Bug 9505

Created attachment 13866 [details]
2.6.24-rc3-dmesg

Created attachment 13867 [details]
2.6.22-dmesg

Firs this bug was reported to ubuntu backtrack, see:
https://bugs.launchpad.net/ubuntu/+source/linux-source-2.6.22/+bug/153096

Created attachment 13868 [details]
debug.patch

Created attachment 13869 [details]
debug.patch

Created attachment 13870 [details]
2.6.24-rc3-dmesg

Created attachment 13879 [details]
Log file with debug patch enabled.

This is a log file of booting the 2.6.24-1 kernel on my PC with the debug patch mentioned above in kernel. I am now using 2.6.20(-16, Ubuntu feisty) and the harddrives both work great. Only one works in newer kernels.

SATA controller is Sil 3112 SATA Controller onboard on a Abit NF7-S v2.0 motherboard. Chipset is NVIDIA nForce2 SPP.

Created attachment 13881 [details]
sata_sil-irq-fix.patch

Please test whether the attached patch fixes the problem.

Created attachment 13888 [details]
dmesg output of problem from 2.6.22 kernel

This is the dmesg from 2.6.22 bootup using a generic SIL3112 card and a HD attached with a SATA->PATA bridge, a SD-ADSAIDE-S1, with a sil3611ct80 chip on board.

The patch works for me.
- kernel 2.6.24-rc4
- Abit NF7-S mobo with onboard sil3112, hard drive attached with a Abit Serillel pata-to-sata bridge (I think it uses the sil3611 chip).

Now I can finally upgrade the kernel... I've been stuck with 2.6.21 for ages :)

Created attachment 13890 [details]
Reject from trying to apply sata_sil-irq-fix patch to 2.6.22 source

Created attachment 13891 [details]
Reject from trying to apply sata_sil-irq-fix patch to 2.6.22 source

The patch didn't apply cleanly, and when I tried to apply the patch by hand and compile, I get the following error:

drivers/ata/libata-core.c: In function 'ata_std_postreset':
drivers/ata/libata-core.c:3583: error: 'link' undeclared (first use in this function)
drivers/ata/libata-core.c:3583: error: (Each undeclared identifier is reported only once
drivers/ata/libata-core.c:3583: error: for each function it appears in.)

@Mark This patch is against 2.6.24-rc4

Alexey, can you confirm the fix too?

Alexey filed it here on our behalf as the bug originally was in Ubuntu's launchpad. He doesn't have this problem himself. I tried to build and install a kernel, but I got trouble when installing it (not related to your code). I will try to build it again in some hours.

Here is my kernel i use on my system for debug purpose, there is no sound and should be no network for you. sata_sil is enabled and patched, at leas this should work.
http://rapidshare.com/files/74878489/linux-image-2.6.24-rc4-patched_2.6.24-rc4-patched-10.00.Custom_i386.deb

check if it complete:
md5sum linux-image-2.6.24-rc4-patched_2.6.24-rc4-patched-10.00.Custom_i386.deb
should be: 8c6198b90c2cf050023d9e4529aab838

to install use:
sudo dpkg -i linux-image-2.6.24-rc4-patched_2.6.24-rc4-patched-10.00.Custom_i386.deb

to remove:
sudo dpkg -r linux-image-2.6.24-rc4-patched

Patch works! :)

Thanks a lot Tejun for the fix and Alexey for all the help debugging and making it work!

Created attachment 13903 [details]
sata_sil-irq-fix patch from my  2.6.22 build environment.

This is the sata_sil-irq-fix patch file, but name and location adjusted for the 2.6.22 kernel.  It works fine there after changing the name in libaba-core.c.

My 2.6.22 kernel is now booting correctly.

Yeah, this one is for -stable too.  I'll forward the patch upstream and -stable.  Resolving as CODE_FIX.  Thanks.