bcachefs: Fix BCH_TRANS_COMMIT_skip_accounting_apply

This was added to avoid double-counting accounting keys in journal
replay. But applied incorrectly (easily done since it applies to the
transaction commit, not a particular update), it leads to skipping
in-mem accounting for real accounting updates, and failure to give them
a version number - which leads to journal replay becoming very confused
the next time around.

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

diff --git a/fs/bcachefs/btree_trans_commit.c b/fs/bcachefs/btree_trans_commit.c
index 0989e2da6e56d6519dc756d94bf061d73597ab81..1a74a1a252ee2384b6b35d8a120549d17fe0e697 100644 (file)
--- a/fs/bcachefs/btree_trans_commit.c
+++ b/fs/bcachefs/btree_trans_commit.c
@@ -700,27 +700,31 @@ bch2_trans_commit_write_locked(struct btree_trans *trans, unsigned flags,
 
        struct jset_entry *entry = trans->journal_entries;
 
-       if (likely(!(flags & BCH_TRANS_COMMIT_skip_accounting_apply))) {
-               percpu_down_read(&c->mark_lock);
-
-               for (entry = trans->journal_entries;
-                    entry != (void *) ((u64 *) trans->journal_entries + trans->journal_entries_u64s);
-                    entry = vstruct_next(entry))
-                       if (jset_entry_is_key(entry) && entry->start->k.type == KEY_TYPE_accounting) {
-                               struct bkey_i_accounting *a = bkey_i_to_accounting(entry->start);
-
-                               a->k.bversion = journal_pos_to_bversion(&trans->journal_res,
-                                                               (u64 *) entry - (u64 *) trans->journal_entries);
-                               BUG_ON(bversion_zero(a->k.bversion));
+       percpu_down_read(&c->mark_lock);
+
+       for (entry = trans->journal_entries;
+            entry != (void *) ((u64 *) trans->journal_entries + trans->journal_entries_u64s);
+            entry = vstruct_next(entry))
+               if (entry->type == BCH_JSET_ENTRY_write_buffer_keys &&
+                   entry->start->k.type == KEY_TYPE_accounting) {
+                       BUG_ON(!trans->journal_res.ref);
+
+                       struct bkey_i_accounting *a = bkey_i_to_accounting(entry->start);
+
+                       a->k.bversion = journal_pos_to_bversion(&trans->journal_res,
+                                                       (u64 *) entry - (u64 *) trans->journal_entries);
+                       BUG_ON(bversion_zero(a->k.bversion));
+
+                       if (likely(!(flags & BCH_TRANS_COMMIT_skip_accounting_apply))) {
                                ret = bch2_accounting_mem_mod_locked(trans, accounting_i_to_s_c(a), BCH_ACCOUNTING_normal);
                                if (ret)
                                        goto revert_fs_usage;
                        }
-               percpu_up_read(&c->mark_lock);
+               }
+       percpu_up_read(&c->mark_lock);
 
-               /* XXX: we only want to run this if deltas are nonzero */
-               bch2_trans_account_disk_usage_change(trans);
-       }
+       /* XXX: we only want to run this if deltas are nonzero */
+       bch2_trans_account_disk_usage_change(trans);
 
        trans_for_each_update(trans, i)
                if (btree_node_type_has_atomic_triggers(i->bkey_type)) {