>Number:         23343
>Category:       port-arm
>Synopsis:       shark hangs processes when accessing disk
>Confidential:   no
>Severity:       critical
>Priority:       high
>Responsible:    port-arm-maintainer
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Sun Nov 02 10:41:00 UTC 2003
>Closed-Date:
>Last-Modified:
>Originator:     Martin Husemann
>Release:        NetBSD 1.6ZE
>Organization:
>Environment:
System: NetBSD drowsy.duskware.de 1.6ZE NetBSD 1.6ZE (DROWSY) #4: Sun Nov 2 11:05:13 CET 2003 martin@night-porter.duskware.de:/usr/src/sys/arch/shark/compile/DROWSY
Architecture: arm
Machine: shark

>Description:
Sorry for this very unprecise report, but this machine (despite running 
-current) is supposed to be in production, so I couldn't test much but had
to fall-back to an older kernel ASAP.

I noticed various problems with -current on my shark, not sure how they
are related. They are not exactly reproducable, but happen very often.

 - When copying a new kernel from an NFS mounted file system to / the
   cp sometimes never finishes and the cp is unkillable
 - If the kernel copy succeeds, sometimes the NFS file system can not
   be unmounted (EBUSY)
 - When trying to send-pr this from the affected machine with the -current
   kernel the send-pr (or probably the postfix process handling the mail)
   got stuck (and I think I hear noise from the disk, but I'm not sure)

In both the first and the last case the process can be stopped, but not
killed. If I stop the process, apparently some directory has it's vnode
locked, since all operations on that directory block too (in the first
case, when copying the new kernel to /, it is / that is locked).

The last kernel I have that works flawlessly is:

NetBSD drowsy.duskware.de 1.6ZC NetBSD 1.6ZC (DROWSY) #0: Sat Sep 27 22:51:45 CST 2003  martin@insomnia.duskware.de:/usr/src/sys/arch/shark/compile/DROWSY shark

>How-To-Repeat:
s/a

>Fix:
n/a
>Release-Note:
>Audit-Trail:
>Unformatted: