summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
Diffstat
-rw-r--r--fs/bcachefs/data_update.c13
-rw-r--r--fs/bcachefs/fsck.c36
-rw-r--r--fs/bcachefs/io_misc.c27
-rw-r--r--fs/bcachefs/io_misc.h2
-rw-r--r--fs/bcachefs/journal_io.c34
-rw-r--r--fs/bcachefs/journal_seq_blacklist.c46
-rw-r--r--fs/bcachefs/journal_seq_blacklist.h3
-rw-r--r--fs/bcachefs/trace.h5
8 files changed, 117 insertions, 49 deletions
diff --git a/fs/bcachefs/data_update.c b/fs/bcachefs/data_update.c
index e848e210a9bf..3968f3be7f3b 100644
--- a/fs/bcachefs/data_update.c
+++ b/fs/bcachefs/data_update.c
@@ -783,6 +783,9 @@ static int can_write_extent(struct bch_fs *c, struct data_update *m)
darray_for_each(m->op.devs_have, i)
__clear_bit(*i, devs.d);
+ CLASS(printbuf, buf)();
+ buf.atomic++;
+
guard(rcu)();
unsigned nr_replicas = 0, i;
@@ -794,7 +797,11 @@ static int can_write_extent(struct bch_fs *c, struct data_update *m)
struct bch_dev_usage usage;
bch2_dev_usage_read_fast(ca, &usage);
- if (!dev_buckets_free(ca, usage, m->op.watermark))
+ u64 nr_free = dev_buckets_free(ca, usage, m->op.watermark);
+
+ prt_printf(&buf, "%s=%llu ", ca->name, nr_free);
+
+ if (!nr_free)
continue;
nr_replicas += ca->mi.durability;
@@ -802,8 +809,10 @@ static int can_write_extent(struct bch_fs *c, struct data_update *m)
break;
}
- if (!nr_replicas)
+ if (!nr_replicas) {
+ trace_data_update_done_no_rw_devs(c, buf.buf);
return bch_err_throw(c, data_update_done_no_rw_devs);
+ }
if (nr_replicas < m->op.nr_replicas)
return bch_err_throw(c, insufficient_devices);
return 0;
diff --git a/fs/bcachefs/fsck.c b/fs/bcachefs/fsck.c
index 00afe0a3593f..471e93a3f00c 100644
--- a/fs/bcachefs/fsck.c
+++ b/fs/bcachefs/fsck.c
@@ -12,6 +12,7 @@
#include "fs.h"
#include "fsck.h"
#include "inode.h"
+#include "io_misc.h"
#include "keylist.h"
#include "namei.h"
#include "recovery_passes.h"
@@ -1637,7 +1638,8 @@ static int check_i_sectors_notnested(struct btree_trans *trans, struct inode_wal
i->count = count2;
}
- if (fsck_err_on(!(i->inode.bi_flags & BCH_INODE_i_sectors_dirty),
+ if (fsck_err_on(!(i->inode.bi_flags & BCH_INODE_i_sectors_dirty) &&
+ i->inode.bi_sectors != i->count,
trans, inode_i_sectors_wrong,
"inode %llu:%u has incorrect i_sectors: got %llu, should be %llu",
w->last_pos.inode, i->inode.bi_snapshot,
@@ -1963,33 +1965,11 @@ static int check_extent(struct btree_trans *trans, struct btree_iter *iter,
"extent type past end of inode %llu:%u, i_size %llu\n%s",
i->inode.bi_inum, i->inode.bi_snapshot, i->inode.bi_size,
(bch2_bkey_val_to_text(&buf, c, k), buf.buf))) {
- struct bkey_i *whiteout = bch2_trans_kmalloc(trans, sizeof(*whiteout));
- ret = PTR_ERR_OR_ZERO(whiteout);
- if (ret)
- goto err;
-
- bkey_init(&whiteout->k);
- whiteout->k.p = SPOS(k.k->p.inode,
- last_block,
- i->inode.bi_snapshot);
- bch2_key_resize(&whiteout->k,
- min(KEY_SIZE_MAX & (~0 << c->block_bits),
- U64_MAX - whiteout->k.p.offset));
-
-
- /*
- * Need a normal (not BTREE_ITER_all_snapshots)
- * iterator, if we're deleting in a different
- * snapshot and need to emit a whiteout
- */
- struct btree_iter iter2;
- bch2_trans_iter_init(trans, &iter2, BTREE_ID_extents,
- bkey_start_pos(&whiteout->k),
- BTREE_ITER_intent);
- ret = bch2_btree_iter_traverse(trans, &iter2) ?:
- bch2_trans_update(trans, &iter2, whiteout,
- BTREE_UPDATE_internal_snapshot_node);
- bch2_trans_iter_exit(trans, &iter2);
+ ret = bch2_fpunch_snapshot(trans,
+ SPOS(i->inode.bi_inum,
+ last_block,
+ i->inode.bi_snapshot),
