kern/40811: crash under disk load

[kefren%ngnetworks.ro@localhost]

>Number:         40811
>Category:       kern
>Synopsis:       crash under disk load
>Confidential:   no
>Severity:       critical
>Priority:       medium
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Sat Feb 28 19:30:00 +0000 2009
>Originator:     Mihai Chelaru
>Release:        NetBSD 5.99.7
>Organization:
        
>Environment:
        
        
System: NetBSD kefren.netbsd.ro 5.99.7 NetBSD 5.99.7 (Home) #16: Sat Feb 28 
19:42:18 EET 2009 
root%kefren.netbsd.ro@localhost:/disk2/work/src/sys/arch/amd64/compile/obj/Home 
amd64
Architecture: x86_64
Machine: amd64
>Description:
        
        Under disk load system become sensitively slow (even mouse is moving in 
steps) and finally crashes after two-three seconds.
        Traceback looks always the same, the following one being generated by 
an install in pkgsrc with PKG_DEVELOPER set. While "Generating pre-install file 
lists" I saw a message that was saying there are too many temporary files. 
After that panic:

uvm_fault(0xffffffff80900580,0xffff80004fc87000,2) -> e
fatal page fault in supervisor mode
trap type 6 code 2 rip ffffffff805c3393 cs 8 rflags 10206 cr2 ffff80004fc87810 
cpl 0 rsp ffff80004fba27d8
kernel: page fault trap, code = 0
Stopped in pid 127.1 (syslogd) at     netbsd: memset+67:     repe stosq
%
es:(%rdi)
18446603337508151296

Traceback:

memset+67
ffs_truncate+3062
ffs_write+2444
VOP_WRITE+102
vn_write+206
dofilewrite+129
sys_write+114
syscall+182

>How-To-Repeat:
        
>Fix:
        

>Unformatted: