aboutsummaryrefslogtreecommitdiffstats
path: root/fs/bcachefs/movinggc.c (follow)
AgeCommit message (Collapse)AuthorFilesLines
2025-09-29Remove bcachefs core codeLinus Torvalds1-476/+0
bcachefs was marked 'externally maintained' in 6.17 but the code remained to make the transition smoother. It's now a DKMS module, making the in-kernel code stale, so remove it to avoid any version confusion. Link: https://lore.kernel.org/linux-bcachefs/yokpt2d2g2lluyomtqrdvmkl3amv3kgnipmenobkpgx537kay7@xgcgjviv3n7x/T/ Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2025-07-16bcachefs: Fix reference to invalid bucket in copygcKent Overstreet1-1/+1
Use bch2_dev_bucket_tryget() instead of bch2_dev_tryget() before checking the bucket bitmap. Reported-by: syzbot+3168625f36f4a539237e@syzkaller.appspotmail.com Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2025-06-11bcachefs: Don't put rhashtable on stackKent Overstreet1-9/+13
Object debugging generally needs special provisions for putting said objects on the stack, which rhashtable does not have. Reported-by: syzbot+bcc38a9556d0324c2ec2@syzkaller.appspotmail.com Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2025-06-01bcachefs: Replace rcu_read_lock() with guardsKent Overstreet1-14/+12
The new guard(), scoped_guard() allow for more natural code. Some of the uses with creative flow control have been left. Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2025-05-21bcachefs: Remove duplicate call to bch2_trans_begin()Alan Huang1-2/+0
There is one in for_each_btree_key_max(). Signed-off-by: Alan Huang <mmpgouride@gmail.com> Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2025-05-21bcachefs: bch2_check_bucket_backpointer_mismatch()Kent Overstreet1-0/+3
Detect buckets with missing backpointers, and run repair on demand. __bch2_move_data_phys() now calls bch2_check_bucket_backpointer_mismatch() as it walks buckets, which checks for missing backpointers by comparing backpointers against bucket sector counts. When missing backpointers are detected, we kick off bch2_check_extents_to_backpointers() asynchronously - right away if we're trying to evacuate, or with a threshold if we're just running copygc. Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2025-05-21bcachefs: Improve bucket_bitmap codeKent Overstreet1-3/+3
Add some more helpers, and mismatches is now a superset of the empty bitmap - simplifies most checks. Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2025-05-21bcachefs: struct bch_fs_recoveryKent Overstreet1-1/+1
bch_fs has gotten obnoxiously big, let's start organizing thins a bit better. Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2025-05-21bcachefs: move_buckets in rhashtable when allocatedKent Overstreet1-19/+19
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2025-05-21bcachefs: Move pending buckets queue to buckets_in_flightKent Overstreet1-13/+11
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2025-05-21bcachefs: kill move_bucket_in_flightKent Overstreet1-43/+39
Small cleanup/simplification, and prep work for the next patch, which will add checking if buckets don't get evacuated because they're missing backpointers. Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2025-05-21bcachefs: buckets_in_flight on stackKent Overstreet1-18/+10
copygc runs with a full stack available, there's no reason to dynamically allocate this. Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2025-05-21bcachefs: bch2_copygc_dev_wait_amount()Kent Overstreet1-22/+34
Factor out the per-device calculations, for better introspection. Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2025-05-21bcachefs: for_each_rw_member_rcu()Kent Overstreet1-1/+3
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2025-04-21bcachefs: Start copygc, rebalance threads earlierKent Overstreet1-0/+7
Previously, copygc and rebalance weren't started until the very end of mounting, after all recvoery passes have finished. But copygc really should be started earlier, since it may be needed for allocations to make forward progress. Additionally, we've been seeing occasional bug reports where starting the kthread fails due to a pending signal - i.e. we're getting timed out by systemd (during a version upgrade), but we're not seeing the signal until mount is about to complete. Additionally, we now have copygc/rebalance explicitly wait for check_snapshots to complete (if being run); they require that for snapshot_is_ancestor() in the data move path. Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2025-04-02bcachefs: bch_dev_usage_fullKent Overstreet1-2/+6
All the fastpaths that need device usage don't need the sector totals or fragmentation, just bucket counts. Split bch_dev_usage up into two different versions, the normal one with just bucket counts. This is also a stack usage improvement, since we have a bch_dev_usage on the stack in the allocation path. Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2025-03-16bcachefs: rebalance, copygc status also print stacktraceKent Overstreet1-0/+11
These are commonly needed when debugging, and saves from having to ask users to dig. Also, rebalance_status now includes pending rebalance work. Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2025-03-14bcachefs: s/BCH_LRU_FRAGMENTATION_START/BCH_LRU_BUCKET_FRAGMENTATION/Kent Overstreet1-2/+2
FRAGMENTATION_START was incorrect, there's currently only one fragmentation LRU (at the end of the reserved bits for LRU type), and we're getting ready to add a stripe fragmentation lru - so give it a better name. Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2025-03-06bcachefs: copygc now skips non-rw devicesKent Overstreet1-13/+12
There's no point in doing copygc on non-rw devices: the fragmentation doesn't matter if we're not writing to them, and we may not have anywhere to put the data on our other devices. Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2025-01-26bcachefs: Fix trace_copygcKent Overstreet1-4/+7
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2024-12-21bcachefs: Rename btree_iter_peek_upto() -> btree_iter_peek_max()Kent Overstreet1-1/+1
We'll be introducing btree_iter_peek_prev_min(), so rename for consistency. Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2024-12-21bcachefs: copygc_enabled, rebalance_enabled now opts.h optionsKent Overstreet1-2/+2
They can now be set at mount time Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2024-10-04bcachefs: Kill alloc_v4.fragmentation_lruKent Overstreet1-3/+9
The fragmentation_lru field hasn't been needed since we reworked the LRU btrees to use the btree write buffer; previously it was used to resolve collisions, but the revised LRU btree uses the backpointer (the bucket) as part of the key. It should have been deleted at the time of the LRU rework; since it wasn't, that left places for bugs to hide, in check/repair. This fixes LRU fsck on a filesystem image helpfully provided by a user who disappeared before I could get his name for the reported-by. Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2024-08-24bcachefs: Fix failure to flush moves before sleeping in copygcKent Overstreet1-1/+1
This fixes an apparent deadlock - rebalance would get stuck trying to take nocow locks because they weren't being released by copygc. Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2024-07-14bcachefs: Improve copygc_wait_to_text()Kent Overstreet1-3/+8
printing the raw values can occasionally be very useful Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2024-06-10bcachefs: Enable automatic shrinking for rhashtablesKent Overstreet1-3/+4
Since the key cache shrinker walks the rhashtable, a mostly empty rhashtable leads to really nasty reclaim performance issues. Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2024-05-08bcachefs: bch2_trans_unlock() must always be followed by relock() or begin()Kent Overstreet1-0/+2
We're about to add new asserts for btree_trans locking consistency, and part of that requires that aren't using the btree_trans while it's unlocked. Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2024-05-08bcachefs: iter/update/trigger/str_hash flag cleanupKent Overstreet1-1/+1
Combine iter/update/trigger/str_hash flags into a single enum, and x-macroize them for a to_text() function later. These flags are all for a specific iter/key/update context, so it makes sense to group them together - iter/update/trigger flags were already given distinct bits, this cleans up and unifies that handling. Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2024-03-18bcachefs: Improve bch2_fatal_error()Kent Overstreet1-2/+1
error messages should always include __func__ Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2024-01-01bcachefs: for_each_member_device() now declares loop iterKent Overstreet1-5/+2
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2024-01-01bcachefs: for_each_btree_key() now declares loop iterKent Overstreet1-2/+0
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2024-01-01bcachefs: for_each_btree_key_upto() -> for_each_btree_key_old_upto()Kent Overstreet1-1/+1
And for_each_btree_key2_upto -> for_each_btree_key_upto Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2024-01-01bcachefs: darray_for_each() now declares loop iterKent Overstreet1-1/+0
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2024-01-01bcachefs: bch_err_(fn|msg) check if should printKent Overstreet1-4/+3
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2024-01-01bcachefs: copygc shouldn't try moving buckets on errorDaniel Hill1-4/+12
Co-developed-by: Kent Overstreet <kent.overstreet@linux.dev> Signed-off-by: Daniel Hill <daniel@gluo.nz> Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2024-01-01bcachefs: copygc should wakeup on shutdown if disabledDaniel Hill1-1/+2
Signed-off-by: Daniel Hill <daniel@gluo.nz> Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2024-01-01bcachefs: remove dead bch2_evacuate_bucket()Daniel Hill1-1/+1
Signed-off-by: Daniel Hill <daniel@gluo.nz> Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2024-01-01bcachefs: bch2_btree_write_buffer_flush() -> bch2_btree_write_buffer_tryflush()Kent Overstreet1-2/+2
More accurate naming. Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2024-01-01bcachefs: Clean up btree write buffer write ref handlingKent Overstreet1-0/+3
__bch2_btree_write_buffer_flush() now assumes a write ref is already held (as called by the transaction commit path); and the wrappers bch2_write_buffer_flush() and flush_sync() take an explicit write ref. This means internally the write buffer code can always use BTREE_INSERT_NOCHECK_RW, instead of in the previous code passing flags around and hoping the NOCHECK_RW flag was always carried around correctly. Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2024-01-01bcachefs: New bucket sector count helpersKent Overstreet1-1/+1
This introduces bch2_bucket_sectors() and bch2_bucket_sectors_dirty(), prep work for separately accounting stripe sectors. Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2023-11-28bcachefs: Extra kthread_should_stop() calls for copygcKent Overstreet1-1/+1
This fixes a bug where going read-only was taking longer than it should have due to copygc forgetting to check kthread_should_stop() Additionally: fix a missing is_kthread check in bch2_move_ratelimit(). Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2023-11-04bcachefs: fix odebug warn and lockdep splat due to on-stack rhashtableBrian Foster1-10/+14
Guenter Roeck reports a lockdep splat and DEBUG_OBJECTS_WORK related warning when bch2_copygc_thread() initializes its rhashtable. The lockdep splat relates to a warning print caused by the fact that the rhashtable exists on the stack but is not annotated as so. This is something that could be addressed by INIT_WORK_ONSTACK(), but rhashtable doesn't expose that control and probably isnt worth the churn for just one user. Instead, dynamically allocate the buckets_in_flight structure and avoid the splat that way. Reported-by: Guenter Roeck <linux@roeck-us.net> Tested-by: Guenter Roeck <linux@roeck-us.net> Signed-off-by: Brian Foster <bfoster@redhat.com> Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2023-11-04bcachefs: Data move path now uses bch2_trans_unlock_long()Kent Overstreet1-2/+2
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2023-11-04bcachefs: Ensure copygc does not spinKent Overstreet1-2/+18
If copygc does no work - finds no fragmented buckets - wait for a bit of IO to happen. Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2023-10-31bcachefs: move: move_stats refactoringKent Overstreet1-0/+1
data_progress_list is gone - it was redundant with moving_context_list The upcoming rebalance rewrite is going to have it using two different move_stats objects with the same moving_context, depending on whether it's scanning or using the rebalance_work btree - this patch plumbs stats around a bit differently so that will work. Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2023-10-31bcachefs: moving_context now owns a btree_transKent Overstreet1-20/+16
btree_trans and moving_context are used together, and having the moving_context owns the transaction object reduces some plumbing. Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2023-10-22bcachefs: Heap allocate btree_transKent Overstreet1-9/+9
We're using more stack than we'd like in a number of functions, and btree_trans is the biggest object that we stack allocate. But we have to do a heap allocatation to initialize it anyways, so there's no real downside to heap allocating the entire thing. Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2023-10-22bcachefs: Fix W=12 build errorsKent Overstreet1-13/+13
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2023-10-22bcachefs: Fix -Wcompare-distinct-pointer-types in bch2_copygc_get_buckets()Nathan Chancellor1-1/+1
When building bcachefs for 32-bit ARM, there is a warning when using max() to compare an expression involving 'size_t' with an 'unsigned long' literal: fs/bcachefs/movinggc.c:159:21: error: comparison of distinct pointer types ('typeof (16UL) *' (aka 'unsigned long *') and 'typeof (buckets_in_flight->nr / 4) *' (aka 'unsigned int *')) [-Werror,-Wcompare-distinct-pointer-types] 159 | size_t nr_to_get = max(16UL, buckets_in_flight->nr / 4); | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ include/linux/minmax.h:76:19: note: expanded from macro 'max' 76 | #define max(x, y) __careful_cmp(x, y, >) | ^~~~~~~~~~~~~~~~~~~~~~ include/linux/minmax.h:38:24: note: expanded from macro '__careful_cmp' 38 | __builtin_choose_expr(__safe_cmp(x, y), \ | ^~~~~~~~~~~~~~~~ include/linux/minmax.h:28:4: note: expanded from macro '__safe_cmp' 28 | (__typecheck(x, y) && __no_side_effects(x, y)) | ^~~~~~~~~~~~~~~~~ include/linux/minmax.h:22:28: note: expanded from macro '__typecheck' 22 | (!!(sizeof((typeof(x) *)1 == (typeof(y) *)1))) | ~~~~~~~~~~~~~~ ^ ~~~~~~~~~~~~~~ 1 error generated. On 64-bit architectures, size_t is 'unsigned long', so there is no warning when comparing these two expressions. Use max_t(size_t, ...) for this situation, eliminating the warning. Fixes: dd49018737d4 ("bcachefs: Rhashtable based buckets_in_flight for copygc") Signed-off-by: Nathan Chancellor <nathan@kernel.org> Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
2023-10-22bcachefs: Break up io.cKent Overstreet1-8/+0
More reorganization, this splits up io.c into - io_read.c - io_misc.c - fallocate, fpunch, truncate - io_write.c Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
generated by cgit v1.2.3 (git 2.39.1) at 2026-01-04 11:27:10 +0000