Today, all callers of qemu_aio_set_fd_handler() pass a valid io_flush
function. However, the function allows the handler to be omitted
and the behavior is a bit strange.
It will still add the file descriptor to the GSource but it will
not consider the source to be "busy". This could lead to aio_flush()
returning prematurely.
Since we never rely on this behavior today, it doesn't matter but
the next patch will start relying on an absent io_flush function
to assume the handler is always busy.
Cc: Paolo Bonzini <bonz...@redhat.com>
Cc: Mike Roth <mdr...@linux.vnet.ibm.com>
Signed-off-by: Anthony Liguori <aligu...@us.ibm.com>
---
aio-posix.c | 4 ++--
aio-win32.c | 4 ++--
2 files changed, 4 insertions(+), 4 deletions(-)
diff --git a/aio-posix.c b/aio-posix.c
index b68eccd..a2349f6 100644
--- a/aio-posix.c
+++ b/aio-posix.c
@@ -208,8 +208,8 @@ bool aio_poll(AioContext *ctx, bool blocking)
* Otherwise, if there are no AIO requests, qemu_aio_wait() would
* wait indefinitely.
*/
- if (!node->deleted && node->io_flush) {
- if (node->io_flush(node->opaque) == 0) {
+ if (!node->deleted) {
+ if (node->io_flush && node->io_flush(node->opaque) == 0) {
continue;
}
busy = true;
diff --git a/aio-win32.c b/aio-win32.c
index 38723bf..b02fd40 100644
--- a/aio-win32.c
+++ b/aio-win32.c
@@ -154,8 +154,8 @@ bool aio_poll(AioContext *ctx, bool blocking)
* Otherwise, if there are no AIO requests, qemu_aio_wait() would
* wait indefinitely.
*/
- if (!node->deleted && node->io_flush) {
- if (node->io_flush(node->e) == 0) {
+ if (!node->deleted) {
+ if (node->io_flush && node->io_flush(node->e) == 0) {
continue;
}
busy = true;
--
1.8.0
