From Fedora Project Wiki

Thu March 12, 2009 All day #fedora-qa)

What to test?[edit]

Today's instalment of Fedora Test Day will focus on the Intel graphics card driver. F11 brings a big update to intel graphics for everything in the graphics stack. We're shipping the new kernel memory manager, GEM, kernel modesetting, DRI2, graphical boot (plymouth) integrating with X, screen hotplugging, GLX1.4 and GL framebuffer objects, better fast user switching (multiple X servers can use DRI). There's a lot of new code and it's all turned on by default, so theres a lot of detail and integration work that needs to be done.

Related feature pages:

If you come to this page after the test day is completed, your testing is still valuable, and you can use the information on this page to test with your graphics card and provide feedback.

Who's available[edit]

The following cast of characters will be available for testing, workarounds, bug fixes, and general discussion ...

What's needed to test[edit]

  • An Intel graphics adapter (i810 or later, except GMA 500 / Poulsbo)
  • Rawhide (tips on installing Rawhide below), or the live CD available for this test day (again see below). Note that the Intel X driver was recently renamed from xorg-x11-drv-i810 to xorg-x11-drv-intel.
  • Remove your xorg.conf, unless you have a really good reason to have one. The Virtual line that was required for working multihead in F10 is no longer necessary and may cause problems - please remove it if you have one.
  • FAS Account - you can create an account in 3 minutes if you don't have one
no animals will be hurt during testing

How to test?[edit]

Update your machine[edit]

See the instructions on the Rawhide page on the various ways in which you can install or update to Rawhide.

Live Image[edit]

Optionally, you may download a non-destructive rawhide live image for your architecture:

Tips on using a live image are available at FedoraLiveCD.

Note: Creating a LiveUSB of this rawhide image requires the latest version of syslinux, which is currently available in Rawhide and in the latest Windows liveusb-creator. Users of F10 and below can yum --enablerepo=rawhide update syslinux


Things to test, roughly in dependency order:

Perform each of the following test cases that you are able to with the resources available to you. Some of the tests depend on the others, so obviously if X fails to run with kernel mode setting for you, you will not be able meaningfully to test video playback or 3D application on top of KMS, for instance.

Miscellaneous notes:

  • A minimal test that modesetting is working is to remove rhgb from the command line and add 3 to boot into text mode. If KMS works, you should have a text mode with a lot more character cells than the standard 80x25 and it will be a little slower.
  • Plymouth on dualhead gives interesting results, typically black borders on one or both displays. Plymouth cheat-codes: Ctrl-T goes to text mode, Ctrl-V enables verbose mode.
  • We're interested in tests on laptops with docking stations, so if you have one, try the tests with it connected.
  • Make sure rhgb is in your kernel boot line.
  • If you get to a point where GDM doesn't seem to work right and when you check from a console, there's 5 X processes running - reboot with your ethernet cable unplugged.

Report your results[edit]

Once you have completed the tests, add your results to the Results table below, following the example results from Adam Williamson as a template. The first column should be your name with a link to your User page in the Wiki if you have one, and the second should be a link to your Smolt hardware profile (see above for a link with instructions on submitting your hardware profile to Smolt). For each test case, if your system worked correctly, simply enter the word PASS. If you had any failures, enter the word FAIL, as a link to the bug report for the failure. If you could not meaningfully run the test due to limitations of your system or because of failures in more basic functionality, enter the word N/A. In the comments column, you can enter the model name and PCI device ID (vendor ID is usually 8086) of your card, if you know it - you can usually find this information in the output of the command lspci -nn.


User Smolt Profile Basic (KMS) Basic (no KMS) Multi-head DPMS DRI2/GLX Xv Fast user switch VT switch Suspend Comments
madcat HW PASS PASS PASS PASS PASS PASS N/A PASS N/A i965, backlight control doesn't work with KMS
saikat HW PASS FAIL bz489892 NA FAIL bz489907 PASS PASS FAIL bz489924 PASS PASS
  • Docking/undocking kills display. Needs hard reboot. bz489938
  • glxgears fps drops while desktop cube is rotating
mwatts HW PASS PASS Partial PASS bz489959 FAIL bz489980 Partial PASS bz489988 FAIL (Unsupported) PASS PASS FAIL 82825/855GM (8086:3582)
  • i855 doesn't support more than 2048 pixels in either direction, so I'm unable to set max resolution for both displays (1280x768 and 1600x1200)
  • KMS suspend = system freeze.
  • nomodeset suspend will suspend and power the system off, but resuming only gets back to the "Freezing blahblahblah" text and then hangs.
jason17055 HW FAIL


PASS NA FAIL FAIL PASS 845G; missing mouse pointer
Mike Bonnet HW PASS PASS N/A FAIL PASS PASS PASS i945 (8086:27a2)
  • DRI2/GLX works but is very slow. glxgears maxes out around 330fps. Moving windows around is jerky, and there is significant visual tearing.
  • DPMS failed, screensaver activated after one minute but the screen never turned off.
Adam Williamson HW PASS PASS PASS FAIL PASS PASS PASS PASS N/A (fails due to non-intel bug) i945 (8086:27a2)
jicama HW PASS* FAIL FAIL PASS PASS PASS FAIL PASS i965 - basic test 'fails' due to non Intel-related bug
  • VT test: Sometimes when I'm switched to a text console the X server dies and gdm pops up on the same VT as the text console I was on. And sometimes it doesn't, I'm not sure what's making the difference.
Rex Dieter HW PASS N/A N/A FAIL PASS (but slow vs. F9) N/A N/A PASS PASS (FAIL using compiz) i945
  • Gpu hang on wrong vbo usage. NoKMS seems to be twice as fast than KMS.
  • When running DRI (glxgears for example) after login to second user X starts crashing and I have to reboot.
  • On resume X a) restarts b) seems to hang on something and I can switch to another vt.
verdurin HW PASS PASS N/A Partial PASS bz489980 Partial PASS* PASS PASS PASS PASS i945GM (27a6)
  • switching workspaces in compiz works for one user but not the other
  • glxgears is only 350 FPS
ceski HW PASS PASS N/A FAIL PASS PASS PASS GMA4500 (8086:2a42)
mdomsch [1] PASS PASS Partial PASS. Laptop panel always worked in proper resolution. External 24" monitor (VGA1) worked until turned off in gnome-display-properties, after which the electrical output was always off regardless of g-d-p or xrandr commands. PASS PASS PASS FAIL. Switching user, left with black screen, cursor in top right corner, and mouse arrow pointer. PASS FAIL, but suspend broken in F10 too. Dell XPS M1330 with integrated graphics: Intel GM965/GL960 (8086:2a02)
  • glxgears ~ 440fps; ~1000fps in F10 with Tiling option tuned.
  • driver for the 945GM does not support > 2048x2048 virtual, but screen mirroring does not work with a 1280x1024 monitor. Mirroring also does not show up correctly with identical 1680x1050 res monitor attached.
bothner HW PASS PASS FAIL Maybe this See also PARTIAL PASS PASS PASS PARTIAL PASS (8086:2a43)
  • Can use external monitor, but not dual-head - this has worked before.
  • Screensaver started after one minute, but screen is never blanked.
  • Can't switch to VT2 on external monitor - this has worked before.
orthob HW FAIL Almost PASS NA FAIL FAIL FAIL ~PASS PASS NA 845G/GL[Brookdale-G] 8086:2562 rev3
  • No cursor w/ KMS
  • Awful 2D accel w/ KMS
  • Corrupt glxgears w/ KMS
  • Backlight not always re-enabled.
ilyes HW PASS PASS N/A N/A PARTIAL bug1 bug2 FAIL N/A PASS N/A intel 82852/855GM (rev 02)
  • glxgears fps drops significantly when KMS is on.
  • Mouse pointer flickers when KMS is on.
  • gstreamer-properties crashes when I try to test the video, it reports a bad allocation on the console. Running it again leads to the entire desktop being reinitialized.
Messerting HW PASS N/A N/A N/A FAIL N/A PASS (but with Compiz enabled, notable tearing on rotating cube) N/A N/A Used livecd x86_64 on Thinkpad T400, GM45

Intel Corporation Mobile 4 Series Chipset Integrated Graphics Controller [8086:2a42] (rev 07)

Messerting HW FAIL N/A N/A N/A N/A N/A N/A N/A N/A G35 8086:2980 (rev 03) Used livecd x86_64 on Asus P5E-VM HDMI using HDMI output

When booting, I did get a text-mode progress bar (LCD-TV reported 1360x768), but login screen never came up. ctrl-alt-f2 worked, and I could log in there.

  • Got warning at lib/dma-debug.c:479, dmesg
liam HW PASS PASS N/A FAIL PASS PASS PASS PASS PASS,but got a kernel warning,"WARNING: at kernel/power/main.c:176 suspend_test_finish+0x5b/0x65() (Tainted: G W )" test with Live CD livecd-fedora-livecd-desktop-200903161319-i386.iso
Waitkw HW PASS PASS Partial, identical behavior to bz489959 Fail, identical behavior to bz489980 PASS PASS Fail, screen goes blank but no prompt to switch users appears. ctrl+alt+f1 brings original session back up PASS Fail, suspends but does not resume *i945 (8086:2592)
  • using livecd
Björn Persson HW partial PASS [1][2] PASS [2] FAIL FAIL PASS PASS PASS PASS PASS 82G35 (8086:2982 and/or 8086:2983, rev 03)

[1] using live USB, text mode progress bar displayed in native resolution

[2] Sometimes there's a long pause after initrd has loaded. Nothing seems to happen except that the Num Lock light blinks slowly. Eventually the progress bar shows up and the boot continues. Other times the progress bar comes quickly after initrd.

mintslice (Rodd Clarkson) HW FAIL PASS N/A PARTIAL

Screensaver works, but screen never sleeps.


Compiz doesn't work, so couldn't test with compiz. Works without.


Compiz doesn't work, so couldn't test with compiz. Works without.

PASS N/A 00:02.0 VGA compatible controller: Intel Corporation 4 Series Chipset Integrated Graphics Controller (rev 03)

00:02.1 Display controller: Intel Corporation 4 Series Chipset Integrated Graphics Controller (rev 03) I get less than 3.0FPS with extremetuxracer, so I'm going to presume that 3D has some serious issues. Turns out there's a kernel patch that should fix the 3d issues with the g41 chipset. see: