id stringlengths 40 40 | instruction stringlengths 68 308 | input stringlengths 17 33.4k | output stringlengths 100 30k |
|---|---|---|---|
c6cf49a95ac7aea3450cacb66b7de1d46671deeb | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix blocking with locks held | Problem Details:
This is a major oopsy - we should always be unlocking before calling
closure_sync(), else we'll cause a deadlock.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev> | ```diff
diff --git a/fs/bcachefs/btree_update_interior.c b/fs/bcachefs/btree_update_interior.c
index 7a682c1751b3..797efa738dc4 100644
--- a/fs/bcachefs/btree_update_interior.c
+++ b/fs/bcachefs/btree_update_interior.c
@@ -1117,10 +1117,10 @@ bch2_btree_update_start(struct btree_trans *trans, struct btree_path *path,
... |
01ed3359b2705d7b862ad0644dabc616774f8779 | Analyze and fix the following issue in the Linux kernel code: bcachefs: btree_update_nodes_written() needs BTREE_INSERT_USE_RESERVE | Problem Details:
This fixes an obvious deadlock - whoops.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev> | ```diff
diff --git a/fs/bcachefs/btree_update_interior.c b/fs/bcachefs/btree_update_interior.c
index 0d87b1c88ed4..7a682c1751b3 100644
--- a/fs/bcachefs/btree_update_interior.c
+++ b/fs/bcachefs/btree_update_interior.c
@@ -625,6 +625,7 @@ static void btree_update_nodes_written(struct btree_update *as)
ret = commit_do... |
d602657cd185784bf4b227ecd1a88ed90735eae5 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix error handling in bch2_btree_update_start() | Problem Details:
We were checking for -EAGAIN, but we're not returned that when we didn't
pass a closure to wait with - oops.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev> | ```diff
diff --git a/fs/bcachefs/btree_update_interior.c b/fs/bcachefs/btree_update_interior.c
index 7028597358d5..0d87b1c88ed4 100644
--- a/fs/bcachefs/btree_update_interior.c
+++ b/fs/bcachefs/btree_update_interior.c
@@ -1110,8 +1110,8 @@ bch2_btree_update_start(struct btree_trans *trans, struct btree_path *path,
... |
8b31e4fc7d9e27fd5dc56bb063acf9e4ea575973 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Kill normalize_read_intent_locks() | Problem Details:
Before we had the deadlock cycle detector, we didn't want to be holding
read locks when taking intent locks, because blocking on an intent lock
while holding a read lock was a lock ordering violation that could
cause a deadlock.
With the cycle detector this is no longer an issue, so this code can be
d... | ```diff
diff --git a/fs/bcachefs/btree_update_leaf.c b/fs/bcachefs/btree_update_leaf.c
index 7b21971fa13d..ad46c887185b 100644
--- a/fs/bcachefs/btree_update_leaf.c
+++ b/fs/bcachefs/btree_update_leaf.c
@@ -740,62 +740,6 @@ bch2_trans_commit_write_locked(struct btree_trans *trans,
return ret;
}
-static inline void... |
2ec254c098da677295c2487ae36e75a26d557222 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Ensure bch2_btree_node_lock_write_nofail() never fails | Problem Details:
In order for bch2_btree_node_lock_write_nofail() to never produce a
deadlock, we must ensure we're never holding read locks when using it.
Fortunately, it's only used from code paths where any read locks may be
safely dropped.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev> | ```diff
diff --git a/fs/bcachefs/btree_locking.c b/fs/bcachefs/btree_locking.c
index 11f83a936dc7..6793d7dd18d7 100644
--- a/fs/bcachefs/btree_locking.c
+++ b/fs/bcachefs/btree_locking.c
@@ -320,6 +320,40 @@ int __bch2_btree_node_lock_write(struct btree_trans *trans, struct btree_path *p
return ret;
}
+void bch2_b... |
0d7009d7ca99ad9261a7cffcecd515108377a6ac | Analyze and fix the following issue in the Linux kernel code: bcachefs: Delete old deadlock avoidance code | Problem Details:
This deletes our old lock ordering based deadlock avoidance code.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/btree_cache.c b/fs/bcachefs/btree_cache.c
index a0e9e14e3fa5..aeb058c800cd 100644
--- a/fs/bcachefs/btree_cache.c
+++ b/fs/bcachefs/btree_cache.c
@@ -151,8 +151,6 @@ void bch2_btree_node_hash_remove(struct btree_cache *bc, struct btree *b)
/* Cause future lookups for this node to fa... |
96d994b37cfcf468bf1d71527ae95ad93a311e38 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Print deadlock cycle in debugfs | Problem Details:
In the event that we're not finished debugging the cycle detector, this
adds a new file to debugfs that shows what the cycle detector finds, if
anything. By comparing this with btree_transactions, which shows held
locks for every btree_transaction, we'll be able to determine if it's
the cycle detector ... | ```diff
diff --git a/fs/bcachefs/btree_iter.c b/fs/bcachefs/btree_iter.c
index 5773b00e69ac..ece80d7914b2 100644
--- a/fs/bcachefs/btree_iter.c
+++ b/fs/bcachefs/btree_iter.c
@@ -2992,7 +2992,6 @@ bch2_btree_path_node_to_text(struct printbuf *out,
c.n[0], c.n[1], c.n[2], pid);
}
-#ifdef CONFIG_BCACHEFS_DEBUG_T... |
33bd5d068603f9e81e0b73dbe50e9b88b2e56d0d | Analyze and fix the following issue in the Linux kernel code: bcachefs: Deadlock cycle detector | Problem Details:
We've outgrown our own deadlock avoidance strategy.
The btree iterator API provides an interface where the user doesn't need
to concern themselves with lock ordering - different btree iterators can
be traversed in any order. Without special care, this will lead to
deadlocks.
Our previous strategy was... | ```diff
diff --git a/fs/bcachefs/bcachefs_format.h b/fs/bcachefs/bcachefs_format.h
index 0e80fe2568f2..5471b797be93 100644
--- a/fs/bcachefs/bcachefs_format.h
+++ b/fs/bcachefs/bcachefs_format.h
@@ -1400,7 +1400,8 @@ struct bch_sb_field_disk_groups {
x(trans_restart_key_cache_upgrade, 70) \
x(trans_traverse_all, ... |
62448afee714354a26db8a0f3c644f58628f0792 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix bch2_btree_node_upgrade() | Problem Details:
Previously, if we were trying to upgrade from a read to an intent lock
but we held an additional read lock via another btree_path,
bch2_btree_node_upgrade() would always fail, in six_lock_tryupgrade().
This patch factors out the code that __bch2_btree_node_lock_write() uses
to temporarily drop extra r... | ```diff
diff --git a/fs/bcachefs/btree_locking.c b/fs/bcachefs/btree_locking.c
index 08dbc799bb35..5b6d8184ea45 100644
--- a/fs/bcachefs/btree_locking.c
+++ b/fs/bcachefs/btree_locking.c
@@ -230,6 +230,7 @@ bool bch2_btree_node_upgrade(struct btree_trans *trans,
struct btree_path *path, unsigned level)
{
st... |
845cffed0d343ecea9f6ff3883cac9a6872d9920 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Add a debug assert | Problem Details:
Chasing down a strange locking bug.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev> | ```diff
diff --git a/fs/bcachefs/btree_iter.c b/fs/bcachefs/btree_iter.c
index 1650ba87ef03..df9949cef907 100644
--- a/fs/bcachefs/btree_iter.c
+++ b/fs/bcachefs/btree_iter.c
@@ -1110,6 +1110,9 @@ static int btree_path_traverse_one(struct btree_trans *trans,
path->level = btree_path_up_until_good_node(trans, path, ... |
e4b7254c754b676a6f4d607fd92cd71d221ff130 | Analyze and fix the following issue in the Linux kernel code: six locks: Fix a lost wakeup | Problem Details:
There was a lost wakeup between a read unlock in percpu mode and a write
lock. The unlock path unlocks, then executes a barrier, then checks for
waiters; correspondingly, the lock side should set the wait bit and
execute a barrier, then attempt to take the lock.
Signed-off-by: Kent Overstreet <kent.ov... | ```diff
diff --git a/fs/bcachefs/six.c b/fs/bcachefs/six.c
index 01ff210ff18c..abdc2414f58b 100644
--- a/fs/bcachefs/six.c
+++ b/fs/bcachefs/six.c
@@ -198,6 +198,14 @@ retry:
atomic64_add(__SIX_VAL(write_locking, 1),
&lock->state.counter);
smp_mb__after_atomic();
+ } else if (!(lock->state.waiters & ... |
0bfb9f42b7b16aa11a7b5d283b0b7b98d11476b7 | Analyze and fix the following issue in the Linux kernel code: six locks: six_lock_waiter() | Problem Details:
This allows passing in the wait list entry - to be used for a deadlock
cycle detector.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev> | ```diff
diff --git a/fs/bcachefs/six.c b/fs/bcachefs/six.c
index d5e09fae1538..82e8d77c3082 100644
--- a/fs/bcachefs/six.c
+++ b/fs/bcachefs/six.c
@@ -439,10 +439,10 @@ static inline bool six_optimistic_spin(struct six_lock *lock, enum six_lock_type
noinline
static int __six_lock_type_slowpath(struct six_lock *lock... |
ebc6f76a667f5fc599a5f76515f6881dfb82af2f | Analyze and fix the following issue in the Linux kernel code: six locks: Simplify wait lists | Problem Details:
This switches to a single list of waiters, instead of separate lists for
read and intent, and switches write locks to also use the wait lists
instead of being handled differently.
Also, removal from the wait list is now done by the process waiting on
the lock, not the process doing the wakeup. This is... | ```diff
diff --git a/fs/bcachefs/six.c b/fs/bcachefs/six.c
index 0ab72f59d23b..d5e09fae1538 100644
--- a/fs/bcachefs/six.c
+++ b/fs/bcachefs/six.c
@@ -89,50 +89,38 @@ static inline unsigned pcpu_read_count(struct six_lock *lock)
return read_count;
}
-struct six_lock_waiter {
- struct list_head list;
- struct task_... |
e4215d0fec777e6516306f5f1b69a45a4205dce0 | Analyze and fix the following issue in the Linux kernel code: bcachefs: All held locks must be in a btree path | Problem Details:
With the new deadlock cycle detector, it's critical that all held locks
be marked in a btree_path, because that's what the cycle detector
traverses - any locks that aren't correctly marked will cause deadlocks.
This changes the btree_path to allocate some btree_paths for the new
nodes, since until the... | ```diff
diff --git a/fs/bcachefs/btree_iter.c b/fs/bcachefs/btree_iter.c
index b1c81278ad75..e65c300ffe40 100644
--- a/fs/bcachefs/btree_iter.c
+++ b/fs/bcachefs/btree_iter.c
@@ -619,61 +619,6 @@ static inline bool btree_path_advance_to_pos(struct btree_path *path,
return true;
}
-/*
- * Verify that iterator for p... |
5a82c7c7d1925f6f060a427f38ea17b53c6945f1 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix sb_field_counters formatting | Problem Details:
We have counters with longer names now, so adjust the tabstop - also,
make sure there's always a space printed between the name and the
number.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev> | ```diff
diff --git a/fs/bcachefs/counters.c b/fs/bcachefs/counters.c
index 745f856e6d3e..edd1b2537f48 100644
--- a/fs/bcachefs/counters.c
+++ b/fs/bcachefs/counters.c
@@ -36,7 +36,7 @@ void bch2_sb_counters_to_text(struct printbuf *out, struct bch_sb *sb,
for (i = 0; i < nr; i++) {
if (i < BCH_COUNTER_NR)
- pr... |
1ffb876fb0f31632b761ee721f633e0d7491ca7b | Analyze and fix the following issue in the Linux kernel code: bcachefs: Kill journal_keys->journal_seq_base | Problem Details:
This removes an optimization that didn't actually save us any memory,
due to alignment, but did make the code more complicated than it needed
to be. We were also seeing a bug where journal_seq_base wasn't getting
correctly initailized, so hopefully it'll fix that too.
Signed-off-by: Kent Overstreet <k... | ```diff
diff --git a/fs/bcachefs/bcachefs.h b/fs/bcachefs/bcachefs.h
index c1d96222f4c3..74da688d994b 100644
--- a/fs/bcachefs/bcachefs.h
+++ b/fs/bcachefs/bcachefs.h
@@ -555,13 +555,13 @@ struct journal_seq_blacklist_table {
struct journal_keys {
struct journal_key {
+ u64 journal_seq;
+ u32 journal_offset;
... |
e87b0e4a7120eeca1850666351b75bf8ceb9d5c9 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix redundant transaction restart | Problem Details:
Little bit of tidying up, this makes the counters a little bit clearer
as to what's happening.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev> | ```diff
diff --git a/fs/bcachefs/btree_locking.c b/fs/bcachefs/btree_locking.c
index bfe9780aea3a..08dbc799bb35 100644
--- a/fs/bcachefs/btree_locking.c
+++ b/fs/bcachefs/btree_locking.c
@@ -413,14 +413,13 @@ int bch2_trans_relock(struct btree_trans *trans)
struct btree_path *path;
if (unlikely(trans->restarted))... |
38474c264252475196a5e3c555b2625a5bc36a00 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Avoid using btree_node_lock_nopath() | Problem Details:
With the upcoming cycle detector, we have to be careful about using
btree_node_lock_nopath - in particular, using it to take write locks can
cause deadlocks.
All held locks need to be tracked in a btree_path, so that the cycle
detector knows about them - unless we know that we cannot cause
deadlocks f... | ```diff
diff --git a/fs/bcachefs/btree_iter.c b/fs/bcachefs/btree_iter.c
index 9c39027513b0..b1c81278ad75 100644
--- a/fs/bcachefs/btree_iter.c
+++ b/fs/bcachefs/btree_iter.c
@@ -716,13 +716,13 @@ void bch2_trans_node_add(struct btree_trans *trans, struct btree *b)
struct btree_path *path;
trans_for_each_path(tra... |
3d21d48e898a2eadc9055c44e0fd51e6087c9e9f | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix usage of six lock's percpu mode, key cache version | Problem Details:
Similar to "bcachefs: Fix usage of six lock's percpu mode", six locks
have a percpu mode, but we can't switch between percpu and non percpu
modes while a lock is in use: threads attempting to take a read lock may
race, and we'll end up with the read count permanently off.
Fixing this the "correct" way... | ```diff
diff --git a/fs/bcachefs/btree_key_cache.c b/fs/bcachefs/btree_key_cache.c
index b3d383ff3d5b..89e540c5a244 100644
--- a/fs/bcachefs/btree_key_cache.c
+++ b/fs/bcachefs/btree_key_cache.c
@@ -13,6 +13,11 @@
#include <linux/sched/mm.h>
+static inline bool btree_uses_pcpu_readers(enum btree_id id)
+{
+ return... |
6b81f194f345d15dd15601ee7b604a0640445895 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix six_lock_readers_add() | Problem Details:
Have to be careful with bit fields - when subtracting, this was
overflowing into the write_locking bit.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev> | ```diff
diff --git a/fs/bcachefs/btree_locking.c b/fs/bcachefs/btree_locking.c
index 24d0ea903380..158cb7ac64f2 100644
--- a/fs/bcachefs/btree_locking.c
+++ b/fs/bcachefs/btree_locking.c
@@ -10,10 +10,12 @@ struct lock_class_key bch2_btree_node_lock_key;
static inline void six_lock_readers_add(struct six_lock *lock,... |
ca7d8fcabf29fae627babb72bda9b51763f9a145 | Analyze and fix the following issue in the Linux kernel code: bcachefs: New locking functions | Problem Details:
In the future, with the new deadlock cycle detector, we won't be using
bare six_lock_* anymore: lock wait entries will all be embedded in
btree_trans, and we will need a btree_trans context whenever locking a
btree node.
This patch plumbs a btree_trans to the few places that need it, and adds
two new ... | ```diff
diff --git a/fs/bcachefs/btree_cache.c b/fs/bcachefs/btree_cache.c
index e09fbf36ebc2..a0e9e14e3fa5 100644
--- a/fs/bcachefs/btree_cache.c
+++ b/fs/bcachefs/btree_cache.c
@@ -959,12 +959,13 @@ lock_node:
return b;
}
-struct btree *bch2_btree_node_get_noiter(struct bch_fs *c,
+struct btree *bch2_btree_node_... |
546180874ade7225676bc0cd5ea4e2388e2374bc | Analyze and fix the following issue in the Linux kernel code: bcachefs: Mark write locks before taking lock | Problem Details:
six locks are unfair: while a thread is blocked trying to take a write
lock, new read locks will fail. The new deadlock cycle detector makes
use of our existing lock tracing, so we need to tell it we're holding a
write lock before we take the lock for it to work correctly.
