General requirements gathering for Xorg in Oneiric
General session to discuss how and when new X.org bits will be integrated in Oneiric.
Should we stay with xserver 1.10.x or plan to move to 1.11? With Oneiric+1 being an LTS, it suggests being aggressive at updating to the newest X to maximize testing. On the other hand, if 1.11 brings ABI changes (like 1.10 did) then we risk breaking the proprietary drivers, since they tend to lag their support weeks or months after an X release. Do you have an opinion? Please come.
Other topics for oft-requested features:
* Range of hardware models to provide development support (declining bug reports for ancient hardware)
* Hybrid graphic system stabilization
* Post-release updates for proprietary drivers (SRUs vs. Jockey+x-updates)
* 3D for -nouveau
* Tablet auto-rotate input with screen (https:/
Blueprint information
- Status:
- Complete
- Approver:
- Chris Halse Rogers
- Priority:
- High
- Drafter:
- Bryce Harrington
- Direction:
- Needs approval
- Assignee:
- Chris Halse Rogers
- Definition:
- Approved
- Series goal:
- Accepted for oneiric
- Implementation:
-
Implemented
- Milestone target:
-
ubuntu-11.10-beta-1
- Started by
- Bryce Harrington
- Completed by
- Bryce Harrington
Whiteboard
Decisions:
* xserver 1.10
* Keep 1.11 in xorg-edgers
* Provide a staging ppa with 1.11 and newer kernel to be uploaded once ubuntu-p opens
* Drivers: Most recent stable releases
* Mesa: Most recent stable release. Include bi-weekly git snapshots starting at mesa's rc1
* Opt-in pockets (or overlay PPAs) for LTS point releases (need to flesh out more)
* For some older hardware, stop taking bug reports (redirect them upstream)
* No specific plans for hybrid gpu switching support
Work items for oneiric-alpha-2:
[raof] Organize nouveau testing day at Dublin sprint: DONE
Work items:
[jibel] Arrange community testing for Alpha-2: DONE
[bryce] Can we have apport detect if the gpu hang did not produce debug info, and then offer to the user to turn debugging on (whatever the kernel team comes up with, or adding “drm.debug=0x06” to the kernel command line), prompt to reboot, and then next time send a bug report in: DONE
[tjaalton] Check which input/video drivers could be dropped from the archive, for either being obsolete or unmaintained: DONE
[raof] Enable 3d nouveau: DONE
[raof] Evaluate whether to retain nouveau 3d, some time after kernel freeze: DONE
[bryce] Ensure xdiagnose has workarounds (where possible) for common broken hybrid breakages: postponed
[bryce] Follow up with kernel engineers about usb livecd kernel/X testing: DONE
[bryce] Talk with cjwatson about grub work and trigger for whether it was a successful boot: DONE
[pitti] In jockey, display new -updates X drivers: DONE
work items for ubuntu-11.10:
[raof] Talk to debian about adding some form of xvfb test to xserver build time checks: DONE
[apw] Make the intel kernel module always print: possible outputs, connected outputs, detected modes, selected mode: POSTPONED
[apw] Understand which kernels are needed for -staging ppa: DONE
[raof] Create a -staging ppa for P with xserver 1.11: DONE
[bryce] Create blueprint for working out details of doing opt-in LTS point updates of X.org: DONE
[bryce] Close out bug reports against old hardware and drivers: DONE
[bryce] Auto-close bug reports about old intel, ati, and nvidia hardware and direct reporters upstream: POSTPONED
[bryce] Organize online testing for nouveau: DROPPED
pitti, 2011-04-27: Copied postponed WIs from https:/
bryce, 2011-05-16: Merged notes/tasks from discussion. Trimmed out some finished/obsolete tasks.
tjaalton, 2011-06-28: Dropped a number of drivers from -video-all, still in the archive though
bryce, 2011-06-28: xdiagnose has support for turning on drm.debug. I'm skeptical that it is of much use for gpu lockup bugs though, so am going to punt on integrating it into the apport hooks themselves. If it proves useful it can be added later, but I've not tended to see upstream request this be turned on for hangs.
bryce, 2011-06-29: Word from Leann is that the iso images are very ad-hoc and the results weren't getting collected reliably enough to be useful, and hasn't been used much in the last year. Brad has a script to generate isos if we want to pursue this style of testing ourselves and roll our own isos, but it appears there isn't an existing ongoing effort we can simply tie into.
raof, 2011-09-29: Bryce added an xvfb test to the xserver packaging. Next cycle we should look at whether xvfb is actually the right choice for testsuites, though - the dummy video and input drivers with the regular DDX might be better. 1.11 xserver is in a PPA, but the multitouch patchset needs refreshing; until then, Unity is broken on it.
bryce, 2011-11-15: Several of the postponed items are now basically done since they were prep-work for Precise; marked as such. Other items are still relevant to be done, and have been copied forward to desktop-p-xorg
