summaryrefslogtreecommitdiff
path: root/fs
diff options
context:
space:
mode:
Diffstat (limited to 'fs')
-rw-r--r--fs/bcachefs/alloc_background.c2
-rw-r--r--fs/bcachefs/btree_gc.c31
-rw-r--r--fs/bcachefs/btree_io.c50
-rw-r--r--fs/bcachefs/btree_node_scan.c29
-rw-r--r--fs/bcachefs/errcode.h6
-rw-r--r--fs/bcachefs/extents.c4
-rw-r--r--fs/bcachefs/fsck.c7
-rw-r--r--fs/bcachefs/io_write.c7
-rw-r--r--fs/bcachefs/journal.c3
-rw-r--r--fs/bcachefs/journal_io.c2
-rw-r--r--fs/bcachefs/recovery.c7
-rw-r--r--fs/bcachefs/super.c2
12 files changed, 83 insertions, 67 deletions
diff --git a/fs/bcachefs/alloc_background.c b/fs/bcachefs/alloc_background.c
index f1d35b7f3fc5..d5cb07c7f4d6 100644
--- a/fs/bcachefs/alloc_background.c
+++ b/fs/bcachefs/alloc_background.c
@@ -389,7 +389,7 @@ void __bch2_alloc_to_v4(struct bkey_s_c k, struct bch_alloc_v4 *out)
if (k.k->type == KEY_TYPE_alloc_v4) {
void *src, *dst;
- *out = *bkey_s_c_to_alloc_v4(k).v;
+ bkey_val_copy(out, bkey_s_c_to_alloc_v4(k));
src = alloc_v4_backpointers(out);
SET_BCH_ALLOC_V4_BACKPOINTERS_START(out, BCH_ALLOC_V4_U64s);
diff --git a/fs/bcachefs/btree_gc.c b/fs/bcachefs/btree_gc.c
index 34cb8a4324dc..e95bb6849aef 100644
--- a/fs/bcachefs/btree_gc.c
+++ b/fs/bcachefs/btree_gc.c
@@ -44,10 +44,6 @@
#include <linux/rcupdate.h>
#include <linux/sched/task.h>
-#define DROP_THIS_NODE 10
-#define DROP_PREV_NODE 11
-#define DID_FILL_FROM_SCAN 12
-
/*
* Returns true if it's a btree we can easily reconstruct, or otherwise won't
* cause data loss if it's missing:
@@ -252,7 +248,7 @@ static int btree_check_node_boundaries(struct btree_trans *trans, struct btree *
return ret;
*pulled_from_scan = cur->data->min_key;
- ret = DID_FILL_FROM_SCAN;
+ ret = bch_err_throw(c, topology_repair_did_fill_from_scan);
} else {
if (mustfix_fsck_err(trans, btree_node_topology_bad_min_key,
"btree node with incorrect min_key%s", buf.buf))
@@ -263,7 +259,7 @@ static int btree_check_node_boundaries(struct btree_trans *trans, struct btree *
if (bpos_ge(prev->data->min_key, cur->data->min_key)) { /* fully? */
if (mustfix_fsck_err(trans, btree_node_topology_overwritten_by_next_node,
"btree node overwritten by next node%s", buf.buf))
- ret = DROP_PREV_NODE;
+ ret = bch_err_throw(c, topology_repair_drop_prev_node);
} else {
if (mustfix_fsck_err(trans, btree_node_topology_bad_max_key,
"btree node with incorrect max_key%s", buf.buf))
@@ -274,7 +270,7 @@ static int btree_check_node_boundaries(struct btree_trans *trans, struct btree *
if (bpos_ge(expected_start, cur->data->max_key)) { /* fully? */
if (mustfix_fsck_err(trans, btree_node_topology_overwritten_by_prev_node,
"btree node overwritten by prev node%s", buf.buf))
- ret = DROP_THIS_NODE;
+ ret = bch_err_throw(c, topology_repair_drop_this_node);
} else {
if (mustfix_fsck_err(trans, btree_node_topology_bad_min_key,
"btree node with incorrect min_key%s", buf.buf))
@@ -314,7 +310,7 @@ static int btree_repair_node_end(struct btree_trans *trans, struct btree *b,
return ret;
*pulled_from_scan = b->key.k.p;
- ret = DID_FILL_FROM_SCAN;
+ ret = bch_err_throw(c, topology_repair_did_fill_from_scan);
} else {
ret = set_node_max(c, child, b->key.k.p);
}
@@ -391,15 +387,15 @@ again:
ret = lockrestart_do(trans,
btree_check_node_boundaries(trans, b, prev, cur, pulled_from_scan));
- if (ret < 0)
+ if (ret && !bch2_err_matches(ret, BCH_ERR_topology_repair))
goto err;
- if (ret == DID_FILL_FROM_SCAN) {
+ if (bch2_err_matches(ret, BCH_ERR_topology_repair_did_fill_from_scan)) {
new_pass = true;
ret = 0;
}
- if (ret == DROP_THIS_NODE) {
+ if (bch2_err_matches(ret, BCH_ERR_topology_repair_drop_this_node)) {
six_unlock_read(&cur->c.lock);
bch2_btree_node_evict(trans, cur_k.k);
ret = bch2_journal_key_delete(c, b->c.btree_id,
@@ -414,7 +410,7 @@ again:
six_unlock_read(&prev->c.lock);
prev = NULL;
- if (ret == DROP_PREV_NODE) {
+ if (bch2_err_matches(ret, BCH_ERR_topology_repair_drop_prev_node)) {
bch_info(c, "dropped prev node");
bch2_btree_node_evict(trans, prev_k.k);
ret = bch2_journal_key_delete(c, b->c.btree_id,
@@ -436,7 +432,7 @@ again:
BUG_ON(cur);
ret = lockrestart_do(trans,
btree_repair_node_end(trans, b, prev, pulled_from_scan));
- if (ret == DID_FILL_FROM_SCAN) {
+ if (bch2_err_matches(ret, BCH_ERR_topology_repair_did_fill_from_scan)) {
new_pass = true;
ret = 0;
}
@@ -477,7 +473,7 @@ again:
six_unlock_read(&cur->c.lock);
cur = NULL;
- if (ret == DROP_THIS_NODE) {
+ if (bch2_err_matches(ret, BCH_ERR_topology_repair_drop_this_node)) {
bch2_btree_node_evict(trans, cur_k.k);
ret = bch2_journal_key_delete(c, b->c.btree_id,
b->c.level, cur_k.k->k.p);
@@ -504,7 +500,7 @@ again:
if (mustfix_fsck_err_on(!have_child,
c, btree_node_topology_interior_node_empty,
"empty interior btree node at %s", buf.buf))
- ret = DROP_THIS_NODE;
+ ret = bch_err_throw(c, topology_repair_drop_this_node);
err:
fsck_err:
if (!IS_ERR_OR_NULL(prev))
@@ -521,7 +517,8 @@ fsck_err:
bch2_bkey_buf_exit(&prev_k, c);
bch2_bkey_buf_exit(&cur_k, c);
- bch_err_fn(c, ret);
+ if (!bch2_err_matches(ret, BCH_ERR_topology_repair))
+ bch_err_fn(c, ret);
return ret;
}
@@ -592,7 +589,7 @@ recover:
ret = bch2_btree_repair_topology_recurse(trans, b, &pulled_from_scan);
six_unlock_read(&b->c.lock);
- if (ret == DROP_THIS_NODE) {
+ if (bch2_err_matches(ret, BCH_ERR_topology_repair_drop_this_node)) {
scoped_guard(mutex, &c->btree_cache.lock)
bch2_btree_node_hash_remove(&c->btree_cache, b);
diff --git a/fs/bcachefs/btree_io.c b/fs/bcachefs/btree_io.c
index bd86dd7151a1..2bd8422eb72c 100644
--- a/fs/bcachefs/btree_io.c
+++ b/fs/bcachefs/btree_io.c
@@ -563,15 +563,11 @@ static int __btree_err(int ret,
struct printbuf *err_msg,
const char *fmt, ...)
{
- if (c->recovery.curr_pass == BCH_RECOVERY_PASS_scan_for_btree_nodes)
- return ret == -BCH_ERR_btree_node_read_err_fixable
- ? bch_err_throw(c, fsck_fix)
- : ret;
-
+ bool in_scan = c->recovery.curr_pass == BCH_RECOVERY_PASS_scan_for_btree_nodes;
bool have_retry = false;
int ret2;
- if (ca) {
+ if (ca && !in_scan) {
bch2_mark_btree_validate_failure(failed, ca->dev_idx);
struct extent_ptr_decoded pick;
@@ -585,12 +581,14 @@ static int __btree_err(int ret,
if (!have_retry && ret == -BCH_ERR_btree_node_read_err_must_retry)
ret = bch_err_throw(c, btree_node_read_err_bad_node);
- bch2_sb_error_count(c, err_type);
+ if (!in_scan)
+ bch2_sb_error_count(c, err_type);
bool print_deferred = err_msg &&
rw == READ &&
- !(test_bit(BCH_FS_in_fsck, &c->flags) &&
- c->opts.fix_errors == FSCK_FIX_ask);
+ (!(test_bit(BCH_FS_in_fsck, &c->flags) &&
+ c->opts.fix_errors == FSCK_FIX_ask) ||
+ in_scan);
CLASS(printbuf, out)();
bch2_log_msg_start(c, &out);
@@ -603,11 +601,17 @@ static int __btree_err(int ret,
va_list args;
va_start(args, fmt);
prt_vprintf(err_msg, fmt, args);
- va_end(args);
+ va_end(args);;
- if (print_deferred) {
+ if (print_deferred)
prt_newline(err_msg);
+ if (in_scan)
+ return ret == -BCH_ERR_btree_node_read_err_fixable
+ ? bch_err_throw(c, fsck_fix)
+ : ret;
+
+ if (print_deferred) {
switch (ret) {
case -BCH_ERR_btree_node_read_err_fixable:
ret2 = bch2_fsck_err_opt(c, FSCK_CAN_FIX, err_type);
@@ -1405,10 +1409,8 @@ static void btree_node_read_work(struct work_struct *work)
ret = bch2_bkey_pick_read_device(c,
bkey_i_to_s_c(&b->key),
&failed, &rb->pick, -1);
- if (ret <= 0) {
- set_btree_node_read_error(b);
+ if (ret <= 0)
break;
- }
ca = bch2_dev_get_ioref(c, rb->pick.ptr.dev, READ, BCH_DEV_READ_REF_btree_node_read);
rb->have_ioref = ca != NULL;
@@ -1442,27 +1444,21 @@ start:
bch2_maybe_corrupt_bio(bio, bch2_btree_read_corrupt_ratio);
ret = bch2_btree_node_read_done(c, ca, b, &failed, &buf);
- if (ret == -BCH_ERR_btree_node_read_err_want_retry ||
- ret == -BCH_ERR_btree_node_read_err_must_retry)
- continue;
-
- if (ret)
- set_btree_node_read_error(b);
-
- break;
+ if (ret != -BCH_ERR_btree_node_read_err_want_retry &&
+ ret != -BCH_ERR_btree_node_read_err_must_retry)
+ break;
}
bch2_io_failures_to_text(&buf, c, &failed);
- if (btree_node_read_error(b))
- bch2_btree_lost_data(c, &buf, b->c.btree_id);
-
/*
* only print retry success if we read from a replica with no errors
*/
- if (btree_node_read_error(b))
+ if (ret) {
+ set_btree_node_read_error(b);
+ bch2_btree_lost_data(c, &buf, b->c.btree_id);
prt_printf(&buf, "ret %s", bch2_err_str(ret));
- else if (failed.nr) {
+ } else if (failed.nr) {
if (!bch2_dev_io_failures(&failed, rb->pick.ptr.dev))
prt_printf(&buf, "retry success");
else
diff --git a/fs/bcachefs/btree_node_scan.c b/fs/bcachefs/btree_node_scan.c
index f4f958f4615d..c518e18ce5dc 100644
--- a/fs/bcachefs/btree_node_scan.c
+++ b/fs/bcachefs/btree_node_scan.c
@@ -155,17 +155,6 @@ static void try_read_btree_node(struct find_btree_nodes *f, struct bch_dev *ca,
if (BTREE_NODE_LEVEL(bn) >= BTREE_MAX_DEPTH)
return;
- if (BTREE_NODE_ID(bn) >= BTREE_ID_NR_MAX)
- return;
-
- bio_reset(bio, ca->disk_sb.bdev, REQ_OP_READ);
- bio->bi_iter.bi_sector = offset;
- bch2_bio_map(bio, b->data, c->opts.btree_node_size);
-
- submit_time = local_clock();
- submit_bio_wait(bio);
- bch2_account_io_completion(ca, BCH_MEMBER_ERROR_read, submit_time, !bio->bi_status);
-
rcu_read_lock();
struct found_btree_node n = {
.btree_id = BTREE_NODE_ID(bn),
@@ -182,10 +171,26 @@ static void try_read_btree_node(struct find_btree_nodes *f, struct bch_dev *ca,
};
rcu_read_unlock();
+ bio_reset(bio, ca->disk_sb.bdev, REQ_OP_READ);
+ bio->bi_iter.bi_sector = offset;
+ bch2_bio_map(bio, b->data, c->opts.btree_node_size);
+
+ submit_time = local_clock();
+ submit_bio_wait(bio);
+ bch2_account_io_completion(ca, BCH_MEMBER_ERROR_read, submit_time, !bio->bi_status);
+
found_btree_node_to_key(&b->key, &n);
CLASS(printbuf, buf)();
- if (!bch2_btree_node_read_done(c, ca, b, NULL, &buf)) {
+
+ found_btree_node_to_text(&buf, c, &n);
+ prt_newline(&buf);
+
+ int ret = bch2_btree_node_read_done(c, ca, b, NULL, &buf);
+
+ bch_verbose(ca, "attempted to read, ret %s\n%s", bch2_err_str(ret), buf.buf);
+
+ if (!ret) {
/* read_done will swap out b->data for another buffer */
bn = b->data;
/*
diff --git a/fs/bcachefs/errcode.h b/fs/bcachefs/errcode.h
index 5db48c01f8e6..cec8b0f47d3d 100644
--- a/fs/bcachefs/errcode.h
+++ b/fs/bcachefs/errcode.h
@@ -218,9 +218,13 @@
x(EINVAL, varint_decode_error) \
x(EINVAL, erasure_coding_found_btree_node) \
x(EINVAL, option_negative) \
+ x(EINVAL, topology_repair) \
+ x(BCH_ERR_topology_repair, topology_repair_drop_this_node) \
+ x(BCH_ERR_topology_repair, topology_repair_drop_prev_node) \
+ x(BCH_ERR_topology_repair, topology_repair_did_fill_from_scan) \
x(EOPNOTSUPP, may_not_use_incompat_feature) \
x(EOPNOTSUPP, no_casefolding_without_utf8) \
- x(EOPNOTSUPP, casefolding_disabled) \
+ x(EOPNOTSUPP, casefolding_disabled) \
x(EOPNOTSUPP, casefold_opt_is_dir_only) \
x(EOPNOTSUPP, unsupported_fsx_flag) \
x(EOPNOTSUPP, unsupported_fa_flag) \
diff --git a/fs/bcachefs/extents.c b/fs/bcachefs/extents.c
index 8152ef1cbbcd..b879a586b7f6 100644
--- a/fs/bcachefs/extents.c
+++ b/fs/bcachefs/extents.c
@@ -282,9 +282,9 @@ int bch2_bkey_pick_read_device(struct bch_fs *c, struct bkey_s_c k,
if (have_pick)
return 1;
- if (!have_dirty_ptrs)
+ if (!have_dirty_ptrs && !bkey_is_btree_ptr(k.k))
return 0;
- if (have_missing_devs)
+ if (have_missing_devs || !have_dirty_ptrs)
return bch_err_throw(c, no_device_to_read_from);
if (have_csum_errors)
return bch_err_throw(c, data_read_csum_err);
diff --git a/fs/bcachefs/fsck.c b/fs/bcachefs/fsck.c
index df0aa2522b18..183b88bbd402 100644
--- a/fs/bcachefs/fsck.c
+++ b/fs/bcachefs/fsck.c
@@ -1975,6 +1975,10 @@ static int check_extent(struct btree_trans *trans, struct btree_iter *iter,
}
}
+ ret = check_extent_overbig(trans, iter, k);
+ if (ret)
+ goto err;
+
ret = bch2_trans_commit(trans, res, NULL, BCH_TRANS_COMMIT_no_enospc);
if (ret)
goto err;
@@ -2021,8 +2025,7 @@ int bch2_check_extents(struct bch_fs *c)
POS(BCACHEFS_ROOT_INO, 0),
BTREE_ITER_prefetch|BTREE_ITER_all_snapshots, k, ({
bch2_disk_reservation_put(c, &res);
- check_extent(trans, &iter, k, &w, &s, &extent_ends, &res) ?:
- check_extent_overbig(trans, &iter, k);
+ check_extent(trans, &iter, k, &w, &s, &extent_ends, &res);
})) ?:
check_i_sectors_notnested(trans, &w);
diff --git a/fs/bcachefs/io_write.c b/fs/bcachefs/io_write.c
index d7620138e038..44b02d4b6502 100644
--- a/fs/bcachefs/io_write.c