Signed-off-by: Kent Overstre... | ```diff
diff --git a/fs/bcachefs/btree_locking.h b/fs/bcachefs/btree_locking.h
index 1e4c81d8084b..ab3161c1b1f4 100644
--- a/fs/bcachefs/btree_locking.h
+++ b/fs/bcachefs/btree_locking.h
@@ -271,10 +271,15 @@ static inline void bch2_btree_node_lock_write(struct btree_trans *trans,
EBUG_ON(path->l[b->c.level].lock_seq... |
c919f53f3bcba3598fc6ce1ee5c5aed75d0834b7 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Don't leak lock pcpu counts memory | Problem Details:
This fixes a small memory leak.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev> | ```diff
diff --git a/fs/bcachefs/btree_key_cache.c b/fs/bcachefs/btree_key_cache.c
index 127cb6edaff5..7349c70f8445 100644
--- a/fs/bcachefs/btree_key_cache.c
+++ b/fs/bcachefs/btree_key_cache.c
@@ -238,8 +238,6 @@ btree_key_cache_create(struct bch_fs *c,
} else {
if (btree_id == BTREE_ID_subvolumes)
six_lock_... |
d97e6aaed60a9c2c727cce2979ca311fe232163f | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix bch2_btree_update_start() to return -BCH_ERR_journal_reclaim_would_deadlock | Problem Details:
On failure to get a journal pre-reservation because we're called from
journal reclaim we're not supposed to return a transaction restart error
- this fixes a livelock.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev> | ```diff
diff --git a/fs/bcachefs/btree_update_interior.c b/fs/bcachefs/btree_update_interior.c
index fc768195be54..dd9405c631f5 100644
--- a/fs/bcachefs/btree_update_interior.c
+++ b/fs/bcachefs/btree_update_interior.c
@@ -1049,6 +1049,11 @@ bch2_btree_update_start(struct btree_trans *trans, struct btree_path *path,
... |
c240c3a94427346f27a7ff48f02cbe03f2c2ebd6 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Print lock counts in debugs btree_transactions | Problem Details:
Improve our debugfs output, to help in debugging deadlocks: this shows,
for every btree node we print, the current number of readers/intent
locks/write locks held.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev> | ```diff
diff --git a/fs/bcachefs/btree_iter.c b/fs/bcachefs/btree_iter.c
index ce9437916cf2..99422e29c704 100644
--- a/fs/bcachefs/btree_iter.c
+++ b/fs/bcachefs/btree_iter.c
@@ -3024,9 +3024,21 @@ bch2_btree_path_node_to_text(struct printbuf *out,
struct btree_bkey_cached_common *b,
bool cached)
{
+... |
02afcb8c26b14ae317754d8c79339f41b3dfeaae | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix adding a device with a label | Problem Details:
Device labels are represented as pointers in the member info section: we
need to get and then set the label for it to be kept correctly.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/disk_groups.c b/fs/bcachefs/disk_groups.c
index 33d2702e6849..5f405d38b3de 100644
--- a/fs/bcachefs/disk_groups.c
+++ b/fs/bcachefs/disk_groups.c
@@ -384,32 +384,34 @@ inval:
prt_printf(out, "invalid label %u", v);
}
-int bch2_dev_group_set(struct bch_fs *c, struct bch_dev *ca, con... |
12043cf1511420ecf38f4925a0089c1ae1aa058b | Analyze and fix the following issue in the Linux kernel code: bcachefs: fsck: Another transaction restart handling fix | Problem Details:
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev> | ```diff
diff --git a/fs/bcachefs/data_update.c b/fs/bcachefs/data_update.c
index c0d6a48d3c72..f9eb147fe229 100644
--- a/fs/bcachefs/data_update.c
+++ b/fs/bcachefs/data_update.c
@@ -248,6 +248,8 @@ next:
}
continue;
nomatch:
+ trace_data_update_fail(&old.k->p);
+
if (m->ctxt) {
BUG_ON(k.k->p.offset <= i... |
42590b53fef427f96fc50da4974923564e9033cd | Analyze and fix the following issue in the Linux kernel code: bcachefs: bch2_btree_delete_range_trans() now returns -BCH_ERR_transaction_restart_nested | Problem Details:
The new convention is that functions that handle transaction restarts
within an existing transaction context should return
-BCH_ERR_transaction_restart_nested when they did so, since they
invalidated the outer transaction context.
This also means bch2_btree_delete_range_trans() is changed to only call... | ```diff
diff --git a/fs/bcachefs/btree_update_leaf.c b/fs/bcachefs/btree_update_leaf.c
index 6bf34853f261..11c3767896aa 100644
--- a/fs/bcachefs/btree_update_leaf.c
+++ b/fs/bcachefs/btree_update_leaf.c
@@ -1713,15 +1713,16 @@ int bch2_btree_delete_range_trans(struct btree_trans *trans, enum btree_id id,
unsigne... |
efd0d03816dd0d5127217220f85f1f0e621974d1 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Minor transaction restart handling fix | Problem Details:
- fsck_inode_rm() wasn't returning BCH_ERR_transaction_restart_nested
- change bch2_trans_verify_not_restarted() to call panic() - we don't
want these errors to be missed
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/btree_iter.c b/fs/bcachefs/btree_iter.c
index 88b2ed1d508a..95bc71dd87af 100644
--- a/fs/bcachefs/btree_iter.c
+++ b/fs/bcachefs/btree_iter.c
@@ -3289,10 +3289,10 @@ u32 bch2_trans_begin(struct btree_trans *trans)
void bch2_trans_verify_not_restarted(struct btree_trans *trans, u32 re... |
23dfb3a2f7757593249b745f2a17b56b756d2874 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix bch2_btree_iter_peek_slot() error path | Problem Details:
iter->k needs to be consistent with iter->pos - required for
bch2_btree_iter_(rewind|advance) to work correctly.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/btree_iter.c b/fs/bcachefs/btree_iter.c
index 0c6e215f46c8..88b2ed1d508a 100644
--- a/fs/bcachefs/btree_iter.c
+++ b/fs/bcachefs/btree_iter.c
@@ -2844,8 +2844,10 @@ struct bkey_s_c bch2_btree_iter_peek_slot(struct btree_iter *iter)
iter->flags & BTREE_ITER_INTENT);
ret = bch2_... |
bbf4288401519a7554201caf9b945c79f29753b3 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Always rebuild aux search trees when node boundaries change | Problem Details:
Topology repair may change btree node min/max keys: when it does so, we
need to always rebuild eytzinger search trees because nodes directly
depend on those values.
This fixes a bug found by the 'kill_btree_node' test, where we'd pop an
assertion in bch2_bset_search_linear().
Signed-off-by: Kent Over... | ```diff
diff --git a/fs/bcachefs/btree_io.c b/fs/bcachefs/btree_io.c
index 6e39c1641b90..bd74bd31dd1f 100644
--- a/fs/bcachefs/btree_io.c
+++ b/fs/bcachefs/btree_io.c
@@ -611,7 +611,6 @@ void bch2_btree_node_drop_keys_outside_node(struct btree *b)
(u64 *) vstruct_end(i) - (u64 *) k);
i->u64s = cpu_to_le16(l... |
15bc0948e73d9a858a6b69fc4eb34d176436044c | Analyze and fix the following issue in the Linux kernel code: bcachefs: Add an overflow check in set_bkey_val_u64s() | Problem Details:
For now this is just a BUG_ON() - we may want to change this to return
an error in the future.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/bkey.h b/fs/bcachefs/bkey.h
index 6a637a408a9f..8e9405f89537 100644
--- a/fs/bcachefs/bkey.h
+++ b/fs/bcachefs/bkey.h
@@ -50,12 +50,15 @@ static inline size_t bkey_val_bytes(const struct bkey *k)
static inline void set_bkey_val_u64s(struct bkey *k, unsigned val_u64s)
{
- k->u64s = B... |
c59d66b51b11064f17d87d5b9695e0216b8ade99 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix bch2_fs_check_snapshots() | Problem Details:
We were iterating starting at BCACHEFS_ROOT_INO, but snapshots start at
POS_MIN - meaning this code was never getting run.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
Reported-by: Olexa Bilaniuk <obilaniu@gmail.com> | ```diff
diff --git a/fs/bcachefs/subvolume.c b/fs/bcachefs/subvolume.c
index 24244bc3d2fb..fb3f8e4074c7 100644
--- a/fs/bcachefs/subvolume.c
+++ b/fs/bcachefs/subvolume.c
@@ -278,8 +278,8 @@ int bch2_fs_check_snapshots(struct bch_fs *c)
bch2_trans_init(&trans, c, 0, 0);
- ret = for_each_btree_key_commit(&trans, i... |
45b033fa1afd35a8eab0af003ffac9413548f476 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix assertion in bch2_btree_key_cache_drop() | Problem Details:
Turns out this assertion was something we could legitimately hit - add a
comment describing what's going on, and handle it.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/btree_key_cache.c b/fs/bcachefs/btree_key_cache.c
index f0055f17381d..0e87c19effeb 100644
--- a/fs/bcachefs/btree_key_cache.c
+++ b/fs/bcachefs/btree_key_cache.c
@@ -639,11 +639,22 @@ bool bch2_btree_insert_key_cached(struct btree_trans *trans,
void bch2_btree_key_cache_drop(struct btr... |
5c0bb66ae341c71e5f62c193ea4d7b0cf278a914 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Track the maximum btree_paths ever allocated by each transaction | Problem Details:
We need a way to check if the machinery for handling btree_paths with in
a transaction is behaving reasonably, as it often has not been - we've
had bugs with transaction path overflows caused by duplicate paths and
plenty of other things.
This patch tracks, per transaction fn, the most btree paths eve... | ```diff
diff --git a/fs/bcachefs/bcachefs.h b/fs/bcachefs/bcachefs.h
index 9fe96516c114..f8b7434534eb 100644
--- a/fs/bcachefs/bcachefs.h
+++ b/fs/bcachefs/bcachefs.h
@@ -531,6 +531,9 @@ struct btree_debug {
struct btree_transaction_stats {
struct bch2_time_stats lock_hold_times;
+ struct mutex lock;
+ unsigned ... |
11c1a62f3b872d2345c97e72700ed4d1b2511888 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Switch bch2_btree_delete_range() to bch2_trans_run() | Problem Details:
This fixes an assertion about unexpected transaction restarts -
bch2_delete_range_trans() handles transaction restarts.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/btree_update_leaf.c b/fs/bcachefs/btree_update_leaf.c
index 1371b7c6ff8b..6bf34853f261 100644
--- a/fs/bcachefs/btree_update_leaf.c
+++ b/fs/bcachefs/btree_update_leaf.c
@@ -1785,9 +1785,8 @@ int bch2_btree_delete_range(struct bch_fs *c, enum btree_id id,
unsigned update_flags,
... |
7c812ab786c4a689989aabf9e865164eb4f8004d | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix btree_path->uptodate inconsistency | Problem Details:
This fixes an assertion in bch2_btree_path_peek_slot().
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/btree_update_leaf.c b/fs/bcachefs/btree_update_leaf.c
index 2d824f6a6fab..1371b7c6ff8b 100644
--- a/fs/bcachefs/btree_update_leaf.c
+++ b/fs/bcachefs/btree_update_leaf.c
@@ -726,8 +726,10 @@ bch2_trans_commit_write_locked(struct btree_trans *trans,
btree_insert_key_leaf(trans, i);
... |
a300261ad19d69e080278ec4950d39caef3ffbf1 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix duplicate paths left by bch2_path_put() | Problem Details:
bch2_path_put() is supposed to drop paths that aren't needed on
transaction restart, or to hold locks that we're supposed to keep until
transaction commit: but it was failing to free paths in some cases that
it should have, leading to transaction path overflows with lots of
duplicate paths.
Signed-off... | ```diff
diff --git a/fs/bcachefs/btree_iter.c b/fs/bcachefs/btree_iter.c
index bd6e35fcbdbd..837cdcc5c77c 100644
--- a/fs/bcachefs/btree_iter.c
+++ b/fs/bcachefs/btree_iter.c
@@ -424,13 +424,17 @@ int bch2_btree_path_relock_intent(struct btree_trans *trans,
return 0;
}
-noinline __flatten
-static int __bch2_btree_... |
fd211bc71c9b4093ed39d4fc93294f9ff423febc | Analyze and fix the following issue in the Linux kernel code: bcachefs: Don't set should_be_locked on paths that aren't locked | Problem Details:
It doesn't make any sense to set should_be_locked on btree_paths that
aren't locked, and is often a bug - this patch adds assertions and fixes
some of those bugs.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/btree_iter.c b/fs/bcachefs/btree_iter.c
index 74978a50023f..1d5d7b639241 100644
--- a/fs/bcachefs/btree_iter.c
+++ b/fs/bcachefs/btree_iter.c
@@ -2103,7 +2103,7 @@ bch2_btree_iter_traverse(struct btree_iter *iter)
if (ret)
return ret;
- iter->path->should_be_locked = true;
+ btre... |
31301dd46975b2423fd38fc64bc58728d89dbcac | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix missing error handling in bch2_subvolume_delete() | Problem Details:
This fixes an assertion when the transaction has been unexpectedly
restarted.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/subvolume.c b/fs/bcachefs/subvolume.c
index b5b0f5e39f97..24244bc3d2fb 100644
--- a/fs/bcachefs/subvolume.c
+++ b/fs/bcachefs/subvolume.c
@@ -877,6 +877,8 @@ int bch2_subvolume_delete(struct btree_trans *trans, u32 subvolid)
goto err;
ret = bch2_snapshot_node_set_deleted(trans, s... |
17047fbced563cf5abe5aa546f6a92af48900b69 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix incorrectly freeing btree_path in alloc path | Problem Details:
Clearing path->preserve means the path will be dropping in
bch2_trans_begin() - but on transaction restart, we're likely to need
that path again.
This fixes a livelock in the allocation path.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/btree_iter.h b/fs/bcachefs/btree_iter.h
index 9a3287da9a12..5ca92b6bb397 100644
--- a/fs/bcachefs/btree_iter.h
+++ b/fs/bcachefs/btree_iter.h
@@ -335,7 +335,8 @@ void bch2_trans_copy_iter(struct btree_iter *, struct btree_iter *);
static inline void set_btree_iter_dontneed(struct btr... |
86b74451931790eafa018021fe900faea3230189 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix bch2_btree_trans_to_text() | Problem Details:
bch2_btree_trans_to_text() is used to print btree_transactions owned by
other threads; thus, it needs to be particularly careful. This fixes a
null ptr deref caused by racing with the owning thread changing
path->l[].b.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/btree_iter.c b/fs/bcachefs/btree_iter.c
index 01ca27ee6314..171894d9347d 100644
--- a/fs/bcachefs/btree_iter.c
+++ b/fs/bcachefs/btree_iter.c
@@ -3417,19 +3417,19 @@ void bch2_trans_exit(struct btree_trans *trans)
static void __maybe_unused
bch2_btree_path_node_to_text(struct printb... |
4a7f7e9e4de665d94e1a7610bf5506265df414d3 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix not punting to worqueue when promoting | Problem Details:
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/io.c b/fs/bcachefs/io.c
index dfa708c0a7fc..0348e2ab6422 100644
--- a/fs/bcachefs/io.c
+++ b/fs/bcachefs/io.c
@@ -2027,6 +2027,7 @@ static void bch2_read_endio(struct bio *bio)
}
if (rbio->narrow_crcs ||
+ rbio->promote ||
crc_is_compressed(rbio->pick.crc) ||
bch2_... |
0763c552e7ef024ce6dbc9cbc828b8715dff251c | Analyze and fix the following issue in the Linux kernel code: bcachefs: fsck: Fix nested transaction handling | Problem Details:
This uses the new trans->restart count to make sure we always correctly
return -BCH_ERR_transaction_restart_nested when we restart a nested
transaction - eliminating some other hacks and preparing for new
assertions.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/fsck.c b/fs/bcachefs/fsck.c
index ea264421fe8f..bb8cab7cb405 100644
--- a/fs/bcachefs/fsck.c
+++ b/fs/bcachefs/fsck.c
@@ -650,6 +650,7 @@ static int __walk_inode(struct btree_trans *trans,
struct bch_fs *c = trans->c;
struct btree_iter iter;
struct bkey_s_c k;
+ u32 restart_count ... |
7903e3d2d7e5266dde40c9db746fafb078f84f4f | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix check_i_sectors() | Problem Details:
bch2_count_inode_sectors() uses for_each_btree_key() internally, which
handles lock restarts - the lockrestart_do() in check_i_sectors() is
redundant, and buggy here since the count that
bch2_count_inode_sectors() returns was interpreted as an error by
lockrestart_do().
Signed-off-by: Kent Overstreet ... | ```diff
diff --git a/fs/bcachefs/fsck.c b/fs/bcachefs/fsck.c
index 306983811c1b..ea264421fe8f 100644
--- a/fs/bcachefs/fsck.c
+++ b/fs/bcachefs/fsck.c
@@ -1122,8 +1122,7 @@ static int check_i_sectors(struct btree_trans *trans, struct inode_walker *w)
if (i->inode.bi_sectors == i->count)
continue;
- count2 = l... |
d7228ecc483e1e104179a24a7e1e4568ae9235cb | Analyze and fix the following issue in the Linux kernel code: bcachefs: Convert debugfs code to for_each_btree_key2() | Problem Details:
This fixes a bug where we were leaking a transaction restart error to
userspace.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/debug.c b/fs/bcachefs/debug.c
index 45f5229f20eb..b9b6cad8cd40 100644
--- a/fs/bcachefs/debug.c
+++ b/fs/bcachefs/debug.c
@@ -188,6 +188,7 @@ struct dump_iter {
struct bch_fs *c;
enum btree_id id;
struct bpos from;
+ struct bpos prev_node;
u64 iter;
struct printbuf bu... |
a0cb8d784f309d22323974e47b103bf01d0b62c0 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Inject transaction restarts in debug mode | Problem Details:
In CONFIG_BCACHEFS_DEBUG mode, we'll now randomly issue transaction
restarts - with a decaying probability based on the number of restarts
we've already had, to ensure that transactions eventually make forward
progress. This should help shake out some bugs.
Signed-off-by: Kent Overstreet <kent.overstr... | ```diff
diff --git a/fs/bcachefs/btree_iter.c b/fs/bcachefs/btree_iter.c
index db247c96298f..777e41a12246 100644
--- a/fs/bcachefs/btree_iter.c
+++ b/fs/bcachefs/btree_iter.c
@@ -17,6 +17,7 @@
#include "subvolume.h"
#include "trace.h"
+#include <linux/random.h>
#include <linux/prefetch.h>
static inline void btr... |
90cecb921cfe95858a32995019f11c20b6339607 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Prevent a btree iter overflow in alloc path | Problem Details:
In bch2_bucket_alloc_trans(), we're iterating over buckets - but not
directly with an iterator, since we're iterating over the freespace
btree.
This means that we need to clear iter->path->preserve, otherwise we'll
end up retaining a btree_path for every alloc key we touched - which is
not what we wan... | ```diff
diff --git a/fs/bcachefs/alloc_foreground.c b/fs/bcachefs/alloc_foreground.c
index 39e3bb5205ca..a9f893361c73 100644
--- a/fs/bcachefs/alloc_foreground.c
+++ b/fs/bcachefs/alloc_foreground.c
@@ -351,6 +351,7 @@ static struct open_bucket *try_alloc_bucket(struct btree_trans *trans, struct bc
if (!ob)
iter.p... |
b962552eabd59f0026dcc21c14775b9d78336baf | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix should_invalidate_buckets() | Problem Details:
Like bch2_copygc_wait_amount, should_invalidate_buckets() needs to try
to ensure that there are always more buckets free than the largest
reserve.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/alloc_background.h b/fs/bcachefs/alloc_background.h
index ff366e61ace5..488db3211ce4 100644
--- a/fs/bcachefs/alloc_background.h
+++ b/fs/bcachefs/alloc_background.h
@@ -134,11 +134,13 @@ void bch2_do_discards(struct bch_fs *);
static inline u64 should_invalidate_buckets(struct bch_dev... |
0d06b4eca687b3a6a07b62fd4ca83d635103c77f | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix repair for extent past end of inode | Problem Details:
When we find an extent past an inode's i_size, we need to do the
deletion in the inode's snapshot (which will emit a whiteout if
necessary); and we also need to note that we now have an a key at that
position and snapshot, so that we don't go into an infinite loop.
