summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
Diffstat
-rw-r--r--fs/bcachefs/acl.c4
-rw-r--r--fs/bcachefs/alloc_background.c18
-rw-r--r--fs/bcachefs/btree_gc.c31
-rw-r--r--fs/bcachefs/btree_io.c50
-rw-r--r--fs/bcachefs/btree_iter.c5
-rw-r--r--fs/bcachefs/btree_key_cache.c8
-rw-r--r--fs/bcachefs/btree_node_scan.c36
-rw-r--r--fs/bcachefs/btree_update_interior.c41
-rw-r--r--fs/bcachefs/errcode.h8
-rw-r--r--fs/bcachefs/extents.c18
-rw-r--r--fs/bcachefs/extents.h2
-rw-r--r--fs/bcachefs/fs-io-direct.c2
-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/move.c16
-rw-r--r--fs/bcachefs/recovery.c7
-rw-r--r--fs/bcachefs/super.c2
-rw-r--r--mm/shrinker.c1
20 files changed, 176 insertions, 92 deletions
diff --git a/fs/bcachefs/acl.c b/fs/bcachefs/acl.c
index 307824d6eccb..8f970dc19dea 100644
--- a/fs/bcachefs/acl.c
+++ b/fs/bcachefs/acl.c
@@ -138,8 +138,8 @@ static struct posix_acl *bch2_acl_from_disk(struct btree_trans *trans,
acl = allocate_dropping_locks(trans, ret,
posix_acl_alloc(count, _gfp));
- if (!acl)
- return ERR_PTR(-ENOMEM);
+ if (!acl && !ret)
+ ret = bch_err_throw(trans->c, ENOMEM_acl);
if (ret) {
kfree(acl);
return ERR_PTR(ret);
diff --git a/fs/bcachefs/alloc_background.c b/fs/bcachefs/alloc_background.c
index 4c1604fd80f9..d5cb07c7f4d6 100644
--- a/fs/bcachefs/alloc_background.c
+++ b/fs/bcachefs/alloc_background.c
@@ -337,9 +337,10 @@ void bch2_alloc_v4_swab(struct bkey_s k)
}
static inline void __bch2_alloc_v4_to_text(struct printbuf *out, struct bch_fs *c,
- unsigned dev, const struct bch_alloc_v4 *a)
+ struct bkey_s_c k,
+ const struct bch_alloc_v4 *a)
{
- struct bch_dev *ca = c ? bch2_dev_tryget_noerror(c, dev) : NULL;
+ struct bch_dev *ca = c ? bch2_dev_tryget_noerror(c, k.k->p.inode) : NULL;
prt_newline(out);
printbuf_indent_add(out, 2);
@@ -348,11 +349,14 @@ static inline void __bch2_alloc_v4_to_text(struct printbuf *out, struct bch_fs *
bch2_prt_data_type(out, a->data_type);
prt_newline(out);
prt_printf(out, "journal_seq_nonempty %llu\n", a->journal_seq_nonempty);
- prt_printf(out, "journal_seq_empty %llu\n", a->journal_seq_empty);
+ if (bkey_val_bytes(k.k) > offsetof(struct bch_alloc_v4, journal_seq_empty))
+ prt_printf(out, "journal_seq_empty %llu\n", a->journal_seq_empty);
+
prt_printf(out, "need_discard %llu\n", BCH_ALLOC_V4_NEED_DISCARD(a));
prt_printf(out, "need_inc_gen %llu\n", BCH_ALLOC_V4_NEED_INC_GEN(a));
prt_printf(out, "dirty_sectors %u\n", a->dirty_sectors);
- prt_printf(out, "stripe_sectors %u\n", a->stripe_sectors);
+ if (bkey_val_bytes(k.k) > offsetof(struct bch_alloc_v4, stripe_sectors))
+ prt_printf(out, "stripe_sectors %u\n", a->stripe_sectors);
prt_printf(out, "cached_sectors %u\n", a->cached_sectors);
prt_printf(out, "stripe %u\n", a->stripe);
prt_printf(out, "stripe_redundancy %u\n", a->stripe_redundancy);
@@ -372,12 +376,12 @@ void bch2_alloc_to_text(struct printbuf *out, struct bch_fs *c, struct bkey_s_c
struct bch_alloc_v4 _a;
const struct bch_alloc_v4 *a = bch2_alloc_to_v4(k, &_a);
- __bch2_alloc_v4_to_text(out, c, k.k->p.inode, a);
+ __bch2_alloc_v4_to_text(out, c, k, a);
}
void bch2_alloc_v4_to_text(struct printbuf *out, struct bch_fs *c, struct bkey_s_c k)
{
- __bch2_alloc_v4_to_text(out, c, k.k->p.inode, bkey_s_c_to_alloc_v4(k).v);
+ __bch2_alloc_v4_to_text(out, c, k, bkey_s_c_to_alloc_v4(k).v);
}
void __bch2_alloc_to_v4(struct bkey_s_c k, struct bch_alloc_v4 *out)
@@ -385,7 +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_iter.c b/fs/bcachefs/btree_iter.c
index cc771affa511..a282c3886168 100644
--- a/fs/bcachefs/btree_iter.c
+++ b/fs/bcachefs/btree_iter.c
@@ -2860,8 +2860,9 @@ struct bkey_s_c bch2_btree_iter_peek_slot(struct btree_trans *trans, struct btre
!bkey_deleted(k.k) &&
(k2 = btree_trans_peek_key_cache(trans, iter, iter->pos)).k) {
k = k2;
- if (!bkey_err(k))
- iter->k = *k.k;
+ if (bkey_err(k))
+ goto out;
+ iter->k = *k.k;
}
if (unlikely(k.k->type == KEY_TYPE_whiteout &&
diff --git a/fs/bcachefs/btree_key_cache.c b/fs/bcachefs/btree_key_cache.c
index ebba14da92b4..d61b782087ce 100644
--- a/fs/bcachefs/btree_key_cache.c
+++ b/fs/bcachefs/btree_key_cache.c
@@ -254,11 +254,13 @@ static int btree_key_cache_create(struct btree_trans *trans,
struct bkey_i *new_k = allocate_dropping_locks(trans, ret,
kmalloc(key_u64s * sizeof(u64), _gfp));
- if (unlikely(!new_k)) {
+ if (unlikely(!new_k && !ret)) {
bch_err(trans->c, "error allocating memory for key cache key, btree %s u64s %u",
bch2_btree_id_str(ck->key.btree_id), key_u64s);
ret = bch_err_throw(c, ENOMEM_btree_key_cache_fill);
- } else if (ret) {
+ }
+
+ if (unlikely(ret)) {
kfree(new_k);
goto err;
}
@@ -407,7 +409,7 @@ int bch2_btree_path_traverse_cached(struct btree_trans *trans,
btree_node_unlock(trans, path, 0);
path->l[0].b = ERR_PTR(ret);
}
- } else {
+ } else if (!(flags & BTREE_ITER_cached_nofill)) {
BUG_ON(path->uptodate);
BUG_ON(!path->nodes_locked);
}
diff --git a/fs/bcachefs/btree_node_scan.c b/fs/bcachefs/btree_node_scan.c
index d997e3818c30..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;
/*
@@ -270,6 +275,9 @@ static int read_btree_nodes(struct find_btree_nodes *f)
int ret = 0;
closure_init_stack(&cl);
+ CLASS(printbuf, buf)();
+
+ prt_printf(&buf, "scanning for btree nodes on");
for_each_online_member(c, ca, BCH_DEV_READ_REF_btree_node_scan) {
if (!(ca->mi.data_allowed & BIT(BCH_DATA_btree)))
@@ -295,10 +303,14 @@ static int read_btree_nodes(struct find_btree_nodes *f)
break;
}
+ prt_printf(&buf, " %s", ca->name);
+
closure_get(&cl);
enumerated_ref_get(&ca->io_ref[READ], BCH_DEV_READ_REF_btree_node_scan);
wake_up_process(t);
}
+
+ bch_notice(c, "%s", buf.buf);
err:
while (closure_sync_timeout(&cl, sysctl_hung_task_timeout_secs * HZ / 2))
;
diff --git a/fs/bcachefs/btree_update_interior.c b/fs/bcachefs/btree_update_interior.c
index 312ef203b27b..e4aa4fa749bc 100644
--- a/fs/bcachefs/btree_update_interior.c
+++ b/fs/bcachefs/btree_update_interior.c
@@ -14,6 +14,7 @@
#include "btree_locking.h"
#include "buckets.h"
#include "clock.h"
+#include "disk_groups.h"
#include "enumerated_ref.h"
#include "error.h"
#include "extents.h"
@@ -277,6 +278,36 @@ static void bch2_btree_node_free_never_used(struct btree_update *as,
bch2_trans_node_drop(trans, b);
}
+static bool can_use_btree_node(struct bch_fs *c,
+ struct disk_reservation *res,
+ unsigned target,
+ struct bkey_s_c k)
+{
+ if (!bch2_bkey_devs_rw(c, k))
+ return false;
+
+ if (target && !bch2_bkey_in_target(c, k, target))
+ return false;
+
+ unsigned durability = bch2_bkey_durability(c, k);
+
