On Wed, Oct 15, 2008 at 7:47 PM, John Levon <levon@movementarian.org> wrote:
>        96193 47.5567 vmlinux
>        60204 29.7642 receiver
>        20059  9.9169 libc-2.5.so
>        11306  5.5896 libACE-5.6.5.so
>         8286  4.0965 anon (tgid:1093 range:0xffffe000-0xfffff000)

That looks awfully like the vsyscall region. Look at /proc/1093/maps

The row matching this range looks as follows:

ffffe000-fffff000 r-xp ffffe000 00:00 0

As expected, it is not backed by a file. I'm not sure how I can interpret this. Can you help me?

> Counted CPU_CLK_UNHALTED events (Cycles outside of halt state) with a
> unit mask of 0x00 (No unit mask) count 100000
> samples  %        image name               symbol name
> 46519    13.2671  vmlinux                  bad_gs

You gave the wrong vmlinux somehow, almost certainly. Compare an 'nm'
against /proc/kallsyms - the symbol values should be the same in every
case.

Both files contain "bad_gs" but not at the same address. Also the total number of symbols ist not the same. I guess this confirms your assumption that I'm using the wrong vmlinux. However I'm not sure how to find the right one. I am on Red Hat and therefore had to download the rpm with the vmlinux. I am sure that I picked the one for the correct kernel version number. What else do I have to keep in mind when looking for the vmlinux?
 
Thanks,

Stefan