8300ffd302
The metairq sample sometimes failed due to stuck because it has a potential race when recording the latencies. The previous solution checks whether the stats.mirq greater than MAX_EVENT. But stats.mirq might also have a race here. So we change to check the msg sequence, its correctness ensured by the kernel's message queue. And also remove the spinlock in previous fix because tests only use 1 CPU, no need to add this. It also improves the fix for #40889, to prevent test failed sporadically on acrn_ehl_crb and ehl_crb. Fix #42874. Signed-off-by: Enjia Mai <enjia.mai@intel.com> |
||
|---|---|---|
| .. | ||
| condition_variables | ||
| metairq_dispatch | ||
| index.rst | ||