NetBSD Problem Report #48196

From www@NetBSD.org  Mon Sep  9 05:03:43 2013
Return-Path: <www@NetBSD.org>
Received: from mail.netbsd.org (mail.netbsd.org [149.20.53.66])
	(using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits))
	(Client CN "mail.NetBSD.org", Issuer "Postmaster NetBSD.org" (verified OK))
	by mollari.NetBSD.org (Postfix) with ESMTPS id D803A7157E
	for <gnats-bugs@gnats.NetBSD.org>; Mon,  9 Sep 2013 05:03:43 +0000 (UTC)
Message-Id: <20130909050342.3D27971581@mollari.NetBSD.org>
Date: Mon,  9 Sep 2013 05:03:42 +0000 (UTC)
From: nathanialsloss@yahoo.com.au
Reply-To: nathanialsloss@yahoo.com.au
To: gnats-bugs@NetBSD.org
Subject: PAE non xen crashes immeadiately on boot up
X-Send-Pr-Version: www-1.0

>Number:         48196
>Category:       port-i386
>Synopsis:       PAE non xen crashes immeadiately on boot up
>Confidential:   no
>Severity:       critical
>Priority:       high
>Responsible:    port-i386-maintainer
>State:          closed
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Mon Sep 09 05:05:00 +0000 2013
>Closed-Date:    Fri Dec 04 05:56:19 +0000 2015
>Last-Modified:  Fri Dec 04 05:56:19 +0000 2015
>Originator:     Nat Sloss
>Release:        NetBSD Current 6.99.23
>Organization:
>Environment:
NetBSD beast 6.99.23 NetBSD 6.99.23 (LOCKDEBUG) #4: Wed Jul 17 22:24:41 EST 2013  build@beast:/usr/src/sys/arch/i386/compile/obj/LOCKDEBUG i386
>Description:
A PAE enabled kernel non xen crashes immediately 

I obtain the following from ddb:

fatal double fault in supervisor mode

trap type 13 code c027ff44 eip 8 cs 246 eflags 20 cr2 0 ilevel 0 esp cf

curlwp 0xc0e6ccc0 pid 0 lid 1 lowest kstack 0xc102c000

kernel: user trap double fault, code=0

Stopped in pid 0.1 (system) at  8:      invalid address

db{0}> bt

panic: lockdebug_lookup: uninitialized lock (lock=0xc0eb0804, from=c0921dce)

fatal breakpoint trap in supervisor mode

trap type 1 code 0 eip c02810a4 cs 8 eflags 246 cr2 8 ilevel 8 esp c102e154

curlwp 0xc0e6ccc0 pid 0 lid 1 lowest kstack 0xc102c000

Stopped in pid 0.1 (system) at  netbsd:breakpoint+0x4:  popl    %ebp

>How-To-Repeat:
Build a i386 kernel with options PAE andd try to boot it.
>Fix:
The problem is due to the fact that interrupts are enabled before they should be.

So to fix it I applied this patch:

--- ./cpu.c     2013-09-06 12:30:20.000000000 +1000
+++ .#cpu.c.1.102       2012-12-20 15:02:28.000000000 +1100
@@ -1276,6 +1276,7 @@
        struct cpu_info *ci = curcpu();
        pd_entry_t *l3_pd = ci->ci_pae_l3_pdir;
        int i;
+       static int intrEnOK = 0;

        /*
         * disable interrupts to block TLB shootdowns, which can reload cr3.
@@ -1286,7 +1287,11 @@
        for (i = 0 ; i < PDP_SIZE; i++) {
                l3_pd[i] = pmap->pm_pdirpa[i] | PG_V;
        }
-       x86_enable_intr();
+       if (intrEnOK > 1)
+               x86_enable_intr();
+       else
+               intrEnOK++;
+
        tlbflush();
 #else /* PAE */
        lcr3(pmap_pdirpa(pmap, 0));

I would like too request that NetBSD-6 be pulled up and that i386/44995 be reopened as it works in NetBSD current and it is critical to NetBSD-6.

Regards,

Nat.

>Release-Note:

>Audit-Trail:
From: scole_mail <scole_mail@gmx.com>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: port-i386/48196
Date: Wed, 16 Sep 2015 10:31:01 -0400

 I am getting the exact same error with 7.0_RC3 and GENERIC with PAE
 option. Without PAE, things seem to work fine.

 Here is my uname:

 NetBSD ugly 7.0_RC3 NetBSD 7.0_RC3 (GENERIC) #0: Wed Sep 16 10:11:01 EDT 2015  root@ugly:/usr/src/sys/arch/i386/compile/GENERIC i386 

 I didn't try it with the suggested fix in this PR, but can if requested.

 Thanks

From: Nathanial Sloss <nat@netbsd.org>
To: gnats-bugs@netbsd.org
Cc: port-i386-maintainer@netbsd.org,
 gnats-admin@netbsd.org,
 netbsd-bugs@netbsd.org,
 scole_mail@gmx.com
Subject: Re: port-i386/48196
Date: Thu, 17 Sep 2015 10:14:26 +1000

 Hi,

 Please try the following patch:
 Index: cpu.c
 ===================================================================
 RCS file: /cvsroot/src/sys/arch/x86/x86/cpu.c,v
 retrieving revision 1.115
 diff -u -r1.115 cpu.c
 --- cpu.c	18 May 2015 13:09:55 -0000	1.115
 +++ cpu.c	16 Sep 2015 23:57:35 -0000
 @@ -137,6 +137,8 @@
  	bool sc_wasonline;
  };

 +#define EFLAGS_IF	__BIT(9)
 +
  #ifdef MULTIPROCESSOR
  int mp_cpu_start(struct cpu_info *, paddr_t); 
  void mp_cpu_start_cleanup(struct cpu_info *);
 @@ -1277,6 +1279,7 @@
  {
  #ifdef PAE
  	struct cpu_info *ci = curcpu();
 +	ulong IntrOK;
  	pd_entry_t *l3_pd = ci->ci_pae_l3_pdir;
  	int i;

 @@ -1285,11 +1288,14 @@
  	 * while this doesn't block NMIs, it's probably ok as NMIs unlikely
  	 * reload cr3.
  	 */
 +	IntrOK = x86_read_flags() & EFLAGS_IF;
  	x86_disable_intr();
  	for (i = 0 ; i < PDP_SIZE; i++) {
  		l3_pd[i] = pmap->pm_pdirpa[i] | PG_V;
  	}
 -	x86_enable_intr();
 +	
 +	if (IntrOK)
 +		x86_enable_intr();
  	tlbflush();
  #else /* PAE */
  	lcr3(pmap_pdirpa(pmap, 0));


 If you are successful with pae please let me know.

 Best regards,

 Nat.

From: matthew green <mrg@eterna.com.au>
To: Nathanial Sloss <nat@netbsd.org>
Cc: port-i386-maintainer@netbsd.org, gnats-admin@netbsd.org,
    netbsd-bugs@netbsd.org, scole_mail@gmx.com, gnats-bugs@netbsd.org
Subject: re: port-i386/48196
Date: Thu, 17 Sep 2015 12:43:46 +1000

 > Please try the following patch:

 this patch works for me (and without it, failure.)

 thanks.


 .mrg.

From: Masanobu SAITOH <msaitoh@execsw.org>
To: Nathanial Sloss <nat@netbsd.org>, gnats-bugs@netbsd.org
Cc: msaitoh@execsw.org, port-i386-maintainer@netbsd.org,
 gnats-admin@netbsd.org, netbsd-bugs@netbsd.org, scole_mail@gmx.com
Subject: Re: port-i386/48196
Date: Thu, 17 Sep 2015 18:28:58 +0900

 Hi.

 On 2015/09/17 9:14, Nathanial Sloss wrote:
 > Hi,
 >
 > Please try the following patch:
 > Index: cpu.c
 > ===================================================================
 > RCS file: /cvsroot/src/sys/arch/x86/x86/cpu.c,v
 > retrieving revision 1.115
 > diff -u -r1.115 cpu.c
 > --- cpu.c	18 May 2015 13:09:55 -0000	1.115
 > +++ cpu.c	16 Sep 2015 23:57:35 -0000
 > @@ -137,6 +137,8 @@
 >   	bool sc_wasonline;
 >   };
 >
 > +#define EFLAGS_IF	__BIT(9)

   I think this flags is PSL_I in x86/include/psl.h




 > +
 >   #ifdef MULTIPROCESSOR
 >   int mp_cpu_start(struct cpu_info *, paddr_t);
 >   void mp_cpu_start_cleanup(struct cpu_info *);
 > @@ -1277,6 +1279,7 @@
 >   {
 >   #ifdef PAE
 >   	struct cpu_info *ci = curcpu();
 > +	ulong IntrOK;
 >   	pd_entry_t *l3_pd = ci->ci_pae_l3_pdir;
 >   	int i;
 >
 > @@ -1285,11 +1288,14 @@
 >   	 * while this doesn't block NMIs, it's probably ok as NMIs unlikely
 >   	 * reload cr3.
 >   	 */
 > +	IntrOK = x86_read_flags() & EFLAGS_IF;
 >   	x86_disable_intr();
 >   	for (i = 0 ; i < PDP_SIZE; i++) {
 >   		l3_pd[i] = pmap->pm_pdirpa[i] | PG_V;
 >   	}
 > -	x86_enable_intr();
 > +	
 > +	if (IntrOK)
 > +		x86_enable_intr();
 >   	tlbflush();
 >   #else /* PAE */
 >   	lcr3(pmap_pdirpa(pmap, 0));
 >
 >
 > If you are successful with pae please let me know.
 >
 > Best regards,
 >
 > Nat.
 >


 -- 
 -----------------------------------------------
                  SAITOH Masanobu (msaitoh@execsw.org
                                   msaitoh@netbsd.org)

From: scole_mail <scole_mail@gmx.com>
To: Nathanial Sloss <nat@netbsd.org>
Cc: gnats-bugs@netbsd.org, port-i386-maintainer@netbsd.org,
	gnats-admin@netbsd.org, netbsd-bugs@netbsd.org
Subject: Re: port-i386/48196
Date: Thu, 17 Sep 2015 09:59:49 -0400

 I tried the patch and built a release, and it worked for me.

 Thanks!

From: "Nathanial Sloss" <nat@netbsd.org>
To: gnats-bugs@gnats.NetBSD.org
Cc: 
Subject: PR/48196 CVS commit: src/sys/arch/x86/x86
Date: Thu, 17 Sep 2015 23:48:02 +0000

 Module Name:	src
 Committed By:	nat
 Date:		Thu Sep 17 23:48:01 UTC 2015

 Modified Files:
 	src/sys/arch/x86/x86: cpu.c

 Log Message:
 Don't disable/re-enable interrupts if they are already disabled.
 Addresses PR 48196.

 This commit was improved and approved by christos@


 To generate a diff of this commit:
 cvs rdiff -u -r1.115 -r1.116 src/sys/arch/x86/x86/cpu.c

 Please note that diffs are not public domain; they are subject to the
 copyright notices on the relevant files.

State-Changed-From-To: open->pending-pullups
State-Changed-By: nat@NetBSD.org
State-Changed-When: Fri, 18 Sep 2015 00:10:10 +0000
State-Changed-Why:
pullup-7 #984


From: "Jeff Rizzo" <riz@netbsd.org>
To: gnats-bugs@gnats.NetBSD.org
Cc: 
Subject: PR/48196 CVS commit: [netbsd-7] src/sys/arch/x86/x86
Date: Fri, 6 Nov 2015 00:40:03 +0000

 Module Name:	src
 Committed By:	riz
 Date:		Fri Nov  6 00:40:02 UTC 2015

 Modified Files:
 	src/sys/arch/x86/x86 [netbsd-7]: cpu.c

 Log Message:
 Pull up following revision(s) (requested by nat in ticket #984):
 	sys/arch/x86/x86/cpu.c: revision 1.116
 Don't disable/re-enable interrupts if they are already disabled.
 Addresses PR 48196.
 This commit was improved and approved by christos@


 To generate a diff of this commit:
 cvs rdiff -u -r1.111.2.1 -r1.111.2.2 src/sys/arch/x86/x86/cpu.c

 Please note that diffs are not public domain; they are subject to the
 copyright notices on the relevant files.

From: "Jeff Rizzo" <riz@netbsd.org>
To: gnats-bugs@gnats.NetBSD.org
Cc: 
Subject: PR/48196 CVS commit: [netbsd-7-0] src/sys/arch/x86/x86
Date: Sun, 8 Nov 2015 00:15:47 +0000

 Module Name:	src
 Committed By:	riz
 Date:		Sun Nov  8 00:15:47 UTC 2015

 Modified Files:
 	src/sys/arch/x86/x86 [netbsd-7-0]: cpu.c

 Log Message:
 Pull up following revision(s) (requested by nat in ticket #984):
 	sys/arch/x86/x86/cpu.c: revision 1.116
 Don't disable/re-enable interrupts if they are already disabled.
 Addresses PR 48196.
 This commit was improved and approved by christos@


 To generate a diff of this commit:
 cvs rdiff -u -r1.111.2.1 -r1.111.2.1.2.1 src/sys/arch/x86/x86/cpu.c

 Please note that diffs are not public domain; they are subject to the
 copyright notices on the relevant files.

State-Changed-From-To: pending-pullups->closed
State-Changed-By: nat@NetBSD.org
State-Changed-When: Fri, 04 Dec 2015 05:56:19 +0000
State-Changed-Why:
Pullups committed.  Thanks to all involved.


>Unformatted:

Home	PR Database Search