I administer several Gentoo Linux computers at work in addition to three more computers at home. Five of these have Intel D865PERL motherboards. The rest have varying Intel compatible motherboards with Pentium chips. Of the five Intel D865PERL motherboards, none can boot a gentoo-dev-sources-2.6.7+ kernel (including 2.6.8) with ACPI enabled in the kernel. The last gentoo-dev-sources kernel that I could boot these computers with ACPI enabled was a 2.6.5-x kernel. I added a comment to <a href="http://bugs.gentoo.org/show_bug.cgi?id=53238#c4">bug report #53238</a>. Boots fail with some sort of "Bad EIP value" remark. A hard reset is required. I can boot these recent kernels if I append a "acpi=off" to the end of the kernel line in grub.conf. Reproducible: Always Steps to Reproduce: 1. Build latest gentoo-dev-sources kernel (2.6.7 and above) on a computer with an Intel D865PERL motherboard with ACPI support built in. 2. Boot fails with a "Bad EIP value." 3. Boot with old, safe kernel. 4. Append "acpi=off" to new kernel line in grub.conf. 5. Boot successfully into new kernel, but without ACPI support.
Have you tried to build development-sources with ACPI? If you haven't, could you please check it out?
Could you try development-sources-2.6.8.1 please and if you still get this issue can you reopen this bug? Thanks!
This bug looks definitely like the bug described in https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=125841 and also in http://bugzilla.kernel.org/show_bug.cgi?id=2941 Looks like especially boards with Intel's 865 chipset and CPUs supporting hyperthreading are affected. I am also suffering from this bug since i own a Intel D865GBF board with 2.6GHz HT CPU. The cause of the problem seems to be identified and a patch which should make you happy exists. Just follow the first link above and take a look at comment #78. I planned to apply and test the patch today - I'm pretty sure it will work - but I don't want to patch every single new gentoo-dev-source over and over again (the problem exists already for quite some time). You either have to wait for 2.6.9 (which will have the mentioned patch applied)or apply the patch yourselve or convince the maintainer of gentoo-dev-sources to release a fixed version ... The latest kernel version that I tested and that shows the bug is 2.6.8.1, didn' try 2.6.8.2 yet. I will test 2.6.8.2 and the reopen the bug if the problem persists. Cheers, Andreas
Oops, sorry for the double post. Dunno what happened. Ok, so after having tested 2.6.8.2 I can confirm that the problem persists. Someone will have to apply that patch, please. Andreas
Which patch? I checked the 2.6.9 tree (as of 2.6.8.1-ck4) and neither of the following two patches have been merged. https://bugzilla.redhat.com/bugzilla/attachment.cgi?id=102665&action=view http://bugzilla.kernel.org/attachment.cgi?id=3508&action=view
Please have a look at https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=125841 and specifically at comment #78: ===================================================================== Additional Comment #78 From Len Brown on 2004-08-16 21:23 ------- This bug is also reported here: http://bugzilla.kernel.org/show_bug.cgi?id=2941 and the fix in (comment #73) is in 2.6.9 tree. ===================================================================== Doesn't this mean that the patch is going to be merged in 2.6.9? Or is it Fedora's 2.6.9? I don't know. Anyway, this evening I applied the patch from http://bugzilla.kernel.org/attachment.cgi?id=3508&action=view to my gentoo-dev-sources-2.6.8-r2 sources and (after having to apply one of the hunks manually bacause of differing base kernel versions) and my problems vanished completely. That means, with ACPI enabled, now both sibling of the HT CPU are detected and the machine works like a charm. Therefore I think it would be a very good idea to have that patch integrated into gentoo-dev-sources. Cheers, Andreas
Ok, this now appears to be in -mm so I'll get this sent up for future releases.
In gentoo-dev-sources-2.6.8-r5
Considering removing this patch, its causing other problems (see bugs #66613 and 66723). My worry is that the livecd uses this kernel, and 2004.3 deadline is coming up quickly. With an issue like this (the fix creates another problem) I'd rather leave the fix out so that any problems experienced aren't our (Gentoo's) fault.