+ if (durability >= res->nr_replicas)
+ return true;
+
+ struct bch_devs_mask devs = target_rw_devs(c, BCH_DATA_btree, target);
+
+ guard(rcu)();
+
+ unsigned durability_available = 0, i;
+ for_each_set_bit(i, devs.d, BCH_SB_MEMBERS_MAX) {
+ struct bch_dev *ca = bch2_dev_rcu_noerror(c, i);
+ if (ca)
+ durability_available += ca->mi.durability;
+ }
+
+ return durability >= durability_available;
+}
+
static struct btree *__bch2_btree_node_alloc(struct btree_trans *trans,
struct disk_reservation *res,
struct closure *cl,
@@ -303,10 +334,14 @@ static struct btree *__bch2_btree_node_alloc(struct btree_trans *trans,
mutex_lock(&c->btree_reserve_cache_lock);
if (c->btree_reserve_cache_nr > nr_reserve) {
for (struct btree_alloc *a = c->btree_reserve_cache;
- a < c->btree_reserve_cache + c->btree_reserve_cache_nr;
- a++) {
- if (target && !bch2_bkey_in_target(c, bkey_i_to_s_c(&a->k), target))
+ a < c->btree_reserve_cache + c->btree_reserve_cache_nr;) {
+ /* check if it has sufficient durability */
+
+ if (!can_use_btree_node(c, res, target, bkey_i_to_s_c(&a->k))) {
+ bch2_open_buckets_put(c, &a->ob);
+ *a = c->btree_reserve_cache[--c->btree_reserve_cache_nr];
continue;
+ }
bkey_copy(&b->key, &a->k);
b->ob = a->ob;
diff --git a/fs/bcachefs/errcode.h b/fs/bcachefs/errcode.h
index 2de0dc91a69e..cec8b0f47d3d 100644
--- a/fs/bcachefs/errcode.h
+++ b/fs/bcachefs/errcode.h
@@ -90,6 +90,8 @@
x(ENOMEM, ENOMEM_disk_accounting) \
x(ENOMEM, ENOMEM_stripe_head_alloc) \
x(ENOMEM, ENOMEM_journal_read_bucket) \
+ x(ENOMEM, ENOMEM_acl) \
+ x(ENOMEM, ENOMEM_move_extent) \
x(ENOSPC, ENOSPC_disk_reservation) \
x(ENOSPC, ENOSPC_bucket_alloc) \
x(ENOSPC, ENOSPC_disk_label_add) \
@@ -216,9 +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 b36ecfc0ab9d..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);
@@ -1006,6 +1006,20 @@ const struct bch_extent_ptr *bch2_bkey_has_device_c(struct bkey_s_c k, unsigned
return NULL;
}
+bool bch2_bkey_devs_rw(struct bch_fs *c, struct bkey_s_c k)
+{
+ struct bkey_ptrs_c ptrs = bch2_bkey_ptrs_c(k);
+
+ guard(rcu)();
+ bkey_for_each_ptr(ptrs, ptr) {
+ CLASS(bch2_dev_tryget, ca)(c, ptr->dev);
+ if (!ca || ca->mi.state != BCH_MEMBER_STATE_rw)
+ return false;
+ }
+
+ return true;
+}
+
bool bch2_bkey_has_target(struct bch_fs *c, struct bkey_s_c k, unsigned target)
{
struct bkey_ptrs_c ptrs = bch2_bkey_ptrs_c(k);
diff --git a/fs/bcachefs/extents.h b/fs/bcachefs/extents.h
index f212f91c278d..35ee03cd5065 100644
--- a/fs/bcachefs/extents.h
+++ b/fs/bcachefs/extents.h
@@ -614,6 +614,8 @@ static inline struct bch_extent_ptr *bch2_bkey_has_device(struct bkey_s k, unsig
return (void *) bch2_bkey_has_device_c(k.s_c, dev);
}
+bool bch2_bkey_devs_rw(struct bch_fs *, struct bkey_s_c);
+
bool bch2_bkey_has_target(struct bch_fs *, struct bkey_s_c, unsigned);
bool bch2_bkey_in_target(struct bch_fs *, struct bkey_s_c, unsigned);
diff --git a/fs/bcachefs/fs-io-direct.c b/fs/bcachefs/fs-io-direct.c
index 73d44875faf2..e53fee0513fd 100644
--- a/fs/bcachefs/fs-io-direct.c
+++ b/fs/bcachefs/fs-io-direct.c
@@ -127,7 +127,7 @@ static int bch2_direct_IO_read(struct kiocb *req, struct iov_iter *iter)
* the dirtying of requests that are internal from the kernel (i.e. from
* loopback), because we'll deadlock on page_lock.
