NetBSD Problem Report #46275

From www@NetBSD.org  Wed Mar 28 15:21:47 2012
Return-Path: <www@NetBSD.org>
Received: from mail.netbsd.org (mail.netbsd.org [149.20.53.66])
	by www.NetBSD.org (Postfix) with ESMTP id 7197A63E3AC
	for <gnats-bugs@gnats.NetBSD.org>; Wed, 28 Mar 2012 15:21:47 +0000 (UTC)
Message-Id: <20120328152146.8FF4363BBEC@www.NetBSD.org>
Date: Wed, 28 Mar 2012 15:21:46 +0000 (UTC)
From: reinoud@NetBSD.org
Reply-To: reinoud@NetBSD.org
To: gnats-bugs@NetBSD.org
Subject: Kernel panics on heavy use, kevent / pipes related?
X-Send-Pr-Version: www-1.0

>Number:         46275
>Category:       kern
>Synopsis:       Kernel panics on heavy use, kevent / pipes related?
>Confidential:   no
>Severity:       critical
>Priority:       high
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Wed Mar 28 15:25:01 +0000 2012
>Last-Modified:  Wed Mar 28 15:40:02 +0000 2012
>Originator:     Reinoud Zandijk
>Release:        NetBSD 6.0_BETA
>Organization:
NetBSD
>Environment:
NetBSD heethoofdje.13thmonkey.org 6.0_BETA NetBSD 6.0_BETA (GENERIC) #0: Sat Mar 24 22:55:30 CET 2012  reinoud@heethoofdje.13thmonkey.org:/usr/sources/cvs.netbsd.org/src-6/sys/arch/i386/compile/GENERIC i386

>Description:
When compiling Haskell packages, the machine can suddenly reset. First i thought it might be memory related but further investigation showed that it kept crashing at exactly the same place. Therefore i inserted assertions to test what NULL pointer got dereferenced and today it struck at the same place as usual, only now in one of my asserts:

panic: kernel assertion "((file_t *) kn->kn_obj)->f_data" failed: file "../../../../kern/sys_pipe.c", line 1387
cpu0: Begin traceback...
kern_assert(c0b706b0,c0c19753,c0c045d0,c0c043f0,56b,4c,dde1bacc,c07abc3d,c57a2b80,c5857d20) at netbsd:kern_assert+0x23
filt_piperead(c5b9caf0,0,c0c40800,c0564afc,c5857d20,0,dde1bafc,c055e78d,c0c40800,c2f91d20) at netbsd:filt_piperead+0x1af
kevent1(dde1bd1c,4,0,0,bb7fc010,40,0,c0b11490,dde1bd48,c0c997cc) at netbsd:kevent1+0x61d
sys___kevent50(c5857d20,dde1bcf4,dde1bd1c,c0652ea7,db7e3dc0,0,c5857d20,c5617440,c068fc65,c5617454) at netbsd:sys___kevent50+0x45
syscall(dde1bd48,99700b3,bb4000ab,bfbf001f,bbad001f,bb723600,bb400000,bb5fdcb4,bbb8f178,9da95fc) at netbsd:syscall+0xad
cpu0: End traceback...

The code is around src/sys/kern/sys_pipe.c:1355

static void
filt_pipedetach(struct knote *kn)
{
        struct pipe *pipe;
        kmutex_t *lock;

        pipe = ((file_t *)kn->kn_obj)->f_data;
        lock = pipe->pipe_lock;
*panic* due to deference of NULL struct pipe*

        mutex_enter(lock);


>How-To-Repeat:
Do heavy compilation work, like ghc Haskell and some Haskell packages and it can show up. No idea why it showed up in Haskell, but maybe its using some form of pipe-snooping.
>Fix:
unknoen

>Audit-Trail:
From: Reinoud Zandijk <reinoud@NetBSD.org>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: kern/46275: Kernel panics on heavy use, kevent / pipes related?
Date: Wed, 28 Mar 2012 17:40:07 +0200

 Sorry for the misunderstanding, i quoted the wrong piece of code though its
 nearly identical. It should have been line 1387:

 ----------
 static int
 filt_piperead(struct knote *kn, long hint)
 {
         assert(kn);
         assert(kn->kn_obj);
         assert(((file_t *) kn->kn_obj)->f_data);	* TRIGGERS *

         struct pipe *rpipe = ((file_t *)kn->kn_obj)->f_data; struct pipe *wpipe;

         if ((hint & NOTE_SUBMIT) == 0) {
                 assert(rpipe);
                 mutex_enter(rpipe->pipe_lock);
         }
 -----------

Home	PR Database Search