- 11 Aug, 2016 3 commits
-
-
Jan Kara authored
We did not count with the padding of xattr value when computing desired shift of xattrs in the inode when expanding i_extra_isize. As a result we could create unaligned start of inline xattrs. Account for alignment properly. CC: stable@vger.kernel.org # 4.4.x- Signed-off-by:Jan Kara <jack@suse.cz>
-
Jan Kara authored
When multiple xattrs need to be moved out of inode, we did not properly recompute total size of xattr headers in the inode and the new header position. Thus when moving the second and further xattr we asked ext4_xattr_shift_entries() to move too much and from the wrong place, resulting in possible xattr value corruption or general memory corruption. CC: stable@vger.kernel.org # 4.4.x Signed-off-by:
Theodore Ts'o <tytso@mit.edu>
-
Jan Kara authored
The code in ext4_expand_extra_isize_ea() treated new_extra_isize argument sometimes as the desired target i_extra_isize and sometimes as the amount by which we need to grow current i_extra_isize. These happen to coincide when i_extra_isize is 0 which used to be the common case and so nobody noticed this until recently when we added i_projid to the inode and so i_extra_isize now needs to grow from 28 to 32 bytes. The result of these bugs was that we sometimes unnecessarily decided to move xattrs out of inode even if there was enough space and we often ended up corrupting in-inode xattrs because arguments to ext4_xattr_shift_entries() were just wrong. This could demonstrate itself as BUG_ON in ext4_xattr_shift_entries() triggering. Fix the problem by introducing new isize_diff variable and use it where appropriate. CC: stable@vger.kernel.org # 4.4.x Reported-by:
Dave Chinner <david@fromorbit.com> Signed-off-by:
-
- 01 Aug, 2016 2 commits
-
-
Theodore Ts'o authored
A number of fuzzing failures seem to be caused by allocation bitmaps or other metadata blocks being pointed at the superblock. This can cause kernel BUG or WARNings once the superblock is overwritten, so validate the group descriptor blocks to make sure this doesn't happen. Cc: stable@vger.kernel.org Signed-off-by: -
Theodore Ts'o authored
The Hurd is using inode fields which restricts it from using more advanced ext4 file system features, due to design choices made over a decade ago. By giving the Hurd an extended attribute index field we allow it to move the translator and author fields out of the core inode fields, and hopefully we can get rid of ugly hacks such as EXT4_OS_HURD and EXT4_MOUNT2_HURD_COMPAT somday. For more information please see: https://summerofcode.withgoogle.com/projects/#5869799859027968Signed-off-by:
-
- 15 Jul, 2016 4 commits
-
-
Vegard Nossum authored
Although the extent tree depth of 5 should enough be for the worst case of 2*32 extents of length 1, the extent tree code does not currently to merge nodes which are less than half-full with a sibling node, or to shrink the tree depth if possible. So it's possible, at least in theory, for the tree depth to be greater than 5. However, even in the worst case, a tree depth of 32 is highly unlikely, and if the file system is maliciously corrupted, an insanely large eh_depth can cause memory allocation failures that will trigger kernel warnings (here, eh_depth = 65280): JBD2: ext4.exe wants too many credits credits:195849 rsv_credits:0 max:256 ------------[ cut here ]------------ WARNING: CPU: 0 PID: 50 at fs/jbd2/transaction.c:293 start_this_handle+0x569/0x580 CPU: 0 PID: 50 Comm: ext4.exe Not tainted 4.7.0-rc5+ #508 Stack: 604a8947 625badd8 0002fd09 00000000 60078643 00000000 62623910 601bf9bc 62623970 6002fc84 626239b0 900000125 Call Trace: [<6001c2dc>] show_stack+0xdc/0x1a0 [<601bf9bc>] dump_stack+0x2a/0x2e [<6002fc84>] __warn+0x114/0x140 [<6002fdff>] warn_slowpath_null+0x1f/0x30 [<60165829>] start_this_handle+0x569/0x580 [<60165d4e>] jbd2__journal_start+0x11e/0x220 [<60146690>] __ext4_journal_start_sb+0x60/0xa0 [<60120a81>] ext4_truncate+0x131/0x3a0 [<60123677>] ext4_setattr+0x757/0x840 [<600d5d0f>] notify_change+0x16f/0x2a0 [<600b2b16>] do_truncate+0x76/0xc0 [<600c3e56>] path_openat+0x806/0x1300 [<600c55c9>] do_filp_open+0x89/0xf0 [<600b4074>] do_sys_open+0x134/0x1e0 [<600b4140>] SyS_open+0x20/0x30 [<6001ea68>] handle_syscall+0x88/0x90 [<600295fd>] userspace+0x3fd/0x500 [<6001ac55>] fork_handler+0x85/0x90 ---[ end trace 08b0b88b6387a244 ]--- [ Commit message modified and the extent tree depath check changed from 5 to 32 -- tytso ] Cc: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by:Vegard Nossum <vegard.nossum@oracle.com> Signed-off-by:
