From 23d79d81b13431756fee428acde49c9b770da132 Mon Sep 17 00:00:00 2001 From: David Sterba Date: Sun, 15 Jun 2014 02:55:29 +0200 Subject: btrfs: use GFP_NOFS in __alloc_extent_buffer directly Same mask from all callers. Signed-off-by: David Sterba --- fs/btrfs/extent_io.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) (limited to 'fs/btrfs/extent_io.c') diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c index 4ebabd2..619592d 100644 --- a/fs/btrfs/extent_io.c +++ b/fs/btrfs/extent_io.c @@ -4598,11 +4598,11 @@ static inline void btrfs_release_extent_buffer(struct extent_buffer *eb) static struct extent_buffer * __alloc_extent_buffer(struct btrfs_fs_info *fs_info, u64 start, - unsigned long len, gfp_t mask) + unsigned long len) { struct extent_buffer *eb = NULL; - eb = kmem_cache_zalloc(extent_buffer_cache, mask); + eb = kmem_cache_zalloc(extent_buffer_cache, GFP_NOFS); if (eb == NULL) return NULL; eb->start = start; @@ -4643,7 +4643,7 @@ struct extent_buffer *btrfs_clone_extent_buffer(struct extent_buffer *src) struct extent_buffer *new; unsigned long num_pages = num_extent_pages(src->start, src->len); - new = __alloc_extent_buffer(NULL, src->start, src->len, GFP_NOFS); + new = __alloc_extent_buffer(NULL, src->start, src->len); if (new == NULL) return NULL; @@ -4672,7 +4672,7 @@ struct extent_buffer *alloc_dummy_extent_buffer(u64 start, unsigned long len) unsigned long num_pages = num_extent_pages(0, len); unsigned long i; - eb = __alloc_extent_buffer(NULL, start, len, GFP_NOFS); + eb = __alloc_extent_buffer(NULL, start, len); if (!eb) return NULL; @@ -4824,7 +4824,7 @@ struct extent_buffer *alloc_extent_buffer(struct btrfs_fs_info *fs_info, if (eb) return eb; - eb = __alloc_extent_buffer(fs_info, start, len, GFP_NOFS); + eb = __alloc_extent_buffer(fs_info, start, len); if (!eb) return NULL; -- cgit v1.1 From 3f556f7853ec4845a7c219d026cbcdf4cfa8cea7 Mon Sep 17 00:00:00 2001 From: David Sterba Date: Sun, 15 Jun 2014 03:20:26 +0200 Subject: btrfs: unify extent buffer allocation api Make the extent buffer allocation interface consistent. Cloned eb will set a valid fs_info. For dummy eb, we can drop the length parameter and set it from fs_info. The built-in sanity checks may pass a NULL fs_info that's queried for nodesize, but we know it's 4096. Signed-off-by: David Sterba --- fs/btrfs/extent_io.c | 23 ++++++++++++++++++----- 1 file changed, 18 insertions(+), 5 deletions(-) (limited to 'fs/btrfs/extent_io.c') diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c index 619592d..dc424e3 100644 --- a/fs/btrfs/extent_io.c +++ b/fs/btrfs/extent_io.c @@ -4643,7 +4643,7 @@ struct extent_buffer *btrfs_clone_extent_buffer(struct extent_buffer *src) struct extent_buffer *new; unsigned long num_pages = num_extent_pages(src->start, src->len); - new = __alloc_extent_buffer(NULL, src->start, src->len); + new = __alloc_extent_buffer(src->fs_info, src->start, src->len); if (new == NULL) return NULL; @@ -4666,13 +4666,26 @@ struct extent_buffer *btrfs_clone_extent_buffer(struct extent_buffer *src) return new; } -struct extent_buffer *alloc_dummy_extent_buffer(u64 start, unsigned long len) +struct extent_buffer *alloc_dummy_extent_buffer(struct btrfs_fs_info *fs_info, + u64 start) { struct extent_buffer *eb; - unsigned long num_pages = num_extent_pages(0, len); + unsigned long len; + unsigned long num_pages; unsigned long i; - eb = __alloc_extent_buffer(NULL, start, len); + if (!fs_info) { + /* + * Called only from tests that don't