aboutsummaryrefslogtreecommitdiff
path: root/src/backend/storage/aio/aio.c
diff options
context:
space:
mode:
authorAndres Freund <andres@anarazel.de>2025-03-18 14:40:05 -0400
committerAndres Freund <andres@anarazel.de>2025-03-26 19:49:13 -0400
commitc325a7633fcb33dbd73f46ddbbe91e95ddf3b227 (patch)
tree0aa7ab5b32bd5dd26eef0aff6b4d4b200a771d75 /src/backend/storage/aio/aio.c
parent8eadd5c73c44708ecd45b9fd3ac54a550511d16f (diff)
downloadpostgresql-c325a7633fcb33dbd73f46ddbbe91e95ddf3b227.tar.gz
postgresql-c325a7633fcb33dbd73f46ddbbe91e95ddf3b227.zip
aio: Add io_method=io_uring
Performing AIO using io_uring can be considerably faster than io_method=worker, particularly when lots of small IOs are issued, as a) the context-switch overhead for worker based AIO becomes more significant b) the number of IO workers can become limiting io_uring, however, is linux specific and requires an additional compile-time dependency (liburing). This implementation is fairly simple and there are substantial optimization opportunities. The description of the existing AIO_IO_COMPLETION wait event is updated to make the difference between it and the new AIO_IO_URING_EXECUTION clearer. Reviewed-by: Noah Misch <noah@leadboat.com> Reviewed-by: Jakub Wartak <jakub.wartak@enterprisedb.com> Discussion: https://postgr.es/m/uvrtrknj4kdytuboidbhwclo4gxhswwcpgadptsjvjqcluzmah%40brqs62irg4dt Discussion: https://postgr.es/m/20210223100344.llw5an2aklengrmn@alap3.anarazel.de Discussion: https://postgr.es/m/stj36ea6yyhoxtqkhpieia2z4krnam7qyetc57rfezgk4zgapf@gcnactj4z56m
Diffstat (limited to 'src/backend/storage/aio/aio.c')
-rw-r--r--src/backend/storage/aio/aio.c41
1 files changed, 41 insertions, 0 deletions
diff --git a/src/backend/storage/aio/aio.c b/src/backend/storage/aio/aio.c
index 116bf97d3ef..91e76113412 100644
--- a/src/backend/storage/aio/aio.c
+++ b/src/backend/storage/aio/aio.c
@@ -65,6 +65,9 @@ static void pgaio_io_wait(PgAioHandle *ioh, uint64 ref_generation);
const struct config_enum_entry io_method_options[] = {
{"sync", IOMETHOD_SYNC, false},
{"worker", IOMETHOD_WORKER, false},
+#ifdef IOMETHOD_IO_URING_ENABLED
+ {"io_uring", IOMETHOD_IO_URING, false},
+#endif
{NULL, 0, false}
};
@@ -82,6 +85,9 @@ PgAioBackend *pgaio_my_backend;
static const IoMethodOps *const pgaio_method_ops_table[] = {
[IOMETHOD_SYNC] = &pgaio_sync_ops,
[IOMETHOD_WORKER] = &pgaio_worker_ops,
+#ifdef IOMETHOD_IO_URING_ENABLED
+ [IOMETHOD_IO_URING] = &pgaio_uring_ops,
+#endif
};
/* callbacks for the configured io_method, set by assign_io_method */
@@ -1118,6 +1124,41 @@ pgaio_closing_fd(int fd)
* it's probably not worth it.
*/
pgaio_submit_staged();
+
+ /*
+ * If requested by the IO method, wait for all IOs that use the
+ * to-be-closed FD.
+ */
+ if (pgaio_method_ops->wait_on_fd_before_close)
+ {
+ /*
+ * As waiting for one IO to complete may complete multiple IOs, we
+ * can't just use a mutable list iterator. The maximum number of
+ * in-flight IOs is fairly small, so just restart the loop after
+ * waiting for an IO.
+ */
+ while (!dclist_is_empty(&pgaio_my_backend->in_flight_ios))
+ {
+ dlist_iter iter;
+ PgAioHandle *ioh = NULL;
+
+ dclist_foreach(iter, &pgaio_my_backend->in_flight_ios)
+ {
+ ioh = dclist_container(PgAioHandle, node, iter.cur);
+
+ if (pgaio_io_uses_fd(ioh, fd))
+ break;
+ else
+ ioh = NULL;
+ }
+
+ if (!ioh)
+ break;
+
+ /* see comment in pgaio_io_wait_for_free() about raciness */
+ pgaio_io_wait(ioh, ioh->generation);
+ }
+ }
}
/*