linux_dsm_epyc7002/fs/xfs/xfs_mount.h

336 lines
13 KiB
C
Raw Normal View History

/*
* Copyright (c) 2000-2005 Silicon Graphics, Inc.
* All Rights Reserved.
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License as
* published by the Free Software Foundation.
*
* This program is distributed in the hope that it would be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write the Free Software Foundation,
* Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
*/
#ifndef __XFS_MOUNT_H__
#define __XFS_MOUNT_H__
struct xlog;
struct xfs_inode;
[XFS] Concurrent Multi-File Data Streams In media spaces, video is often stored in a frame-per-file format. When dealing with uncompressed realtime HD video streams in this format, it is crucial that files do not get fragmented and that multiple files a placed contiguously on disk. When multiple streams are being ingested and played out at the same time, it is critical that the filesystem does not cross the streams and interleave them together as this creates seek and readahead cache miss latency and prevents both ingest and playout from meeting frame rate targets. This patch set creates a "stream of files" concept into the allocator to place all the data from a single stream contiguously on disk so that RAID array readahead can be used effectively. Each additional stream gets placed in different allocation groups within the filesystem, thereby ensuring that we don't cross any streams. When an AG fills up, we select a new AG for the stream that is not in use. The core of the functionality is the stream tracking - each inode that we create in a directory needs to be associated with the directories' stream. Hence every time we create a file, we look up the directories' stream object and associate the new file with that object. Once we have a stream object for a file, we use the AG that the stream object point to for allocations. If we can't allocate in that AG (e.g. it is full) we move the entire stream to another AG. Other inodes in the same stream are moved to the new AG on their next allocation (i.e. lazy update). Stream objects are kept in a cache and hold a reference on the inode. Hence the inode cannot be reclaimed while there is an outstanding stream reference. This means that on unlink we need to remove the stream association and we also need to flush all the associations on certain events that want to reclaim all unreferenced inodes (e.g. filesystem freeze). SGI-PV: 964469 SGI-Modid: xfs-linux-melb:xfs-kern:29096a Signed-off-by: David Chinner <dgc@sgi.com> Signed-off-by: Barry Naujok <bnaujok@sgi.com> Signed-off-by: Donald Douwsma <donaldd@sgi.com> Signed-off-by: Christoph Hellwig <hch@infradead.org> Signed-off-by: Tim Shimmin <tes@sgi.com> Signed-off-by: Vlad Apostolov <vapo@sgi.com>
2007-07-11 08:09:12 +07:00
struct xfs_mru_cache;
struct xfs_nameops;
struct xfs_ail;
struct xfs_quotainfo;
xfs: abstract the differences in dir2/dir3 via an ops vector Lots of the dir code now goes through switches to determine what is the correct on-disk format to parse. It generally involves a "xfs_sbversion_hasfoo" check, deferencing the superblock version and feature fields and hence touching several cache lines per operation in the process. Some operations do multiple checks because they nest conditional operations and they don't pass the information in a direct fashion between each other. Hence, add an ops vector to the xfs_inode structure that is configured when the inode is initialised to point to all the correct decode and encoding operations. This will significantly reduce the branchiness and cacheline footprint of the directory object decoding and encoding. This is the first patch in a series of conversion patches. It will introduce the ops structure, the setup of it and add the first operation to the vector. Subsequent patches will convert directory ops one at a time to keep the changes simple and obvious. Just this patch shows the benefit of such an approach on code size. Just converting the two shortform dir operations as this patch does decreases the built binary size by ~1500 bytes: $ size fs/xfs/xfs.o.orig fs/xfs/xfs.o.p1 text data bss dec hex filename 794490 96802 1096 892388 d9de4 fs/xfs/xfs.o.orig 792986 96802 1096 890884 d9804 fs/xfs/xfs.o.p1 $ That's a significant decrease in the instruction cache footprint of the directory code for such a simple change, and indicates that this approach is definitely worth pursuing further. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Ben Myers <bpm@sgi.com>
2013-10-29 18:11:46 +07:00
struct xfs_dir_ops;
struct xfs_da_geometry;
xfs: dynamic speculative EOF preallocation Currently the size of the speculative preallocation during delayed allocation is fixed by either the allocsize mount option of a default size. We are seeing a lot of cases where we need to recommend using the allocsize mount option to prevent fragmentation when buffered writes land in the same AG. Rather than using a fixed preallocation size by default (up to 64k), make it dynamic by basing it on the current inode size. That way the EOF preallocation will increase as the file size increases. Hence for streaming writes we are much more likely to get large preallocations exactly when we need it to reduce fragementation. For default settings, the size of the initial extents is determined by the number of parallel writers and the amount of memory in the machine. For 4GB RAM and 4 concurrent 32GB file writes: EXT: FILE-OFFSET BLOCK-RANGE AG AG-OFFSET TOTAL 0: [0..1048575]: 1048672..2097247 0 (1048672..2097247) 1048576 1: [1048576..2097151]: 5242976..6291551 0 (5242976..6291551) 1048576 2: [2097152..4194303]: 12583008..14680159 0 (12583008..14680159) 2097152 3: [4194304..8388607]: 25165920..29360223 0 (25165920..29360223) 4194304 4: [8388608..16777215]: 58720352..67108959 0 (58720352..67108959) 8388608 5: [16777216..33554423]: 117440584..134217791 0 (117440584..134217791) 16777208 6: [33554424..50331511]: 184549056..201326143 0 (184549056..201326143) 16777088 7: [50331512..67108599]: 251657408..268434495 0 (251657408..268434495) 16777088 and for 16 concurrent 16GB file writes: EXT: FILE-OFFSET BLOCK-RANGE AG AG-OFFSET TOTAL 0: [0..262143]: 2490472..2752615 0 (2490472..2752615) 262144 1: [262144..524287]: 6291560..6553703 0 (6291560..6553703) 262144 2: [524288..1048575]: 13631592..14155879 0 (13631592..14155879) 524288 3: [1048576..2097151]: 30408808..31457383 0 (30408808..31457383) 1048576 4: [2097152..4194303]: 52428904..54526055 0 (52428904..54526055) 2097152 5: [4194304..8388607]: 104857704..109052007 0 (104857704..109052007) 4194304 6: [8388608..16777215]: 209715304..218103911 0 (209715304..218103911) 8388608 7: [16777216..33554423]: 452984848..469762055 0 (452984848..469762055) 16777208 Because it is hard to take back specualtive preallocation, cases where there are large slow growing log files on a nearly full filesystem may cause premature ENOSPC. Hence as the filesystem nears full, the maximum dynamic prealloc size іs reduced according to this table (based on 4k block size): freespace max prealloc size >5% full extent (8GB) 4-5% 2GB (8GB >> 2) 3-4% 1GB (8GB >> 3) 2-3% 512MB (8GB >> 4) 1-2% 256MB (8GB >> 5) <1% 128MB (8GB >> 6) This should reduce the amount of space held in speculative preallocation for such cases. The allocsize mount option turns off the dynamic behaviour and fixes the prealloc size to whatever the mount option specifies. i.e. the behaviour is unchanged. Signed-off-by: Dave Chinner <dchinner@redhat.com>
