- 18 Nov, 2013 3 commits
-
-
Dave Chinner authored
Michael L Semon reported that generic/069 runtime increased on v5 superblocks by 100% compared to v4 superblocks. his perf-based analysis pointed directly at the timestamp updates being done by the write path in this workload. The append writers are doing 4-byte writes, so there are lots of timestamp updates occurring. The thing is, they aren't being triggered by timestamp changes - they are being triggered by the inode change counter needing to be updated. That is, every write(2) system call needs to bump the inode version count, and it does that through the timestamp update mechanism. Hence for v5 filesystems, test generic/069 is running 3 orders of magnitude more timestmap update transactions on v5 filesystems due to the fact it does a huge number of *4 byte* write(2) calls. This isn't a real world scenario we really need to address - anyone doing such sequential IO should be using fwrite(3), not write(2). i.e. fwrite(3) buffers the writes in userspace to minimise the number of write(2) syscalls, and the problem goes away. However, there is a small change we can make to improve the situation - removing the expensive lock operation on the change counter update. All inode version counter changes in XFS occur under the ip->i_ilock during a transaction, and therefore we don't actually need the spin lock that provides exclusive access to it through inc_inode_iversion(). Hence avoid the lock and just open code the increment ourselves when logging the inode. Reported-by:
Michael L. Semon <mlsemon35@gmail.com> Signed-off-by:
Dave Chinner <dchinner@redhat.com> Reviewed-by:
Christoph Hellwig <hch@lst.de> Signed-off-by:
Ben Myers <bpm@sgi.com>
-
Dave Chinner authored
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:
Eric Sandeen <sandeen@redhat.com> Signed-off-by:
-
Mark Tinguely authored
xfs_trans_ijoin() activates the inode in a transaction and also can specify which lock to free when the transaction is committed or canceled. xfs_bmap_add_attrfork call locks and adds the lock to the transaction but also manually removes the lock. Change the routine to not add the lock to the transaction and manually remove lock on completion. While here, clean up the xfs_trans_cancel flags and goto names. Signed-off-by:
Mark Tinguely <tinguely@sgi.com> Reviewed-by:
-
- 15 Nov, 2013 1 commit
-
-
Ben Myers authored
Add Dave as maintainer of XFS. Signed-off-by:
Ric Wheeler <rwheeler@redhat.com> Reviewed-by:
Dave Chinner <david@fromorbit.com> Reviewed-by:
Alex Elder <elder@kernel.org>
-
- 06 Nov, 2013 3 commits
-
-
Gu Zheng authored
Introduce flag KM_ZERO which is used to alloc zeroed entry, and convert kmem_{zone_}zalloc to call kmem_{zone_}alloc() with KM_ZERO directly, in order to avoid the setting to zero step. And following Dave's suggestion, make kmem_{zone_}zalloc static inline into kmem.h as they're now just a simple wrapper. V2: Make kmem_{zone_}zalloc static inline into kmem.h as Dave suggested. Signed-off-by:Gu Zheng <guz.fnst@cn.fujitsu.com> Reviewed-by:
-
Dave Chinner authored
To help track down AGI/AGF lock ordering issues, I added these tracepoints to tell us when an AGI or AGF is read and locked. With these we can now determine if the lock ordering goes wrong from tracing captures. Signed-off-by:
-
Dave Chinner authored
I debugging a log tail issue on a RHEL6 kernel, I added these trace points to trace log items being added, moved and removed in the AIL and how that affected the log tail LSN that was written to the log. They were very helpful in that they immediately identified the cause of the problem being seen. Hence I'd like to always have them available for use. Signed-off-by:
-
- 04 Nov, 2013 1 commit
-
-
Dave Chinner authored
Removing an inode from the namespace involves removing the directory entry and dropping the link count on the inode. Removing the directory entry can result in locking an AGF (directory blocks were freed) and removing a link count can result in placing the inode on an unlinked list which results in locking an AGI. The big problem here is that we have an ordering constraint on AGF and AGI locking - inode allocation locks the AGI, then can allocate a new extent for new inodes, locking the AGF after the AGI. Similarly, freeing the inode removes the inode from the unlinked list, requiring that we lock the AGI first, and then freeing the inode can result in an inode chunk being freed and hence freeing disk space requiring that we lock an AGF. Hence the ordering that is imposed by other parts of the code is AGI before AGF. This means we cannot remove the directory entry before we drop the inode reference count and put it on the unlinked list as this results in a lock order of AGF then AGI, and this can deadlock against inode allocation and freeing. Therefore we must drop the link counts before we remove the directory entry. This is still safe from a transactional point of view - it is not until we get to xfs_bmap_finish() that we have the possibility of multiple transactions in this operation. Hence as long as we remove the directory entry and drop the link count in the first transaction of the remove operation, there are no transactional constraints on the ordering here. Change the ordering of the operations in the xfs_remove() function to align the ordering of AGI and AGF locking to match that of the rest of the code. Signed-off-by:
-
- 31 Oct, 2013 1 commit
-
-
Jie Liu authored
At xfs_iext_add(), if extent(s) are being appended to the last page in the indirection array and the new extent(s) don't fit in the page, the number of extents(erp->er_extcount) in a new allocated entry should be the minimum value between count and XFS_LINEAR_EXTS, instead of count. For now, there is no existing test case can demonstrates a problem with the er_extcount being set incorrectly here, but it obviously like a bug. Signed-off-by:
Jie Liu <jeff.liu@oracle.com> Reviewed-by:
-
- 30 Oct, 2013 16 commits
-
-
Eric Sandeen authored
Today, if xfs_sb_read_verify encounters a v4 superblock with junk past v4 fields which includes data in sb_crc, it will be treated as a failing checksum and a significant corruption. There are known prior bugs which leave junk at the end of the V4 superblock; we don't need to actually fail the verification in this case if other checks pan out ok. So if this is a secondary superblock, and the primary superblock doesn't indicate that this is a V5 filesystem, don't treat this as an actual checksum failure. We should probably check the garbage condition as we do in xfs_repair, and possibly warn about it or self-heal, but that's a different scope of work. Stable folks: This can go back to v3.10, which is what introduced the sb CRC checking that is tripped up by old, stale, incorrect V4 superblocks w/ unzeroed bits. Cc: stable@vger.kernel.org Signed-off-by:
-
Geyslan G. Bem authored
In xlog_verify_iclog a debug check of the incore log buffers prints an error if icptr is null and then goes on to dereference the pointer regardless. Convert this to an assert so that the intention is clear. This was reported by Coverty. Signed-off-by:
-
Denis Efremov authored
ASSERT on args takes place after args dereference. This assertion is redundant since we are going to panic anyway. Found by Linux Driver Verification project (linuxtesting.org) - PVS-Studio analyzer. Signed-off-by:
Denis Efremov <yefremov.denis@gmail.com> Reviewed-by:
-
Dave Chinner authored
Page cache allocation doesn't always go through ->begin_write and hence we don't always get the opportunity to set the allocation context to GFP_NOFS. Failing to do this means we open up the direct relcaim stack to recurse into the filesystem and consume a significant amount of stack. On RHEL6.4 kernels we are seeing ra_submit() and generic_file_splice_read() from an nfsd context recursing into the filesystem via the inode cache shrinker and evicting inodes. This is causing truncation to be run (e.g EOF block freeing) and causing bmap btree block merges and free space btree block splits to occur. These btree manipulations are occurring with the call chain already 30 functions deep and hence there is not enough stack space to complete such operations. To avoid these specific overruns, we need to prevent the page cache allocation from recursing via direct reclaim. We can do that because the allocation functions take the allocation context from that which is stored in the mapping for the inode. We don't set that right now, so the default is GFP_HIGHUSER_MOVABLE, which is effectively a GFP_KERNEL context. We need it to be the equivalent of GFP_NOFS, so when we initialise an inode, set the mapping gfp mask appropriately. This makes the use of AOP_FLAG_NOFS redundant from other parts of the XFS IO path, so get rid of it. Signed-off-by:
-
Dave Chinner authored
The kbuild test robot indicated that there were some new sparse warnings in fs/xfs/xfs_dquot_buf.c. Actually, there were a lot more that is wasn't warning about, so fix them all up. Reported-by: kbuild test robot Signed-off-by:
-
Dave Chinner authored
The directory block format verifier fails to check that the leaf entry count is in a valid range, and so if it is corrupted then it can lead to derefencing a pointer outside the block buffer. While we can't exactly validate the count without first walking the directory block, we can ensure the count lands in the valid area within the directory block and hence avoid out-of-block references. Signed-off-by:
-
Dave Chinner authored
Rather than hiding the ftype field size accounting inside the dirent padding for the ".." and first entry offset functions for v2 directory formats, add explicit functions that calculate it correctly. Signed-off-by:
-
Dave Chinner authored
Many of the vectorised function calls now take no parameters and return a constant value. There is no reason for these to be vectored functions, so convert them to constants Binary sizes: 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 792350 96802 1096 890248 d9588 fs/xfs/xfs.o.p2 789293 96802 1096 887191 d8997 fs/xfs/xfs.o.p3 789005 96802 1096 886903 d8997 fs/xfs/xfs.o.p4 789061 96802 1096 886959 d88af fs/xfs/xfs.o.p5 789733 96802 1096 887631 d8b4f fs/xfs/xfs.o.p6 791421 96802 1096 889319 d91e7 fs/xfs/xfs.o.p7 791701 96802 1096 889599 d92ff fs/xfs/xfs.o.p8 791205 96802 1096 889103 d91cf fs/xfs/xfs.o.p9 Signed-off-by:
-
Dave Chinner authored
Next step in the vectorisation process is the directory free block encode/decode operations. There are relatively few of these, though there are quite a number of calls to them. Binary sizes: 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 792350 96802 1096 890248 d9588 fs/xfs/xfs.o.p2 789293 96802 1096 887191 d8997 fs/xfs/xfs.o.p3 789005 96802 1096 886903 d8997 fs/xfs/xfs.o.p4 789061 96802 1096 886959 d88af fs/xfs/xfs.o.p5 789733 96802 1096 887631 d8b4f fs/xfs/xfs.o.p6 791421 96802 1096 889319 d91e7 fs/xfs/xfs.o.p7 791701 96802 1096 889599 d92ff fs/xfs/xfs.o.p8 Signed-off-by:
-
Dave Chinner authored
Conversion from on-disk structures to in-core header structures currently relies on magic number checks. If the magic number is wrong, but one of the supported values, we do the wrong thing with the encode/decode operation. Split these functions so that there are discrete operations for the specific directory format we are handling. In doing this, move all the header encode/decode functions to xfs_da_format.c as they are directly manipulating the on-disk format. It should be noted that all the growth in binary size is from xfs_da_format.c - the rest of the code actaully shrinks. 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 792350 96802 1096 890248 d9588 fs/xfs/xfs.o.p2 789293 96802 1096 887191 d8997 fs/xfs/xfs.o.p3 789005 96802 1096 886903 d8997 fs/xfs/xfs.o.p4 789061 96802 1096 886959 d88af fs/xfs/xfs.o.p5 789733 96802 1096 887631 d8b4f fs/xfs/xfs.o.p6 791421 96802 1096 889319 d91e7 fs/xfs/xfs.o.p7 Signed-off-by:
-
Dave Chinner authored
The remaining non-vectorised code for the directory structure is the node format blocks. This is shared with the attribute tree, and so is slightly more complex to vectorise. Introduce a "non-directory" directory ops structure that is attached to all non-directory inodes so that attribute operations can be vectorised for all inodes. Once we do this, we can vectorise all the da btree operations. Because this patch adds more infrastructure than it removes the binary size does not decrease: 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 792350 96802 1096 890248 d9588 fs/xfs/xfs.o.p2 789293 96802 1096 887191 d8997 fs/xfs/xfs.o.p3 789005 96802 1096 886903 d8997 fs/xfs/xfs.o.p4 789061 96802 1096 886959 d88af fs/xfs/xfs.o.p5 789733 96802 1096 887631 d8b4f fs/xfs/xfs.o.p6 Signed-off-by:
-
Dave Chinner authored
