Bug 198810 – linux boot hangs with intel Q35 chipset

• Status: RESOLVED FIXED
• Product: Gentoo Linux

[Description vltg0903 2007-11-11 14:31:54 UTC]

The linux kernel boot will hang at ACPI initialisation unless acpi=off or pci=nommconf are used as boot parameters. The kernel devs are aware of the problem:
http://kerneltrap.org/mailarchive/linux-kernel/2007/8/26/164870
but as of linux-2.6.24_rc2, the patch from the discussion hasn't made it into mainline. There's talk about BIOS updates fixing the issue for G33 chipsets, but not so for Q35 (I'm using the latest BIOS for the intel DQ35JO board). I've tested the patch against vanilla-sources-2.6.24_rc2 and haven't noticed any negative effects so far.

Reproducible: Always

[Comment 1 vltg0903 2007-11-11 20:02:23 UTC]

Created attachment 135761 [details]
Q35 boot fix patch

[Comment 2 Mike Pagano 2007-11-12 14:24:14 UTC]

There seems to still be quite a bit of discussions concerning the validity of this patch. I think it might be worth it to keep an eye on the upstream thread until an accepted patch is proposed and committed.

[Comment 3 Mike Pagano 2007-11-23 18:23:41 UTC]

Please reopen this bug if a patch that fixes this get accepted upstream. We will then attempt to backport the patch to older kernels that we still support at that time.

http://dev.gentoo.org/~dsd/genpatches/faq.htm

[Comment 4 Daniel Drake (RETIRED) 2007-11-25 19:51:10 UTC]

Mike: have you confirmed that upstream has everything under control and doesn't need our help? Unless upstream tell us to go away, we should pester them and offer resources -- namely Andreas who would hopefully test candidate patches should they appear. We shouldn't close this unless there is nothing we can do.

[Comment 5 Mike Pagano 2007-11-26 01:35:02 UTC]

If you have the time, read the thread from the original description. The discussion seems to have lost some steam and stopped.

If you think after reading there is more that we can do, we can reopen this.

[Comment 6 Daniel Drake (RETIRED) 2007-11-26 21:46:24 UTC]

we can at least attempt to breathe some life back into the issue

[Comment 7 vltg0903 2007-12-16 00:58:40 UTC]

The problem has been brought to the attention of the master himself, and with any look above patch will get included soon (preliminarily at least):

http://lkml.org/lkml/2007/12/6/368

[Comment 8 vltg0903 2007-12-16 00:59:16 UTC]

Luck, not look, of course.

[Comment 9 vltg0903 2008-01-26 21:26:42 UTC]

For the record: The fix is in mm-sources and might make it into 2.6.25:
http://www.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.24-rc6/2.6.24-rc6-mm1/broken-out/pci-disable-decoding-during-sizing-of-bars.patch

The only thing that troubles me is that my name is on the top of the page (they took it from this bug), while the patch is by Matthew Wilcox...

[Comment 10 Daniel Drake (RETIRED) 2008-01-26 22:04:46 UTC]

thanks for the update and your efforts, I just sent an email asking Andrew to correct that.

[Comment 11 Daniel Drake (RETIRED) 2008-02-12 16:53:04 UTC]

Created attachment 143323 [details, diff]
new patch

Rumour has it that this patch solves the issue, and it's already upstream. Please could you apply it (*instead* of any others) to 2.6.24 and let us know what happens?

[Comment 12 vltg0903 2008-02-16 21:29:08 UTC]

Daniel, the patch works well here.

[Comment 13 Daniel Drake (RETIRED) 2008-02-16 21:57:39 UTC]

Great, thanks for testing. Queued this up for next 2.6.24 release.
Not going to backport to 2.6.23 because the code the patch touches on has changed too much.

[Comment 14 Daniel Drake (RETIRED) 2008-02-26 17:10:46 UTC]

Fixed in genpatches-2.6.24-4 (gentoo-sources-2.6.24-r3), thanks for testing