IF YOU WOULD LIKE TO GET AN ACCOUNT, please write an
email to Administrator. User accounts are meant only to access repo
and report issues and/or generate pull requests.
This is a purpose-specific Git hosting for
BaseALT
projects. Thank you for your understanding!
Только зарегистрированные пользователи имеют доступ к сервису!
Для получения аккаунта, обратитесь к администратору.
With the code as it stands today, b_retries never increments because
it gets reset to 0 in the error callback.
Remove that, and fix a similar problem where the first retry time
was constantly being overwritten, which defeated the timeout tunable
as well. We now only set first retry time if a non-zero timeout is
set, to match the behavior of only incrementing retries if a retry
value is set.
This way max retries & timeouts consistently take effect after a
tunable is set, rather than acting retroactively on a buffer which
has failed at some point in the past and has accumulated state from
those prior failures.
Thanks to dchinner for talking through this with me.
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Reviewed-by: Carlos Maiolino <cmaiolino@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Fix up a couple places where extra flag manipulation occurs.
In the first case we clear XBF_ASYNC and then immediately reset it -
so don't bother clearing in the first place.
In the 2nd case we are at a point in the function where the buffer
must already be async, so there is no need to reset it.
Add consistent spacing around the " | " while we're at it.
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Reviewed-by: Carlos Maiolino <cmaiolino@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
xfs_error_get_cfg() is called with bp->b_error as an arg, which is
negative, so the switch statement won't ever find any matches.
This results in only the default error handler having any effect, as
EIO/ENOSPC/ENODEV get ignored due to the wrong sign.
It seems simplest to always flip the error sign to positive, so that
we can handle either negative errors in bp->b_error, or possibly a
positive errno via something like xfs_error_get_cfg(EIO) - this
future-proofs the function.
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Reviewed-by: Carlos Maiolino <cmaiolino@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
replace the magic numbers by offsetof(...) and sizeof(...), and add two
extra checks on xfs_check_ondisk_structs()
[dchinner: renamed header structures to be more descriptive]
Signed-off-by: Hou Tao <houtao1@huawei.com>
Reviewed-by: Carlos Maiolino <cmaiolino@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
The indentation in this function is different from the other functions.
Those spacebars are converted to tabs to improve readability.
Signed-off-by: Kaho Ng <ngkaho1234@gmail.com>
Reviewed-by: Carlos Maiolino <cmaiolino@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Errors go from zero which means no error to XFS_ERRTAG_MAX (22). My
static checker complains that xfs_errortag_add() puts an upper bound on
this but not a lower bound. Let's fix it by making it unsigned.
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
When calling fdget() in xfs_ioc_swapext(), we need to verify that
the file descriptors passed into the ioctl point to XFS inodes
before we start operations on them. If we don't do this, we could be
referencing arbitrary kernel memory as an XFS inode. THis could lead
to memory corruption and/or performing locking operations on
attacker-chosen structures in kernel memory.
[dchinner: rewrite commit message ]
[dchinner: add comment explaining new check ]
Signed-off-by: Jann Horn <jann@thejh.net>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Create a common function to calculate the maximum height of a per-AG
btree. This will eventually be used by the rmapbt and refcountbt
code to calculate appropriate maxlevels values for each. This is
important because the verifiers and the transaction block
reservations depend on accurate estimates of how many blocks are
needed to satisfy a btree split.
We were mistakenly using the max bnobt height for all the btrees,
which creates a dangerous situation since the larger records and
keys in an rmapbt make it very possible that the rmapbt will be
taller than the bnobt and so we can run out of transaction block
reservation.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
In struct xfs_bmap_free, convert the open-coded free extent list to
a regular list, then use list_sort to sort it prior to processing.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Break up xfs_free_extent() into a helper that fixes the freelist.
This helper will be used subsequently to ensure the freelist during
deferred rmap processing.
[darrick: refactor to put this at the head of the patchset]
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
This is already in xfsprogs' libxfs, so port it to the kernel.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Currently we don't check the error_tag when someone's trying to set up
error injection testing. If userspace passes in a value we don't know
about, send back an error. This will help xfstests to _notrun a test
that uses error injection to test things like log replay.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Create a second buf_trylock tracepoint so that we can distinguish
between a successful and a failed trylock. With this piece, we can
use a script to look at the ftrace output to detect buffer deadlocks.
