>>>>> "Greg" == Greg Lehey <[EMAIL PROTECTED]> writes:
Greg> On Saturday, 27 November 1999 at 10:26:15 -0500, Viren R.Shah wrote:
>>
>> I'm running a -current system from Nov 26th (approx 4am EST).
>>
>> I can currently reliably crash the system by doing:
>>
>> ln -s /home/users/vshah/public_html/index.html /home/users/vshah/index.html
>>
>>
>> The crash only works when I do it on a NFS mounted filesystem. I'm
>> using NFSv2/UDP. The server is a 3.2-STABLE FreeBSD box, running
>> softupdates on the exported filesystem. I just checked that local
>> filesystem on the server, and it is a 100% full. Can this just be put
>> down to the known "softupdates full filesystem bug"?
Greg> Not based on the (non-existent) evidence you've supplied. Where does
Greg> it crash?
not having DDB compiled into the kernel I can't answer that. However
I'm willing to give any suggestions a go.
>> [BTW: the server hasn't crashed, it's only the FreeBSD client that
>> crashes]
Greg> Do you mean the client process or the client operating system?
The client OS -- immediately after creating the symlink.
This is the panic I got (taken before Eivind's nfs_vnops.c commit
(rev 1.146) :
fatal trap 12: page fault while in kernel mode
fault virtual address = 0x4
fault code = supervisor read, page not present
instruction pointer = 0x8:0xc0163a05
stack pointer = 0x10:0xc9d77e40
code segment = base 0x0; limit 0xfffff, type 0x1b
= DPL0, pres 1, def32.1, gran 1
processor flags = interrupt enabled, resume, 10PL = 0
current process = 1230 (ln)
interrupt mask = none
trap number = 12
panic: page fault
I'm not sure that it helps any...
Greg> Greg
Viren
--
Viren Shah | "You can't trust code that you did not totally
Research Associate, RST Inc. | create yourself. (Especially code from
[EMAIL PROTECTED] | companies that employ people like me.)"
http://www.rstcorp.com/~vshah | - Ken Thompson "Reflections on Trusting Trust"
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