+++ b/fs/bcachefs/io_write.c
@@ -89,7 +89,12 @@ void bch2_latency_acct(struct bch_dev *ca, u64 submit_time, int rw)
new = ewma_add(old, io_latency, 5);
} while (!atomic64_try_cmpxchg(latency, &old, new));
- bch2_congested_acct(ca, io_latency, now, rw);
+ /*
+ * Only track read latency for congestion accounting: writes are subject
+ * to heavy queuing delays from page cache writeback:
+ */
+ if (rw == READ)
+ bch2_congested_acct(ca, io_latency, now, rw);
__bch2_time_stats_update(&ca->io_latency[rw].stats, submit_time, now);
}
diff --git a/fs/bcachefs/journal.c b/fs/bcachefs/journal.c
index 376c9b3b20fe..97760e89e5a3 100644
--- a/fs/bcachefs/journal.c
+++ b/fs/bcachefs/journal.c
@@ -198,7 +198,8 @@ void bch2_journal_do_writes(struct journal *j)
if (!journal_state_seq_count(j, j->reservations, seq)) {
j->seq_write_started = seq;
w->write_started = true;
- closure_call(&w->io, bch2_journal_write, j->wq, &c->cl);
+ closure_get(&c->cl);
+ closure_call(&w->io, bch2_journal_write, j->wq, NULL);
}
break;
diff --git a/fs/bcachefs/journal_io.c b/fs/bcachefs/journal_io.c
index bb4a1a7e48e5..47224666d07e 100644
--- a/fs/bcachefs/journal_io.c
+++ b/fs/bcachefs/journal_io.c
@@ -1821,7 +1821,7 @@ static CLOSURE_CALLBACK(journal_write_done)
if (do_discards)
bch2_do_discards(c);
- closure_return(cl);
+ closure_put(&c->cl);
}
static void journal_write_endio(struct bio *bio)
diff --git a/fs/bcachefs/recovery.c b/fs/bcachefs/recovery.c
index 58c159e5f10d..304473dac268 100644
--- a/fs/bcachefs/recovery.c
+++ b/fs/bcachefs/recovery.c
@@ -67,13 +67,16 @@ int bch2_btree_lost_data(struct bch_fs *c,
ret = __bch2_run_explicit_recovery_pass(c, msg, BCH_RECOVERY_PASS_check_backpointers_to_extents, 0, &write_sb) ?: ret;
#endif
+ write_sb |= !__test_and_set_bit_le64(BCH_FSCK_ERR_lru_entry_bad, ext->errors_silent);
+ write_sb |= !__test_and_set_bit_le64(BCH_FSCK_ERR_backpointer_to_missing_ptr, ext->errors_silent);
+ write_sb |= !__test_and_set_bit_le64(BCH_FSCK_ERR_alloc_key_data_type_wrong, ext->errors_silent);
+ write_sb |= !__test_and_set_bit_le64(BCH_FSCK_ERR_alloc_key_dirty_sectors_wrong, ext->errors_silent);
+
switch (btree) {
case BTREE_ID_alloc:
ret = __bch2_run_explicit_recovery_pass(c, msg, BCH_RECOVERY_PASS_check_alloc_info, 0, &write_sb) ?: ret;
- write_sb |= !__test_and_set_bit_le64(BCH_FSCK_ERR_alloc_key_data_type_wrong, ext->errors_silent);
write_sb |= !__test_and_set_bit_le64(BCH_FSCK_ERR_alloc_key_gen_wrong, ext->errors_silent);
- write_sb |= !__test_and_set_bit_le64(BCH_FSCK_ERR_alloc_key_dirty_sectors_wrong, ext->errors_silent);
write_sb |= !__test_and_set_bit_le64(BCH_FSCK_ERR_alloc_key_cached_sectors_wrong, ext->errors_silent);
write_sb |= !__test_and_set_bit_le64(BCH_FSCK_ERR_alloc_key_stripe_wrong, ext->errors_silent);
write_sb |= !__test_and_set_bit_le64(BCH_FSCK_ERR_alloc_key_stripe_redundancy_wrong, ext->errors_silent);
diff --git a/fs/bcachefs/super.c b/fs/bcachefs/super.c
index 0fc0b2221036..b3b2d8353a36 100644
--- a/fs/bcachefs/super.c
+++ b/fs/bcachefs/super.c
@@ -729,6 +729,8 @@ void __bch2_fs_stop(struct bch_fs *c)
cancel_work_sync(&ca->io_error_work);
cancel_work_sync(&c->read_only_work);
+
+ flush_work(&c->btree_interior_update_work);
}
void bch2_fs_free(struct bch_fs *c)
generated by cgit v1.2.3 (git 2.46.0) at 2025-07-20 10:30:46 +0000