Skip to contentSkip to conversation

syscalls/fanotify07: Add stop_children() to cleanup()

PATCH zhaogongyi via ltp 2 participants 2 messages Nov 23, 2021

lore
Share & tools
lore
#0 lore raw

Hi,

I have test the patch through setting "ulimit -n 20" and while true;do ./fanotify07 &;echo "";done

Or just add a tst_brk as follow:

        /*
         * Create and destroy another instance. This may hang if
         * unanswered fanotify events block notification subsystem.
         */
        newfd = setup_instance();

        SAFE_CLOSE(newfd);
        
		/* inject */
		tst_brk(TBROK, "exit abnormally");

        tst_res(TPASS, "second instance destroyed successfully");

For email CC, I have CC to you and Amir yesterday, please check.

Thanks very much!

Gongyi

Quoted 61 lines · Zhao Gongyi
Quoted 5 lines

When we run the testcase simultaneously, and the ulimit of open files is small, the testcase will fail and remain many while(1) tasks in system, it makes the system very stuck.

This patch looks OK to me. Have you taken the time to verify whether this fix addresses the issue you've outlined in the patch description?

FTR, please remember to also explicitly CC those who you'd like to get feedback from. Most of us are exceptionally busy and miss things that are flying around in the many mailing lists.

Quoted 36 lines

---
v2->v3: replace memset() with assignment statement in loop

 testcases/kernel/syscalls/fanotify/fanotify07.c | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

--
2.17.1

testcases/kernel/syscalls/fanotify/fanotify07.c +7 -1
index cc56d9019..8220213b1 100644
@@ -86,13 +86,19 @@ static int stop_children(void)
 	int child_ret;
 	int i, ret = 0;
 
-	for (i = 0; i < MAX_CHILDREN; i++)
+	for (i = 0; i < MAX_CHILDREN; i++) {
+		if (!child_pid[i])
+			continue;
 		SAFE_KILL(child_pid[i], SIGKILL);
+	}
 
 	for (i = 0; i < MAX_CHILDREN; i++) {
+		if (!child_pid[i])
+			continue;
 		SAFE_WAITPID(child_pid[i], &child_ret, 0);
 		if (!WIFSIGNALED(child_ret))
 			ret = 1;
+		child_pid[i] = 0;
 	}
 
 	return ret;

testcases/kernel/syscalls/fanotify/fanotify07.c static void setup(void) · line 203

Quoted 8 lines
testcases/kernel/syscalls/fanotify/fanotify07.c +2 -0
@@ -190,6 +196,8 @@ static void setup(void)
 
 static void cleanup(void)
 {
+	stop_children();
+
 	if (fd_notify > 0)
 		SAFE_CLOSE(fd_notify);
 }

/M

Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp
#1 lore raw

Hi all,

Quoted 6 lines

Hi,

I have test the patch through setting "ulimit -n 20" and while true;do ./fanotify07 &;echo "";done

Or just add a tst_brk as follow:

Quoted 12 lines
        /*
         * Create and destroy another instance. This may hang if
         * unanswered fanotify events block notification subsystem.
         */
        newfd = setup_instance();

        SAFE_CLOSE(newfd);

		/* inject */
		tst_brk(TBROK, "exit abnormally");

        tst_res(TPASS, "second instance destroyed successfully");

OK, thanks for testing! (suppose you've done this for v3). I'm going to merge it today.

Quoted 1 line

For email CC, I have CC to you and Amir yesterday, please check.

You send extra mail on older version with info that you posted new version. More effective is to use --cc parameter on git format-patch or git send-email in new version (fewer mails + people don't have to search for particular email in LTP mailing list mailbox).

Also, FYI we have archive on lore: https://lore.kernel.org/ltp/
(patchwork sometimes some mails).

Kind regards,
Petr

Quoted 3 lines

Thanks very much!

Gongyi

Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp