Re: 2.6.32.5 regression: page allocation failure. order:1,

[Mark Lord <kernel@teksavvy.com>]

Rafael J. Wysocki wrote:
> Have you attached a (2.6.31.x) dmesg output from the failing box somewhere
> or have I just missed it?
..

The log from failing on 2.6.32.5 was linked to originally:

   http://rtr.ca/2.6.32_page_alloc_fail_logs.txt 

Here's the subsequent revert reboot back to 2.6.32.12:

   http://rtr.ca/2.6.31.12_ok_logs.txt

In the time available thus far, I haven't managed to get 2.6.32.5
to boot/run on anything else, using either a UP kernel, or "nosmp",
along with "mem=512M".

With a bit of luck, I'll have another similar VIA C7 box delivered here by tonight,
and should be able to boot/run that off a cloned disk image from the original system.

Then we can poke at things a bit more.
If the box works, it will be 100% at our disposal.

Note that it's far too slow for practical direct use of git-bisect,
taking hours to rebuild its own kernels.  It might be possible,
with a fair bit of effort, to arrange a quad-core box to do the
kernel compiles for it, though.

Cheers