kern/42104: frequent file system-related crash on suspend/resume

[smb%cs.columbia.edu@localhost]

>Number:         42104
>Category:       kern
>Synopsis:       frequent file system-related crash on suspend/resume
>Confidential:   no
>Severity:       critical
>Priority:       high
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Mon Sep 21 01:15:00 +0000 2009
>Originator:     Steven M. Bellovin
>Release:        NetBSD 5.99.17
>Organization:
Department of Computer Science, Columbia University
>Environment:
        
        
System: NetBSD yellowstone.machshav.com 5.99.17 NetBSD 5.99.17 (YELLOWSTONE) 
#0: Tue Sep 15 08:42:05 EDT 2009 
smb%yellowstone.machshav.com@localhost:/usr/BUILD/obj/sys/arch/amd64/compile/YELLOWSTONE
 amd64
Architecture: x86_64
Machine: amd64
>Description:
        My laptop frequently (i.e., at least one time in three) crashes
        after resuming from a suspend activity.  Usually, the problem
        is file-system related, and automatic fsck can't cope.

        I run fvwm, which creates named pipes and Unix-domain sockets
        under /var/tmp (e.g., /var/tmp/FvwmCommand-bigboy.machshav.com:0.0C
        is a named pipe; I see Unix-domain sockets in /var/tmp/orbit-smb).
        Often -- and I don't know why; I'll be send-pring separately on
        this point -- X exits during the resume.  The i-nodes for these
        "files" are listed as unallocated, or with duplicate blocks.  I've
        been unable to get a crash dump; when I try, I get things like
        page faults during the dump.  I suspect, though, that there's data
        queued that isn't emptied because of an abnormal exit.
>How-To-Repeat:
        Suspend/resume a bunch of times...
>Fix:
        Unknown -- and it's bad enough that I've stopped carrying my
        NetBSD laptop on trips.

>Unformatted: