Revert various i915 patches

Bug #849130 reported by Steve Conklin on 2011-09-13
This bug affects 1 person
Affects Status Importance Assigned to Milestone
linux (Ubuntu)

Bug Description

A number of i915 patches were applied to the Lucid kernel via a stable update.
These resulted in two regressions.

When the first regression was detected, a decision was made to apply a fix instead
of following stable kernel process and reverting the original patches. The fix that was applied was :
756de92 drm/i915: Remove BUG_ON from i915_gem_evict_something

The second regression to be detected requires two upstream patches, and may involve more i915 patches
which are currently queued in the master-next branch for Lucid. These patches are non-trivial, and therefore
all i915 patches applied since the last Lucid stable update will be reverted, except for one quirk.

These patches will be reapplied and sorted as needed on master-next.

The patches to be reverted are:

756de92 drm/i915: Remove BUG_ON from i915_gem_evict_something
9b4fca6 drm/i915: Periodically flush the active lists and requests
53e68e3 drm/i915/evict: Ensure we completely cleanup on failure
38c2e45 drm/i915: Maintain LRU order of inactive objects upon access by CPU (v2)
5ef2204 drm/i915: Implement fair lru eviction across both rings. (v2)
3e69bf2 drm/i915: Move the eviction logic to its own file.
1126f22 drm/i915: prepare for fair lru eviction

Steve Conklin (sconklin) on 2011-09-13
tags: added: kernel-stable-next
Changed in linux (Ubuntu Lucid):
status: New → In Progress
Steve Conklin (sconklin) wrote :

This is the bug reporting the regression:

Steve Conklin (sconklin) wrote :

Also, I dropped the following patches from the Lucid master-next branch, as they depend on the reverted patches:

434094a drm/i915: Fix refleak during eviction.
2ec02ef drm/i915: Hold a reference to the object whilst unbinding the eviction list

Seth Forshee (sforshee) wrote :

I browsed through the 1000+ upstream i915 commits in the changelog since the LRU backport patches for anything that looked tangentially related to these patches, and then I reviewed the identified changesets more closely to see if any were bugfixes related to the patches reverted here. It looks like all bugfixes had been applied to master-next prior to reverting the patches, so if all the same patches are reapplied then all known regressions related to these patches should be fixed.

Steve Conklin (sconklin) on 2012-06-01
Changed in linux (Ubuntu Lucid):
status: In Progress → Fix Released
Tim Gardner (timg-tpi) on 2012-10-12
Changed in linux (Ubuntu):
status: New → Invalid
To post a comment you must log in.
This report contains Public information  Edit
Everyone can see this information.

Other bug subscribers