[REGRESSION] X doesn't work with 2.6.33 (can't find any input devices)
ty...@mit.edu
now finding the next problem --- X doesn't work. If I let X come up,
it doesn't find any mouse or keyboard devices, so my machine becomes
useless and I have to power cycle it unless I can log in remotely and
reboot it.
It worked just *fine* using 2.6.33-rc4. I have no xorg.conf file (X
is autoconfiguring itself) and I'm using an Ubuntu 9.10 userspace.
I've enclosed the dmesg output (which shows the input devices
correctly being detected by the kernel) and the Xorg.0.log file from
the previous boot, where X does not come up.
At this point, I'm not going to be able to dogfood test my ext4
patches I plan to push to Linus, since 2.6.33 is completely useless to
me. Hopefully running the regression tests under KVM is going to have
to be good enough.
Grumble.
- Ted
David Miller
Date: Wed, 3 Mar 2010 10:17:20 -0500
> It worked just *fine* using 2.6.33-rc4. I have no xorg.conf file (X
> is autoconfiguring itself) and I'm using an Ubuntu 9.10 userspace.
X uses udev events to find input devices these days, so it's probably
the same problem preventing your lvm devices from showing up correctly
from initramfs.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majo...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
ty...@mit.edu
> From: ty...@mit.edu
> Date: Wed, 3 Mar 2010 10:17:20 -0500
>
> > It worked just *fine* using 2.6.33-rc4. I have no xorg.conf file (X
> > is autoconfiguring itself) and I'm using an Ubuntu 9.10 userspace.
>
> X uses udev events to find input devices these days, so it's probably
> the same problem preventing your lvm devices from showing up correctly
> from initramfs.
It uses libhal, I believe to find input devices and I was finally able
to work around the problem by configuring an manually recreated
xorg.conf file with the magic option:
Section "ServerFlags"
Option "AllowEmptyInput" "off"
EndSection
... and then making sure that the kbd and mouse input drivers were
installed. There is definitely something wrong though if sometime
between 2;6.33-rc4 and 2.6.33 we've made some incompatible sysfs
change that breaks hal on Ubuntu 9.10 (which is still the latest
Ubuntu community release), such that people can't test 2.6.33 kernels
using it --- at least not easily.
See Linus's complaints over the nouveau driver; it's the same
principle; something is really broken.
- Ted
Dave Airlie
> On Wed, Mar 03, 2010 at 07:20:07AM -0800, David Miller wrote:
>> From: ty...@mit.edu
>> Date: Wed, 3 Mar 2010 10:17:20 -0500
>>
>> > It worked just *fine* using 2.6.33-rc4. �I have no xorg.conf file (X
>> > is autoconfiguring itself) and I'm using an Ubuntu 9.10 userspace.
>>
>> X uses udev events to find input devices these days, so it's probably
>> the same problem preventing your lvm devices from showing up correctly
>> from initramfs.
>
> It uses libhal, I believe to find input devices and I was finally able
> to work around the problem by configuring an manually recreated
> xorg.conf file with the magic option:
>
> Section "ServerFlags"
> � � � �Option "AllowEmptyInput" "off"
> EndSection
>
> ... and then making sure that the kbd and mouse input drivers were
> installed. �There is definitely something wrong though if sometime
> between 2;6.33-rc4 and 2.6.33 we've made some incompatible sysfs
> change that breaks hal on Ubuntu 9.10 (which is still the latest
> Ubuntu community release), such that people can't test 2.6.33 kernels
> using it --- at least not easily.
>
> See Linus's complaints over the nouveau driver; it's the same
> principle; something is really broken.
Not really, this is just somebody doing something stupid one would hope,
does lshal on both kernels give similiar results?
Dave.
Dave Airlie
Are you using 32 on 64 or anything like that?
15e184afa83a45cf8bafdb9dc906b97a8fbc974f is the only thing I can
see that might be close to changing something.
ty...@mit.edu
> On Fri, Mar 5, 2010 at 2:40 PM, <ty...@mit.edu> wrote:
> > See Linus's complaints over the nouveau driver; it's the same
> > principle; something is really broken.
>
> Not really, this is just somebody doing something stupid one would hope,
>
> does lshal on both kernels give similiar results?
No, it's quite different. See attached.
--- lshal-2.6.33 2010-03-06 10:15:44.837265508 -0500
+++ lshal-2.6.32-git4 2010-03-06 18:45:35.782582002 -0500
@@ -1,17 +1,2446 @@
-Dumping 23 device(s) from the Global Device List:
+Dumping 149 device(s) from the Global Device List:
-------------------------------------------------
...
It looks like it's fixed as of 2.6.33-git10. Hopefully whoever fixed
it will backport it to 2.6.33-STABLE, since this will seriously screw
up anyone wanting to use 2.6.33 at least on Ubuntu; I don't know about
other distributions.
- Ted