06 Oct, 2016
2 commits
-
If the admin doesn't set a CoW extent size or a regular extent size
hint, default to creating CoW reservations 32 blocks long to reduce
fragmentation.Signed-off-by: DarricK J. Wong
Reviewed-by: Christoph Hellwig -
Create a per-inode extent size allocator hint for copy-on-write. This
hint is separate from the existing extent size hint so that CoW can
take advantage of the fragmentation-reducing properties of extent size
hints without disabling delalloc for regular writes.The extent size hint that's fed to the allocator during a copy on
write operation is the greater of the cowextsize and regular extsize
hint.During reflink, if we're sharing the entire source file to the entire
destination file and the destination file doesn't already have a
cowextsize hint, propagate the source file's cowextsize hint to the
destination file.Furthermore, zero the bulkstat buffer prior to setting the fields
so that we don't copy kernel memory contents into userspace.Signed-off-by: Darrick J. Wong
Reviewed-by: Christoph Hellwig
05 Oct, 2016
2 commits
-
Introduce a new in-core fork for storing copy-on-write delalloc
reservations and allocated extents that are in the process of being
written out.Signed-off-by: Darrick J. Wong
Reviewed-by: Christoph Hellwig -
Log recovery will iget an inode to replay BUI items and iput the inode
when it's done. Unfortunately, if the inode was unlinked, the iput
will see that i_nlink == 0 and decide to truncate & free the inode,
which prevents us from replaying subsequent BUIs. We can't skip the
BUIs because we have to replay all the redo items to ensure that
atomic operations complete.Since unlinked inode recovery will reap the inode anyway, we can
safely introduce a new inode flag to indicate that an inode is in this
'unlinked recovery' state and should not be auto-reaped in the
drop_inode path.Signed-off-by: Darrick J. Wong
Reviewed-by: Christoph Hellwig
04 Oct, 2016
1 commit
-
Add new per-AG refcount btree definitions to the per-AG structures.
Signed-off-by: Darrick J. Wong
Signed-off-by: Christoph Hellwig
Reviewed-by: Christoph Hellwig
19 Sep, 2016
1 commit
-
For long growing file writes we will usually already have the
eofblocks tag set when adding more speculative preallocations. Add
a flag in the inode to allow us to skip the the fairly expensive
AG-wide spinlocks and multiple radix tree operations in that case.Signed-off-by: Christoph Hellwig
Reviewed-by: Dave Chinner
Signed-off-by: Dave Chinner
03 Aug, 2016
1 commit
-
Drop the compatibility shims that we were using to integrate the new
deferred operation mechanism into the existing code. No new code.Signed-off-by: Darrick J. Wong
Reviewed-by: Brian Foster
Signed-off-by: Dave Chinner
20 Jul, 2016
2 commits
-
Now that we have the direct I/O kiocb flag there is no real need to sample
the value inside of XFS, and the invis flag was always just partially used
and isn't worth keeping this infrastructure around for. This also splits
the read tracepoint into buffered vs direct as we've done for writes a long
time ago.Signed-off-by: Christoph Hellwig
Reviewed-by: Dave Chinner
Signed-off-by: Dave Chinner
21 Jun, 2016
2 commits
-
We'll want to use this code for large offsets now that we're
skipping holes and unwritten extents efficiently. Also rename it to
xfs_zero_range to be a bit more descriptive, and tell the caller if
we actually did any zeroing.Signed-off-by: Christoph Hellwig
Reviewed-by: Dave Chinner
Signed-off-by: Dave Chinner
01 Jun, 2016
1 commit
-
Al Viro noticed that xfs_lock_inodes should be static, and
that led to ... a few more.These are just the easy ones, others require moving functions
higher in source files, so that's not done here to keep
this review simple.Signed-off-by: Eric Sandeen
Reviewed-by: Christoph Hellwig
Signed-off-by: Dave Chinner
06 Apr, 2016
1 commit
-
In the next patch we'll set up different inode operations for inline vs
out of line symlinks, for that we need to make sure the flags are already
set up properly.[dchinner: added xfs_setup_iops() call to xfs_rename_alloc_whiteout()]
Signed-off-by: Christoph Hellwig
Reviewed-by: Dave Chinner
Signed-off-by: Dave Chinner
07 Mar, 2016
1 commit
09 Feb, 2016
3 commits
-
Move the di_mode value from the xfs_icdinode to the VFS inode, reducing
the xfs_icdinode byte another 2 bytes and collapsing another 2 byte hole
in the structure.Signed-off-by: Dave Chinner
Reviewed-by: Brian Foster
Reviewed-by: Christoph Hellwig
Signed-off-by: Dave Chinner -
The VFS tracks the inode nlink just like the xfs_icdinode. We can
remove the variable from the icdinode and use the VFS inode variable
everywhere, reducing the size of the xfs_icdinode by a further 4
bytes.Signed-off-by: Dave Chinner
Reviewed-by: Brian Foster
Reviewed-by: Christoph Hellwig
Signed-off-by: Dave Chinner -
We currently carry around and log an entire inode core in the
struct xfs_inode. A lot of the information in the inode core is
duplicated in the VFS inode, but we cannot remove this duplication
of infomration because the inode core is logged directly in
xfs_inode_item_format().Add a new function xfs_inode_item_format_core() that copies the
inode core data into a struct xfs_icdinode that is pulled directly
from the log vector buffer. This means we no longer directly
copy the inode core, but copy the structures one member at a time.
This will be slightly less efficient than copying, but will allow us
to remove duplicate and unnecessary items from the struct xfs_inode.To enable us to do this, call the new structure a xfs_log_dinode,
so that we know it's different to the physical xfs_dinode and the
in-core xfs_icdinode.Signed-off-by: Dave Chinner
Reviewed-by: Brian Foster
Reviewed-by: Christoph Hellwig
Signed-off-by: Dave Chinner
08 Feb, 2016
1 commit
-
Factor xfs_seek_hole_data into an unlocked helper which takes
an xfs inode rather than a file for internal use.Also allow specification of "end" - the vfs lseek interface is
defined such that any offset past eof/i_size shall return -ENXIO,
but we will use this for quota code which does not maintain i_size,
and we want to be able to SEEK_DATA past i_size as well. So the
lseek path can send in i_size, and the quota code can determine
its own ending offset.Signed-off-by: Eric Sandeen
Reviewed-by: Dave Chinner
Signed-off-by: Dave Chinner
19 Aug, 2015
1 commit
-
Lockdep annotations are a maintenance nightmare. Locking has to be
modified to suit the limitations of the annotations, and we're
always having to fix the annotations because they are unable to
express the complexity of locking heirarchies correctly.So, next up, we've got more issues with lockdep annotations for
inode locking w.r.t. XFS_LOCK_PARENT:- lockdep classes are exclusive and can't be ORed together
to form new classes.
- IOLOCK needs multiple PARENT subclasses to express the
changes needed for the readdir locking rework needed to
stop the endless flow of lockdep false positives involving
readdir calling filldir under the ILOCK.
- there are only 8 unique lockdep subclasses available,
so we can't create a generic solution.IOWs we need to treat the 3-bit space available to each lock type
differently:- IOLOCK uses xfs_lock_two_inodes(), so needs:
- at least 2 IOLOCK subclasses
- at least 2 IOLOCK_PARENT subclasses
- MMAPLOCK uses xfs_lock_two_inodes(), so needs:
- at least 2 MMAPLOCK subclasses
- ILOCK uses xfs_lock_inodes with up to 5 inodes, so needs:
- at least 5 ILOCK subclasses
- one ILOCK_PARENT subclass
- one RTBITMAP subclass
- one RTSUM subclassFor the IOLOCK, split the space into two sets of subclasses.
For the MMAPLOCK, just use half the space for the one subclass to
match the non-parent lock classes of the IOLOCK.
