Message ID | 20200117212602.6737-42-josef@toxicpanda.com (mailing list archive) |
---|---|
State | New, archived |
Headers | show |
Series | Cleanup how we handle root refs, part 1 | expand |
On 17.01.20 г. 23:26 ч., Josef Bacik wrote: > In adding things like eb leak checking and root leak checking there were > a lot of weird corner cases that come from the fact that > > 1) We do not init the fs_info until we get to open_ctree time in the > normal case and > > 2) The test infrastructure half-init's the fs_info for things that it > needs. > > This makes it really annoying to make changes because you have to add > init in two different places, have special cases for testing fs_info's > that may not have certain things init'ed, and cases for fs_info's that > didn't make it to open_ctree and thus are not fully init'ed. > > Fix this by extracting out the non-allocating init of the fs info into > it's own public function and use that to make sure we're all getting > consistent views of an allocated fs_info. > Imo it will be better if btrfs_init_fs_info is called from init_mount_fs_info. And then called explicitly from the test code. That way we keep the initialization close. Otherwise having it in btrfs_mount_root is just confusing and not very obvious what's going on.
On 23.01.20 г. 14:30 ч., Nikolay Borisov wrote: > > > On 17.01.20 г. 23:26 ч., Josef Bacik wrote: >> In adding things like eb leak checking and root leak checking there were >> a lot of weird corner cases that come from the fact that >> >> 1) We do not init the fs_info until we get to open_ctree time in the >> normal case and >> >> 2) The test infrastructure half-init's the fs_info for things that it >> needs. >> >> This makes it really annoying to make changes because you have to add >> init in two different places, have special cases for testing fs_info's >> that may not have certain things init'ed, and cases for fs_info's that >> didn't make it to open_ctree and thus are not fully init'ed. >> >> Fix this by extracting out the non-allocating init of the fs info into >> it's own public function and use that to make sure we're all getting >> consistent views of an allocated fs_info. >> > > > Imo it will be better if btrfs_init_fs_info is called from > init_mount_fs_info. And then called explicitly from the test code. That > way we keep the initialization close. Otherwise having it in > btrfs_mount_root is just confusing and not very obvious what's going on. > Actually now that I think more about it IMO btrfs_init_fs_info should be exported iff btrfs tests are enabled i.e using EXPORT_FOR_TESTS defines
On 1/23/20 7:30 AM, Nikolay Borisov wrote: > > > On 17.01.20 г. 23:26 ч., Josef Bacik wrote: >> In adding things like eb leak checking and root leak checking there were >> a lot of weird corner cases that come from the fact that >> >> 1) We do not init the fs_info until we get to open_ctree time in the >> normal case and >> >> 2) The test infrastructure half-init's the fs_info for things that it >> needs. >> >> This makes it really annoying to make changes because you have to add >> init in two different places, have special cases for testing fs_info's >> that may not have certain things init'ed, and cases for fs_info's that >> didn't make it to open_ctree and thus are not fully init'ed. >> >> Fix this by extracting out the non-allocating init of the fs info into >> it's own public function and use that to make sure we're all getting >> consistent views of an allocated fs_info. >> > > > Imo it will be better if btrfs_init_fs_info is called from > init_mount_fs_info. And then called explicitly from the test code. That > way we keep the initialization close. Otherwise having it in > btrfs_mount_root is just confusing and not very obvious what's going on. > I'm not sure what's confusing about it, we're allocating the structure and then initializing it. And as I point out in the changelog, this is necessary because the error handling is bonkers because we can fail at multiple places _before_ open_ctree, and having random garbage in the fs_info makes proper cleanup a pain. Thus it _has_ to be exported and the fs_info _has_ to have it's static elements properly initialized in order to make cleanup in the error cases clean and sane. Thanks, Josef
diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index b7e4313bdc6f..87bad959b1a5 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -2649,10 +2649,8 @@ static int __cold init_tree_roots(struct btrfs_fs_info *fs_info) return ret; } -static int init_fs_info(struct btrfs_fs_info *fs_info, struct super_block *sb) +void btrfs_init_fs_info(struct btrfs_fs_info *fs_info) { - int ret; - INIT_RADIX_TREE(&fs_info->fs_roots_radix, GFP_ATOMIC); INIT_RADIX_TREE(&fs_info->buffer_radix, GFP_ATOMIC); INIT_LIST_HEAD(&fs_info->trans_list); @@ -2696,7 +2694,6 @@ static int init_fs_info(struct btrfs_fs_info *fs_info, struct super_block *sb) atomic_set(&fs_info->reada_works_cnt, 0); atomic_set(&fs_info->nr_delayed_iputs, 0); atomic64_set(&fs_info->tree_mod_seq, 0); - fs_info->sb = sb; fs_info->max_inline = BTRFS_DEFAULT_MAX_INLINE; fs_info->metadata_ratio = 0; fs_info->defrag_inodes = RB_ROOT; @@ -2722,9 +2719,6 @@ static int init_fs_info(struct btrfs_fs_info *fs_info, struct super_block *sb) btrfs_init_balance(fs_info); btrfs_init_async_reclaim_work(&fs_info->async_reclaim_work); - sb->s_blocksize = BTRFS_BDEV_BLOCKSIZE; - sb->s_blocksize_bits = blksize_bits(BTRFS_BDEV_BLOCKSIZE); - spin_lock_init(&fs_info->block_group_cache_lock); fs_info->block_group_cache_tree = RB_ROOT; fs_info->first_logical_byte = (u64)-1; @@ -2769,6 +2763,15 @@ static int init_fs_info(struct btrfs_fs_info *fs_info, struct super_block *sb) fs_info->swapfile_pins = RB_ROOT; fs_info->send_in_progress = 0; +} + +static int init_mount_fs_info(struct btrfs_fs_info *fs_info, struct super_block *sb) +{ + int ret; + + fs_info->sb = sb; + sb->s_blocksize = BTRFS_BDEV_BLOCKSIZE; + sb->s_blocksize_bits = blksize_bits(BTRFS_BDEV_BLOCKSIZE); ret = init_srcu_struct(&fs_info->subvol_srcu); if (ret) @@ -2833,7 +2836,7 @@ int __cold open_ctree(struct super_block *sb, int clear_free_space_tree = 0; int level; - ret = init_fs_info(fs_info, sb); + ret = init_mount_fs_info(fs_info, sb); if (ret) { err = ret; goto fail; diff --git a/fs/btrfs/disk-io.h b/fs/btrfs/disk-io.h index 97e7ac474a52..2414d572bc9a 100644 --- a/fs/btrfs/disk-io.h +++ b/fs/btrfs/disk-io.h @@ -39,6 +39,7 @@ static inline u64 btrfs_sb_offset(int mirror) struct btrfs_device; struct btrfs_fs_devices; +void btrfs_init_fs_info(struct btrfs_fs_info *fs_info); int btrfs_verify_level_key(struct extent_buffer *eb, int level, struct btrfs_key *first_key, u64 parent_transid); struct extent_buffer *read_tree_block(struct btrfs_fs_info *fs_info, u64 bytenr, diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c index 8ce292a47634..8d5bb8dfed0d 100644 --- a/fs/btrfs/super.c +++ b/fs/btrfs/super.c @@ -1531,6 +1531,7 @@ static struct dentry *btrfs_mount_root(struct file_system_type *fs_type, error = -ENOMEM; goto error_sec_opts; } + btrfs_init_fs_info(fs_info); fs_info->super_copy = kzalloc(BTRFS_SUPER_INFO_SIZE, GFP_KERNEL); fs_info->super_for_commit = kzalloc(BTRFS_SUPER_INFO_SIZE, GFP_KERNEL); diff --git a/fs/btrfs/tests/btrfs-tests.c b/fs/btrfs/tests/btrfs-tests.c index 27f5b662d2cb..683381a692bc 100644 --- a/fs/btrfs/tests/btrfs-tests.c +++ b/fs/btrfs/tests/btrfs-tests.c @@ -120,6 +120,8 @@ struct btrfs_fs_info *btrfs_alloc_dummy_fs_info(u32 nodesize, u32 sectorsize) kfree(fs_info); return NULL; } + INIT_LIST_HEAD(&fs_info->fs_devices->devices); + fs_info->super_copy = kzalloc(sizeof(struct btrfs_super_block), GFP_KERNEL); if (!fs_info->super_copy) { @@ -128,6 +130,8 @@ struct btrfs_fs_info *btrfs_alloc_dummy_fs_info(u32 nodesize, u32 sectorsize) return NULL; } + btrfs_init_fs_info(fs_info); + fs_info->nodesize = nodesize; fs_info->sectorsize = sectorsize; @@ -138,30 +142,6 @@ struct btrfs_fs_info *btrfs_alloc_dummy_fs_info(u32 nodesize, u32 sectorsize) return NULL; } - spin_lock_init(&fs_info->buffer_lock); - spin_lock_init(&fs_info->qgroup_lock); - spin_lock_init(&fs_info->super_lock); - spin_lock_init(&fs_info->fs_roots_radix_lock); - spin_lock_init(&fs_info->tree_mod_seq_lock); - mutex_init(&fs_info->qgroup_ioctl_lock); - mutex_init(&fs_info->qgroup_rescan_lock); - rwlock_init(&fs_info->tree_mod_log_lock); - fs_info->running_transaction = NULL; - fs_info->qgroup_tree = RB_ROOT; - fs_info->qgroup_ulist = NULL; - atomic64_set(&fs_info->tree_mod_seq, 0); - INIT_LIST_HEAD(&fs_info->dirty_qgroups); - INIT_LIST_HEAD(&fs_info->dead_roots); - INIT_LIST_HEAD(&fs_info->tree_mod_seq_list); - INIT_LIST_HEAD(&fs_info->fs_devices->devices); - INIT_RADIX_TREE(&fs_info->buffer_radix, GFP_ATOMIC); - INIT_RADIX_TREE(&fs_info->fs_roots_radix, GFP_ATOMIC); - extent_io_tree_init(fs_info, &fs_info->freed_extents[0], - IO_TREE_FS_INFO_FREED_EXTENTS0, NULL); - extent_io_tree_init(fs_info, &fs_info->freed_extents[1], - IO_TREE_FS_INFO_FREED_EXTENTS1, NULL); - extent_map_tree_init(&fs_info->mapping_tree); - fs_info->pinned_extents = &fs_info->freed_extents[0]; set_bit(BTRFS_FS_STATE_DUMMY_FS_INFO, &fs_info->fs_state); test_mnt->mnt_sb->s_fs_info = fs_info;
In adding things like eb leak checking and root leak checking there were a lot of weird corner cases that come from the fact that 1) We do not init the fs_info until we get to open_ctree time in the normal case and 2) The test infrastructure half-init's the fs_info for things that it needs. This makes it really annoying to make changes because you have to add init in two different places, have special cases for testing fs_info's that may not have certain things init'ed, and cases for fs_info's that didn't make it to open_ctree and thus are not fully init'ed. Fix this by extracting out the non-allocating init of the fs info into it's own public function and use that to make sure we're all getting consistent views of an allocated fs_info. Signed-off-by: Josef Bacik <josef@toxicpanda.com> --- fs/btrfs/disk-io.c | 19 +++++++++++-------- fs/btrfs/disk-io.h | 1 + fs/btrfs/super.c | 1 + fs/btrfs/tests/btrfs-tests.c | 28 ++++------------------------ 4 files changed, 17 insertions(+), 32 deletions(-)