Hi Olivier,
On Fri, Aug 1, 2025 at 5:39 AM Olivier Blin olivier.blin@softathome.com wrote:
Shouldn't you also remove the timerfd variable? It seems to be of no functional use after this change.
The timerfd is still required because it provides the wake-up event for this test. Just before suspend the code programs a CLOCK_BOOTTIME_ALARM to expire in five seconds. While the system is asleep it needs an interrupt to resume; this timer supplies it. The patch only changes how success is checked: we read /sys/power/suspend_stats/success after resume, instead of comparing time diffs on the timerfd. If the timerfd were removed, the machine could suspend with no scheduled event to bring it back, turning the test into a hang rather than yielding a result.
Thanks for the review.
Regards, Moon