From 3cae210fa529d69cb25c2a3c491f29dab687b245 Mon Sep 17 00:00:00 2001 From: Qu Wenruo Date: Tue, 16 Jul 2013 11:19:18 +0800 Subject: btrfs: Cleanup for using BTRFS_SETGET_STACK instead of raw convert Some codes still use the cpu_to_lexx instead of the BTRFS_SETGET_STACK_FUNCS declared in ctree.h. Also added some BTRFS_SETGET_STACK_FUNCS for btrfs_header btrfs_timespec and other structures. Signed-off-by: Qu Wenruo Reviewed-by: Miao Xie Reviewed-by: David Sterba Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/scrub.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) (limited to 'fs/btrfs/scrub.c') diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c index 64a157becbe5..266f24febce3 100644 --- a/fs/btrfs/scrub.c +++ b/fs/btrfs/scrub.c @@ -1345,12 +1345,12 @@ static void scrub_recheck_block_checksum(struct btrfs_fs_info *fs_info, mapped_buffer = kmap_atomic(sblock->pagev[0]->page); h = (struct btrfs_header *)mapped_buffer; - if (sblock->pagev[0]->logical != le64_to_cpu(h->bytenr) || + if (sblock->pagev[0]->logical != btrfs_stack_header_bytenr(h) || memcmp(h->fsid, fs_info->fsid, BTRFS_UUID_SIZE) || memcmp(h->chunk_tree_uuid, fs_info->chunk_tree_uuid, BTRFS_UUID_SIZE)) { sblock->header_error = 1; - } else if (generation != le64_to_cpu(h->generation)) { + } else if (generation != btrfs_stack_header_generation(h)) { sblock->header_error = 1; sblock->generation_error = 1; } @@ -1720,10 +1720,10 @@ static int scrub_checksum_tree_block(struct scrub_block *sblock) * b) the page is already kmapped */ - if (sblock->pagev[0]->logical != le64_to_cpu(h->bytenr)) + if (sblock->pagev[0]->logical != btrfs_stack_header_bytenr(h)) ++fail; - if (sblock->pagev[0]->generation != le64_to_cpu(h->generation)) + if (sblock->pagev[0]->generation != btrfs_stack_header_generation(h)) ++fail; if (memcmp(h->fsid, fs_info->fsid, BTRFS_UUID_SIZE)) @@ -1786,10 +1786,10 @@ static int scrub_checksum_super(struct scrub_block *sblock) s = (struct btrfs_super_block *)mapped_buffer; memcpy(on_disk_csum, s->csum, sctx->csum_size); - if (sblock->pagev[0]->logical != le64_to_cpu(s->bytenr)) + if (sblock->pagev[0]->logical != btrfs_super_bytenr(s)) ++fail_cor; - if (sblock->pagev[0]->generation != le64_to_cpu(s->generation)) + if (sblock->pagev[0]->generation != btrfs_super_generation(s)) ++fail_gen; if (memcmp(s->fsid, fs_info->fsid, BTRFS_UUID_SIZE)) -- cgit v1.2.3 From c1c9ff7c94e83fae89a742df74db51156869bad5 Mon Sep 17 00:00:00 2001 From: Geert Uytterhoeven Date: Tue, 20 Aug 2013 13:20:07 +0200 Subject: Btrfs: Remove superfluous casts from u64 to unsigned long long u64 is "unsigned long long" on all architectures now, so there's no need to cast it when formatting it using the "ll" length modifier. Signed-off-by: Geert Uytterhoeven Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/scrub.c | 16 +++++----------- 1 file changed, 5 insertions(+), 11 deletions(-) (limited to 'fs/btrfs/scrub.c') diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c index 266f24febce3..fcdf67ffacb0 100644 --- a/fs/btrfs/scrub.c +++ b/fs/btrfs/scrub.c @@ -754,8 +754,7 @@ out: num_uncorrectable_read_errors); printk_ratelimited_in_rcu(KERN_ERR "btrfs: unable to fixup (nodatasum) error at logical %llu on dev %s\n", - (unsigned long long)fixup->logical, - rcu_str_deref(fixup->dev->name)); + fixup->logical, rcu_str_deref(fixup->dev->name)); } btrfs_free_path(path); @@ -1154,8 +1153,7 @@ corrected_error: spin_unlock(&sctx->stat_lock); printk_ratelimited_in_rcu(KERN_ERR "btrfs: fixed up error at logical %llu on dev %s\n", - (unsigned long long)logical, - rcu_str_deref(dev->name)); + logical, rcu_str_deref(dev->name)); } } else { did_not_correct_error: @@ -1164,8 +1162,7 @@ did_not_correct_error: spin_unlock(&sctx->stat_lock); printk_ratelimited_in_rcu(KERN_ERR "btrfs: unable to fixup (regular) error at logical %llu on dev %s\n", - (unsigned long long)logical, - rcu_str_deref(dev->name)); + logical, rcu_str_deref(dev->name)); } out: @@ -2455,8 +2452,7 @@ static noinline_for_stack int scrub_stripe(struct scrub_ctx *sctx, printk(KERN_ERR "btrfs scrub: tree block %llu spanning " "stripes, ignored. logical=%llu\n", - (unsigned long long)key.objectid, - (unsigned long long)logical); + key.objectid, logical); goto next; } @@ -3176,9 +3172,7 @@ static void copy_nocow_pages_worker(struct btrfs_work *work) nocow_ctx); if (ret != 0 && ret != -ENOENT) { pr_warn("iterate_inodes_from_logical() failed: log %llu, phys %llu, len %llu, mir %llu, ret %d\n", - (unsigned long long)logical, - (unsigned long long)physical_for_dev_replace, - (unsigned long long)len, + logical, physical_for_dev_replace, len, (unsigned long long)mirror_num, ret); not_written = 1; goto out; -- cgit v1.2.3 From 27f9f02357f2bff96fc5e8a000c78ec5f96d42af Mon Sep 17 00:00:00 2001 From: Geert Uytterhoeven Date: Tue, 20 Aug 2013 13:20:09 +0200 Subject: Btrfs: Format PAGE_SIZE as unsigned long PAGE_SIZE is "unsigned long" everywhere, so there's no need to cast it to "unsigned long long" and format it as a 64-bit number. Signed-off-by: Geert Uytterhoeven Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/scrub.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) (limited to 'fs/btrfs/scrub.c') diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c index fcdf67ffacb0..b2a5f12dc9e7 100644 --- a/fs/btrfs/scrub.c +++ b/fs/btrfs/scrub.c @@ -2859,9 +2859,8 @@ int btrfs_scrub_dev(struct btrfs_fs_info *fs_info, u64 devid, u64 start, if (fs_info->chunk_root->sectorsize != PAGE_SIZE) { /* not supported for data w/o checksums */ printk(KERN_ERR - "btrfs_scrub: size assumption sectorsize != PAGE_SIZE (%d != %lld) fails\n", - fs_info->chunk_root->sectorsize, - (unsigned long long)PAGE_SIZE); + "btrfs_scrub: size assumption sectorsize != PAGE_SIZE (%d != %lu) fails\n", + fs_info->chunk_root->sectorsize, PAGE_SIZE); return -EINVAL; } -- cgit v1.2.3 From 118a0a251425b5ed5d5951097f9ef95f30611b03 Mon Sep 17 00:00:00 2001 From: Geert Uytterhoeven Date: Tue, 20 Aug 2013 13:20:10 +0200 Subject: Btrfs: Format mirror_num as int mirror_num is always "int", hence don't cast it to "unsigned long long" and format it as a 64-bit number. Signed-off-by: Geert Uytterhoeven Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/scrub.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'fs/btrfs/scrub.c') diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c index b2a5f12dc9e7..ec6a33a20200 100644 --- a/fs/btrfs/scrub.c +++ b/fs/btrfs/scrub.c @@ -3170,9 +3170,9 @@ static void copy_nocow_pages_worker(struct btrfs_work *work) copy_nocow_pages_for_inode, nocow_ctx); if (ret != 0 && ret != -ENOENT) { - pr_warn("iterate_inodes_from_logical() failed: log %llu, phys %llu, len %llu, mir %llu, ret %d\n", - logical, physical_for_dev_replace, len, - (unsigned long long)mirror_num, ret); + pr_warn("iterate_inodes_from_logical() failed: log %llu, phys %llu, len %llu, mir %u, ret %d\n", + logical, physical_for_dev_replace, len, mirror_num, + ret); not_written = 1; goto out; } -- cgit v1.2.3 From 23fa76b0ba78b7d84708d9ee683587d8a5bbceef Mon Sep 17 00:00:00 2001 From: Stefan Behrens Date: Fri, 23 Aug 2013 10:34:43 +0200 Subject: Btrf: cleanup: don't check for root_refs == 0 twice btrfs_read_fs_root_no_name() already checks if btrfs_root_refs() is zero and returns ENOENT in this case. There is no need to do it again in three more places. Signed-off-by: Stefan Behrens Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/scrub.c | 5 ----- 1 file changed, 5 deletions(-) (limited to 'fs/btrfs/scrub.c') diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c index ec6a33a20200..0afcd452fcb3 100644 --- a/fs/btrfs/scrub.c +++ b/fs/btrfs/scrub.c @@ -3217,11 +3217,6 @@ static int copy_nocow_pages_for_inode(u64 inum, u64 offset, u64 root, void *ctx) return PTR_ERR(local_root); } - if (btrfs_root_refs(&local_root->root_item) == 0) { - srcu_read_unlock(&fs_info->subvol_srcu, srcu_index); - return -ENOENT; - } - key.type = BTRFS_INODE_ITEM_KEY; key.objectid = inum; key.offset = 0; -- cgit v1.2.3 From 652f25a2921eb32a3e6f88aed3c59b494c1287c4 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Thu, 12 Sep 2013 16:58:28 -0400 Subject: Btrfs: improve replacing nocow extents Various people have hit a deadlock when running btrfs/011. This is because when replacing nocow extents we will take the i_mutex to make sure nobody messes with the file while we are replacing the extent. The problem is we are already holding a transaction open, which is a locking inversion, so instead we need to save these inodes we find and then process them outside of the transaction. Further we can't just lock the inode and assume we are good to go. We need to lock the extent range and then read back the extent cache for the inode to make sure the extent really still points at the physical block we want. If it doesn't we don't have to copy it. Thanks, Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/scrub.c | 112 ++++++++++++++++++++++++++++++++++++++++++++++++------- 1 file changed, 98 insertions(+), 14 deletions(-) (limited to 'fs/btrfs/scrub.c') diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c index 0afcd452fcb3..a18e0e23f6a6 100644 --- a/fs/btrfs/scrub.c +++ b/fs/btrfs/scrub.c @@ -158,12 +158,20 @@ struct scrub_fixup_nodatasum { int mirror_num; }; +struct scrub_nocow_inode { + u64 inum; + u64 offset; + u64 root; + struct list_head list; +}; + struct scrub_copy_nocow_ctx { struct scrub_ctx *sctx; u64 logical; u64 len; int mirror_num; u64 physical_for_dev_replace; + struct list_head inodes; struct btrfs_work work; }; @@ -245,7 +253,7 @@ static void scrub_wr_bio_end_io_worker(struct btrfs_work *work); static int write_page_nocow(struct scrub_ctx *sctx, u64 physical_for_dev_replace, struct page *page); static int copy_nocow_pages_for_inode(u64 inum, u64 offset, u64 root, - void *ctx); + struct scrub_copy_nocow_ctx *ctx); static int copy_nocow_pages(struct scrub_ctx *sctx, u64 logical, u64 len, int mirror_num, u64 physical_for_dev_replace); static void copy_nocow_pages_worker(struct btrfs_work *work); @@ -3126,12 +3134,30 @@ static int copy_nocow_pages(struct scrub_ctx *sctx, u64 logical, u64 len, nocow_ctx->mirror_num = mirror_num; nocow_ctx->physical_for_dev_replace = physical_for_dev_replace; nocow_ctx->work.func = copy_nocow_pages_worker; + INIT_LIST_HEAD(&nocow_ctx->inodes); btrfs_queue_worker(&fs_info->scrub_nocow_workers, &nocow_ctx->work); return 0; } +static int record_inode_for_nocow(u64 inum, u64 offset, u64 root, void *ctx) +{ + struct scrub_copy_nocow_ctx *nocow_ctx = ctx; + struct scrub_nocow_inode *nocow_inode; + + nocow_inode = kzalloc(sizeof(*nocow_inode), GFP_NOFS); + if (!nocow_inode) + return -ENOMEM; + nocow_inode->inum = inum; + nocow_inode->offset = offset; + nocow_inode->root = root; + list_add_tail(&nocow_inode->list, &nocow_ctx->inodes); + return 0; +} + +#define COPY_COMPLETE 1 + static void copy_nocow_pages_worker(struct btrfs_work *work) { struct scrub_copy_nocow_ctx *nocow_ctx = @@ -3167,8 +3193,7 @@ static void copy_nocow_pages_worker(struct btrfs_work *work) } ret = iterate_inodes_from_logical(logical, fs_info, path, - copy_nocow_pages_for_inode, - nocow_ctx); + record_inode_for_nocow, nocow_ctx); if (ret != 0 && ret != -ENOENT) { pr_warn("iterate_inodes_from_logical() failed: log %llu, phys %llu, len %llu, mir %u, ret %d\n", logical, physical_for_dev_replace, len, mirror_num, @@ -3177,7 +3202,33 @@ static void copy_nocow_pages_worker(struct btrfs_work *work) goto out; } + btrfs_end_transaction(trans, root); + trans = NULL; + while (!list_empty(&nocow_ctx->inodes)) { + struct scrub_nocow_inode *entry; + entry = list_first_entry(&nocow_ctx->inodes, + struct scrub_nocow_inode, + list); + list_del_init(&entry->list); + ret = copy_nocow_pages_for_inode(entry->inum, entry->offset, + entry->root, nocow_ctx); + kfree(entry); + if (ret == COPY_COMPLETE) { + ret = 0; + break; + } else if (ret) { + break; + } + } out: + while (!list_empty(&nocow_ctx->inodes)) { + struct scrub_nocow_inode *entry; + entry = list_first_entry(&nocow_ctx->inodes, + struct scrub_nocow_inode, + list); + list_del_init(&entry->list); + kfree(entry); + } if (trans && !IS_ERR(trans)) btrfs_end_transaction(trans, root); if (not_written) @@ -3190,20 +3241,25 @@ out: scrub_pending_trans_workers_dec(sctx); } -static int copy_nocow_pages_for_inode(u64 inum, u64 offset, u64 root, void *ctx) +static int copy_nocow_pages_for_inode(u64 inum, u64 offset, u64 root, + struct scrub_copy_nocow_ctx *nocow_ctx) { - struct scrub_copy_nocow_ctx *nocow_ctx = ctx; struct btrfs_fs_info *fs_info = nocow_ctx->sctx->dev_root->fs_info; struct btrfs_key key; struct inode *inode; struct page *page; struct btrfs_root *local_root; + struct btrfs_ordered_extent *ordered; + struct extent_map *em; + struct extent_state *cached_state = NULL; + struct extent_io_tree *io_tree; u64 physical_for_dev_replace; - u64 len; + u64 len = nocow_ctx->len; + u64 lockstart = offset, lockend = offset + len - 1; unsigned long index; int srcu_index; - int ret; - int err; + int ret = 0; + int err = 0; key.objectid = root; key.type = BTRFS_ROOT_ITEM_KEY; @@ -3229,9 +3285,33 @@ static int copy_nocow_pages_for_inode(u64 inum, u64 offset, u64 root, void *ctx) mutex_lock(&inode->i_mutex); inode_dio_wait(inode); - ret = 0; physical_for_dev_replace = nocow_ctx->physical_for_dev_replace; - len = nocow_ctx->len; + io_tree = &BTRFS_I(inode)->io_tree; + + lock_extent_bits(io_tree, lockstart, lockend, 0, &cached_state); + ordered = btrfs_lookup_ordered_range(inode, lockstart, len); + if (ordered) { + btrfs_put_ordered_extent(ordered); + goto out_unlock; + } + + em = btrfs_get_extent(inode, NULL, 0, lockstart, len, 0); + if (IS_ERR(em)) { + ret = PTR_ERR(em); + goto out_unlock; + } + + /* + * This extent does not actually cover the logical extent anymore, + * move on to the next inode. + */ + if (em->block_start > nocow_ctx->logical || + em->block_start + em->block_len < nocow_ctx->logical + len) { + free_extent_map(em); + goto out_unlock; + } + free_extent_map(em); + while (len >= PAGE_CACHE_SIZE) { index = offset >> PAGE_CACHE_SHIFT; again: @@ -3247,10 +3327,9 @@ again: goto next_page; } else { ClearPageError(page); - err = extent_read_full_page(&BTRFS_I(inode)-> - io_tree, - page, btrfs_get_extent, - nocow_ctx->mirror_num); + err = extent_read_full_page_nolock(io_tree, page, + btrfs_get_extent, + nocow_ctx->mirror_num); if (err) { ret = err; goto next_page; @@ -3264,6 +3343,7 @@ again: * page in the page cache. */ if (page->mapping != inode->i_mapping) { + unlock_page(page); page_cache_release(page); goto again; } @@ -3287,6 +3367,10 @@ next_page: physical_for_dev_replace += PAGE_CACHE_SIZE; len -= PAGE_CACHE_SIZE; } + ret = COPY_COMPLETE; +out_unlock: + unlock_extent_cached(io_tree, lockstart, lockend, &cached_state, + GFP_NOFS); out: mutex_unlock(&inode->i_mutex); iput(inode); -- cgit v1.2.3