kernel.git/fs/ocfs2/suballoc.c, branch linux-6.10.y

ocfs2: speed up chain-list searching

2024-04-26T04:07:04Z

Add short-circuit code to speed up searching Link: https://lkml.kernel.org/r/20240328125203.20892-4-heming.zhao@suse.com Signed-off-by: Heming Zhao Reviewed-by: Joseph Qi Cc: Changwei Ge Cc: Gang He Cc: Joel Becker Cc: Jun Piao Cc: Junxiao Bi Cc: Mark Fasheh Signed-off-by: Andrew Morton

ocfs2: adjust enabling place for la window

2024-04-26T04:07:03Z

Patch series "improve write IO performance when fragmentation is high", v6. This patch (of 4): After introducing gd->bg_contig_free_bits, the code path 'ocfs2_cluster_group_search() => ocfs2_local_alloc_seen_free_bits()' becomes death when all the gd->bg_contig_free_bits are set to the correct value. This patch relocates ocfs2_local_alloc_seen_free_bits() to a more appropriate location. (The new place being ocfs2_block_group_set_bits().) In ocfs2_local_alloc_seen_free_bits(), the scope of the spin-lock has been adjusted to reduce meaningless lock races. e.g: when userspace creates & deletes 1 cluster_size files in parallel, acquiring the spin-lock in ocfs2_local_alloc_seen_free_bits() is totally pointless and impedes IO performance. Link: https://lkml.kernel.org/r/20240328125203.20892-3-heming.zhao@suse.com Signed-off-by: Heming Zhao Reviewed-by: Joseph Qi Cc: Mark Fasheh Cc: Joel Becker Cc: Junxiao Bi Cc: Changwei Ge Cc: Gang He Cc: Jun Piao Signed-off-by: Andrew Morton

ocfs2: improve write IO performance when fragmentation is high

2024-04-26T04:07:03Z

The group_search function ocfs2_cluster_group_search() should bypass groups with insufficient space to avoid unnecessary searches. This patch is particularly useful when ocfs2 is handling huge number small files, and volume fragmentation is very high. In this case, ocfs2 is busy with looking up available la window from //global_bitmap. This patch introduces a new member in the Group Description (gd) struct called 'bg_contig_free_bits', representing the max contigous free bits in this gd. When ocfs2 allocates a new la window from //global_bitmap, 'bg_contig_free_bits' helps expedite the search process. Let's image below path. 1. la state (->local_alloc_state) is set THROTTLED or DISABLED. 2. when user delete a large file and trigger ocfs2_local_alloc_seen_free_bits set osb->local_alloc_state unconditionally. 3. a write IOs thread run and trigger the worst performance path ``` ocfs2_reserve_clusters_with_limit ocfs2_reserve_local_alloc_bits ocfs2_local_alloc_slide_window //[1] + ocfs2_local_alloc_reserve_for_window //[2] + ocfs2_local_alloc_new_window //[3] ocfs2_recalc_la_window ``` [1]: will be called when la window bits used up. [2]: under la state is ENABLED, and this func only check global_bitmap free bits, it will succeed in general. [3]: will use the default la window size to search clusters then fail. ocfs2_recalc_la_window attempts other la window sizes. the timing complexity is O(n^4), resulting in a significant time cost for scanning global bitmap. This leads to a dramatic slowdown in write I/Os (e.g., user space 'dd'). i.e. an ocfs2 partition size: 1.45TB, cluster size: 4KB, la window default size: 106MB. The partition is fragmentation by creating & deleting huge mount of small files. before this patch, the timing of [3] should be (the number got from real world): - la window size change order (size: MB): 106, 53, 26.5, 13, 6.5, 3.25, 1.6, 0.8 only 0.8MB succeed, 0.8MB also triggers la window to disable. ocfs2_local_alloc_new_window retries 8 times, first 7 times totally runs in worst case. - group chain number: 242 ocfs2_claim_suballoc_bits calls for-loop 242 times - each chain has 49 block group ocfs2_search_chain calls while-loop 49 times - each bg has 32256 blocks ocfs2_block_group_find_clear_bits calls while-loop for 32256 bits. for ocfs2_find_next_zero_bit uses ffz() to find zero bit, let's use (32256/64) (this is not worst value) for timing calucation. the loop times: 7*242*49*(32256/64) = 41835024 (~42 million times) In the worst case, user space writes 1MB data will trigger 42M scanning times. under this patch, the timing is '7*242*49 = 83006', reduced by three orders of magnitude. Link: https://lkml.kernel.org/r/20240328125203.20892-2-heming.zhao@suse.com Signed-off-by: Heming Zhao Reviewed-by: Joseph Qi Cc: Changwei Ge Cc: Gang He Cc: Joel Becker Cc: Jun Piao Cc: Junxiao Bi Cc: Mark Fasheh Signed-off-by: Andrew Morton

ocfs2: correctly use ocfs2_find_next_zero_bit()

2024-04-26T04:07:01Z

If no bits are zero, ocfs2_find_next_zero_bit() will return max size, so check the return value with -1 is meaningless. Correct this usage and cleanup the code. Link: https://lkml.kernel.org/r/20240314021713.240796-1-joseph.qi@linux.alibaba.com Signed-off-by: Joseph Qi Reviewed-by: Heming Zhao Cc: Mark Fasheh Cc: Joel Becker Cc: Junxiao Bi Cc: Changwei Ge Cc: Gang He Cc: Jun Piao Signed-off-by: Andrew Morton

ocfs2: fix a deadlock when commit trans

2022-01-30T07:56:58Z

commit 6f1b228529ae introduces a regression which can deadlock as follows: Task1: Task2: jbd2_journal_commit_transaction ocfs2_test_bg_bit_allocatable spin_lock(&jh->b_state_lock) jbd_lock_bh_journal_head __jbd2_journal_remove_checkpoint spin_lock(&jh->b_state_lock) jbd2_journal_put_journal_head jbd_lock_bh_journal_head Task1 and Task2 lock bh->b_state and jh->b_state_lock in different order, which finally result in a deadlock. So use jbd2_journal_[grab|put]_journal_head instead in ocfs2_test_bg_bit_allocatable() to fix it. Link: https://lkml.kernel.org/r/20220121071205.100648-3-joseph.qi@linux.alibaba.com Fixes: 6f1b228529ae ("ocfs2: fix race between searching chunks and release journal_head from buffer_head") Signed-off-by: Joseph Qi Reported-by: Gautham Ananthakrishna Tested-by: Gautham Ananthakrishna Reported-by: Saeed Mirzamohammadi Cc: "Theodore Ts'o" Cc: Andreas Dilger Cc: Changwei Ge Cc: Gang He Cc: Joel Becker Cc: Jun Piao Cc: Junxiao Bi Cc: Mark Fasheh Cc: Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

ocfs2: fix race between searching chunks and release journal_head from buffer_head

2021-10-29T00:18:55Z

Encountered a race between ocfs2_test_bg_bit_allocatable() and jbd2_journal_put_journal_head() resulting in the below vmcore. PID: 106879 TASK: ffff880244ba9c00 CPU: 2 COMMAND: "loop3" Call trace: panic oops_end no_context __bad_area_nosemaphore bad_area_nosemaphore __do_page_fault do_page_fault page_fault [exception RIP: ocfs2_block_group_find_clear_bits+316] ocfs2_block_group_find_clear_bits [ocfs2] ocfs2_cluster_group_search [ocfs2] ocfs2_search_chain [ocfs2] ocfs2_claim_suballoc_bits [ocfs2] __ocfs2_claim_clusters [ocfs2] ocfs2_claim_clusters [ocfs2] ocfs2_local_alloc_slide_window [ocfs2] ocfs2_reserve_local_alloc_bits [ocfs2] ocfs2_reserve_clusters_with_limit [ocfs2] ocfs2_reserve_clusters [ocfs2] ocfs2_lock_refcount_allocators [ocfs2] ocfs2_make_clusters_writable [ocfs2] ocfs2_replace_cow [ocfs2] ocfs2_refcount_cow [ocfs2] ocfs2_file_write_iter [ocfs2] lo_rw_aio loop_queue_work kthread_worker_fn kthread ret_from_fork When ocfs2_test_bg_bit_allocatable() called bh2jh(bg_bh), the bg_bh->b_private NULL as jbd2_journal_put_journal_head() raced and released the jounal head from the buffer head. Needed to take bit lock for the bit 'BH_JournalHead' to fix this race. Link: https://lkml.kernel.org/r/1634820718-6043-1-git-send-email-gautham.ananthakrishna@oracle.com Signed-off-by: Gautham Ananthakrishna Reviewed-by: Joseph Qi Cc: Cc: Mark Fasheh Cc: Joel Becker Cc: Junxiao Bi Cc: Changwei Ge Cc: Gang He Cc: Jun Piao Cc: Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

treewide: remove editor modelines and cruft

2021-05-07T07:26:34Z

The section "19) Editor modelines and other cruft" in Documentation/process/coding-style.rst clearly says, "Do not include any of these in source files." I recently receive a patch to explicitly add a new one. Let's do treewide cleanups, otherwise some people follow the existing code and attempt to upstream their favoriate editor setups. It is even nicer if scripts/checkpatch.pl can check it. If we like to impose coding style in an editor-independent manner, I think editorconfig (patch [1]) is a saner solution. [1] https://lore.kernel.org/lkml/20200703073143.423557-1-danny@kdrag0n.dev/ Link: https://lkml.kernel.org/r/20210324054457.1477489-1-masahiroy@kernel.org Signed-off-by: Masahiro Yamada Acked-by: Geert Uytterhoeven Reviewed-by: Miguel Ojeda [auxdisplay] Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

ocfs2: change slot number type s16 to u16

2020-08-07T18:33:21Z

Dan Carpenter reported the following static checker warning. fs/ocfs2/super.c:1269 ocfs2_parse_options() warn: '(-1)' 65535 can't fit into 32767 'mopt->slot' fs/ocfs2/suballoc.c:859 ocfs2_init_inode_steal_slot() warn: '(-1)' 65535 can't fit into 32767 'osb->s_inode_steal_slot' fs/ocfs2/suballoc.c:867 ocfs2_init_meta_steal_slot() warn: '(-1)' 65535 can't fit into 32767 'osb->s_meta_steal_slot' That's because OCFS2_INVALID_SLOT is (u16)-1. Slot number in ocfs2 can be never negative, so change s16 to u16. Fixes: 9277f8334ffc ("ocfs2: fix value of OCFS2_INVALID_SLOT") Reported-by: Dan Carpenter Signed-off-by: Junxiao Bi Signed-off-by: Andrew Morton Reviewed-by: Joseph Qi Reviewed-by: Gang He Cc: Mark Fasheh Cc: Joel Becker Cc: Junxiao Bi Cc: Changwei Ge Cc: Jun Piao Cc: Link: http://lkml.kernel.org/r/20200627001259.19757-1-junxiao.bi@oracle.com Signed-off-by: Linus Torvalds

ocfs2: fix panic on nfs server over ocfs2

2020-06-26T07:27:37Z

The following kernel panic was captured when running nfs server over ocfs2, at that time ocfs2_test_inode_bit() was checking whether one inode locating at "blkno" 5 was valid, that is ocfs2 root inode, its "suballoc_slot" was OCFS2_INVALID_SLOT(65535) and it was allocted from //global_inode_alloc, but here it wrongly assumed that it was got from per slot inode alloctor which would cause array overflow and trigger kernel panic. BUG: unable to handle kernel paging request at 0000000000001088 IP: [] _raw_spin_lock+0x18/0xf0 PGD 1e06ba067 PUD 1e9e7d067 PMD 0 Oops: 0002 [#1] SMP CPU: 6 PID: 24873 Comm: nfsd Not tainted 4.1.12-124.36.1.el6uek.x86_64 #2 Hardware name: Huawei CH121 V3/IT11SGCA1, BIOS 3.87 02/02/2018 RIP: _raw_spin_lock+0x18/0xf0 RSP: e02b:ffff88005ae97908 EFLAGS: 00010206 RAX: ffff88005ae98000 RBX: 0000000000001088 RCX: 0000000000000000 RDX: 0000000000020000 RSI: 0000000000000009 RDI: 0000000000001088 RBP: ffff88005ae97928 R08: 0000000000000000 R09: ffff880212878e00 R10: 0000000000007ff0 R11: 0000000000000000 R12: 0000000000001088 R13: ffff8800063c0aa8 R14: ffff8800650c27d0 R15: 000000000000ffff FS: 0000000000000000(0000) GS:ffff880218180000(0000) knlGS:ffff880218180000 CS: e033 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 0000000000001088 CR3: 00000002033d0000 CR4: 0000000000042660 Call Trace: igrab+0x1e/0x60 ocfs2_get_system_file_inode+0x63/0x3a0 [ocfs2] ocfs2_test_inode_bit+0x328/0xa00 [ocfs2] ocfs2_get_parent+0xba/0x3e0 [ocfs2] reconnect_path+0xb5/0x300 exportfs_decode_fh+0xf6/0x2b0 fh_verify+0x350/0x660 [nfsd] nfsd4_putfh+0x4d/0x60 [nfsd] nfsd4_proc_compound+0x3d3/0x6f0 [nfsd] nfsd_dispatch+0xe0/0x290 [nfsd] svc_process_common+0x412/0x6a0 [sunrpc] svc_process+0x123/0x210 [sunrpc] nfsd+0xff/0x170 [nfsd] kthread+0xcb/0xf0 ret_from_fork+0x61/0x90 Code: 83 c2 02 0f b7 f2 e8 18 dc 91 ff 66 90 eb bf 0f 1f 40 00 55 48 89 e5 41 56 41 55 41 54 53 0f 1f 44 00 00 48 89 fb ba 00 00 02 00 0f c1 17 89 d0 45 31 e4 45 31 ed c1 e8 10 66 39 d0 41 89 c6 RIP _raw_spin_lock+0x18/0xf0 CR2: 0000000000001088 ---[ end trace 7264463cd1aac8f9 ]--- Kernel panic - not syncing: Fatal exception Link: http://lkml.kernel.org/r/20200616183829.87211-4-junxiao.bi@oracle.com Signed-off-by: Junxiao Bi Reviewed-by: Joseph Qi Cc: Changwei Ge Cc: Gang He Cc: Joel Becker Cc: Jun Piao Cc: Mark Fasheh Cc: Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

ocfs2: there is no need to log twice in several functions

2020-04-02T16:35:25Z

There is no need to log twice in several functions. Signed-off-by: Yan Wang Signed-off-by: Andrew Morton Reviewed-by: Jun Piao Acked-by: Joseph Qi Cc: Mark Fasheh Cc: Joel Becker Cc: Junxiao Bi Cc: Changwei Ge Cc: Gang He Link: http://lkml.kernel.org/r/77eec86a-f634-5b98-4f7d-0cd15185a37b@huawei.com Signed-off-by: Linus Torvalds