Re: npf related panic

[Patrick Welche <prlw1%cam.ac.uk@localhost>]

npf seems blameless: see John Baker's message - we have the same stack trace
below pfil_run_addrhooks. (amd64 stack trace has pfil.c 1.30, the RPI probably
doesn't)

Cheers,

Patrick

On Wed, Jan 04, 2017 at 03:20:13PM +0000, Patrick Welche wrote:
> On Wed, Jan 04, 2017 at 02:36:30PM +0000, Patrick Welche wrote:
> > On Wed, Jan 04, 2017 at 02:20:29PM +0000, Christos Zoulas wrote:
> > > In article <20170104141755.GA11149@quartz>,
> > > Patrick Welche  <prlw1%cam.ac.uk@localhost> wrote:
> > > >Just updated the kernel of a box which had kernel+userland 7.99.51/amd64,
> > > >with a kernel built just now, so 7.99.54. On boot (fat fingered):
> > > >
> > > >...
> > > >Starting dhcpcd.
> > > >uvm_fault(0xffffffe811642e170, 0x0, 1) -> e
> > > >fatal page fault in supervisor mode
> > > >trap type 6 code 0 rip ffffffff804966e9 cs 8 rflags 10246 cr2 18 ilevel
> > > >6 rsp fffffe8116594a10
> > > >curlwp 0xfffffe8116c87040 pid 135.1 lowest kstack 0xfffffe81165912c0
> > > >kernel: page fault trap, code=0
> > > >Stopped in pid 135.1 (dhcpd) at netbsd:npf_ifaddrhook+0x55: movq 18($r12),%rsi
> > > >npf_ifaddrhook() at netbsd:npf_ifaddrhook+0x55
> > > >pfil_run_addrhooks() at netbsd:pfil_run_addrhooks+0x3a
> > > >in6_control() at netbsd:in6_control+0x67d
> > > >udp6_ioctl_wrapper() at netbsd:udp6_ioctl_wrapper+0x32
> > > >doifioctl() at netbsd:doifioctl+0xbb3
> > > >soo_ioctl() at netbsd:soo_ioctl+0x2b5
> > > >sys_ioctl() at netbsd:sys_ioctl+0x101
> > > >syscall() at netbsd:syscall+0x1e8
> > > >--- syscall (number 54) ---
> > > >
> > > >(so didn't manage to boot, to bring userland into sync - this is a DIAGNOSTIC,
> > > >DEBUG, LOCKDEBUG kernel)
> > > 
> > > Ryo committed a change to pfil.c, it might help.
> > 
> > # ident /netbsd | grep pfil
> >      $NetBSD: pfil.c,v 1.30 2017/01/04 13:03:41 ryo Exp $
> 
> Just installed
> 
> http://nyftp.netbsd.org/pub/NetBSD-daily/HEAD/201701040710Z/evbarm-earmv6hf/binary/gzimg/rpi.img.gz
> 
> on a rpi B - this is the first boot, no fiddling:
> 
> Starting dhcpcd.
> 
> uvm_fault(0x9bd295f0, 0, 1) -> e
> Fatal kernel mode data abort: 'Translation Fault (S)'
> trapframe: 0x9b69dc78
> FSR=00000005, FAR=0000000c, spsr=600c0013
> r0 =9bf980c8, r1 =8080696b, r2 =00000000, r3 =8080696b
> r4 =8080696b, r5 =00000000, r6 =9bf93908, r7 =8080696b
> r8 =00000000, r9 =80000000, r10=00000000, r11=9b69dcdc
> r12=9b69dce0, ssp=9b69dcc8, slr=801c3c08, pc =801c3c3c
> 
> Stopped in pid 139.1 (dhcpcd) at        netbsd:npf_ifaddrhook+0x4c:     ldr     r
> 1, [r5, #0x00c]
> db> bt
> 0x9b69dcdc: netbsd:npf_ifaddrhook+0xc
> 0x9b69dd04: netbsd:pfil_run_addrhooks+0x40
> 0x9b69dd64: netbsd:in6_control+0xa90
> 0x9b69de6c: netbsd:doifioctl+0x910
> 0x9b69df34: netbsd:sys_ioctl+0x250
> 0x9b69dfac: netbsd:syscall+0x9c
> 
> so not system specific...
> 
> Cheers,
> 
> Patrick