You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
IPC handler should detect FW crash notifications #537
FW sends an IPC message upon a FW exception notifying host driver an exception has occurred. This is currently ignored by the driver afaict.
[ 9768.126925] sof-audio sof-audio: error: rx list empty but received 0x4200
[ 9768.126927] sof-audio sof-audio: error: can't find message header 0x1004200
This IPC interrupts the host and sends the platform specific offset of the exception log. This should probably be handled in a platform specific detector that could call generic core code to reboot FW etc.
even if specific 0x4200 is not a problem, the point still stands - say
if there is a FW exception like panic, can we reboot the FW?
That's called firmware recovery, where the driver takes the decision to
reset and restart when the DSP panics or there is a watchdog timeout -
and there is no support for it yet. It's a known need that's documented
in issues
#452
That's called firmware recovery, where the driver takes the decision to reset and restart when the DSP panics or there is a watchdog timeout - and there is no support for it yet. It's a known need that's documented in issues #452
Yes, sir, tried to find this issue number to link it, but didn't get it. Thank you!
Activity
xiulipan commentedon Jan 15, 2019
@lgirdwood
This is not exception IPC. It should be purge FW request reply from ROM
@keyonjie found this IPCs are coming from
linux/sound/soc/sof/intel/hda-loader.c
Line 94 in 8ba1140
linux/sound/soc/sof/intel/hda.h
Line 193 in 8ba1140
As there is a
<< 9so you will always see odd message like0x1004200and0x1004400@keyonjie
Please check if I am right about above explanation.
PS: I remember there are fix about this?
keyonjie commentedon Jan 15, 2019
this is actually only a warning, please ignore it, I can submit a patch to remove this warning message later.
It is 4200(stream_tag=2) here because stream_tag 1(4000) is occupied by the playback stream during suspend.
wenqingfu commentedon Feb 2, 2019
even if specific 0x4200 is not a problem, the point still stands - say if there is a FW exception like panic, can we reboot the FW?
plbossart commentedon Feb 2, 2019
wenqingfu commentedon Feb 2, 2019
Yes, sir, tried to find this issue number to link it, but didn't get it. Thank you!
xiulipan commentedon Feb 10, 2019
@lgirdwood I think we have the panic ipc handler inside kernel now:
linux/sound/soc/sof/intel/hda-ipc.c
Lines 194 to 201 in 66ca36b
Maybe we can close this issue? And continue the FW recovery method in #452
plbossart commentedon Mar 24, 2019
@lgirdwood is this issue still current or can it be closed?
plbossart commentedon Jun 16, 2021
closing, this duplicates part of the functionality needed described by #452
Merge branch 'bpf-arm64-indirect-jumps'