- 17 Sep, 2024 2 commits
-
-
I Hsin Cheng authored
Fix test for list_cut_position*() for the missing check of integer "i" after the second loop. The variable should be checked for second time to make sure both lists after the cut operation are formed as expected. Link: https://lkml.kernel.org/r/20240910043531.71343-1-richard120310@gmail.comSigned-off-by:
I Hsin Cheng <richard120310@gmail.com> Cc: David Gow <davidgow@google.com> Signed-off-by:
Andrew Morton <akpm@linux-foundation.org>
-
Alexey Dobriyan authored
Switch away from quite chatty declarations using typeof_member(). In theory this is faster to compile too because there is no macro expansion and there is less type checking. Link: https://lkml.kernel.org/r/81bf02fd-8724-4f4d-a2bb-c59620b7d716@p183Signed-off-by:
Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by:
-
- 09 Sep, 2024 10 commits
-
-
WangYuli authored
There are some spelling mistakes of 'retun' in comments which should be instead of 'return'. Link: https://lkml.kernel.org/r/63D0F870EE8E87A0+20240906054008.390188-1-wangyuli@uniontech.comSigned-off-by:
WangYuli <wangyuli@uniontech.com> Signed-off-by:
-
Joseph Qi authored
Return 0 instead of sizeof(ocfs2_global_disk_dqinfo) that .quota_read returns in normal case. Also cleanup mlog to make code more readable. Link: https://lkml.kernel.org/r/20240904071004.2067695-2-joseph.qi@linux.alibaba.comSigned-off-by:
Joseph Qi <joseph.qi@linux.alibaba.com> Reviewed-by:
Heming Zhao <heming.zhao@suse.com> Cc: Mark Fasheh <mark@fasheh.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Junxiao Bi <junxiao.bi@oracle.com> Cc: Changwei Ge <gechangwei@live.cn> Cc: Gang He <ghe@suse.com> Cc: Jun Piao <piaojun@huawei.com> Signed-off-by:
-
Kunwu Chan authored
Nested unlikely() calls, IS_ERR already uses unlikely() internally Link: https://lkml.kernel.org/r/20240904101618.17716-1-konishi.ryusuke@gmail.comSigned-off-by:
Kunwu Chan <chentao@kylinos.cn> Signed-off-by:
Ryusuke Konishi <konishi.ryusuke@gmail.com> Signed-off-by:
-
Ryusuke Konishi authored
The function nilfs_btree_check_delete(), which checks whether degeneration to direct mapping occurs before deleting a b-tree entry, causes memory access outside the block buffer when retrieving the maximum key if the root node has no entries. This does not usually happen because b-tree mappings with 0 child nodes are never created by mkfs.nilfs2 or nilfs2 itself. However, it can happen if the b-tree root node read from a device is configured that way, so fix this potential issue by adding a check for that case. Link: https://lkml.kernel.org/r/20240904081401.16682-4-konishi.ryusuke@gmail.com Fixes: 17c76b01 ("nilfs2: B-tree based block mapping") Signed-off-by:
-
Ryusuke Konishi authored
Due to the nature of b-trees, nilfs2 itself and admin tools such as mkfs.nilfs2 will never create an intermediate b-tree node block with 0 child nodes, nor will they delete (key, pointer)-entries that would result in such a state. However, it is possible that a b-tree node block is corrupted on the backing device and is read with 0 child nodes. Because operation is not guaranteed if the number of child nodes is 0 for intermediate node blocks other than the root node, modify nilfs_btree_node_broken(), which performs sanity checks when reading a b-tree node block, so that such cases will be judged as metadata corruption. Link: https://lkml.kernel.org/r/20240904081401.16682-3-konishi.ryusuke@gmail.com Fixes: 17c76b01 ("nilfs2: B-tree based block mapping") Signed-off-by:
-
Ryusuke Konishi authored
Patch series "nilfs2: fix potential issues with empty b-tree nodes". This series addresses three potential issues with empty b-tree nodes that can occur with corrupted filesystem images, including one recently discovered by syzbot. This patch (of 3): If a b-tree is broken on the device, and the b-tree height is greater than 2 (the level of the root node is greater than 1) even if the number of child nodes of the b-tree root is 0, a NULL pointer dereference occurs in nilfs_btree_prepare_insert(), which is called from nilfs_btree_insert(). This is because, when the number of child nodes of the b-tree root is 0, nilfs_btree_do_lookup() does not set the block buffer head in any of path[x].bp_bh, leaving it as the initial value of NULL, but if the level of the b-tree root node is greater than 1, nilfs_btree_get_nonroot_node(), which accesses the buffer memory of path[x].bp_bh, is called. Fix this issue by adding a check to nilfs_btree_root_broken(), which performs sanity checks when reading the root node from the device, to detect this inconsistency. Thanks to Lizhi Xu for trying to solve the bug and clarifying the cause early on. Link: https://lkml.kernel.org/r/20240904081401.16682-1-konishi.ryusuke@gmail.com Link: https://lkml.kernel.org/r/20240902084101.138971-1-lizhi.xu@windriver.com Link: https://lkml.kernel.org/r/20240904081401.16682-2-konishi.ryusuke@gmail.com Fixes: 17c76b01 ("nilfs2: B-tree based block mapping") Signed-off-by:
-
Jinjie Ruan authored
Let the kmemdup_array() take care about multiplication and possible overflows. Link: https://lkml.kernel.org/r/20240828072340.1249310-1-ruanjinjie@huawei.comSigned-off-by:
Jinjie Ruan <ruanjinjie@huawei.com> Reviewed-by:
Kees Cook <kees@kernel.org> Cc: Alexey Dobriyan <adobriyan@gmail.com> Cc: Christian Brauner <brauner@kernel.org> Cc: Li zeming <zeming@nfschina.com> Cc: Randy Dunlap <rdunlap@infradead.org> Signed-off-by:
-
zhangjiao authored
rm thp_swap_allocator_test when make clean Link: https://lkml.kernel.org/r/20240829042008.6937-1-zhangjiao2@cmss.chinamobile.comSigned-off-by:
zhangjiao <zhangjiao2@cmss.chinamobile.com> Signed-off-by:
-
Uros Bizjak authored
When strict percpu address space checks are enabled, then current direct casts between the percpu address space and the generic address space fail the compilation on x86_64 with: decompressor_multi_percpu.c: In function `squashfs_decompressor_create': decompressor_multi_percpu.c:49:16: error: cast to generic address space pointer from disjoint `__seg_gs' address space pointer decompressor_multi_percpu.c: In function `squashfs_decompressor_destroy': decompressor_multi_percpu.c:64:25: error: cast to `__seg_gs' address space pointer from disjoint generic address space pointer decompressor_multi_percpu.c: In function `squashfs_decompress': decompressor_multi_percpu.c:82:25: error: cast to `__seg_gs' address space pointer from disjoint generic address space pointer Add intermediate casts to unsigned long, as advised in [1] and [2]. Side note: sparse still requires __force when casting from the percpu address space, although the documentation [2] allows casts to unsigned long without __force attribute. Found by GCC's named address space checks. There were no changes in the resulting object file. [1] https://gcc.gnu.org/onlinedocs/gcc/Named-Address-Spaces.html#x86-Named-Address-Spaces [2] https://sparse.docs.kernel.org/en/latest/annotations.html#address-space-name Link: https://lkml.kernel.org/r/20240830091104.13049-1-ubizjak@gmail.comSigned-off-by:
Uros Bizjak <ubizjak@gmail.com> Cc: Phillip Lougher <phillip@squashfs.org.uk> Signed-off-by:
-
Alok Swaminathan authored
Add null check for character class. Previously, an inverted character class could result in a nul byte being matched and lead to the function reading past the end of the inputted string. Link: https://lkml.kernel.org/r/20240826155709.12383-1-swaminathanalok@gmail.comSigned-off-by:
Alok Swaminathan <swaminathanalok@gmail.com> Signed-off-by:
-
- 02 Sep, 2024 28 commits
-
-
Ryusuke Konishi authored
Simplify nilfs_segctor_thread(), the main loop function of the log writer thread, to make the basic structure easier to understand. In particular, the acquisition and release of the sc_state_lock spinlock was scattered throughout the function, so extract the determination of whether log writing is required into a helper function and make the spinlock lock sections clearer. Link: https://lkml.kernel.org/r/20240826174116.5008-9-konishi.ryusuke@gmail.comSigned-off-by:
-
Ryusuke Konishi authored
By using kthread_create() and kthread_stop() to start and stop the log writer thread, eliminate custom thread start and stop helpers, as well as the wait queue "sc_wait_task" on the "nilfs_sc_info" struct and NILFS_SEGCTOR_QUIT flag that exist only to implement them. Also, update the kernel doc comments of the changed functions as appropriate. Link: https://lkml.kernel.org/r/20240826174116.5008-8-konishi.ryusuke@gmail.comSigned-off-by:
-
Ryusuke Konishi authored
