Sigh. Ok. That's the initramfs DSDT code (again). I tested it myself on a few systems and had a few other people test it as well and didn't run into problems. Figures. This code is consistently a giant headache. While there are still some really old machines for which DSDT workarounds are floating around in the web, it shouldn't be necessary for recent (5 years?) machines to workaround BIOS bugs in such a way. Yes it's hard for some rare people if their old laptop, now serving as a tiny server cannot get updated to 11.3, but this should affect less than 0,01%

... percent of users and mostly boot param workarounds are available. Some problems may also got addressed in the kernel meanwhile and they are trying to fix problems through BIOS table overrides which do not exist anymore. I still consider the ability to override ACPI tables via initrd/boot loader to be important. Not for fixing BIOS bugs on production machines, but for debugging purposes. IMO we should try to push a multi boot grub solution which I poked Peter Anvin two times already, but never got a clear statement why this never made it into the kernel (I could google a patch on lkml around 2005, but there was no thread why it didn't go in). Eventually it's just because Peter Anvin is the x86 maintainer and he is highly biased and does not like grub... I'll poke him again the next days, also adding a grub author and you to CC. While you can override the DSDT via kernel recompilation, it is mostly a no go or a huge overhead if you want someone to test something in bug. A boot loader solution would also easily provide the ability to override arbitrary ACPI tables like Numa, IRQ affinity and all the other, more than a dozen ACPI tables which need to load much earlier than DSDT parsing. As said, nothing for production, but really convenient for debugging and OS/BIOS developement. How to proceed? Shall I remove ACPI_DSDT="" in /etc/sysconfig/kernel for now and we try to make this work again for longterm? Thomas -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-kernel+help@opensuse.org

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 06/16/2010 06:57 AM, richard -rw- weinberger wrote:

Hi,

on my Lenovo Ideapad S10 get the same kernel panic. :-( See attached image.

BTW: What is the benefit of DSDT within initramfs?

Some systems ship with ACPI tables that don't work correctly. Usually this manifests by the fans not working correctly or the battery levels not getting detected properly, but sometimes it can be more severe. As Thomas mentioned in his reply, BIOS vendors have gotten better about testing their tables in the past few years and this feature isn't really needed on newer machines. Unfortunately, there are a bunch of users who want to be able to use the latest openSUSE (and other distributions) on their older hardware and are running into usability issues as a result. Can you test a kernel from here[1]? Your failure is on a 32-bit kernel and in memcpy - which leads me to believe that it's a missing kmap/kunmap causing the failure for you. The package may take some time to build. The kernel rpm changelog should contain "Added potential fix for DSDT initramfs crash." - -Jeff [1] http://download.opensuse.org/repositories/home:/jeff_mahoney/openSUSE_11.2/

Cheers, //richard

On Wed, Jun 16, 2010 at 4:50 AM, Jeff Mahoney wrote: On 06/15/2010 04:36 PM, Christian Hartmann wrote:

...

...

...

Hi,

since I have udpated my two openSUSE 11.2 machines to 2.6.34-45 and now to -46 both with -pae and -desktop from the repository Kernel:HEAD I get the following kernel panic: "Kernel Panic - not syncing: Attempted to kill the idle task".

Booting with different parameters like noapic or acpi=off didn't help. Also the failsafe option from GRUB didn't work.

Find attached an image of the trace.

Sigh. Ok. That's the initramfs DSDT code (again). I tested it myself on a few systems and had a few other people test it as well and didn't run into problems. Figures. This code is consistently a giant headache.

-Jeff

- -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-kernel+help@opensuse.org

...

- -- Jeff Mahoney SUSE Labs -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.15 (GNU/Linux) Comment: Using GnuPG with SUSE - http://enigmail.mozdev.org/ iEYEARECAAYFAkwY6PUACgkQLPWxlyuTD7KyxACeKs5IEP4J907enwwnDwasyeXE 5dEAoJ5jyvoR4PUafhyXoWZBwtYsiFQy =VKtV -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-kernel+help@opensuse.org

On 06/16/2010 03:10 PM, richard -rw- weinberger wrote:

hi,

...

Can you test a kernel from here[1]? Your failure is on a 32-bit kernel and in memcpy - which leads me to believe that it's a missing kmap/kunmap causing the failure for you. The package may take some time to build.

you can also give me the plain patch or a src.rpm. i can compile the kernel on my own machine...

Ok. I've attached the patch. -Jeff -- Jeff Mahoney SuSE Labs

Can you test a kernel from here[1]? Your failure is on a 32-bit kernel and in memcpy - which leads me to believe that it's a missing kmap/kunmap causing the failure for you. The package may take some time to build.

The kernel rpm changelog should contain "Added potential fix for DSDT initramfs crash."

-Jeff

[1] http://download.opensuse.org/repositories/home:/jeff_mahoney/openSUSE_11.2

Will this build be available in your repo soon? By now, there is only one version from 03.06.2010 which doesn't work for me...

On Fri, Jun 18, 2010 at 3:36 PM, Jeff Mahoney wrote:

Thanks for noticing. I'd committed to a branch package so it wasn't getting built there. I've merged the branch and it's building now.

kernel-desktop-2.6.34-5.1 boots fine on my Lenovo IdeaPad. Your fix seems to work. :-) -- Cheers, //richard -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-kernel+help@opensuse.org