From 64a167011bcabc1e855658387c8a4464b71f3138 Mon Sep 17 00:00:00 2001 From: David Woodhouse Date: Wed, 15 Jul 2009 23:29:37 +0100 Subject: Btrfs: add rw argument to merge_bio_hook() We'll want to merge writes so they can fill a full RAID[56] stripe, but not necessarily reads. Signed-off-by: David Woodhouse Signed-off-by: Chris Mason --- fs/btrfs/ctree.h | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'fs/btrfs/ctree.h') diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 547b7b05727f..0ab51be6879f 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -3459,9 +3459,9 @@ int btrfs_writepages(struct address_space *mapping, struct writeback_control *wbc); int btrfs_create_subvol_root(struct btrfs_trans_handle *trans, struct btrfs_root *new_root, u64 new_dirid); -int btrfs_merge_bio_hook(struct page *page, unsigned long offset, - size_t size, struct bio *bio, unsigned long bio_flags); - +int btrfs_merge_bio_hook(int rw, struct page *page, unsigned long offset, + size_t size, struct bio *bio, + unsigned long bio_flags); int btrfs_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf); int btrfs_readpage(struct file *file, struct page *page); void btrfs_evict_inode(struct inode *inode); -- cgit v1.2.3 From 53b381b3abeb86f12787a6c40fee9b2f71edc23b Mon Sep 17 00:00:00 2001 From: David Woodhouse Date: Tue, 29 Jan 2013 18:40:14 -0500 Subject: Btrfs: RAID5 and RAID6 This builds on David Woodhouse's original Btrfs raid5/6 implementation. The code has changed quite a bit, blame Chris Mason for any bugs. Read/modify/write is done after the higher levels of the filesystem have prepared a given bio. This means the higher layers are not responsible for building full stripes, and they don't need to query for the topology of the extents that may get allocated during delayed allocation runs. It also means different files can easily share the same stripe. But, it does expose us to incorrect parity if we crash or lose power while doing a read/modify/write cycle. This will be addressed in a later commit. Scrub is unable to repair crc errors on raid5/6 chunks. Discard does not work on raid5/6 (yet) The stripe size is fixed at 64KiB per disk. This will be tunable in a later commit. Signed-off-by: Chris Mason --- fs/btrfs/Kconfig | 2 + fs/btrfs/Makefile | 2 +- fs/btrfs/ctree.h | 35 +- fs/btrfs/disk-io.c | 62 +- fs/btrfs/disk-io.h | 7 + fs/btrfs/extent-tree.c | 88 ++- fs/btrfs/extent_io.c | 18 +- fs/btrfs/free-space-cache.c | 50 +- fs/btrfs/inode.c | 18 +- fs/btrfs/raid56.c | 1647 +++++++++++++++++++++++++++++++++++++++++++ fs/btrfs/raid56.h | 51 ++ fs/btrfs/scrub.c | 8 + fs/btrfs/transaction.c | 3 + fs/btrfs/volumes.c | 385 ++++++++-- fs/btrfs/volumes.h | 9 +- 15 files changed, 2283 insertions(+), 102 deletions(-) create mode 100644 fs/btrfs/raid56.c create mode 100644 fs/btrfs/raid56.h (limited to 'fs/btrfs/ctree.h') diff --git a/fs/btrfs/Kconfig b/fs/btrfs/Kconfig index d33f01c08b60..4f5dc93fa2f8 100644 --- a/fs/btrfs/Kconfig +++ b/fs/btrfs/Kconfig @@ -6,6 +6,8 @@ config BTRFS_FS select ZLIB_DEFLATE select LZO_COMPRESS select LZO_DECOMPRESS + select RAID6_PQ + help Btrfs is a new filesystem with extents, writable snapshotting, support for multiple devices and many more features. diff --git a/fs/btrfs/Makefile b/fs/btrfs/Makefile index 7df3e0f0ee51..3932224f99e9 100644 --- a/fs/btrfs/Makefile +++ b/fs/btrfs/Makefile @@ -8,7 +8,7 @@ btrfs-y += super.o ctree.o extent-tree.o print-tree.o root-tree.o dir-item.o \ extent_io.o volumes.o async-thread.o ioctl.o locking.o orphan.o \ export.o tree-log.o free-space-cache.o zlib.o lzo.o \ compression.o delayed-ref.o relocation.o delayed-inode.o scrub.o \ - reada.o backref.o ulist.o qgroup.o send.o dev-replace.o + reada.o backref.o ulist.o qgroup.o send.o dev-replace.o raid56.o btrfs-$(CONFIG_BTRFS_FS_POSIX_ACL) += acl.o btrfs-$(CONFIG_BTRFS_FS_CHECK_INTEGRITY) += check-integrity.o diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 0ab51be6879f..0cce3aafbd62 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -502,6 +502,7 @@ struct btrfs_super_block { #define BTRFS_FEATURE_INCOMPAT_BIG_METADATA (1ULL << 5) #define BTRFS_FEATURE_INCOMPAT_EXTENDED_IREF (1ULL << 6) +#define BTRFS_FEATURE_INCOMPAT_RAID56 (1ULL << 7) #define BTRFS_FEATURE_COMPAT_SUPP 0ULL #define BTRFS_FEATURE_COMPAT_RO_SUPP 0ULL @@ -511,6 +512,7 @@ struct btrfs_super_block { BTRFS_FEATURE_INCOMPAT_MIXED_GROUPS | \ BTRFS_FEATURE_INCOMPAT_BIG_METADATA | \ BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO | \ + BTRFS_FEATURE_INCOMPAT_RAID56 | \ BTRFS_FEATURE_INCOMPAT_EXTENDED_IREF) /* @@ -952,8 +954,10 @@ struct btrfs_dev_replace_item { #define BTRFS_BLOCK_GROUP_RAID1 (1ULL << 4) #define BTRFS_BLOCK_GROUP_DUP (1ULL << 5) #define BTRFS_BLOCK_GROUP_RAID10 (1ULL << 6) +#define BTRFS_BLOCK_GROUP_RAID5 (1 << 7) +#define BTRFS_BLOCK_GROUP_RAID6 (1 << 8) #define BTRFS_BLOCK_GROUP_RESERVED BTRFS_AVAIL_ALLOC_BIT_SINGLE -#define BTRFS_NR_RAID_TYPES 5 +#define BTRFS_NR_RAID_TYPES 7 #define BTRFS_BLOCK_GROUP_TYPE_MASK (BTRFS_BLOCK_GROUP_DATA | \ BTRFS_BLOCK_GROUP_SYSTEM | \ @@ -961,6 +965,8 @@ struct btrfs_dev_replace_item { #define BTRFS_BLOCK_GROUP_PROFILE_MASK (BTRFS_BLOCK_GROUP_RAID0 | \ BTRFS_BLOCK_GROUP_RAID1 | \ + BTRFS_BLOCK_GROUP_RAID5 | \ + BTRFS_BLOCK_GROUP_RAID6 | \ BTRFS_BLOCK_GROUP_DUP | \ BTRFS_BLOCK_GROUP_RAID10) /* @@ -1185,6 +1191,10 @@ struct btrfs_block_group_cache { u64 flags; u64 sectorsize; u64 cache_generation; + + /* for raid56, this is a full stripe, without parity */ + unsigned long full_stripe_len; + unsigned int ro:1; unsigned int dirty:1; unsigned int iref:1; @@ -1225,6 +1235,20 @@ struct seq_list { u64 seq; }; +/* used by the raid56 code to lock stripes for read/modify/write */ +struct btrfs_stripe_hash { + struct list_head hash_list; + wait_queue_head_t wait; + spinlock_t lock; +}; + +/* used by the raid56 code to lock stripes for read/modify/write */ +struct btrfs_stripe_hash_table { + struct btrfs_stripe_hash *table; +}; + +#define BTRFS_STRIPE_HASH_TABLE_BITS 11 + /* fs_info */ struct reloc_control; struct btrfs_device; @@ -1307,6 +1331,13 @@ struct btrfs_fs_info { struct mutex cleaner_mutex; struct mutex chunk_mutex; struct mutex volume_mutex; + + /* this is used during read/modify/write to make sure + * no two ios are trying to mod the same stripe at the same + * time + */ + struct btrfs_stripe_hash_table *stripe_hash_table; + /* * this protects the ordered operations list only while we are * processing all of the entries on it. This way we make @@ -1395,6 +1426,8 @@ struct btrfs_fs_info { struct btrfs_workers flush_workers; struct btrfs_workers endio_workers; struct btrfs_workers endio_meta_workers; + struct btrfs_workers endio_raid56_workers; + struct btrfs_workers rmw_workers; struct btrfs_workers endio_meta_write_workers; struct btrfs_workers endio_write_workers; struct btrfs_workers endio_freespace_worker; diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 65f03670a952..e9fa7b4d18e3 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -46,6 +46,7 @@ #include "check-integrity.h" #include "rcu-string.h" #include "dev-replace.h" +#include "raid56.h" #ifdef CONFIG_X86 #include @@ -639,8 +640,15 @@ err: btree_readahead_hook(root, eb, eb->start, ret); } - if (ret) + if (ret) { + /* + * our io error hook is going to dec the io pages + * again, we have to make sure it has something + * to decrement + */ + atomic_inc(&eb->io_pages); clear_extent_buffer_uptodate(eb); + } free_extent_buffer(eb); out: return ret; @@ -654,6 +662,7 @@ static int btree_io_failed_hook(struct page *page, int failed_mirror) eb = (struct extent_buffer *)page->private; set_bit(EXTENT_BUFFER_IOERR, &eb->bflags); eb->read_mirror = failed_mirror; + atomic_dec(&eb->io_pages); if (test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags)) btree_readahead_hook(root, eb, eb->start, -EIO); return -EIO; /* we fixed nothing */ @@ -670,17 +679,23 @@ static void end_workqueue_bio(struct bio *bio, int err) end_io_wq->work.flags = 0; if (bio->bi_rw & REQ_WRITE) { - if (end_io_wq->metadata == 1) + if (end_io_wq->metadata == BTRFS_WQ_ENDIO_METADATA) btrfs_queue_worker(&fs_info->endio_meta_write_workers, &end_io_wq->work); - else if (end_io_wq->metadata == 2) + else if (end_io_wq->metadata == BTRFS_WQ_ENDIO_FREE_SPACE) btrfs_queue_worker(&fs_info->endio_freespace_worker, &end_io_wq->work); + else if (end_io_wq->metadata == BTRFS_WQ_ENDIO_RAID56) + btrfs_queue_worker(&fs_info->endio_raid56_workers, + &end_io_wq->work); else btrfs_queue_worker(&fs_info->endio_write_workers, &end_io_wq->work); } else { - if (end_io_wq->metadata) + if (end_io_wq->metadata == BTRFS_WQ_ENDIO_RAID56) + btrfs_queue_worker(&fs_info->endio_raid56_workers, + &end_io_wq->work); + else if (end_io_wq->metadata) btrfs_queue_worker(&fs_info->endio_meta_workers, &end_io_wq->work); else @@ -695,6 +710,7 @@ static void end_workqueue_bio(struct bio *bio, int err) * 0 - if data * 1 - if normal metadta * 2 - if writing to the free space cache area + * 3 - raid parity work */ int btrfs_bio_wq_end_io(struct btrfs_fs_info *info, struct bio *bio, int metadata) @@ -2165,6 +2181,12 @@ int open_ctree(struct super_block *sb, init_waitqueue_head(&fs_info->transaction_blocked_wait); init_waitqueue_head(&fs_info->async_submit_wait); + ret = btrfs_alloc_stripe_hash_table(fs_info); + if (ret) { + err = -ENOMEM; + goto fail_alloc; + } + __setup_root(4096, 4096, 4096, 4096, tree_root, fs_info, BTRFS_ROOT_TREE_OBJECTID); @@ -2332,6 +2354,12 @@ int open_ctree(struct super_block *sb, btrfs_init_workers(&fs_info->endio_meta_write_workers, "endio-meta-write", fs_info->thread_pool_size, &fs_info->generic_worker); + btrfs_init_workers(&fs_info->endio_raid56_workers, + "endio-raid56", fs_info->thread_pool_size, + &fs_info->generic_worker); + btrfs_init_workers(&fs_info->rmw_workers, + "rmw", fs_info->thread_pool_size, + &fs_info->generic_worker); btrfs_init_workers(&fs_info->endio_write_workers, "endio-write", fs_info->thread_pool_size, &fs_info->generic_worker); @@ -2350,6 +2378,8 @@ int open_ctree(struct super_block *sb, */ fs_info->endio_workers.idle_thresh = 4; fs_info->endio_meta_workers.idle_thresh = 4; + fs_info->endio_raid56_workers.idle_thresh = 4; + fs_info->rmw_workers.idle_thresh = 2; fs_info->endio_write_workers.idle_thresh = 2; fs_info->endio_meta_write_workers.idle_thresh = 2; @@ -2366,6 +2396,8 @@ int open_ctree(struct super_block *sb, ret |= btrfs_start_workers(&fs_info->fixup_workers); ret |= btrfs_start_workers(&fs_info->endio_workers); ret |= btrfs_start_workers(&fs_info->endio_meta_workers); + ret |= btrfs_start_workers(&fs_info->rmw_workers); + ret |= btrfs_start_workers(&fs_info->endio_raid56_workers); ret |= btrfs_start_workers(&fs_info->endio_meta_write_workers); ret |= btrfs_start_workers(&fs_info->endio_write_workers); ret |= btrfs_start_workers(&fs_info->endio_freespace_worker); @@ -2710,6 +2742,8 @@ fail_sb_buffer: btrfs_stop_workers(&fs_info->workers); btrfs_stop_workers(&fs_info->endio_workers); btrfs_stop_workers(&fs_info->endio_meta_workers); + btrfs_stop_workers(&fs_info->endio_raid56_workers); + btrfs_stop_workers(&fs_info->rmw_workers); btrfs_stop_workers(&fs_info->endio_meta_write_workers); btrfs_stop_workers(&fs_info->endio_write_workers); btrfs_stop_workers(&fs_info->endio_freespace_worker); @@ -2728,6 +2762,7 @@ fail_bdi: fail_srcu: cleanup_srcu_struct(&fs_info->subvol_srcu); fail: + btrfs_free_stripe_hash_table(fs_info); btrfs_close_devices(fs_info->fs_devices); return err; @@ -3076,11 +3111,16 @@ int btrfs_calc_num_tolerated_disk_barrier_failures( ((flags & BTRFS_BLOCK_GROUP_PROFILE_MASK) == 0))) num_tolerated_disk_barrier_failures = 0; - else if (num_tolerated_disk_barrier_failures > 1 - && - (flags & (BTRFS_BLOCK_GROUP_RAID1 | - BTRFS_BLOCK_GROUP_RAID10))) - num_tolerated_disk_barrier_failures = 1; + else if (num_tolerated_disk_barrier_failures > 1) { + if (flags & (BTRFS_BLOCK_GROUP_RAID1 | + BTRFS_BLOCK_GROUP_RAID5 | + BTRFS_BLOCK_GROUP_RAID10)) { + num_tolerated_disk_barrier_failures = 1; + } else if (flags & + BTRFS_BLOCK_GROUP_RAID5) { + num_tolerated_disk_barrier_failures = 2; + } + } } } up_read(&sinfo->groups_sem); @@ -3384,6 +3424,8 @@ int close_ctree(struct btrfs_root *root) btrfs_stop_workers(&fs_info->workers); btrfs_stop_workers(&fs_info->endio_workers); btrfs_stop_workers(&fs_info->endio_meta_workers); + btrfs_stop_workers(&fs_info->endio_raid56_workers); + btrfs_stop_workers(&fs_info->rmw_workers); btrfs_stop_workers(&fs_info->endio_meta_write_workers); btrfs_stop_workers(&fs_info->endio_write_workers); btrfs_stop_workers(&fs_info->endio_freespace_worker); @@ -3404,6 +3446,8 @@ int close_ctree(struct btrfs_root *root) bdi_destroy(&fs_info->bdi); cleanup_srcu_struct(&fs_info->subvol_srcu); + btrfs_free_stripe_hash_table(fs_info); + return 0; } diff --git a/fs/btrfs/disk-io.h b/fs/btrfs/disk-io.h index 305c33efb0e3..034d7dc552b2 100644 --- a/fs/btrfs/disk-io.h +++ b/fs/btrfs/disk-io.h @@ -25,6 +25,13 @@ #define BTRFS_SUPER_MIRROR_MAX 3 #define BTRFS_SUPER_MIRROR_SHIFT 12 +enum { + BTRFS_WQ_ENDIO_DATA = 0, + BTRFS_WQ_ENDIO_METADATA = 1, + BTRFS_WQ_ENDIO_FREE_SPACE = 2, + BTRFS_WQ_ENDIO_RAID56 = 3, +}; + static inline u64 btrfs_sb_offset(int mirror) { u64 start = 16 * 1024; diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index d133edfcd449..3345f68fc64b 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -31,6 +31,7 @@ #include "print-tree.h" #include "transaction.h" #include "volumes.h" +#include "raid56.h" #include "locking.h" #include "free-space-cache.h" #include "math.h" @@ -1852,6 +1853,8 @@ static int btrfs_discard_extent(struct btrfs_root *root, u64 bytenr, *actual_bytes = discarded_bytes; + if (ret == -EOPNOTSUPP) + ret = 0; return ret; } @@ -3276,6 +3279,7 @@ u64 btrfs_reduce_alloc_profile(struct btrfs_root *root, u64 flags) u64 num_devices = root->fs_info->fs_devices->rw_devices + root->fs_info->fs_devices->missing_devices; u64 target; + u64 tmp; /* * see if restripe for this chunk_type is in progress, if so @@ -3292,30 +3296,32 @@ u64 btrfs_reduce_alloc_profile(struct btrfs_root *root, u64 flags) } spin_unlock(&root->fs_info->balance_lock); + /* First, mask out the RAID levels which aren't possible */ if (num_devices == 1) - flags &= ~(BTRFS_BLOCK_GROUP_RAID1 | BTRFS_BLOCK_GROUP_RAID0); + flags &= ~(BTRFS_BLOCK_GROUP_RAID1 | BTRFS_BLOCK_GROUP_RAID0 | + BTRFS_BLOCK_GROUP_RAID5); + if (num_devices < 3) + flags &= ~BTRFS_BLOCK_GROUP_RAID6; if (num_devices < 4) flags &= ~BTRFS_BLOCK_GROUP_RAID10; - if ((flags & BTRFS_BLOCK_GROUP_DUP) && - (flags & (BTRFS_BLOCK_GROUP_RAID1 | - BTRFS_BLOCK_GROUP_RAID10))) { - flags &= ~BTRFS_BLOCK_GROUP_DUP; - } - - if ((flags & BTRFS_BLOCK_GROUP_RAID1) && - (flags & BTRFS_BLOCK_GROUP_RAID10)) { - flags &= ~BTRFS_BLOCK_GROUP_RAID1; - } + tmp = flags & (BTRFS_BLOCK_GROUP_DUP | BTRFS_BLOCK_GROUP_RAID0 | + BTRFS_BLOCK_GROUP_RAID1 | BTRFS_BLOCK_GROUP_RAID5 | + BTRFS_BLOCK_GROUP_RAID6 | BTRFS_BLOCK_GROUP_RAID10); + flags &= ~tmp; - if ((flags & BTRFS_BLOCK_GROUP_RAID0) && - ((flags & BTRFS_BLOCK_GROUP_RAID1) | - (flags & BTRFS_BLOCK_GROUP_RAID10) | - (flags & BTRFS_BLOCK_GROUP_DUP))) { - flags &= ~BTRFS_BLOCK_GROUP_RAID0; - } + if (tmp & BTRFS_BLOCK_GROUP_RAID6) + tmp = BTRFS_BLOCK_GROUP_RAID6; + else if (tmp & BTRFS_BLOCK_GROUP_RAID5) + tmp = BTRFS_BLOCK_GROUP_RAID5; + else if (tmp & BTRFS_BLOCK_GROUP_RAID10) + tmp = BTRFS_BLOCK_GROUP_RAID10; + else if (tmp & BTRFS_BLOCK_GROUP_RAID1) + tmp = BTRFS_BLOCK_GROUP_RAID1; + else if (tmp & BTRFS_BLOCK_GROUP_RAID0) + tmp = BTRFS_BLOCK_GROUP_RAID0; - return extended_to_chunk(flags); + return extended_to_chunk(flags | tmp); } static u64 get_alloc_profile(struct btrfs_root *root, u64 flags) @@ -3333,6 +3339,7 @@ static u64 get_alloc_profile(struct btrfs_root *root, u64 flags) u64 btrfs_get_alloc_profile(struct btrfs_root *root, int data) { u64 flags; + u64 ret; if (data) flags = BTRFS_BLOCK_GROUP_DATA; @@ -3341,7 +3348,8 @@ u64 btrfs_get_alloc_profile(struct btrfs_root *root, int data) else flags = BTRFS_BLOCK_GROUP_METADATA; - return get_alloc_profile(root, flags); + ret = get_alloc_profile(root, flags); + return ret; } /* @@ -3516,8 +3524,10 @@ static u64 get_system_chunk_thresh(struct btrfs_root *root, u64 type) { u64 num_dev; - if (type & BTRFS_BLOCK_GROUP_RAID10 || - type & BTRFS_BLOCK_GROUP_RAID0) + if (type & (BTRFS_BLOCK_GROUP_RAID10 | + BTRFS_BLOCK_GROUP_RAID0 | + BTRFS_BLOCK_GROUP_RAID5 | + BTRFS_BLOCK_GROUP_RAID6)) num_dev = root->fs_info->fs_devices->rw_devices; else if (type & BTRFS_BLOCK_GROUP_RAID1) num_dev = 2; @@ -3667,7 +3677,9 @@ static int can_overcommit(struct btrfs_root *root, /* * If we have dup, raid1 or raid10 then only half of the free - * space is actually useable. + * space is actually useable. For raid56, the space info used + * doesn't include the parity drive, so we don't have to + * change the math */ if (profile & (BTRFS_BLOCK_GROUP_DUP | BTRFS_BLOCK_GROUP_RAID1 | @@ -5455,10 +5467,14 @@ int btrfs_free_extent(struct btrfs_trans_handle *trans, struct btrfs_root *root, return ret; } -static u64 stripe_align(struct btrfs_root *root, u64 val) +static u64 stripe_align(struct btrfs_root *root, + struct btrfs_block_group_cache *cache, + u64 val, u64 num_bytes) { - u64 mask = ((u64)root->stripesize - 1); - u64 ret = (val + mask) & ~mask; + u64 mask; + u64 ret; + mask = ((u64)root->stripesize - 1); + ret = (val + mask) & ~mask; return ret; } @@ -5519,9 +5535,12 @@ int __get_raid_index(u64 flags) index = 2; else if (flags & BTRFS_BLOCK_GROUP_RAID0) index = 3; + else if (flags & BTRFS_BLOCK_GROUP_RAID5) + index = 5; + else if (flags & BTRFS_BLOCK_GROUP_RAID6) + index = 6; else - index = 4; - + index = 4; /* BTRFS_BLOCK_GROUP_SINGLE */ return index; } @@ -5665,6 +5684,8 @@ search: if (!block_group_bits(block_group, data)) { u64 extra = BTRFS_BLOCK_GROUP_DUP | BTRFS_BLOCK_GROUP_RAID1 | + BTRFS_BLOCK_GROUP_RAID5 | + BTRFS_BLOCK_GROUP_RAID6 | BTRFS_BLOCK_GROUP_RAID10; /* @@ -5835,7 +5856,8 @@ unclustered_alloc: goto loop; } checks: - search_start = stripe_align(root, offset); + search_start = stripe_align(root, used_block_group, + offset, num_bytes); /* move on to the next group */ if (search_start + num_bytes > @@ -7203,6 +7225,7 @@ static u64 update_block_group_flags(struct btrfs_root *root, u64 flags) root->fs_info->fs_devices->missing_devices; stripped = BTRFS_BLOCK_GROUP_RAID0 | + BTRFS_BLOCK_GROUP_RAID5 | BTRFS_BLOCK_GROUP_RAID6 | BTRFS_BLOCK_GROUP_RAID1 | BTRFS_BLOCK_GROUP_RAID10; if (num_devices == 1) { @@ -7754,7 +7777,9 @@ int btrfs_read_block_groups(struct btrfs_root *root) btrfs_release_path(path); cache->flags = btrfs_block_group_flags(&cache->item); cache->sectorsize = root->sectorsize; - + cache->full_stripe_len = btrfs_full_stripe_len(root, + &root->fs_info->mapping_tree, + found_key.objectid); btrfs_init_free_space_ctl(cache); /* @@ -7808,6 +7833,8 @@ int btrfs_read_block_groups(struct btrfs_root *root) if (!(get_alloc_profile(root, space_info->flags) & (BTRFS_BLOCK_GROUP_RAID10 | BTRFS_BLOCK_GROUP_RAID1 | + BTRFS_BLOCK_GROUP_RAID5 | + BTRFS_BLOCK_GROUP_RAID6 | BTRFS_BLOCK_GROUP_DUP))) continue; /* @@ -7883,6 +7910,9 @@ int btrfs_make_block_group(struct btrfs_trans_handle *trans, cache->key.type = BTRFS_BLOCK_GROUP_ITEM_KEY; cache->sectorsize = root->sectorsize; cache->fs_info = root->fs_info; + cache->full_stripe_len = btrfs_full_stripe_len(root, + &root->fs_info->mapping_tree, + chunk_offset); atomic_set(&cache->count, 1); spin_lock_init(&cache->lock); diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c index 86ecca48c604..3b9fb478b0d1 100644 --- a/fs/btrfs/extent_io.c +++ b/fs/btrfs/extent_io.c @@ -1895,13 +1895,11 @@ static int free_io_failure(struct inode *inode, struct io_failure_record *rec, if (ret) err = ret; - if (did_repair) { - ret = clear_extent_bits(&BTRFS_I(inode)->io_tree, rec->start, - rec->start + rec->len - 1, - EXTENT_DAMAGED, GFP_NOFS); - if (ret && !err) - err = ret; - } + ret = clear_extent_bits(&BTRFS_I(inode)->io_tree, rec->start, + rec->start + rec->len - 1, + EXTENT_DAMAGED, GFP_NOFS); + if (ret && !err) + err = ret; kfree(rec); return err; @@ -1932,10 +1930,15 @@ int repair_io_failure(struct btrfs_fs_info *fs_info, u64 start, u64 map_length = 0; u64 sector; struct btrfs_bio *bbio = NULL; + struct btrfs_mapping_tree *map_tree = &fs_info->mapping_tree; int ret; BUG_ON(!mirror_num); + /* we can't repair anything in raid56 yet */ + if (btrfs_is_parity_mirror(map_tree, logical, length, mirror_num)) + return 0; + bio = bio_alloc(GFP_NOFS, 1); if (!bio) return -EIO; @@ -2052,6 +2055,7 @@ static int clean_io_failure(u64 start, struct page *page) failrec->failed_mirror); did_repair = !ret; } + ret = 0; } out: diff --git a/fs/btrfs/free-space-cache.c b/fs/btrfs/free-space-cache.c index 59ea2e4349c9..62020b7f7036 100644 --- a/fs/btrfs/free-space-cache.c +++ b/fs/btrfs/free-space-cache.c @@ -1463,10 +1463,14 @@ static int search_bitmap(struct btrfs_free_space_ctl *ctl, } static struct btrfs_free_space * -find_free_space(struct btrfs_free_space_ctl *ctl, u64 *offset, u64 *bytes) +find_free_space(struct btrfs_free_space_ctl *ctl, u64 *offset, u64 *bytes, + unsigned long align) { struct btrfs_free_space *entry; struct rb_node *node; + u64 ctl_off; + u64 tmp; + u64 align_off; int ret; if (!ctl->free_space_offset.rb_node) @@ -1481,15 +1485,34 @@ find_free_space(struct btrfs_free_space_ctl *ctl, u64 *offset, u64 *bytes) if (entry->bytes < *bytes) continue; + /* make sure the space returned is big enough + * to match our requested alignment + */ + if (*bytes >= align) { + ctl_off = entry->offset - ctl->start; + tmp = ctl_off + align - 1;; + do_div(tmp, align); + tmp = tmp * align + ctl->start; + align_off = tmp - entry->offset; + } else { + align_off = 0; + tmp = entry->offset; + } + + if (entry->bytes < *bytes + align_off) + continue; + if (entry->bitmap) { - ret = search_bitmap(ctl, entry, offset, bytes); - if (!ret) + ret = search_bitmap(ctl, entry, &tmp, bytes); + if (!ret) { + *offset = tmp; return entry; + } continue; } - *offset = entry->offset; - *bytes = entry->bytes; + *offset = tmp; + *bytes = entry->bytes - align_off; return entry; } @@ -2091,9 +2114,12 @@ u64 btrfs_find_space_for_alloc(struct btrfs_block_group_cache *block_group, struct btrfs_free_space *entry = NULL; u64 bytes_search = bytes + empty_size; u64 ret = 0; + u64 align_gap = 0; + u64 align_gap_len = 0; spin_lock(&ctl->tree_lock); - entry = find_free_space(ctl, &offset, &bytes_search); + entry = find_free_space(ctl, &offset, &bytes_search, + block_group->full_stripe_len); if (!entry) goto out; @@ -2103,9 +2129,15 @@ u64 btrfs_find_space_for_alloc(struct btrfs_block_group_cache *block_group, if (!entry->bytes) free_bitmap(ctl, entry); } else { + unlink_free_space(ctl, entry); - entry->offset += bytes; - entry->bytes -= bytes; + align_gap_len = offset - entry->offset; + align_gap = entry->offset; + + entry->offset = offset + bytes; + WARN_ON(entry->bytes < bytes + align_gap_len); + + entry->bytes -= bytes + align_gap_len; if (!entry->bytes) kmem_cache_free(btrfs_free_space_cachep, entry); else @@ -2115,6 +2147,8 @@ u64 btrfs_find_space_for_alloc(struct btrfs_block_group_cache *block_group, out: spin_unlock(&ctl->tree_lock); + if (align_gap_len) + __btrfs_add_free_space(ctl, align_gap, align_gap_len); return ret; } diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 1b98c4ce3c6f..6f4e41dca970 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -39,6 +39,7 @@ #include #include #include +#include #include "compat.h" #include "ctree.h" #include "disk-io.h" @@ -6386,19 +6387,24 @@ static int btrfs_submit_direct_hook(int rw, struct btrfs_dio_private *dip, int async_submit = 0; map_length = orig_bio->bi_size; - ret = btrfs_map_block(root->fs_info, READ, start_sector << 9, + ret = btrfs_map_block(root->fs_info, rw, start_sector << 9, &map_length, NULL, 0); if (ret) { bio_put(orig_bio); return -EIO; } - if (map_length >= orig_bio->bi_size) { bio = orig_bio; goto submit; } - async_submit = 1; + /* async crcs make it difficult to collect full stripe writes. */ + if (btrfs_get_alloc_profile(root, 1) & + (BTRFS_BLOCK_GROUP_RAID5 | BTRFS_BLOCK_GROUP_RAID6)) + async_submit = 0; + else + async_submit = 1; + bio = btrfs_dio_bio_alloc(orig_bio->bi_bdev, start_sector, GFP_NOFS); if (!bio) return -ENOMEM; @@ -6440,7 +6446,7 @@ static int btrfs_submit_direct_hook(int rw, struct btrfs_dio_private *dip, bio->bi_end_io = btrfs_end_dio_bio; map_length = orig_bio->bi_size; - ret = btrfs_map_block(root->fs_info, READ, + ret = btrfs_map_block(root->fs_info, rw, start_sector << 9, &map_length, NULL, 0); if (ret) { @@ -6583,15 +6589,17 @@ static ssize_t btrfs_direct_IO(int rw, struct kiocb *iocb, { struct file *file = iocb->ki_filp; struct inode *inode = file->f_mapping->host; + ssize_t ret; if (check_direct_IO(BTRFS_I(inode)->root, rw, iocb, iov, offset, nr_segs)) return 0; - return __blockdev_direct_IO(rw, iocb, inode, + ret = __blockdev_direct_IO(rw, iocb, inode, BTRFS_I(inode)->root->fs_info->fs_devices->latest_bdev, iov, offset, nr_segs, btrfs_get_blocks_direct, NULL, btrfs_submit_direct, 0); + return ret; } #define BTRFS_FIEMAP_FLAGS (FIEMAP_FLAG_SYNC) diff --git a/fs/btrfs/raid56.c b/fs/btrfs/raid56.c new file mode 100644 index 000000000000..d02510f34936 --- /dev/null +++ b/fs/btrfs/raid56.c @@ -0,0 +1,1647 @@ +/* + * Copyright (C) 2012 Fusion-io All rights reserved. + * Copyright (C) 2012 Intel Corp. All rights reserved. + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public + * License v2 as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + * + * You should have received a copy of the GNU General Public + * License along with this program; if not, write to the + * Free Software Foundation, Inc., 59 Temple Place - Suite 330, + * Boston, MA 021110-1307, USA. + */ +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include "compat.h" +#include "ctree.h" +#include "extent_map.h" +#include "disk-io.h" +#include "transaction.h" +#include "print-tree.h" +#include "volumes.h" +#include "raid56.h" +#include "async-thread.h" +#include "check-integrity.h" +#include "rcu-string.h" + +/* set when additional merges to this rbio are not allowed */ +#define RBIO_RMW_LOCKED_BIT 1 + +struct btrfs_raid_bio { + struct btrfs_fs_info *fs_info; + struct btrfs_bio *bbio; + + /* + * logical block numbers for the start of each stripe + * The last one or two are p/q. These are sorted, + * so raid_map[0] is the start of our full stripe + */ + u64 *raid_map; + + /* while we're doing rmw on a stripe + * we put it into a hash table so we can + * lock the stripe and merge more rbios + * into it. + */ + struct list_head hash_list; + + /* + * for scheduling work in the helper threads + */ + struct btrfs_work work; + + /* + * bio list and bio_list_lock are used + * to add more bios into the stripe + * in hopes of avoiding the full rmw + */ + struct bio_list bio_list; + spinlock_t bio_list_lock; + + /* + * also protected by the bio_list_lock, the + * stripe locking code uses plug_list to hand off + * the stripe lock to the next pending IO + */ + struct list_head plug_list; + + /* + * flags that tell us if it is safe to + * merge with this bio + */ + unsigned long flags; + + /* size of each individual stripe on disk */ + int stripe_len; + + /* number of data stripes (no p/q) */ + int nr_data; + + /* + * set if we're doing a parity rebuild + * for a read from higher up, which is handled + * differently from a parity rebuild as part of + * rmw + */ + int read_rebuild; + + /* first bad stripe */ + int faila; + + /* second bad stripe (for raid6 use) */ + int failb; + + /* + * number of pages needed to represent the full + * stripe + */ + int nr_pages; + + /* + * size of all the bios in the bio_list. This + * helps us decide if the rbio maps to a full + * stripe or not + */ + int bio_list_bytes; + + atomic_t refs; + + /* + * these are two arrays of pointers. We allocate the + * rbio big enough to hold them both and setup their + * locations when the rbio is allocated + */ + + /* pointers to pages that we allocated for + * reading/writing stripes directly from the disk (including P/Q) + */ + struct page **stripe_pages; + + /* + * pointers to the pages in the bio_list. Stored + * here for faster lookup + */ + struct page **bio_pages; +}; + +static int __raid56_parity_recover(struct btrfs_raid_bio *rbio); +static noinline void finish_rmw(struct btrfs_raid_bio *rbio); +static void rmw_work(struct btrfs_work *work); +static void read_rebuild_work(struct btrfs_work *work); +static void async_rmw_stripe(struct btrfs_raid_bio *rbio); +static void async_read_rebuild(struct btrfs_raid_bio *rbio); +static int fail_bio_stripe(struct btrfs_raid_bio *rbio, struct bio *bio); +static int fail_rbio_index(struct btrfs_raid_bio *rbio, int failed); +static void __free_raid_bio(struct btrfs_raid_bio *rbio); +static void index_rbio_pages(struct btrfs_raid_bio *rbio); +static int alloc_rbio_pages(struct btrfs_raid_bio *rbio); + +/* + * the stripe hash table is used for locking, and to collect + * bios in hopes of making a full stripe + */ +int btrfs_alloc_stripe_hash_table(struct btrfs_fs_info *info) +{ + struct btrfs_stripe_hash_table *table; + struct btrfs_stripe_hash_table *x; + struct btrfs_stripe_hash *cur; + struct btrfs_stripe_hash *h; + int num_entries = 1 << BTRFS_STRIPE_HASH_TABLE_BITS; + int i; + + if (info->stripe_hash_table) + return 0; + + table = kzalloc(sizeof(*table) + sizeof(*h) * num_entries, GFP_NOFS); + if (!table) + return -ENOMEM; + + table->table = (void *)(table + 1); + h = table->table; + + for (i = 0; i < num_entries; i++) { + cur = h + i; + INIT_LIST_HEAD(&cur->hash_list); + spin_lock_init(&cur->lock); + init_waitqueue_head(&cur->wait); + } + + x = cmpxchg(&info->stripe_hash_table, NULL, table); + if (x) + kfree(x); + return 0; +} + +/* + * we hash on the first logical address of the stripe + */ +static int rbio_bucket(struct btrfs_raid_bio *rbio) +{ + u64 num = rbio->raid_map[0]; + + /* + * we shift down quite a bit. We're using byte + * addressing, and most of the lower bits are zeros. + * This tends to upset hash_64, and it consistently + * returns just one or two different values. + * + * shifting off the lower bits fixes things. + */ + return hash_64(num >> 16, BTRFS_STRIPE_HASH_TABLE_BITS); +} + +/* + * merging means we take the bio_list from the victim and + * splice it into the destination. The victim should + * be discarded afterwards. + * + * must be called with dest->rbio_list_lock held + */ +static void merge_rbio(struct btrfs_raid_bio *dest, + struct btrfs_raid_bio *victim) +{ + bio_list_merge(&dest->bio_list, &victim->bio_list); + dest->bio_list_bytes += victim->bio_list_bytes; + bio_list_init(&victim->bio_list); +} + +/* + * free the hash table used by unmount + */ +void btrfs_free_stripe_hash_table(struct btrfs_fs_info *info) +{ + if (!info->stripe_hash_table) + return; + kfree(info->stripe_hash_table); + info->stripe_hash_table = NULL; +} + +/* + * helper function to run the xor_blocks api. It is only + * able to do MAX_XOR_BLOCKS at a time, so we need to + * loop through. + */ +static void run_xor(void **pages, int src_cnt, ssize_t len) +{ + int src_off = 0; + int xor_src_cnt = 0; + void *dest = pages[src_cnt]; + + while(src_cnt > 0) { + xor_src_cnt = min(src_cnt, MAX_XOR_BLOCKS); + xor_blocks(xor_src_cnt, len, dest, pages + src_off); + + src_cnt -= xor_src_cnt; + src_off += xor_src_cnt; + } +} + +/* + * returns true if the bio list inside this rbio + * covers an entire stripe (no rmw required). + * Must be called with the bio list lock held, or + * at a time when you know it is impossible to add + * new bios into the list + */ +static int __rbio_is_full(struct btrfs_raid_bio *rbio) +{ + unsigned long size = rbio->bio_list_bytes; + int ret = 1; + + if (size != rbio->nr_data * rbio->stripe_len) + ret = 0; + + BUG_ON(size > rbio->nr_data * rbio->stripe_len); + return ret; +} + +static int rbio_is_full(struct btrfs_raid_bio *rbio) +{ + unsigned long flags; + int ret; + + spin_lock_irqsave(&rbio->bio_list_lock, flags); + ret = __rbio_is_full(rbio); + spin_unlock_irqrestore(&rbio->bio_list_lock, flags); + return ret; +} + +/* + * returns 1 if it is safe to merge two rbios together. + * The merging is safe if the two rbios correspond to + * the same stripe and if they are both going in the same + * direction (read vs write), and if neither one is + * locked for final IO + * + * The caller is responsible for locking such that + * rmw_locked is safe to test + */ +static int rbio_can_merge(struct btrfs_raid_bio *last, + struct btrfs_raid_bio *cur) +{ + if (test_bit(RBIO_RMW_LOCKED_BIT, &last->flags) || + test_bit(RBIO_RMW_LOCKED_BIT, &cur->flags)) + return 0; + + if (last->raid_map[0] != + cur->raid_map[0]) + return 0; + + /* reads can't merge with writes */ + if (last->read_rebuild != + cur->read_rebuild) { + return 0; + } + + return 1; +} + +/* + * helper to index into the pstripe + */ +static struct page *rbio_pstripe_page(struct btrfs_raid_bio *rbio, int index) +{ + index += (rbio->nr_data * rbio->stripe_len) >> PAGE_CACHE_SHIFT; + return rbio->stripe_pages[index]; +} + +/* + * helper to index into the qstripe, returns null + * if there is no qstripe + */ +static struct page *rbio_qstripe_page(struct btrfs_raid_bio *rbio, int index) +{ + if (rbio->nr_data + 1 == rbio->bbio->num_stripes) + return NULL; + + index += ((rbio->nr_data + 1) * rbio->stripe_len) >> + PAGE_CACHE_SHIFT; + return rbio->stripe_pages[index]; +} + +/* + * The first stripe in the table for a logical address + * has the lock. rbios are added in one of three ways: + * + * 1) Nobody has the stripe locked yet. The rbio is given + * the lock and 0 is returned. The caller must start the IO + * themselves. + * + * 2) Someone has the stripe locked, but we're able to merge + * with the lock owner. The rbio is freed and the IO will + * start automatically along with the existing rbio. 1 is returned. + * + * 3) Someone has the stripe locked, but we're not able to merge. + * The rbio is added to the lock owner's plug list, or merged into + * an rbio already on the plug list. When the lock owner unlocks, + * the next rbio on the list is run and the IO is started automatically. + * 1 is returned + * + * If we return 0, the caller still owns the rbio and must continue with + * IO submission. If we return 1, the caller must assume the rbio has + * already been freed. + */ +static noinline int lock_stripe_add(struct btrfs_raid_bio *rbio) +{ + int bucket = rbio_bucket(rbio); + struct btrfs_stripe_hash *h = rbio->fs_info->stripe_hash_table->table + bucket; + struct btrfs_raid_bio *cur; + struct btrfs_raid_bio *pending; + unsigned long flags; + DEFINE_WAIT(wait); + struct btrfs_raid_bio *freeit = NULL; + int ret = 0; + int walk = 0; + + spin_lock_irqsave(&h->lock, flags); + list_for_each_entry(cur, &h->hash_list, hash_list) { + walk++; + if (cur->raid_map[0] == rbio->raid_map[0]) { + spin_lock(&cur->bio_list_lock); + + /* can we merge into the lock owner? */ + if (rbio_can_merge(cur, rbio)) { + merge_rbio(cur, rbio); + spin_unlock(&cur->bio_list_lock); + freeit = rbio; + ret = 1; + goto out; + } + + /* + * we couldn't merge with the running + * rbio, see if we can merge with the + * pending ones. We don't have to + * check for rmw_locked because there + * is no way they are inside finish_rmw + * right now + */ + list_for_each_entry(pending, &cur->plug_list, + plug_list) { + if (rbio_can_merge(pending, rbio)) { + merge_rbio(pending, rbio); + spin_unlock(&cur->bio_list_lock); + freeit = rbio; + ret = 1; + goto out; + } + } + + /* no merging, put us on the tail of the plug list, + * our rbio will be started with the currently + * running rbio unlocks + */ + list_add_tail(&rbio->plug_list, &cur->plug_list); + spin_unlock(&cur->bio_list_lock); + ret = 1; + goto out; + } + } + + atomic_inc(&rbio->refs); + list_add(&rbio->hash_list, &h->hash_list); +out: + spin_unlock_irqrestore(&h->lock, flags); + if (freeit) + __free_raid_bio(freeit); + return ret; +} + +/* + * called as rmw or parity rebuild is completed. If the plug list has more + * rbios waiting for this stripe, the next one on the list will be started + */ +static noinline void unlock_stripe(struct btrfs_raid_bio *rbio) +{ + int bucket; + struct btrfs_stripe_hash *h; + unsigned long flags; + + bucket = rbio_bucket(rbio); + h = rbio->fs_info->stripe_hash_table->table + bucket; + + spin_lock_irqsave(&h->lock, flags); + spin_lock(&rbio->bio_list_lock); + + if (!list_empty(&rbio->hash_list)) { + + list_del_init(&rbio->hash_list); + atomic_dec(&rbio->refs); + + /* + * we use the plug list to hold all the rbios + * waiting for the chance to lock this stripe. + * hand the lock over to one of them. + */ + if (!list_empty(&rbio->plug_list)) { + struct btrfs_raid_bio *next; + struct list_head *head = rbio->plug_list.next; + + next = list_entry(head, struct btrfs_raid_bio, + plug_list); + + list_del_init(&rbio->plug_list); + + list_add(&next->hash_list, &h->hash_list); + atomic_inc(&next->refs); + spin_unlock(&rbio->bio_list_lock); + spin_unlock_irqrestore(&h->lock, flags); + + if (next->read_rebuild) + async_read_rebuild(next); + else + async_rmw_stripe(next); + + goto done_nolock; + + } else if (waitqueue_active(&h->wait)) { + spin_unlock(&rbio->bio_list_lock); + spin_unlock_irqrestore(&h->lock, flags); + wake_up(&h->wait); + goto done_nolock; + } + } + spin_unlock(&rbio->bio_list_lock); + spin_unlock_irqrestore(&h->lock, flags); + +done_nolock: + return; +} + +static void __free_raid_bio(struct btrfs_raid_bio *rbio) +{ + int i; + + WARN_ON(atomic_read(&rbio->refs) < 0); + if (!atomic_dec_and_test(&rbio->refs)) + return; + + WARN_ON(!list_empty(&rbio->hash_list)); + WARN_ON(!bio_list_empty(&rbio->bio_list)); + + for (i = 0; i < rbio->nr_pages; i++) { + if (rbio->stripe_pages[i]) { + __free_page(rbio->stripe_pages[i]); + rbio->stripe_pages[i] = NULL; + } + } + kfree(rbio->raid_map); + kfree(rbio->bbio); + kfree(rbio); +} + +static void free_raid_bio(struct btrfs_raid_bio *rbio) +{ + unlock_stripe(rbio); + __free_raid_bio(rbio); +} + +/* + * this frees the rbio and runs through all the bios in the + * bio_list and calls end_io on them + */ +static void rbio_orig_end_io(struct btrfs_raid_bio *rbio, int err, int uptodate) +{ + struct bio *cur = bio_list_get(&rbio->bio_list); + struct bio *next; + free_raid_bio(rbio); + + while (cur) { + next = cur->bi_next; + cur->bi_next = NULL; + if (uptodate) + set_bit(BIO_UPTODATE, &cur->bi_flags); + bio_endio(cur, err); + cur = next; + } +} + +/* + * end io function used by finish_rmw. When we finally + * get here, we've written a full stripe + */ +static void raid_write_end_io(struct bio *bio, int err) +{ + struct btrfs_raid_bio *rbio = bio->bi_private; + + if (err) + fail_bio_stripe(rbio, bio); + + bio_put(bio); + + if (!atomic_dec_and_test(&rbio->bbio->stripes_pending)) + return; + + err = 0; + + /* OK, we have read all the stripes we need to. */ + if (atomic_read(&rbio->bbio->error) > rbio->bbio->max_errors) + err = -EIO; + + rbio_orig_end_io(rbio, err, 0); + return; +} + +/* + * the read/modify/write code wants to use the original bio for + * any pages it included, and then use the rbio for everything + * else. This function decides if a given index (stripe number) + * and page number in that stripe fall inside the original bio + * or the rbio. + * + * if you set bio_list_only, you'll get a NULL back for any ranges + * that are outside the bio_list + * + * This doesn't take any refs on anything, you get a bare page pointer + * and the caller must bump refs as required. + * + * You must call index_rbio_pages once before you can trust + * the answers from this function. + */ +static struct page *page_in_rbio(struct btrfs_raid_bio *rbio, + int index, int pagenr, int bio_list_only) +{ + int chunk_page; + struct page *p = NULL; + + chunk_page = index * (rbio->stripe_len >> PAGE_SHIFT) + pagenr; + + spin_lock_irq(&rbio->bio_list_lock); + p = rbio->bio_pages[chunk_page]; + spin_unlock_irq(&rbio->bio_list_lock); + + if (p || bio_list_only) + return p; + + return rbio->stripe_pages[chunk_page]; +} + +/* + * number of pages we need for the entire stripe across all the + * drives + */ +static unsigned long rbio_nr_pages(unsigned long stripe_len, int nr_stripes) +{ + unsigned long nr = stripe_len * nr_stripes; + return (nr + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT; +} + +/* + * allocation and initial setup for the btrfs_raid_bio. Not + * this does not allocate any pages for rbio->pages. + */ +static struct btrfs_raid_bio *alloc_rbio(struct btrfs_root *root, + struct btrfs_bio *bbio, u64 *raid_map, + u64 stripe_len) +{ + struct btrfs_raid_bio *rbio; + int nr_data = 0; + int num_pages = rbio_nr_pages(stripe_len, bbio->num_stripes); + void *p; + + rbio = kzalloc(sizeof(*rbio) + num_pages * sizeof(struct page *) * 2, + GFP_NOFS); + if (!rbio) { + kfree(raid_map); + kfree(bbio); + return ERR_PTR(-ENOMEM); + } + + bio_list_init(&rbio->bio_list); + INIT_LIST_HEAD(&rbio->plug_list); + spin_lock_init(&rbio->bio_list_lock); + INIT_LIST_HEAD(&rbio->hash_list); + rbio->bbio = bbio; + rbio->raid_map = raid_map; + rbio->fs_info = root->fs_info; + rbio->stripe_len = stripe_len; + rbio->nr_pages = num_pages; + rbio->faila = -1; + rbio->failb = -1; + atomic_set(&rbio->refs, 1); + + /* + * the stripe_pages and bio_pages array point to the extra + * memory we allocated past the end of the rbio + */ + p = rbio + 1; + rbio->stripe_pages = p; + rbio->bio_pages = p + sizeof(struct page *) * num_pages; + + if (raid_map[bbio->num_stripes - 1] == RAID6_Q_STRIPE) + nr_data = bbio->num_stripes - 2; + else + nr_data = bbio->num_stripes - 1; + + rbio->nr_data = nr_data; + return rbio; +} + +/* allocate pages for all the stripes in the bio, including parity */ +static int alloc_rbio_pages(struct btrfs_raid_bio *rbio) +{ + int i; + struct page *page; + + for (i = 0; i < rbio->nr_pages; i++) { + if (rbio->stripe_pages[i]) + continue; + page = alloc_page(GFP_NOFS | __GFP_HIGHMEM); + if (!page) + return -ENOMEM; + rbio->stripe_pages[i] = page; + ClearPageUptodate(page); + } + return 0; +} + +/* allocate pages for just the p/q stripes */ +static int alloc_rbio_parity_pages(struct btrfs_raid_bio *rbio) +{ + int i; + struct page *page; + + i = (rbio->nr_data * rbio->stripe_len) >> PAGE_CACHE_SHIFT; + + for (; i < rbio->nr_pages; i++) { + if (rbio->stripe_pages[i]) + continue; + page = alloc_page(GFP_NOFS | __GFP_HIGHMEM); + if (!page) + return -ENOMEM; + rbio->stripe_pages[i] = page; + } + return 0; +} + +/* + * add a single page from a specific stripe into our list of bios for IO + * this will try to merge into existing bios if possible, and returns + * zero if all went well. + */ +int rbio_add_io_page(struct btrfs_raid_bio *rbio, + struct bio_list *bio_list, + struct page *page, + int stripe_nr, + unsigned long page_index, + unsigned long bio_max_len) +{ + struct bio *last = bio_list->tail; + u64 last_end = 0; + int ret; + struct bio *bio; + struct btrfs_bio_stripe *stripe; + u64 disk_start; + + stripe = &rbio->bbio->stripes[stripe_nr]; + disk_start = stripe->physical + (page_index << PAGE_CACHE_SHIFT); + + /* if the device is missing, just fail this stripe */ + if (!stripe->dev->bdev) + return fail_rbio_index(rbio, stripe_nr); + + /* see if we can add this page onto our existing bio */ + if (last) { + last_end = (u64)last->bi_sector << 9; + last_end += last->bi_size; + + /* + * we can't merge these if they are from different + * devices or if they are not contiguous + */ + if (last_end == disk_start && stripe->dev->bdev && + test_bit(BIO_UPTODATE, &last->bi_flags) && + last->bi_bdev == stripe->dev->bdev) { + ret = bio_add_page(last, page, PAGE_CACHE_SIZE, 0); + if (ret == PAGE_CACHE_SIZE) + return 0; + } + } + + /* put a new bio on the list */ + bio = bio_alloc(GFP_NOFS, bio_max_len >> PAGE_SHIFT?:1); + if (!bio) + return -ENOMEM; + + bio->bi_size = 0; + bio->bi_bdev = stripe->dev->bdev; + bio->bi_sector = disk_start >> 9; + set_bit(BIO_UPTODATE, &bio->bi_flags); + + bio_add_page(bio, page, PAGE_CACHE_SIZE, 0); + bio_list_add(bio_list, bio); + return 0; +} + +/* + * while we're doing the read/modify/write cycle, we could + * have errors in reading pages off the disk. This checks + * for errors and if we're not able to read the page it'll + * trigger parity reconstruction. The rmw will be finished + * after we've reconstructed the failed stripes + */ +static void validate_rbio_for_rmw(struct btrfs_raid_bio *rbio) +{ + if (rbio->faila >= 0 || rbio->failb >= 0) { + BUG_ON(rbio->faila == rbio->bbio->num_stripes - 1); + __raid56_parity_recover(rbio); + } else { + finish_rmw(rbio); + } +} + +/* + * these are just the pages from the rbio array, not from anything + * the FS sent down to us + */ +static struct page *rbio_stripe_page(struct btrfs_raid_bio *rbio, int stripe, int page) +{ + int index; + index = stripe * (rbio->stripe_len >> PAGE_CACHE_SHIFT); + index += page; + return rbio->stripe_pages[index]; +} + +/* + * helper function to walk our bio list and populate the bio_pages array with + * the result. This seems expensive, but it is faster than constantly + * searching through the bio list as we setup the IO in finish_rmw or stripe + * reconstruction. + * + * This must be called before you trust the answers from page_in_rbio + */ +static void index_rbio_pages(struct btrfs_raid_bio *rbio) +{ + struct bio *bio; + u64 start; + unsigned long stripe_offset; + unsigned long page_index; + struct page *p; + int i; + + spin_lock_irq(&rbio->bio_list_lock); + bio_list_for_each(bio, &rbio->bio_list) { + start = (u64)bio->bi_sector << 9; + stripe_offset = start - rbio->raid_map[0]; + page_index = stripe_offset >> PAGE_CACHE_SHIFT; + + for (i = 0; i < bio->bi_vcnt; i++) { + p = bio->bi_io_vec[i].bv_page; + rbio->bio_pages[page_index + i] = p; + } + } + spin_unlock_irq(&rbio->bio_list_lock); +} + +/* + * this is called from one of two situations. We either + * have a full stripe from the higher layers, or we've read all + * the missing bits off disk. + * + * This will calculate the parity and then send down any + * changed blocks. + */ +static noinline void finish_rmw(struct btrfs_raid_bio *rbio) +{ + struct btrfs_bio *bbio = rbio->bbio; + void *pointers[bbio->num_stripes]; + int stripe_len = rbio->stripe_len; + int nr_data = rbio->nr_data; + int stripe; + int pagenr; + int p_stripe = -1; + int q_stripe = -1; + struct bio_list bio_list; + struct bio *bio; + int pages_per_stripe = stripe_len >> PAGE_CACHE_SHIFT; + int ret; + + bio_list_init(&bio_list); + + if (bbio->num_stripes - rbio->nr_data == 1) { + p_stripe = bbio->num_stripes - 1; + } else if (bbio->num_stripes - rbio->nr_data == 2) { + p_stripe = bbio->num_stripes - 2; + q_stripe = bbio->num_stripes - 1; + } else { + BUG(); + } + + /* at this point we either have a full stripe, + * or we've read the full stripe from the drive. + * recalculate the parity and write the new results. + * + * We're not allowed to add any new bios to the + * bio list here, anyone else that wants to + * change this stripe needs to do their own rmw. + */ + spin_lock_irq(&rbio->bio_list_lock); + set_bit(RBIO_RMW_LOCKED_BIT, &rbio->flags); + spin_unlock_irq(&rbio->bio_list_lock); + + atomic_set(&rbio->bbio->error, 0); + + /* + * now that we've set rmw_locked, run through the + * bio list one last time and map the page pointers + */ + index_rbio_pages(rbio); + + for (pagenr = 0; pagenr < pages_per_stripe; pagenr++) { + struct page *p; + /* first collect one page from each data stripe */ + for (stripe = 0; stripe < nr_data; stripe++) { + p = page_in_rbio(rbio, stripe, pagenr, 0); + pointers[stripe] = kmap(p); + } + + /* then add the parity stripe */ + p = rbio_pstripe_page(rbio, pagenr); + SetPageUptodate(p); + pointers[stripe++] = kmap(p); + + if (q_stripe != -1) { + + /* + * raid6, add the qstripe and call the + * library function to fill in our p/q + */ + p = rbio_qstripe_page(rbio, pagenr); + SetPageUptodate(p); + pointers[stripe++] = kmap(p); + + raid6_call.gen_syndrome(bbio->num_stripes, PAGE_SIZE, + pointers); + } else { + /* raid5 */ + memcpy(pointers[nr_data], pointers[0], PAGE_SIZE); + run_xor(pointers + 1, nr_data - 1, PAGE_CACHE_SIZE); + } + + + for (stripe = 0; stripe < bbio->num_stripes; stripe++) + kunmap(page_in_rbio(rbio, stripe, pagenr, 0)); + } + + /* + * time to start writing. Make bios for everything from the + * higher layers (the bio_list in our rbio) and our p/q. Ignore + * everything else. + */ + for (stripe = 0; stripe < bbio->num_stripes; stripe++) { + for (pagenr = 0; pagenr < pages_per_stripe; pagenr++) { + struct page *page; + if (stripe < rbio->nr_data) { + page = page_in_rbio(rbio, stripe, pagenr, 1); + if (!page) + continue; + } else { + page = rbio_stripe_page(rbio, stripe, pagenr); + } + + ret = rbio_add_io_page(rbio, &bio_list, + page, stripe, pagenr, rbio->stripe_len); + if (ret) + goto cleanup; + } + } + + atomic_set(&bbio->stripes_pending, bio_list_size(&bio_list)); + BUG_ON(atomic_read(&bbio->stripes_pending) == 0); + + while (1) { + bio = bio_list_pop(&bio_list); + if (!bio) + break; + + bio->bi_private = rbio; + bio->bi_end_io = raid_write_end_io; + BUG_ON(!test_bit(BIO_UPTODATE, &bio->bi_flags)); + submit_bio(WRITE, bio); + } + return; + +cleanup: + rbio_orig_end_io(rbio, -EIO, 0); +} + +/* + * helper to find the stripe number for a given bio. Used to figure out which + * stripe has failed. This expects the bio to correspond to a physical disk, + * so it looks up based on physical sector numbers. + */ +static int find_bio_stripe(struct btrfs_raid_bio *rbio, + struct bio *bio) +{ + u64 physical = bio->bi_sector; + u64 stripe_start; + int i; + struct btrfs_bio_stripe *stripe; + + physical <<= 9; + + for (i = 0; i < rbio->bbio->num_stripes; i++) { + stripe = &rbio->bbio->stripes[i]; + stripe_start = stripe->physical; + if (physical >= stripe_start && + physical < stripe_start + rbio->stripe_len) { + return i; + } + } + return -1; +} + +/* + * helper to find the stripe number for a given + * bio (before mapping). Used to figure out which stripe has + * failed. This looks up based on logical block numbers. + */ +static int find_logical_bio_stripe(struct btrfs_raid_bio *rbio, + struct bio *bio) +{ + u64 logical = bio->bi_sector; + u64 stripe_start; + int i; + + logical <<= 9; + + for (i = 0; i < rbio->nr_data; i++) { + stripe_start = rbio->raid_map[i]; + if (logical >= stripe_start && + logical < stripe_start + rbio->stripe_len) { + return i; + } + } + return -1; +} + +/* + * returns -EIO if we had too many failures + */ +static int fail_rbio_index(struct btrfs_raid_bio *rbio, int failed) +{ + unsigned long flags; + int ret = 0; + + spin_lock_irqsave(&rbio->bio_list_lock, flags); + + /* we already know this stripe is bad, move on */ + if (rbio->faila == failed || rbio->failb == failed) + goto out; + + if (rbio->faila == -1) { + /* first failure on this rbio */ + rbio->faila = failed; + atomic_inc(&rbio->bbio->error); + } else if (rbio->failb == -1) { + /* second failure on this rbio */ + rbio->failb = failed; + atomic_inc(&rbio->bbio->error); + } else { + ret = -EIO; + } +out: + spin_unlock_irqrestore(&rbio->bio_list_lock, flags); + + return ret; +} + +/* + * helper to fail a stripe based on a physical disk + * bio. + */ +static int fail_bio_stripe(struct btrfs_raid_bio *rbio, + struct bio *bio) +{ + int failed = find_bio_stripe(rbio, bio); + + if (failed < 0) + return -EIO; + + return fail_rbio_index(rbio, failed); +} + +/* + * this sets each page in the bio uptodate. It should only be used on private + * rbio pages, nothing that comes in from the higher layers + */ +static void set_bio_pages_uptodate(struct bio *bio) +{ + int i; + struct page *p; + + for (i = 0; i < bio->bi_vcnt; i++) { + p = bio->bi_io_vec[i].bv_page; + SetPageUptodate(p); + } +} + +/* + * end io for the read phase of the rmw cycle. All the bios here are physical + * stripe bios we've read from the disk so we can recalculate the parity of the + * stripe. + * + * This will usually kick off finish_rmw once all the bios are read in, but it + * may trigger parity reconstruction if we had any errors along the way + */ +static void raid_rmw_end_io(struct bio *bio, int err) +{ + struct btrfs_raid_bio *rbio = bio->bi_private; + + if (err) + fail_bio_stripe(rbio, bio); + else + set_bio_pages_uptodate(bio); + + bio_put(bio); + + if (!atomic_dec_and_test(&rbio->bbio->stripes_pending)) + return; + + err = 0; + if (atomic_read(&rbio->bbio->error) > rbio->bbio->max_errors) + goto cleanup; + + /* + * this will normally call finish_rmw to start our write + * but if there are any failed stripes we'll reconstruct + * from parity first + */ + validate_rbio_for_rmw(rbio); + return; + +cleanup: + + rbio_orig_end_io(rbio, -EIO, 0); +} + +static void async_rmw_stripe(struct btrfs_raid_bio *rbio) +{ + rbio->work.flags = 0; + rbio->work.func = rmw_work; + + btrfs_queue_worker(&rbio->fs_info->rmw_workers, + &rbio->work); +} + +static void async_read_rebuild(struct btrfs_raid_bio *rbio) +{ + rbio->work.flags = 0; + rbio->work.func = read_rebuild_work; + + btrfs_queue_worker(&rbio->fs_info->rmw_workers, + &rbio->work); +} + +/* + * the stripe must be locked by the caller. It will + * unlock after all the writes are done + */ +static int raid56_rmw_stripe(struct btrfs_raid_bio *rbio) +{ + int bios_to_read = 0; + struct btrfs_bio *bbio = rbio->bbio; + struct bio_list bio_list; + int ret; + int nr_pages = (rbio->stripe_len + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT; + int pagenr; + int stripe; + struct bio *bio; + + bio_list_init(&bio_list); + + ret = alloc_rbio_pages(rbio); + if (ret) + goto cleanup; + + index_rbio_pages(rbio); + + atomic_set(&rbio->bbio->error, 0); + /* + * build a list of bios to read all the missing parts of this + * stripe + */ + for (stripe = 0; stripe < rbio->nr_data; stripe++) { + for (pagenr = 0; pagenr < nr_pages; pagenr++) { + struct page *page; + /* + * we want to find all the pages missing from + * the rbio and read them from the disk. If + * page_in_rbio finds a page in the bio list + * we don't need to read it off the stripe. + */ + page = page_in_rbio(rbio, stripe, pagenr, 1); + if (page) + continue; + + page = rbio_stripe_page(rbio, stripe, pagenr); + ret = rbio_add_io_page(rbio, &bio_list, page, + stripe, pagenr, rbio->stripe_len); + if (ret) + goto cleanup; + } + } + + bios_to_read = bio_list_size(&bio_list); + if (!bios_to_read) { + /* + * this can happen if others have merged with + * us, it means there is nothing left to read. + * But if there are missing devices it may not be + * safe to do the full stripe write yet. + */ + goto finish; + } + + /* + * the bbio may be freed once we submit the last bio. Make sure + * not to touch it after that + */ + atomic_set(&bbio->stripes_pending, bios_to_read); + while (1) { + bio = bio_list_pop(&bio_list); + if (!bio) + break; + + bio->bi_private = rbio; + bio->bi_end_io = raid_rmw_end_io; + + btrfs_bio_wq_end_io(rbio->fs_info, bio, + BTRFS_WQ_ENDIO_RAID56); + + BUG_ON(!test_bit(BIO_UPTODATE, &bio->bi_flags)); + submit_bio(READ, bio); + } + /* the actual write will happen once the reads are done */ + return 0; + +cleanup: + rbio_orig_end_io(rbio, -EIO, 0); + return -EIO; + +finish: + validate_rbio_for_rmw(rbio); + return 0; +} + +/* + * if the upper layers pass in a full stripe, we thank them by only allocating + * enough pages to hold the parity, and sending it all down quickly. + */ +static int full_stripe_write(struct btrfs_raid_bio *rbio) +{ + int ret; + + ret = alloc_rbio_parity_pages(rbio); + if (ret) + return ret; + + ret = lock_stripe_add(rbio); + if (ret == 0) + finish_rmw(rbio); + return 0; +} + +/* + * partial stripe writes get handed over to async helpers. + * We're really hoping to merge a few more writes into this + * rbio before calculating new parity + */ +static int partial_stripe_write(struct btrfs_raid_bio *rbio) +{ + int ret; + + ret = lock_stripe_add(rbio); + if (ret == 0) + async_rmw_stripe(rbio); + return 0; +} + +/* + * sometimes while we were reading from the drive to + * recalculate parity, enough new bios come into create + * a full stripe. So we do a check here to see if we can + * go directly to finish_rmw + */ +static int __raid56_parity_write(struct btrfs_raid_bio *rbio) +{ + /* head off into rmw land if we don't have a full stripe */ + if (!rbio_is_full(rbio)) + return partial_stripe_write(rbio); + return full_stripe_write(rbio); +} + +/* + * our main entry point for writes from the rest of the FS. + */ +int raid56_parity_write(struct btrfs_root *root, struct bio *bio, + struct btrfs_bio *bbio, u64 *raid_map, + u64 stripe_len) +{ + struct btrfs_raid_bio *rbio; + + rbio = alloc_rbio(root, bbio, raid_map, stripe_len); + if (IS_ERR(rbio)) { + kfree(raid_map); + kfree(bbio); + return PTR_ERR(rbio); + } + bio_list_add(&rbio->bio_list, bio); + rbio->bio_list_bytes = bio->bi_size; + return __raid56_parity_write(rbio); +} + +/* + * all parity reconstruction happens here. We've read in everything + * we can find from the drives and this does the heavy lifting of + * sorting the good from the bad. + */ +static void __raid_recover_end_io(struct btrfs_raid_bio *rbio) +{ + int pagenr, stripe; + void **pointers; + int faila = -1, failb = -1; + int nr_pages = (rbio->stripe_len + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT; + struct page *page; + int err; + int i; + + pointers = kzalloc(rbio->bbio->num_stripes * sizeof(void *), + GFP_NOFS); + if (!pointers) { + err = -ENOMEM; + goto cleanup_io; + } + + faila = rbio->faila; + failb = rbio->failb; + + if (rbio->read_rebuild) { + spin_lock_irq(&rbio->bio_list_lock); + set_bit(RBIO_RMW_LOCKED_BIT, &rbio->flags); + spin_unlock_irq(&rbio->bio_list_lock); + } + + index_rbio_pages(rbio); + + for (pagenr = 0; pagenr < nr_pages; pagenr++) { + /* setup our array of pointers with pages + * from each stripe + */ + for (stripe = 0; stripe < rbio->bbio->num_stripes; stripe++) { + /* + * if we're rebuilding a read, we have to use + * pages from the bio list + */ + if (rbio->read_rebuild && + (stripe == faila || stripe == failb)) { + page = page_in_rbio(rbio, stripe, pagenr, 0); + } else { + page = rbio_stripe_page(rbio, stripe, pagenr); + } + pointers[stripe] = kmap(page); + } + + /* all raid6 handling here */ + if (rbio->raid_map[rbio->bbio->num_stripes - 1] == + RAID6_Q_STRIPE) { + + /* + * single failure, rebuild from parity raid5 + * style + */ + if (failb < 0) { + if (faila == rbio->nr_data) { + /* + * Just the P stripe has failed, without + * a bad data or Q stripe. + * TODO, we should redo the xor here. + */ + err = -EIO; + goto cleanup; + } + /* + * a single failure in raid6 is rebuilt + * in the pstripe code below + */ + goto pstripe; + } + + /* make sure our ps and qs are in order */ + if (faila > failb) { + int tmp = failb; + failb = faila; + faila = tmp; + } + + /* if the q stripe is failed, do a pstripe reconstruction + * from the xors. + * If both the q stripe and the P stripe are failed, we're + * here due to a crc mismatch and we can't give them the + * data they want + */ + if (rbio->raid_map[failb] == RAID6_Q_STRIPE) { + if (rbio->raid_map[faila] == RAID5_P_STRIPE) { + err = -EIO; + goto cleanup; + } + /* + * otherwise we have one bad data stripe and + * a good P stripe. raid5! + */ + goto pstripe; + } + + if (rbio->raid_map[failb] == RAID5_P_STRIPE) { + raid6_datap_recov(rbio->bbio->num_stripes, + PAGE_SIZE, faila, pointers); + } else { + raid6_2data_recov(rbio->bbio->num_stripes, + PAGE_SIZE, faila, failb, + pointers); + } + } else { + void *p; + + /* rebuild from P stripe here (raid5 or raid6) */ + BUG_ON(failb != -1); +pstripe: + /* Copy parity block into failed block to start with */ + memcpy(pointers[faila], + pointers[rbio->nr_data], + PAGE_CACHE_SIZE); + + /* rearrange the pointer array */ + p = pointers[faila]; + for (stripe = faila; stripe < rbio->nr_data - 1; stripe++) + pointers[stripe] = pointers[stripe + 1]; + pointers[rbio->nr_data - 1] = p; + + /* xor in the rest */ + run_xor(pointers, rbio->nr_data - 1, PAGE_CACHE_SIZE); + } + /* if we're doing this rebuild as part of an rmw, go through + * and set all of our private rbio pages in the + * failed stripes as uptodate. This way finish_rmw will + * know they can be trusted. If this was a read reconstruction, + * other endio functions will fiddle the uptodate bits + */ + if (!rbio->read_rebuild) { + for (i = 0; i < nr_pages; i++) { + if (faila != -1) { + page = rbio_stripe_page(rbio, faila, i); + SetPageUptodate(page); + } + if (failb != -1) { + page = rbio_stripe_page(rbio, failb, i); + SetPageUptodate(page); + } + } + } + for (stripe = 0; stripe < rbio->bbio->num_stripes; stripe++) { + /* + * if we're rebuilding a read, we have to use + * pages from the bio list + */ + if (rbio->read_rebuild && + (stripe == faila || stripe == failb)) { + page = page_in_rbio(rbio, stripe, pagenr, 0); + } else { + page = rbio_stripe_page(rbio, stripe, pagenr); + } + kunmap(page); + } + } + + err = 0; +cleanup: + kfree(pointers); + +cleanup_io: + + if (rbio->read_rebuild) { + rbio_orig_end_io(rbio, err, err == 0); + } else if (err == 0) { + rbio->faila = -1; + rbio->failb = -1; + finish_rmw(rbio); + } else { + rbio_orig_end_io(rbio, err, 0); + } +} + +/* + * This is called only for stripes we've read from disk to + * reconstruct the parity. + */ +static void raid_recover_end_io(struct bio *bio, int err) +{ + struct btrfs_raid_bio *rbio = bio->bi_private; + + /* + * we only read stripe pages off the disk, set them + * up to date if there were no errors + */ + if (err) + fail_bio_stripe(rbio, bio); + else + set_bio_pages_uptodate(bio); + bio_put(bio); + + if (!atomic_dec_and_test(&rbio->bbio->stripes_pending)) + return; + + if (atomic_read(&rbio->bbio->error) > rbio->bbio->max_errors) + rbio_orig_end_io(rbio, -EIO, 0); + else + __raid_recover_end_io(rbio); +} + +/* + * reads everything we need off the disk to reconstruct + * the parity. endio handlers trigger final reconstruction + * when the IO is done. + * + * This is used both for reads from the higher layers and for + * parity construction required to finish a rmw cycle. + */ +static int __raid56_parity_recover(struct btrfs_raid_bio *rbio) +{ + int bios_to_read = 0; + struct btrfs_bio *bbio = rbio->bbio; + struct bio_list bio_list; + int ret; + int nr_pages = (rbio->stripe_len + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT; + int pagenr; + int stripe; + struct bio *bio; + + bio_list_init(&bio_list); + + ret = alloc_rbio_pages(rbio); + if (ret) + goto cleanup; + + atomic_set(&rbio->bbio->error, 0); + + /* + * read everything that hasn't failed. + */ + for (stripe = 0; stripe < bbio->num_stripes; stripe++) { + if (rbio->faila == stripe || + rbio->failb == stripe) + continue; + + for (pagenr = 0; pagenr < nr_pages; pagenr++) { + struct page *p; + + /* + * the rmw code may have already read this + * page in + */ + p = rbio_stripe_page(rbio, stripe, pagenr); + if (PageUptodate(p)) + continue; + + ret = rbio_add_io_page(rbio, &bio_list, + rbio_stripe_page(rbio, stripe, pagenr), + stripe, pagenr, rbio->stripe_len); + if (ret < 0) + goto cleanup; + } + } + + bios_to_read = bio_list_size(&bio_list); + if (!bios_to_read) { + /* + * we might have no bios to read just because the pages + * were up to date, or we might have no bios to read because + * the devices were gone. + */ + if (atomic_read(&rbio->bbio->error) <= rbio->bbio->max_errors) { + __raid_recover_end_io(rbio); + goto out; + } else { + goto cleanup; + } + } + + /* + * the bbio may be freed once we submit the last bio. Make sure + * not to touch it after that + */ + atomic_set(&bbio->stripes_pending, bios_to_read); + while (1) { + bio = bio_list_pop(&bio_list); + if (!bio) + break; + + bio->bi_private = rbio; + bio->bi_end_io = raid_recover_end_io; + + btrfs_bio_wq_end_io(rbio->fs_info, bio, + BTRFS_WQ_ENDIO_RAID56); + + BUG_ON(!test_bit(BIO_UPTODATE, &bio->bi_flags)); + submit_bio(READ, bio); + } +out: + return 0; + +cleanup: + if (rbio->read_rebuild) + rbio_orig_end_io(rbio, -EIO, 0); + return -EIO; +} + +/* + * the main entry point for reads from the higher layers. This + * is really only called when the normal read path had a failure, + * so we assume the bio they send down corresponds to a failed part + * of the drive. + */ +int raid56_parity_recover(struct btrfs_root *root, struct bio *bio, + struct btrfs_bio *bbio, u64 *raid_map, + u64 stripe_len, int mirror_num) +{ + struct btrfs_raid_bio *rbio; + int ret; + + rbio = alloc_rbio(root, bbio, raid_map, stripe_len); + if (IS_ERR(rbio)) { + return PTR_ERR(rbio); + } + + rbio->read_rebuild = 1; + bio_list_add(&rbio->bio_list, bio); + rbio->bio_list_bytes = bio->bi_size; + + rbio->faila = find_logical_bio_stripe(rbio, bio); + if (rbio->faila == -1) { + BUG(); + kfree(rbio); + return -EIO; + } + + /* + * reconstruct from the q stripe if they are + * asking for mirror 3 + */ + if (mirror_num == 3) + rbio->failb = bbio->num_stripes - 2; + + ret = lock_stripe_add(rbio); + + /* + * __raid56_parity_recover will end the bio with + * any errors it hits. We don't want to return + * its error value up the stack because our caller + * will end up calling bio_endio with any nonzero + * return + */ + if (ret == 0) + __raid56_parity_recover(rbio); + /* + * our rbio has been added to the list of + * rbios that will be handled after the + * currently lock owner is done + */ + return 0; + +} + +static void rmw_work(struct btrfs_work *work) +{ + struct btrfs_raid_bio *rbio; + + rbio = container_of(work, struct btrfs_raid_bio, work); + raid56_rmw_stripe(rbio); +} + +static void read_rebuild_work(struct btrfs_work *work) +{ + struct btrfs_raid_bio *rbio; + + rbio = container_of(work, struct btrfs_raid_bio, work); + __raid56_parity_recover(rbio); +} diff --git a/fs/btrfs/raid56.h b/fs/btrfs/raid56.h new file mode 100644 index 000000000000..ea5d73bfdfbe --- /dev/null +++ b/fs/btrfs/raid56.h @@ -0,0 +1,51 @@ +/* + * Copyright (C) 2012 Fusion-io All rights reserved. + * Copyright (C) 2012 Intel Corp. All rights reserved. + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public + * License v2 as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + * + * You should have received a copy of the GNU General Public + * License along with this program; if not, write to the + * Free Software Foundation, Inc., 59 Temple Place - Suite 330, + * Boston, MA 021110-1307, USA. + */ + +#ifndef __BTRFS_RAID56__ +#define __BTRFS_RAID56__ +static inline int nr_parity_stripes(struct map_lookup *map) +{ + if (map->type & BTRFS_BLOCK_GROUP_RAID5) + return 1; + else if (map->type & BTRFS_BLOCK_GROUP_RAID6) + return 2; + else + return 0; +} + +static inline int nr_data_stripes(struct map_lookup *map) +{ + return map->num_stripes - nr_parity_stripes(map); +} +#define RAID5_P_STRIPE ((u64)-2) +#define RAID6_Q_STRIPE ((u64)-1) + +#define is_parity_stripe(x) (((x) == RAID5_P_STRIPE) || \ + ((x) == RAID6_Q_STRIPE)) + +int raid56_parity_recover(struct btrfs_root *root, struct bio *bio, + struct btrfs_bio *bbio, u64 *raid_map, + u64 stripe_len, int mirror_num); +int raid56_parity_write(struct btrfs_root *root, struct bio *bio, + struct btrfs_bio *bbio, u64 *raid_map, + u64 stripe_len); + +int btrfs_alloc_stripe_hash_table(struct btrfs_fs_info *info); +void btrfs_free_stripe_hash_table(struct btrfs_fs_info *info); +#endif diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c index bdbb94f245c9..bc35ed4238b8 100644 --- a/fs/btrfs/scrub.c +++ b/fs/btrfs/scrub.c @@ -28,6 +28,7 @@ #include "dev-replace.h" #include "check-integrity.h" #include "rcu-string.h" +#include "raid56.h" /* * This is only the first step towards a full-features scrub. It reads all @@ -2246,6 +2247,13 @@ static noinline_for_stack int scrub_stripe(struct scrub_ctx *sctx, struct btrfs_device *extent_dev; int extent_mirror_num; + if (map->type & (BTRFS_BLOCK_GROUP_RAID5 | + BTRFS_BLOCK_GROUP_RAID6)) { + if (num >= nr_data_stripes(map)) { + return 0; + } + } + nstripes = length; offset = 0; do_div(nstripes, map->stripe_len); diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c index 87fac9a21ea5..a065dec0e330 100644 --- a/fs/btrfs/transaction.c +++ b/fs/btrfs/transaction.c @@ -686,7 +686,9 @@ int btrfs_write_marked_extents(struct btrfs_root *root, struct extent_state *cached_state = NULL; u64 start = 0; u64 end; + struct blk_plug plug; + blk_start_plug(&plug); while (!find_first_extent_bit(dirty_pages, start, &start, &end, mark, &cached_state)) { convert_extent_bit(dirty_pages, start, end, EXTENT_NEED_WAIT, @@ -700,6 +702,7 @@ int btrfs_write_marked_extents(struct btrfs_root *root, } if (err) werr = err; + blk_finish_plug(&plug); return werr; } diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c index 485a5423e3c6..c372264b85bf 100644 --- a/fs/btrfs/volumes.c +++ b/fs/btrfs/volumes.c @@ -25,6 +25,8 @@ #include #include #include +#include +#include #include "compat.h" #include "ctree.h" #include "extent_map.h" @@ -32,6 +34,7 @@ #include "transaction.h" #include "print-tree.h" #include "volumes.h" +#include "raid56.h" #include "async-thread.h" #include "check-integrity.h" #include "rcu-string.h" @@ -1389,6 +1392,14 @@ int btrfs_rm_device(struct btrfs_root *root, char *device_path) } btrfs_dev_replace_unlock(&root->fs_info->dev_replace); + if ((all_avail & (BTRFS_BLOCK_GROUP_RAID5 | + BTRFS_BLOCK_GROUP_RAID6) && num_devices <= 3)) { + printk(KERN_ERR "btrfs: unable to go below three devices " + "on raid5 or raid6\n"); + ret = -EINVAL; + goto out; + } + if ((all_avail & BTRFS_BLOCK_GROUP_RAID10) && num_devices <= 4) { printk(KERN_ERR "btrfs: unable to go below four devices " "on raid10\n"); @@ -1403,6 +1414,21 @@ int btrfs_rm_device(struct btrfs_root *root, char *device_path) goto out; } + if ((all_avail & BTRFS_BLOCK_GROUP_RAID5) && + root->fs_info->fs_devices->rw_devices <= 2) { + printk(KERN_ERR "btrfs: unable to go below two " + "devices on raid5\n"); + ret = -EINVAL; + goto out; + } + if ((all_avail & BTRFS_BLOCK_GROUP_RAID6) && + root->fs_info->fs_devices->rw_devices <= 3) { + printk(KERN_ERR "btrfs: unable to go below three " + "devices on raid6\n"); + ret = -EINVAL; + goto out; + } + if (strcmp(device_path, "missing") == 0) { struct list_head *devices; struct btrfs_device *tmp; @@ -2657,11 +2683,15 @@ static int chunk_drange_filter(struct extent_buffer *leaf, return 0; if (btrfs_chunk_type(leaf, chunk) & (BTRFS_BLOCK_GROUP_DUP | - BTRFS_BLOCK_GROUP_RAID1 | BTRFS_BLOCK_GROUP_RAID10)) - factor = 2; - else - factor = 1; - factor = num_stripes / factor; + BTRFS_BLOCK_GROUP_RAID1 | BTRFS_BLOCK_GROUP_RAID10)) { + factor = num_stripes / 2; + } else if (btrfs_chunk_type(leaf, chunk) & BTRFS_BLOCK_GROUP_RAID5) { + factor = num_stripes - 1; + } else if (btrfs_chunk_type(leaf, chunk) & BTRFS_BLOCK_GROUP_RAID6) { + factor = num_stripes - 2; + } else { + factor = num_stripes; + } for (i = 0; i < num_stripes; i++) { stripe = btrfs_stripe_nr(chunk, i); @@ -2976,6 +3006,7 @@ int btrfs_balance(struct btrfs_balance_control *bctl, int mixed = 0; int ret; u64 num_devices; + int cancel = 0; if (btrfs_fs_closing(fs_info) || atomic_read(&fs_info->balance_pause_req) || @@ -3018,7 +3049,9 @@ int btrfs_balance(struct btrfs_balance_control *bctl, allowed |= (BTRFS_BLOCK_GROUP_RAID0 | BTRFS_BLOCK_GROUP_RAID1); else allowed |= (BTRFS_BLOCK_GROUP_RAID0 | BTRFS_BLOCK_GROUP_RAID1 | - BTRFS_BLOCK_GROUP_RAID10); + BTRFS_BLOCK_GROUP_RAID10 | + BTRFS_BLOCK_GROUP_RAID5 | + BTRFS_BLOCK_GROUP_RAID6); if ((bctl->data.flags & BTRFS_BALANCE_ARGS_CONVERT) && (!alloc_profile_is_valid(bctl->data.target, 1) || @@ -3058,7 +3091,10 @@ int btrfs_balance(struct btrfs_balance_control *bctl, /* allow to reduce meta or sys integrity only if force set */ allowed = BTRFS_BLOCK_GROUP_DUP | BTRFS_BLOCK_GROUP_RAID1 | - BTRFS_BLOCK_GROUP_RAID10; + BTRFS_BLOCK_GROUP_RAID10 | + BTRFS_BLOCK_GROUP_RAID5 | + BTRFS_BLOCK_GROUP_RAID6; + if (((bctl->sys.flags & BTRFS_BALANCE_ARGS_CONVERT) && (fs_info->avail_system_alloc_bits & allowed) && !(bctl->sys.target & allowed)) || @@ -3124,15 +3160,17 @@ int btrfs_balance(struct btrfs_balance_control *bctl, } if ((ret && ret != -ECANCELED && ret != -ENOSPC) || - balance_need_close(fs_info)) { - __cancel_balance(fs_info); - } + balance_need_close(fs_info)) + cancel = 1; if (bctl->sys.flags & BTRFS_BALANCE_ARGS_CONVERT) { fs_info->num_tolerated_disk_barrier_failures = btrfs_calc_num_tolerated_disk_barrier_failures(fs_info); } + if (cancel) + __cancel_balance(fs_info); + wake_up(&fs_info->balance_wait_q); return ret; @@ -3493,13 +3531,45 @@ static int btrfs_cmp_device_info(const void *a, const void *b) } struct btrfs_raid_attr btrfs_raid_array[BTRFS_NR_RAID_TYPES] = { + /* + * sub_stripes info for map, + * dev_stripes -- stripes per dev, 2 for DUP, 1 other wise + * devs_max -- max devices per stripe, 0 for unlimited + * devs_min -- min devices per stripe + * devs_increment -- ndevs must be a multiple of this + * ncopies -- how many copies of the data we have + */ { 2, 1, 0, 4, 2, 2 /* raid10 */ }, { 1, 1, 2, 2, 2, 2 /* raid1 */ }, { 1, 2, 1, 1, 1, 2 /* dup */ }, { 1, 1, 0, 2, 1, 1 /* raid0 */ }, { 1, 1, 0, 1, 1, 1 /* single */ }, + { 1, 1, 0, 2, 1, 2 /* raid5 */ }, + { 1, 1, 0, 3, 1, 3 /* raid6 */ }, }; +static u32 find_raid56_stripe_len(u32 data_devices, u32 dev_stripe_target) +{ + /* TODO allow them to set a preferred stripe size */ + return 64 * 1024; +} + +static void check_raid56_incompat_flag(struct btrfs_fs_info *info, u64 type) +{ + u64 features; + + if (!(type & (BTRFS_BLOCK_GROUP_RAID5 | BTRFS_BLOCK_GROUP_RAID6))) + return; + + features = btrfs_super_incompat_flags(info->super_copy); + if (features & BTRFS_FEATURE_INCOMPAT_RAID56) + return; + + features |= BTRFS_FEATURE_INCOMPAT_RAID56; + btrfs_set_super_incompat_flags(info->super_copy, features); + printk(KERN_INFO "btrfs: setting RAID5/6 feature flag\n"); +} + static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans, struct btrfs_root *extent_root, struct map_lookup **map_ret, @@ -3515,6 +3585,8 @@ static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans, struct btrfs_device_info *devices_info = NULL; u64 total_avail; int num_stripes; /* total number of stripes to allocate */ + int data_stripes; /* number of stripes that count for + block group size */ int sub_stripes; /* sub_stripes info for map */ int dev_stripes; /* stripes per dev */ int devs_max; /* max devs to use */ @@ -3526,6 +3598,7 @@ static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans, u64 max_chunk_size; u64 stripe_size; u64 num_bytes; + u64 raid_stripe_len = BTRFS_STRIPE_LEN; int ndevs; int i; int j; @@ -3651,16 +3724,31 @@ static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans, stripe_size = devices_info[ndevs-1].max_avail; num_stripes = ndevs * dev_stripes; + /* + * this will have to be fixed for RAID1 and RAID10 over + * more drives + */ + data_stripes = num_stripes / ncopies; + if (stripe_size * ndevs > max_chunk_size * ncopies) { stripe_size = max_chunk_size * ncopies; do_div(stripe_size, ndevs); } - + if (type & BTRFS_BLOCK_GROUP_RAID5) { + raid_stripe_len = find_raid56_stripe_len(ndevs - 1, + btrfs_super_stripesize(info->super_copy)); + data_stripes = num_stripes - 1; + } + if (type & BTRFS_BLOCK_GROUP_RAID6) { + raid_stripe_len = find_raid56_stripe_len(ndevs - 2, + btrfs_super_stripesize(info->super_copy)); + data_stripes = num_stripes - 2; + } do_div(stripe_size, dev_stripes); /* align to BTRFS_STRIPE_LEN */ - do_div(stripe_size, BTRFS_STRIPE_LEN); - stripe_size *= BTRFS_STRIPE_LEN; + do_div(stripe_size, raid_stripe_len); + stripe_size *= raid_stripe_len; map = kmalloc(map_lookup_size(num_stripes), GFP_NOFS); if (!map) { @@ -3678,14 +3766,14 @@ static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans, } } map->sector_size = extent_root->sectorsize; - map->stripe_len = BTRFS_STRIPE_LEN; - map->io_align = BTRFS_STRIPE_LEN; - map->io_width = BTRFS_STRIPE_LEN; + map->stripe_len = raid_stripe_len; + map->io_align = raid_stripe_len; + map->io_width = raid_stripe_len; map->type = type; map->sub_stripes = sub_stripes; *map_ret = map; - num_bytes = stripe_size * (num_stripes / ncopies); + num_bytes = stripe_size * data_stripes; *stripe_size_out = stripe_size; *num_bytes_out = num_bytes; @@ -3734,6 +3822,8 @@ static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans, } } + check_raid56_incompat_flag(extent_root->fs_info, type); + kfree(devices_info); return 0; @@ -4003,6 +4093,10 @@ int btrfs_num_copies(struct btrfs_fs_info *fs_info, u64 logical, u64 len) ret = map->num_stripes; else if (map->type & BTRFS_BLOCK_GROUP_RAID10) ret = map->sub_stripes; + else if (map->type & BTRFS_BLOCK_GROUP_RAID5) + ret = 2; + else if (map->type & BTRFS_BLOCK_GROUP_RAID6) + ret = 3; else ret = 1; free_extent_map(em); @@ -4015,6 +4109,52 @@ int btrfs_num_copies(struct btrfs_fs_info *fs_info, u64 logical, u64 len) return ret; } +unsigned long btrfs_full_stripe_len(struct btrfs_root *root, + struct btrfs_mapping_tree *map_tree, + u64 logical) +{ + struct extent_map *em; + struct map_lookup *map; + struct extent_map_tree *em_tree = &map_tree->map_tree; + unsigned long len = root->sectorsize; + + read_lock(&em_tree->lock); + em = lookup_extent_mapping(em_tree, logical, len); + read_unlock(&em_tree->lock); + BUG_ON(!em); + + BUG_ON(em->start > logical || em->start + em->len < logical); + map = (struct map_lookup *)em->bdev; + if (map->type & (BTRFS_BLOCK_GROUP_RAID5 | + BTRFS_BLOCK_GROUP_RAID6)) { + len = map->stripe_len * nr_data_stripes(map); + } + free_extent_map(em); + return len; +} + +int btrfs_is_parity_mirror(struct btrfs_mapping_tree *map_tree, + u64 logical, u64 len, int mirror_num) +{ + struct extent_map *em; + struct map_lookup *map; + struct extent_map_tree *em_tree = &map_tree->map_tree; + int ret = 0; + + read_lock(&em_tree->lock); + em = lookup_extent_mapping(em_tree, logical, len); + read_unlock(&em_tree->lock); + BUG_ON(!em); + + BUG_ON(em->start > logical || em->start + em->len < logical); + map = (struct map_lookup *)em->bdev; + if (map->type & (BTRFS_BLOCK_GROUP_RAID5 | + BTRFS_BLOCK_GROUP_RAID6)) + ret = 1; + free_extent_map(em); + return ret; +} + static int find_live_mirror(struct btrfs_fs_info *fs_info, struct map_lookup *map, int first, int num, int optimal, int dev_replace_is_ongoing) @@ -4052,10 +4192,39 @@ static int find_live_mirror(struct btrfs_fs_info *fs_info, return optimal; } +static inline int parity_smaller(u64 a, u64 b) +{ + return a > b; +} + +/* Bubble-sort the stripe set to put the parity/syndrome stripes last */ +static void sort_parity_stripes(struct btrfs_bio *bbio, u64 *raid_map) +{ + struct btrfs_bio_stripe s; + int i; + u64 l; + int again = 1; + + while (again) { + again = 0; + for (i = 0; i < bbio->num_stripes - 1; i++) { + if (parity_smaller(raid_map[i], raid_map[i+1])) { + s = bbio->stripes[i]; + l = raid_map[i]; + bbio->stripes[i] = bbio->stripes[i+1]; + raid_map[i] = raid_map[i+1]; + bbio->stripes[i+1] = s; + raid_map[i+1] = l; + again = 1; + } + } + } +} + static int __btrfs_map_block(struct btrfs_fs_info *fs_info, int rw, u64 logical, u64 *length, struct btrfs_bio **bbio_ret, - int mirror_num) + int mirror_num, u64 **raid_map_ret) { struct extent_map *em; struct map_lookup *map; @@ -4067,6 +4236,8 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info, int rw, u64 stripe_nr; u64 stripe_nr_orig; u64 stripe_nr_end; + u64 stripe_len; + u64 *raid_map = NULL; int stripe_index; int i; int ret = 0; @@ -4078,6 +4249,7 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info, int rw, int num_alloc_stripes; int patch_the_first_stripe_for_dev_replace = 0; u64 physical_to_patch_in_first_stripe = 0; + u64 raid56_full_stripe_start = (u64)-1; read_lock(&em_tree->lock); em = lookup_extent_mapping(em_tree, logical, *length); @@ -4094,29 +4266,63 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info, int rw, map = (struct map_lookup *)em->bdev; offset = logical - em->start; + if (mirror_num > map->num_stripes) + mirror_num = 0; + + stripe_len = map->stripe_len; stripe_nr = offset; /* * stripe_nr counts the total number of stripes we have to stride * to get to this block */ - do_div(stripe_nr, map->stripe_len); + do_div(stripe_nr, stripe_len); - stripe_offset = stripe_nr * map->stripe_len; + stripe_offset = stripe_nr * stripe_len; BUG_ON(offset < stripe_offset); /* stripe_offset is the offset of this block in its stripe*/ stripe_offset = offset - stripe_offset; - if (rw & REQ_DISCARD) + /* if we're here for raid56, we need to know the stripe aligned start */ + if (map->type & (BTRFS_BLOCK_GROUP_RAID5 | BTRFS_BLOCK_GROUP_RAID6)) { + unsigned long full_stripe_len = stripe_len * nr_data_stripes(map); + raid56_full_stripe_start = offset; + + /* allow a write of a full stripe, but make sure we don't + * allow straddling of stripes + */ + do_div(raid56_full_stripe_start, full_stripe_len); + raid56_full_stripe_start *= full_stripe_len; + } + + if (rw & REQ_DISCARD) { + /* we don't discard raid56 yet */ + if (map->type & + (BTRFS_BLOCK_GROUP_RAID5 | BTRFS_BLOCK_GROUP_RAID6)) { + ret = -EOPNOTSUPP; + goto out; + } *length = min_t(u64, em->len - offset, *length); - else if (map->type & BTRFS_BLOCK_GROUP_PROFILE_MASK) { - /* we limit the length of each bio to what fits in a stripe */ - *length = min_t(u64, em->len - offset, - map->stripe_len - stripe_offset); + } else if (map->type & BTRFS_BLOCK_GROUP_PROFILE_MASK) { + u64 max_len; + /* For writes to RAID[56], allow a full stripeset across all disks. + For other RAID types and for RAID[56] reads, just allow a single + stripe (on a single disk). */ + if (map->type & (BTRFS_BLOCK_GROUP_RAID5 | BTRFS_BLOCK_GROUP_RAID6) && + (rw & REQ_WRITE)) { + max_len = stripe_len * nr_data_stripes(map) - + (offset - raid56_full_stripe_start); + } else { + /* we limit the length of each bio to what fits in a stripe */ + max_len = stripe_len - stripe_offset; + } + *length = min_t(u64, em->len - offset, max_len); } else { *length = em->len - offset; } + /* This is for when we're called from btrfs_merge_bio_hook() and all + it cares about is the length */ if (!bbio_ret) goto out; @@ -4149,7 +4355,7 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info, int rw, u64 physical_of_found = 0; ret = __btrfs_map_block(fs_info, REQ_GET_READ_MIRRORS, - logical, &tmp_length, &tmp_bbio, 0); + logical, &tmp_length, &tmp_bbio, 0, NULL); if (ret) { WARN_ON(tmp_bbio != NULL); goto out; @@ -4215,6 +4421,7 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info, int rw, do_div(stripe_nr_end, map->stripe_len); stripe_end_offset = stripe_nr_end * map->stripe_len - (offset + *length); + if (map->type & BTRFS_BLOCK_GROUP_RAID0) { if (rw & REQ_DISCARD) num_stripes = min_t(u64, map->num_stripes, @@ -4265,6 +4472,65 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info, int rw, dev_replace_is_ongoing); mirror_num = stripe_index - old_stripe_index + 1; } + + } else if (map->type & (BTRFS_BLOCK_GROUP_RAID5 | + BTRFS_BLOCK_GROUP_RAID6)) { + u64 tmp; + + if (bbio_ret && ((rw & REQ_WRITE) || mirror_num > 1) + && raid_map_ret) { + int i, rot; + + /* push stripe_nr back to the start of the full stripe */ + stripe_nr = raid56_full_stripe_start; + do_div(stripe_nr, stripe_len); + + stripe_index = do_div(stripe_nr, nr_data_stripes(map)); + + /* RAID[56] write or recovery. Return all stripes */ + num_stripes = map->num_stripes; + max_errors = nr_parity_stripes(map); + + raid_map = kmalloc(sizeof(u64) * num_stripes, + GFP_NOFS); + if (!raid_map) { + ret = -ENOMEM; + goto out; + } + + /* Work out the disk rotation on this stripe-set */ + tmp = stripe_nr; + rot = do_div(tmp, num_stripes); + + /* Fill in the logical address of each stripe */ + tmp = stripe_nr * nr_data_stripes(map); + for (i = 0; i < nr_data_stripes(map); i++) + raid_map[(i+rot) % num_stripes] = + em->start + (tmp + i) * map->stripe_len; + + raid_map[(i+rot) % map->num_stripes] = RAID5_P_STRIPE; + if (map->type & BTRFS_BLOCK_GROUP_RAID6) + raid_map[(i+rot+1) % num_stripes] = + RAID6_Q_STRIPE; + + *length = map->stripe_len; + stripe_index = 0; + stripe_offset = 0; + } else { + /* + * Mirror #0 or #1 means the original data block. + * Mirror #2 is RAID5 parity block. + * Mirror #3 is RAID6 Q block. + */ + stripe_index = do_div(stripe_nr, nr_data_stripes(map)); + if (mirror_num > 1) + stripe_index = nr_data_stripes(map) + + mirror_num - 2; + + /* We distribute the parity blocks across stripes */ + tmp = stripe_nr + stripe_index; + stripe_index = do_div(tmp, map->num_stripes); + } } else { /* * after this do_div call, stripe_nr is the number of stripes @@ -4373,8 +4639,11 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info, int rw, if (rw & (REQ_WRITE | REQ_GET_READ_MIRRORS)) { if (map->type & (BTRFS_BLOCK_GROUP_RAID1 | BTRFS_BLOCK_GROUP_RAID10 | + BTRFS_BLOCK_GROUP_RAID5 | BTRFS_BLOCK_GROUP_DUP)) { max_errors = 1; + } else if (map->type & BTRFS_BLOCK_GROUP_RAID6) { + max_errors = 2; } } @@ -4475,6 +4744,10 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info, int rw, bbio->stripes[0].physical = physical_to_patch_in_first_stripe; bbio->mirror_num = map->num_stripes + 1; } + if (raid_map) { + sort_parity_stripes(bbio, raid_map); + *raid_map_ret = raid_map; + } out: if (dev_replace_is_ongoing) btrfs_dev_replace_unlock(dev_replace); @@ -4487,7 +4760,7 @@ int btrfs_map_block(struct btrfs_fs_info *fs_info, int rw, struct btrfs_bio **bbio_ret, int mirror_num) { return __btrfs_map_block(fs_info, rw, logical, length, bbio_ret, - mirror_num); + mirror_num, NULL); } int btrfs_rmap_block(struct btrfs_mapping_tree *map_tree, @@ -4501,6 +4774,7 @@ int btrfs_rmap_block(struct btrfs_mapping_tree *map_tree, u64 bytenr; u64 length; u64 stripe_nr; + u64 rmap_len; int i, j, nr = 0; read_lock(&em_tree->lock); @@ -4511,10 +4785,17 @@ int btrfs_rmap_block(struct btrfs_mapping_tree *map_tree, map = (struct map_lookup *)em->bdev; length = em->len; + rmap_len = map->stripe_len; + if (map->type & BTRFS_BLOCK_GROUP_RAID10) do_div(length, map->num_stripes / map->sub_stripes); else if (map->type & BTRFS_BLOCK_GROUP_RAID0) do_div(length, map->num_stripes); + else if (map->type & (BTRFS_BLOCK_GROUP_RAID5 | + BTRFS_BLOCK_GROUP_RAID6)) { + do_div(length, nr_data_stripes(map)); + rmap_len = map->stripe_len * nr_data_stripes(map); + } buf = kzalloc(sizeof(u64) * map->num_stripes, GFP_NOFS); BUG_ON(!buf); /* -ENOMEM */ @@ -4534,8 +4815,11 @@ int btrfs_rmap_block(struct btrfs_mapping_tree *map_tree, do_div(stripe_nr, map->sub_stripes); } else if (map->type & BTRFS_BLOCK_GROUP_RAID0) { stripe_nr = stripe_nr * map->num_stripes + i; - } - bytenr = chunk_start + stripe_nr * map->stripe_len; + } /* else if RAID[56], multiply by nr_data_stripes(). + * Alternatively, just use rmap_len below instead of + * map->stripe_len */ + + bytenr = chunk_start + stripe_nr * rmap_len; WARN_ON(nr >= map->num_stripes); for (j = 0; j < nr; j++) { if (buf[j] == bytenr) @@ -4549,7 +4833,7 @@ int btrfs_rmap_block(struct btrfs_mapping_tree *map_tree, *logical = buf; *naddrs = nr; - *stripe_len = map->stripe_len; + *stripe_len = rmap_len; free_extent_map(em); return 0; @@ -4623,7 +4907,7 @@ static void btrfs_end_bio(struct bio *bio, int err) bio->bi_bdev = (struct block_device *) (unsigned long)bbio->mirror_num; /* only send an error to the higher layers if it is - * beyond the tolerance of the multi-bio + * beyond the tolerance of the btrfs bio */ if (atomic_read(&bbio->error) > bbio->max_errors) { err = -EIO; @@ -4657,13 +4941,18 @@ struct async_sched { * This will add one bio to the pending list for a device and make sure * the work struct is scheduled. */ -static noinline void schedule_bio(struct btrfs_root *root, +noinline void btrfs_schedule_bio(struct btrfs_root *root, struct btrfs_device *device, int rw, struct bio *bio) { int should_queue = 1; struct btrfs_pending_bios *pending_bios; + if (device->missing || !device->bdev) { + bio_endio(bio, -EIO); + return; + } + /* don't bother with additional async steps for reads, right now */ if (!