For the ILOCK, use 0-4 as the ILOCK subclasses, 5-7 for the
remaining individual subclasses.Because they are now all different, modify xfs_lock_inumorder() to
handle the nested subclasses, and to assert fail if passed an
invalid subclass. Further, annotate xfs_lock_inodes() to assert fail
if an invalid combination of lock primitives and inode counts are
passed that would result in a lockdep subclass annotation overflow.Signed-off-by: Dave Chinner
Reviewed-by: Brian Foster
Signed-off-by: Dave Chinner
24 Feb, 2015
2 commits
23 Feb, 2015
3 commits
-
Al Viro noticed a generic set of issues to do with filehandle lookup
racing with dentry cache setup. They involve a filehandle lookup
occurring while an inode is being created and the filehandle lookup
racing with the dentry creation for the real file. This can lead to
multiple dentries for the one path being instantiated. There are a
host of other issues around this same set of paths.The underlying cause is that file handle lookup only waits on inode
cache instantiation rather than full dentry cache instantiation. XFS
is mostly immune to the problems discovered due to it's own internal
inode cache, but there are a couple of corner cases where races can
happen.We currently clear the XFS_INEW flag when the inode is fully set up
after insertion into the cache. Newly allocated inodes are inserted
locked and so aren't usable until the allocation transaction
commits. This, however, occurs before the dentry and security
information is fully initialised and hence the inode is unlocked and
available for lookups to find too early.To solve the problem, only clear the XFS_INEW flag for newly created
inodes once the dentry is fully instantiated. This means lookups
will retry until the XFS_INEW flag is removed from the inode and
hence avoids the race conditions in questions.THis also means that xfs_create(), xfs_create_tmpfile() and
xfs_symlink() need to finish the setup of the inode in their error
paths if we had allocated the inode but failed later in the creation
process. xfs_symlink(), in particular, needed a lot of help to make
it's error handling match that of xfs_create().Signed-off-by: Dave Chinner
Reviewed-by: Brian Foster
Signed-off-by: Dave Chinner -
A new fsync vs power fail test in xfstests indicated that XFS can
have unreliable data consistency when doing extending truncates that
require block zeroing. The blocks beyond EOF get zeroed in memory,
but we never force those changes to disk before we run the
transaction that extends the file size and exposes those blocks to
userspace. This can result in the blocks not being correctly zeroed
after a crash.Because in-memory behaviour is correct, tools like fsx don't pick up
any coherency problems - it's not until the filesystem is shutdown
or the system crashes after writing the truncate transaction to the
journal but before the zeroed data in the page cache is flushed that
the issue is exposed.Fix this by also flushing the dirty data in memory region between
the old size and new size when we've found blocks that need zeroing
in the truncate process.Reported-by: Liu Bo
cc:
Signed-off-by: Dave Chinner
Reviewed-by: Brian Foster
Signed-off-by: Dave Chinner -
Right now we cannot serialise mmap against truncate or hole punch
sanely. ->page_mkwrite is not able to take locks that the read IO
path normally takes (i.e. the inode iolock) because that could
result in lock inversions (read - iolock - page fault - page_mkwrite
- iolock) and so we cannot use an IO path lock to serialise page
write faults against truncate operations.Instead, introduce a new lock that is used *only* in the
->page_mkwrite path that is the equivalent of the iolock. The lock
ordering in a page fault is i_mmaplock -> page lock -> i_ilock,
and so in truncate we can i_iolock -> i_mmaplock and so lock out
new write faults during the process of truncation.Because i_mmap_lock is outside the page lock, we can hold it across
all the same operations we hold the i_iolock for. The only
difference is that we never hold the i_mmaplock in the normal IO
path and so do not ever have the possibility that we can page fault
inside it. Hence there are no recursion issues on the i_mmap_lock
and so we can use it to serialise page fault IO against inode
modification operations that affect the IO path.This patch introduces the i_mmaplock infrastructure, lockdep
annotations and initialisation/destruction code. Use of the new lock
will be in subsequent patches.Signed-off-by: Dave Chinner
Reviewed-by: Brian Foster
Signed-off-by: Dave Chinner
02 Feb, 2015
2 commits
-
This logic is duplicated in xfs_file_fallocate and xfs_ioc_space, and
we'll need another copy of it for pNFS block support.Signed-off-by: Christoph Hellwig
Reviewed-by: Dave Chinner
Signed-off-by: Dave Chinner
24 Dec, 2014
1 commit
-
Adds a new function named xfs_cross_rename(), responsible for
handling requests from sys_renameat2() using RENAME_EXCHANGE flag.Signed-off-by: Carlos Maiolino
Reviewed-by: Brian Foster
Reviewed-by: Dave Chinner
Signed-off-by: Dave Chinner
28 Nov, 2014
1 commit
-
More consolidatation for the on-disk format defintions. Note that the
XFS_IS_REALTIME_INODE moves to xfs_linux.h instead as it is not related
to the on disk format, but depends on a CONFIG_ option.Signed-off-by: Christoph Hellwig
Reviewed-by: Dave Chinner
Signed-off-by: Dave Chinner
02 Oct, 2014
1 commit
-
If we write to the maximum file offset (2^63-2), XFS fails to log the
inode size update when the page is flushed. For example:$ xfs_io -fc "pwrite `echo "2^63-1-1" | bc` 1" /mnt/file
wrote 1/1 bytes at offset 9223372036854775806
1.000000 bytes, 1 ops; 0.0000 sec (22.711 KiB/sec and 23255.8140 ops/sec)
$ stat -c %s /mnt/file
9223372036854775807
$ umount /mnt ; mount /mnt/
$ stat -c %s /mnt/file
0This occurs because XFS calculates the new file size as io_offset +
io_size, I/O occurs in block sized requests, and the maximum supported
file size is not block aligned. Therefore, a write to the max allowable
offset on a 4k blocksize fs results in a write of size 4k to offset
2^63-4096 (e.g., equivalent to round_down(2^63-1, 4096), or IOW the
offset of the block that contains the max file size). The offset plus
size calculation (2^63 - 4096 + 4096 == 2^63) overflows the signed
64-bit variable which goes negative and causes the > comparison to the
on-disk inode size to fail. This returns 0 from xfs_new_eof() and
results in no change to the inode on-disk.Update xfs_new_eof() to explicitly detect overflow of the local
calculation and use the VFS inode size in this scenario. The VFS inode
size is capped to the maximum and thus XFS writes the correct inode size
to disk.Signed-off-by: Brian Foster
Reviewed-by: Dave Chinner
Signed-off-by: Dave Chinner
04 Aug, 2014
1 commit
-
Move the IO flag definitions to xfs_inode.h and kill the header file
as it is now empty.Removing the xfs_vnode.h file showed up an implicit header include
path:
xfs_linux.h -> xfs_vnode.h -> xfs_fs.hAnd so every xfs header file has been inplicitly been including
xfs_fs.h where it is needed or not. Hence the removal of xfs_vnode.h
causes all sorts of build issues because BBTOB() and friends are no
longer automatically included in the build. This also gets fixed.Signed-off-by: Dave Chinner
Reviewed-by: Christoph Hellwig
Signed-off-by: Dave Chinner
20 May, 2014
2 commits
-
Conflicts:
fs/xfs/xfs_inode.c -
mkfs has turned on the XFS_SB_VERSION_NLINKBIT feature bit by
default since November 2007. It's about time we simply made the
kernel code turn it on by default and so always convert v1 inodes to
v2 inodes when reading them in from disk or allocating them. This
This removes needless version checks and modification when bumping
