Bug 200644

Summary: [REGRESSION] run-webkit-tests: No PID defined when searching for simulator crashlogs
Product: WebKit Reporter: Jonathan Bedard <jbedard>
Component: Tools / TestsAssignee: Jonathan Bedard <jbedard>
Status: RESOLVED FIXED    
Severity: Normal CC: aakash_jain, ap, commit-queue, ews-watchlist, glenn, Hironori.Fujii, webkit-bug-importer
Priority: P2 Keywords: InRadar
Version: WebKit Nightly Build   
Hardware: Unspecified   
OS: Unspecified   
Attachments:
Description Flags
Patch none

Description Jonathan Bedard 2019-08-12 14:43:12 PDT
This is a regression from <https://trac.webkit.org/changeset/245544/webkit>. The SimulatorProcess inherits from ServerProcess, so it also needs to define _system_pid. We didn't notice this before because you need the same process to crash multiple times in a given test run for this to actually be an issue.
Comment 1 Jonathan Bedard 2019-08-12 14:55:25 PDT
Created attachment 376096 [details]
Patch
Comment 2 Aakash Jain 2019-08-12 15:29:27 PDT
Comment on attachment 376096 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=376096&action=review

> Tools/Scripts/webkitpy/port/simulator_process.py:97
> +        self._system_pid = self._pid

Where/how is this _system_pid used?
Comment 3 Aakash Jain 2019-08-12 15:30:59 PDT
> Where/how is this _system_pid used?
Just noticed your comment 'SimulatorProcess inherits from ServerProcess'.

That makes sense. Can you please copy/paste the exception which this patch is fixing.
Comment 4 Fujii Hironori 2019-08-12 15:54:26 PDT
Thanks. LGTM.
Comment 5 Jonathan Bedard 2019-08-12 16:17:26 PDT
(In reply to Aakash Jain from comment #3)
> > Where/how is this _system_pid used?
> Just noticed your comment 'SimulatorProcess inherits from ServerProcess'.
> 
> That makes sense. Can you please copy/paste the exception which this patch
> is fixing.

Not really an exception (that's what took us so long to notice the problem), <https://trac.webkit.org/changeset/245544/webkit> does a good job of showing how we use system_pid. Basically, we're using it to piece together crashlogs. If it's not defined, we search by process name and pull the first crash log that matches the process name. Nothing in Buildbot shows the failure at the moment, we need a configuration that's exotic enough to cause multiple crashes in a single run, I can't find any examples of that on build.webkit.org.
Comment 6 WebKit Commit Bot 2019-08-12 16:44:01 PDT
Comment on attachment 376096 [details]
Patch

Clearing flags on attachment: 376096

Committed r248583: <https://trac.webkit.org/changeset/248583>
Comment 7 WebKit Commit Bot 2019-08-12 16:44:03 PDT
All reviewed patches have been landed.  Closing bug.
Comment 8 Radar WebKit Bug Importer 2019-08-12 16:45:17 PDT
<rdar://problem/54233104>