[dchinner: update to if/else as per hch's suggestion]
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Some of the directory/attr structures contain variable-length objects,
so the enclosing structure doesn't have a meaningful fixed size at
compile time. We can check the offsets of the members before the
variable-length member, so do those.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
xfs_reserve_blocks() is responsible to update the XFS reserved block
pool count at mount time or based on user request. When the caller
requests to increase the reserve pool, blocks must be allocated from
the global counters such that they are no longer available for
general purpose use. If the requested reserve pool size is too
large, XFS reserves what blocks are available. The implementation
requires looking at the percpu counters and making an educated guess
as to how many blocks to try and allocate from xfs_mod_fdblocks(),
which can return -ENOSPC if the guess was not accurate due to
counters being modified in parallel.
xfs_reserve_blocks() retries the guess in this scenario until the
allocation succeeds or it is determined that there is no space
available in the fs. While not easily reproducible in the current
form, the retry code doesn't actually work correctly if
xfs_mod_fdblocks() actually fails. The problem is that the percpu
calculations use the m_resblks counter to determine how many blocks
to allocate, but unconditionally update m_resblks before the block
allocation has actually succeeded. Therefore, if xfs_mod_fdblocks()
fails, the code jumps to the retry label and uses the already
updated m_resblks value to determine how many blocks to try and
allocate. If the percpu counters previously suggested that the
entire request was available, fdblocks_delta could end up set to 0.
In that case, m_resblks is updated to the requested value, yet no
blocks have been reserved at all.
Refactor xfs_reserve_blocks() to use an explicit loop and make the
code easier to follow. Since we have to drop the spinlock across the
xfs_mod_fdblocks() call, use a delta value for m_resblks as well and
only apply the delta once allocation succeeds.
[dchinner: convert to do {} while() loop]
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
The filesystem quiesce sequence performs the operations necessary to
drain all background work, push pending transactions through the log
infrastructure and wait on I/O resulting from the final AIL push. We
have had reports of remount,ro hangs in xfs_log_quiesce() ->
xfs_wait_buftarg(), however, and some instrumentation code to detect
transaction commits at this point in the quiesce sequence has inculpated
the eofblocks background scanner as a cause.
While higher level remount code generally prevents user modifications by
the time the filesystem has made it to xfs_log_quiesce(), the background
scanner may still be alive and can perform pending work at any time. If
this occurs between the xfs_log_force() and xfs_wait_buftarg() calls
within xfs_log_quiesce(), this can lead to an indefinite lockup in
xfs_wait_buftarg().
To prevent this problem, cancel the background eofblocks scan worker
during the remount read-only quiesce sequence. This suspends background
trimming when a filesystem is remounted read-only. This is only done in
the remount path because the freeze codepath has already locked out new
transactions by the time the filesystem attempts to quiesce (and thus
waiting on an active work item could deadlock). Kick the eofblocks
worker to pick up where it left off once an fs is remounted back to
read-write.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Instead punch the whole first, and the use the our zeroing helper
to punch out the edge blocks.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
We now skip holes in it, so no need to have the caller do it as well.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
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 <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Note that this removes support for the untested FIEMAP_FLAG_XATTR. It
could be added relatively easily with iomap ops for the attr fork, but
without test coverage I don't feel safe doing this.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Convert XFS to use the new iomap based multipage write path. This involves
implementing the ->iomap_begin and ->iomap_end methods, and switching the
buffered file write, page_mkwrite and xfs_iozero paths to the new iomap
helpers.
With this change __xfs_get_blocks will never be used for buffered writes,
and the code handling them can be removed.
Based on earlier code from Dave Chinner.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Currently zeroing out blocks and waiting for writeout is a bit of a mess in
truncate. This patch gives it a clear order in preparation for the iomap
path:
(1) we first wait for any direct I/O to complete to prevent any races
for it
(2) we then perform the actual zeroing, and only use the truncate_page
helpers for truncating down. The truncate up case already is
handled by the separate call to xfs_zero_eof.
