Re: Was: boot failure, "DWARF2 unwinder stuck at 0xc0100199"

From: Jan Beulich
Date: Thu Aug 31 2006 - 03:45:20 EST


>>> Andi Kleen <ak@xxxxxxx> 31.08.06 09:41 >>>
>On Thursday 31 August 2006 09:35, Jan Beulich wrote:
>> Andi submitted a fix for this to Linus, but that's post-rc5. Jan
>
>I assume you mean the fallback validation fix. Linus unfortunately
>didn't merge any of my new patches yet :/

Actually, the same patch, but other pieces of it ...

>But did you ever work out why the stack backtrace completely restarted?
>I never got this. In theory the RSP gotten out of the unwind
>context and used for the fallback should have been already near the end
>and the old unwinder shouldn't have found much.

In the old (up to -rc5) code, we had

if (unw_ret > 0 && !arch_unw_user_mode(&info)) {
< all the fallback handling>
}

with no else, thus just falling through (without even changing the
stack pointer, which was wrong when unw_ret > 0 but we reached
a user mode address (i.e. as in the example here, after unwinding
out of a syscall frame).

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