+ POS(i->inode.bi_inum, U64_MAX));
if (ret)
goto err;
diff --git a/fs/bcachefs/io_misc.c b/fs/bcachefs/io_misc.c
index bf72b1d2e2cb..07023667a475 100644
--- a/fs/bcachefs/io_misc.c
+++ b/fs/bcachefs/io_misc.c
@@ -135,6 +135,33 @@ err_noprint:
return ret;
}
+/* For fsck */
+int bch2_fpunch_snapshot(struct btree_trans *trans, struct bpos start, struct bpos end)
+{
+ u32 restart_count = trans->restart_count;
+ struct bch_fs *c = trans->c;
+ struct disk_reservation disk_res = bch2_disk_reservation_init(c, 0);
+ unsigned max_sectors = KEY_SIZE_MAX & (~0 << c->block_bits);
+ struct bkey_i delete;
+
+ int ret = for_each_btree_key_max_commit(trans, iter, BTREE_ID_extents,
+ start, end, 0, k,
+ &disk_res, NULL, BCH_TRANS_COMMIT_no_enospc, ({
+ bkey_init(&delete.k);
+ delete.k.p = iter.pos;
+
+ /* create the biggest key we can */
+ bch2_key_resize(&delete.k, max_sectors);
+ bch2_cut_back(end, &delete);
+
+ bch2_extent_trim_atomic(trans, &iter, &delete) ?:
+ bch2_trans_update(trans, &iter, &delete, 0);
+ }));
+
+ bch2_disk_reservation_put(c, &disk_res);
+ return ret ?: trans_was_restarted(trans, restart_count);
+}
+
/*
* Returns -BCH_ERR_transacton_restart if we had to drop locks:
*/
diff --git a/fs/bcachefs/io_misc.h b/fs/bcachefs/io_misc.h
index 9cb44a7c43c1..b93e4d4b3c0c 100644
--- a/fs/bcachefs/io_misc.h
+++ b/fs/bcachefs/io_misc.h
@@ -5,6 +5,8 @@
int bch2_extent_fallocate(struct btree_trans *, subvol_inum, struct btree_iter *,
u64, struct bch_io_opts, s64 *,
struct write_point_specifier);
+
+int bch2_fpunch_snapshot(struct btree_trans *, struct bpos, struct bpos);
int bch2_fpunch_at(struct btree_trans *, struct btree_iter *,
subvol_inum, u64, s64 *);
int bch2_fpunch(struct bch_fs *c, subvol_inum, u64, u64, s64 *);
diff --git a/fs/bcachefs/journal_io.c b/fs/bcachefs/journal_io.c
index f3cf48193398..2d6ce4348a22 100644
--- a/fs/bcachefs/journal_io.c
+++ b/fs/bcachefs/journal_io.c
@@ -1245,6 +1245,8 @@ noinline_for_stack
static void bch2_journal_print_checksum_error(struct bch_fs *c, struct journal_replay *j)
{
struct printbuf buf = PRINTBUF;
+ bch2_log_msg_start(c, &buf);
+
enum bch_csum_type csum_type = JSET_CSUM_TYPE(&j->j);
bool have_good = false;
@@ -1279,20 +1281,14 @@ struct u64_range bch2_journal_entry_missing_range(struct bch_fs *c, u64 start, u
if (start == end)
return (struct u64_range) {};
- while (start < end &&
- bch2_journal_seq_is_blacklisted(c, start, false))
- start++;
-
- if (start == end)
+ start = bch2_journal_seq_next_nonblacklisted(c, start);
+ if (start >= end)
return (struct u64_range) {};
- struct u64_range missing = { .start = start };
-
- while (start < end &&
- !bch2_journal_seq_is_blacklisted(c, start, false))
- start++;
-
- missing.end = start - 1;
+ struct u64_range missing = {
+ .start = start,
+ .end = min(end, bch2_journal_seq_next_blacklisted(c, start)),
+ };
if (missing.start == missing.end)
return (struct u64_range) {};
@@ -1308,6 +1304,7 @@ static int bch2_journal_check_for_missing(struct bch_fs *c, u64 start_seq, u64 e
struct genradix_iter radix_iter;
struct journal_replay *i, **_i, *prev = NULL;
+ /* Sequence number we expect to find next, to check for missing entries */
u64 seq = start_seq;
genradix_for_each(&c->journal_entries, radix_iter, _i) {
@@ -1323,27 +1320,26 @@ static int bch2_journal_check_for_missing(struct bch_fs *c, u64 start_seq, u64 e
while ((missing = bch2_journal_entry_missing_range(c, seq, le64_to_cpu(i->j.seq))).start) {
printbuf_reset(&buf);
