From 8238b4579866b7c1bb99883cfe102a43db5506ff Mon Sep 17 00:00:00 2001 From: Mikulas Patocka Date: Fri, 26 Aug 2022 09:17:08 -0400 Subject: wait_on_bit: add an acquire memory barrier There are several places in the kernel where wait_on_bit is not followed by a memory barrier (for example, in drivers/md/dm-bufio.c:new_read). On architectures with weak memory ordering, it may happen that memory accesses that follow wait_on_bit are reordered before wait_on_bit and they may return invalid data. Fix this class of bugs by introducing a new function "test_bit_acquire" that works like test_bit, but has acquire memory ordering semantics. Signed-off-by: Mikulas Patocka Acked-by: Will Deacon Cc: stable@vger.kernel.org Signed-off-by: Linus Torvalds --- include/linux/buffer_head.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'include/linux/buffer_head.h') diff --git a/include/linux/buffer_head.h b/include/linux/buffer_head.h index def8b8d30ccc..089c9ade4325 100644 --- a/include/linux/buffer_head.h +++ b/include/linux/buffer_head.h @@ -156,7 +156,7 @@ static __always_inline int buffer_uptodate(const struct buffer_head *bh) * make it consistent with folio_test_uptodate * pairs with smp_mb__before_atomic in set_buffer_uptodate */ - return (smp_load_acquire(&bh->b_state) & (1UL << BH_Uptodate)) != 0; + return test_bit_acquire(BH_Uptodate, &bh->b_state); } #define bh_offset(bh) ((unsigned long)(bh)->b_data & ~PAGE_MASK) -- cgit v1.2.3 From 2f79cdfe58c13949bbbb65ba5926abfe9561d0ec Mon Sep 17 00:00:00 2001 From: Linus Torvalds Date: Wed, 31 Aug 2022 09:46:12 -0700 Subject: fs: only do a memory barrier for the first set_buffer_uptodate() Commit d4252071b97d ("add barriers to buffer_uptodate and set_buffer_uptodate") added proper memory barriers to the buffer head BH_Uptodate bit, so that anybody who tests a buffer for being up-to-date will be guaranteed to actually see initialized state. However, that commit didn't _just_ add the memory barrier, it also ended up dropping the "was it already set" logic that the BUFFER_FNS() macro had. That's conceptually the right thing for a generic "this is a memory barrier" operation, but in the case of the buffer contents, we really only care about the memory barrier for the _first_ time we set the bit, in that the only memory ordering protection we need is to avoid anybody seeing uninitialized memory contents. Any other access ordering wouldn't be about the BH_Uptodate bit anyway, and would require some other proper lock (typically BH_Lock or the folio lock). A reader that races with somebody invalidating the buffer head isn't an issue wrt the memory ordering, it's a serialization issue. Now, you'd think that the buffer head operations don't matter in this day and age (and I certainly thought so), but apparently some loads still end up being heavy users of buffer heads. In particular, the kernel test robot reported that not having this bit access optimization in place caused a noticeable direct IO performance regression on ext4: fxmark.ssd_ext4_no_jnl_DWTL_54_directio.works/sec -26.5% regression although you presumably need a fast disk and a lot of cores to actually notice. Link: https://lore.kernel.org/all/Yw8L7HTZ%2FdE2%2Fo9C@xsang-OptiPlex-9020/ Reported-by: kernel test robot Tested-by: Fengwei Yin Cc: Mikulas Patocka Cc: Matthew Wilcox (Oracle) Cc: stable@kernel.org Signed-off-by: Linus Torvalds --- include/linux/buffer_head.h | 11 +++++++++++ 1 file changed, 11 insertions(+) (limited to 'include/linux/buffer_head.h') diff --git a/include/linux/buffer_head.h b/include/linux/buffer_head.h index 089c9ade4325..df518c429667 100644 --- a/include/linux/buffer_head.h +++ b/include/linux/buffer_head.h @@ -137,6 +137,17 @@ BUFFER_FNS(Defer_Completion, defer_completion) static __always_inline void set_buffer_uptodate(struct buffer_head *bh) { + /* + * If somebody else already set this uptodate, they will + * have done the memory barrier, and a reader will thus + * see *some* valid buffer state. + * + * Any other serialization (with IO errors or whatever that + * might clear the bit) has to come from other state (eg BH_Lock). + */ + if (test_bit(BH_Uptodate, &bh->b_state)) + return; + /* * make it consistent with folio_mark_uptodate * pairs with smp_load_acquire in buffer_uptodate -- cgit v1.2.3 From 214f8796907b8015b778badf4710a4701472779a Mon Sep 17 00:00:00 2001 From: Zhang Yi Date: Thu, 1 Sep 2022 21:34:52 +0800 Subject: fs/buffer: remove __breadahead_gfp() Patch series "fs/buffer: remove ll_rw_block()", v2. ll_rw_block() will skip locked buffer before submitting IO, it assumes that locked buffer means it is under IO. This assumption is not always true because we cannot guarantee every buffer lock path would submit IO. After commit 88dbcbb3a484 ("blkdev: avoid migration stalls for blkdev pages"), buffer_migrate_folio_norefs() becomes one exceptional case, and there may be others. So ll_rw_block() is not safe on the sync read path, we could get false positive EIO return value when filesystem reading metadata. It seems that it could be only used on the readahead path. Unfortunately, many filesystem misuse the ll_rw_block() on the sync read path. This patch set just remove ll_rw_block() and add new friendly helpers, which could prevent false positive EIO on the read metadata path. Thanks for the suggestion from Jan, the original discussion is at [1]. patch 1: remove unused helpers in fs/buffer.c patch 2: add new bh_read_[*] helpers patch 3-11: remove all ll_rw_block() calls in filesystems patch 12-14: do some leftover cleanups. [1]. https://lore.kernel.org/linux-mm/20220825080146.2021641-1-chengzhihao1@huawei.com/ This patch (of 14): No one use __breadahead_gfp() and sb_breadahead_unmovable() any more, remove them. Link: https://lkml.kernel.org/r/20220901133505.2510834-1-yi.zhang@huawei.com Link: https://lkml.kernel.org/r/20220901133505.2510834-2-yi.zhang@huawei.com Signed-off-by: Zhang Yi Reviewed-by: Jan Kara Reviewed-by: Christoph Hellwig Cc: Alexander Viro Cc: Andreas Gruenbacher Cc: Bob Peterson Cc: Evgeniy Dushistov Cc: Heming Zhao Cc: Jens Axboe Cc: Konstantin Komarov Cc: Mark Fasheh Cc: Theodore Ts'o Cc: Yu Kuai Cc: Zhihao Cheng Signed-off-by: Andrew Morton --- include/linux/buffer_head.h | 8 -------- 1 file changed, 8 deletions(-) (limited to 'include/linux/buffer_head.h') diff --git a/include/linux/buffer_head.h b/include/linux/buffer_head.h index 089c9ade4325..c3863c417b00 100644 --- a/include/linux/buffer_head.h +++ b/include/linux/buffer_head.h @@ -214,8 +214,6 @@ struct buffer_head *__getblk_gfp(struct block_device *bdev, sector_t block, void __brelse(struct buffer_head *); void __bforget(struct buffer_head *); void __breadahead(struct block_device *, sector_t block, unsigned int size); -void __breadahead_gfp(struct block_device *, sector_t block, unsigned int size, - gfp_t gfp); struct buffer_head *__bread_gfp(struct block_device *, sector_t block, unsigned size, gfp_t gfp); void invalidate_bh_lrus(void); @@ -340,12 +338,6 @@ sb_breadahead(struct super_block *sb, sector_t block) __breadahead(sb->s_bdev, block, sb->s_blocksize); } -static inline void -sb_breadahead_unmovable(struct super_block *sb, sector_t block) -{ - __breadahead_gfp(sb->s_bdev, block, sb->s_blocksize, 0); -} - static inline struct buffer_head * sb_getblk(struct super_block *sb, sector_t block) { -- cgit v1.2.3 From fdee117ee86479fd2644bcd9ac2b2469e55722d1 Mon Sep 17 00:00:00 2001 From: Zhang Yi Date: Thu, 1 Sep 2022 21:34:53 +0800 Subject: fs/buffer: add some new buffer read helpers Current ll_rw_block() helper is fragile because it assumes that locked buffer means it's under IO which is submitted by some other who holds the lock, it skip buffer if it failed to get the lock, so it's only safe on the readahead path. Unfortunately, now that most filesystems still use this helper mistakenly on the sync metadata read path. There is no guarantee that the one who holds the buffer lock always submit IO (e.g. buffer_migrate_folio_norefs() after commit 88dbcbb3a484 ("blkdev: avoid migration stalls for blkdev pages"), it could lead to false positive -EIO when submitting reading IO. This patch add some friendly buffer read helpers to prepare replacing ll_rw_block() and similar calls. We can only call bh_readahead_[] helpers for the readahead paths. Link: https://lkml.kernel.org/r/20220901133505.2510834-3-yi.zhang@huawei.com Signed-off-by: Zhang Yi Reviewed-by: Jan Kara Reviewed-by: Christoph Hellwig Signed-off-by: Andrew Morton --- include/linux/buffer_head.h | 38 ++++++++++++++++++++++++++++++++++++++ 1 file changed, 38 insertions(+) (limited to 'include/linux/buffer_head.h') diff --git a/include/linux/buffer_head.h b/include/linux/buffer_head.h index c3863c417b00..6d09785bed9f 100644 --- a/include/linux/buffer_head.h +++ b/include/linux/buffer_head.h @@ -232,6 +232,9 @@ void write_boundary_block(struct block_device *bdev, sector_t bblock, unsigned blocksize); int bh_uptodate_or_lock(struct buffer_head *bh); int bh_submit_read(struct buffer_head *bh); +int __bh_read(struct buffer_head *bh, blk_opf_t op_flags, bool wait); +void __bh_read_batch(int nr, struct buffer_head *bhs[], + blk_opf_t op_flags, bool force_lock); extern int buffer_heads_over_limit; @@ -399,6 +402,41 @@ static inline struct buffer_head *__getblk(struct block_device *bdev, return __getblk_gfp(bdev, block, size, __GFP_MOVABLE); } +static inline void bh_readahead(struct buffer_head *bh, blk_opf_t op_flags) +{ + if (!buffer_uptodate(bh) && trylock_buffer(bh)) { + if (!buffer_uptodate(bh)) + __bh_read(bh, op_flags, false); + else + unlock_buffer(bh); + } +} + +static inline void bh_read_nowait(struct buffer_head *bh, blk_opf_t op_flags) +{ + if (!bh_uptodate_or_lock(bh)) + __bh_read(bh, op_flags, false); +} + +/* Returns 1 if buffer uptodated, 0 on success, and -EIO on error. */ +static inline int bh_read(struct buffer_head *bh, blk_opf_t op_flags) +{ + if (bh_uptodate_or_lock(bh)) + return 1; + return __bh_read(bh, op_flags, true); +} + +static inline void bh_read_batch(int nr, struct buffer_head *bhs[]) +{ + __bh_read_batch(nr, bhs, 0, true); +} + +static inline void bh_readahead_batch(int nr, struct buffer_head *bhs[], + blk_opf_t op_flags) +{ + __bh_read_batch(nr, bhs, op_flags, false); +} + /** * __bread() - reads a specified block and returns the bh * @bdev: the block_device to read from -- cgit v1.2.3 From 79f5978420691fb84593f98557ea56f8b32228f2 Mon Sep 17 00:00:00 2001 From: Zhang Yi Date: Thu, 1 Sep 2022 21:35:03 +0800 Subject: fs/buffer: remove ll_rw_block() helper Now that all ll_rw_block() users has been replaced to new safe helpers, we just remove it here. Link: https://lkml.kernel.org/r/20220901133505.2510834-13-yi.zhang@huawei.com Signed-off-by: Zhang Yi Reviewed-by: Jan Kara Reviewed-by: Christoph Hellwig Signed-off-by: Andrew Morton --- include/linux/buffer_head.h | 1 - 1 file changed, 1 deletion(-) (limited to 'include/linux/buffer_head.h') diff --git a/include/linux/buffer_head.h b/include/linux/buffer_head.h index 6d09785bed9f..b415d8bc2a09 100644 --- a/include/linux/buffer_head.h +++ b/include/linux/buffer_head.h @@ -223,7 +223,6 @@ struct buffer_head *alloc_buffer_head(gfp_t gfp_flags); void free_buffer_head(struct buffer_head * bh); void unlock_buffer(struct buffer_head *bh); void __lock_buffer(struct buffer_head *bh); -void ll_rw_block(blk_opf_t, int, struct buffer_head * bh[]); int sync_dirty_buffer(struct buffer_head *bh); int __sync_dirty_buffer(struct buffer_head *bh, blk_opf_t op_flags); void write_dirty_buffer(struct buffer_head *bh, blk_opf_t op_flags); -- cgit v1.2.3 From 454552d0145486cf82572e73cca266ab6a56e86b Mon Sep 17 00:00:00 2001 From: Zhang Yi Date: Thu, 1 Sep 2022 21:35:05 +0800 Subject: fs/buffer: remove bh_submit_read() helper bh_submit_read() has no user anymore, just remove it. Link: https://lkml.kernel.org/r/20220901133505.2510834-15-yi.zhang@huawei.com Signed-off-by: Zhang Yi Reviewed-by: Jan Kara Reviewed-by: Christoph Hellwig Signed-off-by: Andrew Morton --- include/linux/buffer_head.h | 1 - 1 file changed, 1 deletion(-) (limited to 'include/linux/buffer_head.h') diff --git a/include/linux/buffer_head.h b/include/linux/buffer_head.h index b415d8bc2a09..9b6556d3f110 100644 --- a/include/linux/buffer_head.h +++ b/include/linux/buffer_head.h @@ -230,7 +230,6 @@ int submit_bh(blk_opf_t, struct buffer_head *); void write_boundary_block(struct block_device *bdev, sector_t bblock, unsigned blocksize); int bh_uptodate_or_lock(struct buffer_head *bh); -int bh_submit_read(struct buffer_head *bh); int __bh_read(struct buffer_head *bh, blk_opf_t op_flags, bool wait); void __bh_read_batch(int nr, struct buffer_head *bhs[], blk_opf_t op_flags, bool force_lock); -- cgit v1.2.3 From 5bdf402a05fafc55c876b9993b4c948bb2bc2361 Mon Sep 17 00:00:00 2001 From: "Ritesh Harjani (IBM)" Date: Thu, 18 Aug 2022 10:34:40 +0530 Subject: fs/buffer: make submit_bh & submit_bh_wbc return type as void submit_bh/submit_bh_wbc are non-blocking functions which just submit the bio and return. The caller of submit_bh/submit_bh_wbc needs to wait on buffer till I/O completion and then check buffer head's b_state field to know if there was any I/O error. Hence there is no need for these functions to have any return type. Even now they always returns 0. Hence drop the return value and make their return type as void to avoid any confusion. Reviewed-by: Christoph Hellwig Reviewed-by: Jan Kara Signed-off-by: Ritesh Harjani (IBM) Link: https://lore.kernel.org/r/cb66ef823374cdd94d2d03083ce13de844fffd41.1660788334.git.ritesh.list@gmail.com Signed-off-by: Theodore Ts'o --- include/linux/buffer_head.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'include/linux/buffer_head.h') diff --git a/include/linux/buffer_head.h b/include/linux/buffer_head.h index 089c9ade4325..dcc0e90d8979 100644 --- a/include/linux/buffer_head.h +++ b/include/linux/buffer_head.h @@ -229,7 +229,7 @@ void ll_rw_block(blk_opf_t, int, struct buffer_head * bh[]); int sync_dirty_buffer(struct buffer_head *bh); int __sync_dirty_buffer(struct buffer_head *bh, blk_opf_t op_flags); void write_dirty_buffer(struct buffer_head *bh, blk_opf_t op_flags); -int submit_bh(blk_opf_t, struct buffer_head *); +void submit_bh(blk_opf_t, struct buffer_head *); void write_boundary_block(struct block_device *bdev, sector_t bblock, unsigned blocksize); int bh_uptodate_or_lock(struct buffer_head *bh); -- cgit v1.2.3