-
Vegard Nossum authored
If we encounter a filesystem error during orphan cleanup, we should stop. Otherwise, we may end up in an infinite loop where the same inode is processed again and again. EXT4-fs (loop0): warning: checktime reached, running e2fsck is recommended EXT4-fs error (device loop0): ext4_mb_generate_buddy:758: group 2, block bitmap and bg descriptor inconsistent: 6117 vs 0 free clusters Aborting journal on device loop0-8. EXT4-fs (loop0): Remounting filesystem read-only EXT4-fs error (device loop0) in ext4_free_blocks:4895: Journal has aborted EXT4-fs error (device loop0) in ext4_do_update_inode:4893: Journal has aborted EXT4-fs error (device loop0) in ext4_do_update_inode:4893: Journal has aborted EXT4-fs error (device loop0) in ext4_ext_remove_space:3068: IO failure EXT4-fs error (device loop0) in ext4_ext_truncate:4667: Journal has aborted EXT4-fs error (device loop0) in ext4_orphan_del:2927: Journal has aborted EXT4-fs error (device loop0) in ext4_do_update_inode:4893: Journal has aborted EXT4-fs (loop0): Inode 16 (00000000618192a0): orphan list check failed! [...] EXT4-fs (loop0): Inode 16 (0000000061819748): orphan list check failed! [...] EXT4-fs (loop0): Inode 16 (0000000061819bf0): orphan list check failed! [...] See-also: c9eb13a9 ("ext4: fix hang when processing corrupted orphaned inode list") Cc: Jan Kara <jack@suse.cz> Signed-off-by: -
Vegard Nossum authored
If we hit this error when mounted with errors=continue or errors=remount-ro: EXT4-fs error (device loop0): ext4_mb_mark_diskspace_used:2940: comm ext4.exe: Allocating blocks 5090-6081 which overlap fs metadata then ext4_mb_new_blocks() will call ext4_mb_release_context() and try to continue. However, ext4_mb_release_context() is the wrong thing to call here since we are still actually using the allocation context. Instead, just error out. We could retry the allocation, but there is a possibility of getting stuck in an infinite loop instead, so this seems safer. [ Fixed up so we don't return EAGAIN to userspace. --tytso ] Fixes: 8556e8f3 ("ext4: Don't allow new groups to be added during block allocation") Signed-off-by: -
Theodore Ts'o authored
Signed-off-by:
-
- 10 Jul, 2016 1 commit
-
-
Jaegeuk Kim authored
This patch removes the most parts of internal crypto codes. And then, it modifies and adds some ext4-specific crypt codes to use the generic facility. Signed-off-by:
Jaegeuk Kim <jaegeuk@kernel.org> Signed-off-by:
-
- 06 Jul, 2016 3 commits
-
-
Carlos Maiolino authored
This bug can be reproducible with fsfuzzer, although, I couldn't reproduce it 100% of my tries, it is quite easily reproducible. During the deletion of an inode, ext2_xattr_delete_inode() does not check if the block pointed by EXT2_I(inode)->i_file_acl is a valid data block, this might lead to a deadlock, when i_file_acl == 1, and the filesystem block size is 1024. In that situation, ext2_xattr_delete_inode, will load the superblock's buffer head (instead of a valid i_file_acl block), and then lock that buffer head, which, ext2_sync_super will also try to lock, making the filesystem deadlock in the following stack trace: root 17180 0.0 0.0 113660 660 pts/0 D+ 07:08 0:00 rmdir /media/test/dir1 [<ffffffff8125da9f>] __sync_dirty_buffer+0xaf/0x100 [<ffffffff8125db03>] sync_dirty_buffer+0x13/0x20 [<ffffffffa03f0d57>] ext2_sync_super+0xb7/0xc0 [ext2] [<ffffffffa03f10b9>] ext2_error+0x119/0x130 [ext2] [<ffffffffa03e9d93>] ext2_free_blocks+0x83/0x350 [ext2] [<ffffffffa03f3d03>] ext2_xattr_delete_inode+0x173/0x190 [ext2] [<ffffffffa03ee9e9>] ext2_evict_inode+0xc9/0x130 [ext2] [<ffffffff8123fd23>] evict+0xb3/0x180 [<ffffffff81240008>] iput+0x1b8/0x240 [<ffffffff8123c4ac>] d_delete+0x11c/0x150 [<ffffffff8122fa7e>] vfs_rmdir+0xfe/0x120 [<ffffffff812340ee>] do_rmdir+0x17e/0x1f0 [<ffffffff81234dd6>] SyS_rmdir+0x16/0x20 [<ffffffff81838cf2>] entry_SYSCALL_64_fastpath+0x1a/0xa4 [<ffffffffffffffff>] 0xffffffffffffffff Fix this by using the same approach ext4 uses to test data blocks validity, implementing ext2_data_block_valid. An another possibility when the superblock is very corrupted, is that i_file_acl is 1, block_count is 1 and first_data_block is 0. For such situations, we might have i_file_acl pointing to a 'valid' block, but still step over the superblock. The approach I used was to also test if the superblock is not in the range described by ext2_data_block_valid() arguments Signed-off-by:
