| Summary: | [i965 bisected] Incorrect texture rendering in ut2004 with drm-intel-next | ||
|---|---|---|---|
| Product: | Mesa | Reporter: | fangxun <xunx.fang> |
| Component: | Drivers/DRI/i965 | Assignee: | Eric Anholt <eric> |
| Status: | VERIFIED FIXED | QA Contact: | |
| Severity: | normal | ||
| Priority: | high | CC: | chris, daniel, jian.j.zhao, yi.sun |
| Version: | unspecified | ||
| Hardware: | All | ||
| OS: | Linux (All) | ||
| Whiteboard: | |||
| i915 platform: | i915 features: | ||
| Attachments: |
screenshot of incorrect text rendering
Xorg log release_mmap when changing from untiled to tiled |
||
Created attachment 33919 [details]
Xorg log
If it's a regression, then a bisect is required. unblocking Q1 release with bisect I found the first bad commit is commit 10ae9bd25acf394c8fa2f9d795dfa9cec4d19ed6 Author: Daniel Vetter <daniel.vetter@ffwll.ch> Date: Mon Feb 1 13:59:17 2010 +0100 drm/i915: blow away userspace mappings before fence change This aligns it with the other user of i915_gem_clear_fence_reg, which blows away the mapping before changing the fence reg. Only affects userspace if it races against itself when changing tiling parameters, i.e. behaviour is undefined, anyway. Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Eric Anholt <eric@anholt.net> Can you please try http://bugs.freedesktop.org/attachment.cgi?id=33877 which reverts the effects of this patch? Just reverting this commit as-is is gona break stuff elsewhere. The obvious problem with this commit when I look at it now is how does the appropriate fence get installed when going untiled -> tiled and the object is already bound at an appropriate location and is already GTT mapped untiled? Created attachment 34164 [details]
release_mmap when changing from untiled to tiled
Thanks for the hint, Eric. I'm just feeling embarrassingly stupid now ;)
Please test this patch (and forget about the other one I've mentioned).
Daniel's patch seems to fix my corruption also http://www.sh0n.net/spstarr/intel-corruption.png verified with this patch. let's keep this open until the fix is committed. *** Bug 26766 has been marked as a duplicate of this bug. *** commit fe305198d4bf481d6dd017df35c566c9d477fada Author: Daniel Vetter <daniel.vetter@ffwll.ch> Date: Thu Mar 18 09:22:12 2010 +0100 drm/intel: fix up set_tiling for untiled->tiled transition (In reply to comment #12) > commit fe305198d4bf481d6dd017df35c566c9d477fada > Author: Daniel Vetter <daniel.vetter@ffwll.ch> > Date: Thu Mar 18 09:22:12 2010 +0100 > > drm/intel: fix up set_tiling for untiled->tiled transition > Eric, can you put this fix to drm-intel-next? verified with drm-intel-next. |
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.
Created attachment 33918 [details] screenshot of incorrect text rendering System Environment: -------------------------- Arch: x86_64 Platform: G45 Libdrm: (master)04fd3872ee8bd8d5e2c27740508c67c2d51dbc11 Mesa: (master)6f4ce4a4fed9f0f0f0ee89a63e406ab86dae7150 Xserver: (master)bbae92795c7eab062e6722c42fa7915e0cee5d69 Xf86_video_intel: (master)318aa9ed799197810e2039dbe3ec51559dcc888c Kernel: (drm-intel-next)338d762fc2dc2c1493813123fc4cea998bb3e683 Bug detailed description: ------------------------- Texture rendering in ut2004 is incorrect(as shown in the attached screenshot). This issue is regression. It happens on i965 platform with current unstable kernel. It works fine with current stable kernel. Reproduce steps: ---------------- 1.xinit& 2.ut2004