*/
- dio->should_dirty = iter_is_iovec(iter);
+ dio->should_dirty = user_backed_iter(iter);
blk_start_plug(&plug);
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 de03e20f6e30..97760e89e5a3 100644
--- a/fs/bcachefs/journal.c
+++ b/fs/bcachefs/journal.c
@@ -182,6 +182,8 @@ journal_error_check_stuck(struct journal *j, int error, unsigned flags)
void bch2_journal_do_writes(struct journal *j)
{
+ struct bch_fs *c = container_of(j, struct bch_fs, journal);
+
for (u64 seq = journal_last_unwritten_seq(j);
seq <= journal_cur_seq(j);
seq++) {
@@ -196,6 +198,7 @@ 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_get(&c->cl);
closure_call(&w->io, bch2_journal_write, j->wq, NULL);
}
diff --git a/fs/bcachefs/journal_io.c b/fs/bcachefs/journal_io.c
index 2835250a14c4..47224666d07e 100644
--- a/fs/bcachefs/journal_io.c
+++ b/fs/bcachefs/journal_io.c
@@ -1820,6 +1820,8 @@ static CLOSURE_CALLBACK(journal_write_done)
if (do_discards)
bch2_do_discards(c);
+
+ closure_put(&c->cl);
}
static void journal_write_endio(struct bio *bio)
diff --git a/fs/bcachefs/move.c b/fs/bcachefs/move.c
index 54dd6fec81db..84a228c42f06 100644
--- a/fs/bcachefs/move.c
+++ b/fs/bcachefs/move.c
@@ -330,7 +330,7 @@ int bch2_move_extent(struct moving_context *ctxt,
{
struct btree_trans *trans = ctxt->trans;
struct bch_fs *c = trans->c;
- int ret = -ENOMEM;
+ int ret = 0;
if (trace_io_move_enabled())
trace_io_move2(c, k, &io_opts, &data_opts);
@@ -351,11 +351,10 @@ int bch2_move_extent(struct moving_context *ctxt,
struct moving_io *io = allocate_dropping_locks(trans, ret,
kzalloc(sizeof(struct moving_io), _gfp));
- if (!io)
- goto err;
-
+ if (!io && !ret)
+ ret = bch_err_throw(c, ENOMEM_move_extent);
if (ret)
- goto err_free;
+ goto err;
INIT_LIST_HEAD(&io->io_list);
io->write.ctxt = ctxt;
@@ -366,7 +365,7 @@ int bch2_move_extent(struct moving_context *ctxt,
ret = bch2_data_update_init(trans, iter, ctxt, &io->write, ctxt->wp,
&io_opts, data_opts, iter->btree_id, k);
if (ret)
- goto err_free;
+ goto err;
io->write.op.end_io = move_write_done;
} else {
@@ -380,7 +379,7 @@ int bch2_move_extent(struct moving_context *ctxt,
ret = bch2_data_update_bios_init(&io->write, c, &io_opts);
if (ret)
- goto err_free;
+ goto err;
}
io->write.rbio.bio.bi_end_io = move_read_endio;
@@ -423,9 +422,8 @@ int bch2_move_extent(struct moving_context *ctxt,
BCH_READ_last_fragment,
data_opts.scrub ? data_opts.read_dev : -1);
return 0;
-err_free:
- kfree(io);
err:
+ kfree(io);
if (bch2_err_matches(ret, EROFS) ||
bch2_err_matches(ret, BCH_ERR_transaction_restart))
return ret;
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)
diff --git a/mm/shrinker.c b/mm/shrinker.c
index cbadd4d4437c..c94eedf2cfd8 100644
--- a/mm/shrinker.c
+++ b/mm/shrinker.c
@@ -833,7 +833,6 @@ void shrinker_to_text(struct seq_buf *out, struct shrinker *shrinker)
};
unsigned long nr_freed = atomic_long_read(&shrinker->objects_freed);
-
seq_buf_puts(out, shrinker->name);
seq_buf_putc(out, '\n');
generated by cgit v1.2.3 (git 2.46.0) at 2025-07-20 07:49:20 +0000