Re: Interesting lock issue mtx_flags = 0x6?

From: Randall Stewart <rrs_at_cisco.com>
Date: Mon, 04 Sep 2006 08:04:59 -0400
Never mind.. I think I just figiured it out.. its
a foo-bar on my part :-D

R

Randall Stewart wrote:
> Hi all:
> 
> I am hunting a different bug (crash) in my SCTP code.. and I ran
> across an interesting issue.
> 
> The kernel crashed in mtx_lock() at line 544.. looking
> at the kernel core... I see the mtx_flags is set to 0x6
> (MTX_RELEASED|MTX_CONTESTED)... The turnstile was
> empty as well..
> 
> My last update to kern_mutex.c was about August 15th..
> 
> Is this a known issue? Should I try cvsup'ing and see
> if I can reproduce it.. or does anyone with a lot
> more knowledge on how the locking is supposed to work
> want to look at this before I cvsup and loose the ability
> to kgdb it (I would be glad to give access to anyone that
> wants to poke around on the machine).
> 
>  From my limited look at the atomic's that are going on
> here I can't see how you could get 0x6 ever set in
> the flags.. and of course this is also assumed in
> the lock code and is why it crashes...
> 
> Let me know if anyone is interested in looking at this..
> 
> R


-- 
Randall Stewart
NSSTG - Cisco Systems Inc.
803-345-0369 <or> 815-342-5222 (cell)
Received on Mon Sep 04 2006 - 10:10:29 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:38:59 UTC