[Skiboot] [PATCH RFC 05/12] opal-prd: implement basic heuristic for hservice_puts loglevels

[Daniel M Crowell]

> be at the beginning of the puts() string?
Yes, those are always at the beginning.

> Would it be worthwhile removing the strcasestr(.., "error"/"warning"),
Yeah, not sure what you'll get with those.

> Or is FAIL_MRK more of a warning? Or a notice ("normal but significant 
condition")?
I've honestly never seen a F> in any trace I've looked at.  I suspect we 
copied these in from the FSP and just never used it.  I would equate FAIL 
with ERR

--
Dan Crowell
Senior Software Engineer - Power Systems Enablement Firmware
IBM Rochester: t/l 553-2987
dcrowell at us.ibm.com



From:   Jeremy Kerr <jk at ozlabs.org>
To:     Daniel M Crowell <dcrowell at us.ibm.com>
Cc:     Frederic Bonnard <FREDERIC at fr.ibm.com>, skiboot at lists.ozlabs.org
Date:   06/03/2017 12:20 AM
Subject:        Re: [PATCH RFC 05/12] opal-prd: implement basic heuristic 
for hservice_puts loglevels



Hi Dan,

>>  - all log output will go to a new log file, /var/log/opal-prd.log
>>
>>  - log messages of LOG_WARNING and LOG_ERROR *also* go to the standard
>>    syslog paths (generally /var/log/syslog).
> 
> I'm good with that.  We just need to get PE to update their FFDC
> collection scripts to grab the new file once it shows up.
> 
> 
> I would suggest you add a couple more cases in your compares.  We've got
> these constant in HB that are used a lot (though not all over) as the
> first 2 characters in our traces. 
> #define ERR_MRK   "E>"
> #define FAIL_MRK  "F>"
> #define WARN_MRK  "W>"
> #define INFO_MRK  "I>"

Oh, that's great then. If one of these is present, is it going to
be at the beginning of the puts() string?

Would it be worthwhile removing the strcasestr(.., "error"/"warning"),
and *just* using these instead? That would prevent the potential for
false classification.

We'd probably map both FAIL_MRK and ERR_MRK to LOG_ERROR. Or is FAIL_MRK
more of a warning? Or a notice ("normal but significant condition")?

Cheers,


Jeremy





-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ozlabs.org/pipermail/skiboot/attachments/20170603/ed098d09/attachment.html>