On Fri, 10 Jan 1997 10:38:29 -0800 (PST) 
 Curt Sampson <cjs@portal.ca> wrote:

 > It's not a showstopper. We could change /proc so that we can hand
 > it (as mount options) a kernel file and a core file, and then we
 > can mount another proc filesystem (somewhere other than /proc, of
 > course) that would give us the appropriate information for that
 > kernel and dump. The -N option on the tools would be changed to
 > have them look at a specified filesystem.

...we seem to have this discussion every few months... so, let's try
to squish it now before it gets too much out of hand...

Even doing what Curt suggests is bad...

	- Adds cruft to procfs.
	- Makes ps(1) depend on procfs being in the kernel
	  and /proc being mounted.
	- Requires a new file system to be mounted just to read a
	  kernel crash dump.  (Yuck!)

...if you want ps(1) to work in the absense of procfs, then you
_still_ have to keep the kmem grovelling code there.

"Why bother?"

Jason R. Thorpe                                       thorpej@nas.nasa.gov
NASA Ames Research Center                               Home: 408.866.1912
NAS: M/S 258-6                                          Work: 415.604.0935
Moffett Field, CA 94035                                Pager: 415.428.6939