Thread (10 messages) 10 messages, 5 authors, 2007-05-12

Re: APIC error on 32-bit kernel

From: Eric W. Biederman <hidden>
Date: 2007-03-28 01:04:57

Jay Cliburn [off-list ref] writes:
On Tue, 27 Mar 2007 14:42:20 -0600
ebiederm@xmission.com (Eric W. Biederman) wrote:

Thanks for replying, Eric.  I've added atl1-devel to the cc list.
quoted
Do you have msi's working in 2.6.21-rc4 in the x86_64 kernel?
I can't personally verify it anymore because I removed x86_64 to
duplicate the MSI problem on i386, but the driver was working fine under
x86_64 in earlier versions of 2.6.21-rcX.  The first hint of a problem
was a report on March 12 by a user running a 32-bit 2.6.19 Fedora 6
kernel who encountered a kernel panic on network startup.
Just FYI: You can run a 64bit kernel with a 32bit user space...

That this happens as long ago as 2.6.19 is interesting.  That indicates
this is not a brand new problem, and it also makes the 64bit kernel
test a little less interesting.

quoted
quoted
We also do not see this problem on Intel-based motherboards, with
either 32- or 64-bit kernels.
Can you confirm MSI is enabled in those kernels as well?
Absolutely, yes.  MSI is enabled and working for me on a 64-bit
kernel on an Intel-based motherboard, and Luca Tettamanti reports no
problems running a 32-bit kernel on a similar motherboard.  (Luca wrote
the MSI patch for the atl1 driver.)

We enable MSI by default in the driver.

I can now stimulate a kernel oops by pinging my router.  Here's the
console output.

...snip (nothing but a flood of APIC errors above here)...
[  103.052000] APIC error on CPU1: 08(08)
[  103.052000] APIC error on CPU0: 08(08)
[  103.154000] APIC error on CPU1: 08(08)
[  103.154000] APIC error on CPU0: 08(08)
[  103.256000] APIC error on CPU1: 08(08)
[  103.256000] APIC error on CPU0: 08(08)
[  103.359000] APIC error on CPU1: 08(08)
[  103.359000] APIC error on CPU0: 08(08)
[  103.461000] APIC error on CPU1: 08(08)
[  103.461000] APIC error on CPU0: 08(08)

pinged router somewhere about here...
Ok.  I need to look back the initial APIC errors had a different error
code and I think if we decode that it may give us a clue about what
is going on.

Eric
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help