lkml.org 
[lkml]   [2002]   [Oct]   [2]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
Date
SubjectRe: [RFC][PATCH] 4KB stack + irq stack for x86
On Oct 02, 2002  14:18 -0700, Dave Hansen wrote:
> I've resynced Ben's patch against 2.5.40. However, I'm getting some
> strange failures. The patch is good enough to pass LTP, but
> consistently freezes when I run tcpdump on it.
>
> Although I don't have CONFIG_PREEMPT on, I have the feeling that I
> need to disable preemption in common_interrupt() like it was before.
> Any insights would be appreciated.

I'm a little bit worried about this patch. Have you tried something
like NFS-over-ext3-over-LVM-over-MD or so, which can have a deep stack?

We hit a bunch of deep stack problems like this (overflowing an 8kB stack)
even without interrupts involved when developing Lustre. Granted, we
fixed some large stack allocations in the ext3 indexed-directory code
and in our own code, but I'm still worried that a 4kB stack is too small.

The Stanford checker folks would probably be able to run a test for
large stack allocations in 2.5.40 if you asked them nicely, and maybe
even do stack depths for call chains.

Alternately, you could set up an 8kB stack + IRQ stack and "red-zone"
the high page of the current 8kB stack and see if it is ever used.

Cheers, Andreas
--
Andreas Dilger
http://www-mddsp.enel.ucalgary.ca/People/adilger/
http://sourceforge.net/projects/ext2resize/

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:29    [W:0.043 / U:0.392 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site