always have a fs_info + * available, but we know that nodesize is 4096 + */ + len = 4096; + } else { + len = fs_info->tree_root->nodesize; + } + num_pages = num_extent_pages(0, len); + + eb = __alloc_extent_buffer(fs_info, start, len); if (!eb) return NULL; @@ -4770,7 +4783,7 @@ struct extent_buffer *alloc_test_extent_buffer(struct btrfs_fs_info *fs_info, eb = find_extent_buffer(fs_info, start); if (eb) return eb; - eb = alloc_dummy_extent_buffer(start, len); + eb = alloc_dummy_extent_buffer(fs_info, start); if (!eb) return NULL; eb->fs_info = fs_info; -- cgit v1.1 From ce3e69847e3ec79a38421bfd3d6f554d5e481231 Mon Sep 17 00:00:00 2001 From: David Sterba Date: Sun, 15 Jun 2014 03:00:04 +0200 Subject: btrfs: sink parameter len to alloc_extent_buffer Because we're using globally known nodesize. Do the same for the sanity test function variant. Signed-off-by: David Sterba --- fs/btrfs/extent_io.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'fs/btrfs/extent_io.c') diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c index dc424e3..c4ca90a 100644 --- a/fs/btrfs/extent_io.c +++ b/fs/btrfs/extent_io.c @@ -4775,7 +4775,7 @@ struct extent_buffer *find_extent_buffer(struct btrfs_fs_info *fs_info, #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS struct extent_buffer *alloc_test_extent_buffer(struct btrfs_fs_info *fs_info, - u64 start, unsigned long len) + u64 start) { struct extent_buffer *eb, *exists = NULL; int ret; @@ -4821,8 +4821,9 @@ free_eb: #endif struct extent_buffer *alloc_extent_buffer(struct btrfs_fs_info *fs_info, - u64 start, unsigned long len) + u64 start) { + unsigned long len = fs_info->tree_root->nodesize; unsigned long num_pages = num_extent_pages(start, len); unsigned long i; unsigned long index = start >> PAGE_CACHE_SHIFT; -- cgit v1.1 From 9ee49a047dc53fd21808cbb7f3b6a3345463e834 Mon Sep 17 00:00:00 2001 From: David Sterba Date: Wed, 14 Jan 2015 19:52:13 +0100 Subject: btrfs: switch extent_state state to unsigned Currently there's a 4B hole in the structure between refs and state and there are only 16 bits used so we can make it unsigned. This will get a better packing and may save some stack space for local variables. The size of extent_state gets reduced by 8B and there are usually a lot of slab objects. struct extent_state { u64 start; /* 0 8 */ u64 end; /* 8 8 */ struct rb_node rb_node; /* 16 24 */ wait_queue_head_t wq; /* 40 24 */ /* --- cacheline 1 boundary (64 bytes) --- */ atomic_t refs; /* 64 4 */ /* XXX 4 bytes hole, try to pack */ long unsigned int state; /* 72 8 */ u64 private; /* 80 8 */ /* size: 88, cachelines: 2, members: 7 */ /* sum members: 84, holes: 1, sum holes: 4 */ /* last cacheline: 24 bytes */ }; Signed-off-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/extent_io.c | 45 +++++++++++++++++++++++---------------------- 1 file changed, 23 insertions(+), 22 deletions(-) (limited to 'fs/btrfs/extent_io.c') diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c index c4ca90a..65bd285 100644 --- a/fs/btrfs/extent_io.c +++ b/fs/btrfs/extent_io.c @@ -64,7 +64,7 @@ void btrfs_leak_debug_check(void) while (!list_empty(&states)) { state = list_entry(states.next, struct extent_state, leak_list); - pr_err("BTRFS: state leak: start %llu end %llu state %lu in tree %d refs %d\n", + pr_err("BTRFS: state leak: start %llu end %llu state %u in tree %d refs %d\n", state->start, state->end, state->state, extent_state_in_tree(state), atomic_read(&state->refs)); @@ -396,21 +396,21 @@ static void merge_state(struct extent_io_tree *tree, } static void set_state_cb(struct extent_io_tree *tree, - struct extent_state *state, unsigned long *bits) + struct extent_state *state, unsigned *bits) { if (tree->ops && tree->ops->set_bit_hook) tree->ops->set_bit_hook(tree->mapping->host, state, bits); } static void clear_state_cb(struct extent_io_tree *tree, - struct extent_state *state, unsigned long *bits) + struct extent_state *state, unsigned *bits) { if (tree->ops && tree->ops->clear_bit_hook) tree->ops->clear_bit_hook(tree->mapping->host, state, bits); } static void set_state_bits(struct extent_io_tree *tree, - struct extent_state *state, unsigned long *bits); + struct extent_state *state, unsigned *bits); /* * insert an extent_state struct into the tree. 'bits' are set on the @@ -426,7 +426,7 @@ static int insert_state(struct extent_io_tree *tree, struct extent_state *state, u64 start, u64 end, struct rb_node ***p, struct rb_node **parent, - unsigned long *bits) + unsigned *bits) { struct rb_node *node; @@ -511,10 +511,10 @@ static struct extent_state *next_state(struct extent_state *state) */ static struct extent_state *clear_state_bit(struct extent_io_tree *tree, struct extent_state *state, - unsigned long *bits, int wake) + unsigned *bits, int wake) { struct extent_state *next; - unsigned long bits_to_clear = *bits & ~EXTENT_CTLBITS; + unsigned bits_to_clear = *bits & ~EXTENT_CTLBITS; if ((bits_to_clear & EXTENT_DIRTY) && (state->state & EXTENT_DIRTY)) { u64 range = state->end - state->start + 1; @@ -570,7 +570,7 @@ static void extent_io_tree_panic(struct extent_io_tree *tree, int err) * This takes the tree lock, and returns 0 on success and < 0 on error. */ int clear_extent_bit(struct extent_io_tree *tree, u64 start, u64 end, - unsigned long bits, int wake, int delete, + unsigned bits, int wake, int delete, struct extent_state **cached_state, gfp_t mask) { @@ -789,9 +789,9 @@ out: static void set_state_bits(struct extent_io_tree *tree, struct extent_state *state, - unsigned long *bits) + unsigned *bits) { - unsigned long bits_to_set = *bits & ~EXTENT_CTLBITS; + unsigned bits_to_set = *bits & ~EXTENT_CTLBITS; set_state_cb(tree, state, bits); if ((bits_to_set & EXTENT_DIRTY) && !(state->state & EXTENT_DIRTY)) { @@ -803,7 +803,7 @@ static void set_state_bits(struct extent_io_tree *tree, static void cache_state_if_flags(struct extent_state *state, struct extent_state **cached_ptr, - const u64 flags) + unsigned flags) { if (cached_ptr && !(*cached_ptr)) { if (!flags || (state->state & flags)) { @@ -833,7 +833,7 @@ static void cache_state(struct extent_state *state, static int __must_check __set_extent_bit(struct extent_io_tree *tree, u64 start, u64 end, - unsigned long bits, unsigned long exclusive_bits, + unsigned bits, unsigned exclusive_bits, u64 *failed_start, struct extent_state **cached_state, gfp_t mask) { @@ -1034,7 +1034,7 @@ search_again: } int set_extent_bit(struct extent_io_tree *tree, u64 start, u64 end, - unsigned long bits, u64 * failed_start, + unsigned bits, u64 * failed_start, struct extent_state **cached_state, gfp_t mask) { return __set_extent_bit(tree, start, end, bits, 0, failed_start, @@ -1060,7 +1060,7 @@ int set_extent_bit(struct extent_io_tree *tree, u64 start, u64 end, * boundary bits like LOCK. */ int convert_extent_bit(struct extent_io_tree *tree, u64 start, u64 end, - unsigned long bits, unsigned long clear_bits, + unsigned bits, unsigned clear_bits, struct extent_state **cached_state, gfp_t mask) { struct extent_state *state; @@ -1268,14 +1268,14 @@ int set_extent_dirty(struct extent_io_tree *tree, u64 start, u64 end, } int set_extent_bits(struct extent_io_tree *tree, u64 start, u64 end, - unsigned long bits, gfp_t mask) + unsigned bits, gfp_t mask) { return set_extent_bit(tree, start, end, bits, NULL, NULL, mask); } int clear_extent_bits(struct extent_io_tree *tree, u64 start, u64 end, - unsigned long bits, gfp_t mask) + unsigned bits, gfp_t mask) { return clear_extent_bit(tree, start, end, bits, 0, 0, NULL, mask); } @@ -1330,10 +1330,11 @@ int clear_extent_uptodate(struct extent_io_tree *tree, u64 start, u64 end, * us if waiting is desired. */ int lock_extent_bits(struct extent_io_tree *tree, u64 start, u64 end, - unsigned long bits, struct extent_state **cached_state) + unsigned bits, struct extent_state **cached_state) { int err; u64 failed_start; + while (1) { err = __set_extent_bit(tree, start, end, EXTENT_LOCKED | bits, EXTENT_LOCKED, &failed_start, @@ -1440,7 +1441,7 @@ static int set_range_writeback(struct extent_io_tree *tree, u64 start, u64 end) */ static struct extent_state * find_first_extent_bit_state(struct extent_io_tree *tree, - u64 start, unsigned long bits) + u64 start, unsigned bits) { struct rb_node *node; struct extent_state *state; @@ -1474,7 +1475,7 @@ out: * If nothing was found, 1 is returned. If found something, return 0. */ int find_first_extent_bit(struct extent_io_tree *tree, u64 start, - u64 *start_ret, u64 *end_ret, unsigned long bits, + u64 *start_ret, u64 *end_ret, unsigned bits, struct extent_state **cached_state) { struct extent_state *state; @@ -1753,7 +1754,7 @@ out_failed: int extent_clear_unlock_delalloc(struct inode *inode, u64 start, u64 end, struct page *locked_page, - unsigned long clear_bits, + unsigned clear_bits, unsigned long page_ops) { struct extent_io_tree *tree = &BTRFS_I(inode)->io_tree; @@ -1810,7 +1811,7 @@ int extent_clear_unlock_delalloc(struct inode *inode, u64 start, u64 end, */ u64 count_range_bits(struct extent_io_tree *tree, u64 *start, u64 search_end, u64 max_bytes, - unsigned long bits, int contig) + unsigned bits, int contig) { struct rb_node *node; struct extent_state *state; @@ -1928,7 +1929,7 @@ out: * range is found set. */ int test_range_bit(struct extent_io_tree *tree, u64 start, u64 end, - unsigned long bits, int filled, struct extent_state *cached) + unsigned bits, int filled, struct extent_state *cached) { struct extent_state *state = NULL; struct rb_node *node; -- cgit v1.1 From 6e9606d2a2dce098c1739fb3cd82a1c34fd73d3a Mon Sep 17 00:00:00 2001 From: Zhao Lei Date: Tue, 20 Jan 2015 15:11:34 +0800 Subject: Btrfs: add ref_count and free function for btrfs_bio 1: ref_count is simple than current RBIO_HOLD_BBIO_MAP_BIT flag to keep btrfs_bio's memory in raid56 recovery implement. 2: free function for bbio will make code clean and flexible, plus forced data type checking in compile. Changelog v1->v2: Rename following by David Sterba's suggestion: put_btrfs_bio() -> btrfs_put_bio() get_btrfs_bio() -> btrfs_get_bio() bbio->ref_count -> bbio->refs Signed-off-by: Zhao Lei Signed-off-by: Miao Xie Signed-off-by: Chris Mason --- fs/btrfs/extent_io.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs/btrfs/extent_io.c') diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c index 65bd285..dab8af4 100644 --- a/fs/btrfs/extent_io.c +++ b/fs/btrfs/extent_io.c @@ -2058,7 +2058,7 @@ int repair_io_failure(struct inode *inode, u64 start, u64 length, u64 logical, sector = bbio->stripes[mirror_num-1].physical >> 9; bio->bi_iter.bi_sector = sector; dev = bbio->stripes[mirror_num-1].dev; - kfree(bbio); + btrfs_put_bbio(bbio); if (!dev || !dev->bdev || !dev->writeable) { bio_put(bio); return -EIO; -- cgit v1.1 From 289454ad26a2d752e04b07234a175feda9ec0f4e Mon Sep 17 00:00:00 2001 From: Naohiro Aota Date: Tue, 6 Jan 2015 01:01:03 +0900 Subject: btrfs: clear bio reference after submit_one_bio() After submit_one_bio(), `bio' can go away. However submit_extent_page() leave `bio' referable if submit_one_bio() failed (e.g. -ENOMEM on OOM). It will cause invalid paging request when submit_extent_page() is called next time. I reproduced ENOMEM case with the following script (need CONFIG_FAIL_PAGE_ALLOC, and CONFIG_FAULT_INJECTION_DEBUG_FS). #!/bin/bash dmesgout=dmesg.txt start=100000 end=300000 step=1000 # btrfs options device=/dev/vdb1 directory=/mnt/btrfs # fault-injection options percent=100 times=3 mkdir -p $directory || exit 1 mount -o compress $device $directory || exit 1 rm -f $directory/file || exit 1 dd if=/dev/zero of=$directory/file bs=1M count=512 || exit 1 for interval in `seq $start $step $end`; do dmesg -C echo 1 > /proc/sys/vm/drop_caches sync export FAILCMD_TYPE=fail_page_alloc ./failcmd.sh -p $percent -t $times -i $interval \ --ignore-gfp-highmem=N --ignore-gfp-wait=N --min-order=0 \ -- \ cat $directory/file > /dev/null dmesg > ${dmesgout} if grep -q BUG: ${dmesgout}; then cat ${dmesgout} exit 1 fi done umount $directory exit 0 Signed-off-by: Naohiro Aota Tested-by: Satoru Takeuchi Signed-off-by: Chris Mason --- fs/btrfs/extent_io.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'fs/btrfs/extent_io.c') diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c index dab8af4..a7f6600 100644 --- a/fs/btrfs/extent_io.c +++ b/fs/btrfs/extent_io.c @@ -2817,8 +2817,10 @@ static int submit_extent_page(int rw, struct extent_io_tree *tree, bio_add_page(bio, page, page_size, offset) < page_size) { ret = submit_one_bio(rw, bio, mirror_num, prev_bio_flags); - if (ret < 0) + if (ret < 0) { + *bio_ret = NULL; return ret; + } bio = NULL; } else { return 0; -- cgit v1.1 From dcab6a3b2ae657a2017637083c28ee303b6b1b8e Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Wed, 11 Feb 2015 15:08:59 -0500 Subject: Btrfs: account for large extents with enospc On our gluster boxes we stream large tar balls of backups onto our fses. With 160gb of ram this means we get really large contiguous ranges of dirty data, but the way our ENOSPC stuff works is that as long as it's contiguous we only hold metadata reservation for one extent. The problem is we limit our extents to 128mb, so we'll end up with at least 800 extents so our enospc accounting is quite a bit lower than what we need. To keep track of this make sure we increase outstanding_extents for every multiple of the max extent size so we can be sure to have enough reserved metadata space. Thanks, Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/extent_io.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs/btrfs/extent_io.c') diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c index a7f6600..29850d4 100644 --- a/fs/btrfs/extent_io.c +++ b/fs/btrfs/extent_io.c @@ -3242,7 +3242,7 @@ static noinline_for_stack int writepage_delalloc(struct inode *inode, page, &delalloc_start, &delalloc_end, - 128 * 1024 * 1024); + BTRFS_MAX_EXTENT_SIZE); if (nr_delalloc == 0) { delalloc_start = delalloc_end + 1; continue; -- cgit v1.1