Feature changed by: Stefan Dirsch (sndirsch) Feature #307139, revision 21 Title: Enable KMS for Intel Graphics cards openSUSE-11.2: Rejected by Stephan Kulow (coolo) reject date: 2009-08-12 11:02:48 reject reason: see comments. Priority Requester: Important - openSUSE-11.3: Candidate + openSUSE-11.3: Done Priority Requester: Desirable Projectmanager: Mandatory Requested by: Kshitij Kulshreshtha (polyconvex) Description: Since in the Intel Graphics Release 2009Q2 kernel modesetting is the recommended method, please enable CONFIG_DRM_I915_KMS in the kernel config. In the current factory kernel (August 03) CONFIG_DRM_RADEON_KMS is set even though the code is considered unstable and depends on CONFIG_STAGING, however the more stable and recommended CONFIG_DRM_I915_KMS is not set. Please enable this option too. Discussion: #1: Andreas Jaeger (a_jaeger) (2009-08-12 10:36:48) Stefan, could you comment on this, please? #2: Stefan Dirsch (sndirsch) (2009-08-12 10:56:07) We didn't enable KMS for radeon by intention. See Bug #527910 for more details. #3: Stephan Kulow (coolo) (2009-08-12 11:02:41) kms support is in our kernels, CONFIG_DRM_I915_KMS is about the default and we won't change the default not yet. Boot with i195.modeset=1 if you know your machine supports it. For those machines we might add it automatically, but not via kernel default. #4: Stefan Dirsch (sndirsch) (2009-08-12 11:11:05) KMS on my 945GM machine at home still gave me a black screen last time I tried (Kernel 2.6.30, xorg-server 1.6.2, xf86-video-intel 2.8.0, libdrm 2.4.11, Mesa 7.5.0) whereas UMS works. Thus I'm not sure whether we want to make KMS the default now. I'm going to try again with latest X/Kernel stack soon. #5: Stefan Dirsch (sndirsch) (2009-08-13 02:39:50) Apparently KMS on 945GM improved a lot with latest X.Org/Kernel (xorg- server 1.6.3/Kernel 2.6.31-rc5). No longer black screen. Even switching to Linux console and back works. I'm trying to use this one regularly now. #6: Ruchir Brahmbhatt (ruchir) (2009-08-14 21:54:28) It would be nice to make it default if it seems stable enough. #7: Matthias Hopf (mhopf) (2009-08-17 12:02:35) (reply to #6) How many different systems have actually been tested successfully so far? Your statement is pretty bold... But we should consider to make KMS the default on i915,i945, i965, and G33. I haven't tested G33 yet, though. i8** are IMHO too little tested, maybe they don't work at all, i815 at least probably doesn't work. #8: Rastislav Krupansky (ra100) (2009-09-30 12:32:01) Any progress here for M8? Can we expect enabled KMS (especially i´m interested in i915) Or still should i activate by adding i915. modeset=1? And i´d like to ask, how about is with UXA? Is it default already? #9: Stefan Dirsch (sndirsch) (2009-09-30 12:38:18) No changes here to expect for final release, since we're afraid of a lot of regressions if making KMS the default. So you still need to specify i915.modeset=1. There is nothing left in the driver than UXA. So yes, it's the default meanwhile. #10: Kshitij Kulshreshtha (polyconvex) (2009-11-17 11:55:34) I am currently running openSUSE 11.2 with drm.ko and i915.ko added to the initrd and the following kernel command line: root=/dev/disk/by-label/rootfs resume=/dev/disk/by-label/swapfs i915. modeset=1 splash=silent quiet showopts My chipset is i915GM. The only problem I see at the moment is that the splash=silent parameter isn't regarded and I see a framebuffer console with boot messages instead of the splash graphic. The dmesg has the following lines Linux agpgart interface v0.103 agpgart-intel 0000:00:00.0: Intel 915GM Chipset agpgart-intel 0000:00:00.0: detected 7932K stolen memory agpgart-intel 0000:00:00.0: AGP aperture is 256M @ 0xc0000000 [drm] Initialized drm 1.1.0 20060810 i915 0000:00:02.0: PCI INT A -> GSI 16 (level, low) -> IRQ 16 i915 0000:00:02.0: setting latency timer to 64 i2c-adapter i2c-1: unable to read EDID block. i915 0000:00:02.0: LVDS-1: no EDID data [drm] DAC-6: set mode 640x480 0 i2c-adapter i2c-1: unable to read EDID block. i915 0000:00:02.0: LVDS-1: no EDID data [drm] TV-12: set mode NTSC 480i 0 bootsplash 3.1.6-2004/03/31: looking for picture... bootsplash: silentjpeg size 115769 bytes bootsplash: does not fit into framebuffer. bootsplash: ...found (1280x800, 27301 bytes, v3). render error detected, EIR: 0x00000010 page table error PGTBL_ER: 0x00000100 [drm:i915_handle_error] *ERROR* EIR stuck: 0x00000010, masking render error detected, EIR: 0x00000010 page table error PGTBL_ER: 0x00000100 [drm] LVDS-8: set mode 1280x800 16 Console: switching to colour frame buffer device 156x46 [drm] fb0: inteldrmfb frame buffer device [drm] Initialized i915 1.6.0 20080730 for 0000:00:02.0 on minor 0 #11: Kshitij Kulshreshtha (polyconvex) (2009-11-17 11:59:23) (reply to #10) Ehhh... the textbox mangled messages Linux agpgart interface v0.103 agpgart-intel 0000:00:00.0: Intel 915GM Chipset agpgart-intel 0000:00:00.0: detected 7932K stolen memory agpgart-intel 0000:00:00.0: AGP aperture is 256M @ 0xc0000000 [drm] Initialized drm 1.1.0 20060810 i915 0000:00:02.0: PCI INT A -> GSI 16 (level, low) -> IRQ 16 i915 0000:00:02.0: setting latency timer to 64 i2c-adapter i2c-1: unable to read EDID block. i915 0000:00:02.0: LVDS-1: no EDID data [drm] DAC-6: set mode 640x480 0 i2c-adapter i2c-1: unable to read EDID block. i915 0000:00:02.0: LVDS-1: no EDID data [drm] TV-12: set mode NTSC 480i 0 bootsplash 3.1.6-2004/03/31: looking for picture... bootsplash: silentjpeg size 115769 bytes bootsplash: does not fit into framebuffer. bootsplash: ...found (1280x800, 27301 bytes, v3). render error detected, EIR: 0x00000010 page table error PGTBL_ER: 0x00000100 [drm:i915_handle_error] *ERROR* EIR stuck: 0x00000010, masking render error detected, EIR: 0x00000010 page table error PGTBL_ER: 0x00000100 [drm] LVDS-8: set mode 1280x800 16 Console: switching to colour frame buffer device 156x46 [drm] fb0: inteldrmfb frame buffer device [drm] Initialized i915 1.6.0 20080730 for 0000:00:02.0 on minor 0 #12: Matthias Hopf (mhopf) (2009-11-17 12:24:00) Suspend to disk is currently broken when KMS is enabled on many i915 machines. Issue is known upstream, yet there hasn't been progress for months now. KMS doesn't improve anything, at least on i915 machines. Resume from StR isn't faster (which was claimed), most machines have the native resolution as a VESA mode for the console, framebuffer reallocation works with user mode setting as well. So there's currently no reason to switch the default. #13: Kshitij Kulshreshtha (polyconvex) (2009-11-17 13:13:43) On my machine I am able to suspend and successfully resume from disk any number of times if I set HIBERNATION_METHOD=kernel in a file in /etc/pm/config.d/. With HIBERNATION_METHOD=userspace (which is the default) I get a GPU Hang when splashy displays on resume, which requires a reboot to get a working system again. If I recompile the suspend-0.80 package with --disable-splashy, I am able to resume from StD successfully again. So I would suggest that the problem lies in the userspace resume (espacially splashy) and not in the kernel. #14: Andreas Jaeger (a_jaeger) (2010-03-02 13:51:51) AFAIK Egbert is working on this, so setting this to implementation. If this is false, please send it back to me... + #15: Stefan Dirsch (sndirsch) (2010-03-02 15:00:51) + This has already been implemented for openSUSE 11.3. -- openSUSE Feature: https://features.opensuse.org/307139