Changes in bmap.c

e4a99d48 - nilfs2: correct return value kernel-doc descriptions for bmap functions

Fri, 10 Jan 2025 01:01:45 +0000

nilfs2: correct return value kernel-doc descriptions for bmap functionsSimilar to the previous patch to fix the ioctl return value descriptions,this fixes the format of the return value descriptions for bmap (andbtree)-related functions, which was causing the kernel-doc script to emita number of warnings.Link: https://lkml.kernel.org/r/20250110010530.21872-3-konishi.ryusuke@gmail.comSigned-off-by: Ryusuke Konishi Cc: "Brian G ." Signed-off-by: Andrew Morton List of files: /linux-6.15/fs/nilfs2/bmap.c

caaab566 - nilfs2: fix missing initial short descriptions of kernel-doc comments

Fri, 16 Aug 2024 07:43:19 +0000

nilfs2: fix missing initial short descriptions of kernel-doc commentsUpdate some kernel-doc comments that are missing the initial shortdescription and fix the following warnings output by the kernel-docscript: fs/nilfs2/bmap.c:353: warning: missing initial short description on line: * nilfs_bmap_lookup_dirty_buffers - fs/nilfs2/cpfile.c:708: warning: missing initial short description on line: * nilfs_cpfile_delete_checkpoint - fs/nilfs2/cpfile.c:972: warning: missing initial short description on line: * nilfs_cpfile_is_snapshot - fs/nilfs2/dat.c:275: warning: missing initial short description on line: * nilfs_dat_mark_dirty - fs/nilfs2/sufile.c:844: warning: missing initial short description on line: * nilfs_sufile_get_suinfo -Link: https://lkml.kernel.org/r/20240816074319.3253-9-konishi.ryusuke@gmail.comSigned-off-by: Ryusuke Konishi Signed-off-by: Andrew Morton List of files: /linux-6.15/fs/nilfs2/bmap.c

1f49c147 - nilfs2: drop usage of page_index

Tue, 21 May 2024 17:58:44 +0000

nilfs2: drop usage of page_indexPatch series "mm/swap: clean up and optimize swap cache index", v6.Currently we use one swap_address_space for every 64M chunk to reduce lockcontention, this is like having a set of smaller files inside a swapdevice. But when doing swap cache look up or insert, we are still usingthe offset of the whole large swap device. This is OK for correctness, asthe offset (key) is unique.But Xarray is specially optimized for small indexes, it creates the redixtree levels lazily to be just enough to fit the largest key stored in oneXarray. So we are wasting tree nodes unnecessarily.For 64M chunk it should only take at most 3 level to contain everything. But if we are using the offset from the whole swap device, the offset(key) value will be way beyond 64M, and so will the tree level.Optimize this by reduce the swap cache search space into 64M scope.Test with `time memhog 128G` inside a 8G memcg using 128G swap (ramdiskwith SWP_SYNCHRONOUS_IO dropped, tested 3 times, results are stable. Thetest result is similar but the improvement is smaller ifSWP_SYNCHRONOUS_IO is enabled, as swap out path can never skip swapcache):Before:6.07user 250.74system 4:17.26elapsed 99%CPU (0avgtext+0avgdata 8373376maxresident)k0inputs+0outputs (55major+33555018minor)pagefaults 0swapsAfter (+1.8% faster):6.08user 246.09system 4:12.58elapsed 99%CPU (0avgtext+0avgdata 8373248maxresident)k0inputs+0outputs (54major+33555027minor)pagefaults 0swapsSimilar result with MySQL and sysbench using swap:Before:94055.61 qpsAfter (+0.8% faster):94834.91 qpsThere is alse a very slight drop of radix tree node slab usage:Before: 303952KAfter: 302224KFor this series:There are multiple places that expect mixed type of pages (page cache orswap cache), eg. migration, huge memory split; There are four helpersfor that:- page_index- page_file_offset- folio_index- folio_file_posTo keep the code clean and compatible, this series first cleaned up usageof them.page_file_offset and folio_file_pos are historical helpes that can besimply dropped after clean up. And page_index can be all converted tofolio_index or folio->index.Then introduce two new helpers swap_cache_index and swap_dev_pos for swap.Replace swp_offset with swap_cache_index when used to retrieve folio fromswap cache, and use swap_dev_pos when needed to retrieve the deviceposition of a swap entry. This way, swap_cache_index can return theoptimized value with no compatibility issue.The result is better performance and reduced LOC.Idealy, in the future, we may want to reduce SWAP_ADDRESS_SPACE_SHIFT from14 to 12: Default Xarray chunk offset is 6, so we have 3 level treesinstead of 2 level trees just for 2 extra bits. But swap cache is basedon address_space struct, with 4 times more metadata sparsely distributedin memory it waste more cacheline, the performance gain from this seriesis almost canceled according to my test. So first, just have a cleanerseperation of offsets and smaller search space.This patch (of 10):page_index is only for mixed usage of page cache and swap cache, for purepage cache usage, the caller can just use page->index instead.It can't be a swap cache page here (being part of buffer head), so justdrop it. And while we are at it, optimize the code by retrieving theoffset of the buffer head within the folio directly using bh_offset, andget rid of the loop and usage of page helpers.Link: https://lkml.kernel.org/r/20240521175854.96038-1-ryncsn@gmail.comLink: https://lkml.kernel.org/r/20240521175854.96038-3-ryncsn@gmail.comSuggested-by: Matthew Wilcox Signed-off-by: Kairui Song Acked-by: Ryusuke Konishi Cc: Ryusuke Konishi Cc: Anna Schumaker Cc: Barry Song Cc: Chao Yu Cc: Chris Li Cc: David Hildenbrand Cc: David Howells Cc: "Huang, Ying" Cc: Hugh Dickins Cc: Ilya Dryomov Cc: Jaegeuk Kim Cc: Jeff Layton Cc: Marc Dionne Cc: Minchan Kim Cc: NeilBrown Cc: Ryan Roberts Cc: Trond Myklebust Cc: Xiubo Li Cc: Yosry Ahmed Signed-off-by: Andrew Morton List of files: /linux-6.15/fs/nilfs2/bmap.c

97d3e940 - nilfs2: do not acquire rwsem in nilfs_bmap_write()

Mon, 22 Jan 2024 14:01:56 +0000

nilfs2: do not acquire rwsem in nilfs_bmap_write()It is now clear that nilfs_bmap_write() is only used to finalize logswritten to disk. Concurrent bmap modification operations are notperformed on bmaps in this context. Additionally, this function does notmodify data used in read-only operations such as bmap lookups.Therefore, there is no need to acquire bmap->b_sem in nilfs_bmap_write(),so delete it.Link: https://lkml.kernel.org/r/20240122140202.6950-10-konishi.ryusuke@gmail.comSigned-off-by: Ryusuke Konishi Signed-off-by: Andrew Morton List of files: /linux-6.15/fs/nilfs2/bmap.c

a6a491c0 - nilfs2: fix infinite loop in nilfs_mdt_get_block()

Sun, 30 Apr 2023 19:30:46 +0000

nilfs2: fix infinite loop in nilfs_mdt_get_block()If the disk image that nilfs2 mounts is corrupted and a virtual blockaddress obtained by block lookup for a metadata file is invalid,nilfs_bmap_lookup_at_level() may return the same internal return code as-ENOENT, meaning the block does not exist in the metadata file.This duplication of return codes confuses nilfs_mdt_get_block(), causingit to read and create a metadata block indefinitely.In particular, if this happens to the inode metadata file, ifile,semaphore i_rwsem can be left held, causing task hangs in lock_mount.Fix this issue by making nilfs_bmap_lookup_at_level() treat virtual blockaddress translation failures with -ENOENT as metadata corruption insteadof returning the error code.Link: https://lkml.kernel.org/r/20230430193046.6769-1-konishi.ryusuke@gmail.comSigned-off-by: Ryusuke Konishi Tested-by: Ryusuke Konishi Reported-by: syzbot+221d75710bde87fa0e97@syzkaller.appspotmail.com Link: https://syzkaller.appspot.com/bug?extid=221d75710bde87fa0e97Cc: Signed-off-by: Andrew Morton List of files: /linux-6.15/fs/nilfs2/bmap.c

94ee1d91 - nilfs2: remove filenames from file comments

Tue, 09 Nov 2021 02:35:01 +0000

nilfs2: remove filenames from file commentsRemove filenames that are not particularly useful in file comments, andsuppress checkpatch warnings WARNING: It's generally not useful to have the filename in the fileLink: https://lkml.kernel.org/r/1635151862-11547-3-git-send-email-konishi.ryusuke@gmail.comSigned-off-by: Ryusuke Konishi Cc: Qing Wang Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds List of files: /linux-6.15/fs/nilfs2/bmap.c

64ead520 - nilfs2: fix some kernel-doc warnings for nilfs2

Fri, 16 Oct 2020 03:13:11 +0000

nilfs2: fix some kernel-doc warnings for nilfs2Fixes the following W=1 kernel build warning(s):fs/nilfs2/bmap.c:378: warning: Excess function parameter 'bhp' description in 'nilfs_bmap_assign'fs/nilfs2/cpfile.c:907: warning: Excess function parameter 'status' description in 'nilfs_cpfile_change_cpmode'fs/nilfs2/cpfile.c:946: warning: Excess function parameter 'stat' description in 'nilfs_cpfile_get_stat'fs/nilfs2/page.c:76: warning: Excess function parameter 'inode' description in 'nilfs_forget_buffer'fs/nilfs2/sufile.c:563: warning: Excess function parameter 'stat' description in 'nilfs_sufile_get_stat'Signed-off-by: Wang Hai Signed-off-by: Ryusuke Konishi Signed-off-by: Andrew Morton Link: https://lkml.kernel.org/r/1601386269-2423-1-git-send-email-konishi.ryusuke@gmail.comSigned-off-by: Linus Torvalds List of files: /linux-6.15/fs/nilfs2/bmap.c

df561f66 - treewide: Use fallthrough pseudo-keyword

Sun, 23 Aug 2020 22:36:59 +0000

treewide: Use fallthrough pseudo-keywordReplace the existing /* fall through */ comments and its variants withthe new pseudo-keyword macro fallthrough[1]. Also, remove unnecessaryfall-through markings when it is the case.[1] https://www.kernel.org/doc/html/v5.7/process/deprecated.html?highlight=fallthrough#implicit-switch-case-fall-throughSigned-off-by: Gustavo A. R. Silva List of files: /linux-6.15/fs/nilfs2/bmap.c

ae98043f - nilfs2: convert to SPDX license tags

Tue, 04 Sep 2018 22:46:30 +0000

nilfs2: convert to SPDX license tagsRemove the verbose license text from NILFS2 files and replace them withSPDX tags. This does not change the license of any of the code.Link: http://lkml.kernel.org/r/1535624528-5982-1-git-send-email-konishi.ryusuke@lab.ntt.co.jpSigned-off-by: Ryusuke Konishi Reviewed-by: Andrew Morton Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds List of files: /linux-6.15/fs/nilfs2/bmap.c

cae3d4ca - nilfs2: hide function name argument from nilfs_error()

Tue, 02 Aug 2016 21:05:00 +0000

nilfs2: hide function name argument from nilfs_error()Simplify nilfs_error(), an output function used to report criticalissues in file system. This renames the original nilfs_error() functionto __nilfs_error() and redefines it as a macro to hide its function nameargument within the macro.Every call site of nilfs_error() is changed to strip __func__ argumentexcept nilfs_bmap_convert_error(); nilfs_bmap_convert_error() directlycalls __nilfs_error() because it inherits caller's function name.Link: http://lkml.kernel.org/r/1464875891-5443-2-git-send-email-konishi.ryusuke@lab.ntt.co.jpSigned-off-by: Ryusuke Konishi Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds List of files: /linux-6.15/fs/nilfs2/bmap.c

