NetBSD Problem Report #57839
From martin@duskware.de Fri Jan 12 18:07:24 2024
Return-Path: <martin@duskware.de>
Received: from mail.netbsd.org (mail.netbsd.org [199.233.217.200])
(using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits))
(Client CN "mail.NetBSD.org", Issuer "mail.NetBSD.org CA" (not verified))
by mollari.NetBSD.org (Postfix) with ESMTPS id 1B5101A9238
for <gnats-bugs@gnats.NetBSD.org>; Fri, 12 Jan 2024 18:07:24 +0000 (UTC)
From: martin@NetBSD.org
Reply-To: martin@NetBSD.org
To: gnats-bugs@NetBSD.org
Subject: Reproducable VAX pmap bug
X-Send-Pr-Version: 3.95
>Number: 57839
>Category: port-vax
>Synopsis: Reproducable VAX pmap bug
>Confidential: no
>Severity: critical
>Priority: high
>Responsible: port-vax-maintainer
>State: feedback
>Class: sw-bug
>Submitter-Id: net
>Arrival-Date: Fri Jan 12 18:10:00 +0000 2024
>Closed-Date:
>Last-Modified: Wed Nov 11 18:30:03 +0000 2026
>Originator: Martin Husemann
>Release: NetBSD 10.99.10
>Organization:
The NetBSD Foundation, Inc.
>Environment:
System: NetBSD dead-to-the-world.duskware.de 10.99.10 NetBSD 10.99.10 (DEAD) #19: Wed Jan 10 15:59:25 CET 2024 martin@seven-days-to-the-wolves.aprisoft.de:/work/src/sys/arch/vax/compile/DEAD vax
Architecture: vax
Machine: vax
>Description:
Running the atf test suite crashes the kernel:
[ 2155.0184746] panic: pmap_destroy used pmap
[ 2155.0309698] cpu0: Begin traceback...
[ 2155.0447079] panic: pmap_destroy used pmap
[ 2155.0537111] Stack traceback :
[ 2155.0611692] Process is executing in user space.
[ 2155.0684572] cpu0: End traceback...
Stopped in pid 337.337 (t_lwp_create) at netbsd:vpanic+0x179: pushl
$0
db> bt
panic: pmap_destroy used pmap
Stack traceback :
0x8b9b8d40: vpanic+0x179(0x802116e5,0x8b9b8dd8)
0x8b9b8d60: printf_nostamp+0x0(0x802116e5)
0x8b9b8d8c: pmap_kenter_pa+0x0(0x87687d40)
0x8b9b8ddc: uvmspace_free+0x73(0x867d2900)
0x8b9b8e04: uvm_proc_exit+0x53(0x85a09530)
0x8b9b8e28: exit1+0x15b(0x877a6ac0,0,0)
0x8b9b8ed8: sys_exit+0x31(0x877a6ac0,0x8b9b8f54,0x8b9b8f4c)
0x8b9b8f18: syscall+0x126(0x8b9b8fb4)
>How-To-Repeat:
cd /usr/tests/lib/libc/sys && atf-run t_lwp_create
>Fix:
n/a
>Release-Note:
>Audit-Trail:
From: Jan-Benedict Glaw <jbglaw@lug-owl.de>
To: gnats-bugs@netbsd.org
Cc:
Subject: Re: port-vax/57839
Date: Thu, 5 Nov 2026 23:45:01 +0100
--oNku53dHTcHtLiB4
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable
Hi,
first of all, this issue is still reproducible with an
DIAGNOSTIC-enabled kernel (here in SIMH; needed to upsize the miniroot
fs from 8m to 10m):
[ 1398.1366770] panic: pmap_destroy used pmap
[ 1398.1366770] cpu0: Begin traceback...
[ 1398.1466792] Stack traceback:
[ 1398.3567255] 0xa71e9d44: vpanic+0x17c(0x8039afcf,0xa71e9ddc)
[ 1398.5867762] 0xa71e9d64: aprint_verbose+0x0(0x8039afcf)
[ 1398.7968225] 0xa71e9d90: pmap_kenter_pa+0x0(0x9f525740)
[ 1399.0168710] 0xa71e9de0: uvmspace_free+0x7b(0x9fea4d80)
[ 1399.2069129] 0xa71e9e08: uvm_proc_exit+0x53(0x9f8ca640)
[ 1399.2869305] 0xa71e9e2c: exit1+0x170(0x9f8ae880,0,0)
[ 1399.3669481] 0xa71e9ed8: sys_exit+0x31(0x9f8ae880,0xa71e9f54,0xa71e9f4c)
[ 1399.5169812] 0xa71e9f18: syscall+0xec(0xa71e9fb4)
[ 1399.5169812] cpu0: End traceback...
[ 1399.5169812] dump to dev 9,1 not possible
HALT instruction, PC: 80013FFC (PUSHAB 8039B207)
=2E..while on a non-DIAG kernel, the test just passes.
I think there's still a pm_pcbs referencing the pmap when it is fed to
pmap_destroy, It's always linked in through cpu_lwp_fork(), and
cleared through pmap_deactivate() called through mi_switch(). But that
doesn't happen if the LWP never runs (eg. through error pathes.) The
testcase (lwp_create_bad_lid_ptr) provokes such an error path.
Right now, I'm building a new kernel that invokes pmap_deactivate()
in cpu_lwp_free() IFF l !=3D curlwp. That should be enough I hope...
Martin, ISTR that you stated something like you'd like to run ATF
more often on VAX, but you tend to have DIAGNOSTIC enabled. I really
hope this catches it!
MfG, JBG
--=20
--oNku53dHTcHtLiB4
Content-Type: application/pgp-signature; name="signature.asc"
-----BEGIN PGP SIGNATURE-----
iF0EABECAB0WIQQlDTvPcScNjKREqWEdvV51g5nhuwUCau0G6QAKCRAdvV51g5nh
u6gsAJ4vOkf+lTixb/sCvzFZcqdWdxayHgCdFZn421agAm00YQH9f16/NE8lW14=
=hldp
-----END PGP SIGNATURE-----
--oNku53dHTcHtLiB4--
From: Jan-Benedict Glaw <jbglaw@lug-owl.de>
To: gnats-bugs@netbsd.org
Cc:
Subject: Re: port-vax/57839
Date: Sun, 8 Nov 2026 08:44:45 +0100
--whbJPoPwx1YUiGt1
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable
Hi!
With the following patch applied, I can sucessfully run this test
case:
vax-template# cd /usr/tests/lib/libc/sys && atf-run t_lwp_create
Content-Type: application/X-atf-tps; version=3D"3"
info: atf.version, Automated Testing Framework 0.20 (atf-0.20)
info: tests.root, /usr/tests/lib/libc/sys
info: time.start, Sun Nov 8 07:32:08 UTC 2026
info: uname.sysname, NetBSD
info: uname.nodename, vax-template.classiccmp
info: uname.release, 11.99.9
info: uname.version, NetBSD 11.99.9 (GENERIC) #0: Wed Oct 28 09:49:53 UTC 2=
026 mkrepro@mkrepro.NetBSD.org:/usr/src/sys/arch/vax/compile/GENERIC
info: uname.machine, vax
info: env, ENV=3D/root/.shrc
info: env, BLOCKSIZE=3D1k
info: env, PWD=3D/usr/tests/lib/libc/sys
info: env, MAIL=3D/var/mail/root
info: env, HOME=3D/root
info: env, PATH=3D/sbin:/usr/sbin:/bin:/usr/bin:/usr/pkg/sbin:/usr/pkg/bin:=
/usr/games:/usr/X11R7/bin:/usr/local/sbin:/usr/local/bin
info: env, HOST=3Dvax-template.classiccmp
info: env, SSH_CONNECTION=3D10.234.0.1 54068 10.234.2.7 22
info: env, SSH_TTY=3D/dev/pts/0
info: env, TERM=3Dscreen5
info: env, OLDPWD=3D/root
info: env, USER=3Droot
info: env, SSH_CLIENT=3D10.234.0.1 54068 22
info: env, PKG_PATH=3Dhttp://ftp.netbsd.org/pub/pkgsrc/packages/NetBSD/vax/=
10.0_2022Q4/All
info: env, LOGNAME=3Droot
info: env, SHELL=3D/bin/sh
tps-count: 1
tp-start: 1794123148.564015, t_lwp_create, 6
tc-start: 1794123148.584015, lwp_create_bad_lid_ptr
tc-end: 1794123150.504058, lwp_create_bad_lid_ptr, passed
tc-start: 1794123150.574059, lwp_create_generic_fail_no_uc_cpu
tc-end: 1794123151.524081, lwp_create_generic_fail_no_uc_cpu, passed
tc-start: 1794123151.584082, lwp_create_vax_fail_psl_0
tc-end: 1794123152.574104, lwp_create_vax_fail_psl_0, passed
tc-start: 1794123152.644105, lwp_create_vax_fail_psl_1
tc-end: 1794123153.624127, lwp_create_vax_fail_psl_1, passed
tc-start: 1794123153.694129, lwp_create_vax_fail_psl_cm
tc-end: 1794123154.504147, lwp_create_vax_fail_psl_cm, passed
tc-start: 1794123154.594149, lwp_create_works
tc-end: 1794123155.574171, lwp_create_works, passed
tp-end: 1794123155.644172, t_lwp_create
info: time.end, Sun Nov 8 07:32:39 UTC 2026
vax-template#=20
=2E..without crashing, that is.
root@VPS2264-jenkins-node1:/var/cache/git/buildhelpers/patches [master] # c=
at generic-netbsd--jbglaw-pr57839-enable-DIAGNOSTIC.patch generic-netbsd--j=
bglaw-pr57839-lwp-free-pcb.patch=20
diff --git a/distrib/vax/miniroot/Makefile.inc b/distrib/vax/miniroot/Makef=
ile.inc
index 37fbd138a3c1..f272794ee7e0 100644
--- a/distrib/vax/miniroot/Makefile.inc
+++ b/distrib/vax/miniroot/Makefile.inc
@@ -1,6 +1,6 @@
# $NetBSD: Makefile.inc,v 1.25 2017/01/24 18:04:05 christos Exp $
=20
-IMAGESIZE=3D 8m
+IMAGESIZE=3D 10m
MAKEFS_FLAGS+=3D -o density=3D3k
IMAGEENDIAN=3D le
MAKEDEVTARGETS=3D all ipty
diff --git a/sys/arch/vax/conf/GENERIC b/sys/arch/vax/conf/GENERIC
index b9955a0e5f0f..04a97e03fb72 100644
--- a/sys/arch/vax/conf/GENERIC
+++ b/sys/arch/vax/conf/GENERIC
@@ -51,7 +51,7 @@ makeoptions DEBUG=3D"-g"
options DDB
options DDB_HISTORY_SIZE=3D512 # enable history editing in DDB
options KTRACE
-#options DIAGNOSTIC
+options DIAGNOSTIC
#options DEBUG
#options PMAPDEBUG
#options TRAPDEBUG
diff --git a/sys/arch/vax/vax/vm_machdep.c b/sys/arch/vax/vax/vm_machdep.c
index fe138a16c2f4..5c2b9855f675 100644
--- a/sys/arch/vax/vax/vm_machdep.c
+++ b/sys/arch/vax/vax/vm_machdep.c
@@ -175,8 +175,15 @@ void
cpu_lwp_free(struct lwp *l, int proc)
{
=20
- (void)l;
(void)proc;
+
+ /*
+ * Unlink the PCB here, before the u-area holding it is freed. THis=
is
+ * only needed for LWPs than never ran (due to errors while creating
+ * them) as in all other cases, mi_switch() will do the job.
+ */
+ if (l !=3D curlwp)
+ pmap_deactivate(l);
}
=20
void
I've stored it as two patches: One enabling DIAGNOSTIC (and sizing up
the miniroot fs as it wouldn't fit otherwise), the second part is the
actual fix.
MfG, JBG
--=20
--whbJPoPwx1YUiGt1
Content-Type: application/pgp-signature; name="signature.asc"
-----BEGIN PGP SIGNATURE-----
iF0EABECAB0WIQQlDTvPcScNjKREqWEdvV51g5nhuwUCavAoagAKCRAdvV51g5nh
uz/WAJ4pcg6QmKqdEzKqkKu73uCm1DEMvACbBXJAgJ+Y5FQqFbxf6n5sXfRz9Cw=
=bBdj
-----END PGP SIGNATURE-----
--whbJPoPwx1YUiGt1--
From: Jan-Benedict Glaw <jbglaw@lug-owl.de>
To: gnats-bugs@netbsd.org
Cc:
Subject: Re: port-vax/57839
Date: Tue, 10 Nov 2026 10:55:38 +0100
--negmOx2LioZftMgl
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable
Hi!
I believe the patch is correct and fixes the issue. (By now, it's also
tested by me.) Who'd review it?
Thanks,
Jan-Benedict
--=20
--negmOx2LioZftMgl
Content-Type: application/pgp-signature; name="signature.asc"
-----BEGIN PGP SIGNATURE-----
iF0EABECAB0WIQQlDTvPcScNjKREqWEdvV51g5nhuwUCavLqGAAKCRAdvV51g5nh
u1gSAJ4m59DpzTOHcnkYsrTF8uVqUxTpfACghmjG4ObgxoBG4oEB4Ew0JwopUvw=
=q/ER
-----END PGP SIGNATURE-----
--negmOx2LioZftMgl--
From: Kalvis Duckmanton <kalvisd@gmail.com>
To: gnats-bugs@NetBSD.org
Cc:
Subject: Re: port-vax/57839
Date: Tue, 10 Nov 2026 23:22:51 +1100
To recap:
The ATF test /usr/tests/lib/libc/sys/t_lwp_create:lwp_create_bad_lid_ptr
invokes sys__lwp_create() with a bogus lwpid_t *.
In sys__lwp_create(), the creation of the new LWP by do_lwp_create()
succeeds, but the subsequent copyout() of the LWP ID fails due to the
bogus pointer and the newly created LWP, having never run, is then
destroyed by lwp_exit().
do_lwp_create() ends up calling vax/vm_machdep.c:cpu_lwp_fork(), which
in turn calls pmap_activate(). pmap_activate() does a few things -- it
stores values for the P0 and P1 base and length registers in the new
LWP's PCB and links that PCB into the list of PCBs using the pmap.
While pmap_activate() can also load the CPU's P0 and P1 registers, it
doesn't at this time because the new LWP is not the one currently active.
Later, because the LWP never ran, the matching pmap_deactivate() is
never called and the reference in the pmap to the PCB remains, leading
to a panic when the pmap is destroyed.
After investigation, the consensus is that the call to pmap_activate()
in cpu_lwp_fork() is not needed; pmap_activate() (and the matching
pmap_deactivate()) will be called later as the LWP is scheduled.
However, one of the side-effects of pmap_activate() is still required --
the values for the P0 and P1 registers in the new LWP's PCB need to be
initialised, and in particular the initial AST level needs to be set to
AST_PCB (no ASTs pending)
cheers
kalvis
From: "Kalvis Duckmanton" <kalvisd@netbsd.org>
To: gnats-bugs@gnats.NetBSD.org
Cc:
Subject: PR/57839 CVS commit: src/sys/arch/vax/vax
Date: Tue, 10 Nov 2026 12:34:23 +0000
Module Name: src
Committed By: kalvisd
Date: Tue Nov 10 12:34:23 UTC 2026
Modified Files:
src/sys/arch/vax/vax: vm_machdep.c
Log Message:
vax/vm_machdep.c: avoid calling pmap_activate() from cpu_lwp_fork()
In cpu_lwp_fork(), set initial values for the P0 and P1 base and
length registers and AST level rather than calling pmap_activate() to
do so. pmap_activate() will be called later when the LWP runs.