Also, switch to walking inodes in re... | ```diff
diff --git a/fs/bcachefs/fsck.c b/fs/bcachefs/fsck.c
index b401c0913bdc..5cec55edb483 100644
--- a/fs/bcachefs/fsck.c
+++ b/fs/bcachefs/fsck.c
@@ -489,6 +489,28 @@ static inline void snapshots_seen_init(struct snapshots_seen *s)
memset(s, 0, sizeof(*s));
}
+static int snapshots_seen_add(struct bch_fs *c, s... |
4ab35c34d5ab258fdd7325315fe5d94699e51eb4 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix subvol/snapshot deleting in recovery | Problem Details:
fsck doesn't want to run while we're cleaning up deleted snapshots - if
that work needs to be done, we want it to have finished before fsck
runs, otherwise fsck will get confused when it finds multiple keys in
the same snapshot ID equivalence class (i.e. the mechanism that
snapshot deletion uses for cl... | ```diff
diff --git a/fs/bcachefs/bcachefs.h b/fs/bcachefs/bcachefs.h
index c07ea9af561d..7020eee5de21 100644
--- a/fs/bcachefs/bcachefs.h
+++ b/fs/bcachefs/bcachefs.h
@@ -514,6 +514,8 @@ enum {
BCH_FS_INITIAL_GC_UNFIXED, /* kill when we enumerate fsck errors */
BCH_FS_NEED_ANOTHER_GC,
+ BCH_FS_HAVE_DELETED_SNAPSH... |
416cc426c0d79c65d85de52d3548a32de06ab3e8 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix snapshot deletion | Problem Details:
Snapshots being deleted won't in general have a corresponding subvolume:
this fixes a spurious fsck error where we'd complain about a snapshot
pointing to a missing subvolume - but the subvolume had been deleted,
and the snapshot was pending deletion as well.
Signed-off-by: Kent Overstreet <kent.overs... | ```diff
diff --git a/fs/bcachefs/subvolume.c b/fs/bcachefs/subvolume.c
index d74dc9843028..083eb4324583 100644
--- a/fs/bcachefs/subvolume.c
+++ b/fs/bcachefs/subvolume.c
@@ -17,7 +17,7 @@ void bch2_snapshot_to_text(struct printbuf *out, struct bch_fs *c,
{
struct bkey_s_c_snapshot s = bkey_s_c_to_snapshot(k);
- p... |
7c0732b88dfb87b5f32dcf5fdd7984d2acb992b9 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix move path when move_stats == NULL | Problem Details:
This isn't done very often, but it is legitimate
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/move.c b/fs/bcachefs/move.c
index eae93c65e1c7..8b44d95c32ce 100644
--- a/fs/bcachefs/move.c
+++ b/fs/bcachefs/move.c
@@ -140,13 +140,15 @@ void bch2_moving_ctxt_exit(struct moving_context *ctxt)
{
move_ctxt_wait_event(ctxt, NULL, list_empty(&ctxt->reads));
closure_sync(&ctxt->cl);... |
47ab0c5f6a1e6ac1e9387181585fb39393fec4ea | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix bch2_check_alloc_key() | Problem Details:
bch2_check_alloc_key() was failing to check buckets that didn't have
alloc keys yet (because they'd never been used) - they still need to be
added to the freespace btree.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/alloc_background.c b/fs/bcachefs/alloc_background.c
index 9ede98a3dc64..f4457d62d75e 100644
--- a/fs/bcachefs/alloc_background.c
+++ b/fs/bcachefs/alloc_background.c
@@ -618,7 +618,9 @@ static int bch2_check_alloc_key(struct btree_trans *trans,
struct printbuf buf = PRINTBUF;
int re... |
7a47d0993be95bf68e2e04653f40d311e3c25bed | Analyze and fix the following issue in the Linux kernel code: bcachefs: Always descend to leaf nodes it btree_gc | Problem Details:
If a btree node is unreadable, it's the topology repair that fixes that
and it's kicked off by btree_gc, so btree_gc needs to touch every node
and very that they can be read.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/btree_gc.c b/fs/bcachefs/btree_gc.c
index 2de5d97a9d2c..0447f5a51b5e 100644
--- a/fs/bcachefs/btree_gc.c
+++ b/fs/bcachefs/btree_gc.c
@@ -851,10 +851,7 @@ static int bch2_gc_btree(struct btree_trans *trans, enum btree_id btree_id,
struct bch_fs *c = trans->c;
struct btree_iter iter;... |
58aaa0836be3be279f17db9ba07f3a58da033a9f | Analyze and fix the following issue in the Linux kernel code: bcachefs: fix __dev_available(). | Problem Details:
__dev_available() now calculates available buckets correctly. Previously
it would almost always return 0 when we have cached data.
Signed-off-by: Daniel Hill <daniel@gluo.nz>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev> | ```diff
diff --git a/fs/bcachefs/buckets.h b/fs/bcachefs/buckets.h
index 080bcb20a5b0..b4cf10a47c52 100644
--- a/fs/bcachefs/buckets.h
+++ b/fs/bcachefs/buckets.h
@@ -159,12 +159,12 @@ static inline u64 __dev_buckets_available(struct bch_dev *ca,
enum alloc_reserve reserve)
{
return max_t(s64, 0,
- usa... |
2817d453819a10654ac72e5f4937d4653dea1648 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix assertion in topology repair | Problem Details:
If we were at the end of the node, when breaking out of the loop we'd
pop the assertion on line 446 when cur wasn't NULL.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/btree_gc.c b/fs/bcachefs/btree_gc.c
index 123644ffe93c..2de5d97a9d2c 100644
--- a/fs/bcachefs/btree_gc.c
+++ b/fs/bcachefs/btree_gc.c
@@ -395,6 +395,7 @@ again:
bch2_btree_node_evict(c, cur_k.k);
ret = bch2_journal_key_delete(c, b->c.btree_id,
b->c.level, cur_k.k->k... |
5a3c24714c8bd5e0d01d1547c0848147f2f79f69 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Make verbose option settable at runtime | Problem Details:
-o verbose is very useful, and we're starting to use it more for runtime
debug statements - making it possible to enable at runtime is a no
brainer.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/opts.h b/fs/bcachefs/opts.h
index 43b4488b8c6f..30cf9a2d9dc1 100644
--- a/fs/bcachefs/opts.h
+++ b/fs/bcachefs/opts.h
@@ -269,7 +269,7 @@ enum opt_type {
BCH2_NO_SB_OPT, true, \
NULL, "Enable discard/TRIM support") \
x(verbose, u8, \
- OPT_FS|OPT_MOUNT, \
+ ... |
1c6ff39445553f84f9a46c2c60a4768c7f4ef226 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix refcount leak in bch2_do_invalidates() | Problem Details:
If we fail to queue the work item because it's already in process, we
need to drop the ref we just took.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/alloc_background.c b/fs/bcachefs/alloc_background.c
index 3084081966b6..933334fed960 100644
--- a/fs/bcachefs/alloc_background.c
+++ b/fs/bcachefs/alloc_background.c
@@ -1192,8 +1192,9 @@ static void bch2_do_invalidates_work(struct work_struct *work)
void bch2_do_invalidates(struct b... |
2ed6248ab3a58dbbe5819cbd0e60e4e3a5b72c47 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix assertion in bch2_dev_list_add_dev() | Problem Details:
We were only allowing 4 devices in a dev_list, not 16.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/super.h b/fs/bcachefs/super.h
index 359fa1e7fc18..d66de6f589ac 100644
--- a/fs/bcachefs/super.h
+++ b/fs/bcachefs/super.h
@@ -89,7 +89,7 @@ static inline void bch2_dev_list_add_dev(struct bch_devs_list *devs,
unsigned dev)
{
BUG_ON(bch2_dev_list_has_dev(*devs, dev));
- BUG_ON(... |
c9bd67321e9b9bae0a9ba151a3906086878159b6 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix btree node read retries | Problem Details:
b->written wasn't being reset to 0 in the btree node read retry path,
causing decrypting & validation of previously read bsets to not be
re-run - ouch.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/btree_io.c b/fs/bcachefs/btree_io.c
index 598c30b7ab8b..b7441677dc33 100644
--- a/fs/bcachefs/btree_io.c
+++ b/fs/bcachefs/btree_io.c
@@ -879,6 +879,8 @@ int bch2_btree_node_read_done(struct bch_fs *c, struct bch_dev *ca,
int ret, retry_read = 0, write = READ;
b->version_ondisk = ... |
a5d18f9ec0e2cfce1e0dc50f07d614ba21edaae9 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Improved human readable integer parsing | Problem Details:
Printbufs recently switched to using string_get_size() for printing
integers in human readable units. This updates __bch2_strtoh() to parse
numbers printed by string_get_size() - we now have to handle floating
point numbers, and new unit suffixes.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.... | ```diff
diff --git a/fs/bcachefs/util.c b/fs/bcachefs/util.c
index dac7dfd0b806..a2e42ae77371 100644
--- a/fs/bcachefs/util.c
+++ b/fs/bcachefs/util.c
@@ -27,16 +27,19 @@
static const char si_units[] = "?kMGTPEZY";
-static int __bch2_strtoh(const char *cp, u64 *res,
- u64 t_max, bool t_signed)
+/* string_get_si... |
df8c2ccb9309f342724f3104fd9d68d7e4ce65da | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix freespace initialization | Problem Details:
bch2_dev_freespace_init() was using __bch2_trans_do() incorrectly, and
calling bch2_bucket_do_index() with a stale alloc key.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/alloc_background.c b/fs/bcachefs/alloc_background.c
index bffbddbdacea..982cd185b1af 100644
--- a/fs/bcachefs/alloc_background.c
+++ b/fs/bcachefs/alloc_background.c
@@ -1190,12 +1190,26 @@ void bch2_do_invalidates(struct bch_fs *c)
queue_work(system_long_wq, &c->invalidate_work);
}... |
652018d66190412669a898c2dc3e75073eac8679 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix btree node read error path | Problem Details:
We were forgetting to clear the read_in_flight flag - oops. This also
fixes it to not call bch2_fatal_error() before topology repair has had a
chance to do its thing.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/btree_io.c b/fs/bcachefs/btree_io.c
index a1043492dbd9..9d4d6a65e70c 100644
--- a/fs/bcachefs/btree_io.c
+++ b/fs/bcachefs/btree_io.c
@@ -1483,23 +1483,32 @@ void bch2_btree_node_read(struct bch_fs *c, struct btree *b,
struct btree_read_bio *rb;
struct bch_dev *ca;
struct bio *bio... |
576179021c90bea808ac12c491bd9b239ca80c2e | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix btree_and_journal_iter | Problem Details:
We had a bug where btree_and_journal_iter would return the same key
twice - after deleting it (perhaps because it was present in both the
btree and the journal?)
This reworks btree_and_journal_iter to track the current position, much
like btree_paths, which makes the logic considerably simpler and mor... | ```diff
diff --git a/fs/bcachefs/recovery.c b/fs/bcachefs/recovery.c
index 5fe7595d36be..d755da42d6c5 100644
--- a/fs/bcachefs/recovery.c
+++ b/fs/bcachefs/recovery.c
@@ -159,21 +159,17 @@ static void journal_iters_fix(struct bch_fs *c)
{
struct journal_keys *keys = &c->journal_keys;
/* The key we just inserted is... |
f2aa02657561b0e6d96089eb8ee44e4154f4acad | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix for cmd_list_journal | Problem Details:
cmd_list_journal wasn't correctly listing the most recent journal
entries as blacklisted - because in the recovery path when just reading
the journal, we were failing to add those to the blacklist table.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/recovery.c b/fs/bcachefs/recovery.c
index edb04f65a148..5fe7595d36be 100644
--- a/fs/bcachefs/recovery.c
+++ b/fs/bcachefs/recovery.c
@@ -1175,9 +1175,6 @@ use_clean:
blacklist_seq = journal_seq = le64_to_cpu(clean->journal_seq) + 1;
}
- if (c->opts.read_journal_only)
- goto out... |
636d4eef1eefe447deef134bdf8e34c979ff009e | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix memory corruption in encryption path | Problem Details:
When do_encrypt() was passed a vmalloc address and the buffer spanned
more than a single page, we were encrypting/decrypting completely
different pages than the ones intended.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/checksum.c b/fs/bcachefs/checksum.c
index 317efd047a46..e9a444f75b93 100644
--- a/fs/bcachefs/checksum.c
+++ b/fs/bcachefs/checksum.c
@@ -114,15 +114,41 @@ static inline int do_encrypt(struct crypto_sync_skcipher *tfm,
struct nonce nonce,
void *buf, size_t len)
{
- ... |
9b688da35072910ec205697d2f51226cd3fd9f90 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix error checking in bch2_fs_alloc() | Problem Details:
One of the init calls had a ; instead of a ?:, and errors after that got
dropped - oops.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/super.c b/fs/bcachefs/super.c
index 159d47d129a2..8d02a8158520 100644
--- a/fs/bcachefs/super.c
+++ b/fs/bcachefs/super.c
@@ -780,7 +780,7 @@ static struct bch_fs *bch2_fs_alloc(struct bch_sb *sb, struct bch_opts opts)
bch2_fs_btree_key_cache_init(&c->btree_key_cache) ?:
bch... |
30525f68633740e071c0960c11c4380f1f6851af | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix journal_keys_search() overhead | Problem Details:
Previously, on every btree_iter_peek() operation we were searching the
journal keys, doing a full binary search - which was slow.
This patch fixes that by saving our position in the journal keys, so
that we only do a full binary search when moving our position backwards
or a large jump forwards.
Sign... | ```diff
diff --git a/fs/bcachefs/btree_iter.c b/fs/bcachefs/btree_iter.c
index b840035dca55..3ce0571651b5 100644
--- a/fs/bcachefs/btree_iter.c
+++ b/fs/bcachefs/btree_iter.c
@@ -2242,13 +2242,30 @@ static inline struct bkey_i *btree_trans_peek_updates(struct btree_iter *iter)
: NULL;
}
+struct bkey_i *bch2_btree... |
c346def9af1d3890ee604905fb08d689e8383855 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix encryption path on arm | Problem Details:
flush_dcache_page() is not a noop on arm, but we were using
virt_to_page() instead of vmalloc_to_page() for an address on the kernel
stack - vmalloc memory, leading to an oops in flush_dcache_page().
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/checksum.c b/fs/bcachefs/checksum.c
index 50157b4013a5..317efd047a46 100644
--- a/fs/bcachefs/checksum.c
+++ b/fs/bcachefs/checksum.c
@@ -116,7 +116,12 @@ static inline int do_encrypt(struct crypto_sync_skcipher *tfm,
{
struct scatterlist sg;
- sg_init_one(&sg, buf, len);
+ sg_init... |
e320b42dfeb5797a618edbba5186060b3907ba89 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix extent merging | Problem Details:
When merging extents, we have to check that we won't overflow size
fields in any CRC entries - but the check for this was wrong, because in
the loop it was in we weren't keeping a pointer to the (packed, encoded)
CRC field.
Fix this by moving it to its own loop.
Signed-off-by: Kent Overstreet <kent.o... | ```diff
diff --git a/fs/bcachefs/extents.c b/fs/bcachefs/extents.c
index c56925d94bfe..d8f429ffe57c 100644
--- a/fs/bcachefs/extents.c
+++ b/fs/bcachefs/extents.c
@@ -307,8 +307,20 @@ bool bch2_extent_merge(struct bch_fs *c, struct bkey_s l, struct bkey_s_c r)
lp.crc.uncompressed_size +
rp.crc.uncompres... |
099989c1b230e0f36ee7146d1df948822c999f6a | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix journal_iters_fix() | Problem Details:
journal_iters_fix() was incorrectly rewinding iterators past keys they
had already returned, leading to those keys being double counted in the
bch2_gc() path - oops.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/recovery.c b/fs/bcachefs/recovery.c
index 5831ab53a982..f54859b49416 100644
--- a/fs/bcachefs/recovery.c
+++ b/fs/bcachefs/recovery.c
@@ -147,7 +147,7 @@ static void journal_iters_fix(struct bch_fs *c)
/*
* If an iterator points one after the key we just inserted,
- * and the ke... |
fd4cecd2583e784ff28b851de2ff5046201c57d7 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Lock ordering fix | Problem Details:
Can't take btree node locks while holding btree_reserve_cache_lock - it
would be nice if we could check this with lockdep.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/btree_update_interior.c b/fs/bcachefs/btree_update_interior.c
index 27ab1cde2217..9696eb2b91e7 100644
--- a/fs/bcachefs/btree_update_interior.c
+++ b/fs/bcachefs/btree_update_interior.c
@@ -383,16 +383,13 @@ static void bch2_btree_reserve_put(struct btree_update *as)
struct bch_fs *c ... |
d8f31407c842331a13c48404bc030f49d60f25aa | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix hash_check_key() | Problem Details:
hash_check_key() was incorrectly handling transaction restarts - switch
it to for_each_btree_key_norestart().
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/fsck.c b/fs/bcachefs/fsck.c
index d507b9fdd32e..f1abec95a740 100644
--- a/fs/bcachefs/fsck.c
+++ b/fs/bcachefs/fsck.c
@@ -740,8 +740,9 @@ static int hash_check_key(struct btree_trans *trans,
if (hash_k.k->p.offset < hash)
goto bad_hash;
- for_each_btree_key(trans, iter, desc.btre... |
7419646b254a599d7881e7815fc5e61b18c94b7e | Analyze and fix the following issue in the Linux kernel code: bcachefs: btree_update_interior.c prep for backpointers | Problem Details:
Previously, btree_update_interior.c passed keys to bch2_trans_mark_*
that hadn't been fully initialized - they didn't have the key field
filled out, just the value.
