Subject: kern/25031: panic during build.sh with nfs-mounted sources
To: None <gnats-bugs@gnats.netbsd.org>
From: Hauke Fath <hauke@Espresso.Rhein-Neckar.DE>
List: netbsd-bugs
Date: 04/02/2004 21:56:28
>Number: 25031
>Category: kern
>Synopsis: panic during build.sh with nfs-mounted sources
>Confidential: no
>Severity: critical
>Priority: medium
>Responsible: kern-bug-people
>State: open
>Class: sw-bug
>Submitter-Id: net
>Arrival-Date: Fri Apr 02 19:58:00 UTC 2004
>Closed-Date:
>Last-Modified:
>Originator: Hauke Fath <hauke@Espresso.Rhein-Neckar.DE>
>Release: NetBSD 1.6ZL
>Organization:
Mahlzeit!
>Environment:
System: NetBSD mara 1.6ZL NetBSD 1.6ZL (MARA) #0: Sun Mar 28 20:28:15 CEST 2004
hauke@mara:/usr/src/sys/arch/mac68k/compile/MARA mac68k
Architecture: m68k
Machine: mac68k
>Description:
Early during a build.sh build on a -current mac68k (tried with a
2004-03-18 1.6ZK kernel and a 1.6ZL kernel from today's sources),
I have repeatedly run into an nfs related kernel panic:
uvm_fault(0x167890, 0xfffff000, 0, 0x1) -> 0xe
type 8, code [mmu,,ssw]: 505
trap type 8, code = 0x505, v = 0xffffffd4
kernel program counter = 0xc6a54
kernel: MMU fault trap
pid = 383, lid = 1, pc = 000C6A54, ps = 2000, sfc = 1, dfc = 1
Registers:
0 1 2 3 4 5 6 7
dreg: 0000001F FFFF7004 00010002 0000050C 00000023 043D9668 043FAEA4 0086EC00
areg: FFFFFF58 046BEC88 000030BC 000030BC 0111E800 0111E874 043FAD14 FFFFC444
Kernel stack (043FABF8):
3FABF8: 00114726 043FAC78 00000080 00010002 0000050C 00000023 043D9668 043FAEA4
3FAC18: 0086EC00 000030BC 000030BC 0111E800 0111E874 FFFFF000 00000001 00000001
3FAC38: 00000000 00000000 00000000 00000001 00000000 00000000 00000008 00000000
3FAC58: 00000000 00000000 FFFFFFFF 043FAD14 00003076 00000008 00000505 FFFFFFD4
3FAC78: 0000001F FFFF7004 00010002 0000050C 00000023 043D9668 043FAEA4 0086EC00
3FAC98: FFFFFF58 046BEC88 000030BC 000030BC 0111E800 0111E874 043FAD14 FFFFC444
3FACB8: 00000000 2000000C 6A547008 043FACB8 05050005 00050005 FFFFFFD4 043FACF8
3FACD8: 043FAD04 043FACF0 000C7BC8 046BECF8 00000006 046BEC88 043FAD00 00058774
3FACF8: 043FAD04 00010002 000030BC 0012DC4C 000030BC 00010002 043FAEA4 043FAD2C
3FAD18: 000BDF3E 000030BC 00010002 00000000 00000000 043FADCC 00032ABC 000030BC
3FAD38: 00000002 00000004 043D9668 00000200 00000000 00000000 00000000 0111E800
3FAD58: 043D9668 043FAF44 043FAEA4 FFFFFFFF 043D9668 043D9668 00000200 00000000
3FAD78: 00000000 00000000 043FADD0 043D9668 00000000 043FADA0 000C7BC8 043DD07C
3FAD98: 00000000 043DD00C 043FADB0 0111E800 043FADD0 00000080 043FADE0 000BE0A6
3FADB8: 043FADD0 00000000 FFFFFFFF 00000000 04032000 043FAF0C 000484A4 0111E800
3FADD8: 0111DF80 043FAEA4 00000100 FFFFC4F8 0086EC00 043FAFB4 043D9668 00000200
panic: MMU fault
Stopped in pid 383.1 (mount_nfs) at netbsd:cpu_Debugger+0x6: unlk a6
db> t
cpu_Debugger(2000,8,43fac78,43fac64,114746) + 6
panic(13c0bc,10002,50c,23,43d9668) + f8
trap(8,505,ffffffd4) + 254
vn_lock(30bc,10002) + 82
vget(30bc,2) + 92
nqnfs_clientd(111e800,111df80,43faea4,100,ffffc4f8,86ec00) + 272
sys_nfssvc(86ec00,43faf44,43faf3c) + 4e2
syscall_plain(9b,86ec00,43fafb4,100,ffffc73a) + 82
syscall(9b) + 4c
trap0() + e
db> ps
PID PPID PGRP UID S FLAGS LWPS COMMAND WAIT
5605 4964 413 1000 2 0x4002 1 nbmake
4964 6936 413 1000 2 0x4002 1 sh wait
6936 6604 413 1000 2 0x4002 1 nbmake wait
6604 1963 413 1000 2 0x4002 1 sh wait
1963 436 413 1000 2 0x4002 1 nbmake wait
2248 1260 2248 1000 2 0x4102 1 systat ttyin
1260 1200 1260 1000 2 0x4003 1 tcsh pause
1200 270 270 0 2 0x4000 1 rlogind poll
366 286 366 1000 2 0x4002 1 less nanosle
361 413 413 1000 2 0x4002 1 tee piperd
436 413 413 1000 2 0x4002 1 sh wait
413 286 413 1000 2 0x4002 1 sh wait
367 1 367 0 2 0 1 mount_nfs
>How-To-Repeat:
'mount -r -t nfs -o -i,-U,-C,-q,-3 server:/netbsd/sources /local/src',
then build.sh release (with appropriate options). See the machine
crash, typically while "make clean" or creating the objdir hierarchy.
>Fix:
No idea.
>Release-Note:
>Audit-Trail:
>Unformatted:
>383 1 383 0 2 0 1 mount_nfs
375 0 0 0 2 0x20200 1 nfsio nfsidl
379 0 0 0 2 0x20200 1 nfsio nfsidl
374 0 0 0 2 0x20200 1 nfsio nfsidl
103 0 0 0 2 0x20200 1 nfsio nfsidl
286 301 286 1000 2 0x4003 1 tcsh pause
301 270 270 0 2 0x4000 1 rlogind poll
273 1 273 0 2 0x4002 1 getty ttyin
275 1 275 0 2 0 1 cron nanosle
270 1 270 0 2 0 1 inetd kqread
217 1 217 0 2 0 1 sshd select
213 1 213 0 2 0 1 timed poll
124 1 124 0 2 0 1 mount_mfs mfsidl
106 1 106 0 2 0 1 syslogd
75 1 75 0 2 0 1 dhclient select
6 0 0 0 2 0x20200 1 aiodoned aiodone
5 0 0 0 2 0x20200 1 ioflush syncer
4 0 0 0 2 0x20200 1 pagedaemon pgdaemo
3 0 0 0 2 0x20200 1 scsibus0 sccomp
2 0 0 0 2 0x20200 1 kcont kcont
1 0 1 0 2 0x4000 1 init wait
0 -1 0 0 2 0x20200 1 swapper schedul
db>
The sources were nfs-mounted with '-o -i,-U,-3,-C,-q' at the
time. Frederick Bruckman suspected the -q (nqnfs extension) as the
cause, but I have in the meantime seen panics (different stacktrace)
with vanilla nfs mounts, too; at least one on an i386 1.6ZL machine.