PR port-vax/57839
To generate a diff of this commit:
cvs rdiff -u -r1.119 -r1.120 src/sys/arch/vax/vax/vm_machdep.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->feedback
State-Changed-By: kalvisd@NetBSD.org
State-Changed-When: Tue, 10 Nov 2026 12:42:09 +0000
State-Changed-Why:
Committed fix
From: Jan-Benedict Glaw <jbglaw@lug-owl.de>
To: gnats-bugs@netbsd.org
Cc:
Subject: Re: port-vax/57839
Date: Wed, 11 Nov 2026 19:27:01 +0100
--7WezKbWAZ44urU7c
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable
Build a new release (-current) with your patch included (@kalvisd) and
it works, too:
vax-template# cd /usr/tests/lib/libc/sys && atf-run t_lwp_create
Content-Type: application/X-atf-tps; version=3D"3"
info: atf.version, Automated Testing Framework 0.20 (atf-0.20)
info: tests.root, /usr/tests/lib/libc/sys
info: time.start, Wed Nov 11 18:22:30 UTC 2026
info: uname.sysname, NetBSD
info: uname.nodename, vax-template.classiccmp
info: uname.release, 11.99.9
info: uname.version, NetBSD 11.99.9 (GENERIC) #0: Wed Oct 28 09:49:53 UTC 2=
026 mkrepro@mkrepro.NetBSD.org:/usr/src/sys/arch/vax/compile/GENERIC
info: uname.machine, vax
info: env, ENV=3D/root/.shrc
info: env, BLOCKSIZE=3D1k
info: env, PWD=3D/usr/tests/lib/libc/sys
info: env, MAIL=3D/var/mail/root
info: env, HOME=3D/root
info: env, PATH=3D/sbin:/usr/sbin:/bin:/usr/bin:/usr/pkg/sbin:/usr/pkg/bin:=
/usr/games:/usr/X11R7/bin:/usr/local/sbin:/usr/local/bin
info: env, HOST=3Dvax-template.classiccmp
info: env, SSH_CONNECTION=3D10.234.0.1 42918 10.234.2.12 22
info: env, SSH_TTY=3D/dev/pts/0
info: env, TERM=3Dscreen5
info: env, OLDPWD=3D/root
info: env, USER=3Droot
info: env, SSH_CLIENT=3D10.234.0.1 42918 22
info: env, PKG_PATH=3Dhttp://ftp.netbsd.org/pub/pkgsrc/packages/NetBSD/vax/=
10.0_2022Q4/All
info: env, LOGNAME=3Droot
info: env, SHELL=3D/bin/sh
tps-count: 1
tp-start: 1794421371.512482, t_lwp_create, 6
tc-start: 1794421371.542485, lwp_create_bad_lid_ptr
tc-end: 1794421373.752723, lwp_create_bad_lid_ptr, passed
tc-start: 1794421373.832731, lwp_create_generic_fail_no_uc_cpu
tc-end: 1794421374.732828, lwp_create_generic_fail_no_uc_cpu, passed
tc-start: 1794421374.842840, lwp_create_vax_fail_psl_0
tc-end: 1794421375.802943, lwp_create_vax_fail_psl_0, passed
tc-start: 1794421375.992964, lwp_create_vax_fail_psl_1
tc-end: 1794421377.13073, lwp_create_vax_fail_psl_1, passed
tc-start: 1794421377.83081, lwp_create_vax_fail_psl_cm
tc-end: 1794421378.93189, lwp_create_vax_fail_psl_cm, passed
tc-start: 1794421378.163197, lwp_create_works
tc-end: 1794421378.983285, lwp_create_works, passed
tp-end: 1794421379.63293, t_lwp_create
info: time.end, Wed Nov 11 18:23:03 UTC 2026
So I guess this ticket can now be either closed directly, or prepare
pull-ups to 9 and 10 as this is a real kernel bug that any of the
numerous VAX users could use to crash the system.
--=20
--7WezKbWAZ44urU7c
Content-Type: application/pgp-signature; name="signature.asc"
-----BEGIN PGP SIGNATURE-----
iF0EABECAB0WIQQlDTvPcScNjKREqWEdvV51g5nhuwUCavSzcQAKCRAdvV51g5nh
u8j7AKCER4/YDrIY8JmnUolXi333KplqCwCfZ5y8uWG+7clqxpXkR7pWPbo/aKM=
=toBi
-----END PGP SIGNATURE-----
--7WezKbWAZ44urU7c--
>Unformatted:
(Contact us)
$NetBSD: query-full-pr,v 1.51 2026/08/10 02:28:17 riastradh Exp $
$NetBSD: gnats_config.sh,v 1.10 2026/05/13 22:00:09 riastradh Exp $
Copyright © 1994-2026
The NetBSD Foundation, Inc. ALL RIGHTS RESERVED.