[arch-dev-public] [signoff] kernel 2.6.32.2-1
Hi guys, Upstream changes: http://kernelnewbies.org/LinuxChanges Arch Linux bugfixes/feature requests: http://bugs.archlinux.org/task/16855 # added CONFIG_PM_DEBUG http://bugs.archlinux.org/task/17106 # added CONFIG_MMIOTRACE Arch Linux changes: - splitted kernel-headers to extra package If you want to build external modules please install: pacman -S kernel26-headers Please change your PKGBUILDS to makedepend on this package. - added xen support to 64bit kernel - changed intel kms enabled by default - changed to new firewire subsystem: http://ieee1394.wiki.kernel.org/index.php/Juju_Migration Please signoff both arches, greetings tpowa -- Tobias Powalowski Archlinux Developer & Package Maintainer (tpowa) http://www.archlinux.org tpowa@archlinux.org
On 12/20/2009 10:32 PM, Tobias Powalowski wrote:
Hi guys,
Upstream changes: http://kernelnewbies.org/LinuxChanges
Arch Linux bugfixes/feature requests: http://bugs.archlinux.org/task/16855 # added CONFIG_PM_DEBUG http://bugs.archlinux.org/task/17106 # added CONFIG_MMIOTRACE
Arch Linux changes: - splitted kernel-headers to extra package If you want to build external modules please install: pacman -S kernel26-headers Please change your PKGBUILDS to makedepend on this package. - added xen support to 64bit kernel - changed intel kms enabled by default - changed to new firewire subsystem: http://ieee1394.wiki.kernel.org/index.php/Juju_Migration
Please signoff both arches,
greetings tpowa
signoff x86_64. note that i'm not using any kms driver. -- Ionut
I think we need a news item how to fix Radeon kms. What's way you want to recommend? -Andy
On Mon, 2009-12-21 at 09:12 +0100, Andreas Radke wrote:
I think we need a news item how to fix Radeon kms. What's way you want to recommend?
-Andy
First of all, Radeon KMS is not ready for production. Like with the intel driver, modesetting can be enabled by using radeon.modeset=1 in the bootflags if KMS is not enabled by default for Radeon. CONFIG_DRM_RADEON_KMS=y should be disabled. Next issue is problems with Intel KMS. The 2.6.32 kernel contains powermanagement features for intel. These new features are utilized by the driver I uploaded to testing last weekend on hardware that supports it. However, this causes GPU lockups, requiring a reboot now and then. This was reverted in 2.6.33, so we should also revert this. A patch has been posted here: http://patchwork.kernel.org/patch/67768/ This patch is also committed to the drm-intel git tree, maybe it's easier to fetch the patch from git.kernel.org. Besides these changes, I haven't seen problems with this new kernel, everything has been stable for me for a while now. If we change these two things, I will give a signoff for both architectures.
Am 22.12.2009 15:59, schrieb Jan de Groot:
On Mon, 2009-12-21 at 09:12 +0100, Andreas Radke wrote:
I think we need a news item how to fix Radeon kms. What's way you want to recommend?
-Andy
First of all, Radeon KMS is not ready for production. Like with the intel driver, modesetting can be enabled by using radeon.modeset=1 in the bootflags if KMS is not enabled by default for Radeon. CONFIG_DRM_RADEON_KMS=y should be disabled.
Last time I check this was not true: Disabling the CONFIG_DRM_RADEON_KMS option would not only disable it by default, but remove all KMS code entirely.
On Tue, 2009-12-22 at 16:35 +0100, Thomas Bächler wrote:
Am 22.12.2009 15:59, schrieb Jan de Groot:
On Mon, 2009-12-21 at 09:12 +0100, Andreas Radke wrote:
I think we need a news item how to fix Radeon kms. What's way you want to recommend?
-Andy
First of all, Radeon KMS is not ready for production. Like with the intel driver, modesetting can be enabled by using radeon.modeset=1 in the bootflags if KMS is not enabled by default for Radeon. CONFIG_DRM_RADEON_KMS=y should be disabled.
Last time I check this was not true: Disabling the CONFIG_DRM_RADEON_KMS option would not only disable it by default, but remove all KMS code entirely.
Andy claimed that also, but with 2.6.32.2, it just disables the module aliases for autoloading and the default setting of KMS. Check the radeon driver for the define set by the Kconfig file and you'll see what I mean.
Am 22.12.2009 16:44, schrieb Jan de Groot:
Last time I check this was not true: Disabling the CONFIG_DRM_RADEON_KMS option would not only disable it by default, but remove all KMS code entirely.
Andy claimed that also, but with 2.6.32.2, it just disables the module aliases for autoloading and the default setting of KMS. Check the radeon driver for the define set by the Kconfig file and you'll see what I mean.
If they changed that, we can also disable that option. In 2.6.31 it was definitley like I said.
First of all, Radeon KMS is not ready for production. Like with the intel driver, modesetting can be enabled by using radeon.modeset=1 in the bootflags if KMS is not enabled by default for Radeon. CONFIG_DRM_RADEON_KMS=y should be disabled. Will try this in next build then.
Next issue is problems with Intel KMS. The 2.6.32 kernel contains powermanagement features for intel. These new features are utilized by the driver I uploaded to testing last weekend on hardware that supports it. However, this causes GPU lockups, requiring a reboot now and then. This was reverted in 2.6.33, so we should also revert this. A patch has been posted here: http://patchwork.kernel.org/patch/67768/
When i apply this patch i get an error: patching file drivers/gpu/drm/i915/intel_display.c Reversed (or previously applied) patch detected! Skipping patch. Could it be they solved this already in 32.2? greetings tpowa -- Tobias Powalowski Archlinux Developer & Package Maintainer (tpowa) http://www.archlinux.org tpowa@archlinux.org
On Tue, Dec 22, 2009 at 10:59 PM, Tobias Powalowski
First of all, Radeon KMS is not ready for production. Like with the intel driver, modesetting can be enabled by using radeon.modeset=1 in the bootflags if KMS is not enabled by default for Radeon. CONFIG_DRM_RADEON_KMS=y should be disabled. Will try this in next build then.
Next issue is problems with Intel KMS. The 2.6.32 kernel contains powermanagement features for intel. These new features are utilized by the driver I uploaded to testing last weekend on hardware that supports it. However, this causes GPU lockups, requiring a reboot now and then. This was reverted in 2.6.33, so we should also revert this. A patch has been posted here: http://patchwork.kernel.org/patch/67768/
When i apply this patch i get an error: patching file drivers/gpu/drm/i915/intel_display.c Reversed (or previously applied) patch detected! Skipping patch. Could it be they solved this already in 32.2?
http://git.kernel.org/?p=linux/kernel/git/stable/linux-2.6.32.y.git;a=blob_p... This still contains all the references to renderclock , that the patch is supposed to remove. But anyway, just look at the file and the patch side by side to see whether your local kernel tree has this patch or not.
On Sun, 20 Dec 2009 21:32:10 +0100, Tobias Powalowski
Hi guys,
Upstream changes: http://kernelnewbies.org/LinuxChanges
Arch Linux bugfixes/feature requests: http://bugs.archlinux.org/task/16855 # added CONFIG_PM_DEBUG http://bugs.archlinux.org/task/17106 # added CONFIG_MMIOTRACE
Arch Linux changes: - splitted kernel-headers to extra package If you want to build external modules please install: pacman -S kernel26-headers Please change your PKGBUILDS to makedepend on this package. - added xen support to 64bit kernel - changed intel kms enabled by default - changed to new firewire subsystem: http://ieee1394.wiki.kernel.org/index.php/Juju_Migration
Please signoff both arches,
greetings tpowa
signed off botrh arches. Did not notice any real problem for a while. I would suggest to not use autodetect in mkinitcpio.conf with software raid as it does not detect the raid module and the klibc binary (forgot its name) which detects the filesystem just segfault (for me since .32). We shouldn't wait that long with the move; our kernel in core has (afaik) some security issues. -- Pierre Schmitz, https://users.archlinux.de/~pierre
participants (7)
-
Andreas Radke
-
Ionut Biru
-
Jan de Groot
-
Pierre Schmitz
-
Thomas Bächler
-
Tobias Powalowski
-
Xavier