Am 25.10.2011 17:12, schrieb Tobias Powalowski:

Archlinux Changes: - $(uname -r) reports now the full kernel version - all external modules should be installed in /lib/modules/extramodules-3.1-ARCH now

I made a mistake here, regarding the 'depmod' commands in the external module post_install/upgrade procedures. We'll need a rebuild of everything to fix that. You shouldn't notice the problem now, but we still need to fix it before moving to core. Thanks to Ionut for spotting this.

On Tue, Oct 25, 2011 at 11:14 AM, Tom Gundersen wrote:

On Tue, Oct 25, 2011 at 5:12 PM, Tobias Powalowski wrote:

...

- disabled the old IDE subsystem,  marked as deprecated for quite some time  All users of the old ide hook in mkinitcpio  should make the transition to pata hook now!

Did Dan's patches (or equivalent) land in time for 3.1? I did not notice them flying by, but I might have missed them in the kernel git log. Nope, but that's OK by me. If someone else has my exact hardware, I'll be glad to provide a rebuilt module that I'm going to have to build locally for now.

Of course, they haven't landed post-3.1 yet as of v3.1-2828-g7670c70...

For me, it boots, "dmesg --kernel --level err" is empty, and the warnings I get seem to be the kernel telling me I have crappy hardware (which I believe). So:

signoff x86_64

-t

Am 25.10.2011 17:12, schrieb Tobias Powalowski:

Hi guys, please signoff 3.1 series for both arches.

Upstream changes: http://kernelnewbies.org/LinuxChanges

Archlinux Changes: - $(uname -r) reports now the full kernel version - all external modules should be installed in /lib/modules/extramodules-3.1-ARCH now - disabled the old IDE subsystem, marked as deprecated for quite some time All users of the old ide hook in mkinitcpio should make the transition to pata hook now! - disabled the old ACPI /proc files, marked as deprecated for quite some time, everything is available in /sys

Another release, mostly identical, but I fixed the depmod on kernel update in nvidia, lirc, fcpcmcia and fcpci. All external module package maintainers in AUR/community should look at the PKGBUILD and install files for those to see what needs changing.

On Wed, Oct 26, 2011 at 3:26 PM, Evangelos Foutras wrote:

On 25/10/11 21:48, Thomas Bächler wrote:

...

Am 25.10.2011 17:12, schrieb Tobias Powalowski:

...

Hi guys, please signoff 3.1 series for both arches.

Upstream changes: http://kernelnewbies.org/LinuxChanges

Archlinux Changes: - $(uname -r) reports now the full kernel version - all external modules should be installed in   /lib/modules/extramodules-3.1-ARCH now - disabled the old IDE subsystem,   marked as deprecated for quite some time   All users of the old ide hook in mkinitcpio   should make the transition to pata hook now! - disabled the old ACPI /proc files,   marked as deprecated for quite some time,   everything is available in /sys

Another release, mostly identical, but I fixed the depmod on kernel update in nvidia, lirc, fcpcmcia and fcpci. All external module package maintainers in AUR/community should look at the PKGBUILD and install files for those to see what needs changing.

Signoff x86_64.

And another for i686.

On Sat, Oct 29, 2011 at 6:03 PM, Jan de Groot wrote:

1. The fix-i915.patch was removed, this causes my laptop to see a connected television again, while my laptop doesn't even have an output for that

Is there an upstream bug report for this? If we add the patch back it might be worth linking to the bug report so we can follow up on this when it gets resolved upstream. -t

On za, 2011-10-29 at 18:29 +0200, Thomas Bächler wrote:

Am 29.10.2011 18:03, schrieb Jan de Groot:

...

-1 from me. Reasons for this: 1. The fix-i915.patch was removed, this causes my laptop to see a connected television again, while my laptop doesn't even have an output for that 2. Massive screen flicker when starting up X. Something has regressed seriously in the intel driver.

I see Thomas removed the fix-i915.patch file, because it was "not accepted upstream". The only reference I can find for this patch is that nobody ever tested if TV-Out actually works when this patch is applied. As we haven't got any bugreports for this and nobody ever tested the patch with a hooked up TV, I'd say just add it back.

We do not add patches to our kernel that have not been merged upstream, will not be merged upstream, or have not been submitted upstream. I do not know why an exception was made in this case. This patch has been there for 3 or 4 major kernel versions (or longer?). Where is the bug report? The mailing list thread? The upstream commit?

I will not add this patch back unless it appears in the appropriate subsystem tree or in the Linus tree.

http://lists.freedesktop.org/archives/intel-gfx/2011-April/010371.html That's the whole discussion about this one. Seems to have been applied, but reverted because the reasoning behind the patch was unclear. Someone claims it's documented in some errata, but doesn't say which one. So far I've been able to find the bug in our tracker, with a link to the kernel.org bugtracker that's down: https://bugs.archlinux.org/task/19234

Am 29.10.2011 21:58, schrieb Thomas Bächler:

Am 29.10.2011 20:53, schrieb Jan de Groot:

...

http://lists.freedesktop.org/archives/intel-gfx/2011-April/010371.html

That's the whole discussion about this one. Seems to have been applied, but reverted because the reasoning behind the patch was unclear. Someone claims it's documented in some errata, but doesn't say which one.

That's unfortunate.

...

So far I've been able to find the bug in our tracker, with a link to the kernel.org bugtracker that's down: https://bugs.archlinux.org/task/19234

Oh yeah, the kernel.org bugzilla, still down. We could add it back I guess, but this time with some proper references in the PKGBUILD (like the above two links).

Forgot to say this: Jan, can you please re-submit that patch to the appropriate mailing lists, and explain the whole situation? IMO it is a critical error in the upstream driver and must be taken care of upstream.