0c6c44cb - nilfs2: avoid bare use of 'unsigned'

Mon, 23 May 2016 23:23:39 +0000

nilfs2: avoid bare use of 'unsigned'This fixes checkpatch.pl warning "WARNING: Prefer 'unsigned int' tobare use of 'unsigned'".Link: http://lkml.kernel.org/r/1462886671-3521-5-git-send-email-konishi.ryusuke@lab.ntt.co.jpSigned-off-by: Ryusuke Konishi Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds List of files: /linux-6.15/fs/nilfs2/bmap.c

06f4abf6 - nilfs2: do not emit extra newline on nilfs_warning() and nilfs_error()

Mon, 23 May 2016 23:23:31 +0000

nilfs2: do not emit extra newline on nilfs_warning() and nilfs_error()This updates call sites of nilfs_warning() and nilfs_error() so that theydon't add a duplicate newline. These output functions are alreadydesigned to add a trailing newline to the message.Link: http://lkml.kernel.org/r/1462886671-3521-2-git-send-email-konishi.ryusuke@lab.ntt.co.jpSigned-off-by: Ryusuke Konishi Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds List of files: /linux-6.15/fs/nilfs2/bmap.c

4b420ab4 - nilfs2: clean up old e-mail addresses

Mon, 23 May 2016 23:23:09 +0000

nilfs2: clean up old e-mail addressesE-mail addresses of osrg.net domain are no longer available. Thisremoves them from authorship notices and prevents reporters from beingconfused.Link: http://lkml.kernel.org/r/1461935747-10380-5-git-send-email-konishi.ryusuke@lab.ntt.co.jpSigned-off-by: Ryusuke Konishi Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds List of files: /linux-6.15/fs/nilfs2/bmap.c

5726d0b4 - nilfs2: remove FSF mailing address from GPL notices

Mon, 23 May 2016 23:23:06 +0000

nilfs2: remove FSF mailing address from GPL noticesThis removes the extra paragraph which mentions FSF address in GPLnotices from source code of nilfs2 and avoids the checkpatch.pl errorrelated to it.Link: http://lkml.kernel.org/r/1461935747-10380-4-git-send-email-konishi.ryusuke@lab.ntt.co.jpSigned-off-by: Ryusuke Konishi Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds List of files: /linux-6.15/fs/nilfs2/bmap.c

09cbfeaf - mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros

Fri, 01 Apr 2016 12:29:47 +0000

mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macrosPAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* timeago with promise that one day it will be possible to implement pagecache with bigger chunks than PAGE_SIZE.This promise never materialized. And unlikely will.We have many places where PAGE_CACHE_SIZE assumed to be equal toPAGE_SIZE. And it's constant source of confusion on whetherPAGE_CACHE_* or PAGE_* constant should be used in a particular case,especially on the border between fs and mm.Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to muchbreakage to be doable.Let's stop pretending that pages in page cache are special. They arenot.The changes are pretty straight-forward: - << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> ; - >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> ; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page();This patch contains automated changes generated with coccinelle usingscript below. For some reason, coccinelle doesn't patch header files.I've called spatch for them manually.The only adjustment after coccinelle is revert of changes toPAGE_CAHCE_ALIGN definition: we are going to drop it later.There are few places in the code where coccinelle didn't reach. I'llfix them manually in a separate patch. Comments and documentation alsowill be addressed with the separate patch.virtual patch@@expression E;@@- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)+ E@@expression E;@@- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)+ E@@@@- PAGE_CACHE_SHIFT+ PAGE_SHIFT@@@@- PAGE_CACHE_SIZE+ PAGE_SIZE@@@@- PAGE_CACHE_MASK+ PAGE_MASK@@expression E;@@- PAGE_CACHE_ALIGN(E)+ PAGE_ALIGN(E)@@expression E;@@- page_cache_get(E)+ get_page(E)@@expression E;@@- page_cache_release(E)+ put_page(E)Signed-off-by: Kirill A. Shutemov Acked-by: Michal Hocko Signed-off-by: Linus Torvalds List of files: /linux-6.15/fs/nilfs2/bmap.c

