vnode = AFS_FS_I(inode);
|
|
struct dentry *alias;
|
|
int ret;
|
|
|
|
DECLARE_WAIT_QUEUE_HEAD_ONSTACK(wq);
|
|
|
|
_enter("%p{%pd},%llx", dentry, dentry, vnode->fid.vnode);
|
|
|
|
down_read(&dvnode->rmdir_lock);
|
|
|
|
alias = d_alloc_parallel(dentry->d_parent, &dentry->d_name, &wq);
|
|
if (IS_ERR(alias)) {
|
|
up_read(&dvnode->rmdir_lock);
|
|
return 0;
|
|
}
|
|
|
|
if (!d_in_lookup(alias)) {
|
|
/* We raced with lookup... See if we need to transfer the
|
|
* sillyrename information to the aliased dentry.
|
|
*/
|
|
ret = 0;
|
|
spin_lock(&alias->d_lock);
|
|
if (d_really_is_positive(alias) &&
|
|
!(alias->d_flags & DCACHE_NFSFS_RENAMED)) {
|
|
alias->d_flags |= DCACHE_NFSFS_RENAMED;
|
|
ret = 1;
|
|
}
|
|
spin_unlock(&alias->d_lock);
|
|
up_read(&dvnode->rmdir_lock);
|
|
dput(alias);
|
|
return ret;
|
|
}
|
|
|
|
/* Stop lock-release from complaining. */
|
|
spin_lock(&vnode->lock);
|
|
vnode->lock_state = AFS_VNODE_LOCK_DELETED;
|
|
trace_afs_flock_ev(vnode, NULL, afs_flock_silly_delete, 0);
|
|
spin_unlock(&vnode->lock);
|
|
|
|
afs_do_silly_unlink(dvnode, vnode, dentry, dvnode->silly_key);
|
|
up_read(&dvnode->rmdir_lock);
|
|
d_lookup_done(alias);
|
|
dput(alias);
|
|
return 1;
|
|
}
|