Bug 13302

Summary: valgrind ptest results are inconsistent per image
Product: [Build System, Metadata & Runtime] OE-Core Reporter: Joshua Watt <JPEWhacker>
Component: coreAssignee: Randy MacLeod <randy.macleod>
Status: RESOLVED FIXED QA Contact:
Severity: normal    
Priority: Medium+ CC: alex.kanavin, meta.mr.watcher, meta.watcher, randy.macleod
Version: 2.7   
Target Milestone: 2.8 M3   
Hardware: x86   
OS: Multiple   
Whiteboard:
OS type for building Yocto: --- Type of Regression: ---
Verified: Documentation change: No (bug/feature does not impact docs)
Attachments:
Description Flags
ptest log from core-image-sato build
none
ptest log from core-image-sato-sdk build
none
ptest log from core-image-minimal build none

Description Joshua Watt 2019-04-18 18:43:13 UTC 
valgrind is generating different ptest results between the core-image-sato and core-image-sato-sdk images. Specifically:

core-image-sato: https://autobuilder.yocto.io/pub/non-release/20190417-10/testresults/qemux86-64-ptest/testresults.json

Recipe                       | Passed      | Failed   | Skipped   | 
valgrind                     | 0           | 0        | 0         | 

core-image-sato-sdk: https://autobuilder.yocto.io/pub/non-release/20190417-14/testresults/qemux86-64-ptest/testresults.json

Recipe                       | Passed      | Failed   | Skipped   | 
valgrind                     | 297         | 251      | 31        | 

Not that no tests are reported at all in core-image-sato
Comment 1 Joshua Watt 2019-04-18 18:43:52 UTC 
Created attachment 4507 [details]
ptest log from core-image-sato build
Comment 2 Joshua Watt 2019-04-18 18:44:06 UTC 
Created attachment 4508 [details]
ptest log from core-image-sato-sdk build
Comment 3 Trevor Gamblin 2019-08-23 13:19:16 UTC 
Created attachment 4557 [details]
ptest log from core-image-minimal build
Comment 4 Randy MacLeod 2019-09-05 22:57:35 UTC 
I've done the work to make the tests independent of the image. 
I've also enable ~500 more tests, most of which pass. On qemuarm64, some tests are causing valgrind to segfault and since the valgrind test perl script doesn't reap defunct processes, it eventually causes the oom-killer to run. I'm working on fixing the arm64 bug or if that takes too long, I'll skip the problematic tests and send the patches that I have.
Comment 6 Randy MacLeod 2019-09-17 00:52:09 UTC 
Fixed by:
  http://git.openembedded.org/openembedded-core/commit/?id=6dbaaeec17eae8329031188b688b33306a871870
and the two previous commits.

Test results from master-next that includes this work:
https://autobuilder.yocto.io/pub/non-release/20190916-9/testresults/testresult-report.txt

There is more work to do for arm64 and likely other arches. The arm64 problems are tracked by:
   https://bugzilla.yoctoproject.org/show_bug.cgi?id=13520