(3) only then we write back dirty data, as zeroing block may cause
dirty pages when using either xfs_zero_eof or the new iomap
infrastructure.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
And ensure it works for RT subvolume files an set the block device,
both of which will be needed to be able to use the function in the
buffered write path.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Add a simple fiemap implementation based on iomap_ops, partially based
on a previous implementation from Bob Peterson <rpeterso@redhat.com>.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
This avoid needing a separate inefficient get_block based DAX zero_range
implementation in file systems.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Add infrastructure for multipage buffered writes. This is implemented
using an main iterator that applies an actor function to a range that
can be written.
This infrastucture is used to implement a buffered write helper, one
to zero file ranges and one to implement the ->page_mkwrite VM
operations. All of them borrow a fair amount of code from fs/buffers.
for now by using an internal version of __block_write_begin that
gets passed an iomap and builds the corresponding buffer head.
The file system is gets a set of paired ->iomap_begin and ->iomap_end
calls which allow it to map/reserve a range and get a notification
once the write code is finished with it.
Based on earlier code from Dave Chinner.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
When we have a lot of metadata to flush from the AIL, the buffer
list can get very long. The current submission code tries to batch
submission to optimise IO order of the metadata (i.e. ascending
block order) to maximise block layer merging or IO to adjacent
metadata blocks.
Unfortunately, the method used can result in long lock times
occurring as buffers locked early on in the buffer list might not be
dispatched until the end of the IO licst processing. This is because
sorting does not occur util after the buffer list has been processed
and the buffers that are going to be submitted are locked. Hence
when the buffer list is several thousand buffers long, the lock hold
times before IO dispatch can be significant.
To fix this, sort the buffer list before we start trying to lock and
submit buffers. This means we can now submit buffers immediately
after they are locked, allowing merging to occur immediately on the
plug and dispatch to occur as quickly as possible. This means there
is minimal delay between locking the buffer and IO submission
occuring, hence reducing the worst case lock hold times seen during
delayed write buffer IO submission signficantly.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Carlos Maiolino <cmaiolino@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
And the same for XFS_IOC_THAW. Just because we now have a common
version of the ioctl we still need to provide the old name for it
for anyone using those.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
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 <sandeen@sandeen.net>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
The static checker reports that after commit 8d99fe92fed0 ("xfs: fix
efi/efd error handling to avoid fs shutdown hangs"), the code has been
reworked such that error == -EFSCORRUPTED is not possible in this
codepath.
Remove the spurious error check and just use SHUTDOWN_META_IO_ERROR
unconditionally.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Carlos Maiolino <cmaiolino@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Multi-block buffers are logged based on buffer offset in
xfs_trans_log_buf(). xfs_buf_item_log() ultimately walks each mapping in
the buffer and marks the associated range to be logged in the
xfs_buf_log_format bitmap for that mapping. This code is broken,
however, in that it marks the actual buffer offsets of the associated
range in each bitmap rather than shifting to the byte range for that
particular mapping.
For example, on a 4k fsb fs, buffer offset 4096 refers to the first byte
of the second mapping in the buffer. This means byte 0 of the second log
format bitmap should be tagged as dirty. Instead, the current code marks
byte offset 4096 of the second log format bitmap, which is invalid and
potentially out of range of the mapping.
As a result of this, the log item format code invoked at transaction
commit time is not be able to correctly identify what parts of the
buffer to copy into log vectors. This can lead to NULL log vector
pointer dereferences in CIL push context if the item format code was not
able to locate any dirty ranges at all. This crash has been reproduced
on a 4k FSB filesystem using 16k directory blocks where an unlink
operation happened not to log anything in the first block of the
mapping. The logged offsets were all over 4k, marked as such in the
subsequent log format mappings, and thus left the transaction with an
xfs_log_item that is marked DIRTY but without any logged regions.
Further, even when the logged regions are marked correctly in the buffer
log format bitmaps, the format code doesn't copy the correct ranges of
the buffer into the log. This means that any logged region beyond the
first block of a multi-block buffer is subject to corruption after a
crash and log recovery sequence. This is due to a failure to convert the
mapping bm_len field from basic blocks to bytes in the buffer offset
tracking code in xfs_buf_item_format().
Update xfs_buf_item_log() to convert buffer offsets to segment relative
offsets when logging multi-block buffers. This ensures that the modified
regions of a buffer are logged correctly and avoids the aforementioned
crash. Also update xfs_buf_item_format() to correctly track the source
offset into the buffer for the log vector formatting code. This ensures
that the correct data is copied into the log.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Eric Sandeen <sandeen@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
The self-test was updated to cover zero-length strings; the function
needs to be updated, too.
