When flushing the btree write buffer, we pull write buffer keys directly from the journal instead of letting the journal write path copy them to the write buffer. When flushing from the currently open journal buffer, we have to block new reservations and wait for outstanding reservations to complete. Recheck the reservation state after blocking new reservations: previously, we were checking the reservation count from before calling __journal_block().
Unified · Split
| @@ -1082,6 +1082,7 @@ static struct journal_buf *__bch2_next_write_buffer_flush_journal_buf(struct jou · 1081 unmodified lines | |||
| 1082 | 1082 | ||
| 1083 | if (open && !*blocked) { | 1083 | if (open && !*blocked) { |
| 1084 | __bch2_journal_block(j); | 1084 | __bch2_journal_block(j); |
| 1085 | s.v = atomic64_read_acquire(&j->reservations.counter); | ||
| 1085 | *blocked = true; | 1086 | *blocked = true; |
| 1086 | } | 1087 | } |
| 1087 | 1088 | ||