Next step in the vectorisation process is the leaf block encode/decode operations. Most of the operations on leaves are handled by the data block vectors, so there are relatively few of them here. Because of all the shuffling of code and having to pass more state to some functions, this patch doesn't directly reduce the size of the binary. It does open up many more opportunities for factoring and optimisation, however. 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 792350 96802 1096 890248 d9588 fs/xfs/xfs.o.p2 789293 96802 1096 887191 d8997 fs/xfs/xfs.o.p3 789005 96802 1096 886903 d8997 fs/xfs/xfs.o.p4 789061 96802 1096 886959 d88af fs/xfs/xfs.o.p5 Signed-off-by:
-
Dave Chinner authored
Convert the rest of the directory data block encode/decode operations to vector format. This further reduces the size of the built binary: 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 792350 96802 1096 890248 d9588 fs/xfs/xfs.o.p2 789293 96802 1096 887191 d8997 fs/xfs/xfs.o.p3 789005 96802 1096 886903 d8997 fs/xfs/xfs.o.p4 Signed-off-by:
-
Dave Chinner authored
Following from the initial patches to vectorise the shortform directory encode/decode operations, convert half the data block operations to use the vector. The rest will be done in a second patch. This further reduces the size of the built binary: 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 792350 96802 1096 890248 d9588 fs/xfs/xfs.o.p2 789293 96802 1096 887191 d8997 fs/xfs/xfs.o.p3 Signed-off-by:
-
Dave Chinner authored
Following from the initial patch to introduce the directory operations vector, convert the rest of the shortform directory operations to use vectored ops rather than superblock feature checks. This further reduces the size of the built binary: 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 792350 96802 1096 890248 d9588 fs/xfs/xfs.o.p2 Signed-off-by:
-
Dave Chinner authored
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:
-
- 23 Oct, 2013 7 commits
-
-
Dave Chinner authored
xfs_rtalloc.c is partially shared with userspace. Split the file up into two parts - one that is kernel private and the other which is wholly shared with userspace. Signed-off-by:
-
Dave Chinner authored
Currently the xfs_inode.h header has a dependency on the definition of the BMAP btree records as the inode fork includes an array of xfs_bmbt_rec_host_t objects in it's definition. Move all the btree format definitions from xfs_btree.h, xfs_bmap_btree.h, xfs_alloc_btree.h and xfs_ialloc_btree.h to xfs_format.h to continue the process of centralising the on-disk format definitions. With this done, the xfs inode definitions are no longer dependent on btree header files. The enables a massive culling of unnecessary includes, with close to 200 #include directives removed from the XFS kernel code base. Signed-off-by:
-
Dave Chinner authored
xfs_trans.h has a dependency on xfs_log.h for a couple of structures. Most code that does transactions doesn't need to know anything about the log, but this dependency means that they have to include xfs_log.h. Decouple the xfs_trans.h and xfs_log.h header files and clean up the includes to be in dependency order. In doing this, remove the direct include of xfs_trans_reserve.h from xfs_trans.h so that we remove the dependency between xfs_trans.h and xfs_mount.h. Hence the xfs_trans.h include can be moved to the indicate the actual dependencies other header files have on it. Note that these are kernel only header files, so this does not translate to any userspace changes at all. Signed-off-by:
-
Dave Chinner authored
We don't do callbacks at transaction commit time, no do we have any infrastructure to set up or run such callbacks, so remove the variables and typedefs for these operations. If we ever need to add callbacks, we can reintroduce the variables at that time. Signed-off-by:
-
Dave Chinner authored
Parts of userspace want to be able to read and modify dquot buffers (e.g. xfs_db) so we need to split out the reading and writing of these buffers so it is easy to shared code with libxfs in userspace. Signed-off-by:
-
Dave Chinner authored