2011-01-04 07:35:03 +07:00
/* dynamic preallocation free space thresholds, 5% down to 1% */
enum {
XFS_LOWSP_1_PCNT = 0,
XFS_LOWSP_2_PCNT,
XFS_LOWSP_3_PCNT,
XFS_LOWSP_4_PCNT,
XFS_LOWSP_5_PCNT,
XFS_LOWSP_MAX,
};
typedef struct xfs_mount {
struct super_block *m_super;
xfs_tid_t m_tid; /* next unused tid for fs */
struct xfs_ail *m_ail; /* fs active log item list */
struct xfs_sb m_sb; /* copy of fs superblock */
spinlock_t m_sb_lock; /* sb counter lock */
struct percpu_counter m_icount; /* allocated inodes counter */
struct percpu_counter m_ifree; /* free inodes counter */
struct percpu_counter m_fdblocks; /* free block counter */
struct xfs_buf *m_sb_bp; /* buffer for superblock */
char *m_fsname; /* filesystem name */
int m_fsname_len; /* strlen of fs name */
char *m_rtname; /* realtime device name */
char *m_logname; /* external log device name */
int m_bsize; /* fs logical block size */
xfs_agnumber_t m_agfrotor; /* last ag where space found */
xfs_agnumber_t m_agirotor; /* last ag dir inode alloced */
spinlock_t m_agirotor_lock;/* .. and lock protecting it */
xfs_agnumber_t m_maxagi; /* highest inode alloc group */
uint m_readio_log; /* min read size log bytes */
uint m_readio_blocks; /* min read size blocks */
uint m_writeio_log; /* min write size log bytes */
uint m_writeio_blocks; /* min write size blocks */
struct xfs_da_geometry *m_dir_geo; /* directory block geometry */
struct xfs_da_geometry *m_attr_geo; /* attribute block geometry */
struct xlog *m_log; /* log specific stuff */
int m_logbufs; /* number of log buffers */
int m_logbsize; /* size of each log buffer */
uint m_rsumlevels; /* rt summary levels */
uint m_rsumsize; /* size of rt summary, bytes */
struct xfs_inode *m_rbmip; /* pointer to bitmap inode */
struct xfs_inode *m_rsumip; /* pointer to summary inode */
struct xfs_inode *m_rootip; /* pointer to root directory */
struct xfs_quotainfo *m_quotainfo; /* disk quota information */
xfs_buftarg_t *m_ddev_targp; /* saves taking the address */
xfs_buftarg_t *m_logdev_targp;/* ptr to log device */
xfs_buftarg_t *m_rtdev_targp; /* ptr to rt device */
__uint8_t m_blkbit_log; /* blocklog + NBBY */
__uint8_t m_blkbb_log; /* blocklog - BBSHIFT */
__uint8_t m_agno_log; /* log #ag's */
__uint8_t m_agino_log; /* #bits for agino in inum */
xfs: increase inode cluster size for v5 filesystems v5 filesystems use 512 byte inodes as a minimum, so read inodes in clusters that are effectively half the size of a v4 filesystem with 256 byte inodes. For v5 fielsystems, scale the inode cluster size with the size of the inode so that we keep a constant 32 inodes per cluster ratio for all inode IO. This only works if mkfs.xfs sets the inode alignment appropriately for larger inode clusters, so this functionality is made conditional on mkfs doing the right thing. xfs_repair needs to know about the inode alignment changes, too. Wall time: create bulkstat find+stat ls -R unlink v4 237s 161s 173s 201s 299s v5 235s 163s 205s 31s 356s patched 234s 160s 182s 29s 317s System time: create bulkstat find+stat ls -R unlink v4 2601s 2490s 1653s 1656s 2960s v5 2637s 2497s 1681s 20s 3216s patched 2613s 2451s 1658s 20s 3007s So, wall time same or down across the board, system time same or down across the board, and cache hit rates all improve except for the ls -R case which is a pure cold cache directory read workload on v5 filesystems... So, this patch removes most of the performance and CPU usage differential between v4 and v5 filesystems on traversal related workloads. Note: while this patch is currently for v5 filesystems only, there is no reason it can't be ported back to v4 filesystems. This hasn't been done here because bringing the code back to v4 requires forwards and backwards kernel compatibility testing. i.e. to deterine if older kernels(*) do the right thing with larger inode alignments but still only using 8k inode cluster sizes. None of this testing and validation on v4 filesystems has been done, so for the moment larger inode clusters is limited to v5 superblocks. (*) a current default config v4 filesystem should mount just fine on 2.6.23 (when lazy-count support was introduced), and so if we change the alignment emitted by mkfs without a feature bit then we have to make sure it works properly on all kernels since 2.6.23. And if we allow it to be changed when the lazy-count bit is not set, then it's all kernels since v2 logs were introduced that need to be tested for compatibility... Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Eric Sandeen <sandeen@redhat.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2013-11-01 11:27:20 +07:00
uint m_inode_cluster_size;/* min inode buf size */
uint m_blockmask; /* sb_blocksize-1 */
uint m_blockwsize; /* sb_blocksize in words */
uint m_blockwmask; /* blockwsize-1 */
uint m_alloc_mxr[2]; /* max alloc btree records */
uint m_alloc_mnr[2]; /* min alloc btree records */
uint m_bmap_dmxr[2]; /* max bmap btree records */
uint m_bmap_dmnr[2]; /* min bmap btree records */
uint m_inobt_mxr[2]; /* max inobt btree records */
uint m_inobt_mnr[2]; /* min inobt btree records */
uint m_ag_maxlevels; /* XFS_AG_MAXLEVELS */
uint m_bm_maxlevels[2]; /* XFS_BM_MAXLEVELS */
uint m_in_maxlevels; /* max inobt btree levels. */
xfs: Replace per-ag array with a radix tree The use of an array for the per-ag structures requires reallocation of the array when growing the filesystem. This requires locking access to the array to avoid use after free situations, and the locking is difficult to get right. To avoid needing to reallocate an array, change the per-ag structures to an allocated object per ag and index them using a tree structure. The AGs are always densely indexed (hence the use of an array), but the number supported is 2^32 and lookups tend to be random and hence indexing needs to scale. A simple choice is a radix tree - it works well with this sort of index. This change also removes another large contiguous allocation from the mount/growfs path in XFS. The growing process now needs to change to only initialise the new AGs required for the extra space, and as such only needs to exclusively lock the tree for inserts. The rest of the code only needs to lock the tree while doing lookups, and hence this will remove all the deadlocks that currently occur on the m_perag_lock as it is now an innermost lock. The lock is also changed to a spinlock from a read/write lock as the hold time is now extremely short. To complete the picture, the per-ag structures will need to be reference counted to ensure that we don't free/modify them while they are still in use. This will be done in subsequent patch. Signed-off-by: Dave Chinner <david@fromorbit.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Alex Elder <aelder@sgi.com>
2010-01-11 18:47:44 +07:00
struct radix_tree_root m_perag_tree; /* per-ag accounting info */
spinlock_t m_perag_lock; /* lock for m_perag_tree */
struct mutex m_growlock; /* growfs mutex */
int m_fixedfsid[2]; /* unchanged for life of FS */
uint m_dmevmask; /* DMI events for this FS */
__uint64_t m_flags; /* global mount flags */
int m_ialloc_inos; /* inodes in inode allocation */
int m_ialloc_blks; /* blocks in inode allocation */
int m_inoalign_mask;/* mask sb_inoalignmt if used */
uint m_qflags; /* quota status flags */
struct xfs_trans_resv m_resv; /* precomputed res values */
__uint64_t m_maxicount; /* maximum inode count */
__uint64_t m_resblks; /* total reserved blocks */
__uint64_t m_resblks_avail;/* available reserved blocks */
__uint64_t m_resblks_save; /* reserved blks @ remount,ro */
int m_dalign; /* stripe unit */
int m_swidth; /* stripe width */
int m_sinoalign; /* stripe unit inode alignment */
__uint8_t m_sectbb_log; /* sectlog - BBSHIFT */
const struct xfs_nameops *m_dirnameops; /* vector of dir name ops */
xfs: abstract the differences in dir2/dir3 via an ops vector Lots of the dir code now goes through switches to determine what is the correct on-disk format to parse. It generally involves a "xfs_sbversion_hasfoo" check, deferencing the superblock version and feature fields and hence touching several cache lines per operation in the process. Some operations do multiple checks because they nest conditional operations and they don't pass the information in a direct fashion between each other. Hence, add an ops vector to the xfs_inode structure that is configured when the inode is initialised to point to all the correct decode and encoding operations. This will significantly reduce the branchiness and cacheline footprint of the directory object decoding and encoding. This is the first patch in a series of conversion patches. It will introduce the ops structure, the setup of it and add the first operation to the vector. Subsequent patches will convert directory ops one at a time to keep the changes simple and obvious. Just this patch shows the benefit of such an approach on code size. Just converting the two shortform dir operations as this patch does decreases the built binary size by ~1500 bytes: $ size fs/xfs/xfs.o.orig fs/xfs/xfs.o.p1 text data bss dec hex filename 794490 96802 1096 892388 d9de4 fs/xfs/xfs.o.orig 792986 96802 1096 890884 d9804 fs/xfs/xfs.o.p1 $ That's a significant decrease in the instruction cache footprint of the directory code for such a simple change, and indicates that this approach is definitely worth pursuing further. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Ben Myers <bpm@sgi.com>
2013-10-29 18:11:46 +07:00
const struct xfs_dir_ops *m_dir_inode_ops; /* vector of dir inode ops */
const struct xfs_dir_ops *m_nondir_inode_ops; /* !dir inode ops */
uint m_chsize; /* size of next field */
atomic_t m_active_trans; /* number trans frozen */
[XFS] Concurrent Multi-File Data Streams In media spaces, video is often stored in a frame-per-file format. When dealing with uncompressed realtime HD video streams in this format, it is crucial that files do not get fragmented and that multiple files a placed contiguously on disk. When multiple streams are being ingested and played out at the same time, it is critical that the filesystem does not cross the streams and interleave them together as this creates seek and readahead cache miss latency and prevents both ingest and playout from meeting frame rate targets. This patch set creates a "stream of files" concept into the allocator to place all the data from a single stream contiguously on disk so that RAID array readahead can be used effectively. Each additional stream gets placed in different allocation groups within the filesystem, thereby ensuring that we don't cross any streams. When an AG fills up, we select a new AG for the stream that is not in use. The core of the functionality is the stream tracking - each inode that we create in a directory needs to be associated with the directories' stream. Hence every time we create a file, we look up the directories' stream object and associate the new file with that object. Once we have a stream object for a file, we use the AG that the stream object point to for allocations. If we can't allocate in that AG (e.g. it is full) we move the entire stream to another AG. Other inodes in the same stream are moved to the new AG on their next allocation (i.e. lazy update). Stream objects are kept in a cache and hold a reference on the inode. Hence the inode cannot be reclaimed while there is an outstanding stream reference. This means that on unlink we need to remove the stream association and we also need to flush all the associations on certain events that want to reclaim all unreferenced inodes (e.g. filesystem freeze). SGI-PV: 964469 SGI-Modid: xfs-linux-melb:xfs-kern:29096a Signed-off-by: David Chinner <dgc@sgi.com> Signed-off-by: Barry Naujok <bnaujok@sgi.com> Signed-off-by: Donald Douwsma <donaldd@sgi.com> Signed-off-by: Christoph Hellwig <hch@infradead.org> Signed-off-by: Tim Shimmin <tes@sgi.com> Signed-off-by: Vlad Apostolov <vapo@sgi.com>
2007-07-11 08:09:12 +07:00
struct xfs_mru_cache *m_filestream; /* per-mount filestream data */
xfs: introduce background inode reclaim work Background inode reclaim needs to run more frequently that the XFS syncd work is run as 30s is too long between optimal reclaim runs. Add a new periodic work item to the xfs syncd workqueue to run a fast, non-blocking inode reclaim scan. Background inode reclaim is kicked by the act of marking inodes for reclaim. When an AG is first marked as having reclaimable inodes, the background reclaim work is kicked. It will continue to run periodically untill it detects that there are no more reclaimable inodes. It will be kicked again when the first inode is queued for reclaim. To ensure shrinker based inode reclaim throttles to the inode cleaning and reclaim rate but still reclaim inodes efficiently, make it kick the background inode reclaim so that when we are low on memory we are trying to reclaim inodes as efficiently as possible. This kick shoul d not be necessary, but it will protect against failures to kick the background reclaim when inodes are first dirtied. To provide the rate throttling, make the shrinker pass do synchronous inode reclaim so that it blocks on inodes under IO. This means that the shrinker will reclaim inodes rather than just skipping over them, but it does not adversely affect the rate of reclaim because most dirty inodes are already under IO due to the background reclaim work the shrinker kicked. These two modifications solve one of the two OOM killer invocations Chris Mason reported recently when running a stress testing script. The particular workload trigger for the OOM killer invocation is where there are more threads than CPUs all unlinking files in an extremely memory constrained environment. Unlike other solutions, this one does not have a performance impact on performance when memory is not constrained or the number of concurrent threads operating is <= to the number of CPUs. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Alex Elder <aelder@sgi.com>
2011-04-08 09:45:07 +07:00
struct delayed_work m_reclaim_work; /* background inode reclaim */
struct delayed_work m_eofblocks_work; /* background eof blocks
trimming */
bool m_update_sb; /* sb needs update in mount */
xfs: dynamic speculative EOF preallocation Currently the size of the speculative preallocation during delayed allocation is fixed by either the allocsize mount option of a default size. We are seeing a lot of cases where we need to recommend using the allocsize mount option to prevent fragmentation when buffered writes land in the same AG. Rather than using a fixed preallocation size by default (up to 64k), make it dynamic by basing it on the current inode size. That way the EOF preallocation will increase as the file size increases. Hence for streaming writes we are much more likely to get large preallocations exactly when we need it to reduce fragementation. For default settings, the size of the initial extents is determined by the number of parallel writers and the amount of memory in the machine. For 4GB RAM and 4 concurrent 32GB file writes: EXT: FILE-OFFSET BLOCK-RANGE AG AG-OFFSET TOTAL 0: [0..1048575]: 1048672..2097247 0 (1048672..2097247) 1048576 1: [1048576..2097151]: 5242976..6291551 0 (5242976..6291551) 1048576 2: [2097152..4194303]: 12583008..14680159 0 (12583008..14680159) 2097152 3: [4194304..8388607]: 25165920..29360223 0 (25165920..29360223) 4194304 4: [8388608..16777215]: 58720352..67108959 0 (58720352..67108959) 8388608 5: [16777216..33554423]: 117440584..134217791 0 (117440584..134217791) 16777208 6: [33554424..50331511]: 184549056..201326143 0 (184549056..201326143) 16777088 7: [50331512..67108599]: 251657408..268434495 0 (251657408..268434495) 16777088 and for 16 concurrent 16GB file writes: EXT: FILE-OFFSET BLOCK-RANGE AG AG-OFFSET TOTAL 0: [0..262143]: 2490472..2752615 0 (2490472..2752615) 262144 1: [262144..524287]: 6291560..6553703 0 (6291560..6553703) 262144 2: [524288..1048575]: 13631592..14155879 0 (13631592..14155879) 524288 3: [1048576..2097151]: 30408808..31457383 0 (30408808..31457383) 1048576 4: [2097152..4194303]: 52428904..54526055 0 (52428904..54526055) 2097152 5: [4194304..8388607]: 104857704..109052007 0 (104857704..109052007) 4194304 6: [8388608..16777215]: 209715304..218103911 0 (209715304..218103911) 8388608 7: [16777216..33554423]: 452984848..469762055 0 (452984848..469762055) 16777208 Because it is hard to take back specualtive preallocation, cases where there are large slow growing log files on a nearly full filesystem may cause premature ENOSPC. Hence as the filesystem nears full, the maximum dynamic prealloc size іs reduced according to this table (based on 4k block size): freespace max prealloc size >5% full extent (8GB) 4-5% 2GB (8GB >> 2) 3-4% 1GB (8GB >> 3) 2-3% 512MB (8GB >> 4) 1-2% 256MB (8GB >> 5) <1% 128MB (8GB >> 6) This should reduce the amount of space held in speculative preallocation for such cases. The allocsize mount option turns off the dynamic behaviour and fixes the prealloc size to whatever the mount option specifies. i.e. the behaviour is unchanged. Signed-off-by: Dave Chinner <dchinner@redhat.com>
2011-01-04 07:35:03 +07:00
int64_t m_low_space[XFS_LOWSP_MAX];
/* low free space thresholds */
struct xfs_kobj m_kobj;
xfs: replace global xfslogd wq with per-mount wq The xfslogd workqueue is a global, single-job workqueue for buffer ioend processing. This means we allow for a single work item at a time for all possible XFS mounts on a system. fsstress testing in loopback XFS over XFS configurations has reproduced xfslogd deadlocks due to the single threaded nature of the queue and dependencies introduced between the separate XFS instances by online discard (-o discard). Discard over a loopback device converts the discard request to a hole punch (fallocate) on the underlying file. Online discard requests are issued synchronously and from xfslogd context in XFS, hence the xfslogd workqueue is blocked in the upper fs waiting on a hole punch request to be servied in the lower fs. If the lower fs issues I/O that depends on xfslogd to complete, both filesystems end up hung indefinitely. This is reproduced reliabily by generic/013 on XFS->loop->XFS test devices with the '-o discard' mount option. Further, docker implementations appear to use this kind of configuration for container instance filesystems by default (container fs->dm-> loop->base fs) and therefore are subject to this deadlock when running on XFS. Replace the global xfslogd workqueue with a per-mount variant. This guarantees each mount access to a single worker and prevents deadlocks due to inter-fs dependencies introduced by discard. Since the queue is only responsible for buffer iodone processing at this point in time, rename xfslogd to xfs-buf. Signed-off-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Dave Chinner <dchinner@redhat.com> Signed-off-by: Dave Chinner <david@fromorbit.com>
2014-11-28 09:59:58 +07:00
struct workqueue_struct *m_buf_workqueue;
struct workqueue_struct *m_data_workqueue;
struct workqueue_struct *m_unwritten_workqueue;
struct workqueue_struct *m_cil_workqueue;
struct workqueue_struct *m_reclaim_workqueue;
struct workqueue_struct *m_log_workqueue;
struct workqueue_struct *m_eofblocks_workqueue;
/*
* Generation of the filesysyem layout. This is incremented by each
* growfs, and used by the pNFS server to ensure the client updates
* its view of the block device once it gets a layout that might
* reference the newly added blocks. Does not need to be persistent
* as long as we only allow file system size increments, but if we
* ever support shrinks it would have to be persisted in addition
* to various other kinds of pain inflicted on the pNFS server.
