On Thu, 1 Oct 1998, Greg A. Woods wrote:

 * > Another thing is that if we have a limit of 8 bits for the minor
 * > number, there can be only 32 nailed-down devices.
 * 
 * Indeed.

"Indeed", indeed! :-)

Haven't we migrated to larger dev_ts with an asymmetric split between
major and minor devices?

{To be honest, I don't see more than 1024 totally different types of
devices out there just now; splitting 10/22 seems more than adequate
if we're going to sit on an unsigned long dev_t;}

 * > HP-UX includes "ioscan" which will find all attached hardware for which
 * > drivers are in the kernel, whether or not they have device files.
 * 
 * AT&T SysVr3 on the 3b2 had such probe capabilities, and finally with
 * SysVr4 they automated everything quite a bit such that on multiuser
 * boot the probe would run and would create any missing device nodes.

I'm still waiting for on-line device probing which happens on accesses
and not just at boot time.

 * 
 * 

-- 

				--*greywolf;
--
Friends don't let friends use Microsoft.