Automated rotation test failed, but can rotate it manually

Bug #1172667 reported by Po-Hsu Lin
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
Checkbox
Fix Released
Medium
Alberto Milone

Bug Description

CID: 201206-11410 (HP Pavilion 23 AIO)

The rotation test will be stuck at the first rotation (left), if the nVidia proprietary driver was in use.

Step:
1. Install the nVidia proprietary driver and reboot.
2. Run the rotation test.

Expected Result:
The screen will rotate through all four directions, and go back to normal at the end.

Actual Result:
The screen will rotate to "left" and stayed there, won't back to normal at the end. Test Failed.
But I can freely to rotate it with xrandr -o "directions" or by using the Display utility.
This make me wondering if this is a bug of checkbox.

Note:
1. Since we don't have any available systems that with a same video card (GeForce 610M) here, unable to get results for cross-references.
System list: https://certification.canonical.com/devices/26959/
2. Tried with all three drivers (current / current-update / experimental) and the one downloaded the the nVidia website.
3. BTW, why we're using the xrandr from checkbox.contrib instead of the xrandr -o command to do the rotation directly?

Related branches

Revision history for this message
Po-Hsu Lin (cypressyew) wrote :
Revision history for this message
Po-Hsu Lin (cypressyew) wrote :
Revision history for this message
Po-Hsu Lin (cypressyew) wrote :
Revision history for this message
Brendan Donegan (brendan-donegan) wrote :

"BTW, why we're using the xrandr from checkbox.contrib instead of the xrandr -o command to do the rotation directly?"

This is a good question - this module was actually contributed by someone from outside the team, so we should try and find out what their rationale was.

Changed in checkbox:
assignee: Checkbox Developers (checkbox-dev) → nobody
status: New → Confirmed
importance: Undecided → Medium
Revision history for this message
Brendan Donegan (brendan-donegan) wrote :

I'll have a look at this for now though

Changed in checkbox:
assignee: nobody → Brendan Donegan (brendan-donegan)
status: Confirmed → In Progress
Revision history for this message
Jeff Lane  (bladernr) wrote :

I know there was a reason, but you'll want to ask Alberto Milone, about that one. He had reasons for wrapping the xrandr calls in a library, but I don't recall now what those were.

Changed in checkbox:
status: In Progress → Fix Committed
assignee: Brendan Donegan (brendan-donegan) → Alberto Milone (albertomilone)
Revision history for this message
Brian Sullivan (bsullivan-n) wrote :

This issue is caused by a bug in the NVIDIA driver's timestamp tracking. After the first rotation the eventual call to XRRSetScreenConfigAndRate using the timestamp returned by XRRTimes is incorrectly rejected. Using timestamp 0 or CurrentTime works around issue, or as done here starting over resetting screen also works.

The driver bug is now fixed and will be in a future release.

Changed in checkbox:
status: Fix Committed → Fix Released
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.