Reported-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: George Spelvin <linux@sciencehorizons.net>
Fixes: fcfd2fbf22d2 ("fs/namei.c: Add hashlen_string() function")
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The original name was simply hash_string(), but that conflicted with a
function with that name in drivers/base/power/trace.c, and I decided
that calling it "hashlen_" was better anyway.
But you have to do it in two places.
[ This caused build errors for architectures that don't define
CONFIG_DCACHE_WORD_ACCESS - Linus ]
Signed-off-by: George Spelvin <linux@sciencehorizons.net>
Reported-by: Guenter Roeck <linux@roeck-us.net>
Fixes: fcfd2fbf22d2 ("fs/namei.c: Add hashlen_string() function")
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The HPFS filesystem used generic_show_options to produce string that is
displayed in /proc/mounts. However, there is a problem that the options
may disappear after remount. If we mount the filesystem with option1
and then remount it with option2, /proc/mounts should show both option1
and option2, however it only shows option2 because the whole option
string is replaced with replace_mount_options in hpfs_remount_fs.
To fix this bug, implement the hpfs_show_options function that prints
options that are currently selected.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Commit c8f33d0bec99 ("affs: kstrdup() memory handling") checks if the
kstrdup function returns NULL due to out-of-memory condition.
However, if we are remounting a filesystem with no change to
filesystem-specific options, the parameter data is NULL. In this case,
kstrdup returns NULL (because it was passed NULL parameter), although no
out of memory condition exists. The mount syscall then fails with
ENOMEM.
This patch fixes the bug. We fail with ENOMEM only if data is non-NULL.
The patch also changes the call to replace_mount_options - if we didn't
pass any filesystem-specific options, we don't call
replace_mount_options (thus we don't erase existing reported options).
Fixes: c8f33d0bec99 ("affs: kstrdup() memory handling")
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Cc: stable@vger.kernel.org # v4.1+
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Commit ce657611baf9 ("hpfs: kstrdup() out of memory handling") checks if
the kstrdup function returns NULL due to out-of-memory condition.
However, if we are remounting a filesystem with no change to
filesystem-specific options, the parameter data is NULL. In this case,
kstrdup returns NULL (because it was passed NULL parameter), although no
out of memory condition exists. The mount syscall then fails with
ENOMEM.
This patch fixes the bug. We fail with ENOMEM only if data is non-NULL.
The patch also changes the call to replace_mount_options - if we didn't
pass any filesystem-specific options, we don't call
replace_mount_options (thus we don't erase existing reported options).
Fixes: ce657611baf9 ("hpfs: kstrdup() out of memory handling")
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Various builds (such as i386:allmodconfig) fail with
fs/binfmt_aout.c:133:2: error: expected identifier or '(' before 'return'
fs/binfmt_aout.c:134:1: error: expected identifier or '(' before '}' token
[ Oops. My bad, I had stupidly thought that "allmodconfig" covered this
on x86-64 too, but it obviously doesn't. Egg on my face. - Linus ]
Fixes: 5d22fc25d4fc ("mm: remove more IS_ERR_VALUE abuses")
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Pull string hash improvements from George Spelvin:
"This series does several related things:
- Makes the dcache hash (fs/namei.c) useful for general kernel use.
(Thanks to Bruce for noticing the zero-length corner case)
- Converts the string hashes in <linux/sunrpc/svcauth.h> to use the
above.
- Avoids 64-bit multiplies in hash_64() on 32-bit platforms. Two
32-bit multiplies will do well enough.
- Rids the world of the bad hash multipliers in hash_32.
This finishes the job started in commit 689de1d6ca95 ("Minimal
fix-up of bad hashing behavior of hash_64()")
The vast majority of Linux architectures have hardware support for
32x32-bit multiply and so derive no benefit from "simplified"
multipliers.
The few processors that do not (68000, h8/300 and some models of
Microblaze) have arch-specific implementations added. Those
patches are last in the series.
- Overhauls the dcache hash mixing.
The patch in commit 0fed3ac866ea ("namei: Improve hash mixing if
CONFIG_DCACHE_WORD_ACCESS") was an off-the-cuff suggestion.
Replaced with a much more careful design that's simultaneously
faster and better. (My own invention, as there was noting suitable
in the literature I could find. Comments welcome!)
- Modify the hash_name() loop to skip the initial HASH_MIX(). This
would let us salt the hash if we ever wanted to.
- Sort out partial_name_hash().
The hash function is declared as using a long state, even though
it's truncated to 32 bits at the end and the extra internal state
contributes nothing to the result. And some callers do odd things:
- fs/hfs/string.c only allocates 32 bits of state
- fs/hfsplus/unicode.c uses it to hash 16-bit unicode symbols not bytes
- Modify bytemask_from_count to handle inputs of 1..sizeof(long)
rather than 0..sizeof(long)-1. This would simplify users other
than full_name_hash"
Special thanks to Bruce Fields for testing and finding bugs in v1. (I
learned some humbling lessons about "obviously correct" code.)
On the arch-specific front, the m68k assembly has been tested in a
standalone test harness, I've been in contact with the Microblaze
maintainers who mostly don't care, as the hardware multiplier is never
omitted in real-world applications, and I haven't heard anything from
the H8/300 world"
* 'hash' of git://ftp.sciencehorizons.net/linux:
h8300: Add <asm/hash.h>
microblaze: Add <asm/hash.h>
m68k: Add <asm/hash.h>
<linux/hash.h>: Add support for architecture-specific functions
fs/namei.c: Improve dcache hash function
Eliminate bad hash multipliers from hash_32() and hash_64()
Change hash_64() return value to 32 bits
<linux/sunrpc/svcauth.h>: Define hash_str() in terms of hashlen_string()
fs/namei.c: Add hashlen_string() function
Pull out string hash to <linux/stringhash.h>
This is just the infrastructure; there are no users yet.
This is modelled on CONFIG_ARCH_RANDOM; a CONFIG_ symbol declares
the existence of <asm/hash.h>.
That file may define its own versions of various functions, and define
HAVE_* symbols (no CONFIG_ prefix!) to suppress the generic ones.
Included is a self-test (in lib/test_hash.c) that verifies the basics.
It is NOT in general required that the arch-specific functions compute
the same thing as the generic, but if a HAVE_* symbol is defined with
the value 1, then equality is tested.
Signed-off-by: George Spelvin <linux@sciencehorizons.net>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Greg Ungerer <gerg@linux-m68k.org>
Cc: Andreas Schwab <schwab@linux-m68k.org>
Cc: Philippe De Muyter <phdm@macq.eu>
Cc: linux-m68k@lists.linux-m68k.org
Cc: Alistair Francis <alistai@xilinx.com>
Cc: Michal Simek <michal.simek@xilinx.com>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Cc: uclinux-h8-devel@lists.sourceforge.jp
Patch 0fed3ac866 improved the hash mixing, but the function is slower
than necessary; there's a 7-instruction dependency chain (10 on x86)
each loop iteration.
Word-at-a-time access is a very tight loop (which is good, because
link_path_walk() is one of the hottest code paths in the entire kernel),
and the hash mixing function must not have a longer latency to avoid
slowing it down.
There do not appear to be any published fast hash functions that:
1) Operate on the input a word at a time, and
2) Don't need to know the length of the input beforehand, and
3) Have a single iterated mixing function, not needing conditional
branches or unrolling to distinguish different loop iterations.
One of the algorithms which comes closest is Yann Collet's xxHash, but
that's two dependent multiplies per word, which is too much.
The key insights in this design are:
1) Barring expensive ops like multiplies, to diffuse one input bit
across 64 bits of hash state takes at least log2(64) = 6 sequentially
dependent instructions. That is more cycles than we'd like.
2) An operation like "hash ^= hash << 13" requires a second temporary
register anyway, and on a 2-operand machine like x86, it's three
instructions.
3) A better use of a second register is to hold a two-word hash state.
With careful design, no temporaries are needed at all, so it doesn't
increase register pressure. And this gets rid of register copying
on 2-operand machines, so the code is smaller and faster.
4) Using two words of state weakens the requirement for one-round mixing;
we now have two rounds of mixing before cancellation is possible.
5) A two-word hash state also allows operations on both halves to be
done in parallel, so on a superscalar processor we get more mixing
in fewer cycles.