With backpointers, we need to make sure keys are fully initialized
before marking them - because the backpointer points back to the
origi... | ```diff
diff --git a/fs/bcachefs/btree_update_interior.c b/fs/bcachefs/btree_update_interior.c
index 29de7afa9616..7a092d852930 100644
--- a/fs/bcachefs/btree_update_interior.c
+++ b/fs/bcachefs/btree_update_interior.c
@@ -509,18 +509,6 @@ static void btree_update_add_key(struct btree_update *as,
bch2_keylist_push(ke... |
c609947b5eae4bee096fde660a53a719ee65e191 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix for getting stuck in journal replay | Problem Details:
In journal replay, we weren't immediately dropping journal pins when we
start doing updates that ewern't from journal replay - leading to
journal reclaim getting stuck.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/recovery.c b/fs/bcachefs/recovery.c
index 16ba5d24a86d..dc11eae1bcaa 100644
--- a/fs/bcachefs/recovery.c
+++ b/fs/bcachefs/recovery.c
@@ -556,7 +556,8 @@ static int journal_keys_sort(struct bch_fs *c)
static void replay_now_at(struct journal *j, u64 seq)
{
BUG_ON(seq < j->replay_jou... |
e296b1f9cadfc4ee7ebe5933fb98497263ce9999 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix inode_backpointer_exists() | Problem Details:
If the dirent an inode points to doesn't exist, we shouldn't be
returning an error - just 0/false.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/fsck.c b/fs/bcachefs/fsck.c
index 6a89b0694e50..ccbf3ac4b1d6 100644
--- a/fs/bcachefs/fsck.c
+++ b/fs/bcachefs/fsck.c
@@ -1092,7 +1092,7 @@ static int inode_backpointer_exists(struct btree_trans *trans,
SPOS(inode->bi_dir, inode->bi_dir_offset, snapshot));
ret = bkey_err(d.s_c);
... |
5650bb46be89a1254609d47e4c87d1e9cf9121fb | Analyze and fix the following issue in the Linux kernel code: bcachefs: Introduce bch2_journal_keys_peek_(upto|slot)() | Problem Details:
When many journal replay keys have been overwritten,
bch2_journal_keys_peek() was taking excessively long to scan before it
found a key to return.
Fix this by introducing bch2_journal_keys_peek_upto() which takes a
parameter for the end of the range we want, so that we can terminate the
search much so... | ```diff
diff --git a/fs/bcachefs/btree_iter.c b/fs/bcachefs/btree_iter.c
index 1c0560ecd120..80ea97e21cf3 100644
--- a/fs/bcachefs/btree_iter.c
+++ b/fs/bcachefs/btree_iter.c
@@ -2225,10 +2225,10 @@ static noinline
struct bkey_s_c btree_trans_peek_slot_journal(struct btree_trans *trans,
struct btree_iter *... |
9b93596c33f6c23de96c05dce82b9aead271a286 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Improve error message when alloc key doesn't match lru entry | Problem Details:
Error messages should always print out the full key when available -
this gives us a starting point when looking through the journal to debug
what went wrong.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/alloc_background.c b/fs/bcachefs/alloc_background.c
index 17b147d15320..1e6283b55e3b 100644
--- a/fs/bcachefs/alloc_background.c
+++ b/fs/bcachefs/alloc_background.c
@@ -1082,6 +1082,7 @@ static int invalidate_one_bucket(struct btree_trans *trans, struct bch_dev *ca)
struct bkey_s_c k... |
ec7ccbde6baa50760c2679a5c54b2fccfd5ca64e | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix CPU usage in journal read path | Problem Details:
In journal_entry_add(), we were repeatedly scanning the journal entries
radix tree to scan for old entries that can be freed, with O(n^2)
behaviour. This patch tweaks things to remember the previous last_seq,
so we don't have to scan for entries to free from the start.
Signed-off-by: Kent Overstreet <... | ```diff
diff --git a/fs/bcachefs/journal_io.c b/fs/bcachefs/journal_io.c
index 64341da75963..9e43914ebd6a 100644
--- a/fs/bcachefs/journal_io.c
+++ b/fs/bcachefs/journal_io.c
@@ -16,17 +16,17 @@
#include "replicas.h"
#include "trace.h"
-static inline u32 journal_entry_radix_idx(struct bch_fs *c,
- struct jset... |
6e811bbbc2ea3b20854c45fa29811640881bb534 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix a null ptr deref | Problem Details:
We start doing allocations before the GC thread is created, which means
we need to check for that to avoid a null ptr deref.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/btree_gc.h b/fs/bcachefs/btree_gc.h
index 8de54005e4ea..95d803b5743d 100644
--- a/fs/bcachefs/btree_gc.h
+++ b/fs/bcachefs/btree_gc.h
@@ -105,7 +105,8 @@ static inline bool gc_visited(struct bch_fs *c, struct gc_pos pos)
static inline void bch2_do_gc_gens(struct bch_fs *c)
{
atomic_... |
a9c0a4cbf1ceb9842fee5d7084817509a5e962aa | Analyze and fix the following issue in the Linux kernel code: bcachefs: Minor device removal fixes | Problem Details:
- We weren't clearing the LRU btree
- bch2_alloc_read() runs before bch2_check_alloc_key() deletes alloc
keys for devices/buckets that don't exists, so it needs to check for
that
- bch2_check_lrus() needs to check that buckets exists
- improve some error messages
Signed-off-by: Kent Overstree... | ```diff
diff --git a/fs/bcachefs/alloc_background.c b/fs/bcachefs/alloc_background.c
index 6110d4ce4e5f..108b98c9fe45 100644
--- a/fs/bcachefs/alloc_background.c
+++ b/fs/bcachefs/alloc_background.c
@@ -446,6 +446,13 @@ int bch2_alloc_read(struct bch_fs *c)
for_each_btree_key(&trans, iter, BTREE_ID_alloc, POS_MIN,
... |
502f973dba660ed04f295e5ba129f2d369cc1aa6 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix a few warnings on 32 bit | Problem Details:
These showed up when building for mips.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/bcachefs_format.h b/fs/bcachefs/bcachefs_format.h
index a84a8e088953..969507c42c55 100644
--- a/fs/bcachefs/bcachefs_format.h
+++ b/fs/bcachefs/bcachefs_format.h
@@ -851,10 +851,9 @@ struct bch_dirent {
#define DT_SUBVOL 16
#define BCH_DT_MAX 17
-#define BCH_NAME_MAX (U8_MAX * sizeo... |
ce6201c456571d919e722eec3c17f868f0575b05 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Use a genradix for reading journal entries | Problem Details:
Previously, the journal read path used a linked list for storing the
journal entries we read from disk. But there's been a bug that's been
causing journal_flush_delay to incorrectly be set to 0, leading to far
more journal entries than is normal being written out, which then means
filesystems are no lo... | ```diff
diff --git a/fs/bcachefs/bcachefs.h b/fs/bcachefs/bcachefs.h
index f2bb23162b4a..43e921b91d85 100644
--- a/fs/bcachefs/bcachefs.h
+++ b/fs/bcachefs/bcachefs.h
@@ -894,7 +894,8 @@ mempool_t bio_bounce_pages;
mempool_t btree_bounce_pool;
struct journal journal;
- struct list_head journal_entries;
+ GENRA... |
48620e5177ae7cd91722f5c504c5138160b90df4 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Topology repair fixes | Problem Details:
- We were failing to start topology repair, because we hadn't set the
superblock flag indicating it needed to run
- set_node_min() forget to update the btree node's key
- bch2_gc_alloc_reset() didn't reset data type, leading to inserting an
invalid key that was empty but had nonzero data type
... | ```diff
diff --git a/fs/bcachefs/btree_gc.c b/fs/bcachefs/btree_gc.c
index 747667ce131d..14b772cd8fe5 100644
--- a/fs/bcachefs/btree_gc.c
+++ b/fs/bcachefs/btree_gc.c
@@ -214,7 +214,7 @@ static int set_node_min(struct bch_fs *c, struct btree *b, struct bpos new_min)
}
bch2_btree_node_drop_keys_outside_node(b);
-
... |
1d8a268940045b73f43f86e6332bdfdfe64d0fa0 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Improve btree_bad_header() | Problem Details:
In the future printbufs will be mempool-ified, so we shouldn't be using
more than one at a time if we don't have to.
This also fixes an extra trailing newline.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/btree_cache.c b/fs/bcachefs/btree_cache.c
index 0e3db9ee65d2..6557fcb24b21 100644
--- a/fs/bcachefs/btree_cache.c
+++ b/fs/bcachefs/btree_cache.c
@@ -767,31 +767,29 @@ static int lock_node_check_fn(struct six_lock *lock, void *p)
static noinline void btree_bad_header(struct bch_fs *c... |
c6b6d416126da015e4b6b6a66b4c6fd3eda40f1f | Analyze and fix the following issue in the Linux kernel code: bcachefs: gc mark fn fixes, cleanups | Problem Details:
mark_stripe_bucket() was busted; it was using @new unitialized.