Carlos Maiolino <cmaiolino@redhat.com> Signed-off-by:
-
Wang Shilong authored
We should always transfer quota accounting, regardless of whether quota limits are enabled. Steps to reproduce: # mkfs.ext4 /dev/sda4 -O quota,project # mount /dev/sda4 /mnt/test # cp /bin/bash /mnt/test # chattr -p 123 /mnt/test/bash # quota -v -P 123 Signed-off-by:
Wang Shilong <wshilong@ddn.com> Signed-off-by:
-
Theodore Ts'o authored
If s_reserved_gdt_blocks is extremely large, it's possible for ext4_init_block_bitmap(), which is called when ext4 sets up an uninitialized block bitmap, to corrupt random kernel memory. Add the same checks which e2fsck has --- it must never be larger than blocksize / sizeof(__u32) --- and then add a backup check in ext4_init_block_bitmap() in case the superblock gets modified after the file system is mounted. Reported-by:
-
- 05 Jul, 2016 1 commit
-
-
yalin wang authored
Signed-off-by:
yalin wang <yalin.wang2010@gmail.com> Signed-off-by:
Jan Kara <jack@suse.com>
-
- 04 Jul, 2016 4 commits
-
-
Vegard Nossum authored
If ext4_fill_super() fails early, it's possible for ext4_evict_inode() to call ext4_should_journal_data() before superblock options and flags are fully set up. In that case, the iput() on the journal inode can end up causing a BUG(). Work around this problem by reordering the tests so we only call ext4_should_journal_data() after we know it's not the journal inode. Fixes: 2d859db3 ("ext4: fix data corruption in inodes with journalled data") Fixes: 2b405bfa ("ext4: fix data=journal fast mount/umount hang") Cc: Jan Kara <jack@suse.cz> Cc: stable@vger.kernel.org Signed-off-by:
-
Pranay Kr. Srivastava authored
If there are racing calls to ext4_commit_super() it's possible for another writeback of the superblock to result in the buffer being marked with an error after we check if the buffer is marked as having a write error and the buffer up-to-date flag is set again. If that happens mark_buffer_dirty() can end up throwing a WARN_ON_ONCE. Fix this by moving this check to write before we call write_buffer_dirty(), and keeping the buffer locked during this whole sequence. Signed-off-by:
Pranay Kr. Srivastava <pranjas@gmail.com> Signed-off-by:
-
Jan Kara authored
Commit 06bd3c36 (ext4: fix data exposure after a crash) uncovered a deadlock in ext4_writepages() which was previously much harder to hit. After this commit xfstest generic/130 reproduces the deadlock on small filesystems. The problem happens when ext4_do_update_inode() sets LARGE_FILE feature and marks current inode handle as synchronous. That subsequently results in ext4_journal_stop() called from ext4_writepages() to block waiting for transaction commit while still holding page locks, reference to io_end, and some prepared bio in mpd structure each of which can possibly block transaction commit from completing and thus results in deadlock. Fix the problem by releasing page locks, io_end reference, and submitting prepared bio before calling ext4_journal_stop(). [ Changed to defer the call to ext4_journal_stop() only if the handle is synchronous. --tytso ] Reported-and-tested-by:
Eryu Guan <eguan@redhat.com> Signed-off-by:
-
Daeho Jeong authored
ext4_dx_csum_verify() returns the success return value in two checksum verification failure cases. We need to set the return values to zero as failure like ext4_dirent_csum_verify() returning zero when failing to find a checksum dirent at the tail. Signed-off-by:
Daeho Jeong <daeho.jeong@samsung.com> Signed-off-by:
Darrick J. Wong <darrick.wong@oracle.com>
-
- 03 Jul, 2016 1 commit
