Created attachment 39216 [details] backtrace Setting the EXANoUploadToScreen option in xorg.conf makes the xserver segfault. git bisect says: f8fb9312d791af1f77020e8c2d35bb30841ed9aa is the first bad commit commit f8fb9312d791af1f77020e8c2d35bb30841ed9aa Author: Karl Tomlinson <karlt+@karlt.net> Date: Sun Aug 22 22:46:33 2010 +1200 RADEONPrepareAccess_CS: fallback to DFS when pixmap is in VRAM This avoids costly CPU VRAM reads and lets EXA manage a system memory cache of the portions of pixmaps needed for unaccelerated operations. https://bugs.freedesktop.org/show_bug.cgi?id=27139
I'm inclined to resolve this as WONTFIX. This is a useful optimization, and the driver has required the UploadToScreen hook on big endian machines for as long as KMS has worked there. Why are you trying to use Option "EXANoUploadToScreen"?
I agree it's a useful optimization and WONTFIX is fine with me, although I think it would be cleaner to disable EXANoUploadToScreen option, or at least mention it in exa(4) that this option causes segfault with radeon driver. I was trying this option because of bug 30188, which gives me EVERGREENUploadToScreen in the backtrace, and was hoping EXANoUploadToScreen option would fix it, unfortunately it doesn't.
(In reply to comment #2) > I agree it's a useful optimization and WONTFIX is fine with me, although I > think it would be cleaner to disable EXANoUploadToScreen option, Not possible, it's an option of the "exa" module, which doesn't know that the driver really needs the UploadToScreen hook. > or at least mention it in exa(4) that this option causes segfault with radeon > driver. If anything it should rather be documented with the driver. > I was trying this option because of bug 30188, which gives me > EVERGREENUploadToScreen in the backtrace, and was hoping EXANoUploadToScreen > option would fix it, unfortunately it doesn't. Trying to work around that problem really requires the UploadTo/DownloadFromScreen hooks. The issue is that there is still sometimes direct CPU access to VRAM.
*** Bug 30897 has been marked as a duplicate of this bug. ***
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.