Aksal/nuttx-apps
1
0
Fork 1
mirror of https://github.com/apache/nuttx-apps.git synced 2026-08-01 20:29:00 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

ostest/hrtimer: Increase the tolerent latency.

Browse source 
For QEMU, virtual CPUs can be preempted by any high priority thread in
test environments. This can cause the timer to be triggered later than
expected. This commit increased the tolerant latency to avoid the test
failures.

Signed-off-by: ouyangxiangzhen <ouyangxiangzhen@xiaomi.com>
This commit is contained in:
ouyangxiangzhen 2026-02-02 16:17:22 +08:00 • committed by Matteo Golin
parent 0bbf5dc092
commit d33830e83f
1 changed files with 9 additions and 5 deletions
Download patch file Download diff file Expand all files Collapse all files

14
testing/ostest/hrtimer.c
View file

@ -45,18 +45,22 @@
#define HRTIMER_THREAD_LOOP_NR 50
#define HRTIMER_TEST_THREAD_NR (CONFIG_SMP_NCPUS * 5)
/* Set a 1ms latency to allow hrtimertest to pass in QEMU.
/* Set the tolerent latency to 10ms to allow hrtimer_test to pass
* in QEMU.
*
* QEMU is a virtual platform, and its timer resolution and scheduling
* latency may be less precise than on real hardware. Using a larger
* latency ensures that tests do not fail due to timing inaccuracies.
* QEMU is a virtual platform, vCPUs can be preempted by any
* high priority thread. This can cause the timer to be triggered
* later than expected. This is especially true on QEMU because it
* has a lot of overhead. The timer resolution is also less precise
* than on real hardware. Using a larger latency ensures that tests
* do not fail due to timing inaccuracies.
*
* On real hardware (verified on the a2g-tc397-5v-tft board), this
* latency can be reduced to less than 5 ns because timers are precise
* and deterministic.
*/
#define HRTIMER_TEST_TOLERENT_LATENCY (NSEC_PER_MSEC)
#define HRTIMER_TEST_TOLERENT_LATENCY (10 * NSEC_PER_MSEC)
/****************************************************************************
* Private Types