prt_printf(&buf, "journal entries %llu-%llu missing! (replaying %llu-%llu)",
- missing.start, missing.end,
+ missing.start, missing.end - 1,
start_seq, end_seq);
- prt_printf(&buf, "\nprev at ");
if (prev) {
+ prt_printf(&buf, "\n%llu at ", le64_to_cpu(prev->j.seq));
bch2_journal_ptrs_to_text(&buf, c, prev);
prt_printf(&buf, " size %zu", vstruct_sectors(&prev->j, c->block_bits));
- } else
- prt_printf(&buf, "(none)");
+ }
- prt_printf(&buf, "\nnext at ");
+ prt_printf(&buf, "\n%llu at ", le64_to_cpu(i->j.seq));
bch2_journal_ptrs_to_text(&buf, c, i);
prt_printf(&buf, ", continue?");
fsck_err(c, journal_entries_missing, "%s", buf.buf);
- seq = missing.end + 1;
+ seq = missing.end;
}
prev = i;
- seq++;
+ seq = le64_to_cpu(i->j.seq) + 1;
}
fsck_err:
printbuf_exit(&buf);
diff --git a/fs/bcachefs/journal_seq_blacklist.c b/fs/bcachefs/journal_seq_blacklist.c
index af4fe416d9ec..6361809b5e2e 100644
--- a/fs/bcachefs/journal_seq_blacklist.c
+++ b/fs/bcachefs/journal_seq_blacklist.c
@@ -103,6 +103,52 @@ static int journal_seq_blacklist_table_cmp(const void *_l, const void *_r)
return cmp_int(l->start, r->start);
}
+static int journal_seq_blacklist_table_end_cmp(const void *_l, const void *_r)
+{
+ const struct journal_seq_blacklist_table_entry *l = _l;
+ const struct journal_seq_blacklist_table_entry *r = _r;
+
+ return cmp_int(l->end, r->end);
+}
+
+u64 bch2_journal_seq_next_blacklisted(struct bch_fs *c, u64 seq)
+{
+ struct journal_seq_blacklist_table *t = c->journal_seq_blacklist_table;
+
+ if (!t)
+ return U64_MAX;
+
+ struct journal_seq_blacklist_table_entry search = { .end = seq };
+ int idx = eytzinger0_find_gt(t->entries, t->nr,
+ sizeof(t->entries[0]),
+ journal_seq_blacklist_table_end_cmp,
+ &search);
+ if (idx < 0)
+ return U64_MAX;
+
+ return max(seq, t->entries[idx].start);
+}
+
+u64 bch2_journal_seq_next_nonblacklisted(struct bch_fs *c, u64 seq)
+{
+ struct journal_seq_blacklist_table *t = c->journal_seq_blacklist_table;
+
+ if (!t)
+ return seq;
+
+ while (true) {
+ struct journal_seq_blacklist_table_entry search = { .start = seq };
+ int idx = eytzinger0_find_le(t->entries, t->nr,
+ sizeof(t->entries[0]),
+ journal_seq_blacklist_table_cmp,
+ &search);
+ if (idx < 0 || t->entries[idx].end <= seq)
+ return seq;
+
+ seq = t->entries[idx].end;
+ }
+}
+
bool bch2_journal_seq_is_blacklisted(struct bch_fs *c, u64 seq,
bool dirty)
{
diff --git a/fs/bcachefs/journal_seq_blacklist.h b/fs/bcachefs/journal_seq_blacklist.h
index f06942ccfcdd..389b789b26f4 100644
--- a/fs/bcachefs/journal_seq_blacklist.h
+++ b/fs/bcachefs/journal_seq_blacklist.h
@@ -11,6 +11,9 @@ blacklist_nr_entries(struct bch_sb_field_journal_seq_blacklist *bl)
: 0;
}
+u64 bch2_journal_seq_next_blacklisted(struct bch_fs *, u64);
+u64 bch2_journal_seq_next_nonblacklisted(struct bch_fs *, u64);
+
bool bch2_journal_seq_is_blacklisted(struct bch_fs *, u64, bool);
u64 bch2_journal_last_blacklisted_seq(struct bch_fs *);
int bch2_journal_seq_blacklist_add(struct bch_fs *c, u64, u64);
diff --git a/fs/bcachefs/trace.h b/fs/bcachefs/trace.h
index b5dae1145afa..9324ef32903d 100644
--- a/fs/bcachefs/trace.h
+++ b/fs/bcachefs/trace.h
@@ -1330,6 +1330,11 @@ DEFINE_EVENT(fs_str, data_update,
TP_ARGS(c, str)
);
+DEFINE_EVENT(fs_str, data_update_done_no_rw_devs,
+ TP_PROTO(struct bch_fs *c, const char *str),
+ TP_ARGS(c, str)
+);
+
DEFINE_EVENT(fs_str, io_move_pred,
TP_PROTO(struct bch_fs *c, const char *str),
TP_ARGS(c, str)
generated by cgit v1.2.3 (git 2.46.0) at 2025-07-03 03:02:11 +0000