-
-
Daeho Jeong authored
We temporally change checksum fields in buffers of some types of metadata into '0' for verifying the checksum values. By doing this without locking the buffer, some metadata's checksums, which are being committed or written back to the storage, could be damaged. In our test, several metadata blocks were found with damaged metadata checksum value during recovery process. When we only verify the checksum value, we have to avoid modifying checksum fields directly. Signed-off-by:
Youngjin Gil <youngjin.gil@samsung.com> Signed-off-by:
-
- 30 Jun, 2016 5 commits
-
-
Vegard Nossum authored
An extent with lblock = 4294967295 and len = 1 will pass the ext4_valid_extent() test: ext4_lblk_t last = lblock + len - 1; if (len == 0 || lblock > last) return 0; since last = 4294967295 + 1 - 1 = 4294967295. This would later trigger the BUG_ON(es->es_lblk + es->es_len < es->es_lblk) in ext4_es_end(). We can simplify it by removing the - 1 altogether and changing the test to use lblock + len <= lblock, since now if len = 0, then lblock + 0 == lblock and it fails, and if len > 0 then lblock + len > lblock in order to pass (i.e. it doesn't overflow). Fixes: 5946d089 ("ext4: check for overlapping extents in ext4_valid_extent_entries()") Fixes: 2f974865 ("ext4: check for zero length extent explicitly") Cc: Eryu Guan <guaneryu@gmail.com> Cc: stable@vger.kernel.org Signed-off-by:
Phil Turnbull <phil.turnbull@oracle.com> Signed-off-by:
-
Arnd Bergmann authored
The jbd2 journal stores the commit time in 64-bit seconds and 32-bit nanoseconds, which avoids an overflow in 2038, but it gets the numbers from current_kernel_time(), which uses 'long' seconds on 32-bit architectures. This simply changes the code to call current_kernel_time64() so we use 64-bit seconds consistently. Signed-off-by:
Arnd Bergmann <arnd@arndb.de> Signed-off-by:
-
Jan Kara authored
So far we were tracking only dependency on transaction commit due to starting a new handle (which may require commit to start a new transaction). Now add tracking also for other cases where we wait for transaction commit. This way lockdep can catch deadlocks e. g. because we call jbd2_journal_stop() for a synchronous handle with some locks held which rank below transaction start. Signed-off-by:
-
Jan Kara authored
Currently lockdep map is tracked in each journal handle. To be able to expand lockdep support to cover also other cases where we depend on transaction commit and where handle is not available, move lockdep map into struct journal_s. Since this makes the lockdep map shared for all handles, we have to use rwsem_acquire_read() for acquisitions now. Signed-off-by:
-
Jan Kara authored
The transaction the handle references is free to commit once we've decremented t_updates counter. Move the lockdep instrumentation to that place. Currently it was a bit later which did not really matter but subsequent improvements to lockdep instrumentation would cause false positives with it. Signed-off-by:
-
- 26 Jun, 2016 2 commits
-
-
Theodore Ts'o authored
Also, if we are going to issue the barrier, we should do this after we write out the parent directories if necessary. Signed-off-by:
-
Theodore Ts'o authored
If there are no pending blocks to be released after a commit, forcing a journal commit has no hope of helping. It's possible that a commit had just completed, so if there are now free blocks available for allocation, it's worth retrying the commit. Reported-by:
Chao Yu <yuchao0@huawei.com> Signed-off-by:
-
- 29 May, 2016 3 commits
-
-
Linus Torvalds authored
-
George Spelvin authored
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: fcfd2fbf ("fs/namei.c: Add hashlen_string() function") Signed-off-by:
Linus Torvalds <torvalds@linux-foundation.org>
-
George Spelvin authored
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:
Guenter Roeck <linux@roeck-us.net> Fixes: fcfd2fbf ("fs/namei.c: Add hashlen_string() function") Signed-off-by:
-
- 28 May, 2016 11 commits
-
-
Mikulas Patocka authored
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:
-
Mikulas Patocka authored
Commit c8f33d0b ("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: c8f33d0b ("affs: kstrdup() memory handling") Signed-off-by:
-
Mikulas Patocka authored
Commit ce657611 ("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: ce657611 ("hpfs: kstrdup() out of memory handling") Signed-off-by:
-
git://git.linux-mips.org/pub/scm/ralf/upstream-linusLinus Torvalds authored
Pull more MIPS updates from Ralf Baechle: "This is the secondnd batch of MIPS patches for 4.7. Summary: CPS: - Copy EVA configuration when starting secondary VPs. EIC: - Clear Status IPL. Lasat: - Fix a few off by one bugs. lib: - Mark intrinsics notrace. Not only are the intrinsics uninteresting, it would cause infinite recursion. MAINTAINERS: - Add file patterns for MIPS BRCM device tree bindings. - Add file patterns for mips device tree bindings. MT7628: - Fix MT7628 pinmux typos. - wled_an pinmux gpio. - EPHY LEDs pinmux support. Pistachio: - Enable KASLR VDSO: - Build microMIPS VDSO for microMIPS kernels. - Fix aliasing warning by building with `-fno-strict-aliasing' for debugging but also tracing them might result in recursion. Misc: - Add missing FROZEN hotplug notifier transitions. - Fix clk binding example for varioius PIC32 devices. - Fix cpu interrupt controller node-names in the DT files. - Fix XPA CPU feature separation. - Fix write_gc0_* macros when writing zero. - Add inline asm encoding helpers. - Add missing VZ accessor microMIPS encodings. - Fix little endian microMIPS MSA encodings. - Add 64-bit HTW fields and fix its configuration. - Fix sigreturn via VDSO on microMIPS kernel. - Lots of typo fixes. - Add definitions of SegCtl registers and use them" * 'upstream' of git://git.linux-mips.org/pub/scm/ralf/upstream-linus: (49 commits) MIPS: Add missing FROZEN hotplug notifier transitions MIPS: Build microMIPS VDSO for microMIPS kernels MIPS: Fix sigreturn via VDSO on microMIPS kernel MIPS: devicetree: fix cpu interrupt controller node-names MIPS: VDSO: Build with `-fno-strict-aliasing' MIPS: Pistachio: Enable KASLR MIPS: lib: Mark intrinsics notrace MIPS: Fix 64-bit HTW configuration MIPS: Add 64-bit HTW fields MAINTAINERS: Add file patterns for mips device tree bindings MAINTAINERS: Add file patterns for mips brcm device tree bindings MIPS: Simplify DSP instruction encoding macros MIPS: Add missing tlbinvf/XPA microMIPS encodings MIPS: Fix little endian microMIPS MSA encodings MIPS: Add missing VZ accessor microMIPS encodings MIPS: Add inline asm encoding helpers MIPS: Spelling fix lets -> let's MIPS: VR41xx: Fix typo MIPS: oprofile: Fix typo MIPS: math-emu: Fix typo ... -
Guenter Roeck authored
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: 5d22fc25 ("mm: remove more IS_ERR_VALUE abuses") Signed-off-by: -
git://ftp.sciencehorizons.net/linuxLinus Torvalds authored
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 689de1d6 ("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 0fed3ac8 ("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> -
George Spelvin authored
This will improve the performance of hash_32() and hash_64(), but due to complete lack of multi-bit shift instructions on H8, performance will still be bad in surrounding code. Designing H8-specific hash algorithms to work around that is a separate project. (But if the maintainers would like to get in touch...) Signed-off-by: -
George Spelvin authored
Microblaze is an FPGA soft core that can be configured various ways. If it is configured without a multiplier, the standard __hash_32() will require a call to __mulsi3, which is a slow software loop. Instead, use a shift-and-add sequence for the constant multiply. GCC knows how to do this, but it's not as clever as some. Signed-off-by: -
George Spelvin authored
This provides a multiply by constant GOLDEN_RATIO_32 = 0x61C88647 for the original mc68000, which lacks a 32x32-bit multiply instruction. Yes, the amount of optimization effort put in is excessive. :-) Shift-add chain found by Yevgen Voronenko's Hcub algorithm at http://spiral.ece.cmu.edu/mcm/gen.htmlSigned-off-by:
-
George Spelvin authored
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 authored
Patch 0fed3ac8 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:
J. Bruce Fields <bfields@redhat.com>
-