5b20384f - nilfs2: add bmap function to seek a valid key

Thu, 16 Apr 2015 19:46:36 +0000

nilfs2: add bmap function to seek a valid keyAdd a new bmap function, nilfs_bmap_seek_key(), which seeks a validentry and returns its key starting from a given key. This functioncan be used to skip hole blocks efficiently.Signed-off-by: Ryusuke Konishi Cc: Dan Carpenter Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds List of files: /linux-6.15/fs/nilfs2/bmap.c

3568a13f - nilfs2: unify type of key arguments in bmap interface

Thu, 16 Apr 2015 19:46:34 +0000

nilfs2: unify type of key arguments in bmap interfaceThe type of key arguments in block mapping interface varies dependingon function. For instance, nilfs_bmap_lookup_at_level() takes "__u64"for its key argument whereas nilfs_bmap_lookup() takes "unsignedlong".This fits them to "__u64" to eliminate the variation.Signed-off-by: Ryusuke Konishi Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds List of files: /linux-6.15/fs/nilfs2/bmap.c

0ef28f9a - nilfs2: get rid of NILFS_I_NILFS

Thu, 05 May 2011 03:56:51 +0000

nilfs2: get rid of NILFS_I_NILFSThis replaces all references of NILFS_I_NILFS(inode)->ns_bdev withinode->i_sb->s_bdev and unfolds remaining uses of NILFS_I_NILFS inlinefunction.Before 2.6.37, referring to a nilfs object from inodes needed aconditional judgement, and NILFS_I_NILFS was helpful to simplify it.But now we can simply do it by going through a super block instancelike inode->i_sb->s_fs_info.Signed-off-by: Ryusuke Konishi List of files: /linux-6.15/fs/nilfs2/bmap.c

e3154e97 - nilfs2: get rid of nilfs_sb_info structure

Wed, 09 Mar 2011 02:05:08 +0000

nilfs2: get rid of nilfs_sb_info structureThis directly uses sb->s_fs_info to keep a nilfs filesystem object andfully removes the intermediate nilfs_sb_info structure. With thischange, the hierarchy of on-memory structures of nilfs will besimplified as follows:Before: super_block -> nilfs_sb_info -> the_nilfs -> cptree --+-> nilfs_root (current file system) +-> nilfs_root (snapshot A) +-> nilfs_root (snapshot B) : -> nilfs_sc_info (log writer structure)After: super_block -> the_nilfs -> cptree --+-> nilfs_root (current file system) +-> nilfs_root (snapshot A) +-> nilfs_root (snapshot B) : -> nilfs_sc_info (log writer structure)The reason why we didn't design so from the beginning is because theinitial shape also differed from the above. The early hierachy wascomposed of "per-mount-point" super_block -> nilfs_sb_info pairs and ashared nilfs object. On the kernel 2.6.37, it was changed to thecurrent shape in order to unify super block instances into one perdevice, and this cleanup became applicable as the result.Signed-off-by: Ryusuke Konishi List of files: /linux-6.15/fs/nilfs2/bmap.c

be667377 - nilfs2: record used amount of each checkpoint in checkpoint list

Fri, 04 Mar 2011 15:19:32 +0000

nilfs2: record used amount of each checkpoint in checkpoint listThis records the number of used blocks per checkpoint in eachcheckpoint entry of cpfile. Even though userland tools can get theblock count via nilfs_get_cpinfo ioctl, it was not updated by thenilfs2 kernel code. This fixes the issue and makes it available foruserland tools to calculate used amount per checkpoint.Signed-off-by: Ryusuke Konishi Cc: Jiro SEKIBA List of files: /linux-6.15/fs/nilfs2/bmap.c