I ended up using a mixing function inspired by the ChaCha and Speck
round functions. It is 6 simple instructions and 3 cycles per iteration
(assuming multiply by 9 can be done by an "lea" instruction):
x ^= *input++;
y ^= x; x = ROL(x, K1);
x += y; y = ROL(y, K2);
y *= 9;
Not only is this reversible, two consecutive rounds are reversible:
if you are given the initial and final states, but not the intermediate
state, it is possible to compute both input words. This means that at
least 3 words of input are required to create a collision.
(It also has the property, used by hash_name() to avoid a branch, that
it hashes all-zero to all-zero.)
The rotate constants K1 and K2 were found by experiment. The search took
a sample of random initial states (I used 1023) and considered the effect
of flipping each of the 64 input bits on each of the 128 output bits two
rounds later. Each of the 8192 pairs can be considered a biased coin, and
adding up the Shannon entropy of all of them produces a score.
The best-scoring shifts also did well in other tests (flipping bits in y,
trying 3 or 4 rounds of mixing, flipping all 64*63/2 pairs of input bits),
so the choice was made with the additional constraint that the sum of the
shifts is odd and not too close to the word size.
The final state is then folded into a 32-bit hash value by a less carefully
optimized multiply-based scheme. This also has to be fast, as pathname
components tend to be short (the most common case is one iteration!), but
there's some room for latency, as there is a fair bit of intervening logic
before the hash value is used for anything.
(Performance verified with "bonnie++ -s 0 -n 1536:-2" on tmpfs. I need
a better benchmark; the numbers seem to show a slight dip in performance
between 4.6.0 and this patch, but they're too noisy to quote.)
Special thanks to Bruce fields for diligent testing which uncovered a
nasty fencepost error in an earlier version of this patch.
[checkpatch.pl formatting complaints noted and respectfully disagreed with.]
Signed-off-by: George Spelvin <linux@sciencehorizons.net>
Tested-by: J. Bruce Fields <bfields@redhat.com>
We'd like to make more use of the highly-optimized dcache hash functions
throughout the kernel, rather than have every subsystem create its own,
and a function that hashes basic null-terminated strings is required
for that.
(The name is to emphasize that it returns both hash and length.)
It's actually useful in the dcache itself, specifically d_alloc_name().
Other uses in the next patch.
full_name_hash() is also tweaked to make it more generally useful:
1) Take a "char *" rather than "unsigned char *" argument, to
be consistent with hash_name().
2) Handle zero-length inputs. If we want more callers, we don't want
to make them worry about corner cases.
Signed-off-by: George Spelvin <linux@sciencehorizons.net>
improvements of UBI and UBIFS.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2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=L16y
-----END PGP SIGNATURE-----
Merge tag 'upstream-4.7-rc1' of git://git.infradead.org/linux-ubifs
Pull UBI/UBIFS updates from Richard Weinberger:
"This contains mostly cleanups and minor improvements of UBI and UBIFS"
* tag 'upstream-4.7-rc1' of git://git.infradead.org/linux-ubifs:
ubifs: ubifs_dump_inode: Fix dumping field bulk_read
UBI: Fix static volume checks when Fastmap is used
UBI: Set free_count to zero before walking through erase list
UBI: Silence an unintialized variable warning
UBI: Clean up return in ubi_remove_volume()
UBI: Modify wrong comment in ubi_leb_map function.
UBI: Don't read back all data in ubi_eba_copy_leb()
UBI: Add ro-mode sysfs attribute
Older versions of gcc don't understand named initializers inside a
anonymous structure or union member. It can be worked around by adding
the bracin gin the initializer for the anonymous member.
Without this, gcc 4.4.4 will fail the build with
CC fs/nfs/nfs4state.o
fs/nfs/nfs4state.c:69: error: unknown field ‘data’ specified in initializer
fs/nfs/nfs4state.c:69: warning: missing braces around initializer
fs/nfs/nfs4state.c:69: warning: (near initialization for ‘zero_stateid.<anonymous>.data’)
make[2]: *** [fs/nfs/nfs4state.o] Error 1
introduced in commit 93b717fd81bf ("NFSv4: Label stateids with the type")
Reported-and-tested-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Cc: Anna Schumaker <Anna.Schumaker@netapp.com>
Cc: Trond Myklebust <trond.myklebust@primarydata.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
