Subject: Re: PCMCIA & ATAPI
To: None <firstname.lastname@example.org>
From: John Kohl <email@example.com>
Date: 03/04/1997 10:55:32
>>>>> "LA" == Lennart Augustsson <firstname.lastname@example.org> writes:
>> The PCMCIA code, although functional, is considered "bad".
LA> Maybe the people who consider it bad could explain what's bad about it?
LA> I've not really looked at it myself yet.
[These comments are all concerning the PCMCIA code I've maintained,
based on Stefan Grefen's initial work.]
Its configuration model is not proper (needs some work, especially to
handle things like: multiple-unit cards such as combo ethernet/modem
cardssome; a way to leave preconfigured cards alone and treating them as
ISA devices; dealing with card insertion/removal; etc.). The code
itself is in need of some serious cleanup, slimming down, and comparison
with the specification--there are places where it can loop due to
bogus/faulty PC card configuration details, etc.
There also is a need for some additional work in the generic
autoconfiguration code to handle dynamic detachment of devices and in
other parts of the kernel (e.g. networking) to deal with removed devices.
Basically, it needs a rewrite from scratch.
Note that there is another set of pcmcia changes floating around, done
in part (in whole?) by Rob Baron of CMU; it doesn't have quite the same
problems but last I heard it likewise isn't ready for prime time.