> From a design and security standpoint, is it a good idea to permit userland
> binaries to panic the kernel?  I can picture a cracker with an old version
> of mountd giving me a denial of service headache.

Unless i'm mistaken (and i very well might be), a 'random' user
couldn't use an old mountd to crash the kernel.  mountd is command
used by root, and only by root.

If my assertion above is true (and again, it might not be; i've not
tried it), then couldn't such a cracker cause more disruption by using
the 'halt' command?

i mean, the system reboots after a panic...



chris