The on-disk format definitions for the directory and attribute structures are spread across 3 header files right now, only one of which is dedicated to defining on-disk structures and their manipulation (xfs_dir2_format.h). Pull all the format definitions into a single header file - xfs_da_format.h - and switch all the code over to point at that. Signed-off-by:
-
Dave Chinner authored
All of the buffer operations structures are needed to be exported for xfs_db, so move them all to a common location rather than spreading them all over the place. They are verifying the on-disk format, so while xfs_format.h might be a good place, it is not part of the on disk format. Hence we need to create a new header file that we centralise these related definitions. Start by moving the bffer operations structures, and then also move all the other definitions that have crept into xfs_log_format.h and xfs_format.h as there was no other shared header file to put them in. Signed-off-by:
-
- 21 Oct, 2013 5 commits
-
-
Christoph Hellwig authored
Now that only one caller of xfs_change_file_space is left it can be merged into said caller. Signed-off-by:
-
Christoph Hellwig authored
Call xfs_alloc_file_space or xfs_free_file_space directly from xfs_file_fallocate instead of going through xfs_change_file_space. This simplified the code by removing the unessecary marshalling of the arguments into an xfs_flock64_t structure and allows removing checks that are already done in the VFS code. Signed-off-by:
-
Christoph Hellwig authored
Currently fallocate always holds the iolock when calling into xfs_change_file_space, while the ioctl path lets some of the lower level functions take it, but leave it out in others. This patch makes sure the ioctl path also always holds the iolock and thus introduces consistent locking for the preallocation operations while simplifying the code and allowing to kill the now unused XFS_ATTR_NOLOCK flag. Signed-off-by:
-
Christoph Hellwig authored
Signed-off-by:
-
Christoph Hellwig authored
There is no reason to conditionally take the iolock inside xfs_setattr_size when we can let the caller handle it unconditionally, which just incrases the lock hold time for the case where it was previously taken internally by a few instructions. Signed-off-by:
-
- 17 Oct, 2013 3 commits
-
-
Eric Sandeen authored
When xfs_growfs_data_private() is updating backup superblocks, it bails out on the first error encountered, whether reading or writing: * If we get an error writing out the alternate superblocks, * just issue a warning and continue. The real work is * already done and committed. This can cause a problem later during repair, because repair looks at all superblocks, and picks the most prevalent one as correct. If we bail out early in the backup superblock loop, we can end up with more "bad" matching superblocks than good, and a post-growfs repair may revert the filesystem to the old geometry. With the combination of superblock verifiers and old bugs, we're more likely to encounter read errors due to verification. And perhaps even worse, we don't even properly write any of the newly-added superblocks in the new AGs. Even with this change, growfs will still say: xfs_growfs: XFS_IOC_FSGROWFSDATA xfsctl failed: Structure needs cleaning data blocks changed from 319815680 to 335216640 which might be confusing to the user, but it at least communicates that something has gone wrong, and dmesg will probably highlight the need for an xfs_repair. And this is still best-effort; if verifiers fail on more than half the backup supers, they may still "win" - but that's probably best left to repair to more gracefully handle by doing its own strict verification as part of the backup super "voting." Signed-off-by:
-
Eric Sandeen authored
If we get EWRONGFS due to probing of non-xfs filesystems, there's no need to issue the scary corruption error and backtrace. Signed-off-by:
-
Eric Sandeen authored
__xfs_printk adds its own "\n". Having it in the original string leads to unintentional blank lines from these messages. Most format strings have no newline, but a few do, leading to i.e.: [ 7347.119911] XFS (sdb2): Access to block zero in inode 132 start_block: 0 start_off: 0 blkcnt: 0 extent-state: 0 lastx: 1a05 [ 7347.119911] [ 7347.119919] XFS (sdb2): Access to block zero in inode 132 start_block: 0 start_off: 0 blkcnt: 0 extent-state: 0 lastx: 1a05 [ 7347.119919] Fix them all. Signed-off-by:
-
