| Summary: | [bisected] pilgit glx/glx-make-glxdrawable-current regressed | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Product: | xorg | Reporter: | fangxun <xunx.fang> | ||||||||
| Component: | Driver/intel | Assignee: | Julien Cristau <jcristau> | ||||||||
| Status: | VERIFIED FIXED | QA Contact: | Xorg Project Team <xorg-team> | ||||||||
| Severity: | major | ||||||||||
| Priority: | high | CC: | jcristau, keithp | ||||||||
| Version: | git | ||||||||||
| Hardware: | All | ||||||||||
| OS: | Linux (All) | ||||||||||
| Whiteboard: | |||||||||||
| i915 platform: | i915 features: | ||||||||||
| Attachments: |
|
||||||||||
|
Description
fangxun
2011-01-21 00:25:53 UTC
On Fri, Jan 21, 2011 at 00:25:54 -0800, bugzilla-daemon@freedesktop.org wrote: > Bug detailed description: > ------------------------- > The regression happens on all platforms. Bisect shows > 361128389e5cb0101cbd091ff8de77cf34608f6c is the first bad commit. > commit 361128389e5cb0101cbd091ff8de77cf34608f6c > Merge: 65ceaad d9225b9 > Author: Keith Packard <keithp@keithp.com> > Date: Tue Jan 18 15:18:08 2011 -0800 > > Merge remote branch 'jcristau/for-keith' > Does this mean d9225b9602c85603ae616a7381c784f5cf5e811c is ok, and 361128389e5cb0101cbd091ff8de77cf34608f6c (the merge commit) isn't? That sounds weird... I'm getting: X Error of failed request: BadLength (poly request too large or internal Xlib length error) Major opcode of failed request: 153 (GLX) Minor opcode of failed request: 32 (X_GLXDestroyWindow) Serial number of failed request: 38 Current serial number in output stream: 39 That looks like a mesa bug. src/glx/glx_pbuffer.c::DestroyDrawable (called by glXDestroyWindow) has GetReqExtra(GLXDestroyPbuffer, 4, req);, which sets the request length to 3 instead of 2, for some reason. The extra 4 bytes don't seem needed by either glXDestroyPixmap or glXDestroyWindow. Both requests should be 8 bytes, not 12. Created attachment 42332 [details] [review] proposed patch This fixes it for me. Created attachment 42338 [details] [review] fix request length check for CreateGLXPbufferSGIX I found a couple more similar bugs in mesa, and one in my xserver patch. Created attachment 42339 [details] [review] work around wrong request lengths from mesa Works fine with fixed patch in xserver. BTW, without the fixed patch, it also works fine after updating mesa to 02d7d9ec363c7955105497c518296630cf2b638a. It now works well with newest code on master. Tested with: Libdrm: (master)2.4.23-6-g550fe2ca3b29ad2191eab4fdfbed9ed21e25492d Mesa: (master)d3df641f0aba99b0b65ecd4d9b06798bca090a29 Xserver: (master)xorg-server-1.9.99.901-165-gbe3be7580b6f6fd2f7fa4d4abfe5e1ab19470223 Xf86_video_intel: (master)2.14.0-13-gc6dc27562abbc8ca9e873ad502ca49ae010461d2 Cairo: (master)a8e8d2aba811487dbb5b682c4f55065008e7ebbd Kernel: (drm-intel-next) 1a3665c81df32b23c38d4ba8a74761551d5673b1 |
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.