| Summary: | Graphics - ABAT test fails | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Product: | [QA/Testing] Runtime Testing | Reporter: | Cristina Agurida <cristina-danielax.agurida> | ||||||||||
| Component: | general | Assignee: | Saul Wold <sgw> | ||||||||||
| Status: | VERIFIED NOTABUG | QA Contact: | Cristina Agurida <cristina-danielax.agurida> | ||||||||||
| Severity: | normal | ||||||||||||
| Priority: | Medium+ | CC: | alexandru.c.georgescu, georgex.l.musat, yp.kernel.watcher, yp.watcher | ||||||||||
| Version: | 1.9 | ||||||||||||
| Target Milestone: | 2.0 | ||||||||||||
| Hardware: | NUC | ||||||||||||
| OS: | x86_64 | ||||||||||||
| Whiteboard: | |||||||||||||
| OS type for building Yocto: | --- | Type of Regression: | Regression (Used to work) | ||||||||||
| Verified: | Documentation change: | No (bug/feature does not impact docs) | |||||||||||
| Attachments: |
|
||||||||||||
This issue happens also on genericx86, core-image-sato-sdk image on MinnowMax. It appears that X did not start, can you confirm that. Could you provide both the dmesg output and /var/log/Xorg.0.log. I tried both 32 and 64 bit sato-sdk images from the autobuilder and they both succeed to start X and I verified glxinfo and glxgears work 32bit NUC DC3217IYE Intel(R) Core(TM) i3-3217U CPU @ 1.80GHz 64bit NUC D54250WYK Intel(R) Core(TM) i5-4250U CPU @ 1.30GHz Currently, I only have 64bit MinnowMax, did it work correctly on that machine? Created attachment 2753 [details]
screenshot
Created attachment 2754 [details]
Xorg.0.log
In the attached screenshot you can see the message that I receive in terminal after running abat.sh file. Also I attached the dmesg and Xorg.0.log
Created attachment 2755 [details]
abat_dmesg
This issue happens also on 1.9_M3.rc1 meta-intel: cea00968b858b60222d68103491f076067d73876, corei7-64, core-image-sato-sdk image on NUC(model: D54250WYK) It appears that the abat test is attempting to start X on the current display and failing due to X/Sato already running on that display, this will show as a failure, please confirm that you run ABAT under a terminal window and have run it like this before? I can not access the actual abat test script and the testopia test case (269) does not provide the information about where you run it from. The screen shot very clearly shows the Fatal server error that a server is already active for display 0, this will cause a cascade failure for the remaining tests. After some additional testing, the root cause is the "ps" command and it's
difference in output from busybox vs procps.
The standard procps output just lists items attached to a given user / terminal
pair as follows;
PID TTY TIME CMD
568 pts/0 00:00:00 sh
966 pts/0 00:00:00 vi
968 pts/0 00:00:01 glxgears
969 pts/0 00:00:00 ps
The busybox version of ps provides a full listing of all processes (procps "ps
ax")
In x_start.sh it uses ps and looks for the 6th, but with procps you need to
look for the 5th item
note 4 columns for busybox,, abat's glxgears_check.sh uses awk '{print $1,
$5}', which looks at the 5th column, which is true for busybox's version
Some additional checks around which ps is being used and ensure you use one
consistently, for example if procps is found, run the busybox version instead
by calling "busybox ps" instead.
That way the the awk print out can use a consistent column.
After modifying the script as per last comment the tests passed without failures. Set this issue as "resolved notabug" because the problem was with the script. As per last comments set it as verified. |
Created attachment 2738 [details] log STATUS: Failed BUILD: 1.9_M3.rc1: ea562de57590c966cd5a75fda8defecd397e6436 ENVIRONMENT: genericx86-64 on NUC IMAGE: core-image-sato-sdk STEPS TO REPRODUCE: Run abat.sh script. EXPECTED OUTCOME: All ABAT tests should pass. Actual outcome: 3/8 tests failed. Check attached log.