With overlayfs test case on kernel v4.19, we get extra events.
If we do not close the event->fd file descriptros of all the events
that we read, overlayfs will fail to unmount at the end of the test
and test will retry un-mounting for a while.
Signed-off-by: Amir Goldstein <amir73il@xxxxxxxxx>
---
.../kernel/syscalls/fanotify/fanotify06.c | 18 ++++++++++++++----
1 file changed, 14 insertions(+), 4 deletions(-)
diff --git a/testcases/kernel/syscalls/fanotify/fanotify06.c b/testcases/kernel/syscalls/fanotify/fanotify06.c
index e053da0e5..273c1f0c4 100644
--- a/testcases/kernel/syscalls/fanotify/fanotify06.c
+++ b/testcases/kernel/syscalls/fanotify/fanotify06.c
@@ -148,6 +148,17 @@ static void verify_event(int group, struct fanotify_event_metadata *event)
}
}
+/* Close all file descriptors of read events */
+static void close_events_fd(struct fanotify_event_metadata *event, int buflen)
+{
+ while (buflen >= (int)FAN_EVENT_METADATA_LEN) {
+ if (event->fd != FAN_NOFD)
+ SAFE_CLOSE(event->fd);
+ buflen -= (int)FAN_EVENT_METADATA_LEN;
+ event++;
+ }
+}
+
void test_fanotify(unsigned int n)
{
int ret;
@@ -198,17 +209,16 @@ void test_fanotify(unsigned int n)
} else {
verify_event(i, event);
}
- if (event->fd != FAN_NOFD)
- SAFE_CLOSE(event->fd);
+ close_events_fd(event, ret);
}
+
for (p = 1; p < FANOTIFY_PRIORITIES; p++) {
for (i = 0; i < GROUPS_PER_PRIO; i++) {
ret = read(fd_notify[p][i], event_buf, EVENT_BUF_LEN);
if (ret > 0) {
tst_res(TFAIL, "group %d got event",
p*GROUPS_PER_PRIO + i);
- if (event->fd != FAN_NOFD)
- SAFE_CLOSE(event->fd);
+ close_events_fd((void *)event_buf, ret);
} else if (ret == 0) {
tst_brk(TBROK, "zero length "
"read from fanotify fd");
--
2.17.1