link counts on inodes, and will take code out of a few common code
paths.text data bss dec hex filename
783251 100867 616 884734 d7ffe fs/xfs/xfs.o.orig
782664 100867 616 884147 d7db3 fs/xfs/xfs.o.patchedSigned-off-by: Dave Chinner
Reviewed-by: Christoph Hellwig
Signed-off-by: Dave Chinner
15 May, 2014
1 commit
23 Apr, 2014
1 commit
-
We never test the flag except in xfs_inode_is_filestream, but that
function already tests the on-disk flag or filesystem wide flags,
and is used to decide if we want to set XFS_IFILESTREAM in the
first place.Signed-off-by: Christoph Hellwig
Reviewed-by: Dave Chinner
Signed-off-by: Dave Chinner
17 Apr, 2014
1 commit
-
xfstests generic/004 reproduces an ilock deadlock using the tmpfile
interface when selinux is enabled. This occurs because
xfs_create_tmpfile() takes the ilock and then calls d_tmpfile(). The
latter eventually calls into xfs_xattr_get() which attempts to get the
lock again. E.g.:xfs_io D ffffffff81c134c0 4096 3561 3560 0x00000080
ffff8801176a1a68 0000000000000046 ffff8800b401b540 ffff8801176a1fd8
00000000001d5800 00000000001d5800 ffff8800b401b540 ffff8800b401b540
ffff8800b73a6bd0 fffffffeffffffff ffff8800b73a6bd8 ffff8800b5ddb480
Call Trace:
[] schedule+0x29/0x70
[] rwsem_down_read_failed+0xc5/0x120
[] ? xfs_ilock_attr_map_shared+0x1f/0x50 [xfs]
[] call_rwsem_down_read_failed+0x14/0x30
[] ? down_read_nested+0x89/0xa0
[] ? xfs_ilock+0x122/0x250 [xfs]
[] xfs_ilock+0x122/0x250 [xfs]
[] xfs_ilock_attr_map_shared+0x1f/0x50 [xfs]
[] xfs_attr_get+0x90/0xe0 [xfs]
[] xfs_xattr_get+0x37/0x50 [xfs]
[] generic_getxattr+0x4f/0x70
[] inode_doinit_with_dentry+0x1ae/0x650
[] selinux_d_instantiate+0x1c/0x20
[] security_d_instantiate+0x1b/0x30
[] d_instantiate+0x50/0x70
[] d_tmpfile+0xb5/0xc0
[] xfs_create_tmpfile+0x362/0x410 [xfs]
[] xfs_vn_tmpfile+0x18/0x20 [xfs]
[] path_openat+0x228/0x6a0
[] ? sched_clock+0x9/0x10
[] ? kvm_clock_read+0x27/0x40
[] ? __alloc_fd+0xaf/0x1f0
[] do_filp_open+0x3a/0x90
[] ? _raw_spin_unlock+0x27/0x40
[] ? __alloc_fd+0xaf/0x1f0
[] do_sys_open+0x12e/0x210
[] SyS_open+0x1e/0x20
[] system_call_fastpath+0x16/0x1bxfs_vn_tmpfile() also fails to initialize security on the newly created
inode.Pull the d_tmpfile() call up into xfs_vn_tmpfile() after the transaction
has been committed and the inode unlocked. Also, initialize security on
the inode based on the parent directory provided via the tmpfile call.Signed-off-by: Brian Foster
Reviewed-by: Christoph Hellwig
Signed-off-by: Dave Chinner
13 Mar, 2014
1 commit
-
Conflicts:
fs/xfs/xfs_trans_resv.c
- fix for XFS_INODE_CLUSTER_SIZE macro removal
07 Jan, 2014
2 commits
-
Add two functions xfs_create_tmpfile() and xfs_vn_tmpfile()
to support O_TMPFILE file creation.In contrast to xfs_create(), xfs_create_tmpfile() has a different
log reservation to the regular file creation because there is no
directory modification, and doesn't check if an entry can be added
to the directory, but the reservation quotas is required appropriately,
and finally its inode is added to the unlinked list.xfs_vn_tmpfile() add one O_TMPFILE method to VFS interface and directly
invoke xfs_create_tmpfile().Signed-off-by: Zhi Yong Wu
Reviewed-by: Dave Chinner
Reviewed-by: Christoph Hellwig
Signed-off-by: Ben Myers -
It will be reused by the O_TMPFILE creation function.
Reviewed-by: Christoph Hellwig
Signed-off-by: Zhi Yong Wu
Signed-off-by: Ben Myers
19 Dec, 2013
2 commits
-
Equivalent to xfs_ilock_data_map_shared, except for the attribute fork.
Make xfs_getbmap use it if called for the attribute fork instead of
xfs_ilock_data_map_shared.Signed-off-by: Christoph Hellwig
Reviewed-by: Ben Myers
Signed-off-by: Ben Myers -
Make it clear that we're only locking against the extent map on the data
fork. Also clean the function up a little bit.Signed-off-by: Christoph Hellwig
Reviewed-by: Dave Chinner
Signed-off-by: Ben Myers
