Bug description: using xf86-video-intel 2.99.911, X is 10 to 100 times slower than normal to start, or never starts and displays a black or garbled screen. Switching to a virtual console is still possible, but sometimes garbled until X is closed. Even when displaying a black screen, all the components of the desktop are running as are auto-started applications such as Google Chrome. If X starts, after taking close to 100 times longer than normal, and it's not displaying a garbled screen, everything works normally for that session. Downgrading to xf86-video-intel 2.99.910 solves all issues. System environment: -- chipset: Core i5-3570K / Intel Xeon E3-1200 v2/3rd Gen Core processor Graphics Controller -- system architecture: 64-bit -- xf86-video-intel: 2.99.911 -- xserver: X.Org 1.15.0 -- mesa: 10.1.0_devel.60267-1 (package intel-dri-git in AUR flagged out of date back on 2013-10-16) -- libdrm: 2.4.50-1 (package libdrm-git in AUR last updated 2013-04-23) -- kernel: 3.13.6 -- Linux distribution: Arch -- Machine or mobo model: GIGABYTE GA-Z77N-WIFI -- Display connector: HDMI Reproducing steps: I tried starting both Xfce and DWM. I tried DWM just because it's minimal. Xfce was auto-starting a lot of stuff. DWM wasn't auto-starting anything. DWM didn't work either. Additional info: The errors generated are not listed in journalctl because I get the following journalctl error: systemd-journald[131]: Failed to write entry (26 items, 81589132 bytes) despite vacuuming, ignoring: Argument list too long
In filing this bug report
In filing this bug report, I discovered that I had an out of date mesa package. Switching to the standard intel-dri in the Arch repositories solved the problem. I also switched to the standard package for libdrm at the same time, so that I'm not sure which one caused the problem though.
So, to be clear, I have no issues with xf86-video-intel 2.99.911, now that I'm using -- mesa: 10.1.0 (instead of 10.1.0_devel.60267) -- libdrm: 2.4.52 (instead of 2.4.50) I don't know which of the old packages caused the problem. Some testing probably needs to be done with libdrm 2.4.50 to see if that was the problem.
Whilst this does sound like the regressions in mesa-10.1, this is the first I've heard of an issue becoming more frequent with 2.99.911. So please can you attach Xorg.0.log and /sys/class/drm/card/0/error, and if at all possible bisect between 910 and 911 to see if there was a particular commit that makes it worse.
Would be nice to know exactly what combination of factors caused the slow down, but if all is well now, let's just be happy.
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.