*/
__uint32_t m_generation;
} xfs_mount_t;
/*
* Flags for m_flags.
*/
[XFS] Lazy Superblock Counters When we have a couple of hundred transactions on the fly at once, they all typically modify the on disk superblock in some way. create/unclink/mkdir/rmdir modify inode counts, allocation/freeing modify free block counts. When these counts are modified in a transaction, they must eventually lock the superblock buffer and apply the mods. The buffer then remains locked until the transaction is committed into the incore log buffer. The result of this is that with enough transactions on the fly the incore superblock buffer becomes a bottleneck. The result of contention on the incore superblock buffer is that transaction rates fall - the more pressure that is put on the superblock buffer, the slower things go. The key to removing the contention is to not require the superblock fields in question to be locked. We do that by not marking the superblock dirty in the transaction. IOWs, we modify the incore superblock but do not modify the cached superblock buffer. In short, we do not log superblock modifications to critical fields in the superblock on every transaction. In fact we only do it just before we write the superblock to disk every sync period or just before unmount. This creates an interesting problem - if we don't log or write out the fields in every transaction, then how do the values get recovered after a crash? the answer is simple - we keep enough duplicate, logged information in other structures that we can reconstruct the correct count after log recovery has been performed. It is the AGF and AGI structures that contain the duplicate information; after recovery, we walk every AGI and AGF and sum their individual counters to get the correct value, and we do a transaction into the log to correct them. An optimisation of this is that if we have a clean unmount record, we know the value in the superblock is correct, so we can avoid the summation walk under normal conditions and so mount/recovery times do not change under normal operation. One wrinkle that was discovered during development was that the blocks used in the freespace btrees are never accounted for in the AGF counters. This was once a valid optimisation to make; when the filesystem is full, the free space btrees are empty and consume no space. Hence when it matters, the "accounting" is correct. But that means the when we do the AGF summations, we would not have a correct count and xfs_check would complain. Hence a new counter was added to track the number of blocks used by the free space btrees. This is an *on-disk format change*. As a result of this, lazy superblock counters are a mkfs option and at the moment on linux there is no way to convert an old filesystem. This is possible - xfs_db can be used to twiddle the right bits and then xfs_repair will do the format conversion for you. Similarly, you can convert backwards as well. At some point we'll add functionality to xfs_admin to do the bit twiddling easily.... SGI-PV: 964999 SGI-Modid: xfs-linux-melb:xfs-kern:28652a Signed-off-by: David Chinner <dgc@sgi.com> Signed-off-by: Christoph Hellwig <hch@infradead.org> Signed-off-by: Tim Shimmin <tes@sgi.com>
2007-05-24 12:26:31 +07:00
#define XFS_MOUNT_WSYNC (1ULL << 0) /* for nfs - all metadata ops
must be synchronous except
for space allocations */
[XFS] Lazy Superblock Counters When we have a couple of hundred transactions on the fly at once, they all typically modify the on disk superblock in some way. create/unclink/mkdir/rmdir modify inode counts, allocation/freeing modify free block counts. When these counts are modified in a transaction, they must eventually lock the superblock buffer and apply the mods. The buffer then remains locked until the transaction is committed into the incore log buffer. The result of this is that with enough transactions on the fly the incore superblock buffer becomes a bottleneck. The result of contention on the incore superblock buffer is that transaction rates fall - the more pressure that is put on the superblock buffer, the slower things go. The key to removing the contention is to not require the superblock fields in question to be locked. We do that by not marking the superblock dirty in the transaction. IOWs, we modify the incore superblock but do not modify the cached superblock buffer. In short, we do not log superblock modifications to critical fields in the superblock on every transaction. In fact we only do it just before we write the superblock to disk every sync period or just before unmount. This creates an interesting problem - if we don't log or write out the fields in every transaction, then how do the values get recovered after a crash? the answer is simple - we keep enough duplicate, logged information in other structures that we can reconstruct the correct count after log recovery has been performed. It is the AGF and AGI structures that contain the duplicate information; after recovery, we walk every AGI and AGF and sum their individual counters to get the correct value, and we do a transaction into the log to correct them. An optimisation of this is that if we have a clean unmount record, we know the value in the superblock is correct, so we can avoid the summation walk under normal conditions and so mount/recovery times do not change under normal operation. One wrinkle that was discovered during development was that the blocks used in the freespace btrees are never accounted for in the AGF counters. This was once a valid optimisation to make; when the filesystem is full, the free space btrees are empty and consume no space. Hence when it matters, the "accounting" is correct. But that means the when we do the AGF summations, we would not have a correct count and xfs_check would complain. Hence a new counter was added to track the number of blocks used by the free space btrees. This is an *on-disk format change*. As a result of this, lazy superblock counters are a mkfs option and at the moment on linux there is no way to convert an old filesystem. This is possible - xfs_db can be used to twiddle the right bits and then xfs_repair will do the format conversion for you. Similarly, you can convert backwards as well. At some point we'll add functionality to xfs_admin to do the bit twiddling easily.... SGI-PV: 964999 SGI-Modid: xfs-linux-melb:xfs-kern:28652a Signed-off-by: David Chinner <dgc@sgi.com> Signed-off-by: Christoph Hellwig <hch@infradead.org> Signed-off-by: Tim Shimmin <tes@sgi.com>
2007-05-24 12:26:31 +07:00
#define XFS_MOUNT_WAS_CLEAN (1ULL << 3)
#define XFS_MOUNT_FS_SHUTDOWN (1ULL << 4) /* atomic stop of all filesystem
operations, typically for
disk errors in metadata */
#define XFS_MOUNT_DISCARD (1ULL << 5) /* discard unused blocks */
#define XFS_MOUNT_NOALIGN (1ULL << 7) /* turn off stripe alignment
allocations */
#define XFS_MOUNT_ATTR2 (1ULL << 8) /* allow use of attr2 format */
#define XFS_MOUNT_GRPID (1ULL << 9) /* group-ID assigned from directory */
#define XFS_MOUNT_NORECOVERY (1ULL << 10) /* no recovery - dirty fs */
#define XFS_MOUNT_DFLT_IOSIZE (1ULL << 12) /* set default i/o size */
#define XFS_MOUNT_32BITINODES (1ULL << 14) /* do not create inodes above
* 32 bits in size */
#define XFS_MOUNT_SMALL_INUMS (1ULL << 15) /* users wants 32bit inodes */
#define XFS_MOUNT_NOUUID (1ULL << 16) /* ignore uuid during mount */
#define XFS_MOUNT_BARRIER (1ULL << 17)
#define XFS_MOUNT_IKEEP (1ULL << 18) /* keep empty inode clusters*/
#define XFS_MOUNT_SWALLOC (1ULL << 19) /* turn on stripe width
* allocation */
#define XFS_MOUNT_RDONLY (1ULL << 20) /* read-only fs */
#define XFS_MOUNT_DIRSYNC (1ULL << 21) /* synchronous directory ops */
#define XFS_MOUNT_COMPAT_IOSIZE (1ULL << 22) /* don't report large preferred
* I/O size in stat() */
[XFS] Concurrent Multi-File Data Streams In media spaces, video is often stored in a frame-per-file format. When dealing with uncompressed realtime HD video streams in this format, it is crucial that files do not get fragmented and that multiple files a placed contiguously on disk. When multiple streams are being ingested and played out at the same time, it is critical that the filesystem does not cross the streams and interleave them together as this creates seek and readahead cache miss latency and prevents both ingest and playout from meeting frame rate targets. This patch set creates a "stream of files" concept into the allocator to place all the data from a single stream contiguously on disk so that RAID array readahead can be used effectively. Each additional stream gets placed in different allocation groups within the filesystem, thereby ensuring that we don't cross any streams. When an AG fills up, we select a new AG for the stream that is not in use. The core of the functionality is the stream tracking - each inode that we create in a directory needs to be associated with the directories' stream. Hence every time we create a file, we look up the directories' stream object and associate the new file with that object. Once we have a stream object for a file, we use the AG that the stream object point to for allocations. If we can't allocate in that AG (e.g. it is full) we move the entire stream to another AG. Other inodes in the same stream are moved to the new AG on their next allocation (i.e. lazy update). Stream objects are kept in a cache and hold a reference on the inode. Hence the inode cannot be reclaimed while there is an outstanding stream reference. This means that on unlink we need to remove the stream association and we also need to flush all the associations on certain events that want to reclaim all unreferenced inodes (e.g. filesystem freeze). SGI-PV: 964469 SGI-Modid: xfs-linux-melb:xfs-kern:29096a Signed-off-by: David Chinner <dgc@sgi.com> Signed-off-by: Barry Naujok <bnaujok@sgi.com> Signed-off-by: Donald Douwsma <donaldd@sgi.com> Signed-off-by: Christoph Hellwig <hch@infradead.org> Signed-off-by: Tim Shimmin <tes@sgi.com> Signed-off-by: Vlad Apostolov <vapo@sgi.com>
2007-07-11 08:09:12 +07:00
#define XFS_MOUNT_FILESTREAMS (1ULL << 24) /* enable the filestreams
allocator */
#define XFS_MOUNT_NOATTR2 (1ULL << 25) /* disable use of attr2 format */
/*
* Default minimum read and write sizes.
*/
#define XFS_READIO_LOG_LARGE 16
#define XFS_WRITEIO_LOG_LARGE 16
/*
* Max and min values for mount-option defined I/O
* preallocation sizes.
*/
#define XFS_MAX_IO_LOG 30 /* 1G */
#define XFS_MIN_IO_LOG PAGE_SHIFT
/*
* Synchronous read and write sizes. This should be
* better for NFSv2 wsync filesystems.
*/
#define XFS_WSYNC_READIO_LOG 15 /* 32k */
#define XFS_WSYNC_WRITEIO_LOG 14 /* 16k */
/*
* Allow large block sizes to be reported to userspace programs if the
* "largeio" mount option is used.
*
* If compatibility mode is specified, simply return the basic unit of caching
* so that we don't get inefficient read/modify/write I/O from user apps.
* Otherwise....
*
* If the underlying volume is a stripe, then return the stripe width in bytes
* as the recommended I/O size. It is not a stripe and we've set a default
* buffered I/O size, return that, otherwise return the compat default.
*/
static inline unsigned long
xfs_preferred_iosize(xfs_mount_t *mp)
{
if (mp->m_flags & XFS_MOUNT_COMPAT_IOSIZE)
return PAGE_CACHE_SIZE;
return (mp->m_swidth ?
(mp->m_swidth << mp->m_sb.sb_blocklog) :
((mp->m_flags & XFS_MOUNT_DFLT_IOSIZE) ?