(rw & REQ_WRITE)) { bio_get(bio); @@ -4761,7 +5050,7 @@ static void submit_stripe_bio(struct btrfs_root *root, struct btrfs_bio *bbio, #endif bio->bi_bdev = dev->bdev; if (async) - schedule_bio(root, dev, rw, bio); + btrfs_schedule_bio(root, dev, rw, bio); else btrfsic_submit_bio(rw, bio); } @@ -4820,6 +5109,7 @@ int btrfs_map_bio(struct btrfs_root *root, int rw, struct bio *bio, u64 logical = (u64)bio->bi_sector << 9; u64 length = 0; u64 map_length; + u64 *raid_map = NULL; int ret; int dev_nr = 0; int total_devs = 1; @@ -4828,12 +5118,30 @@ int btrfs_map_bio(struct btrfs_root *root, int rw, struct bio *bio, length = bio->bi_size; map_length = length; - ret = btrfs_map_block(root->fs_info, rw, logical, &map_length, &bbio, - mirror_num); - if (ret) + ret = __btrfs_map_block(root->fs_info, rw, logical, &map_length, &bbio, + mirror_num, &raid_map); + if (ret) /* -ENOMEM */ return ret; total_devs = bbio->num_stripes; + bbio->orig_bio = first_bio; + bbio->private = first_bio->bi_private; + bbio->end_io = first_bio->bi_end_io; + atomic_set(&bbio->stripes_pending, bbio->num_stripes); + + if (raid_map) { + /* In this case, map_length has been set to the length of + a single stripe; not the whole write */ + if (rw & WRITE) { + return raid56_parity_write(root, bio, bbio, + raid_map, map_length); + } else { + return raid56_parity_recover(root, bio, bbio, + raid_map, map_length, + mirror_num); + } + } + if (map_length < length) { printk(KERN_CRIT "btrfs: mapping failed logical %llu bio len %llu " "len %llu\n", (unsigned long long)logical, @@ -4842,11 +5150,6 @@ int btrfs_map_bio(struct btrfs_root *root, int rw, struct bio *bio, BUG(); } - bbio->orig_bio = first_bio; - bbio->private = first_bio->bi_private; - bbio->end_io = first_bio->bi_end_io; - atomic_set(&bbio->stripes_pending, bbio->num_stripes); - while (dev_nr < total_devs) { dev = bbio->stripes[dev_nr].dev; if (!dev || !dev->bdev || (rw & WRITE && !dev->writeable)) { diff --git a/fs/btrfs/volumes.h b/fs/btrfs/volumes.h index d3c3939ac751..0c2b856ecd98 100644 --- a/fs/btrfs/volumes.h +++ b/fs/btrfs/volumes.h @@ -321,7 +321,14 @@ void btrfs_destroy_dev_replace_tgtdev(struct btrfs_fs_info *fs_info, void btrfs_init_dev_replace_tgtdev_for_resume(struct btrfs_fs_info *fs_info, struct btrfs_device *tgtdev); int btrfs_scratch_superblock(struct btrfs_device *device); - +void btrfs_schedule_bio(struct btrfs_root *root, + struct btrfs_device *device, + int rw, struct bio *bio); +int btrfs_is_parity_mirror(struct btrfs_mapping_tree *map_tree, + u64 logical, u64 len, int mirror_num); +unsigned long btrfs_full_stripe_len(struct btrfs_root *root, + struct btrfs_mapping_tree *map_tree, + u64 logical); static inline void btrfs_dev_stat_inc(struct btrfs_device *dev, int index) { -- cgit v1.2.3 From 4ae10b3a133e1147f3c818fe2ebaf005b217b7bf Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Thu, 31 Jan 2013 14:42:09 -0500 Subject: Btrfs: Add a stripe cache to raid56 The stripe cache allows us to avoid extra read/modify/write cycles by caching the pages we read off the disk. Pages are cached when: * They are read in during a read/modify/write cycle * They are written during a read/modify/write cycle * They are involved in a parity rebuild Pages are not cached if we're doing a full stripe write. We're assuming that a full stripe write won't be followed by another partial stripe write any time soon. This provides a substantial boost in performance for workloads that synchronously modify adjacent offsets in the file, and for the parity rebuild use case in general. The size of the stripe cache isn't tunable (yet) and is set at 1024 entries. Example on flash: dd if=/dev/zero of=/mnt/xxx bs=4K oflag=direct Without the stripe cache -- 2.1MB/s With the stripe cache 21MB/s Signed-off-by: Chris Mason --- fs/btrfs/ctree.h | 5 +- fs/btrfs/raid56.c | 327 ++++++++++++++++++++++++++++++++++++++++++++++++++++-- 2 files changed, 324 insertions(+), 8 deletions(-) (limited to 'fs/btrfs/ctree.h') diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 0cce3aafbd62..e3a4fd70f55a 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -1244,7 +1244,10 @@ struct btrfs_stripe_hash { /* used by the raid56 code to lock stripes for read/modify/write */ struct btrfs_stripe_hash_table { - struct btrfs_stripe_hash *table; + struct list_head stripe_cache; + spinlock_t cache_lock; + int cache_size; + struct btrfs_stripe_hash table[]; }; #define BTRFS_STRIPE_HASH_TABLE_BITS 11 diff --git a/fs/btrfs/raid56.c b/fs/btrfs/raid56.c index d02510f34936..7ccddca9ee71 100644 --- a/fs/btrfs/raid56.c +++ b/fs/btrfs/raid56.c @@ -47,6 +47,20 @@ /* set when additional merges to this rbio are not allowed */ #define RBIO_RMW_LOCKED_BIT 1 +/* + * set when this rbio is sitting in the hash, but it is just a cache + * of past RMW + */ +#define RBIO_CACHE_BIT 2 + +/* + * set when it is safe to trust the stripe_pages for caching + */ +#define RBIO_CACHE_READY_BIT 3 + + +#define RBIO_CACHE_SIZE 1024 + struct btrfs_raid_bio { struct btrfs_fs_info *fs_info; struct btrfs_bio *bbio; @@ -65,6 +79,11 @@ struct btrfs_raid_bio { */ struct list_head hash_list; + /* + * LRU list for the stripe cache + */ + struct list_head stripe_cache; + /* * for scheduling work in the helper threads */ @@ -176,7 +195,9 @@ int btrfs_alloc_stripe_hash_table(struct btrfs_fs_info *info) if (!table) return -ENOMEM; - table->table = (void *)(table + 1); + spin_lock_init(&table->cache_lock); + INIT_LIST_HEAD(&table->stripe_cache); + h = table->table; for (i = 0; i < num_entries; i++) { @@ -192,6 +213,42 @@ int btrfs_alloc_stripe_hash_table(struct btrfs_fs_info *info) return 0; } +/* + * caching an rbio means to copy anything from the + * bio_pages array into the stripe_pages array. We + * use the page uptodate bit in the stripe cache array + * to indicate if it has valid data + * + * once the caching is done, we set the cache ready + * bit. + */ +static void cache_rbio_pages(struct btrfs_raid_bio *rbio) +{ + int i; + char *s; + char *d; + int ret; + + ret = alloc_rbio_pages(rbio); + if (ret) + return; + + for (i = 0; i < rbio->nr_pages; i++) { + if (!rbio->bio_pages[i]) + continue; + + s = kmap(rbio->bio_pages[i]); + d = kmap(rbio->stripe_pages[i]); + + memcpy(d, s, PAGE_CACHE_SIZE); + + kunmap(rbio->bio_pages[i]); + kunmap(rbio->stripe_pages[i]); + SetPageUptodate(rbio->stripe_pages[i]); + } + set_bit(RBIO_CACHE_READY_BIT, &rbio->flags); +} + /* * we hash on the first logical address of the stripe */ @@ -210,6 +267,34 @@ static int rbio_bucket(struct btrfs_raid_bio *rbio) return hash_64(num >> 16, BTRFS_STRIPE_HASH_TABLE_BITS); } +/* + * stealing an rbio means taking all the uptodate pages from the stripe + * array in the source rbio and putting them into the destination rbio + */ +static void steal_rbio(struct btrfs_raid_bio *src, struct btrfs_raid_bio *dest) +{ + int i; + struct page *s; + struct page *d; + + if (!test_bit(RBIO_CACHE_READY_BIT, &src->flags)) + return; + + for (i = 0; i < dest->nr_pages; i++) { + s = src->stripe_pages[i]; + if (!s || !PageUptodate(s)) { + continue; + } + + d = dest->stripe_pages[i]; + if (d) + __free_page(d); + + dest->stripe_pages[i] = s; + src->stripe_pages[i] = NULL; + } +} + /* * merging means we take the bio_list from the victim and * splice it into the destination. The victim should @@ -226,16 +311,170 @@ static void merge_rbio(struct btrfs_raid_bio *dest, } /* - * free the hash table used by unmount + * used to prune items that are in the cache. The caller + * must hold the hash table lock. + */ +static void __remove_rbio_from_cache(struct btrfs_raid_bio *rbio) +{ + int bucket = rbio_bucket(rbio); + struct btrfs_stripe_hash_table *table; + struct btrfs_stripe_hash *h; + int freeit = 0; + + /* + * check the bit again under the hash table lock. + */ + if (!test_bit(RBIO_CACHE_BIT, &rbio->flags)) + return; + + table = rbio->fs_info->stripe_hash_table; + h = table->table + bucket; + + /* hold the lock for the bucket because we may be + * removing it from the hash table + */ + spin_lock(&h->lock); + + /* + * hold the lock for the bio list because we need + * to make sure the bio list is empty + */ + spin_lock(&rbio->bio_list_lock); + + if (test_and_clear_bit(RBIO_CACHE_BIT, &rbio->flags)) { + list_del_init(&rbio->stripe_cache); + table->cache_size -= 1; + freeit = 1; + + /* if the bio list isn't empty, this rbio is + * still involved in an IO. We take it out + * of the cache list, and drop the ref that + * was held for the list. + * + * If the bio_list was empty, we also remove + * the rbio from the hash_table, and drop + * the corresponding ref + */ + if (bio_list_empty(&rbio->bio_list)) { + if (!list_empty(&rbio->hash_list)) { + list_del_init(&rbio->hash_list); + atomic_dec(&rbio->refs); + BUG_ON(!list_empty(&rbio->plug_list)); + } + } + } + + spin_unlock(&rbio->bio_list_lock); + spin_unlock(&h->lock); + + if (freeit) + __free_raid_bio(rbio); +} + +/* + * prune a given rbio from the cache + */ +static void remove_rbio_from_cache(struct btrfs_raid_bio *rbio) +{ + struct btrfs_stripe_hash_table *table; + unsigned long flags; + + if (!test_bit(RBIO_CACHE_BIT, &rbio->flags)) + return; + + table = rbio->fs_info->stripe_hash_table; + + spin_lock_irqsave(&table->cache_lock, flags); + __remove_rbio_from_cache(rbio); + spin_unlock_irqrestore(&table->cache_lock, flags); +} + +/* + * remove everything in the cache + */ +void btrfs_clear_rbio_cache(struct btrfs_fs_info *info) +{ + struct btrfs_stripe_hash_table *table; + unsigned long flags; + struct btrfs_raid_bio *rbio; + + table = info->stripe_hash_table; + + spin_lock_irqsave(&table->cache_lock, flags); + while (!list_empty(&table->stripe_cache)) { + rbio = list_entry(table->stripe_cache.next, + struct btrfs_raid_bio, + stripe_cache); + __remove_rbio_from_cache(rbio); + } + spin_unlock_irqrestore(&table->cache_lock, flags); +} + +/* + * remove all cached entries and free the hash table + * used by unmount */ void btrfs_free_stripe_hash_table(struct btrfs_fs_info *info) { if (!info->stripe_hash_table) return; + btrfs_clear_rbio_cache(info); kfree(info->stripe_hash_table); info->stripe_hash_table = NULL; } +/* + * insert an rbio into the stripe cache. It + * must have already been prepared by calling + * cache_rbio_pages + * + * If this rbio was already cached, it gets + * moved to the front of the lru. + * + * If the size of the rbio cache is too big, we + * prune an item. + */ +static void cache_rbio(struct btrfs_raid_bio *rbio) +{ + struct btrfs_stripe_hash_table *table; + unsigned long flags; + + if (!test_bit(RBIO_CACHE_READY_BIT, &rbio->flags)) + return; + + table = rbio->fs_info->stripe_hash_table; + + spin_lock_irqsave(&table->cache_lock, flags); + spin_lock(&rbio->bio_list_lock); + + /* bump our ref if we were not in the list before */ + if (!test_and_set_bit(RBIO_CACHE_BIT, &rbio->flags)) + atomic_inc(&rbio->refs); + + if (!list_empty(&rbio->stripe_cache)){ + list_move(&rbio->stripe_cache, &table->stripe_cache); + } else { + list_add(&rbio->stripe_cache, &table->stripe_cache); + table->cache_size += 1; + } + + spin_unlock(&rbio->bio_list_lock); + + if (table->cache_size > RBIO_CACHE_SIZE) { + struct btrfs_raid_bio *found; + + found = list_entry(table->stripe_cache.prev, + struct btrfs_raid_bio, + stripe_cache); + + if (found != rbio) + __remove_rbio_from_cache(found); + } + + spin_unlock_irqrestore(&table->cache_lock, flags); + return; +} + /* * helper function to run the xor_blocks api. It is only * able to do MAX_XOR_BLOCKS at a time, so we need to @@ -303,6 +542,17 @@ static int rbio_can_merge(struct btrfs_raid_bio *last, test_bit(RBIO_RMW_LOCKED_BIT, &cur->flags)) return 0; + /* + * we can't merge with cached rbios, since the + * idea is that when we merge the destination + * rbio is going to run our IO for us. We can + * steal from cached rbio's though, other functions + * handle that. + */ + if (test_bit(RBIO_CACHE_BIT, &last->flags) || + test_bit(RBIO_CACHE_BIT, &cur->flags)) + return 0; + if (last->raid_map[0] != cur->raid_map[0]) return 0; @@ -370,6 +620,7 @@ static noinline int lock_stripe_add(struct btrfs_raid_bio *rbio) unsigned long flags; DEFINE_WAIT(wait); struct btrfs_raid_bio *freeit = NULL; + struct btrfs_raid_bio *cache_drop = NULL; int ret = 0; int walk = 0; @@ -379,6 +630,21 @@ static noinline int lock_stripe_add(struct btrfs_raid_bio *rbio) if (cur->raid_map[0] == rbio->raid_map[0]) { spin_lock(&cur->bio_list_lock); + /* can we steal this cached rbio's pages? */ + if (bio_list_empty(&cur->bio_list) && + list_empty(&cur->plug_list) && + test_bit(RBIO_CACHE_BIT, &cur->flags) && + !test_bit(RBIO_RMW_LOCKED_BIT, &cur->flags)) { + list_del_init(&cur->hash_list); + atomic_dec(&cur->refs); + + steal_rbio(cur, rbio); + cache_drop = cur; + spin_unlock(&cur->bio_list_lock); + + goto lockit; + } + /* can we merge into the lock owner? */ if (rbio_can_merge(cur, rbio)) { merge_rbio(cur, rbio); @@ -388,6 +654,7 @@ static noinline int lock_stripe_add(struct btrfs_raid_bio *rbio) goto out; } + /* * we couldn't merge with the running * rbio, see if we can merge with the @@ -417,11 +684,13 @@ static noinline int lock_stripe_add(struct btrfs_raid_bio *rbio) goto out; } } - +lockit: atomic_inc(&rbio->refs); list_add(&rbio->hash_list, &h->hash_list); out: spin_unlock_irqrestore(&h->lock, flags); + if (cache_drop) + remove_rbio_from_cache(cache_drop); if (freeit) __free_raid_bio(freeit); return ret; @@ -436,14 +705,30 @@ static noinline void unlock_stripe(struct btrfs_raid_bio *rbio) int bucket; struct btrfs_stripe_hash *h; unsigned long flags; + int keep_cache = 0; bucket = rbio_bucket(rbio); h = rbio->fs_info->stripe_hash_table->table + bucket; + if (list_empty(&rbio->plug_list)) + cache_rbio(rbio); + spin_lock_irqsave(&h->lock, flags); spin_lock(&rbio->bio_list_lock); if (!list_empty(&rbio->hash_list)) { + /* + * if we're still cached and there is no other IO + * to perform, just leave this rbio here for others + * to steal from later + */ + if (list_empty(&rbio->plug_list) && + test_bit(RBIO_CACHE_BIT, &rbio->flags)) { + keep_cache = 1; + clear_bit(RBIO_RMW_LOCKED_BIT, &rbio->flags); + BUG_ON(!bio_list_empty(&rbio->bio_list)); + goto done; + } list_del_init(&rbio->hash_list); atomic_dec(&rbio->refs); @@ -469,11 +754,12 @@ static noinline void unlock_stripe(struct btrfs_raid_bio *rbio) if (next->read_rebuild) async_read_rebuild(next); - else + else { + steal_rbio(rbio, next); async_rmw_stripe(next); + } goto done_nolock; - } else if (waitqueue_active(&h->wait)) { spin_unlock(&rbio->bio_list_lock); spin_unlock_irqrestore(&h->lock, flags); @@ -481,11 +767,13 @@ static noinline void unlock_stripe(struct btrfs_raid_bio *rbio) goto done_nolock; } } +done: spin_unlock(&rbio->bio_list_lock); spin_unlock_irqrestore(&h->lock, flags); done_nolock: - return; + if (!keep_cache) + remove_rbio_from_cache(rbio); } static void __free_raid_bio(struct btrfs_raid_bio *rbio) @@ -496,6 +784,7 @@ static void __free_raid_bio(struct btrfs_raid_bio *rbio) if (!atomic_dec_and_test(&rbio->refs)) return; + WARN_ON(!list_empty(&rbio->stripe_cache)); WARN_ON(!list_empty(&rbio->hash_list)); WARN_ON(!bio_list_empty(&rbio->bio_list)); @@ -630,6 +919,7 @@ static struct btrfs_raid_bio *alloc_rbio(struct btrfs_root *root, bio_list_init(&rbio->bio_list); INIT_LIST_HEAD(&rbio->plug_list); spin_lock_init(&rbio->bio_list_lock); + INIT_LIST_HEAD(&rbio->stripe_cache); INIT_LIST_HEAD(&rbio->hash_list); rbio->bbio = bbio; rbio->raid_map = raid_map; @@ -864,8 +1154,17 @@ static noinline void finish_rmw(struct btrfs_raid_bio *rbio) /* * now that we've set rmw_locked, run through the * bio list one last time and map the page pointers + * + * We don't cache full rbios because we're assuming + * the higher layers are unlikely to use this area of + * the disk again soon. If they do use it again, + * hopefully they will send another full bio. */ index_rbio_pages(rbio); + if (!rbio_is_full(rbio)) + cache_rbio_pages(rbio); + else + clear_bit(RBIO_CACHE_READY_BIT, &rbio->flags); for (pagenr = 0; pagenr < pages_per_stripe; pagenr++) { struct page *p; @@ -1155,6 +1454,13 @@ static int raid56_rmw_stripe(struct btrfs_raid_bio *rbio) continue; page = rbio_stripe_page(rbio, stripe, pagenr); + /* + * the bio cache may have handed us an uptodate + * page. If so, be happy and use it + */ + if (PageUptodate(page)) + continue; + ret = rbio_add_io_page(rbio, &bio_list, page, stripe, pagenr, rbio->stripe_len); if (ret) @@ -1440,6 +1746,11 @@ cleanup: cleanup_io: if (rbio->read_rebuild) { + if (err == 0) + cache_rbio_pages(rbio); + else + clear_bit(RBIO_CACHE_READY_BIT, &rbio->flags); + rbio_orig_end_io(rbio, err, err == 0); } else if (err == 0) { rbio->faila = -1; @@ -1505,7 +1816,9 @@ static int __raid56_parity_recover(struct btrfs_raid_bio *rbio) atomic_set(&rbio->bbio->error, 0); /* - * read everything that hasn't failed. + * read everything that hasn't failed. Thanks to the + * stripe cache, it is possible that some or all of these + * pages are going to be uptodate. */ for (stripe = 0; stripe < bbio->num_stripes; stripe++) { if (rbio->faila == stripe || -- cgit v1.2.3 From 2ab28f322f9896782da904f5942f3873432addc8 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Fri, 12 Oct 2012 15:27:49 -0400 Subject: Btrfs: wait on ordered extents at the last possible moment Since we don't actually copy the extent information from the source tree in the fast case we don't need to wait for ordered io to be completed in order to fsync, we just need to wait for the io to be completed. So when we're logging our file just attach all of the ordered extents to the log, and then when the log syncs just wait for IO_DONE on the ordered extents and then write the super. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/ctree.h | 3 ++ fs/btrfs/disk-io.c | 4 ++ fs/btrfs/file.c | 30 +++++++++-- fs/btrfs/inode.c | 8 +++ fs/btrfs/ordered-data.c | 68 +++++++++++++++++++++++++ fs/btrfs/ordered-data.h | 11 ++++ fs/btrfs/tree-log.c | 132 ++++++++++++++++++++++++++++++++++++++++++++++-- 7 files changed, 247 insertions(+), 9 deletions(-) (limited to 'fs/btrfs/ctree.h') diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 547b7b05727f..411c8d97074e 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -1623,6 +1623,9 @@ struct btrfs_root { struct list_head root_list; + spinlock_t log_extents_lock[2]; + struct list_head logged_list[2]; + spinlock_t orphan_lock; atomic_t orphan_inodes; struct btrfs_block_rsv *orphan_block_rsv; diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index a762f9137610..1db8a9938829 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -1178,9 +1178,13 @@ static void __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize, INIT_LIST_HEAD(&root->dirty_list); INIT_LIST_HEAD(&root->root_list); + INIT_LIST_HEAD(&root->logged_list[0]); + INIT_LIST_HEAD(&root->logged_list[1]); spin_lock_init(&root->orphan_lock); spin_lock_init(&root->inode_lock); spin_lock_init(&root->accounting_lock); + spin_lock_init(&root->log_extents_lock[0]); + spin_lock_init(&root->log_extents_lock[1]); mutex_init(&root->objectid_mutex); mutex_init(&root->log_mutex); init_waitqueue_head(&root->log_writer_wait); diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c index b06d289f998f..083abca56055 100644 --- a/fs/btrfs/file.c +++ b/fs/btrfs/file.c @@ -1655,16 +1655,21 @@ int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync) struct btrfs_root *root = BTRFS_I(inode)->root; int ret = 0; struct btrfs_trans_handle *trans; + bool full_sync = 0; trace_btrfs_sync_file(file, datasync); /* * We write the dirty pages in the range and wait until they complete * out of the ->i_mutex. If so, we can flush the dirty pages by - * multi-task, and make the performance up. + * multi-task, and make the performance up. See + * btrfs_wait_ordered_range for an explanation of the ASYNC check. */ atomic_inc(&BTRFS_I(inode)->sync_writers); - ret = filemap_write_and_wait_range(inode->i_mapping, start, end); + ret = filemap_fdatawrite_range(inode->i_mapping, start, end); + if (!ret && test_bit(BTRFS_INODE_HAS_ASYNC_EXTENT, + &BTRFS_I(inode)->runtime_flags)) + ret = filemap_fdatawrite_range(inode->i_mapping, start, end); atomic_dec(&BTRFS_I(inode)->sync_writers); if (ret) return ret; @@ -1676,7 +1681,10 @@ int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync) * range being left. */ atomic_inc(&root->log_batch); - btrfs_wait_ordered_range(inode, start, end - start + 1); + full_sync = test_bit(BTRFS_INODE_NEEDS_FULL_SYNC, + &BTRFS_I(inode)->runtime_flags); + if (full_sync) + btrfs_wait_ordered_range(inode, start, end - start + 1); atomic_inc(&root->log_batch); /* @@ -1743,13 +1751,25 @@ int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync) if (ret != BTRFS_NO_LOG_SYNC) { if (ret > 0) { + /* + * If we didn't already wait for ordered extents we need + * to do that now. + */ + if (!full_sync) + btrfs_wait_ordered_range(inode, start, + end - start + 1); ret = btrfs_commit_transaction(trans, root); } else { ret = btrfs_sync_log(trans, root); - if (ret == 0) + if (ret == 0) { ret = btrfs_end_transaction(trans, root); - else + } else { + if (!full_sync) + btrfs_wait_ordered_range(inode, start, + end - + start + 1); ret = btrfs_commit_transaction(trans, root); + } } } else { ret = btrfs_end_transaction(trans, root); diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 35d152444932..31a871ec48f2 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -700,6 +700,8 @@ retry: em->start = async_extent->start; em->len = async_extent->ram_size; em->orig_start = em->start; + em->mod_start = em->start; + em->mod_len = em->len; em->block_start = ins.objectid; em->block_len = ins.offset; @@ -892,6 +894,8 @@ static noinline int __cow_file_range(struct btrfs_trans_handle *trans, em->orig_start = em->start; ram_size = ins.offset; em->len = ins.offset; + em->mod_start = em->start; + em->mod_len = em->len; em->block_start = ins.objectid; em->block_len = ins.offset; @@ -1338,6 +1342,8 @@ out_check: em->block_start = disk_bytenr; em->orig_block_len = disk_num_bytes; em->bdev = root->fs_info->fs_devices->latest_bdev; + em->mod_start = em->start; + em->mod_len = em->len; set_bit(EXTENT_FLAG_PINNED, &em->flags); set_bit(EXTENT_FLAG_FILLING, &em->flags); em->generation = -1; @@ -5966,6 +5972,8 @@ static struct extent_map *create_pinned_em(struct inode *inode, u64 start, em->start = start; em->orig_start = orig_start; + em->mod_start = start; + em->mod_len = len; em->len = len; em->block_len = block_len; em->block_start = block_start; diff --git a/fs/btrfs/ordered-data.c b/fs/btrfs/ordered-data.c index e5ed56729607..f14b17432117 100644 --- a/fs/btrfs/ordered-data.c +++ b/fs/btrfs/ordered-data.c @@ -196,6 +196,9 @@ static int __btrfs_add_ordered_extent(struct inode *inode, u64 file_offset, entry->file_offset = file_offset; entry->start = start; entry->len = len; + if (!(BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM) && + !(type == BTRFS_ORDERED_NOCOW)) + entry->csum_bytes_left = disk_len; entry->disk_len = disk_len; entry->bytes_left = len; entry->inode = igrab(inode); @@ -213,6 +216,7 @@ static int __btrfs_add_ordered_extent(struct inode *inode, u64 file_offset, INIT_LIST_HEAD(&entry->root_extent_list); INIT_LIST_HEAD(&entry->work_list); init_completion(&entry->completion); + INIT_LIST_HEAD(&entry->log_list); trace_btrfs_ordered_extent_add(inode, entry); @@ -270,6 +274,10 @@ void btrfs_add_ordered_sum(struct inode *inode, tree = &BTRFS_I(inode)->ordered_tree; spin_lock_irq(&tree->lock); list_add_tail(&sum->list, &entry->list); + WARN_ON(entry->csum_bytes_left < sum->len); + entry->csum_bytes_left -= sum->len; + if (entry->csum_bytes_left == 0) + wake_up(&entry->wait); spin_unlock_irq(&tree->lock); } @@ -405,6 +413,66 @@ out: return ret == 0; } +/* Needs to either be called under a log transaction or the log_mutex */ +void btrfs_get_logged_extents(struct btrfs_root *log, struct inode *inode) +{ + struct btrfs_ordered_inode_tree *tree; + struct btrfs_ordered_extent *ordered; + struct rb_node *n; + int index = log->log_transid % 2; + + tree = &BTRFS_I(inode)->ordered_tree; + spin_lock_irq(&tree->lock); + for (n = rb_first(&tree->tree); n; n = rb_next(n)) { + ordered = rb_entry(n, struct btrfs_ordered_extent, rb_node); + spin_lock(&log->log_extents_lock[index]); + if (list_empty(&ordered->log_list)) { + list_add_tail(&ordered->log_list, &log->logged_list[index]); + atomic_inc(&ordered->refs); + } + spin_unlock(&log->log_extents_lock[index]); + } + spin_unlock_irq(&tree->lock); +} + +void btrfs_wait_logged_extents(struct btrfs_root *log, u64 transid) +{ + struct btrfs_ordered_extent *ordered; + int index = transid % 2; + + spin_lock_irq(&log->log_extents_lock[index]); + while (!list_empty(&log->logged_list[index])) { + ordered = list_first_entry(&log->logged_list[index], + struct btrfs_ordered_extent, + log_list); + list_del_init(&ordered->log_list); + spin_unlock_irq(&log->log_extents_lock[index]); + wait_event(ordered->wait, test_bit(BTRFS_ORDERED_IO_DONE, + &ordered->flags)); + btrfs_put_ordered_extent(ordered); + spin_lock_irq(&log->log_extents_lock[index]); + } + spin_unlock_irq(&log->log_extents_lock[index]); +} + +void btrfs_free_logged_extents(struct btrfs_root *log, u64 transid) +{ + struct btrfs_ordered_extent *ordered; + int index = transid % 2; + + spin_lock_irq(&log->log_extents_lock[index]); + while (!list_empty(&log->logged_list[index])) { + ordered = list_first_entry(&log->logged_list[index], + struct btrfs_ordered_extent, + log_list); + list_del_init(&ordered->log_list); + spin_unlock_irq(&log->log_extents_lock[index]); + btrfs_put_ordered_extent(ordered); + spin_lock_irq(&log->log_extents_lock[index]); + } + spin_unlock_irq(&log->log_extents_lock[index]); +} + /* * used to drop a reference on an ordered extent. This will free * the extent if the last reference is dropped diff --git a/fs/btrfs/ordered-data.h b/fs/btrfs/ordered-data.h index efc7c2930c17..d523dbd2314d 100644 --- a/fs/btrfs/ordered-data.h +++ b/fs/btrfs/ordered-data.h @@ -79,6 +79,8 @@ struct btrfs_ordered_sum { #define BTRFS_ORDERED_UPDATED_ISIZE 7 /* indicates wether this ordered extent * has done its due diligence in updating * the isize. */ +#define BTRFS_ORDERED_LOGGED_CSUM 8 /* We've logged the csums on this ordered + ordered extent */ struct btrfs_ordered_extent { /* logical offset in the file */ @@ -96,6 +98,9 @@ struct btrfs_ordered_extent { /* number of bytes that still need writing */ u64 bytes_left; + /* number of bytes that still need csumming */ + u64 csum_bytes_left; + /* * the end of the ordered extent which is behind it but * didn't update disk_i_size. Please see the comment of @@ -118,6 +123,9 @@ struct btrfs_ordered_extent { /* list of checksums for insertion when the extent io is done */ struct list_head list; + /* If we need to wait on this to be done */ + struct list_head log_list; + /* used to wait for the BTRFS_ORDERED_COMPLETE bit */ wait_queue_head_t wait; @@ -194,6 +202,9 @@ void btrfs_add_ordered_operation(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct inode *inode); void btrfs_wait_ordered_extents(struct btrfs_root *root, int delay_iput); +void btrfs_get_logged_extents(struct btrfs_root *log, struct inode *inode); +void btrfs_wait_logged_extents(struct btrfs_root *log, u64 transid); +void btrfs_free_logged_extents(struct btrfs_root *log, u64 transid); int __init ordered_data_init(void); void ordered_data_exit(void); #endif diff --git a/fs/btrfs/tree-log.c b/fs/btrfs/tree-log.c index 9027bb1e7466..7de720d22b74 100644 --- a/fs/btrfs/tree-log.c +++ b/fs/btrfs/tree-log.c @@ -2281,6 +2281,7 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans, unsigned long log_transid = 0; mutex_lock(&root->log_mutex); + log_transid = root->log_transid; index1 = root->log_transid % 2; if (atomic_read(&root->log_commit[index1])) { wait_log_commit(trans, root, root->log_transid); @@ -2308,11 +2309,11 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans, /* bail out if we need to do a full commit */ if (root->fs_info->last_trans_log_full_commit == trans->transid) { ret = -EAGAIN; + btrfs_free_logged_extents(log, log_transid); mutex_unlock(&root->log_mutex); goto out; } - log_transid = root->log_transid; if (log_transid % 2 == 0) mark = EXTENT_DIRTY; else @@ -2324,6 +2325,7 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans, ret = btrfs_write_marked_extents(log, &log->dirty_log_pages, mark); if (ret) { btrfs_abort_transaction(trans, root, ret); + btrfs_free_logged_extents(log, log_transid); mutex_unlock(&root->log_mutex); goto out; } @@ -2363,6 +2365,7 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans, } root->fs_info->last_trans_log_full_commit = trans->transid; btrfs_wait_marked_extents(log, &log->dirty_log_pages, mark); + btrfs_free_logged_extents(log, log_transid); mutex_unlock(&log_root_tree->log_mutex); ret = -EAGAIN; goto out; @@ -2373,6 +2376,7 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans, btrfs_wait_marked_extents(log, &log->dirty_log_pages, mark); wait_log_commit(trans, log_root_tree, log_root_tree->log_transid); + btrfs_free_logged_extents(log, log_transid); mutex_unlock(&log_root_tree->log_mutex); ret = 0; goto out; @@ -2392,6 +2396,7 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans, */ if (root->fs_info->last_trans_log_full_commit == trans->transid) { btrfs_wait_marked_extents(log, &log->dirty_log_pages, mark); + btrfs_free_logged_extents(log, log_transid); mutex_unlock(&log_root_tree->log_mutex); ret = -EAGAIN; goto out_wake_log_root; @@ -2402,10 +2407,12 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans, EXTENT_DIRTY | EXTENT_NEW); if (ret) { btrfs_abort_transaction(trans, root, ret); + btrfs_free_logged_extents(log, log_transid); mutex_unlock(&log_root_tree->log_mutex); goto out_wake_log_root; } btrfs_wait_marked_extents(log, &log->dirty_log_pages, mark); + btrfs_wait_logged_extents(log, log_transid); btrfs_set_super_log_root(root->fs_info->super_for_commit, log_root_tree->node->start); @@ -2475,6 +2482,14 @@ static void free_log_tree(struct btrfs_trans_handle *trans, EXTENT_DIRTY | EXTENT_NEW, GFP_NOFS); } + /* + * We may have short-circuited the log tree with the full commit logic + * and left ordered extents on our list, so clear these out to keep us + * from leaking inodes and memory. + */ + btrfs_free_logged_extents(log, 0); + btrfs_free_logged_extents(log, 1); + free_extent_buffer(log->node); kfree(log); } @@ -3271,14 +3286,18 @@ static int log_one_extent(struct btrfs_trans_handle *trans, struct btrfs_root *log = root->log_root; struct btrfs_file_extent_item *fi; struct extent_buffer *leaf; + struct btrfs_ordered_extent *ordered; struct list_head ordered_sums; struct btrfs_map_token token; struct btrfs_key key; - u64 csum_offset = em->mod_start - em->start; - u64 csum_len = em->mod_len; + u64 mod_start = em->mod_start; + u64 mod_len = em->mod_len; + u64 csum_offset; + u64 csum_len; u64 extent_offset = em->start - em->orig_start; u64 block_len; int ret; + int index = log->log_transid % 2; bool skip_csum = BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM; INIT_LIST_HEAD(&ordered_sums); @@ -3362,6 +3381,92 @@ static int log_one_extent(struct btrfs_trans_handle *trans, csum_len = block_len; } + /* + * First check and see if our csums are on our outstanding ordered + * extents. + */ +again: + spin_lock_irq(&log->log_extents_lock[index]); + list_for_each_entry(ordered, &log->logged_list[index], log_list) { + struct btrfs_ordered_sum *sum; + + if (!mod_len) + break; + + if (ordered->inode != inode) + continue; + + if (ordered->file_offset + ordered->len <= mod_start || + mod_start + mod_len <= ordered->file_offset) + continue; + + /* + * We are going to copy all the csums on this ordered extent, so + * go ahead and adjust mod_start and mod_len in case this + * ordered extent has already been logged. + */ + if (ordered->file_offset > mod_start) { + if (ordered->file_offset + ordered->len >= + mod_start + mod_len) + mod_len = ordered->file_offset - mod_start; + /* + * If we have this case + * + * |--------- logged extent ---------| + * |----- ordered extent ----| + * + * Just don't mess with mod_start and mod_len, we'll + * just end up logging more csums than we need and it + * will be ok. + */ + } else { + if (ordered->file_offset + ordered->len < + mod_start + mod_len) { + mod_len = (mod_start + mod_len) - + (ordered->file_offset + ordered->len); + mod_start = ordered->file_offset + + ordered->len; + } else { + mod_len = 0; + } + } + + /* + * To keep us from looping for the above case of an ordered + * extent that falls inside of the logged extent. + */ + if (test_and_set_bit(BTRFS_ORDERED_LOGGED_CSUM, + &ordered->flags)) + continue; + atomic_inc(&ordered->refs); + spin_unlock_irq(&log->log_extents_lock[index]); + /* + * we've dropped the lock, we must either break or + * start over after this. + */ + + wait_event(ordered->wait, ordered->csum_bytes_left == 0); + + list_for_each_entry(sum, &ordered->list, list) { + ret = btrfs_csum_file_blocks(trans, log, sum); + if (ret) { + btrfs_put_ordered_extent(ordered); + goto unlocked; + } + } + btrfs_put_ordered_extent(ordered); + goto again; + + } + spin_unlock_irq(&log->log_extents_lock[index]); +unlocked: + + if (!mod_len || ret) + return ret; + + csum_offset = mod_start - em->start; + csum_len = mod_len; + /* block start is already adjusted for the file extent offset. */ ret = btrfs_lookup_csums_range(log->fs_info->csum_root, em->block_start + csum_offset, @@ -3393,6 +3498,7 @@ static int btrfs_log_changed_extents(struct btrfs_trans_handle *trans, struct extent_map_tree *tree = &BTRFS_I(inode)->extent_tree; u64 test_gen; int ret = 0; + int num = 0; INIT_LIST_HEAD(&extents); @@ -3401,16 +3507,31 @@ static int btrfs_log_changed_extents(struct btrfs_trans_handle *trans, list_for_each_entry_safe(em, n, &tree->modified_extents, list) { list_del_init(&em->list); + + /* + * Just an arbitrary number, this can be really CPU intensive + * once we start getting a lot of extents, and really once we + * have a bunch of extents we just want to commit since it will + * be faster. + */ + if (++num > 32768) { + list_del_init(&tree->modified_extents); + ret = -EFBIG; + goto process; + } + if (em->generation <= test_gen) continue; /* Need a ref to keep it from getting evicted from cache */ atomic_inc(&em->refs); set_bit(EXTENT_FLAG_LOGGING, &em->flags); list_add_tail(&em->list, &extents); + num++; } list_sort(NULL, &extents, extent_cmp); +process: while (!list_empty(&extents)) { em = list_entry(extents.next, struct extent_map, list); @@ -3513,6 +3634,8 @@ static int btrfs_log_inode(struct btrfs_trans_handle *trans, mutex_lock(&BTRFS_I(inode)->log_mutex); + btrfs_get_logged_extents(log, inode); + /* * a brute force approach to making sure we get the most uptodate * copies of everything. @@ -3656,6 +3779,8 @@ log_extents: BTRFS_I(inode)->logged_trans = trans->transid; BTRFS_I(inode)->last_log_commit = BTRFS_I(inode)->last_sub_trans; out_unlock: + if (err) + btrfs_free_logged_extents(log, log->log_transid); mutex_unlock(&BTRFS_I(inode)->log_mutex); btrfs_free_path(path); @@ -3822,7 +3947,6 @@ int btrfs_log_inode_parent(struct btrfs_trans_handle *trans, end_trans: dput(old_parent); if (ret < 0) { - WARN_ON(ret != -ENOSPC); root->fs_info->last_trans_log_full_commit = trans->transid; ret = 1; } -- cgit v1.2.3 From dcfac4156fa102c1bab0e4e31df37e47278292f6 Mon Sep 17 00:00:00 2001 From: Liu Bo Date: Thu, 27 Dec 2012 09:01:20 +0000 Subject: Btrfs: kill unused argument of btrfs_pin_extent_for_log_replay Argument 'trans' is not used any more. Signed-off-by: Liu Bo Signed-off-by: Josef Bacik --- fs/btrfs/ctree.h | 3 +-- fs/btrfs/extent-tree.c | 3 +-- fs/btrfs/tree-log.c | 3 +-- 3 files changed, 3 insertions(+), 6 deletions(-) (limited to 'fs/btrfs/ctree.h') diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 411c8d97074e..22f012d41fd0 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -2939,8 +2939,7 @@ int btrfs_lookup_extent_info(struct btrfs_trans_handle *trans, u64 num_bytes, u64 *refs, u64 *flags); int btrfs_pin_extent(struct btrfs_root *root, u64 bytenr, u64 num, int reserved); -int btrfs_pin_extent_for_log_replay(struct btrfs_trans_handle *trans, - struct btrfs_root *root, +int btrfs_pin_extent_for_log_replay(struct btrfs_root *root, u64 bytenr, u64 num_bytes); int btrfs_cross_ref_exist(struct btrfs_trans_handle *trans, struct btrfs_root *root, diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index a162c7cb8fb5..825f23b13b58 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -4898,8 +4898,7 @@ int btrfs_pin_extent(struct btrfs_root *root, /* * this function must be called within transaction */ -int btrfs_pin_extent_for_log_replay(struct btrfs_trans_handle *trans, - struct btrfs_root *root, +int btrfs_pin_extent_for_log_replay(struct btrfs_root *root, u64 bytenr, u64 num_bytes) { struct btrfs_block_group_cache *cache; diff --git a/fs/btrfs/tree-log.c b/fs/btrfs/tree-log.c index 7de720d22b74..deb7f66356bd 100644 --- a/fs/btrfs/tree-log.c +++ b/fs/btrfs/tree-log.c @@ -278,8 +278,7 @@ static int process_one_buffer(struct btrfs_root *log, struct walk_control *wc, u64 gen) { if (wc->pin) - btrfs_pin_extent_for_log_replay(wc->trans, - log->fs_info->extent_root, + btrfs_pin_extent_for_log_replay(log->fs_info->extent_root, eb->start, eb->len); if (btrfs_buffer_uptodate(eb, gen, 0)) { -- cgit v1.2.3 From a1897fddd28daf6b23d05a30dc2a18836f77f8e3 Mon Sep 17 00:00:00 2001 From: Liu Bo Date: Thu, 27 Dec 2012 09:01:23 +0000 Subject: Btrfs: record first logical byte in memory This'd save us a rbtree search which may become expensive in large filesystem. Signed-off-by: Liu Bo Signed-off-by: Josef Bacik --- fs/btrfs/ctree.h | 1 + fs/btrfs/disk-io.c | 1 + fs/btrfs/extent-tree.c | 19 ++++++++++++++++++- 3 files changed, 20 insertions(+), 1 deletion(-) (limited to 'fs/btrfs/ctree.h') diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 22f012d41fd0..9ee099f3f834 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -1250,6 +1250,7 @@ struct btrfs_fs_info { /* block group cache stuff */ spinlock_t block_group_cache_lock; + u64 first_logical_byte; struct rb_root block_group_cache_tree; /* keep track of unallocated space */ diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 1db8a9938829..04f98e3ffd90 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -2130,6 +2130,7 @@ int open_ctree(struct super_block *sb, spin_lock_init(&fs_info->block_group_cache_lock); fs_info->block_group_cache_tree = RB_ROOT; + fs_info->first_logical_byte = (u64)-1; extent_io_tree_init(&fs_info->freed_extents[0], fs_info->btree_inode->i_mapping); diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 825f23b13b58..82400b2b2517 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -161,6 +161,10 @@ static int btrfs_add_block_group_cache(struct btrfs_fs_info *info, rb_link_node(&block_group->cache_node, parent, p); rb_insert_color(&block_group->cache_node, &info->block_group_cache_tree); + + if (info->first_logical_byte > block_group->key.objectid) + info->first_logical_byte = block_group->key.objectid; + spin_unlock(&info->block_group_cache_lock); return 0; @@ -202,8 +206,11 @@ block_group_cache_tree_search(struct btrfs_fs_info *info, u64 bytenr, break; } } - if (ret) + if (ret) { btrfs_get_block_group(ret); + if (bytenr == 0 && info->first_logical_byte > ret->key.objectid) + info->first_logical_byte = ret->key.objectid; + } spin_unlock(&info->block_group_cache_lock); return ret; @@ -4848,6 +4855,13 @@ static u64 first_logical_byte(struct btrfs_root *root, u64 search_start) struct btrfs_block_group_cache *cache; u64 bytenr; + spin_lock(&root->fs_info->block_group_cache_lock); + bytenr = root->fs_info->first_logical_byte; + spin_unlock(&root->fs_info->block_group_cache_lock); + + if (bytenr < (u64)-1) + return bytenr; + cache = btrfs_lookup_first_block_group(root->fs_info, search_start); if (!cache) return 0; @@ -8059,6 +8073,9 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans, spin_lock(&root->fs_info->block_group_cache_lock); rb_erase(&block_group->cache_node, &root->fs_info->block_group_cache_tree); + + if (root->fs_info->first_logical_byte == block_group->key.objectid) + root->fs_info->first_logical_byte = (u64)-1; spin_unlock(&root->fs_info->block_group_cache_lock); down_write(&block_group->space_info->groups_sem); -- cgit v1.2.3 From e6ec716f0ddbe51741ef261d0804f0c28038dda4 Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Thu, 17 Jan 2013 05:38:51 +0000 Subject: Btrfs: make raid attr array more readable The current code of raid attr arry is hard to understand and it is easy to introduce some problem if we modify the array. So I changed it and made it more readable. Cc: Liu Bo Signed-off-by: Miao Xie Signed-off-by: Josef Bacik --- fs/btrfs/ctree.h | 10 +++++++++- fs/btrfs/extent-tree.c | 22 +++++++++------------- fs/btrfs/volumes.c | 47 +++++++++++++++++++++++++++++++++++++++++------ 3 files changed, 59 insertions(+), 20 deletions(-) (limited to 'fs/btrfs/ctree.h') diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 9ee099f3f834..541ce9a9949e 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -953,7 +953,15 @@ struct btrfs_dev_replace_item { #define BTRFS_BLOCK_GROUP_DUP (1ULL << 5) #define BTRFS_BLOCK_GROUP_RAID10 (1ULL << 6) #define BTRFS_BLOCK_GROUP_RESERVED BTRFS_AVAIL_ALLOC_BIT_SINGLE -#define BTRFS_NR_RAID_TYPES 5 + +enum btrfs_raid_types { + BTRFS_RAID_RAID10, + BTRFS_RAID_RAID1, + BTRFS_RAID_DUP, + BTRFS_RAID_RAID0, + BTRFS_RAID_SINGLE, + BTRFS_NR_RAID_TYPES +}; #define BTRFS_BLOCK_GROUP_TYPE_MASK (BTRFS_BLOCK_GROUP_DATA | \ BTRFS_BLOCK_GROUP_SYSTEM | \ diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 82400b2b2517..174c4d5c692c 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -5545,20 +5545,16 @@ wait_block_group_cache_done(struct btrfs_block_group_cache *cache) int __get_raid_index(u64 flags) { - int index; - if (flags & BTRFS_BLOCK_GROUP_RAID10) - index = 0; + return BTRFS_RAID_RAID10; else if (flags & BTRFS_BLOCK_GROUP_RAID1) - index = 1; + return BTRFS_RAID_RAID1; else if (flags & BTRFS_BLOCK_GROUP_DUP) - index = 2; + return BTRFS_RAID_DUP; else if (flags & BTRFS_BLOCK_GROUP_RAID0) - index = 3; + return BTRFS_RAID_RAID0; else - index = 4; - - return index; + return BTRFS_RAID_SINGLE; } static int get_block_group_index(struct btrfs_block_group_cache *cache) @@ -7518,16 +7514,16 @@ int btrfs_can_relocate(struct btrfs_root *root, u64 bytenr) index = get_block_group_index(block_group); } - if (index == 0) { + if (index == BTRFS_RAID_RAID10) { dev_min = 4; /* Divide by 2 */ min_free >>= 1; - } else if (index == 1) { + } else if (index == BTRFS_RAID_RAID1) { dev_min = 2; - } else if (index == 2) { + } else if (index == BTRFS_RAID_DUP) { /* Multiply by 2 */ min_free <<= 1; - } else if (index == 3) { + } else if (index == BTRFS_RAID_RAID0) { dev_min = fs_devices->rw_devices; do_div(min_free, dev_min); } diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c index 5349e17d8863..1ce581d58d87 100644 --- a/fs/btrfs/volumes.c +++ b/fs/btrfs/volumes.c @@ -3562,13 +3562,48 @@ static int btrfs_cmp_device_info(const void *a, const void *b) } struct btrfs_raid_attr btrfs_raid_array[BTRFS_NR_RAID_TYPES] = { - { 2, 1, 0, 4, 2, 2 /* raid10 */ }, - { 1, 1, 2, 2, 2, 2 /* raid1 */ }, - { 1, 2, 1, 1, 1, 2 /* dup */ }, - { 1, 1, 0, 2, 1, 1 /* raid0 */ }, - { 1, 1, 1, 1, 1, 1 /* single */ }, + [BTRFS_RAID_RAID10] = { + .sub_stripes = 2, + .dev_stripes = 1, + .devs_max = 0, /* 0 == as many as possible */ + .devs_min = 4, + .devs_increment = 2, + .ncopies = 2, + }, + [BTRFS_RAID_RAID1] = { + .sub_stripes = 1, + .dev_stripes = 1, + .devs_max = 2, + .devs_min = 2, + .devs_increment = 2, + .ncopies = 2, + }, + [BTRFS_RAID_DUP] = { + .sub_stripes = 1, + .dev_stripes = 2, + .devs_max = 1, + .devs_min = 1, + .devs_increment = 1, + .ncopies = 2, + }, + [BTRFS_RAID_RAID0] = { + .sub_stripes = 1, + .dev_stripes = 1, + .devs_max = 0, + .devs_min = 2, + .devs_increment = 1, + .ncopies = 1, + }, + [BTRFS_RAID_SINGLE] = { + .sub_stripes = 1, + .dev_stripes = 1, + .devs_max = 1, + .devs_min = 1, + .devs_increment = 1, + .ncopies = 1, + }, }; - + static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans, struct btrfs_root *extent_root, struct map_lookup **map_ret, -- cgit v1.2.3 From 55e301fd57a6239ec14b91a1cf2e70b3dd135194 Mon Sep 17 00:00:00 2001 From: Filipe Brandenburger Date: Tue, 29 Jan 2013 06:04:50 +0000 Subject: Btrfs: move fs/btrfs/ioctl.h to include/uapi/linux/btrfs.h The header file will then be installed under /usr/include/linux so that userspace applications can refer to Btrfs ioctls by name and use the same structs used internally in the kernel. Signed-off-by: Filipe Brandenburger Signed-off-by: Josef Bacik --- fs/btrfs/backref.h | 2 +- fs/btrfs/ctree.h | 2 +- fs/btrfs/file.c | 2 +- fs/btrfs/inode.c | 2 +- fs/btrfs/ioctl.c | 2 +- fs/btrfs/ioctl.h | 502 -------------------------------------------- fs/btrfs/qgroup.c | 2 +- fs/btrfs/super.c | 2 +- fs/btrfs/volumes.h | 2 +- include/linux/btrfs.h | 6 + include/uapi/linux/Kbuild | 1 + include/uapi/linux/btrfs.h | 503 +++++++++++++++++++++++++++++++++++++++++++++ 12 files changed, 518 insertions(+), 510 deletions(-) delete mode 100644 fs/btrfs/ioctl.h create mode 100644 include/linux/btrfs.h create mode 100644 include/uapi/linux/btrfs.h (limited to 'fs/btrfs/ctree.h') diff --git a/fs/btrfs/backref.h b/fs/btrfs/backref.h index d61feca79455..310a7f6d09b1 100644 --- a/fs/btrfs/backref.h +++ b/fs/btrfs/backref.h @@ -19,7 +19,7 @@ #ifndef __BTRFS_BACKREF__ #define __BTRFS_BACKREF__ -#include "ioctl.h" +#include #include "ulist.h" #include "extent_io.h" diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 541ce9a9949e..69321013683c 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -31,10 +31,10 @@ #include #include #include +#include #include "extent_io.h" #include "extent_map.h" #include "async-thread.h" -#include "ioctl.h" struct btrfs_trans_handle; struct btrfs_transaction; diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c index 083abca56055..13c78ea3ebce 100644 --- a/fs/btrfs/file.c +++ b/fs/btrfs/file.c @@ -30,11 +30,11 @@ #include #include #include +#include #include "ctree.h" #include "disk-io.h" #include "transaction.h" #include "btrfs_inode.h" -#include "ioctl.h" #include "print-tree.h" #include "tree-log.h" #include "locking.h" diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 60ec7589900c..fc8aa8bf80a1 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -39,12 +39,12 @@ #include #include #include +#include #include "compat.h" #include "ctree.h" #include "disk-io.h" #include "transaction.h" #include "btrfs_inode.h" -#include "ioctl.h" #include "print-tree.h" #include "ordered-data.h" #include "xattr.h" diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index 1b554b47e814..96ecefc1724f 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -42,12 +42,12 @@ #include #include #include +#include #include "compat.h" #include "ctree.h" #include "disk-io.h" #include "transaction.h" #include "btrfs_inode.h" -#include "ioctl.h" #include "print-tree.h" #include "volumes.h" #include "locking.h" diff --git a/fs/btrfs/ioctl.h b/fs/btrfs/ioctl.h deleted file mode 100644 index dabca9cc8c2e..000000000000 --- a/fs/btrfs/ioctl.h +++ /dev/null @@ -1,502 +0,0 @@ -/* - * Copyright (C) 2007 Oracle. All rights reserved. - * - * This program is free software; you can redistribute it and/or - * modify it under the terms of the GNU General Public - * License v2 as published by the Free Software Foundation. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU - * General Public License for more details. - * - * You should have received a copy of the GNU General Public - * License along with this program; if not, write to the - * Free Software Foundation, Inc., 59 Temple Place - Suite 330, - * Boston, MA 021110-1307, USA. - */ - -#ifndef __IOCTL_ -#define __IOCTL_ -#include - -#define BTRFS_IOCTL_MAGIC 0x94 -#define BTRFS_VOL_NAME_MAX 255 - -/* this should be 4k */ -#define BTRFS_PATH_NAME_MAX 4087 -struct btrfs_ioctl_vol_args { - __s64 fd; - char name[BTRFS_PATH_NAME_MAX + 1]; -}; - -#define BTRFS_DEVICE_PATH_NAME_MAX 1024 - -#define BTRFS_SUBVOL_CREATE_ASYNC (1ULL << 0) -#define BTRFS_SUBVOL_RDONLY (1ULL << 1) -#define BTRFS_SUBVOL_QGROUP_INHERIT (1ULL << 2) -#define BTRFS_FSID_SIZE 16 -#define BTRFS_UUID_SIZE 16 - -#define BTRFS_QGROUP_INHERIT_SET_LIMITS (1ULL << 0) - -struct btrfs_qgroup_limit { - __u64 flags; - __u64 max_rfer; - __u64 max_excl; - __u64 rsv_rfer; - __u64 rsv_excl; -}; - -struct btrfs_qgroup_inherit { - __u64 flags; - __u64 num_qgroups; - __u64 num_ref_copies; - __u64 num_excl_copies; - struct btrfs_qgroup_limit lim; - __u64 qgroups[0]; -}; - -struct btrfs_ioctl_qgroup_limit_args { - __u64 qgroupid; - struct btrfs_qgroup_limit lim; -}; - -#define BTRFS_SUBVOL_NAME_MAX 4039 -struct btrfs_ioctl_vol_args_v2 { - __s64 fd; - __u64 transid; - __u64 flags; - union { - struct { - __u64 size; - struct btrfs_qgroup_inherit __user *qgroup_inherit; - }; - __u64 unused[4]; - }; - char name[BTRFS_SUBVOL_NAME_MAX + 1]; -}; - -/* - * structure to report errors and progress to userspace, either as a - * result of a finished scrub, a canceled scrub or a progress inquiry - */ -struct btrfs_scrub_progress { - __u64 data_extents_scrubbed; /* # of data extents scrubbed */ - __u64 tree_extents_scrubbed; /* # of tree extents scrubbed */ - __u64 data_bytes_scrubbed; /* # of data bytes scrubbed */ - __u64 tree_bytes_scrubbed; /* # of tree bytes scrubbed */ - __u64 read_errors; /* # of read errors encountered (EIO) */ - __u64 csum_errors; /* # of failed csum checks */ - __u64 verify_errors; /* # of occurences, where the metadata - * of a tree block did not match the - * expected values, like generation or - * logical */ - __u64 no_csum; /* # of 4k data block for which no csum - * is present, probably the result of - * data written with nodatasum */ - __u64 csum_discards; /* # of csum for which no data was found - * in the extent tree. */ - __u64 super_errors; /* # of bad super blocks encountered */ - __u64 malloc_errors; /* # of internal kmalloc errors. These - * will likely cause an incomplete - * scrub */ - __u64 uncorrectable_errors; /* # of errors where either no intact - * copy was found or the writeback - * failed */ - __u64 corrected_errors; /* # of errors corrected */ - __u64 last_physical; /* last physical address scrubbed. In - * case a scrub was aborted, this can - * be used to restart the scrub */ - __u64 unverified_errors; /* # of occurences where a read for a - * full (64k) bio failed, but the re- - * check succeeded for each 4k piece. - * Intermittent error. */ -}; - -#define BTRFS_SCRUB_READONLY 1 -struct btrfs_ioctl_scrub_args { - __u64 devid; /* in */ - __u64 start; /* in */ - __u64 end; /* in */ - __u64 flags; /* in */ - struct btrfs_scrub_progress progress; /* out */ - /* pad to 1k */ - __u64 unused[(1024-32-sizeof(struct btrfs_scrub_progress))/8]; -}; - -#define BTRFS_IOCTL_DEV_REPLACE_CONT_READING_FROM_SRCDEV_MODE_ALWAYS 0 -#define BTRFS_IOCTL_DEV_REPLACE_CONT_READING_FROM_SRCDEV_MODE_AVOID 1 -struct btrfs_ioctl_dev_replace_start_params { - __u64 srcdevid; /* in, if 0, use srcdev_name instead */ - __u64 cont_reading_from_srcdev_mode; /* in, see #define - * above */ - __u8 srcdev_name[BTRFS_DEVICE_PATH_NAME_MAX + 1]; /* in */ - __u8 tgtdev_name[BTRFS_DEVICE_PATH_NAME_MAX + 1]; /* in */ -}; - -#define BTRFS_IOCTL_DEV_REPLACE_STATE_NEVER_STARTED 0 -#define BTRFS_IOCTL_DEV_REPLACE_STATE_STARTED 1 -#define BTRFS_IOCTL_DEV_REPLACE_STATE_FINISHED 2 -#define BTRFS_IOCTL_DEV_REPLACE_STATE_CANCELED 3 -#define BTRFS_IOCTL_DEV_REPLACE_STATE_SUSPENDED 4 -struct btrfs_ioctl_dev_replace_status_params { - __u64 replace_state; /* out, see #define above */ - __u64 progress_1000; /* out, 0 <= x <= 1000 */ - __u64 time_started; /* out, seconds since 1-Jan-1970 */ - __u64 time_stopped; /* out, seconds since 1-Jan-1970 */ - __u64 num_write_errors; /* out */ - __u64 num_uncorrectable_read_errors; /* out */ -}; - -#define BTRFS_IOCTL_DEV_REPLACE_CMD_START 0 -#define BTRFS_IOCTL_DEV_REPLACE_CMD_STATUS 1 -#define BTRFS_IOCTL_DEV_REPLACE_CMD_CANCEL 2 -#define BTRFS_IOCTL_DEV_REPLACE_RESULT_NO_ERROR 0 -#define BTRFS_IOCTL_DEV_REPLACE_RESULT_NOT_STARTED 1 -#define BTRFS_IOCTL_DEV_REPLACE_RESULT_ALREADY_STARTED 2 -struct btrfs_ioctl_dev_replace_args { - __u64 cmd; /* in */ - __u64 result; /* out */ - - union { - struct btrfs_ioctl_dev_replace_start_params start; - struct btrfs_ioctl_dev_replace_status_params status; - }; /* in/out */ - - __u64 spare[64]; -}; - -struct btrfs_ioctl_dev_info_args { - __u64 devid; /* in/out */ - __u8 uuid[BTRFS_UUID_SIZE]; /* in/out */ - __u64 bytes_used; /* out */ - __u64 total_bytes; /* out */ - __u64 unused[379]; /* pad to 4k */ - __u8 path[BTRFS_DEVICE_PATH_NAME_MAX]; /* out */ -}; - -struct btrfs_ioctl_fs_info_args { - __u64 max_id; /* out */ - __u64 num_devices; /* out */ - __u8 fsid[BTRFS_FSID_SIZE]; /* out */ - __u64 reserved[124]; /* pad to 1k */ -}; - -/* balance control ioctl modes */ -#define BTRFS_BALANCE_CTL_PAUSE 1 -#define BTRFS_BALANCE_CTL_CANCEL 2 - -/* - * this is packed, because it should be exactly the same as its disk - * byte order counterpart (struct btrfs_disk_balance_args) - */ -struct btrfs_balance_args { - __u64 profiles; - __u64 usage; - __u64 devid; - __u64 pstart; - __u64 pend; - __u64 vstart; - __u64 vend; - - __u64 target; - - __u64 flags; - - __u64 unused[8]; -} __attribute__ ((__packed__)); - -/* report balance progress to userspace */ -struct btrfs_balance_progress { - __u64 expected; /* estimated # of chunks that will be - * relocated to fulfill the request */ - __u64 considered; /* # of chunks we have considered so far */ - __u64 completed; /* # of chunks relocated so far */ -}; - -#define BTRFS_BALANCE_STATE_RUNNING (1ULL << 0) -#define BTRFS_BALANCE_STATE_PAUSE_REQ (1ULL << 1) -#define BTRFS_BALANCE_STATE_CANCEL_REQ (1ULL << 2) - -struct btrfs_ioctl_balance_args { - __u64 flags; /* in/out */ - __u64 state; /* out */ - - struct btrfs_balance_args data; /* in/out */ - struct btrfs_balance_args meta; /* in/out */ - struct btrfs_balance_args sys; /* in/out */ - - struct btrfs_balance_progress stat; /* out */ - - __u64 unused[72]; /* pad to 1k */ -}; - -#define BTRFS_INO_LOOKUP_PATH_MAX 4080 -struct btrfs_ioctl_ino_lookup_args { - __u64 treeid; - __u64 objectid; - char name[BTRFS_INO_LOOKUP_PATH_MAX]; -}; - -struct btrfs_ioctl_search_key { - /* which root are we searching. 0 is the tree of tree roots */ - __u64 tree_id; - - /* keys returned will be >= min and <= max */ - __u64 min_objectid; - __u64 max_objectid; - - /* keys returned will be >= min and <= max */ - __u64 min_offset; - __u64 max_offset; - - /* max and min transids to search for */ - __u64 min_transid; - __u64 max_transid; - - /* keys returned will be >= min and <= max */ - __u32 min_type; - __u32 max_type; - - /* - * how many items did userland ask for, and how many are we - * returning - */ - __u32 nr_items; - - /* align to 64 bits */ - __u32 unused; - - /* some extra for later */ - __u64 unused1; - __u64 unused2; - __u64 unused3; - __u64 unused4; -}; - -struct btrfs_ioctl_search_header { - __u64 transid; - __u64 objectid; - __u64 offset; - __u32 type; - __u32 len; -}; - -#define BTRFS_SEARCH_ARGS_BUFSIZE (4096 - sizeof(struct btrfs_ioctl_search_key)) -/* - * the buf is an array of search headers where - * each header is followed by the actual item - * the type field is expanded to 32 bits for alignment - */ -struct btrfs_ioctl_search_args { - struct btrfs_ioctl_search_key key; - char buf[BTRFS_SEARCH_ARGS_BUFSIZE]; -}; - -struct btrfs_ioctl_clone_range_args { - __s64 src_fd; - __u64 src_offset, src_length; - __u64 dest_offset; -}; - -/* flags for the defrag range ioctl */ -#define BTRFS_DEFRAG_RANGE_COMPRESS 1 -#define BTRFS_DEFRAG_RANGE_START_IO 2 - -struct btrfs_ioctl_space_info { - __u64 flags; - __u64 total_bytes; - __u64 used_bytes; -}; - -struct btrfs_ioctl_space_args { - __u64 space_slots; - __u64 total_spaces; - struct btrfs_ioctl_space_info spaces[0]; -}; - -struct btrfs_data_container { - __u32 bytes_left; /* out -- bytes not needed to deliver output */ - __u32 bytes_missing; /* out -- additional bytes needed for result */ - __u32 elem_cnt; /* out */ - __u32 elem_missed; /* out */ - __u64 val[0]; /* out */ -}; - -struct btrfs_ioctl_ino_path_args { - __u64 inum; /* in */ - __u64 size; /* in */ - __u64 reserved[4]; - /* struct btrfs_data_container *fspath; out */ - __u64 fspath; /* out */ -}; - -struct btrfs_ioctl_logical_ino_args { - __u64 logical; /* in */ - __u64 size; /* in */ - __u64 reserved[4]; - /* struct btrfs_data_container *inodes; out */ - __u64 inodes; -}; - -enum btrfs_dev_stat_values { - /* disk I/O failure stats */ - BTRFS_DEV_STAT_WRITE_ERRS, /* EIO or EREMOTEIO from lower layers */ - BTRFS_DEV_STAT_READ_ERRS, /* EIO or EREMOTEIO from lower layers */ - BTRFS_DEV_STAT_FLUSH_ERRS, /* EIO or EREMOTEIO from lower layers */ - - /* stats for indirect indications for I/O failures */ - BTRFS_DEV_STAT_CORRUPTION_ERRS, /* checksum error, bytenr error or - * contents is illegal: this is an - * indication that the block was damaged - * during read or write, or written to - * wrong location or read from wrong - * location */ - BTRFS_DEV_STAT_GENERATION_ERRS, /* an indication that blocks have not - * been written */ - - BTRFS_DEV_STAT_VALUES_MAX -}; - -/* Reset statistics after reading; needs SYS_ADMIN capability */ -#define BTRFS_DEV_STATS_RESET (1ULL << 0) - -struct btrfs_ioctl_get_dev_stats { - __u64 devid; /* in */ - __u64 nr_items; /* in/out */ - __u64 flags; /* in/out */ - - /* out values: */ - __u64 values[BTRFS_DEV_STAT_VALUES_MAX]; - - __u64 unused[128 - 2 - BTRFS_DEV_STAT_VALUES_MAX]; /* pad to 1k */ -}; - -#define BTRFS_QUOTA_CTL_ENABLE 1 -#define BTRFS_QUOTA_CTL_DISABLE 2 -#define BTRFS_QUOTA_CTL_RESCAN 3 -struct btrfs_ioctl_quota_ctl_args { - __u64 cmd; - __u64 status; -}; - -struct btrfs_ioctl_qgroup_assign_args { - __u64 assign; - __u64 src; - __u64 dst; -}; - -struct btrfs_ioctl_qgroup_create_args { - __u64 create; - __u64 qgroupid; -}; -struct btrfs_ioctl_timespec { - __u64 sec; - __u32 nsec; -}; - -struct btrfs_ioctl_received_subvol_args { - char uuid[BTRFS_UUID_SIZE]; /* in */ - __u64 stransid; /* in */ - __u64 rtransid; /* out */ - struct btrfs_ioctl_timespec stime; /* in */ - struct btrfs_ioctl_timespec rtime; /* out */ - __u64 flags; /* in */ - __u64 reserved[16]; /* in */ -}; - -struct btrfs_ioctl_send_args { - __s64 send_fd; /* in */ - __u64 clone_sources_count; /* in */ - __u64 __user *clone_sources; /* in */ - __u64 parent_root; /* in */ - __u64 flags; /* in */ - __u64 reserved[4]; /* in */ -}; - -#define BTRFS_IOC_SNAP_CREATE _IOW(BTRFS_IOCTL_MAGIC, 1, \ - struct btrfs_ioctl_vol_args) -#define BTRFS_IOC_DEFRAG _IOW(BTRFS_IOCTL_MAGIC, 2, \ - struct btrfs_ioctl_vol_args) -#define BTRFS_IOC_RESIZE _IOW(BTRFS_IOCTL_MAGIC, 3, \ - struct btrfs_ioctl_vol_args) -#define BTRFS_IOC_SCAN_DEV _IOW(BTRFS_IOCTL_MAGIC, 4, \ - struct btrfs_ioctl_vol_args) -/* trans start and trans end are dangerous, and only for - * use by applications that know how to avoid the - * resulting deadlocks - */ -#define BTRFS_IOC_TRANS_START _IO(BTRFS_IOCTL_MAGIC, 6) -#define BTRFS_IOC_TRANS_END _IO(BTRFS_IOCTL_MAGIC, 7) -#define BTRFS_IOC_SYNC _IO(BTRFS_IOCTL_MAGIC, 8) - -#define BTRFS_IOC_CLONE _IOW(BTRFS_IOCTL_MAGIC, 9, int) -#define BTRFS_IOC_ADD_DEV _IOW(BTRFS_IOCTL_MAGIC, 10, \ - struct btrfs_ioctl_vol_args) -#define BTRFS_IOC_RM_DEV _IOW(BTRFS_IOCTL_MAGIC, 11, \ - struct btrfs_ioctl_vol_args) -#define BTRFS_IOC_BALANCE _IOW(BTRFS_IOCTL_MAGIC, 12, \ - struct btrfs_ioctl_vol_args) - -#define BTRFS_IOC_CLONE_RANGE _IOW(BTRFS_IOCTL_MAGIC, 13, \ - struct btrfs_ioctl_clone_range_args) - -#define BTRFS_IOC_SUBVOL_CREATE _IOW(BTRFS_IOCTL_MAGIC, 14, \ - struct btrfs_ioctl_vol_args) -#define BTRFS_IOC_SNAP_DESTROY _IOW(BTRFS_IOCTL_MAGIC, 15, \ - struct btrfs_ioctl_vol_args) -#define BTRFS_IOC_DEFRAG_RANGE _IOW(BTRFS_IOCTL_MAGIC, 16, \ - struct btrfs_ioctl_defrag_range_args) -#define BTRFS_IOC_TREE_SEARCH _IOWR(BTRFS_IOCTL_MAGIC, 17, \ - struct btrfs_ioctl_search_args) -#define BTRFS_IOC_INO_LOOKUP _IOWR(BTRFS_IOCTL_MAGIC, 18, \ - struct btrfs_ioctl_ino_lookup_args) -#define BTRFS_IOC_DEFAULT_SUBVOL _IOW(BTRFS_IOCTL_MAGIC, 19, u64) -#define BTRFS_IOC_SPACE_INFO _IOWR(BTRFS_IOCTL_MAGIC, 20, \ - struct btrfs_ioctl_space_args) -#define BTRFS_IOC_START_SYNC _IOR(BTRFS_IOCTL_MAGIC, 24, __u64) -#define BTRFS_IOC_WAIT_SYNC _IOW(BTRFS_IOCTL_MAGIC, 22, __u64) -#define BTRFS_IOC_SNAP_CREATE_V2 _IOW(BTRFS_IOCTL_MAGIC, 23, \ - struct btrfs_ioctl_vol_args_v2) -#define BTRFS_IOC_SUBVOL_CREATE_V2 _IOW(BTRFS_IOCTL_MAGIC, 24, \ - struct btrfs_ioctl_vol_args_v2) -#define BTRFS_IOC_SUBVOL_GETFLAGS _IOR(BTRFS_IOCTL_MAGIC, 25, __u64) -#define BTRFS_IOC_SUBVOL_SETFLAGS _IOW(BTRFS_IOCTL_MAGIC, 26, __u64) -#define BTRFS_IOC_SCRUB _IOWR(BTRFS_IOCTL_MAGIC, 27, \ - struct btrfs_ioctl_scrub_args) -#define BTRFS_IOC_SCRUB_CANCEL _IO(BTRFS_IOCTL_MAGIC, 28) -#define BTRFS_IOC_SCRUB_PROGRESS _IOWR(BTRFS_IOCTL_MAGIC, 29, \ - struct btrfs_ioctl_scrub_args) -#define BTRFS_IOC_DEV_INFO _IOWR(BTRFS_IOCTL_MAGIC, 30, \ - struct btrfs_ioctl_dev_info_args) -#define BTRFS_IOC_FS_INFO _IOR(BTRFS_IOCTL_MAGIC, 31, \ - struct btrfs_ioctl_fs_info_args) -#define BTRFS_IOC_BALANCE_V2 _IOWR(BTRFS_IOCTL_MAGIC, 32, \ - struct btrfs_ioctl_balance_args) -#define BTRFS_IOC_BALANCE_CTL _IOW(BTRFS_IOCTL_MAGIC, 33, int) -#define BTRFS_IOC_BALANCE_PROGRESS _IOR(BTRFS_IOCTL_MAGIC, 34, \ - struct btrfs_ioctl_balance_args) -#define BTRFS_IOC_INO_PATHS _IOWR(BTRFS_IOCTL_MAGIC, 35, \ - struct btrfs_ioctl_ino_path_args) -#define BTRFS_IOC_LOGICAL_INO _IOWR(BTRFS_IOCTL_MAGIC, 36, \ - struct btrfs_ioctl_ino_path_args) -#define BTRFS_IOC_SET_RECEIVED_SUBVOL _IOWR(BTRFS_IOCTL_MAGIC, 37, \ - struct btrfs_ioctl_received_subvol_args) -#define BTRFS_IOC_SEND _IOW(BTRFS_IOCTL_MAGIC, 38, struct btrfs_ioctl_send_args) -#define BTRFS_IOC_DEVICES_READY _IOR(BTRFS_IOCTL_MAGIC, 39, \ - struct btrfs_ioctl_vol_args) -#define BTRFS_IOC_QUOTA_CTL _IOWR(BTRFS_IOCTL_MAGIC, 40, \ - struct btrfs_ioctl_quota_ctl_args) -#define BTRFS_IOC_QGROUP_ASSIGN _IOW(BTRFS_IOCTL_MAGIC, 41, \ - struct btrfs_ioctl_qgroup_assign_args) -#define BTRFS_IOC_QGROUP_CREATE _IOW(BTRFS_IOCTL_MAGIC, 42, \ - struct btrfs_ioctl_qgroup_create_args) -#define BTRFS_IOC_QGROUP_LIMIT _IOR(BTRFS_IOCTL_MAGIC, 43, \ - struct btrfs_ioctl_qgroup_limit_args) -#define BTRFS_IOC_GET_DEV_STATS _IOWR(BTRFS_IOCTL_MAGIC, 52, \ - struct btrfs_ioctl_get_dev_stats) -#define BTRFS_IOC_DEV_REPLACE _IOWR(BTRFS_IOCTL_MAGIC, 53, \ - struct btrfs_ioctl_dev_replace_args) - -#endif diff --git a/fs/btrfs/qgroup.c b/fs/btrfs/qgroup.c index a5c856234323..a0d6368249fa 100644 --- a/fs/btrfs/qgroup.c +++ b/fs/btrfs/qgroup.c @@ -23,13 +23,13 @@ #include #include #include +#include #include "ctree.h" #include "transaction.h" #include "disk-io.h" #include "locking.h" #include "ulist.h" -#include "ioctl.h" #include "backref.h" /* TODO XXX FIXME diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c index 67b373bf3ff9..6846ededfe95 100644 --- a/fs/btrfs/super.c +++ b/fs/btrfs/super.c @@ -41,13 +41,13 @@ #include #include #include +#include #include "compat.h" #include "delayed-inode.h" #include "ctree.h" #include "disk-io.h" #include "transaction.h" #include "btrfs_inode.h" -#include "ioctl.h" #include "print-tree.h" #include "xattr.h" #include "volumes.h" diff --git a/fs/btrfs/volumes.h b/fs/btrfs/volumes.h index d3c3939ac751..12bb84166a5f 100644 --- a/fs/btrfs/volumes.h +++ b/fs/btrfs/volumes.h @@ -21,8 +21,8 @@ #include #include +#include #include "async-thread.h" -#include "ioctl.h" #define BTRFS_STRIPE_LEN (64 * 1024) diff --git a/include/linux/btrfs.h b/include/linux/btrfs.h new file mode 100644 index 000000000000..22d799147db2 --- /dev/null +++ b/include/linux/btrfs.h @@ -0,0 +1,6 @@ +#ifndef _LINUX_BTRFS_H +#define _LINUX_BTRFS_H + +#include + +#endif /* _LINUX_BTRFS_H */ diff --git a/include/uapi/linux/Kbuild b/include/uapi/linux/Kbuild index 19e765fbfef7..896ee1247294 100644 --- a/include/uapi/linux/Kbuild +++ b/include/uapi/linux/Kbuild @@ -68,6 +68,7 @@ header-y += blkpg.h header-y += blktrace_api.h header-y += bpqether.h header-y += bsg.h +header-y += btrfs.h header-y += can.h header-y += capability.h header-y += capi.h diff --git a/include/uapi/linux/btrfs.h b/include/uapi/linux/btrfs.h new file mode 100644 index 000000000000..cffbb582dd90 --- /dev/null +++ b/include/uapi/linux/btrfs.h @@ -0,0 +1,503 @@ +/* + * Copyright (C) 2007 Oracle. All rights reserved. + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public + * License v2 as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + * + * You should have received a copy of the GNU General Public + * License along with this program; if not, write to the + * Free Software Foundation, Inc., 59 Temple Place - Suite 330, + * Boston, MA 021110-1307, USA. + */ + +#ifndef _UAPI_LINUX_BTRFS_H +#define _UAPI_LINUX_BTRFS_H +#include +#include + +#define BTRFS_IOCTL_MAGIC 0x94 +#define BTRFS_VOL_NAME_MAX 255 + +/* this should be 4k */ +#define BTRFS_PATH_NAME_MAX 4087 +struct btrfs_ioctl_vol_args { + __s64 fd; + char name[BTRFS_PATH_NAME_MAX + 1]; +}; + +#define BTRFS_DEVICE_PATH_NAME_MAX 1024 + +#define BTRFS_SUBVOL_CREATE_ASYNC (1ULL << 0) +#define BTRFS_SUBVOL_RDONLY (1ULL << 1) +#define BTRFS_SUBVOL_QGROUP_INHERIT (1ULL << 2) +#define BTRFS_FSID_SIZE 16 +#define BTRFS_UUID_SIZE 16 + +#define BTRFS_QGROUP_INHERIT_SET_LIMITS (1ULL << 0) + +struct btrfs_qgroup_limit { + __u64 flags; + __u64 max_rfer; + __u64 max_excl; + __u64 rsv_rfer; + __u64 rsv_excl; +}; + +struct btrfs_qgroup_inherit { + __u64 flags; + __u64 num_qgroups; + __u64 num_ref_copies; + __u64 num_excl_copies; + struct btrfs_qgroup_limit lim; + __u64 qgroups[0]; +}; + +struct btrfs_ioctl_qgroup_limit_args { + __u64 qgroupid; + struct btrfs_qgroup_limit lim; +}; + +#define BTRFS_SUBVOL_NAME_MAX 4039 +struct btrfs_ioctl_vol_args_v2 { + __s64 fd; + __u64 transid; + __u64 flags; + union { + struct { + __u64 size; + struct btrfs_qgroup_inherit __user *qgroup_inherit; + }; + __u64 unused[4]; + }; + char name[BTRFS_SUBVOL_NAME_MAX + 1]; +}; + +/* + * structure to report errors and progress to userspace, either as a + * result of a finished scrub, a canceled scrub or a progress inquiry + */ +struct btrfs_scrub_progress { + __u64 data_extents_scrubbed; /* # of data extents scrubbed */ + __u64 tree_extents_scrubbed; /* # of tree extents scrubbed */ + __u64 data_bytes_scrubbed; /* # of data bytes scrubbed */ + __u64 tree_bytes_scrubbed; /* # of tree bytes scrubbed */ + __u64 read_errors; /* # of read errors encountered (EIO) */ + __u64 csum_errors; /* # of failed csum checks */ + __u64 verify_errors; /* # of occurences, where the metadata + * of a tree block did not match the + * expected values, like generation or + * logical */ + __u64 no_csum; /* # of 4k data block for which no csum + * is present, probably the result of + * data written with nodatasum */ + __u64 csum_discards; /* # of csum for which no data was found + * in the extent tree. */ + __u64 super_errors; /* # of bad super blocks encountered */ + __u64 malloc_errors; /* # of internal kmalloc errors. These + * will likely cause an incomplete + * scrub */ + __u64 uncorrectable_errors; /* # of errors where either no intact + * copy was found or the writeback + * failed */ + __u64 corrected_errors; /* # of errors corrected */ + __u64 last_physical; /* last physical address scrubbed. In + * case a scrub was aborted, this can + * be used to restart the scrub */ + __u64 unverified_errors; /* # of occurences where a read for a + * full (64k) bio failed, but the re- + * check succeeded for each 4k piece. + * Intermittent error. */ +}; + +#define BTRFS_SCRUB_READONLY 1 +struct btrfs_ioctl_scrub_args { + __u64 devid; /* in */ + __u64 start; /* in */ + __u64 end; /* in */ + __u64 flags; /* in */ + struct btrfs_scrub_progress progress; /* out */ + /* pad to 1k */ + __u64 unused[(1024-32-sizeof(struct btrfs_scrub_progress))/8]; +}; + +#define BTRFS_IOCTL_DEV_REPLACE_CONT_READING_FROM_SRCDEV_MODE_ALWAYS 0 +#define BTRFS_IOCTL_DEV_REPLACE_CONT_READING_FROM_SRCDEV_MODE_AVOID 1 +struct btrfs_ioctl_dev_replace_start_params { + __u64 srcdevid; /* in, if 0, use srcdev_name instead */ + __u64 cont_reading_from_srcdev_mode; /* in, see #define + * above */ + __u8 srcdev_name[BTRFS_DEVICE_PATH_NAME_MAX + 1]; /* in */ + __u8 tgtdev_name[BTRFS_DEVICE_PATH_NAME_MAX + 1]; /* in */ +}; + +#define BTRFS_IOCTL_DEV_REPLACE_STATE_NEVER_STARTED 0 +#define BTRFS_IOCTL_DEV_REPLACE_STATE_STARTED 1 +#define BTRFS_IOCTL_DEV_REPLACE_STATE_FINISHED 2 +#define BTRFS_IOCTL_DEV_REPLACE_STATE_CANCELED 3 +#define BTRFS_IOCTL_DEV_REPLACE_STATE_SUSPENDED 4 +struct btrfs_ioctl_dev_replace_status_params { + __u64 replace_state; /* out, see #define above */ + __u64 progress_1000; /* out, 0 <= x <= 1000 */ + __u64 time_started; /* out, seconds since 1-Jan-1970 */ + __u64 time_stopped; /* out, seconds since 1-Jan-1970 */ + __u64 num_write_errors; /* out */ + __u64 num_uncorrectable_read_errors; /* out */ +}; + +#define BTRFS_IOCTL_DEV_REPLACE_CMD_START 0 +#define BTRFS_IOCTL_DEV_REPLACE_CMD_STATUS 1 +#define BTRFS_IOCTL_DEV_REPLACE_CMD_CANCEL 2 +#define BTRFS_IOCTL_DEV_REPLACE_RESULT_NO_ERROR 0 +#define BTRFS_IOCTL_DEV_REPLACE_RESULT_NOT_STARTED 1 +#define BTRFS_IOCTL_DEV_REPLACE_RESULT_ALREADY_STARTED 2 +struct btrfs_ioctl_dev_replace_args { + __u64 cmd; /* in */ + __u64 result; /* out */ + + union { + struct btrfs_ioctl_dev_replace_start_params start; + struct btrfs_ioctl_dev_replace_status_params status; + }; /* in/out */ + + __u64 spare[64]; +}; + +struct btrfs_ioctl_dev_info_args { + __u64 devid; /* in/out */ + __u8 uuid[BTRFS_UUID_SIZE]; /* in/out */ + __u64 bytes_used; /* out */ + __u64 total_bytes; /* out */ + __u64 unused[379]; /* pad to 4k */ + __u8 path[BTRFS_DEVICE_PATH_NAME_MAX]; /* out */ +}; + +struct btrfs_ioctl_fs_info_args { + __u64 max_id; /* out */ + __u64 num_devices; /* out */ + __u8 fsid[BTRFS_FSID_SIZE]; /* out */ + __u64 reserved[124]; /* pad to 1k */ +}; + +/* balance control ioctl modes */ +#define BTRFS_BALANCE_CTL_PAUSE 1 +#define BTRFS_BALANCE_CTL_CANCEL 2 + +/* + * this is packed, because it should be exactly the same as its disk + * byte order counterpart (struct btrfs_disk_balance_args) + */ +struct btrfs_balance_args { + __u64 profiles; + __u64 usage; + __u64 devid; + __u64 pstart; + __u64 pend; + __u64 vstart; + __u64 vend; + + __u64 target; + + __u64 flags; + + __u64 unused[8]; +} __attribute__ ((__packed__)); + +/* report balance progress to userspace */ +struct btrfs_balance_progress { + __u64 expected; /* estimated # of chunks that will be + * relocated to fulfill the request */ + __u64 considered; /* # of chunks we have considered so far */ + __u64 completed; /* # of chunks relocated so far */ +}; + +#define BTRFS_BALANCE_STATE_RUNNING (1ULL << 0) +#define BTRFS_BALANCE_STATE_PAUSE_REQ (1ULL << 1) +#define BTRFS_BALANCE_STATE_CANCEL_REQ (1ULL << 2) + +struct btrfs_ioctl_balance_args { + __u64 flags; /* in/out */ + __u64 state; /* out */ + + struct btrfs_balance_args data; /* in/out */ + struct btrfs_balance_args meta; /* in/out */ + struct btrfs_balance_args sys; /* in/out */ + + struct btrfs_balance_progress stat; /* out */ + + __u64 unused[72]; /* pad to 1k */ +}; + +#define BTRFS_INO_LOOKUP_PATH_MAX 4080 +struct btrfs_ioctl_ino_lookup_args { + __u64 treeid; + __u64 objectid; + char name[BTRFS_INO_LOOKUP_PATH_MAX]; +}; + +struct btrfs_ioctl_search_key { + /* which root are we searching. 0 is the tree of tree roots */ + __u64 tree_id; + + /* keys returned will be >= min and <= max */ + __u64 min_objectid; + __u64 max_objectid; + + /* keys returned will be >= min and <= max */ + __u64 min_offset; + __u64 max_offset; + + /* max and min transids to search for */ + __u64 min_transid; + __u64 max_transid; + + /* keys returned will be >= min and <= max */ + __u32 min_type; + __u32 max_type; + + /* + * how many items did userland ask for, and how many are we + * returning + */ + __u32 nr_items; + + /* align to 64 bits */ + __u32 unused; + + /* some extra for later */ + __u64 unused1; + __u64 unused2; + __u64 unused3; + __u64 unused4; +}; + +struct btrfs_ioctl_search_header { + __u64 transid; + __u64 objectid; + __u64 offset; + __u32 type; + __u32 len; +}; + +#define BTRFS_SEARCH_ARGS_BUFSIZE (4096 - sizeof(struct btrfs_ioctl_search_key)) +/* + * the buf is an array of search headers where + * each header is followed by the actual item + * the type field is expanded to 32 bits for alignment + */ +struct btrfs_ioctl_search_args { + struct btrfs_ioctl_search_key key; + char buf[BTRFS_SEARCH_ARGS_BUFSIZE]; +}; + +struct btrfs_ioctl_clone_range_args { + __s64 src_fd; + __u64 src_offset, src_length; + __u64 dest_offset; +}; + +/* flags for the defrag range ioctl */ +#define BTRFS_DEFRAG_RANGE_COMPRESS 1 +#define BTRFS_DEFRAG_RANGE_START_IO 2 + +struct btrfs_ioctl_space_info { + __u64 flags; + __u64 total_bytes; + __u64 used_bytes; +}; + +struct btrfs_ioctl_space_args { + __u64 space_slots; + __u64 total_spaces; + struct btrfs_ioctl_space_info spaces[0]; +}; + +struct btrfs_data_container { + __u32 bytes_left; /* out -- bytes not needed to deliver output */ + __u32 bytes_missing; /* out -- additional bytes needed for result */ + __u32 elem_cnt; /* out */ + __u32 elem_missed; /* out */ + __u64 val[0]; /* out */ +}; + +struct btrfs_ioctl_ino_path_args { + __u64 inum; /* in */ + __u64 size; /* in */ + __u64 reserved[4]; + /* struct btrfs_data_container *fspath; out */ + __u64 fspath; /* out */ +}; + +struct btrfs_ioctl_logical_ino_args { + __u64 logical; /* in */ + __u64 size; /* in */ + __u64 reserved[4]; + /* struct btrfs_data_container *inodes; out */ + __u64 inodes; +}; + +enum btrfs_dev_stat_values { + /* disk I/O failure stats */ + BTRFS_DEV_STAT_WRITE_ERRS, /* EIO or EREMOTEIO from lower layers */ + BTRFS_DEV_STAT_READ_ERRS, /* EIO or EREMOTEIO from lower layers */ + BTRFS_DEV_STAT_FLUSH_ERRS, /* EIO or EREMOTEIO from lower layers */ + + /* stats for indirect indications for I/O failures */ + BTRFS_DEV_STAT_CORRUPTION_ERRS, /* checksum error, bytenr error or + * contents is illegal: this is an + * indication that the block was damaged + * during read or write, or written to + * wrong location or read from wrong + * location */ + BTRFS_DEV_STAT_GENERATION_ERRS, /* an indication that blocks have not + * been written */ + + BTRFS_DEV_STAT_VALUES_MAX +}; + +/* Reset statistics after reading; needs SYS_ADMIN capability */ +#define BTRFS_DEV_STATS_RESET (1ULL << 0) + +struct btrfs_ioctl_get_dev_stats { + __u64 devid; /* in */ + __u64 nr_items; /* in/out */ + __u64 flags; /* in/out */ + + /* out values: */ + __u64 values[BTRFS_DEV_STAT_VALUES_MAX]; + + __u64 unused[128 - 2 - BTRFS_DEV_STAT_VALUES_MAX]; /* pad to 1k */ +}; + +#define BTRFS_QUOTA_CTL_ENABLE 1 +#define BTRFS_QUOTA_CTL_DISABLE 2 +#define BTRFS_QUOTA_CTL_RESCAN 3 +struct btrfs_ioctl_quota_ctl_args { + __u64 cmd; + __u64 status; +}; + +struct btrfs_ioctl_qgroup_assign_args { + __u64 assign; + __u64 src; + __u64 dst; +}; + +struct btrfs_ioctl_qgroup_create_args { + __u64 create; + __u64 qgroupid; +}; +struct btrfs_ioctl_timespec { + __u64 sec; + __u32 nsec; +}; + +struct btrfs_ioctl_received_subvol_args { + char uuid[BTRFS_UUID_SIZE]; /* in */ + __u64 stransid; /* in */ + __u64 rtransid; /* out */ + struct btrfs_ioctl_timespec stime; /* in */ + struct btrfs_ioctl_timespec rtime; /* out */ + __u64 flags; /* in */ + __u64 reserved[16]; /* in */ +}; + +struct btrfs_ioctl_send_args { + __s64 send_fd; /* in */ + __u64 clone_sources_count; /* in */ + __u64 __user *clone_sources; /* in */ + __u64 parent_root; /* in */ + __u64 flags; /* in */ + __u64 reserved[4]; /* in */ +}; + +#define BTRFS_IOC_SNAP_CREATE _IOW(BTRFS_IOCTL_MAGIC, 1, \ + struct btrfs_ioctl_vol_args) +#define BTRFS_IOC_DEFRAG _IOW(BTRFS_IOCTL_MAGIC, 2, \ + struct btrfs_ioctl_vol_args) +#define BTRFS_IOC_RESIZE _IOW(BTRFS_IOCTL_MAGIC, 3, \ + struct btrfs_ioctl_vol_args) +#define BTRFS_IOC_SCAN_DEV _IOW(BTRFS_IOCTL_MAGIC, 4, \ + struct btrfs_ioctl_vol_args) +/* trans start and trans end are dangerous, and only for + * use by applications that know how to avoid the + * resulting deadlocks + */ +#define BTRFS_IOC_TRANS_START _IO(BTRFS_IOCTL_MAGIC, 6) +#define BTRFS_IOC_TRANS_END _IO(BTRFS_IOCTL_MAGIC, 7) +#define BTRFS_IOC_SYNC _IO(BTRFS_IOCTL_MAGIC, 8) + +#define BTRFS_IOC_CLONE _IOW(BTRFS_IOCTL_MAGIC, 9, int) +#define BTRFS_IOC_ADD_DEV _IOW(BTRFS_IOCTL_MAGIC, 10, \ + struct btrfs_ioctl_vol_args) +#define BTRFS_IOC_RM_DEV _IOW(BTRFS_IOCTL_MAGIC, 11, \ + struct btrfs_ioctl_vol_args) +#define BTRFS_IOC_BALANCE _IOW(BTRFS_IOCTL_MAGIC, 12, \ + struct btrfs_ioctl_vol_args) + +#define BTRFS_IOC_CLONE_RANGE _IOW(BTRFS_IOCTL_MAGIC, 13, \ + struct btrfs_ioctl_clone_range_args) + +#define BTRFS_IOC_SUBVOL_CREATE _IOW(BTRFS_IOCTL_MAGIC, 14, \ + struct btrfs_ioctl_vol_args) +#define BTRFS_IOC_SNAP_DESTROY _IOW(BTRFS_IOCTL_MAGIC, 15, \ + struct btrfs_ioctl_vol_args) +#define BTRFS_IOC_DEFRAG_RANGE _IOW(BTRFS_IOCTL_MAGIC, 16, \ + struct btrfs_ioctl_defrag_range_args) +#define BTRFS_IOC_TREE_SEARCH _IOWR(BTRFS_IOCTL_MAGIC, 17, \ + struct btrfs_ioctl_search_args) +#define BTRFS_IOC_INO_LOOKUP _IOWR(BTRFS_IOCTL_MAGIC, 18, \ + struct btrfs_ioctl_ino_lookup_args) +#define BTRFS_IOC_DEFAULT_SUBVOL _IOW(BTRFS_IOCTL_MAGIC, 19, u64) +#define BTRFS_IOC_SPACE_INFO _IOWR(BTRFS_IOCTL_MAGIC, 20, \ + struct btrfs_ioctl_space_args) +#define BTRFS_IOC_START_SYNC _IOR(BTRFS_IOCTL_MAGIC, 24, __u64) +#define BTRFS_IOC_WAIT_SYNC _IOW(BTRFS_IOCTL_MAGIC, 22, __u64) +#define BTRFS_IOC_SNAP_CREATE_V2 _IOW(BTRFS_IOCTL_MAGIC, 23, \ + struct btrfs_ioctl_vol_args_v2) +#define BTRFS_IOC_SUBVOL_CREATE_V2 _IOW(BTRFS_IOCTL_MAGIC, 24, \ + struct btrfs_ioctl_vol_args_v2) +#define BTRFS_IOC_SUBVOL_GETFLAGS _IOR(BTRFS_IOCTL_MAGIC, 25, __u64) +#define BTRFS_IOC_SUBVOL_SETFLAGS _IOW(BTRFS_IOCTL_MAGIC, 26, __u64) +#define BTRFS_IOC_SCRUB _IOWR(BTRFS_IOCTL_MAGIC, 27, \ + struct btrfs_ioctl_scrub_args) +#define BTRFS_IOC_SCRUB_CANCEL _IO(BTRFS_IOCTL_MAGIC, 28) +#define BTRFS_IOC_SCRUB_PROGRESS _IOWR(BTRFS_IOCTL_MAGIC, 29, \ + struct btrfs_ioctl_scrub_args) +#define BTRFS_IOC_DEV_INFO _IOWR(BTRFS_IOCTL_MAGIC, 30, \ + struct btrfs_ioctl_dev_info_args) +#define BTRFS_IOC_FS_INFO _IOR(BTRFS_IOCTL_MAGIC, 31, \ + struct btrfs_ioctl_fs_info_args) +#define BTRFS_IOC_BALANCE_V2 _IOWR(BTRFS_IOCTL_MAGIC, 32, \ + struct btrfs_ioctl_balance_args) +#define BTRFS_IOC_BALANCE_CTL _IOW(BTRFS_IOCTL_MAGIC, 33, int) +#define BTRFS_IOC_BALANCE_PROGRESS _IOR(BTRFS_IOCTL_MAGIC, 34, \ + struct btrfs_ioctl_balance_args) +#define BTRFS_IOC_INO_PATHS _IOWR(BTRFS_IOCTL_MAGIC, 35, \ + struct btrfs_ioctl_ino_path_args) +#define BTRFS_IOC_LOGICAL_INO _IOWR(BTRFS_IOCTL_MAGIC, 36, \ + struct btrfs_ioctl_ino_path_args) +#define BTRFS_IOC_SET_RECEIVED_SUBVOL _IOWR(BTRFS_IOCTL_MAGIC, 37, \ + struct btrfs_ioctl_received_subvol_args) +#define BTRFS_IOC_SEND _IOW(BTRFS_IOCTL_MAGIC, 38, struct btrfs_ioctl_send_args) +#define BTRFS_IOC_DEVICES_READY _IOR(BTRFS_IOCTL_MAGIC, 39, \ + struct btrfs_ioctl_vol_args) +#define BTRFS_IOC_QUOTA_CTL _IOWR(BTRFS_IOCTL_MAGIC, 40, \ + struct btrfs_ioctl_quota_ctl_args) +#define BTRFS_IOC_QGROUP_ASSIGN _IOW(BTRFS_IOCTL_MAGIC, 41, \ + struct btrfs_ioctl_qgroup_assign_args) +#define BTRFS_IOC_QGROUP_CREATE _IOW(BTRFS_IOCTL_MAGIC, 42, \ + struct btrfs_ioctl_qgroup_create_args) +#define BTRFS_IOC_QGROUP_LIMIT _IOR(BTRFS_IOCTL_MAGIC, 43, \ + struct btrfs_ioctl_qgroup_limit_args) +#define BTRFS_IOC_GET_DEV_STATS _IOWR(BTRFS_IOCTL_MAGIC, 52, \ + struct btrfs_ioctl_get_dev_stats) +#define BTRFS_IOC_DEV_REPLACE _IOWR(BTRFS_IOCTL_MAGIC, 53, \ + struct btrfs_ioctl_dev_replace_args) + +#endif /* _UAPI_LINUX_BTRFS_H */ -- cgit v1.2.3 From 8c6a3ee6dbd564d05019d396ecb5dc5b35cbc273 Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Tue, 29 Jan 2013 10:05:05 +0000 Subject: Btrfs: add a comment for fs_info->max_inline Though ->max_inline is a 64bit variant, and may be accessed by multi-task, but it is just suggestive number, so we needn't add anything to protect fs_info->max_inline, just add a comment to explain wny we don't use a lock to protect it. Signed-off-by: Miao Xie Signed-off-by: Josef Bacik --- fs/btrfs/ctree.h | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'fs/btrfs/ctree.h') diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 69321013683c..169baa817d96 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -1297,6 +1297,12 @@ struct btrfs_fs_info { u64 last_trans_log_full_commit; unsigned long mount_opt; unsigned long compress_type:4; + /* + * It is a suggestive number, the read side is safe even it gets a + * wrong number because we will write out the data into a regular + * extent. The write side(mount/remount) is under ->s_umount lock, + * so it is also safe. + */ u64 max_inline; u64 alloc_start; struct btrfs_transaction *running_transaction; -- cgit v1.2.3 From c018daecead7a46a575e2a1397fea850b83396c8 Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Tue, 29 Jan 2013 10:07:33 +0000 Subject: Btrfs: protect fs_info->alloc_start fs_info->alloc_start is a 64bits variant, can be accessed by multi-task, but it is not protected strictly, it can be changed while we are accessing it. On 32bit machine, we will get wrong value because we access it by two instructions.(In fact, it is also possible that the same problem happens on the 64bit machine, because the compiler may split the 64bit operation into two 32bit operation.) For example: Assuming -> alloc_start is 0x0000 0000 0001 0000 at the beginning, then we remount and set ->alloc_start to 0x0000 0100 0000 0000. Task0 Task1 load high 32 bits set high 32 bits set low 32 bits load low 32 bits Task1 will get 0. This patch fixes this problem by using two locks to protect it fs_info->chunk_mutex sb->s_umount On the read side, we just need get one of these two locks, and on the write side, we must lock all of them. Signed-off-by: Miao Xie Signed-off-by: Josef Bacik --- fs/btrfs/ctree.h | 10 ++++++++++ fs/btrfs/super.c | 4 ++++ 2 files changed, 14 insertions(+) (limited to 'fs/btrfs/ctree.h') diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 169baa817d96..d8e539fe5544 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -1304,6 +1304,16 @@ struct btrfs_fs_info { * so it is also safe. */ u64 max_inline; + /* + * Protected by ->chunk_mutex and sb->s_umount. + * + * The reason that we use two lock to protect it is because only + * remount and mount operations can change it and these two operations + * are under sb->s_umount, but the read side (chunk allocation) can not + * acquire sb->s_umount or the deadlock would happen. So we use two + * locks to protect it. On the write side, we must acquire two locks, + * and on the read side, we just need acquire one of them. + */ u64 alloc_start; struct btrfs_transaction *running_transaction; wait_queue_head_t transaction_throttle; diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c index 6846ededfe95..eda330df45a4 100644 --- a/fs/btrfs/super.c +++ b/fs/btrfs/super.c @@ -519,7 +519,9 @@ int btrfs_parse_options(struct btrfs_root *root, char *options) case Opt_alloc_start: num = match_strdup(&args[0]); if (num) { + mutex_lock(&info->chunk_mutex); info->alloc_start = memparse(num, NULL); + mutex_unlock(&info->chunk_mutex); kfree(num); printk(KERN_INFO "btrfs: allocations start at %llu\n", @@ -1289,7 +1291,9 @@ restore: fs_info->mount_opt = old_opts; fs_info->compress_type = old_compress_type; fs_info->max_inline = old_max_inline; + mutex_lock(&fs_info->chunk_mutex); fs_info->alloc_start = old_alloc_start; + mutex_unlock(&fs_info->chunk_mutex); btrfs_resize_thread_pool(fs_info, old_thread_pool_size, fs_info->thread_pool_size); fs_info->metadata_ratio = old_metadata_ratio; -- cgit v1.2.3 From e2d845211eda9cf296e8edf6724b3d541f4fbfd5 Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Tue, 29 Jan 2013 10:09:20 +0000 Subject: Btrfs: use percpu counter for dirty metadata count ->dirty_metadata_bytes is accessed very frequently, so use percpu counter instead of the u64 variant to reduce the contention of the lock. This patch also fixed the problem that we access it without lock protection in __btrfs_btree_balance_dirty(), which may cause we skip the dirty pages flush. Signed-off-by: Miao Xie Signed-off-by: Josef Bacik --- fs/btrfs/ctree.h | 9 ++++---- fs/btrfs/disk-io.c | 64 ++++++++++++++++++++++++++++------------------------ fs/btrfs/extent_io.c | 9 +++----- 3 files changed, 42 insertions(+), 40 deletions(-) (limited to 'fs/btrfs/ctree.h') diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index d8e539fe5544..4c476281b66b 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -191,6 +191,8 @@ static int btrfs_csum_sizes[] = { 4, 0 }; /* ioprio of readahead is set to idle */ #define BTRFS_IOPRIO_READA (IOPRIO_PRIO_VALUE(IOPRIO_CLASS_IDLE, 0)) +#define BTRFS_DIRTY_METADATA_THRESH (32 * 1024 * 1024) + /* * The key defines the order in the tree, and so it also defines (optimal) * block layout. @@ -1448,10 +1450,9 @@ struct btrfs_fs_info { u64 total_pinned; - /* protected by the delalloc lock, used to keep from writing - * metadata until there is a nice batch - */ - u64 dirty_metadata_bytes; + /* used to keep from writing metadata until there is a nice batch */ + struct percpu_counter dirty_metadata_bytes; + s32 dirty_metadata_batch; struct list_head dirty_cowonly_roots; struct btrfs_fs_devices *fs_devices; diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 04f98e3ffd90..34ace168eebc 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -946,18 +946,20 @@ static int btree_writepages(struct address_space *mapping, struct writeback_control *wbc) { struct extent_io_tree *tree; + struct btrfs_fs_info *fs_info; + int ret; + tree = &BTRFS_I(mapping->host)->io_tree; if (wbc->sync_mode == WB_SYNC_NONE) { - struct btrfs_root *root = BTRFS_I(mapping->host)->root; - u64 num_dirty; - unsigned long thresh = 32 * 1024 * 1024; if (wbc->for_kupdate) return 0; + fs_info = BTRFS_I(mapping->host)->root->fs_info; /* this is a bit racy, but that's ok */ - num_dirty = root->fs_info->dirty_metadata_bytes; - if (num_dirty < thresh) + ret = percpu_counter_compare(&fs_info->dirty_metadata_bytes, + BTRFS_DIRTY_METADATA_THRESH); + if (ret < 0) return 0; } return btree_write_cache_pages(mapping, wbc); @@ -1125,24 +1127,16 @@ struct extent_buffer *read_tree_block(struct btrfs_root *root, u64 bytenr, void clean_tree_block(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct extent_buffer *buf) { + struct btrfs_fs_info *fs_info = root->fs_info; + if (btrfs_header_generation(buf) == - root->fs_info->running_transaction->transid) { + fs_info->running_transaction->transid) { btrfs_assert_tree_locked(buf); if (test_and_clear_bit(EXTENT_BUFFER_DIRTY, &buf->bflags)) { - spin_lock(&root->fs_info->delalloc_lock); - if (root->fs_info->dirty_metadata_bytes >= buf->len) - root->fs_info->dirty_metadata_bytes -= buf->len; - else { - spin_unlock(&root->fs_info->delalloc_lock); - btrfs_panic(root->fs_info, -EOVERFLOW, - "Can't clear %lu bytes from " - " dirty_mdatadata_bytes (%llu)", - buf->len, - root->fs_info->dirty_metadata_bytes); - } - spin_unlock(&root->fs_info->delalloc_lock); - + __percpu_counter_add(&fs_info->dirty_metadata_bytes, + -buf->len, + fs_info->dirty_metadata_batch); /* ugh, clear_extent_buffer_dirty needs to lock the page */ btrfs_set_lock_blocking(buf); clear_extent_buffer_dirty(buf); @@ -2008,10 +2002,18 @@ int open_ctree(struct super_block *sb, goto fail_srcu; } + ret = percpu_counter_init(&fs_info->dirty_metadata_bytes, 0); + if (ret) { + err = ret; + goto fail_bdi; + } + fs_info->dirty_metadata_batch = PAGE_CACHE_SIZE * + (1 + ilog2(nr_cpu_ids)); + fs_info->btree_inode = new_inode(sb); if (!fs_info->btree_inode) { err = -ENOMEM; - goto fail_bdi; + goto fail_dirty_metadata_bytes; } mapping_set_gfp_mask(fs_info->btree_inode->i_mapping, GFP_NOFS); @@ -2266,6 +2268,7 @@ int open_ctree(struct super_block *sb, leafsize = btrfs_super_leafsize(disk_super); sectorsize = btrfs_super_sectorsize(disk_super); stripesize = btrfs_super_stripesize(disk_super); + fs_info->dirty_metadata_batch = leafsize * (1 + ilog2(nr_cpu_ids)); /* * mixed block groups end up with duplicate but slightly offset @@ -2728,6 +2731,8 @@ fail_iput: invalidate_inode_pages2(fs_info->btree_inode->i_mapping); iput(fs_info->btree_inode); +fail_dirty_metadata_bytes: + percpu_counter_destroy(&fs_info->dirty_metadata_bytes); fail_bdi: bdi_destroy(&fs_info->bdi); fail_srcu: @@ -3406,6 +3411,7 @@ int close_ctree(struct btrfs_root *root) btrfs_close_devices(fs_info->fs_devices); btrfs_mapping_tree_free(&fs_info->mapping_tree); + percpu_counter_destroy(&fs_info->dirty_metadata_bytes); bdi_destroy(&fs_info->bdi); cleanup_srcu_struct(&fs_info->subvol_srcu); @@ -3448,11 +3454,10 @@ void btrfs_mark_buffer_dirty(struct extent_buffer *buf) (unsigned long long)transid, (unsigned long long)root->fs_info->generation); was_dirty = set_extent_buffer_dirty(buf); - if (!was_dirty) { - spin_lock(&root->fs_info->delalloc_lock); - root->fs_info->dirty_metadata_bytes += buf->len; - spin_unlock(&root->fs_info->delalloc_lock); - } + if (!was_dirty) + __percpu_counter_add(&root->fs_info->dirty_metadata_bytes, + buf->len, + root->fs_info->dirty_metadata_batch); } static void __btrfs_btree_balance_dirty(struct btrfs_root *root, @@ -3462,8 +3467,7 @@ static void __btrfs_btree_balance_dirty(struct btrfs_root *root, * looks as though older kernels can get into trouble with * this code, they end up stuck in balance_dirty_pages forever */ - u64 num_dirty; - unsigned long thresh = 32 * 1024 * 1024; + int ret; if (current->flags & PF_MEMALLOC) return; @@ -3471,9 +3475,9 @@ static void __btrfs_btree_balance_dirty(struct btrfs_root *root, if (flush_delayed) btrfs_balance_delayed_items(root); - num_dirty = root->fs_info->dirty_metadata_bytes; - - if (num_dirty > thresh) { + ret = percpu_counter_compare(&root->fs_info->dirty_metadata_bytes, + BTRFS_DIRTY_METADATA_THRESH); + if (ret > 0) { balance_dirty_pages_ratelimited_nr( root->fs_info->btree_inode->i_mapping, 1); } diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c index bda36fef301e..b9019c847c58 100644 --- a/fs/btrfs/extent_io.c +++ b/fs/btrfs/extent_io.c @@ -3122,12 +3122,9 @@ static int lock_extent_buffer_for_io(struct extent_buffer *eb, set_bit(EXTENT_BUFFER_WRITEBACK, &eb->bflags); spin_unlock(&eb->refs_lock); btrfs_set_header_flag(eb, BTRFS_HEADER_FLAG_WRITTEN); - spin_lock(&fs_info->delalloc_lock); - if (fs_info->dirty_metadata_bytes >= eb->len) - fs_info->dirty_metadata_bytes -= eb->len; - else - WARN_ON(1); - spin_unlock(&fs_info->delalloc_lock); + __percpu_counter_add(&fs_info->dirty_metadata_bytes, + -eb->len, + fs_info->dirty_metadata_batch); ret = 1; } else { spin_unlock(&eb->refs_lock); -- cgit v1.2.3 From 963d678b0f7649300e3a67f2513ca9d830c6e303 Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Tue, 29 Jan 2013 10:10:51 +0000 Subject: Btrfs: use percpu counter for fs_info->delalloc_bytes fs_info->delalloc_bytes is accessed very frequently, so use percpu counter instead of the u64 variant for it to reduce the lock contention. This patch also fixed the problem that we access the variant without the lock protection.At worst, we would not flush the delalloc inodes, and just return ENOSPC error when we still have some free space in the fs. Signed-off-by: Miao Xie Signed-off-by: Josef Bacik --- fs/btrfs/ctree.h | 7 ++++--- fs/btrfs/disk-io.c | 18 ++++++++++++++---- fs/btrfs/extent-tree.c | 6 ++++-- fs/btrfs/inode.c | 6 ++++-- 4 files changed, 26 insertions(+), 11 deletions(-) (limited to 'fs/btrfs/ctree.h') diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 4c476281b66b..50def99f5379 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -1392,6 +1392,7 @@ struct btrfs_fs_info { */ struct list_head ordered_extents; + spinlock_t delalloc_lock; /* * all of the inodes that have delalloc bytes. It is possible for * this list to be empty even when there is still dirty data=ordered @@ -1452,7 +1453,10 @@ struct btrfs_fs_info { /* used to keep from writing metadata until there is a nice batch */ struct percpu_counter dirty_metadata_bytes; + struct percpu_counter delalloc_bytes; s32 dirty_metadata_batch; + s32 delalloc_batch; + struct list_head dirty_cowonly_roots; struct btrfs_fs_devices *fs_devices; @@ -1468,9 +1472,6 @@ struct btrfs_fs_info { struct reloc_control *reloc_ctl; - spinlock_t delalloc_lock; - u64 delalloc_bytes; - /* data_alloc_cluster is only used in ssd mode */ struct btrfs_free_cluster data_alloc_cluster; diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 34ace168eebc..2c9498aefe86 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -2010,10 +2010,16 @@ int open_ctree(struct super_block *sb, fs_info->dirty_metadata_batch = PAGE_CACHE_SIZE * (1 + ilog2(nr_cpu_ids)); + ret = percpu_counter_init(&fs_info->delalloc_bytes, 0); + if (ret) { + err = ret; + goto fail_dirty_metadata_bytes; + } + fs_info->btree_inode = new_inode(sb); if (!fs_info->btree_inode) { err = -ENOMEM; - goto fail_dirty_metadata_bytes; + goto fail_delalloc_bytes; } mapping_set_gfp_mask(fs_info->btree_inode->i_mapping, GFP_NOFS); @@ -2269,6 +2275,7 @@ int open_ctree(struct super_block *sb, sectorsize = btrfs_super_sectorsize(disk_super); stripesize = btrfs_super_stripesize(disk_super); fs_info->dirty_metadata_batch = leafsize * (1 + ilog2(nr_cpu_ids)); + fs_info->delalloc_batch = sectorsize * 512 * (1 + ilog2(nr_cpu_ids)); /* * mixed block groups end up with duplicate but slightly offset @@ -2731,6 +2738,8 @@ fail_iput: invalidate_inode_pages2(fs_info->btree_inode->i_mapping); iput(fs_info->btree_inode); +fail_delalloc_bytes: + percpu_counter_destroy(&fs_info->delalloc_bytes); fail_dirty_metadata_bytes: percpu_counter_destroy(&fs_info->dirty_metadata_bytes); fail_bdi: @@ -3362,9 +3371,9 @@ int