Also, clean up all the gc mark functions, and convert them to the same
style.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/buckets.c b/fs/bcachefs/buckets.c
index dcb2ea3de4b8..5b78e8f983a1 100644
--- a/fs/bcachefs/buckets.c
+++ b/fs/bcachefs/buckets.c
@@ -624,13 +624,13 @@ int bch2_mark_alloc(struct btree_trans *trans,
return 0;
}
-void bch2_mark_metadata_bucket(struct bch_fs *c, struct bch_dev *ca,
-... |
c32fc674d4ca13f3b889693a2b59365d93a77144 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix pr_buf() calls | Problem Details:
In a few places we were passing a variable to pr_buf() for the format
string - oops.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/opts.c b/fs/bcachefs/opts.c
index 77fbb7d2194e..385451ef865e 100644
--- a/fs/bcachefs/opts.c
+++ b/fs/bcachefs/opts.c
@@ -324,7 +324,7 @@ void bch2_opt_to_text(struct printbuf *out,
if (flags & OPT_SHOW_FULL_LIST)
bch2_string_opt_to_text(out, opt->choices, v);
else
- pr_buf(... |
70a9953c424ccba616a3b74368780de13a80dabd | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix bch2_journal_pin_set() | Problem Details:
When bch2_journal_pin_set() is updating an existing pin, we shouldn't
call bch2_journal_reclaim_fast() after dropping the old pin and before
dropping the new pin - that could reclaim the entry we're trying to pin.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev> | ```diff
diff --git a/fs/bcachefs/journal_reclaim.c b/fs/bcachefs/journal_reclaim.c
index b30730ce58c5..f55fc0b11977 100644
--- a/fs/bcachefs/journal_reclaim.c
+++ b/fs/bcachefs/journal_reclaim.c
@@ -346,13 +346,13 @@ void bch2_journal_pin_put(struct journal *j, u64 seq)
}
}
-static inline void __journal_pin_drop(s... |
d864842581e70d2280f2a51ceb2ad6e7cefeed94 | Analyze and fix the following issue in the Linux kernel code: bcachefs: btree_path_make_mut() clears should_be_locked | Problem Details:
This fixes a bug where __bch2_btree_node_update_key() wasn't clearing
should_be_locked, leading to bch2_btree_path_traverse() always failing -
all callers of btree_path_make_mut() want should_be_locked cleared, so
do it there.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/btree_iter.c b/fs/bcachefs/btree_iter.c
index bfba10b1c127..3a4ed2f70cc2 100644
--- a/fs/bcachefs/btree_iter.c
+++ b/fs/bcachefs/btree_iter.c
@@ -1662,9 +1662,8 @@ static struct btree_path *btree_path_clone(struct btree_trans *trans, struct btr
return new;
}
-struct btree_path * __... |
81cdc8f3070561786a906c66d697666e6a3319b9 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix error path in bch2_snapshot_set_equiv() | Problem Details:
We weren't properly catching errors from snapshot_live() - oops.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/subvolume.c b/fs/bcachefs/subvolume.c
index 2c5f7e7793a7..20c6b21e54d3 100644
--- a/fs/bcachefs/subvolume.c
+++ b/fs/bcachefs/subvolume.c
@@ -139,7 +139,7 @@ static int bch2_snapshots_set_equiv(struct btree_trans *trans)
for_each_btree_key(trans, iter, BTREE_ID_snapshots,
POS_M... |
7fda0f08fa86731f057367ca36054d29d0c0344c | Analyze and fix the following issue in the Linux kernel code: bcachefs: Work around a journal self-deadlock | Problem Details:
bch2_journal_space_available -> bch2_journal_halt() self deadlocks on
journal lock; work around this by dropping/retaking journal lock before
we call bch2_fatal_error().
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/journal_reclaim.c b/fs/bcachefs/journal_reclaim.c
index 56b0c018ac26..b30730ce58c5 100644
--- a/fs/bcachefs/journal_reclaim.c
+++ b/fs/bcachefs/journal_reclaim.c
@@ -216,7 +216,14 @@ void bch2_journal_space_available(struct journal *j)
bch_err(c, "journal stuck\n%s", buf.buf);
pri... |
30690c441a6b481aadb0284d1b9bf487f8d28bfc | Analyze and fix the following issue in the Linux kernel code: bcachefs: Heap code fix | Problem Details:
When deleting an entry from a heap that was at entry h->used - 1, we'd
end up calling heap_sift() on an entry outside the heap - the entry we
just removed - which would end up re-adding it to the heap and deleting
something we didn't want to delete. Oops...
Signed-off-by: Kent Overstreet <kent.overstr... | ```diff
diff --git a/fs/bcachefs/util.h b/fs/bcachefs/util.h
index f2df4d7fbec9..fd776fb281b7 100644
--- a/fs/bcachefs/util.h
+++ b/fs/bcachefs/util.h
@@ -210,9 +210,11 @@ do { \
\
BUG_ON(_i >= (h)->used); \
(h)->used--; \
- heap_swap(h, _i, (h)->used, set_backpointer); \
- heap_sift_u... |
fd1e9c69959ec2dd0946f3d2285548c87f9d6ea3 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix an unitialized var warning in userspace | Problem Details:
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/io.c b/fs/bcachefs/io.c
index c9204cab055d..2cc56979fcb3 100644
--- a/fs/bcachefs/io.c
+++ b/fs/bcachefs/io.c
@@ -1141,7 +1141,7 @@ static void __bch2_write(struct closure *cl)
struct bch_write_op *op = container_of(cl, struct bch_write_op, cl);
struct bch_fs *c = op->c;
struct wr... |
b8559f1a212a7035b430b83e0a01e94a872adc23 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Reset journal flush delay to default value if zeroed | Problem Details:
We've been seeing a very strange bug where journal flush & reclaim delay
end up getting inexplicably zeroed, in the superblock. We're now
validating all the options in bch2_validate_super(), and 0 is no longer
a valid value for those options, but we need to be careful not to
prevent people's filesystem... | ```diff
diff --git a/fs/bcachefs/super-io.c b/fs/bcachefs/super-io.c
index 224653f129f8..6d54319a95e6 100644
--- a/fs/bcachefs/super-io.c
+++ b/fs/bcachefs/super-io.c
@@ -253,7 +253,8 @@ static int validate_sb_layout(struct bch_sb_layout *layout, struct printbuf *out
return 0;
}
-static int bch2_sb_validate(struct... |
b0be2fcfb425022025203c27ed75a20db8c6feda | Analyze and fix the following issue in the Linux kernel code: bcachefs: Change journal_io.c assertion to error message | Problem Details:
Something funny is going on with the new code for restoring the journal
write point, and it's hard to reproduce.
We do want to debug this because resuming writing to the journal in the
wrong spot could be something serious. For now, replace the assertion
with an error message and revert to old behavio... | ```diff
diff --git a/fs/bcachefs/journal_io.c b/fs/bcachefs/journal_io.c
index ba43e5771c7c..b2c3ee336c1f 100644
--- a/fs/bcachefs/journal_io.c
+++ b/fs/bcachefs/journal_io.c
@@ -964,8 +964,16 @@ static void bch2_journal_read_device(struct closure *cl)
}
mutex_unlock(&jlist->lock);
- BUG_ON(ja->bucket_seq[ja->cur... |
7a6f4411aeaae888a7b2880f0c046f9efb0d83cc | Analyze and fix the following issue in the Linux kernel code: bcachefs: Make minimum journal_flush_delay nonzero | Problem Details:
We're seeing a very strange bug where journal_flush_delay sometimes gets
set to 0 in the superblock. Together with the preceding patch, this
should help us track it down.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/opts.h b/fs/bcachefs/opts.h
index 69ca75429943..b45740ec3c67 100644
--- a/fs/bcachefs/opts.h
+++ b/fs/bcachefs/opts.h
@@ -275,7 +275,7 @@ enum opt_type {
NULL, "Extra debugging information during mount/recovery")\
x(journal_flush_delay, u32, \
OPT_FS|OPT_MOUNT|OPT_RUNTIME... |
3a306f3c2c303febffefea4caf09b2326107507a | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix large key cache keys | Problem Details:
Previously, we'd go into an infinite loop when attempting to cache a
bkey in the key cache larger than 128 u64s - since we were only using a
u8 for the size field, it'd get rounded up to 256 then truncated to 0.
Oops.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/btree_types.h b/fs/bcachefs/btree_types.h
index 51eb686331bf..a8b08955cedc 100644
--- a/fs/bcachefs/btree_types.h
+++ b/fs/bcachefs/btree_types.h
@@ -323,7 +323,7 @@ struct bkey_cached {
struct btree_bkey_cached_common c;
unsigned long flags;
- u8 u64s;
+ u16 u64s;
bool v... |
5d93a842c1eb292e2cde9f5025628269d7d386e0 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Fix BTREE_TRIGGER_WANTS_OLD_AND_NEW | Problem Details:
BTREE_TRIGGER_WANTS_OLD_AND_NEW didn't work correctly when the old and
new key were both alloc keys, but different versions - it required old
and new key type to be identical, and this bug is a problem for the new
allocator rewrite.
This patch fixes it by checking if the old and new key have the same
... | ```diff
diff --git a/fs/bcachefs/btree_update_leaf.c b/fs/bcachefs/btree_update_leaf.c
index a789a7cf4c74..cb0cab71b99b 100644
--- a/fs/bcachefs/btree_update_leaf.c
+++ b/fs/bcachefs/btree_update_leaf.c
@@ -442,7 +442,8 @@ static int run_one_mem_trigger(struct btree_trans *trans,
if (!btree_node_type_needs_gc(i->btre... |
880e2275f9f1461c87cd113a8da291861cc01400 | Analyze and fix the following issue in the Linux kernel code: bcachefs: Move trigger fns to bkey_ops | Problem Details:
This replaces the switch statements in bch2_mark_key(),
bch2_trans_mark_key() with new bkey methods - prep work for the next
patch, which fixes BTREE_TRIGGER_WANTS_OLD_AND_NEW.
Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> | ```diff
diff --git a/fs/bcachefs/alloc_background.h b/fs/bcachefs/alloc_background.h
index 98c7866e20b5..3eaa6d204286 100644
--- a/fs/bcachefs/alloc_background.h
+++ b/fs/bcachefs/alloc_background.h
@@ -65,16 +65,19 @@ void bch2_alloc_to_text(struct printbuf *, struct bch_fs *, struct bkey_s_c);
#define bch2_bkey_ops_... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.