>Number:         25886
>Category:       kern
>Synopsis:       panic: pipe_write during ls(1).
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Wed Jun 09 18:55:00 UTC 2004
>Closed-Date:
>Last-Modified:
>Originator:     Jonathan Perkin
>Release:        2.0F
>Organization:
BBC Internet Services
>Environment:
NetBSD octoblob.perkin.org.uk 2.0F NetBSD 2.0F (GENERIC_LAPTOP) #0: Wed Jun  9 17:19:12 BST 2004  sketch@tucarambo.perkin.org.uk:/store/netbsd/obj/sys/arch/i386/compile/GENERIC_LAPTOP i386
>Description:
Local machine is laptop running the above kernel.  /store is
mounted NFS from NetBSD/sparc64 machine running -current as of
yesterday.  /home is local ffsv1 partition

While copying a freshly built cross-release from /store to 
/home/install, an ls(1) in /home (to see how much of the file
had been transferred) caused:

  uvm_fault(0xc5b37ee0, 0, 0, 1) -> 0xe
  kernel: page fault trap, code=0
  Stopped in pid 3943.1 (ls) at netbsd:pipe_write+0xf: movl 0x84(%eax),%ebx
  db>

I had previously done a few ls -l in the directory which worked
fine, but first noticed a problem when using ksh filename
completion attempting to "cd /home/ins<tab>" which hung.  I then
did an explicit "ls -l comp.tgz" which is the one above.

Both processes hung for a while (maybe a couple of minutes) while
the cp was still running and the sets were still coming across the
wire, before the kernel panicked (while I was doing some basic tests
to see what the processes were up to).
>How-To-Repeat:
I'll see if I can reproduce the panic again, this was the first time
I've attempted the above, but will be using it a lot in the future to
move my -current updates onto the bigger sparc64 box before installing
onto the laptop.
>Fix:

>Release-Note:
>Audit-Trail:
>Unformatted: