Bug fix: sof-kernel-log-check DSP reset

[Bin-QA]

After patch: thesofproject/linux#1676 merged
add "error: status" and "error: cl_dsp_init" into ignore string
to avoid catch DPS reset which can be ignored

Signed-off-by: Wu, BinX binx.wu@intel.com

[bardliao]

How about changing the message on kernel?
https://github.com/thesofproject/linux/blob/topic/sof-dev/sound/soc/sof/intel/hda-loader.c#L331
Maybe use dev_warn and "warning:" instead of dev_err and "error:".

[Bin-QA]

How about changing the message on kernel?
https://github.com/thesofproject/linux/blob/topic/sof-dev/sound/soc/sof/intel/hda-loader.c#L331
Maybe use dev_warn and "warning:" instead of dev_err and "error:".

 # 0 error: cl_dsp_init
# 1 error: status
# 2 error: Error
# 3 error: iteration

If you want modify the source code those information also need to modify:
Line 0 error: cl_dsp_init
https://github.com/thesofproject/linux/blob/topic/sof-dev/sound/soc/sof/intel/hda-loader.c#L81
Line 1 error: status
https://github.com/thesofproject/linux/blob/topic/sof-dev/sound/soc/sof/intel/hda.c#L204
Line 2 error: Error
https://github.com/thesofproject/linux/blob/topic/sof-dev/sound/soc/sof/intel/hda-loader.c#L331
Line 3 error: iteration
https://github.com/thesofproject/linux/blob/topic/sof-dev/sound/soc/sof/intel/hda-loader.c#L336

Because from the kernel log, catch 4 error message when DSP reset
On the other hand, I format the error message to avoid miss some really error

dsp_count=$(echo "$err"|grep 'error: cl_dsp_init' -c)
allow_count=$(echo "$err"|grep 'error: cl_dsp_init' -A3|grep 'iteration [01]' -c)

@bardliao

[xiulipan]

@bardliao@Bin-QA Let wait for the final discussion about the dsp init message. I think this error is not user friendly.

[wenqingfu]

@bardliao@Bin-QA Let wait for the final discussion about the dsp init message. I think this error is not user friendly.

Which discussion do you refer to @xiulipan ?

I don't see this discussion pattern is finalizing. I agree with @bardliao. Similarly, the unexpected reply shouldn't be called error, either. If they are not real errors, then don't call it error. It freak people out.

small fix, big good, please just act with your patch. @bardliao , can you please fix it?

[bardliao]

@bardliao@Bin-QA Let wait for the final discussion about the dsp init message. I think this error is not user friendly.

Which discussion do you refer to @xiulipan ?

I don't see this discussion pattern is finalizing. I agree with @bardliao. Similarly, the unexpected reply shouldn't be called error, either. If they are not real errors, then don't call it error. It freak people out.

small fix, big good, please just act with your patch. @bardliao , can you please fix it?

Yes, I will send a kernel PR to fix it.

[xiulipan]

@wenqingfu It seems the vacation break the discussion. I also agree with @bardliao that we should not output error when we have retry method.
@ranj063 what is your opinion here, even the S0ix refine thesofproject/linux#1631 may have benefit on this. But the error with retry will still make people get panic.

[wenqingfu]

Thanks, @xiulipan@bardliao . connect the dot with thesofproject/linux#1676 , Seems we have overwhelmingly consensus of "let's don't freak people out" there. :)

[Bin-QA]

Update for kernel patch merged
@xiulipan@libinyang please help to review this patch

[xiulipan]

Let keep this as temp solution here. I think the error will be very misleading in none-debug build.

[stripes416]

This LGTM.

[Bin-QA]

@xiulipan
ignore_str="no reply expected,|error: debugfs write failed to idle -16|error: status |error: cl_dsp_init|iteration [01]"

How about remove 'failed' keyword in err_str
if here just use 'failed' to catch DSP reset

[xiulipan]

@Bin-QA I think there will also be IPC failed, so let keep it to see if we got any other issues.
Please send the patch to the change
ignore_str="no reply expected,|error: debugfs write failed to idle -16|error: status |error: cl_dsp_init|iteration [01]"