(1 << (int)MAX(mp->m_readio_log, mp->m_writeio_log)) :
PAGE_CACHE_SIZE));
}
[XFS] Lazy Superblock Counters When we have a couple of hundred transactions on the fly at once, they all typically modify the on disk superblock in some way. create/unclink/mkdir/rmdir modify inode counts, allocation/freeing modify free block counts. When these counts are modified in a transaction, they must eventually lock the superblock buffer and apply the mods. The buffer then remains locked until the transaction is committed into the incore log buffer. The result of this is that with enough transactions on the fly the incore superblock buffer becomes a bottleneck. The result of contention on the incore superblock buffer is that transaction rates fall - the more pressure that is put on the superblock buffer, the slower things go. The key to removing the contention is to not require the superblock fields in question to be locked. We do that by not marking the superblock dirty in the transaction. IOWs, we modify the incore superblock but do not modify the cached superblock buffer. In short, we do not log superblock modifications to critical fields in the superblock on every transaction. In fact we only do it just before we write the superblock to disk every sync period or just before unmount. This creates an interesting problem - if we don't log or write out the fields in every transaction, then how do the values get recovered after a crash? the answer is simple - we keep enough duplicate, logged information in other structures that we can reconstruct the correct count after log recovery has been performed. It is the AGF and AGI structures that contain the duplicate information; after recovery, we walk every AGI and AGF and sum their individual counters to get the correct value, and we do a transaction into the log to correct them. An optimisation of this is that if we have a clean unmount record, we know the value in the superblock is correct, so we can avoid the summation walk under normal conditions and so mount/recovery times do not change under normal operation. One wrinkle that was discovered during development was that the blocks used in the freespace btrees are never accounted for in the AGF counters. This was once a valid optimisation to make; when the filesystem is full, the free space btrees are empty and consume no space. Hence when it matters, the "accounting" is correct. But that means the when we do the AGF summations, we would not have a correct count and xfs_check would complain. Hence a new counter was added to track the number of blocks used by the free space btrees. This is an *on-disk format change*. As a result of this, lazy superblock counters are a mkfs option and at the moment on linux there is no way to convert an old filesystem. This is possible - xfs_db can be used to twiddle the right bits and then xfs_repair will do the format conversion for you. Similarly, you can convert backwards as well. At some point we'll add functionality to xfs_admin to do the bit twiddling easily.... SGI-PV: 964999 SGI-Modid: xfs-linux-melb:xfs-kern:28652a Signed-off-by: David Chinner <dgc@sgi.com> Signed-off-by: Christoph Hellwig <hch@infradead.org> Signed-off-by: Tim Shimmin <tes@sgi.com>
2007-05-24 12:26:31 +07:00
#define XFS_LAST_UNMOUNT_WAS_CLEAN(mp) \
((mp)->m_flags & XFS_MOUNT_WAS_CLEAN)
#define XFS_FORCED_SHUTDOWN(mp) ((mp)->m_flags & XFS_MOUNT_FS_SHUTDOWN)
void xfs_do_force_shutdown(struct xfs_mount *mp, int flags, char *fname,
int lnnum);
#define xfs_force_shutdown(m,f) \
xfs_do_force_shutdown(m, f, __FILE__, __LINE__)
#define SHUTDOWN_META_IO_ERROR 0x0001 /* write attempt to metadata failed */
#define SHUTDOWN_LOG_IO_ERROR 0x0002 /* write attempt to the log failed */
#define SHUTDOWN_FORCE_UMOUNT 0x0004 /* shutdown from a forced unmount */
#define SHUTDOWN_CORRUPT_INCORE 0x0008 /* corrupt in-memory data structures */
#define SHUTDOWN_REMOTE_REQ 0x0010 /* shutdown came from remote cell */
#define SHUTDOWN_DEVICE_REQ 0x0020 /* failed all paths to the device */
/*
* Flags for xfs_mountfs
*/
#define XFS_MFSI_QUIET 0x40 /* Be silent if mount errors found */
static inline xfs_agnumber_t
xfs_daddr_to_agno(struct xfs_mount *mp, xfs_daddr_t d)
{
xfs_daddr_t ld = XFS_BB_TO_FSBT(mp, d);
do_div(ld, mp->m_sb.sb_agblocks);
return (xfs_agnumber_t) ld;
}
static inline xfs_agblock_t
xfs_daddr_to_agbno(struct xfs_mount *mp, xfs_daddr_t d)
{
xfs_daddr_t ld = XFS_BB_TO_FSBT(mp, d);
return (xfs_agblock_t) do_div(ld, mp->m_sb.sb_agblocks);
}
/*
* Per-ag incore structure, copies of information in agf and agi, to improve the
* performance of allocation group selection.
*/
typedef struct xfs_perag {
struct xfs_mount *pag_mount; /* owner filesystem */
xfs_agnumber_t pag_agno; /* AG this structure belongs to */
atomic_t pag_ref; /* perag reference count */
char pagf_init; /* this agf's entry is initialized */
char pagi_init; /* this agi's entry is initialized */
char pagf_metadata; /* the agf is preferred to be metadata */
char pagi_inodeok; /* The agi is ok for inodes */
__uint8_t pagf_levels[XFS_BTNUM_AGF];
/* # of levels in bno & cnt btree */
__uint32_t pagf_flcount; /* count of blocks in freelist */
xfs_extlen_t pagf_freeblks; /* total free blocks */
xfs_extlen_t pagf_longest; /* longest free space */
__uint32_t pagf_btreeblks; /* # of blocks held in AGF btrees */
xfs_agino_t pagi_freecount; /* number of free inodes */
xfs_agino_t pagi_count; /* number of allocated inodes */
/*
* Inode allocation search lookup optimisation.
* If the pagino matches, the search for new inodes
* doesn't need to search the near ones again straight away
*/
xfs_agino_t pagl_pagino;
xfs_agino_t pagl_leftrec;
xfs_agino_t pagl_rightrec;
spinlock_t pagb_lock; /* lock for pagb_tree */
struct rb_root pagb_tree; /* ordered tree of busy extents */
atomic_t pagf_fstrms; /* # of filestreams active in this AG */
spinlock_t pag_ici_lock; /* incore inode cache lock */
struct radix_tree_root pag_ici_root; /* incore inode cache root */
int pag_ici_reclaimable; /* reclaimable inodes */
struct mutex pag_ici_reclaim_lock; /* serialisation point */
unsigned long pag_ici_reclaim_cursor; /* reclaim restart point */
/* buffer cache index */
spinlock_t pag_buf_lock; /* lock for pag_buf_tree */
struct rb_root pag_buf_tree; /* ordered tree of active buffers */
/* for rcu-safe freeing */
struct rcu_head rcu_head;
int pagb_count; /* pagb slots in use */
} xfs_perag_t;
extern int xfs_log_sbcount(xfs_mount_t *);
extern __uint64_t xfs_default_resblks(xfs_mount_t *mp);
extern int xfs_mountfs(xfs_mount_t *mp);
extern int xfs_initialize_perag(xfs_mount_t *mp, xfs_agnumber_t agcount,
xfs_agnumber_t *maxagi);
extern void xfs_unmountfs(xfs_mount_t *);
extern int xfs_mod_icount(struct xfs_mount *mp, int64_t delta);
extern int xfs_mod_ifree(struct xfs_mount *mp, int64_t delta);
extern int xfs_mod_fdblocks(struct xfs_mount *mp, int64_t delta,
bool reserved);
extern int xfs_mod_frextents(struct xfs_mount *mp, int64_t delta);
xfs: remove bitfield based superblock updates When we log changes to the superblock, we first have to write them to the on-disk buffer, and then log that. Right now we have a complex bitfield based arrangement to only write the modified field to the buffer before we log it. This used to be necessary as a performance optimisation because we logged the superblock buffer in every extent or inode allocation or freeing, and so performance was extremely important. We haven't done this for years, however, ever since the lazy superblock counters pulled the superblock logging out of the transaction commit fast path. Hence we have a bunch of complexity that is not necessary that makes writing the in-core superblock to disk much more complex than it needs to be. We only need to log the superblock now during management operations (e.g. during mount, unmount or quota control operations) so it is not a performance critical path anymore. As such, remove the complex field based logging mechanism and replace it with a simple conversion function similar to what we use for all other on-disk structures. This means we always log the entirity of the superblock, but again because we rarely modify the superblock this is not an issue for log bandwidth or CPU time. Indeed, if we do log the superblock frequently, delayed logging will minimise the impact of this overhead. [Fixed gquota/pquota inode sharing regression noticed by bfoster.] Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Brian Foster <bfoster@redhat.com> Signed-off-by: Dave Chinner <david@fromorbit.com>
2015-01-22 05:10:26 +07:00
extern int xfs_mount_log_sb(xfs_mount_t *);
extern struct xfs_buf *xfs_getsb(xfs_mount_t *, int);
extern int xfs_readsb(xfs_mount_t *, int);
extern void xfs_freesb(xfs_mount_t *);
extern bool xfs_fs_writable(struct xfs_mount *mp, int level);
extern int xfs_sb_validate_fsb_count(struct xfs_sb *, __uint64_t);
extern int xfs_dev_is_read_only(struct xfs_mount *, char *);
xfs: dynamic speculative EOF preallocation Currently the size of the speculative preallocation during delayed allocation is fixed by either the allocsize mount option of a default size. We are seeing a lot of cases where we need to recommend using the allocsize mount option to prevent fragmentation when buffered writes land in the same AG. Rather than using a fixed preallocation size by default (up to 64k), make it dynamic by basing it on the current inode size. That way the EOF preallocation will increase as the file size increases. Hence for streaming writes we are much more likely to get large preallocations exactly when we need it to reduce fragementation. For default settings, the size of the initial extents is determined by the number of parallel writers and the amount of memory in the machine. For 4GB RAM and 4 concurrent 32GB file writes: EXT: FILE-OFFSET BLOCK-RANGE AG AG-OFFSET TOTAL 0: [0..1048575]: 1048672..2097247 0 (1048672..2097247) 1048576 1: [1048576..2097151]: 5242976..6291551 0 (5242976..6291551) 1048576 2: [2097152..4194303]: 12583008..14680159 0 (12583008..14680159) 2097152 3: [4194304..8388607]: 25165920..29360223 0 (25165920..29360223) 4194304 4: [8388608..16777215]: 58720352..67108959 0 (58720352..67108959) 8388608 5: [16777216..33554423]: 117440584..134217791 0 (117440584..134217791) 16777208 6: [33554424..50331511]: 184549056..201326143 0 (184549056..201326143) 16777088 7: [50331512..67108599]: 251657408..268434495 0 (251657408..268434495) 16777088 and for 16 concurrent 16GB file writes: EXT: FILE-OFFSET BLOCK-RANGE AG AG-OFFSET TOTAL 0: [0..262143]: 2490472..2752615 0 (2490472..2752615) 262144 1: [262144..524287]: 6291560..6553703 0 (6291560..6553703) 262144 2: [524288..1048575]: 13631592..14155879 0 (13631592..14155879) 524288 3: [1048576..2097151]: 30408808..31457383 0 (30408808..31457383) 1048576 4: [2097152..4194303]: 52428904..54526055 0 (52428904..54526055) 2097152 5: [4194304..8388607]: 104857704..109052007 0 (104857704..109052007) 4194304 6: [8388608..16777215]: 209715304..218103911 0 (209715304..218103911) 8388608 7: [16777216..33554423]: 452984848..469762055 0 (452984848..469762055) 16777208 Because it is hard to take back specualtive preallocation, cases where there are large slow growing log files on a nearly full filesystem may cause premature ENOSPC. Hence as the filesystem nears full, the maximum dynamic prealloc size іs reduced according to this table (based on 4k block size): freespace max prealloc size >5% full extent (8GB) 4-5% 2GB (8GB >> 2) 3-4% 1GB (8GB >> 3) 2-3% 512MB (8GB >> 4) 1-2% 256MB (8GB >> 5) <1% 128MB (8GB >> 6) This should reduce the amount of space held in speculative preallocation for such cases. The allocsize mount option turns off the dynamic behaviour and fixes the prealloc size to whatever the mount option specifies. i.e. the behaviour is unchanged. Signed-off-by: Dave Chinner <dchinner@redhat.com>
2011-01-04 07:35:03 +07:00
extern void xfs_set_low_space_thresholds(struct xfs_mount *);
#endif /* __XFS_MOUNT_H__ */