On Tue, 6 Jun 2000, Brett Lymn wrote:

# According to David Brownlee:
# >
# >	That would be a definite step up on what we have now.
# >	How much extra work would it be to add the locator information
# >	to the in-kernel decive list, and how would that interact
# >	with the proposal to have a boot time kernel config interface?
# >
# 
# ``A LOT''  AFAIK adding the locator information will touch every
# single driver in the tree, they all will have to stash the locator
# information _somewhere_ and be told to do it.  I am not sure if all
# the boot time kernel config interface is just feeding the appropriate
# bits into the driver probe routine or it does something more
# sophisticated.  If the kernel config interface forces an update of all
# the drivers then adding locator info would be good.
# 
# With some slight changes (which are almost done) prtconf could be used
# to document the running kernel's device probe parameters.

The locator information sounds like a good candidate for pageable
kernel information.

Speaking of which:  Has our kernel entered the new age and has some parts
of it which can actually be paged out, or are we still entirely resident?
[And NO, process paging doesn't count for what I HOPE are obvious reasons!]

# 
# 


				--*greywolf;
--
BSD: No Sh;t!