When running a newer kernel with KMS enabled on my Macbook 2,1 it corrupts the screen of my VT and makes the running of the Xserver impossible. I have enabled under Graphics (not as modules, but build in) only the AGPART for Intel Chipsets, the Intel i915 driver (with KMS default) and the basic support for framebuffers (no framebuffers selected in the list below). As soon as my VT hits the lowest line it somewhat corrupts the lowest few pixels of the screen. Also I tried all Xorgs in combination with all dri2proto, libdrm, mesa, intel drivers (also X11 with 9999 builds) but all show me only a massively corrupted mouse and hardlock the Xorg. This is independent of using UXA or EXA. But Im not sure whether this is related to this problem. Reproducible: Always Steps to Reproduce: Unsure. I spoke with another MacBook 2,1 owner (running Archlinux) and he had a KMS Kernel running just fine.
Created attachment 191030 [details] Picture of the Screen 1
Created attachment 191031 [details] Picture of the Screen 2
Hum, I've never heard of such a bug before. Please file a bug in FreeDesktop's bugzilla [1] and please paste the url here. Thanks [1] http://intellinuxgraphics.org/how_to_report_bug.html
Bug reported opened. https://bugs.freedesktop.org/show_bug.cgi?id=21710
Which is the last known working kernel?
Every Kernel is working. As long as I dont activate KMS I am currently running 2.6.30-rc5 UXA is running fast & stable over days, with suspend2ram. As long as I dont use KMS. Then I cant even start the X.
Could you make sure that with xorg-server 1.6, the intel driver initializes with DRI2? Xorg.0.log should have something like this : (II) intel(0): [DRI2] Setup complete And please try using the latest versions of libdrm, mesa, xorg-server and xf86-video-intel. Thanks
Anything to report here?
(In reply to comment #8) > Anything to report here? > I think there are 2 issues here. I have a macbook 2,1. I had the exact problem with the bottom edge of the console becoming corrupt at system startup as soon as the text output of the boot process reached the bottom edge of the screen. This occurred with KMS enabled on both the 2.6.29 and 2.6.30-rc8 kernels (and not with KMS disabled). It was fixed with the 2.6.30 final kernel. However, with both the 2.6.29 and 2.6.30 kernels, when kms is enabled, X essentially hangs at startup, with "massive mouse corruption." The mouse is movable, but the system is otherwise completely hung - cntrl-alt-backspace does not kill the X-server. Nothing useful in the logs, and hard-reset is required. Without KMS, X works on the 2.6.30 kernel with only minor graphics corruption. This appears to be a separate bug, and could consider filing another bug report for it specifically. I'm not much of a computer person, but if I get more info in the future I'll consider filing it myself.
(In reply to comment #9) > (In reply to comment #8) > > Anything to report here? > > > > I think there are 2 issues here. I have a macbook 2,1. l. > > However, with both the 2.6.29 and 2.6.30 kernels, when kms is enabled, X > essentially hangs at startup, with "massive mouse corruption." The mouse is > movable, but the system is otherwise completely hung it was the same with me. also on macbook 2,1 this is solved lately with a patch for drm/i915 https://bugs.freedesktop.org/show_bug.cgi?id=21710 it is in Linux 2.6.31-rc2 i tried it with gentoo-sources-2.6.30-r2. kms now works fine. i had to set: Option "DRI" "false" in the Device section. UXA works though. And 3D-accel seems also nice: glxgears is about 995fps and tuxracer up to 40 (1280x800). consider to add this patch to gentoo-sources
Created attachment 196707 [details, diff] init fence registers from: https://bugs.freedesktop.org/show_bug.cgi?id=21710 drm/i915: initialize fence registers to zero when loading GEM Unitialized fence register could leads to corrupted display. Problem encountered on MacBooks (revision 1 and 2), directly booting from EFI or through BIOS emulation.
Confirmed - running vanilla 2.6.31-rc2 fixes both problems. The KMS virtual terminal has no screen corruption at the lower edge. The X server starts and works beautifully.
(In reply to comment #11) > Created an attachment (id=196707) [edit] > init fence registers > > from: https://bugs.freedesktop.org/show_bug.cgi?id=21710 > > drm/i915: initialize fence registers to zero when loading GEM > > Unitialized fence register could leads to corrupted display. Problem > encountered on MacBooks (revision 1 and 2), directly booting from EFI > or through BIOS emulation. > I noticed that the patch submitted on the mailing list is slightly different from the patch from the bug report, don't know if it matters. http://lists.freedesktop.org/archives/intel-gfx/2009-June/002975.html
Created attachment 199012 [details, diff] init_fence_registers patch added the patch commited to git, and included in 2.6.31-rc2... i think it can be added to genpatches...
http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff_plain;h=b5aa8a0fc132dd512c33e7c2621d075e3b77a65e
Released in gentoo-sources-2.6.30-r4