Skip to contentSkip to conversation

A few small fixups to the alarmtimer-suspend test

PATCH In review John Stultz 2 participants 8 messages May 13, 2015 – May 14, 2015

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

cover A few small fixups to the alarmtimer-suspend test

Hey Shuah,
Wanted to send along a few small fixups to the alarmtimer-suspend
test that I had pending.

Let me know if you have any objections or feedback.

thanks
-john

Cc Shuah Khan <[email protected]>
Cc Prarit Bhargava <[email protected]>
Cc Thomas Gleixner <[email protected]>
Cc Richard Cochran <[email protected]>

John Stultz (3):
kselftests: timers: Increase delay between suspends in
alarmtimer-suspend
kselftests: timers: Ease alarmtimer-suspend unreasonable latency value
kselftests: timers: Check _ALARM clockids are supported before
suspending

 tools/testing/selftests/timers/alarmtimer-suspend.c | 10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)
Signature · 1 line
1.9.1
#1 lore raw

PATCH 1/3 kselftests: timers: Increase delay between suspends in alarmtimer-suspend

When testing on some hardware, waiting only a second before re-triggering suspend can keep TCP connections from re-establishing which after a number of cycles can cause TCP connections to close while the test is running.

So extend the delay between suspend calls to 3 seconds to let the connections stay alive.

Cc Shuah Khan <[email protected]>
Cc Prarit Bhargava <[email protected]>
Cc Thomas Gleixner <[email protected]>
Cc Richard Cochran <[email protected]>
---
 tools/testing/selftests/timers/alarmtimer-suspend.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tools/testing/selftests/timers/alarmtimer-suspend.c +1 -1
index aaffbde..99a1ad3 100644
@@ -172,7 +172,7 @@ int main(void)
 		while (alarmcount < 10) {
 			int ret;
 
-			sleep(1);
+			sleep(3);
 			ret = system("echo mem > /sys/power/state");
 			if (ret)
 				break;
Signature · 1 line
1.9.1
#2 lore raw

PATCH 2/3 kselftests: timers: Ease alarmtimer-suspend unreasonable latency value

In reply to John Stultz

On the hardware I have, resume latency from an alarm is often 2-3 seconds (with a fair amount of variability due to the RTC's single second granularity). Having four seconds be the pass/fail bar is maybe a little too tight, so extend this to 5 seconds.

Cc Shuah Khan <[email protected]>
Cc Prarit Bhargava <[email protected]>
Cc Thomas Gleixner <[email protected]>
Cc Richard Cochran <[email protected]>
---
 tools/testing/selftests/timers/alarmtimer-suspend.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tools/testing/selftests/timers/alarmtimer-suspend.c +1 -1
index 99a1ad3..13586b6 100644
@@ -57,7 +57,7 @@ static inline int ksft_exit_fail(void)
 
 
 #define NSEC_PER_SEC 1000000000ULL
-#define UNREASONABLE_LAT (NSEC_PER_SEC * 4) /* hopefully we resume in 4secs */
+#define UNREASONABLE_LAT (NSEC_PER_SEC * 5) /* hopefully we resume in 5 secs */
 
 #define SUSPEND_SECS 15
 int alarmcount;
Signature · 1 line
1.9.1
#3 lore raw

PATCH 3/3 kselftests: timers: Check _ALARM clockids are supported before suspending

In reply to John Stultz

It was reported that the alarmtimer-suspend test hangs on older systems that don't support _ALARM clockids.

This is due to the fact that we don't check if the timer_create fails, and thus when we suspend, the system will not programatically resume.

Fix this by checking the timer_create call for errors.

Cc Shuah Khan <[email protected]>
Cc Prarit Bhargava <[email protected]>
Cc Thomas Gleixner <[email protected]>
Cc Richard Cochran <[email protected]>
---
 tools/testing/selftests/timers/alarmtimer-suspend.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)
tools/testing/selftests/timers/alarmtimer-suspend.c +5 -1
index 13586b6..72cacf5 100644
@@ -152,7 +152,11 @@ int main(void)
 			alarm_clock_id++) {
 
 		alarmcount = 0;
-		timer_create(alarm_clock_id, &se, &tm1);
+		if (timer_create(alarm_clock_id, &se, &tm1) == -1) {
+			printf("timer_create failled, %s unspported?\n",
+					clockstring(alarm_clock_id));
+			break;
+		}
 
 		clock_gettime(alarm_clock_id, &start_time);
 		printf("Start time (%s): %ld:%ld\n", clockstring(alarm_clock_id),
Signature · 1 line
1.9.1
#4 lore raw
In reply to John Stultz
Quoted 6 lines · John Stultz

Hey Shuah,
Wanted to send along a few small fixups to the alarmtimer-suspend
test that I had pending.

Let me know if you have any objections or feedback.

Hi John,

No issues with these. We are tweaking sleep and suspend times. The nature of the beast I am guessing. I gave it test run and looks good. I would like to get these into rc5. Patches 1 and 3 look like critical fixes to me. I will wait a couple of days for any reviews comments before I apply them to linux-kselftest fixes.

thanks,
-- Shuah

Signature · 5 lines
Shuah Khan
Sr. Linux Kernel Developer
Open Source Innovation Group
Samsung Research America (Silicon Valley)
[email protected] | (970) 217-8978
#5 lore raw
Quoted 8 lines · Shuah Khan
Quoted 6 lines · John Stultz

Hey Shuah,
Wanted to send along a few small fixups to the alarmtimer-suspend
test that I had pending.

Let me know if you have any objections or feedback.

Quoted 8 lines

Hi John,

No issues with these. We are tweaking sleep and suspend times. The nature of the beast I am guessing. I gave it test run and looks good. I would like to get these into rc5. Patches 1 and 3 look like critical fixes to me. I will wait a couple of days for any reviews comments before I apply them to linux-kselftest fixes.

I'd not claim any of them are actually urgent. So if you'd rather queue these for 4.2 I'm fine with it.

thanks
-john

#6 lore raw
Quoted 18 lines · John Stultz
Quoted 8 lines · Shuah Khan
Quoted 6 lines · John Stultz

Hey Shuah,
Wanted to send along a few small fixups to the alarmtimer-suspend
test that I had pending.

Let me know if you have any objections or feedback.

Quoted 8 lines

Hi John,

No issues with these. We are tweaking sleep and suspend times. The nature of the beast I am guessing. I gave it test run and looks good. I would like to get these into rc5. Patches 1 and 3 look like critical fixes to me. I will wait a couple of days for any reviews comments before I apply them to linux-kselftest fixes.

Quoted 3 lines

I'd not claim any of them are actually urgent. So if you'd rather queue these for 4.2 I'm fine with it.

ok I will get them into 4.2 then.

thanks,
-- Shuah

Signature · 5 lines
Shuah Khan
Sr. Linux Kernel Developer
Open Source Innovation Group
Samsung Research America (Silicon Valley)
[email protected] | (970) 217-8978
#7 lore raw
In reply to John Stultz
Quoted 18 lines · John Stultz
Quoted 8 lines · Shuah Khan
Quoted 6 lines · John Stultz

Hey Shuah,
Wanted to send along a few small fixups to the alarmtimer-suspend
test that I had pending.

Let me know if you have any objections or feedback.

Quoted 8 lines

Hi John,

No issues with these. We are tweaking sleep and suspend times. The nature of the beast I am guessing. I gave it test run and looks good. I would like to get these into rc5. Patches 1 and 3 look like critical fixes to me. I will wait a couple of days for any reviews comments before I apply them to linux-kselftest fixes.

Quoted 3 lines

I'd not claim any of them are actually urgent. So if you'd rather queue these for 4.2 I'm fine with it.

Applied to linux-kselftest next for 4.2

thanks,
-- Shuah

Signature · 5 lines
Shuah Khan
Sr. Linux Kernel Developer
Open Source Innovation Group
Samsung Research America (Silicon Valley)
[email protected] | (970) 217-8978