kernel: Update k_wakeup() #69255
kernel: Update k_wakeup() #69255
Conversation
This commit does two things to k_wakeup(): 1. It locks the scheduler before marking the thread as not suspended. As the the clearing of the _THREAD_SUSPENDED bit is not atomic, this helps ensure that neither another thread nor ISR interrupts this action (resulting in a corrupted thread_state). 2. The call to flag_ipi() has been removed as it is already being made within ready_thread(). Signed-off-by: Peter Mitsis <peter.mitsis@intel.com>
There was a problem hiding this comment.
Nice, was about to look at exactly the same bug. Looks correct to me.
Might be worth nothing that a root-ish cause of this messup is that (1) k_wakeup() is really obscure and (2) it really should have shared a backend implementation with k_thread_resume(), which is a much more common (and in this case, correct) path.
|
FYI @peter-mitsis @ceolin -- we saw new cases of "__ASSERT_NO_MSG(!z_smp_cpu_mobile()" firing form zephyr/kernel/pm.c:pm_system_suspend(). Example fail log https://sof-ci.01.org/sofpr/PR8901/build3107/devicetest/index.html?model=MTLP_RVP_NOCODEC&testcase=check-playback-10sec . FYI to @lyakh as well. We are debugging this in context of thesofproject/sof#8818 |
This commit does two things to k_wakeup():
It locks the scheduler before marking the thread as not suspended. As the the clearing of the _THREAD_SUSPENDED bit is not atomic, this helps ensure that neither another thread nor ISR interrupts this action (resulting in a corrupted thread_state).
The call to flag_ipi() has been removed as it is already being made within ready_thread().
Fixes #69148