close_ctree(struct btrfs_root *root) btrfs_free_qgroup_config(root->fs_info); - if (fs_info->delalloc_bytes) { - printk(KERN_INFO "btrfs: at unmount delalloc count %llu\n", - (unsigned long long)fs_info->delalloc_bytes); + if (percpu_counter_sum(&fs_info->delalloc_bytes)) { + printk(KERN_INFO "btrfs: at unmount delalloc count %lld\n", + percpu_counter_sum(&fs_info->delalloc_bytes)); } free_extent_buffer(fs_info->extent_root->node); @@ -3412,6 +3421,7 @@ int close_ctree(struct btrfs_root *root) btrfs_mapping_tree_free(&fs_info->mapping_tree); percpu_counter_destroy(&fs_info->dirty_metadata_bytes); + percpu_counter_destroy(&fs_info->delalloc_bytes); bdi_destroy(&fs_info->bdi); cleanup_srcu_struct(&fs_info->subvol_srcu); diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 174c4d5c692c..115d1646bf50 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3760,7 +3760,8 @@ static void shrink_delalloc(struct btrfs_root *root, u64 to_reclaim, u64 orig, space_info = block_rsv->space_info; smp_mb(); - delalloc_bytes = root->fs_info->delalloc_bytes; + delalloc_bytes = percpu_counter_sum_positive( + &root->fs_info->delalloc_bytes); if (delalloc_bytes == 0) { if (trans) return; @@ -3799,7 +3800,8 @@ static void shrink_delalloc(struct btrfs_root *root, u64 to_reclaim, u64 orig, break; } smp_mb(); - delalloc_bytes = root->fs_info->delalloc_bytes; + delalloc_bytes = percpu_counter_sum_positive( + &root->fs_info->delalloc_bytes); } } diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index fc8aa8bf80a1..24c0b7805fe1 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -1516,7 +1516,8 @@ static void btrfs_set_bit_hook(struct inode *inode, spin_lock(&root->fs_info->delalloc_lock); BTRFS_I(inode)->delalloc_bytes += len; - root->fs_info->delalloc_bytes += len; + __percpu_counter_add(&root->fs_info->delalloc_bytes, len, + root->fs_info->delalloc_batch); if (do_list && list_empty(&BTRFS_I(inode)->delalloc_inodes)) { list_add_tail(&BTRFS_I(inode)->delalloc_inodes, &root->fs_info->delalloc_inodes); @@ -1557,7 +1558,8 @@ static void btrfs_clear_bit_hook(struct inode *inode, btrfs_free_reserved_data_space(inode, len); spin_lock(&root->fs_info->delalloc_lock); - root->fs_info->delalloc_bytes -= len; + __percpu_counter_add(&root->fs_info->delalloc_bytes, -len, + root->fs_info->delalloc_batch); BTRFS_I(inode)->delalloc_bytes -= len; if (do_list && BTRFS_I(inode)->delalloc_bytes == 0 && -- cgit v1.2.3 From de98ced9e743656d108de41841797def0f5cb951 Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Tue, 29 Jan 2013 10:13:12 +0000 Subject: Btrfs: use seqlock to protect fs_info->avail_{data, metadata, system}_alloc_bits There is no lock to protect fs_info->avail_{data, metadata, system}_alloc_bits, it may introduce some problem, such as the wrong profile information, so we add a seqlock to protect them. Signed-off-by: Zhao Lei Signed-off-by: Miao Xie Signed-off-by: Josef Bacik --- fs/btrfs/ctree.h | 2 ++ fs/btrfs/disk-io.c | 1 + fs/btrfs/extent-tree.c | 22 ++++++++++++++------ fs/btrfs/volumes.c | 56 +++++++++++++++++++++++++++----------------------- 4 files changed, 49 insertions(+), 32 deletions(-) (limited to 'fs/btrfs/ctree.h') diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 50def99f5379..5f5c30aaef36 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -1483,6 +1483,8 @@ struct btrfs_fs_info { struct rb_root defrag_inodes; atomic_t defrag_running; + /* Used to protect avail_{data, metadata, system}_alloc_bits */ + seqlock_t profiles_lock; /* * these three are in extended format (availability of single * chunks is denoted by BTRFS_AVAIL_ALLOC_BIT_SINGLE bit, other diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 11f6dbcb1191..00b6742fdde7 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -2040,6 +2040,7 @@ int open_ctree(struct super_block *sb, spin_lock_init(&fs_info->tree_mod_seq_lock); rwlock_init(&fs_info->tree_mod_log_lock); mutex_init(&fs_info->reloc_mutex); + seqlock_init(&fs_info->profiles_lock); init_completion(&fs_info->kobj_unregister); INIT_LIST_HEAD(&fs_info->dirty_cowonly_roots); diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 115d1646bf50..faff98f720de 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3227,12 +3227,14 @@ static void set_avail_alloc_bits(struct btrfs_fs_info *fs_info, u64 flags) u64 extra_flags = chunk_to_extended(flags) & BTRFS_EXTENDED_PROFILE_MASK; + write_seqlock(&fs_info->profiles_lock); if (flags & BTRFS_BLOCK_GROUP_DATA) fs_info->avail_data_alloc_bits |= extra_flags; if (flags & BTRFS_BLOCK_GROUP_METADATA) fs_info->avail_metadata_alloc_bits |= extra_flags; if (flags & BTRFS_BLOCK_GROUP_SYSTEM) fs_info->avail_system_alloc_bits |= extra_flags; + write_sequnlock(&fs_info->profiles_lock); } /* @@ -3324,12 +3326,18 @@ u64 btrfs_reduce_alloc_profile(struct btrfs_root *root, u64 flags) static u64 get_alloc_profile(struct btrfs_root *root, u64 flags) { - if (flags & BTRFS_BLOCK_GROUP_DATA) - flags |= root->fs_info->avail_data_alloc_bits; - else if (flags & BTRFS_BLOCK_GROUP_SYSTEM) - flags |= root->fs_info->avail_system_alloc_bits; - else if (flags & BTRFS_BLOCK_GROUP_METADATA) - flags |= root->fs_info->avail_metadata_alloc_bits; + unsigned seq; + + do { + seq = read_seqbegin(&root->fs_info->profiles_lock); + + if (flags & BTRFS_BLOCK_GROUP_DATA) + flags |= root->fs_info->avail_data_alloc_bits; + else if (flags & BTRFS_BLOCK_GROUP_SYSTEM) + flags |= root->fs_info->avail_system_alloc_bits; + else if (flags & BTRFS_BLOCK_GROUP_METADATA) + flags |= root->fs_info->avail_metadata_alloc_bits; + } while (read_seqretry(&root->fs_info->profiles_lock, seq)); return btrfs_reduce_alloc_profile(root, flags); } @@ -7967,12 +7975,14 @@ static void clear_avail_alloc_bits(struct btrfs_fs_info *fs_info, u64 flags) u64 extra_flags = chunk_to_extended(flags) & BTRFS_EXTENDED_PROFILE_MASK; + write_seqlock(&fs_info->profiles_lock); if (flags & BTRFS_BLOCK_GROUP_DATA) fs_info->avail_data_alloc_bits &= ~extra_flags; if (flags & BTRFS_BLOCK_GROUP_METADATA) fs_info->avail_metadata_alloc_bits &= ~extra_flags; if (flags & BTRFS_BLOCK_GROUP_SYSTEM) fs_info->avail_system_alloc_bits &= ~extra_flags; + write_sequnlock(&fs_info->profiles_lock); } int btrfs_remove_block_group(struct btrfs_trans_handle *trans, diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c index 1ce581d58d87..8c9ea4cd66bb 100644 --- a/fs/btrfs/volumes.c +++ b/fs/btrfs/volumes.c @@ -1430,14 +1430,19 @@ int btrfs_rm_device(struct btrfs_root *root, char *device_path) u64 devid; u64 num_devices; u8 *dev_uuid; + unsigned seq; int ret = 0; bool clear_super = false; mutex_lock(&uuid_mutex); - all_avail = root->fs_info->avail_data_alloc_bits | - root->fs_info->avail_system_alloc_bits | - root->fs_info->avail_metadata_alloc_bits; + do { + seq = read_seqbegin(&root->fs_info->profiles_lock); + + all_avail = root->fs_info->avail_data_alloc_bits | + root->fs_info->avail_system_alloc_bits | + root->fs_info->avail_metadata_alloc_bits; + } while (read_seqretry(&root->fs_info->profiles_lock, seq)); num_devices = root->fs_info->fs_devices->num_devices; btrfs_dev_replace_lock(&root->fs_info->dev_replace); @@ -3043,6 +3048,7 @@ int btrfs_balance(struct btrfs_balance_control *bctl, int mixed = 0; int ret; u64 num_devices; + unsigned seq; if (btrfs_fs_closing(fs_info) || atomic_read(&fs_info->balance_pause_req) || @@ -3126,22 +3132,26 @@ int btrfs_balance(struct btrfs_balance_control *bctl, /* allow to reduce meta or sys integrity only if force set */ allowed = BTRFS_BLOCK_GROUP_DUP | BTRFS_BLOCK_GROUP_RAID1 | BTRFS_BLOCK_GROUP_RAID10; - if (((bctl->sys.flags & BTRFS_BALANCE_ARGS_CONVERT) && - (fs_info->avail_system_alloc_bits & allowed) && - !(bctl->sys.target & allowed)) || - ((bctl->meta.flags & BTRFS_BALANCE_ARGS_CONVERT) && - (fs_info->avail_metadata_alloc_bits & allowed) && - !(bctl->meta.target & allowed))) { - if (bctl->flags & BTRFS_BALANCE_FORCE) { - printk(KERN_INFO "btrfs: force reducing metadata " - "integrity\n"); - } else { - printk(KERN_ERR "btrfs: balance will reduce metadata " - "integrity, use force if you want this\n"); - ret = -EINVAL; - goto out; + do { + seq = read_seqbegin(&fs_info->profiles_lock); + + if (((bctl->sys.flags & BTRFS_BALANCE_ARGS_CONVERT) && + (fs_info->avail_system_alloc_bits & allowed) && + !(bctl->sys.target & allowed)) || + ((bctl->meta.flags & BTRFS_BALANCE_ARGS_CONVERT) && + (fs_info->avail_metadata_alloc_bits & allowed) && + !(bctl->meta.target & allowed))) { + if (bctl->flags & BTRFS_BALANCE_FORCE) { + printk(KERN_INFO "btrfs: force reducing metadata " + "integrity\n"); + } else { + printk(KERN_ERR "btrfs: balance will reduce metadata " + "integrity, use force if you want this\n"); + ret = -EINVAL; + goto out; + } } - } + } while (read_seqretry(&fs_info->profiles_lock, seq)); if (bctl->sys.flags & BTRFS_BALANCE_ARGS_CONVERT) { int num_tolerated_disk_barrier_failures; @@ -3980,10 +3990,7 @@ static noinline int init_first_rw_device(struct btrfs_trans_handle *trans, if (ret) return ret; - alloc_profile = BTRFS_BLOCK_GROUP_METADATA | - fs_info->avail_metadata_alloc_bits; - alloc_profile = btrfs_reduce_alloc_profile(root, alloc_profile); - + alloc_profile = btrfs_get_alloc_profile(extent_root, 0); ret = __btrfs_alloc_chunk(trans, extent_root, &map, &chunk_size, &stripe_size, chunk_offset, alloc_profile); if (ret) @@ -3991,10 +3998,7 @@ static noinline int init_first_rw_device(struct btrfs_trans_handle *trans, sys_chunk_offset = chunk_offset + chunk_size; - alloc_profile = BTRFS_BLOCK_GROUP_SYSTEM | - fs_info->avail_system_alloc_bits; - alloc_profile = btrfs_reduce_alloc_profile(root, alloc_profile); - + alloc_profile = btrfs_get_alloc_profile(fs_info->chunk_root, 0); ret = __btrfs_alloc_chunk(trans, extent_root, &sys_map, &sys_chunk_size, &sys_stripe_size, sys_chunk_offset, alloc_profile); -- cgit v1.2.3 From 87533c475187c1420794a2e164bc67a7974f1327 Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Tue, 29 Jan 2013 10:14:48 +0000 Subject: Btrfs: use bit operation for ->fs_state There is no lock to protect fs_info->fs_state, it will introduce some problems, such as the value may be covered by the other task when several tasks modify it. For example: Task0 - CPU0 Task1 - CPU1 mov %fs_state rax or $0x1 rax mov %fs_state rax or $0x2 rax mov rax %fs_state mov rax %fs_state The expected value is 3, but in fact, it is 2. Though this problem doesn't happen now (because there is only one flag currently), the code is error prone, if we add other flags, the above problem will happen to a certainty. Now we use bit operation for it to fix the above problem. In this way, we can make the code more robust and be easy to add new flags. Signed-off-by: Miao Xie Signed-off-by: Josef Bacik --- fs/btrfs/ctree.h | 4 +++- fs/btrfs/disk-io.c | 5 +++-- fs/btrfs/file.c | 2 +- fs/btrfs/scrub.c | 2 +- fs/btrfs/super.c | 4 ++-- fs/btrfs/transaction.c | 9 ++++----- 6 files changed, 14 insertions(+), 12 deletions(-) (limited to 'fs/btrfs/ctree.h') diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 5f5c30aaef36..78fba44ad3d9 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -338,7 +338,9 @@ static inline unsigned long btrfs_chunk_item_size(int num_stripes) /* * File system states */ +#define BTRFS_FS_STATE_ERROR 0 +/* Super block flags */ /* Errors detected */ #define BTRFS_SUPER_FLAG_ERROR (1ULL << 2) @@ -1549,7 +1551,7 @@ struct btrfs_fs_info { u64 qgroup_seq; /* filesystem state */ - u64 fs_state; + unsigned long fs_state; struct btrfs_delayed_root *delayed_root; diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 00b6742fdde7..8e58a1f90546 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -2201,7 +2201,8 @@ int open_ctree(struct super_block *sb, goto fail_alloc; /* check FS state, whether FS is broken. */ - fs_info->fs_state |= btrfs_super_flags(disk_super); + if (btrfs_super_flags(disk_super) & BTRFS_SUPER_FLAG_ERROR) + set_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state); ret = btrfs_check_super_valid(fs_info, sb->s_flags & MS_RDONLY); if (ret) { @@ -3359,7 +3360,7 @@ int close_ctree(struct btrfs_root *root) printk(KERN_ERR "btrfs: commit super ret %d\n", ret); } - if (fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR) + if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) btrfs_error_commit_super(root); btrfs_put_block_group_cache(fs_info); diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c index 13c78ea3ebce..75d0fe134be3 100644 --- a/fs/btrfs/file.c +++ b/fs/btrfs/file.c @@ -1545,7 +1545,7 @@ static ssize_t btrfs_file_aio_write(struct kiocb *iocb, * although we have opened a file as writable, we have * to stop this write operation to ensure FS consistency. */ - if (root->fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR) { + if (test_bit(BTRFS_FS_STATE_ERROR, &root->fs_info->fs_state)) { mutex_unlock(&inode->i_mutex); err = -EROFS; goto out; diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c index 67783e03d121..c78b2a3fc335 100644 --- a/fs/btrfs/scrub.c +++ b/fs/btrfs/scrub.c @@ -2708,7 +2708,7 @@ static noinline_for_stack int scrub_supers(struct scrub_ctx *sctx, int ret; struct btrfs_root *root = sctx->dev_root; - if (root->fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR) + if (test_bit(BTRFS_FS_STATE_ERROR, &root->fs_info->fs_state)) return -EIO; gen = root->fs_info->last_trans_committed; diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c index eda330df45a4..4152f9ea34f5 100644 --- a/fs/btrfs/super.c +++ b/fs/btrfs/super.c @@ -98,7 +98,7 @@ static void __save_error_info(struct btrfs_fs_info *fs_info) * today we only save the error info into ram. Long term we'll * also send it down to the disk */ - fs_info->fs_state = BTRFS_SUPER_FLAG_ERROR; + set_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state); } static void save_error_info(struct btrfs_fs_info *fs_info) @@ -114,7 +114,7 @@ static void btrfs_handle_error(struct btrfs_fs_info *fs_info) if (sb->s_flags & MS_RDONLY) return; - if (fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR) { + if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) { sb->s_flags |= MS_RDONLY; printk(KERN_INFO "btrfs is forced readonly\n"); /* diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c index 34610dc6d140..baf6d74fd0f2 100644 --- a/fs/btrfs/transaction.c +++ b/fs/btrfs/transaction.c @@ -61,7 +61,7 @@ static noinline int join_transaction(struct btrfs_root *root, int type) spin_lock(&fs_info->trans_lock); loop: /* The file system has been taken offline. No new transactions. */ - if (fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR) { + if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) { spin_unlock(&fs_info->trans_lock); return -EROFS; } @@ -113,7 +113,7 @@ loop: kmem_cache_free(btrfs_transaction_cachep, cur_trans); cur_trans = fs_info->running_transaction; goto loop; - } else if (fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR) { + } else if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) { spin_unlock(&fs_info->trans_lock); kmem_cache_free(btrfs_transaction_cachep, cur_trans); return -EROFS; @@ -301,7 +301,7 @@ start_transaction(struct btrfs_root *root, u64 num_items, int type, int ret; u64 qgroup_reserved = 0; - if (root->fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR) + if (test_bit(BTRFS_FS_STATE_ERROR, &root->fs_info->fs_state)) return ERR_PTR(-EROFS); if (current->journal_info) { @@ -645,9 +645,8 @@ static int __btrfs_end_transaction(struct btrfs_trans_handle *trans, btrfs_run_delayed_iputs(root); if (trans->aborted || - root->fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR) { + test_bit(BTRFS_FS_STATE_ERROR, &root->fs_info->fs_state)) err = -EIO; - } assert_qgroups_uptodate(trans); kmem_cache_free(btrfs_trans_handle_cachep, trans); -- cgit v1.2.3 From aa43a17c21cf35329b1a495dd6876798dd8b016b Mon Sep 17 00:00:00 2001 From: Eric Sandeen Date: Thu, 31 Jan 2013 00:54:55 +0000 Subject: btrfs: handle null fs_info in btrfs_panic() At least backref_tree_panic() can apparently pass in a null fs_info, so handle that in __btrfs_panic to get the message out on the console. The btrfs_panic macro also uses fs_info, but that's largely pointless; it's testing to see if BTRFS_MOUNT_PANIC_ON_FATAL_ERROR is not set. But if it *were* set, __btrfs_panic() would have, well, paniced and we wouldn't be here, testing it! So just BUG() at this point. And since we only use fs_info once now, just use it directly. Signed-off-by: Eric Sandeen Signed-off-by: Josef Bacik --- fs/btrfs/ctree.h | 9 ++++++--- fs/btrfs/super.c | 2 +- 2 files changed, 7 insertions(+), 4 deletions(-) (limited to 'fs/btrfs/ctree.h') diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 78fba44ad3d9..1e3a994e1899 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -3653,11 +3653,14 @@ __printf(5, 6) void __btrfs_panic(struct btrfs_fs_info *fs_info, const char *function, unsigned int line, int errno, const char *fmt, ...); +/* + * If BTRFS_MOUNT_PANIC_ON_FATAL_ERROR is in mount_opt, __btrfs_panic + * will panic(). Otherwise we BUG() here. + */ #define btrfs_panic(fs_info, errno, fmt, args...) \ do { \ - struct btrfs_fs_info *_i = (fs_info); \ - __btrfs_panic(_i, __func__, __LINE__, errno, fmt, ##args); \ - BUG_ON(!(_i->mount_opt & BTRFS_MOUNT_PANIC_ON_FATAL_ERROR)); \ + __btrfs_panic(fs_info, __func__, __LINE__, errno, fmt, ##args); \ + BUG(); \ } while (0) /* acl.c */ diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c index fd6a07c30ee9..06537d217419 100644 --- a/fs/btrfs/super.c +++ b/fs/btrfs/super.c @@ -289,7 +289,7 @@ void __btrfs_panic(struct btrfs_fs_info *fs_info, const char *function, vaf.va = &args; errstr = btrfs_decode_error(errno, nbuf); - if (fs_info->mount_opt & BTRFS_MOUNT_PANIC_ON_FATAL_ERROR) + if (fs_info && (fs_info->mount_opt & BTRFS_MOUNT_PANIC_ON_FATAL_ERROR)) panic(KERN_CRIT "BTRFS panic (device %s) in %s:%d: %pV (%s)\n", s_id, function, line, &vaf, errstr); -- cgit v1.2.3 From de78b51a2852bddccd6535e9e12de65f92787a1e Mon Sep 17 00:00:00 2001 From: Eric Sandeen Date: Thu, 31 Jan 2013 18:21:12 +0000 Subject: btrfs: remove cache only arguments from defrag path The entry point at the defrag ioctl always sets "cache only" to 0; the codepaths haven't run for a long time as far as I can tell. Chris says they're dead code, so remove them. Signed-off-by: Eric Sandeen Signed-off-by: Josef Bacik --- fs/btrfs/ctree.c | 65 ++++++++------------------------------------------ fs/btrfs/ctree.h | 8 +++---- fs/btrfs/ioctl.c | 8 +++---- fs/btrfs/transaction.c | 8 +++---- fs/btrfs/transaction.h | 2 +- fs/btrfs/tree-defrag.c | 19 ++++++--------- fs/btrfs/tree-log.c | 4 ++-- 7 files changed, 32 insertions(+), 82 deletions(-) (limited to 'fs/btrfs/ctree.h') diff --git a/fs/btrfs/ctree.c b/fs/btrfs/ctree.c index 35444013f0cf..ecd25a1b4e51 100644 --- a/fs/btrfs/ctree.c +++ b/fs/btrfs/ctree.c @@ -1442,7 +1442,7 @@ int btrfs_comp_cpu_keys(struct btrfs_key *k1, struct btrfs_key *k2) */ int btrfs_realloc_node(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct extent_buffer *parent, - int start_slot, int cache_only, u64 *last_ret, + int start_slot, u64 *last_ret, struct btrfs_key *progress) { struct extent_buffer *cur; @@ -1462,8 +1462,6 @@ int btrfs_realloc_node(struct btrfs_trans_handle *trans, struct btrfs_disk_key disk_key; parent_level = btrfs_header_level(parent); - if (cache_only && parent_level != 1) - return 0; WARN_ON(trans->transaction != root->fs_info->running_transaction); WARN_ON(trans->transid != root->fs_info->generation); @@ -1509,10 +1507,6 @@ int btrfs_realloc_node(struct btrfs_trans_handle *trans, else uptodate = 0; if (!cur || !uptodate) { - if (cache_only) { - free_extent_buffer(cur); - continue; - } if (!cur) { cur = read_tree_block(root, blocknr, blocksize, gen); @@ -4826,8 +4820,8 @@ int btrfs_prev_leaf(struct btrfs_root *root, struct btrfs_path *path) /* * A helper function to walk down the tree starting at min_key, and looking - * for nodes or leaves that are either in cache or have a minimum - * transaction id. This is used by the btree defrag code, and tree logging + * for nodes or leaves that are have a minimum transaction id. + * This is used by the btree defrag code, and tree logging * * This does not cow, but it does stuff the starting key it finds back * into min_key, so you can call btrfs_search_slot with cow=1 on the @@ -4848,7 +4842,7 @@ int btrfs_prev_leaf(struct btrfs_root *root, struct btrfs_path *path) */ int btrfs_search_forward(struct btrfs_root *root, struct btrfs_key *min_key, struct btrfs_key *max_key, - struct btrfs_path *path, int cache_only, + struct btrfs_path *path, u64 min_trans) { struct extent_buffer *cur; @@ -4888,15 +4882,12 @@ again: if (sret && slot > 0) slot--; /* - * check this node pointer against the cache_only and - * min_trans parameters. If it isn't in cache or is too - * old, skip to the next one. + * check this node pointer against the min_trans parameters. + * If it is too old, old, skip to the next one. */ while (slot < nritems) { u64 blockptr; u64 gen; - struct extent_buffer *tmp; - struct btrfs_disk_key disk_key; blockptr = btrfs_node_blockptr(cur, slot); gen = btrfs_node_ptr_generation(cur, slot); @@ -4904,27 +4895,7 @@ again: slot++; continue; } - if (!cache_only) - break; - - if (max_key) { - btrfs_node_key(cur, &disk_key, slot); - if (comp_keys(&disk_key, max_key) >= 0) { - ret = 1; - goto out; - } - } - - tmp = btrfs_find_tree_block(root, blockptr, - btrfs_level_size(root, level - 1)); - - if (tmp && btrfs_buffer_uptodate(tmp, gen, 1) > 0) { - free_extent_buffer(tmp); - break; - } - if (tmp) - free_extent_buffer(tmp); - slot++; + break; } find_next_key: /* @@ -4935,7 +4906,7 @@ find_next_key: path->slots[level] = slot; btrfs_set_path_blocking(path); sret = btrfs_find_next_key(root, path, min_key, level, - cache_only, min_trans); + min_trans); if (sret == 0) { btrfs_release_path(path); goto again; @@ -5400,8 +5371,7 @@ out: /* * this is similar to btrfs_next_leaf, but does not try to preserve * and fixup the path. It looks for and returns the next key in the - * tree based on the current path and the cache_only and min_trans - * parameters. + * tree based on the current path and the min_trans parameters. * * 0 is returned if another key is found, < 0 if there are any errors * and 1 is returned if there are no higher keys in the tree @@ -5410,8 +5380,7 @@ out: * calling this function. */ int btrfs_find_next_key(struct btrfs_root *root, struct btrfs_path *path, - struct btrfs_key *key, int level, - int cache_only, u64 min_trans) + struct btrfs_key *key, int level, u64 min_trans) { int slot; struct extent_buffer *c; @@ -5462,22 +5431,8 @@ next: if (level == 0) btrfs_item_key_to_cpu(c, key, slot); else { - u64 blockptr = btrfs_node_blockptr(c, slot); u64 gen = btrfs_node_ptr_generation(c, slot); - if (cache_only) { - struct extent_buffer *cur; - cur = btrfs_find_tree_block(root, blockptr, - btrfs_level_size(root, level - 1)); - if (!cur || - btrfs_buffer_uptodate(cur, gen, 1) <= 0) { - slot++; - if (cur) - free_extent_buffer(cur); - goto next; - } - free_extent_buffer(cur); - } if (gen < min_trans) { slot++; goto next; diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 1e3a994e1899..7e2cffd2a5d8 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -3125,10 +3125,10 @@ struct extent_buffer *btrfs_root_node(struct btrfs_root *root); struct extent_buffer *btrfs_lock_root_node(struct btrfs_root *root); int btrfs_find_next_key(struct btrfs_root *root, struct btrfs_path *path, struct btrfs_key *key, int lowest_level, - int cache_only, u64 min_trans); + u64 min_trans); int btrfs_search_forward(struct btrfs_root *root, struct btrfs_key *min_key, struct btrfs_key *max_key, - struct btrfs_path *path, int cache_only, + struct btrfs_path *path, u64 min_trans); enum btrfs_compare_tree_result { BTRFS_COMPARE_TREE_NEW, @@ -3181,7 +3181,7 @@ int btrfs_search_slot_for_read(struct btrfs_root *root, int find_higher, int return_any); int btrfs_realloc_node(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct extent_buffer *parent, - int start_slot, int cache_only, u64 *last_ret, + int start_slot, u64 *last_ret, struct btrfs_key *progress); void btrfs_release_path(struct btrfs_path *p); struct btrfs_path *btrfs_alloc_path(void); @@ -3576,7 +3576,7 @@ int btrfs_dirty_pages(struct btrfs_root *root, struct inode *inode, /* tree-defrag.c */ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans, - struct btrfs_root *root, int cache_only); + struct btrfs_root *root); /* sysfs.c */ int btrfs_init_sysfs(void); diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index 96ecefc1724f..daea831f3d36 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -818,7 +818,7 @@ static int find_new_extents(struct btrfs_root *root, while(1) { ret = btrfs_search_forward(root, &min_key, &max_key, - path, 0, newer_than); + path, newer_than); if (ret != 0) goto none; if (min_key.objectid != ino) @@ -1864,7 +1864,7 @@ static noinline int search_ioctl(struct inode *inode, path->keep_locks = 1; while(1) { - ret = btrfs_search_forward(root, &key, &max_key, path, 0, + ret = btrfs_search_forward(root, &key, &max_key, path, sk->min_transid); if (ret != 0) { if (ret > 0) @@ -2212,10 +2212,10 @@ static int btrfs_ioctl_defrag(struct file *file, void __user *argp) ret = -EPERM; goto out; } - ret = btrfs_defrag_root(root, 0); + ret = btrfs_defrag_root(root); if (ret) goto out; - ret = btrfs_defrag_root(root->fs_info->extent_root, 0); + ret = btrfs_defrag_root(root->fs_info->extent_root); break; case S_IFREG: if (!(file->f_mode & FMODE_WRITE)) { diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c index 5144ad19ef47..60481a53e004 100644 --- a/fs/btrfs/transaction.c +++ b/fs/btrfs/transaction.c @@ -959,10 +959,10 @@ static noinline int commit_fs_roots(struct btrfs_trans_handle *trans, } /* - * defrag a given btree. If cacheonly == 1, this won't read from the disk, - * otherwise every leaf in the btree is read and defragged. + * defrag a given btree. + * Every leaf in the btree is read and defragged. */ -int btrfs_defrag_root(struct btrfs_root *root, int cacheonly) +int btrfs_defrag_root(struct btrfs_root *root) { struct btrfs_fs_info *info = root->fs_info; struct btrfs_trans_handle *trans; @@ -976,7 +976,7 @@ int btrfs_defrag_root(struct btrfs_root *root, int cacheonly) if (IS_ERR(trans)) return PTR_ERR(trans); - ret = btrfs_defrag_leaves(trans, root, cacheonly); + ret = btrfs_defrag_leaves(trans, root); btrfs_end_transaction(trans, root); btrfs_btree_balance_dirty(info->tree_root); diff --git a/fs/btrfs/transaction.h b/fs/btrfs/transaction.h index 69700f7b20ac..46628210e5d8 100644 --- a/fs/btrfs/transaction.h +++ b/fs/btrfs/transaction.h @@ -117,7 +117,7 @@ int btrfs_write_and_wait_transaction(struct btrfs_trans_handle *trans, struct btrfs_root *root); int btrfs_add_dead_root(struct btrfs_root *root); -int btrfs_defrag_root(struct btrfs_root *root, int cacheonly); +int btrfs_defrag_root(struct btrfs_root *root); int btrfs_clean_old_snapshots(struct btrfs_root *root); int btrfs_commit_transaction(struct btrfs_trans_handle *trans, struct btrfs_root *root); diff --git a/fs/btrfs/tree-defrag.c b/fs/btrfs/tree-defrag.c index 3b580ee8ab1d..94e05c1f118a 100644 --- a/fs/btrfs/tree-defrag.c +++ b/fs/btrfs/tree-defrag.c @@ -23,13 +23,14 @@ #include "transaction.h" #include "locking.h" -/* defrag all the leaves in a given btree. If cache_only == 1, don't read - * things from disk, otherwise read all the leaves and try to get key order to +/* + * Defrag all the leaves in a given btree. + * Read all the leaves and try to get key order to * better reflect disk order */ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans, - struct btrfs_root *root, int cache_only) + struct btrfs_root *root) { struct btrfs_path *path = NULL; struct btrfs_key key; @@ -41,9 +42,6 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans, u64 last_ret = 0; u64 min_trans = 0; - if (cache_only) - goto out; - if (root->fs_info->extent_root == root) { /* * there's recursion here right now in the tree locking, @@ -86,11 +84,8 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans, } path->keep_locks = 1; - if (cache_only) - min_trans = root->defrag_trans_start; - ret = btrfs_search_forward(root, &key, NULL, path, - cache_only, min_trans); + ret = btrfs_search_forward(root, &key, NULL, path, min_trans); if (ret < 0) goto out; if (ret > 0) { @@ -109,11 +104,11 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans, goto out; } path->slots[1] = btrfs_header_nritems(path->nodes[1]); - next_key_ret = btrfs_find_next_key(root, path, &key, 1, cache_only, + next_key_ret = btrfs_find_next_key(root, path, &key, 1, min_trans); ret = btrfs_realloc_node(trans, root, path->nodes[1], 0, - cache_only, &last_ret, + &last_ret, &root->defrag_progress); if (ret) { WARN_ON(ret == -EAGAIN); diff --git a/fs/btrfs/tree-log.c b/fs/btrfs/tree-log.c index deb7f66356bd..1a79087c4575 100644 --- a/fs/btrfs/tree-log.c +++ b/fs/btrfs/tree-log.c @@ -2738,7 +2738,7 @@ static noinline int log_dir_items(struct btrfs_trans_handle *trans, path->keep_locks = 1; ret = btrfs_search_forward(root, &min_key, &max_key, - path, 0, trans->transid); + path, trans->transid); /* * we didn't find anything from this transaction, see if there @@ -3680,7 +3680,7 @@ static int btrfs_log_inode(struct btrfs_trans_handle *trans, while (1) { ins_nr = 0; ret = btrfs_search_forward(root, &min_key, &max_key, - path, 0, trans->transid); + path, trans->transid); if (ret != 0) break; again: -- cgit v1.2.3 From 5d80366e9b5e56b3ffc1923b4995e83bbbf605e3 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Thu, 7 Feb 2013 16:06:02 -0500 Subject: Btrfs: steal from global reserve if we are cleaning up orphans Sometimes xfstest 83 will fail to remount the scratch device because we've gotten ourselves so full that we cannot cleanup the orphan items. In this case check to see if we're doing the orphan cleanup and if we are allow us to steal our reservation from the global block rsv. With this patch I've not been able to reproduce the failed mount problem. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/ctree.h | 5 +++++ fs/btrfs/extent-tree.c | 11 +++++++++++ fs/btrfs/inode.c | 5 ----- 3 files changed, 16 insertions(+), 5 deletions(-) (limited to 'fs/btrfs/ctree.h') diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 7e2cffd2a5d8..f1cc247f3178 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -1237,6 +1237,11 @@ struct seq_list { u64 seq; }; +enum btrfs_orphan_cleanup_state { + ORPHAN_CLEANUP_STARTED = 1, + ORPHAN_CLEANUP_DONE = 2, +}; + /* fs_info */ struct reloc_control; struct btrfs_device; diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 81aa7cf3ae86..1818dd90c27e 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -102,6 +102,8 @@ static void dump_space_info(struct btrfs_space_info *info, u64 bytes, int dump_block_groups); static int btrfs_update_reserved_bytes(struct btrfs_block_group_cache *cache, u64 num_bytes, int reserve); +static int block_rsv_use_bytes(struct btrfs_block_rsv *block_rsv, + u64 num_bytes); static noinline int block_group_cache_done(struct btrfs_block_group_cache *cache) @@ -4099,6 +4101,15 @@ again: goto again; out: + if (ret == -ENOSPC && + unlikely(root->orphan_cleanup_state == ORPHAN_CLEANUP_STARTED)) { + struct btrfs_block_rsv *global_rsv = + &root->fs_info->global_block_rsv; + + if (block_rsv != global_rsv && + !block_rsv_use_bytes(global_rsv, orig_bytes)) + ret = 0; + } if (flushing) { spin_lock(&space_info->lock); space_info->flush = 0; diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 348b7bb920ef..cf26778085e0 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -2218,11 +2218,6 @@ void btrfs_run_delayed_iputs(struct btrfs_root *root) } } -enum btrfs_orphan_cleanup_state { - ORPHAN_CLEANUP_STARTED = 1, - ORPHAN_CLEANUP_DONE = 2, -}; - /* * This is called in transaction commit time. If there are no orphan * files in the subvolume, it removes orphan item and frees block_rsv -- cgit v1.2.3 From 210549ebe9047ae5a8cc47487203d3ee16a7749b Mon Sep 17 00:00:00 2001 From: David Sterba Date: Sat, 9 Feb 2013 23:38:06 +0000 Subject: btrfs: add cancellation points to defrag The defrag operation can take very long, we want to have a way how to cancel it. The code checks for a pending signal at safe points in the defrag loops and returns EAGAIN. This means a user can press ^C after running 'btrfs fi defrag', woks for both defrag modes, files and root. Returning from the command was instant in my light tests, but may take longer depending on the aging factor of the filesystem. Signed-off-by: David Sterba Signed-off-by: Josef Bacik --- fs/btrfs/ctree.h | 7 +++++++ fs/btrfs/ioctl.c | 6 ++++++ fs/btrfs/transaction.c | 6 ++++++ 3 files changed, 19 insertions(+) (limited to 'fs/btrfs/ctree.h') diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index f1cc247f3178..14f01dc70ea6 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -3786,4 +3786,11 @@ static inline int is_fstree(u64 rootid) return 1; return 0; } + +static inline int btrfs_defrag_cancelled(struct btrfs_fs_info *fs_info) +{ + return signal_pending(current); +} + + #endif diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index 6d6314406e27..d02ec577f70f 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -1202,6 +1202,12 @@ int btrfs_defrag_file(struct inode *inode, struct file *file, if (!(inode->i_sb->s_flags & MS_ACTIVE)) break; + if (btrfs_defrag_cancelled(root->fs_info)) { + printk(KERN_DEBUG "btrfs: defrag_file cancelled\n"); + ret = -EAGAIN; + break; + } + if (!should_defrag_range(inode, (u64)i << PAGE_CACHE_SHIFT, extent_thresh, &last_len, &skip, &defrag_end, range->flags & diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c index 60481a53e004..d574d830a1c4 100644 --- a/fs/btrfs/transaction.c +++ b/fs/btrfs/transaction.c @@ -984,6 +984,12 @@ int btrfs_defrag_root(struct btrfs_root *root) if (btrfs_fs_closing(root->fs_info) || ret != -EAGAIN) break; + + if (btrfs_defrag_cancelled(root->fs_info)) { + printk(KERN_DEBUG "btrfs: defrag_root cancelled\n"); + ret = -EAGAIN; + break; + } } root->defrag_running = 0; return ret; -- cgit v1.2.3 From 569e0f358c0c37f6733702d4a5d2c412860f7169 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Wed, 13 Feb 2013 11:09:14 -0500 Subject: Btrfs: place ordered operations on a per transaction list Miao made the ordered operations stuff run async, which introduced a deadlock where we could get somebody (sync) racing in and committing the transaction while a commit was already happening. The new committer would try and flush ordered operations which would hang waiting for the commit to finish because it is done asynchronously and no longer inherits the callers trans handle. To fix this we need to make the ordered operations list a per transaction list. We can get new inodes added to the ordered operation list by truncating them and then having another process writing to them, so this makes it so that anybody trying to add an ordered operation _must_ start a transaction in order to add itself to the list, which will keep new inodes from getting added to the ordered operations list after we start committing. This should fix the deadlock and also keeps us from doing a lot more work than we need to during commit. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/ctree.h | 7 ------- fs/btrfs/disk-io.c | 11 ++++++----- fs/btrfs/file.c | 15 ++++++++++++++- fs/btrfs/ordered-data.c | 13 ++++++++----- fs/btrfs/ordered-data.h | 3 ++- fs/btrfs/transaction.c | 5 +++-- fs/btrfs/transaction.h | 1 + 7 files changed, 34 insertions(+), 21 deletions(-) (limited to 'fs/btrfs/ctree.h') diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 14f01dc70ea6..961ff2986341 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -1407,13 +1407,6 @@ struct btrfs_fs_info { */ struct list_head delalloc_inodes; - /* - * special rename and truncate targets that must be on disk before - * we're allowed to commit. This is basically the ext3 style - * data=ordered list. - */ - struct list_head ordered_operations; - /* * there is a pool of worker threads for checksumming during writes * and a pool for checksumming after reads. This is because readers diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 9fcce54ecde4..e511d9f78c19 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -56,7 +56,8 @@ static void end_workqueue_fn(struct btrfs_work *work); static void free_fs_root(struct btrfs_root *root); static int btrfs_check_super_valid(struct btrfs_fs_info *fs_info, int read_only); -static void btrfs_destroy_ordered_operations(struct btrfs_root *root); +static void btrfs_destroy_ordered_operations(struct btrfs_transaction *t, + struct btrfs_root *root); static void btrfs_destroy_ordered_extents(struct btrfs_root *root); static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans, struct btrfs_root *root); @@ -2029,7 +2030,6 @@ int open_ctree(struct super_block *sb, INIT_LIST_HEAD(&fs_info->dead_roots); INIT_LIST_HEAD(&fs_info->delayed_iputs); INIT_LIST_HEAD(&fs_info->delalloc_inodes); - INIT_LIST_HEAD(&fs_info->ordered_operations); INIT_LIST_HEAD(&fs_info->caching_block_groups); spin_lock_init(&fs_info->delalloc_lock); spin_lock_init(&fs_info->trans_lock); @@ -3538,7 +3538,8 @@ void btrfs_error_commit_super(struct btrfs_root *root) btrfs_cleanup_transaction(root); } -static void btrfs_destroy_ordered_operations(struct btrfs_root *root) +static void btrfs_destroy_ordered_operations(struct btrfs_transaction *t, + struct btrfs_root *root) { struct btrfs_inode *btrfs_inode; struct list_head splice; @@ -3548,7 +3549,7 @@ static void btrfs_destroy_ordered_operations(struct btrfs_root *root) mutex_lock(&root->fs_info->ordered_operations_mutex); spin_lock(&root->fs_info->ordered_extent_lock); - list_splice_init(&root->fs_info->ordered_operations, &splice); + list_splice_init(&t->ordered_operations, &splice); while (!list_empty(&splice)) { btrfs_inode = list_entry(splice.next, struct btrfs_inode, ordered_operations); @@ -3829,7 +3830,7 @@ int btrfs_cleanup_transaction(struct btrfs_root *root) while (!list_empty(&list)) { t = list_entry(list.next, struct btrfs_transaction, list); - btrfs_destroy_ordered_operations(root); + btrfs_destroy_ordered_operations(t, root); btrfs_destroy_ordered_extents(root); diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c index 75d0fe134be3..b12ba52c4505 100644 --- a/fs/btrfs/file.c +++ b/fs/btrfs/file.c @@ -1628,7 +1628,20 @@ int btrfs_release_file(struct inode *inode, struct file *filp) */ if (test_and_clear_bit(BTRFS_INODE_ORDERED_DATA_CLOSE, &BTRFS_I(inode)->runtime_flags)) { - btrfs_add_ordered_operation(NULL, BTRFS_I(inode)->root, inode); + struct btrfs_trans_handle *trans; + struct btrfs_root *root = BTRFS_I(inode)->root; + + /* + * We need to block on a committing transaction to keep us from + * throwing a ordered operation on to the list and causing + * something like sync to deadlock trying to flush out this + * inode. + */ + trans = btrfs_start_transaction(root, 0); + if (IS_ERR(trans)) + return PTR_ERR(trans); + btrfs_add_ordered_operation(trans, BTRFS_I(inode)->root, inode); + btrfs_end_transaction(trans, root); if (inode->i_size > BTRFS_ORDERED_OPERATIONS_FLUSH_LIMIT) filemap_flush(inode->i_mapping); } diff --git a/fs/btrfs/ordered-data.c b/fs/btrfs/ordered-data.c index 9489fa96e3ed..dc08d77b717e 100644 --- a/fs/btrfs/ordered-data.c +++ b/fs/btrfs/ordered-data.c @@ -612,10 +612,12 @@ void btrfs_wait_ordered_extents(struct btrfs_root *root, int delay_iput) * extra check to make sure the ordered operation list really is empty * before we return */ -int btrfs_run_ordered_operations(struct btrfs_root *root, int wait) +int btrfs_run_ordered_operations(struct btrfs_trans_handle *trans, + struct btrfs_root *root, int wait) { struct btrfs_inode *btrfs_inode; struct inode *inode; + struct btrfs_transaction *cur_trans = trans->transaction; struct list_head splice; struct list_head works; struct btrfs_delalloc_work *work, *next; @@ -626,7 +628,7 @@ int btrfs_run_ordered_operations(struct btrfs_root *root, int wait) mutex_lock(&root->fs_info->ordered_operations_mutex); spin_lock(&root->fs_info->ordered_extent_lock); - list_splice_init(&root->fs_info->ordered_operations, &splice); + list_splice_init(&cur_trans->ordered_operations, &splice); while (!list_empty(&splice)) { btrfs_inode = list_entry(splice.next, struct btrfs_inode, ordered_operations); @@ -643,7 +645,7 @@ int btrfs_run_ordered_operations(struct btrfs_root *root, int wait) if (!wait) list_add_tail(&BTRFS_I(inode)->ordered_operations, - &root->fs_info->ordered_operations); + &cur_trans->ordered_operations); spin_unlock(&root->fs_info->ordered_extent_lock); work = btrfs_alloc_delalloc_work(inode, wait, 1); @@ -653,7 +655,7 @@ int btrfs_run_ordered_operations(struct btrfs_root *root, int wait) list_add_tail(&btrfs_inode->ordered_operations, &splice); list_splice_tail(&splice, - &root->fs_info->ordered_operations); + &cur_trans->ordered_operations); spin_unlock(&root->fs_info->ordered_extent_lock); ret = -ENOMEM; goto out; @@ -1033,6 +1035,7 @@ out: void btrfs_add_ordered_operation(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct inode *inode) { + struct btrfs_transaction *cur_trans = trans->transaction; u64 last_mod; last_mod = max(BTRFS_I(inode)->generation, BTRFS_I(inode)->last_trans); @@ -1047,7 +1050,7 @@ void btrfs_add_ordered_operation(struct btrfs_trans_handle *trans, spin_lock(&root->fs_info->ordered_extent_lock); if (list_empty(&BTRFS_I(inode)->ordered_operations)) { list_add_tail(&BTRFS_I(inode)->ordered_operations, - &root->fs_info->ordered_operations); + &cur_trans->ordered_operations); } spin_unlock(&root->fs_info->ordered_extent_lock); } diff --git a/fs/btrfs/ordered-data.h b/fs/btrfs/ordered-data.h index d523dbd2314d..267ac99095f6 100644 --- a/fs/btrfs/ordered-data.h +++ b/fs/btrfs/ordered-data.h @@ -197,7 +197,8 @@ struct btrfs_ordered_extent *btrfs_lookup_ordered_range(struct inode *inode, int btrfs_ordered_update_i_size(struct inode *inode, u64 offset, struct btrfs_ordered_extent *ordered); int btrfs_find_ordered_sum(struct inode *inode, u64 offset, u64 disk_bytenr, u32 *sum); -int btrfs_run_ordered_operations(struct btrfs_root *root, int wait); +int btrfs_run_ordered_operations(struct btrfs_trans_handle *trans, + struct btrfs_root *root, int wait); void btrfs_add_ordered_operation(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct inode *inode); diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c index d574d830a1c4..0c87d18d1881 100644 --- a/fs/btrfs/transaction.c +++ b/fs/btrfs/transaction.c @@ -157,6 +157,7 @@ loop: spin_lock_init(&cur_trans->delayed_refs.lock); INIT_LIST_HEAD(&cur_trans->pending_snapshots); + INIT_LIST_HEAD(&cur_trans->ordered_operations); list_add_tail(&cur_trans->list, &fs_info->trans_list); extent_io_tree_init(&cur_trans->dirty_pages, fs_info->btree_inode->i_mapping); @@ -1456,7 +1457,7 @@ static int btrfs_flush_all_pending_stuffs(struct btrfs_trans_handle *trans, * it here and no for sure that nothing new will be added * to the list */ - ret = btrfs_run_ordered_operations(root, 1); + ret = btrfs_run_ordered_operations(trans, root, 1); return ret; } @@ -1479,7 +1480,7 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans, int should_grow = 0; unsigned long now = get_seconds(); - ret = btrfs_run_ordered_operations(root, 0); + ret = btrfs_run_ordered_operations(trans, root, 0); if (ret) { btrfs_abort_transaction(trans, root, ret); btrfs_end_transaction(trans, root); diff --git a/fs/btrfs/transaction.h b/fs/btrfs/transaction.h index 46628210e5d8..3f772fd0191a 100644 --- a/fs/btrfs/transaction.h +++ b/fs/btrfs/transaction.h @@ -43,6 +43,7 @@ struct btrfs_transaction { wait_queue_head_t writer_wait; wait_queue_head_t commit_wait; struct list_head pending_snapshots; + struct list_head ordered_operations; struct btrfs_delayed_ref_root delayed_refs; int aborted; }; -- cgit v1.2.3 From cdb4c5748cb3ac533889a6b0b95aa10651e68785 Mon Sep 17 00:00:00 2001 From: Zach Brown Date: Wed, 20 Feb 2013 00:55:13 +0000 Subject: btrfs: define BTRFS_MAGIC as a u64 value super.magic is an le64 but it's treated as an unterminated string when compared against BTRFS_MAGIC which is defined as a string. Instead define BTRFS_MAGIC as a normal hex value and use endian helpers to compare it to the super's magic. I tested this by mounting an fs made before the change and made sure that it didn't introduce sparse errors. This matches a similar cleanup that is pending in btrfs-progs. David Sterba pointed out that we should fix the kernel side as well :). Signed-off-by: Zach Brown Signed-off-by: Josef Bacik --- fs/btrfs/check-integrity.c | 3 +-- fs/btrfs/ctree.h | 2 +- fs/btrfs/disk-io.c | 6 ++---- fs/btrfs/volumes.c | 3 +-- 4 files changed, 5 insertions(+), 9 deletions(-) (limited to 'fs/btrfs/ctree.h') diff --git a/fs/btrfs/check-integrity.c b/fs/btrfs/check-integrity.c index 11d47bfb62b4..18af6f48781a 100644 --- a/fs/btrfs/check-integrity.c +++ b/fs/btrfs/check-integrity.c @@ -813,8 +813,7 @@ static int btrfsic_process_superblock_dev_mirror( (bh->b_data + (dev_bytenr & 4095)); if (btrfs_super_bytenr(super_tmp) != dev_bytenr || - strncmp((char *)(&(super_tmp->magic)), BTRFS_MAGIC, - sizeof(super_tmp->magic)) || + super_tmp->magic != cpu_to_le64(BTRFS_MAGIC) || memcmp(device->uuid, super_tmp->dev_item.uuid, BTRFS_UUID_SIZE) || btrfs_super_nodesize(super_tmp) != state->metablock_size || btrfs_super_leafsize(super_tmp) != state->metablock_size || diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 961ff2986341..1679051f4d39 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -46,7 +46,7 @@ extern struct kmem_cache *btrfs_path_cachep; extern struct kmem_cache *btrfs_free_space_cachep; struct btrfs_ordered_sum; -#define BTRFS_MAGIC "_BHRfS_M" +#define BTRFS_MAGIC 0x4D5F53665248425FULL /* ascii _BHRfS_M, no null */ #define BTRFS_MAX_MIRRORS 3 diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index e511d9f78c19..39ff34a62a24 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -2407,8 +2407,7 @@ int open_ctree(struct super_block *sb, sb->s_blocksize = sectorsize; sb->s_blocksize_bits = blksize_bits(sectorsize); - if (strncmp((char *)(&disk_super->magic), BTRFS_MAGIC, - sizeof(disk_super->magic))) { + if (disk_super->magic != cpu_to_le64(BTRFS_MAGIC)) { printk(KERN_INFO "btrfs: valid FS not found on %s\n", sb->s_id); goto fail_sb_buffer; } @@ -2815,8 +2814,7 @@ struct buffer_head *btrfs_read_dev_super(struct block_device *bdev) super = (struct btrfs_super_block *)bh->b_data; if (btrfs_super_bytenr(super) != bytenr || - strncmp((char *)(&super->magic), BTRFS_MAGIC, - sizeof(super->magic))) { + super->magic != cpu_to_le64(BTRFS_MAGIC)) { brelse(bh); continue; } diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c index 567076542d1d..72b1cf1b2b5e 100644 --- a/fs/btrfs/volumes.c +++ b/fs/btrfs/volumes.c @@ -855,8 +855,7 @@ int btrfs_scan_one_device(const char *path, fmode_t flags, void *holder, disk_super = p + (bytenr & ~PAGE_CACHE_MASK); if (btrfs_super_bytenr(disk_super) != bytenr || - strncmp((char *)(&disk_super->magic), BTRFS_MAGIC, - sizeof(disk_super->magic))) + disk_super->magic != cpu_to_le64(BTRFS_MAGIC)) goto error_unmap; devid = btrfs_stack_device_id(&disk_super->dev_item); -- cgit v1.2.3 From dc81cdc58ad2f413b96b9004f8d681e5dc554473 Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Wed, 20 Feb 2013 23:32:52 -0700 Subject: Btrfs: fix remount vs autodefrag If we remount the fs to close the auto defragment or make the fs R/O, we should stop the auto defragment. Signed-off-by: Miao Xie Signed-off-by: Chris Mason --- fs/btrfs/ctree.h | 2 ++ fs/btrfs/file.c | 5 +++++ fs/btrfs/super.c | 40 ++++++++++++++++++++++++++++++++++++++-- 3 files changed, 45 insertions(+), 2 deletions(-) (limited to 'fs/btrfs/ctree.h') diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 3dcedfe4f759..ae8dcc406805 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -339,6 +339,7 @@ static inline unsigned long btrfs_chunk_item_size(int num_stripes) * File system states */ #define BTRFS_FS_STATE_ERROR 0 +#define BTRFS_FS_STATE_REMOUNTING 1 /* Super block flags */ /* Errors detected */ @@ -1902,6 +1903,7 @@ struct btrfs_ioctl_defrag_range_args { #define btrfs_clear_opt(o, opt) ((o) &= ~BTRFS_MOUNT_##opt) #define btrfs_set_opt(o, opt) ((o) |= BTRFS_MOUNT_##opt) +#define btrfs_raw_test_opt(o, opt) ((o) & BTRFS_MOUNT_##opt) #define btrfs_test_opt(root, opt) ((root)->fs_info->mount_opt & \ BTRFS_MOUNT_##opt) /* diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c index 9f67e623206d..6e6dd8cdad92 100644 --- a/fs/btrfs/file.c +++ b/fs/btrfs/file.c @@ -374,6 +374,11 @@ int btrfs_run_defrag_inodes(struct btrfs_fs_info *fs_info) atomic_inc(&fs_info->defrag_running); while(1) { + /* Pause the auto defragger. */ + if (test_bit(BTRFS_FS_STATE_REMOUNTING, + &fs_info->fs_state)) + break; + if (!__need_auto_defrag(fs_info->tree_root)) break; diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c index db1ba9a2ed64..68a29a1ea068 100644 --- a/fs/btrfs/super.c +++ b/fs/btrfs/super.c @@ -1202,6 +1202,38 @@ static void btrfs_resize_thread_pool(struct btrfs_fs_info *fs_info, new_pool_size); } +static inline void btrfs_remount_prepare(struct btrfs_fs_info *fs_info, + unsigned long old_opts, int flags) +{ + set_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state); + + if (btrfs_raw_test_opt(old_opts, AUTO_DEFRAG) && + (!btrfs_raw_test_opt(fs_info->mount_opt, AUTO_DEFRAG) || + (flags & MS_RDONLY))) { + /* wait for any defraggers to finish */ + wait_event(fs_info->transaction_wait, + (atomic_read(&fs_info->defrag_running) == 0)); + if (flags & MS_RDONLY) + sync_filesystem(fs_info->sb); + } +} + +static inline void btrfs_remount_cleanup(struct btrfs_fs_info *fs_info, + unsigned long old_opts) +{ + /* + * We need cleanup all defragable inodes if the autodefragment is + * close or the fs is R/O. + */ + if (btrfs_raw_test_opt(old_opts, AUTO_DEFRAG) && + (!btrfs_raw_test_opt(fs_info->mount_opt, AUTO_DEFRAG) || + (fs_info->sb->s_flags & MS_RDONLY))) { + btrfs_cleanup_defrag_inodes(fs_info); + } + + clear_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state); +} + static int btrfs_remount(struct super_block *sb, int *flags, char *data) { struct btrfs_fs_info *fs_info = btrfs_sb(sb); @@ -1215,6 +1247,8 @@ static int btrfs_remount(struct super_block *sb, int *flags, char *data) unsigned int old_metadata_ratio = fs_info->metadata_ratio; int ret; + btrfs_remount_prepare(fs_info, old_opts, *flags); + ret = btrfs_parse_options(root, data); if (ret) { ret = -EINVAL; @@ -1225,7 +1259,7 @@ static int btrfs_remount(struct super_block *sb, int *flags, char *data) fs_info->thread_pool_size, old_thread_pool_size); if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY)) - return 0; + goto out; if (*flags & MS_RDONLY) { /* @@ -1280,7 +1314,8 @@ static int btrfs_remount(struct super_block *sb, int *flags, char *data) } sb->s_flags &= ~MS_RDONLY; } - +out: + btrfs_remount_cleanup(fs_info, old_opts); return 0; restore: @@ -1297,6 +1332,7 @@ restore: btrfs_resize_thread_pool(fs_info, old_thread_pool_size, fs_info->thread_pool_size); fs_info->metadata_ratio = old_metadata_ratio; + btrfs_remount_cleanup(fs_info, old_opts); return ret; } -- cgit v1.2.3 From d5c1207017cd8387b4d3224dd7ab6cf5cd7f1c9a Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Thu, 28 Feb 2013 10:04:33 +0000 Subject: Btrfs: fix wrong reserved space in qgroup during snap/subv creation There are two problems in the space reservation of the snapshot/ subvolume creation. - don't reserve the space for the root item insertion - the space which is reserved in the qgroup is different with the free space reservation. we need reserve free space for 7 items, but in qgroup reservation, we need reserve space only for 3 items. So we implement new metadata reservation functions for the snapshot/subvolume creation. Signed-off-by: Miao Xie Signed-off-by: Josef Bacik --- fs/btrfs/ctree.h | 9 +++++-- fs/btrfs/extent-tree.c | 65 ++++++++++++++++++++++++++++++++++++++++---------- fs/btrfs/ioctl.c | 62 ++++++++++++++++++++++++++++++++--------------- fs/btrfs/transaction.c | 4 +--- fs/btrfs/transaction.h | 1 + 5 files changed, 105 insertions(+), 36 deletions(-) (limited to 'fs/btrfs/ctree.h') diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index ae8dcc406805..0d82922179db 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -3106,8 +3106,13 @@ void btrfs_trans_release_metadata(struct btrfs_trans_handle *trans, int btrfs_orphan_reserve_metadata(struct btrfs_trans_handle *trans, struct inode *inode); void btrfs_orphan_release_metadata(struct inode *inode); -int btrfs_snap_reserve_metadata(struct btrfs_trans_handle *trans, - struct btrfs_pending_snapshot *pending); +int btrfs_subvolume_reserve_metadata(struct btrfs_root *root, + struct btrfs_block_rsv *rsv, + int nitems, + u64 *qgroup_reserved); +void btrfs_subvolume_release_metadata(struct btrfs_root *root, + struct btrfs_block_rsv *rsv, + u64 qgroup_reserved); int btrfs_delalloc_reserve_metadata(struct inode *inode, u64 num_bytes); void btrfs_delalloc_release_metadata(struct inode *inode, u64 num_bytes); int btrfs_delalloc_reserve_space(struct inode *inode, u64 num_bytes); diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 5681a91ed400..7cb9d734f6e5 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -4560,19 +4560,60 @@ void btrfs_orphan_release_metadata(struct inode *inode) btrfs_block_rsv_release(root, root->orphan_block_rsv, num_bytes); } -int btrfs_snap_reserve_metadata(struct btrfs_trans_handle *trans, - struct btrfs_pending_snapshot *pending) +/* + * btrfs_subvolume_reserve_metadata() - reserve space for subvolume operation + * root: the root of the parent directory + * rsv: block reservation + * items: the number of items that we need do reservation + * qgroup_reserved: used to return the reserved size in qgroup + * + * This function is used to reserve the space for snapshot/subvolume + * creation and deletion. Those operations are different with the + * common file/directory operations, they change two fs/file trees + * and root tree, the number of items that the qgroup reserves is + * different with the free space reservation. So we can not use + * the space reseravtion mechanism in start_transaction(). + */ +int btrfs_subvolume_reserve_metadata(struct btrfs_root *root, + struct btrfs_block_rsv *rsv, + int items, + u64 *qgroup_reserved) { - struct btrfs_root *root = pending->root; - struct btrfs_block_rsv *src_rsv = get_block_rsv(trans, root); - struct btrfs_block_rsv *dst_rsv = &pending->block_rsv; - /* - * two for root back/forward refs, two for directory entries, - * one for root of the snapshot and one for parent inode. - */ - u64 num_bytes = btrfs_calc_trans_metadata_size(root, 6); - dst_rsv->space_info = src_rsv->space_info; - return block_rsv_migrate_bytes(src_rsv, dst_rsv, num_bytes); + u64 num_bytes; + int ret; + + if (root->fs_info->quota_enabled) { + /* One for parent inode, two for dir entries */ + num_bytes = 3 * root->leafsize; + ret = btrfs_qgroup_reserve(root, num_bytes); + if (ret) + return ret; + } else { + num_bytes = 0; + } + + *qgroup_reserved = num_bytes; + + num_bytes = btrfs_calc_trans_metadata_size(root, items); + rsv->space_info = __find_space_info(root->fs_info, + BTRFS_BLOCK_GROUP_METADATA); + ret = btrfs_block_rsv_add(root, rsv, num_bytes, + BTRFS_RESERVE_FLUSH_ALL); + if (ret) { + if (*qgroup_reserved) + btrfs_qgroup_free(root, *qgroup_reserved); + } + + return ret; +} + +void btrfs_subvolume_release_metadata(struct btrfs_root *root, + struct btrfs_block_rsv *rsv, + u64 qgroup_reserved) +{ + btrfs_block_rsv_release(root, rsv, (u64)-1); + if (qgroup_reserved) + btrfs_qgroup_free(root, qgroup_reserved); } /** diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index 8dcd4ff0c3a5..56d92549389c 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -363,7 +363,7 @@ static noinline int btrfs_ioctl_fitrim(struct file *file, void __user *arg) return 0; } -static noinline int create_subvol(struct btrfs_root *root, +static noinline int create_subvol(struct inode *dir, struct dentry *dentry, char *name, int namelen, u64 *async_transid, @@ -374,32 +374,39 @@ static noinline int create_subvol(struct btrfs_root *root, struct btrfs_root_item root_item; struct btrfs_inode_item *inode_item; struct extent_buffer *leaf; + struct btrfs_root *root = BTRFS_I(dir)->root; struct btrfs_root *new_root; - struct dentry *parent = dentry->d_parent; - struct inode *dir; + struct btrfs_block_rsv block_rsv; struct timespec cur_time = CURRENT_TIME; int ret; int err; u64 objectid; u64 new_dirid = BTRFS_FIRST_FREE_OBJECTID; u64 index = 0; + u64 qgroup_reserved; uuid_le new_uuid; ret = btrfs_find_free_objectid(root->fs_info->tree_root, &objectid); if (ret) return ret; - dir = parent->d_inode; - + btrfs_init_block_rsv(&block_rsv, BTRFS_BLOCK_RSV_TEMP); /* - * 1 - inode item - * 2 - refs - * 1 - root item - * 2 - dir items + * The same as the snapshot creation, please see the comment + * of create_snapshot(). */ - trans = btrfs_start_transaction(root, 6); - if (IS_ERR(trans)) - return PTR_ERR(trans); + ret = btrfs_subvolume_reserve_metadata(root, &block_rsv, + 7, &qgroup_reserved); + if (ret) + return ret; + + trans = btrfs_start_transaction(root, 0); + if (IS_ERR(trans)) { + ret = PTR_ERR(trans); + goto out; + } + trans->block_rsv = &block_rsv; + trans->bytes_reserved = block_rsv.size; ret = btrfs_qgroup_inherit(trans, root->fs_info, 0, objectid, inherit); if (ret) @@ -515,6 +522,8 @@ static noinline int create_subvol(struct btrfs_root *root, BUG_ON(ret); fail: + trans->block_rsv = NULL; + trans->bytes_reserved = 0; if (async_transid) { *async_transid = trans->transid; err = btrfs_commit_transaction_async(trans, root, 1); @@ -526,7 +535,8 @@ fail: if (!ret) d_instantiate(dentry, btrfs_lookup_dentry(dir, dentry)); - +out: + btrfs_subvolume_release_metadata(root, &block_rsv, qgroup_reserved); return ret; } @@ -549,21 +559,31 @@ static int create_snapshot(struct btrfs_root *root, struct inode *dir, btrfs_init_block_rsv(&pending_snapshot->block_rsv, BTRFS_BLOCK_RSV_TEMP); + /* + * 1 - parent dir inode + * 2 - dir entries + * 1 - root item + * 2 - root ref/backref + * 1 - root of snapshot + */ + ret = btrfs_subvolume_reserve_metadata(BTRFS_I(dir)->root, + &pending_snapshot->block_rsv, 7, + &pending_snapshot->qgroup_reserved); + if (ret) + goto out; + pending_snapshot->dentry = dentry; pending_snapshot->root = root; pending_snapshot->readonly = readonly; pending_snapshot->dir = dir; pending_snapshot->inherit = inherit; - trans = btrfs_start_transaction(root->fs_info->extent_root, 6); + trans = btrfs_start_transaction(root, 0); if (IS_ERR(trans)) { ret = PTR_ERR(trans); goto fail; } - ret = btrfs_snap_reserve_metadata(trans, pending_snapshot); - BUG_ON(ret); - spin_lock(&root->fs_info->trans_lock); list_add(&pending_snapshot->list, &trans->transaction->pending_snapshots); @@ -600,6 +620,10 @@ static int create_snapshot(struct btrfs_root *root, struct inode *dir, d_instantiate(dentry, inode); ret = 0; fail: + btrfs_subvolume_release_metadata(BTRFS_I(dir)->root, + &pending_snapshot->block_rsv, + pending_snapshot->qgroup_reserved); +out: kfree(pending_snapshot); return ret; } @@ -733,8 +757,8 @@ static noinline int btrfs_mksubvol(struct path *parent, error = create_snapshot(snap_src, dir, dentry, name, namelen, async_transid, readonly, inherit); } else { - error = create_subvol(BTRFS_I(dir)->root, dentry, - name, namelen, async_transid, inherit); + error = create_subvol(dir, dentry, name, namelen, + async_transid, inherit); } if (!error) fsnotify_mkdir(dir, dentry); diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c index 71de435a291e..f11c2e0a3746 100644 --- a/fs/btrfs/transaction.c +++ b/fs/btrfs/transaction.c @@ -1082,7 +1082,7 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans, path = btrfs_alloc_path(); if (!path) { ret = pending->error = -ENOMEM; - goto path_alloc_fail; + return ret; } new_root_item = kmalloc(sizeof(*new_root_item), GFP_NOFS); @@ -1279,8 +1279,6 @@ no_free_objectid: kfree(new_root_item); root_item_alloc_fail: btrfs_free_path(path); -path_alloc_fail: - btrfs_block_rsv_release(root, &pending->block_rsv, (u64)-1); return ret; } diff --git a/fs/btrfs/transaction.h b/fs/btrfs/transaction.h index 5f67fba07ab4..3c8e0d25c8e4 100644 --- a/fs/btrfs/transaction.h +++ b/fs/btrfs/transaction.h @@ -90,6 +90,7 @@ struct btrfs_pending_snapshot { struct btrfs_qgroup_inherit *inherit; /* block reservation for the operation */ struct btrfs_block_rsv block_rsv; + u64 qgroup_reserved; /* extra metadata reseration for relocation */ int error; bool readonly; -- cgit v1.2.3