btrfs: zoned: fixes for data relocation
Get series with b4
b4 am [email protected]Versions newest first · linked by author and title
- v2 Latest2 patches 3 repliesIn review
- v12 patches 2 repliesSuperseded
Changes from the cover letter
- v2:
- fix block_group->zoned_data_reloc_ongoing to be a bit.
Patches in v2 No applies recorded yet
| # | Subject and files | Size | Replies | Applied |
|---|---|---|---|---|
| 1/2 | btrfs: zoned: prevent allocation from previous data relocation BGfs/ fs/ fs/ +2 more | +53 | 0 replies | |
| 2/2 | btrfs: zoned: fix critical section of relocation inode writebackfs/ | +2 | 3 replies |
Reviews trailers per patch
1 of 2 patches has a Reviewed-by or Acked-by.
| Patch | Johannes Thumshirn |
|---|---|
| 1/2 | |
| 2/2 | R-b |
Trailers given in replies are filled; those carried in the patches from earlier versions are outlined with dashes.
Cover letter [PATCH v2 0/2] btrfs: zoned: fixes for data relocation
Changes:
- v2:
- fix block_group->zoned_data_reloc_ongoing to be a bit.
There are two long-standing potential bugs in the data relocation path of zoned btrfs. They are recently revealed by commit 5f0addf7b890 ("btrfs:
zoned: use dedicated lock for data relocation"). One is a mixed issue of WRITE (for relocation extents) and ZONE APPEND (for regular extent) at the same time, which confuses the write pointer. The other one is a too short critical section, which can cause an out-of-order issue of the IOs.
Actually, these bugs are easily reproducible with a smaller zone size (e.g, 128 MB) with fstests btrfs/232. For example, IO failures occurs like this:
[99909.031820][T4038707] WARNING: CPU: 3 PID: 4038707 at fs/btrfs/extent-tree.c:2381 btrfs_cross_ref_exist+0xfc/0x120 [btrfs] <snip> [99909.268769][T4038707] Call Trace: [99909.272105][T4038707] <TASK> [99909.275093][T4038707] run_delalloc_nocow+0x7f1/0x11a0 [btrfs] [99909.280996][T4038707] ? test_range_bit+0x174/0x320 [btrfs] [99909.286622][T4038707] ? fallback_to_cow+0x980/0x980 [btrfs] [99909.292333][T4038707] ? find_lock_delalloc_range+0x33e/0x3e0 [btrfs] [99909.298825][T4038707] btrfs_run_delalloc_range+0x445/0x1320 [btrfs] [99909.305222][T4038707] ? test_range_bit+0x320/0x320 [btrfs] [99909.310844][T4038707] ? lock_downgrade+0x6a0/0x6a0 [99909.315732][T4038707] ? orc_find.part.0+0x1ed/0x300 [99909.320705][T4038707] ? __module_address.part.0+0x25/0x300 [99909.326280][T4038707] writepage_delalloc+0x159/0x310 [btrfs] <snip> [99909.883814][ C3] sd 10:0:1:0: [sde] tag#2620 FAILED Result: hostbyte=DID_OK driverbyte=DRIVER_OK cmd_age=0s [99909.893855][ C3] sd 10:0:1:0: [sde] tag#2620 Sense Key : Illegal Request [current] [99909.901819][ C3] sd 10:0:1:0: [sde] tag#2620 Add. Sense: Unaligned write command [99909.909525][ C3] sd 10:0:1:0: [sde] tag#2620 CDB: Write(16) 8a 00 00 00 00 00 02 f3 63 87 00 00 00 2c 00 00 [99909.919544][ C3] critical target error, dev sde, sector 396041272 op 0x1:(WRITE) flags 0x800 phys_seg 3 prio class 0 [99909.930329][ C3] BTRFS error (device dm-1): bdev /dev/mapper/dml_102_2 errs: wr 1, rd 0, flush 0, corrupt 0, gen 0
Or, an assertion failure occur like this:
[ 12.527832] assertion failed: start >= found_start && end <= found_end, in fs/btrfs/free-space-tree.c:737 <snip> [ 12.533391] Call Trace: [ 12.533391] <TASK> [ 12.533391] __remove_from_free_space_tree.cold+0x11/0x22 [btrfs] [ 12.542073] ? setup_items_for_insert.isra.0+0x2bf/0x3f0 [btrfs] [ 12.542073] remove_from_free_space_tree+0x80/0x110 [btrfs] [ 12.542073] alloc_reserved_file_extent+0x1b4/0x240 [btrfs] [ 12.542073] __btrfs_run_delayed_refs+0x692/0xf30 [btrfs] [ 12.542073] ? btrfs_btree_balance_dirty+0x2f/0x50 [btrfs] [ 12.542073] btrfs_run_delayed_refs+0x81/0x1e0 [btrfs] [ 12.542073] btrfs_commit_transaction+0x54/0xaf0 [btrfs] [ 12.542073] ? start_transaction+0xc2/0x5b0 [btrfs] [ 12.542073] ? _raw_read_lock_irqsave+0x20/0x40 [ 12.542073] relocate_block_group+0x320/0x550 [btrfs] [ 12.542073] btrfs_relocate_block_group+0x1f9/0x3a0 [btrfs] [ 12.542073] btrfs_relocate_chunk+0x36/0xf0 [btrfs] [ 12.542073] btrfs_reclaim_bgs_work.cold+0x4f/0x74 [btrfs] [ 12.542073] process_one_work+0x1b0/0x310 [ 12.542073] worker_thread+0x48/0x3d0 [ 12.542073] ? rescuer_thread+0x3a0/0x3a0 [ 12.542073] kthread+0xed/0x120 [ 12.550506] ? kthread_complete_and_exit+0x20/0x20 [ 12.550506] ret_from_fork+0x22/0x30 [ 12.550506] </TASK>
This series fixes the two issues. The first one is fixed by introducing a new btrfs_block_group bit to disallow extent allocation but still allow nocow writes to start.
The second one is simply fixed by extending the critical section.
Naohiro Aota (2):
btrfs: zoned: prevent allocation from previous data relocation BG
btrfs: zoned: fix critical section of relocation inode writeback
fs/btrfs/block-group.h | 1 + fs/btrfs/extent-tree.c | 20 ++++++++++++++++++-- fs/btrfs/extent_io.c | 3 ++- fs/btrfs/inode.c | 2 ++ fs/btrfs/zoned.c | 27 +++++++++++++++++++++++++++ fs/btrfs/zoned.h | 5 +++++ 6 files changed, 55 insertions(+), 3 deletions(-)
Signature · 1 line
2.35.1