After commit f5d4e046 ("nilfs2: fix use-after-free of timer for log writer thread") is applied, nilfs_construct_timeout(), which is called by a timer and wakes up the log writer thread, is never called after the log writer thread has terminated. As a result, the member variable "sc_timer_task" of the "nilfs_sc_info" structure, which was added when timer_setup() was adopted to retain a reference to the log writer thread's task even after it had terminated, is no longer needed, as it should be; we can simply use "sc_task" instead, which holds a reference to the log writer thread's task for its lifetime. So, eliminate "sc_timer_task" by this means. Link: https://lkml.kernel.org/r/20240826174116.5008-7-konishi.ryusuke@gmail.comSigned-off-by:
-
Ryusuke Konishi authored
After commit 93aef9ed ("nilfs2: fix incorrect inode allocation from reserved inodes") is applied, the inode number returned by nilfs_ifile_create_inode() is guaranteed to always be greater than or equal to NILFS_USER_INO, so if the inode number is a reserved inode number (less than NILFS_USER_INO), the code to repair the bitmap immediately following it is no longer executed. So, delete it. Link: https://lkml.kernel.org/r/20240826174116.5008-6-konishi.ryusuke@gmail.comSigned-off-by:
-
Ryusuke Konishi authored
Use get_random_u32() as the source for inode->i_generation for new inodes, and eliminate the original source, the shared counter ns_next_generation along with its exclusive access spinlock ns_next_gen_lock. Link: https://lkml.kernel.org/r/20240826174116.5008-5-konishi.ryusuke@gmail.comSigned-off-by:
-
Ryusuke Konishi authored
In nilfs_iget_locked() and nilfs_ilookup(), which are used to find or obtain nilfs2 inodes, the nilfs_iget_args structure used to identify inodes has type information divided into multiple booleans, making type determination complicated. Simplify inode type determination by consolidating inode type information into an unsigned integer represented by a comibination of flags and by separating the type identification information for on-memory inodes from the i_state member in the nilfs_inode_info structure. Link: https://lkml.kernel.org/r/20240826174116.5008-4-konishi.ryusuke@gmail.comSigned-off-by:
-
Ryusuke Konishi authored
The macros NILFS_BMAP_KEY_BIT and NILFS_BMAP_NEW_PTR_INIT calculate, within their definitions, the number of bits in an unsigned long variable. Use the BITS_PER_LONG macro to make them simpler. Link: https://lkml.kernel.org/r/20240826174116.5008-3-konishi.ryusuke@gmail.comSigned-off-by:
-
Huang Xiaojia authored
Patch series "nilfs2: assorted cleanups". This is a collection of cleanup patches, with only the last three focused on the log writer thread, the rest are miscellaneous. Patches 1/8, 4/8, and 7/8 adopt common implementations, 2/8 uses a generic macro, 5/8 removes dead code, 6/8 removes an unnecessary reference, and 3/8 and 8/8 each simplify a paticular messy implementation. This patch (of 8): Deduplicate the nilfs2 file type conversion implementation. Link: https://lkml.kernel.org/r/20240826174116.5008-1-konishi.ryusuke@gmail.com Link: https://lkml.kernel.org/r/20240815013442.1220909-1-huangxiaojia2@huawei.com Link: https://lkml.kernel.org/r/20240826174116.5008-2-konishi.ryusuke@gmail.comSigned-off-by:
Huang Xiaojia <huangxiaojia2@huawei.com> Signed-off-by:
-
Hongbo Li authored
The helper str_false_true() was introduced to return "false/true" string literal. We can simplify this format by str_false_true. Link: https://lkml.kernel.org/r/20240827024517.914100-4-lihongbo22@huawei.comSigned-off-by:
Hongbo Li <lihongbo22@huawei.com> Cc: Andy Shevchenko <andy@kernel.org> Cc: Anna Schumaker <anna@kernel.org> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Kees Cook <kees@kernel.org> Cc: Trond Myklebust <trondmy@kernel.org> Cc: Matthew Wilcox <willy@infradead.org> Signed-off-by:
-
Hongbo Li authored
The helper str_true_false() was introduced to return "true/false" string literal. We can simplify this format by str_true_false. Link: https://lkml.kernel.org/r/20240827024517.914100-3-lihongbo22@huawei.comSigned-off-by:
-
Hongbo Li authored
Add str_true_false()/str_false_true() helper to retur a "true" or "false" string literal. We found more than 10 cases currently exist in the tree. So these helpers can be used for these cases. This patch (of 3): Add str_true_false()/str_false_true() helper to return "true" or "false" string literal. Link: https://lkml.kernel.org/r/20240827024517.914100-1-lihongbo22@huawei.com Link: https://lkml.kernel.org/r/20240827024517.914100-2-lihongbo22@huawei.comSigned-off-by:
-
Chen Yu authored
When analyzing a kernel waring message, Peter pointed out that there is a race condition when the kworker is being frozen and falls into try_to_freeze() with TASK_INTERRUPTIBLE, which could trigger a might_sleep() warning in try_to_freeze(). Although the root cause is not related to freeze()[1], it is still worthy to fix this issue ahead. One possible race scenario: CPU 0 CPU 1 ----- ----- // kthread_worker_fn set_current_state(TASK_INTERRUPTIBLE); suspend_freeze_processes() freeze_processes static_branch_inc(&freezer_active); freeze_kernel_threads pm_nosig_freezing = true; if (work) { //false __set_current_state(TASK_RUNNING); } else if (!freezing(current)) //false, been frozen freezing(): if (static_branch_unlikely(&freezer_active)) if (pm_nosig_freezing) return true; schedule() } // state is still TASK_INTERRUPTIBLE try_to_freeze() might_sleep() <--- warning Fix this by explicitly set the TASK_RUNNING before entering try_to_freeze(). Link: https://lore.kernel.org/lkml/Zs2ZoAcUsZMX2B%2FI@chenyu5-mobl2/ [1] Link: https://lkml.kernel.org/r/20240827112308.181081-1-yu.c.chen@intel.com Fixes: b56c0d89 ("kthread: implement kthread_worker") Signed-off-by:Chen Yu <yu.c.chen@intel.com> Suggested-by:
Peter Zijlstra <peterz@infradead.org> Suggested-by:
-
Luca Ceresoli authored
When no parameters are passed, the usage instructions are presented only when debuginfod-find is not found. This makes sense because with debuginfod none of the positional parameters are needed. However it means that users having debuginfod-find installed will have no chance of reading the usage text without opening the file. Many programs have a '-h' flag to get the usage, so add such a flag. Invoking 'scripts/decode_stacktrace.sh -h' will now show the usage text and exit. Link: https://lkml.kernel.org/r/20240823-decode_stacktrace-find_module-improvements-v2-3-d7a57d35558b@bootlin.comSigned-off-by:
Luca Ceresoli <luca.ceresoli@bootlin.com> Reviewed-by:
Stephen Boyd <swboyd@chromium.org> Cc: Alexis Lothoré (eBPF Foundation) <alexis.lothore@bootlin.com> Cc: Konstantin Khlebnikov <koct9i@gmail.com> Cc: Sasha Levin <sashal@kernel.org> Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com> Signed-off-by:
-
Luca Ceresoli authored
The syntax as expressed by usage() is not entirely correct: "<modules path>" cannot be passed without "<base path>|auto". Additionally human reading of this syntax can be subject to misunderstanding due the mixture of '|' and '[]'. Improve readability in various ways: * rewrite using two lines for the two allowed usages * add square brackets around "<vmlinux>" as it is optional when using debuginfod-find * move "<modules path>" to inside the square brackets of the 2nd positional parameter * use underscores instead of spaces in <...> strings Link: https://lkml.kernel.org/r/20240823-decode_stacktrace-find_module-improvements-v2-2-d7a57d35558b@bootlin.comSigned-off-by:
-
Luca Ceresoli authored
Patch series "scripts/decode_stacktrace.sh: improve error reporting and usability", v2. This small series improves usability of scripts/decode_stacktrace.sh by improving the usage text and correctly reporting when modules are built without debugging symbols. This patch (of 3): The find_module() function can fail for two reasons: * the module was not found * the module was found but without debugging info In both cases the user is reported the same error: WARNING! Modules path isn't set, but is needed to parse this symbol This is misleading in case the modules path is set correctly. find_module() is currently implemented as a recursive function based on global variables in order to check up to 4 different paths. This is not straightforward to read and even less to modify. Besides, the debuginfo code at the beginning of find_module() is executed identically every time the function is entered, i.e. up to 4 times per each module search due to recursion. To be able to improve error reporting, first rewrite the find_module() function to remove recursion. The new version of the function iterates over all the same (up to 4) paths as before and for each of them does the same checks as before. At the end of the iteration it is now able to print an appropriate error message, so that has been moved from the caller into find_module(). Finally, when the module is found but without debugging info, mention the two Kconfig variables one needs to set in order to have the needed debugging symbols. Link: https://lkml.kernel.org/r/20240823-decode_stacktrace-find_module-improvements-v2-0-d7a57d35558b@bootlin.com Link: https://lkml.kernel.org/r/20240823-decode_stacktrace-find_module-improvements-v2-1-d7a57d35558b@bootlin.comSigned-off-by:
-
Yang Ruibin authored
debugfs_create_dir() returns error pointers. It never returns NULL. So use IS_ERR() to check it. Link: https://lkml.kernel.org/r/20240821073441.9701-1-11162571@vivo.comSigned-off-by:
Yang Ruibin <11162571@vivo.com> Signed-off-by:
-
Andy Shevchenko authored
*-objs suffix is reserved rather for (user-space) host programs while usually *-y suffix is used for kernel drivers (although *-objs works for that purpose for now). Let's correct the old usages of *-objs in Makefiles. Link: https://lkml.kernel.org/r/20240821155140.611514-1-andriy.shevchenko@linux.intel.comSigned-off-by:
Andy Shevchenko <andriy.shevchenko@linux.intel.com> Acked-by:
Florian Fainelli <florian.fainelli@broadcom.com> Reviewed-by:
Alexander Lobakin <aleksander.lobakin@intel.com> Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk> Cc: Tal Gilboa <talgi@nvidia.com> Signed-off-by:
-
Ryusuke Konishi authored
nilfs_sufile_mark_dirty(), which marks a block in the sufile metadata file as dirty in preparation for log writing, returns -ENOENT to the caller if the block containing the segment usage of the specified segment is missing. This internal code can propagate through the log writer to system calls such as fsync. To prevent this, treat this case as a filesystem error and return -EIO instead. Link: https://lkml.kernel.org/r/20240821154627.11848-6-konishi.ryusuke@gmail.comSigned-off-by:
-
Ryusuke Konishi authored
nilfs_sufile_freev(), which is used to free segments in GC, aborts with -ENOENT if the target segment usage is on a hole block. This error only occurs if one of the segment numbers to be freed passed by the GC ioctl is invalid, so return -EINVAL instead. To avoid impairing readability, introduce a wrapper function that encapsulates error handling including the error code conversion (and error message output). Link: https://lkml.kernel.org/r/20240821154627.11848-5-konishi.ryusuke@gmail.comSigned-off-by:
-
Ryusuke Konishi authored
nilfs_sufile_free() returns the error code -ENOENT when the block where the segment usage should be placed does not exist (hole block case), but this error should not be propagated upwards to the mount system call. In nilfs_prepare_segment_for_recovery(), one of the recovery steps during mount, nilfs_sufile_free() is used and may return -ENOENT as is, so in that case return -EINVAL instead. Link: https://lkml.kernel.org/r/20240821154627.11848-4-konishi.ryusuke@gmail.comSigned-off-by:
-
Ryusuke Konishi authored
The cpfile, a metadata file that holds metadata for checkpoint management, also has statistical information in its first block, and if reading this block fails, it receives the internal code -ENOENT and returns that code to the callers. As with sufile, to prevent this -ENOENT from being propagated to system calls, return -EIO instead when reading the header block fails. Link: https://lkml.kernel.org/r/20240821154627.11848-3-konishi.ryusuke@gmail.comSigned-off-by:
-
Ryusuke Konishi authored
Patch series "nilfs2: prevent unexpected ENOENT propagation". This series fixes potential issues where the result code -ENOENT, which is returned internally when a metadata file operation encouters a hole block, is exposed to user space without being properly handled. Several issues with the same cause leading to hangs or WARN_ON check failures have been reported by syzbot and fixed each time in the past. This collectively fixes the missing -ENOENT conversions that do not cause stability issues and are not covered by syzbot. This patch (of 5): The sufile, a metadata file that holds metadata for segment management, has statistical information in its first block, but if reading this block fails, it receives the internal code -ENOENT and returns it unchanged to the callers. To prevent this -ENOENT from being propagated to system calls, if reading the header block fails, return -EIO (or -EINVAL depending on the context) instead. Link: https://lkml.kernel.org/r/20240821154627.11848-1-konishi.ryusuke@gmail.com Link: https://lkml.kernel.org/r/20240821154627.11848-2-konishi.ryusuke@gmail.comSigned-off-by:
-
Thorsten Blum authored
Use the max() macro to simplify the ocfs2_dlm_seq_show() function and improve its readability. Link: https://lkml.kernel.org/r/20240820021605.97887-3-thorsten.blum@toblux.comSigned-off-by:
Thorsten Blum <thorsten.blum@toblux.com> Reviewed-by:
-
qasdev authored
This patch addresses a shift-out-of-bounds error in the ocfs2_verify_volume() function, identified by UBSAN. The bug was triggered by an invalid s_clustersize_bits value (e.g., 1548), which caused the expression "1 << le32_to_cpu(di->id2.i_super.s_clustersize_bits)" to exceed the limits of a 32-bit integer, leading to an out-of-bounds shift. Link: https://lkml.kernel.org/r/ZsPvwQAXd5R/jNY+@hostnameSigned-off-by:
Qasim Ijaz <qasdev00@gmail.com> Reported-by:
syzbot <syzbot+f3fff775402751ebb471@syzkaller.appspotmail.com> Closes: https://syzkaller.appspot.com/bug?extid=f3fff775402751ebb471Tested-by:
-
Mateusz Guzik authored
Only bit 1 is used, making an unsigned long a total overkill. This brings it from 40 to 32 bytes, which in turn shrinks user_struct from 136 to 128 bytes. Since the latter is allocated with hwalign, this means the total usage goes down from 192 to 128 bytes per object. No functional changes. Link: https://lkml.kernel.org/r/20240817123754.240924-1-mjguzik@gmail.comSigned-off-by:
Mateusz Guzik <mjguzik@gmail.com> Signed-off-by:
-
Kent Overstreet authored
nix only puts /usr/bin/env at the standard location (as required by posix), so shebangs have to be tweaked. Link: https://lkml.kernel.org/r/20240817215025.161628-1-kent.overstreet@linux.devSigned-off-by:
Kent Overstreet <kent.overstreet@linux.dev> Cc: Bjorn Andersson <quic_bjorande@quicinc.com> Cc: Elliot Berman <quic_eberman@quicinc.com> Cc: Xiong Nandi <xndchn@gmail.com> Signed-off-by:
-
Chi Zhiling authored
In a guest virtual machine, we found that there is unexpected data zeroing problem detected occassionly: XFS (vdb): Mounting V5 Filesystem XFS (vdb): Ending clean mount XFS (vdb): Metadata CRC error detected at xfs_refcountbt_read_verify+0x2c/0xf0, xfs_refcountbt block 0x200028 XFS (vdb): Unmount and run xfs_repair XFS (vdb): First 128 bytes of corrupted metadata buffer: 00000000e0cd2f5e: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ 00000000cafd57f5: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ 00000000d0298d7d: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ 00000000f0698484: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ 00000000adb789a7: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ 000000005292b878: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ 00000000885b4700: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ 00000000fd4b4df7: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ XFS (vdb): metadata I/O error in "xfs_trans_read_buf_map" at daddr 0x200028 len 8 error 74 XFS (vdb): Error -117 recovering leftover CoW allocations. XFS (vdb): xfs_do_force_shutdown(0x8) called from line 994 of file fs/xfs/xfs_mount.c. Return address = 000000003a53523a XFS (vdb): Corruption of in-memory data detected. Shutting down filesystem XFS (vdb): Please umount the filesystem and rectify the problem(s) It turns out that the root cause is from the physical host machine. More specifically, it is caused by the ocfs2. when the page_size is 64k, the block should advance by 16 each time instead of 1. This will lead to a wrong mapping from the page to the disk, which will zero some adjacent part of the disk. Link: https://lkml.kernel.org/r/20240815092141.1223238-1-chizhiling@163.comSigned-off-by:
Chi Zhiling <chizhiling@kylinos.cn> Suggested-by:
Shida Zhang <zhangshida@kylinos.cn> Reviewed-by:
-
Kuan-Wei Chiu authored
The custom swap functions used in ocfs2 do not perform any special operations and can be replaced with the built-in swap function of sort. This change not only reduces code size but also improves efficiency, especially in scenarios where CONFIG_RETPOLINE is enabled, as it makes indirect function calls more expensive. By using the built-in swap, we avoid these costly indirect function calls, leading to better performance. Link: https://lkml.kernel.org/r/20240810195316.186504-1-visitorckw@gmail.comSigned-off-by:
Kuan-Wei Chiu <visitorckw@gmail.com> Reviewed-by:
-
