Re: Integration of pcnet_cs into drivers/pcmcia

David Hinds (dhinds@zen.stanford.edu)
Fri, 10 Sep 1999 15:30:28 -0700


On Thu, Sep 09, 1999 at 09:44:49PM +0200, Pavel Machek wrote:
>
> pcmcia is initialized _after_ initcall()'s are done, so I suspect it
> would not work. There's probably good reason: pcmcia_init() needs to
> know what areas are allocated.

Normally it should not, actually. pcmcia_init() doesn't check or
allocate any resources until the cardmgr daemon is started. I also
try not to rely on other drivers correctly allocating device resources
anyway, since the Linux resource maps are generally incomplete.

> Integrating it was rather easy... I do not think much thinking is
> required. And - bulkmem is already there, is it? ;-). I can not see
> anything wrong with one_driver_at_a_time approach.

bulkmem is not a driver, it is a middle-layer for accessing memory
cards.

The wrong thing about one_driver_at_a_time is that it will be quite
confusing for people who want to use PCMCIA, if one (or a few) drivers
are found in the kernel tree, and the rest are elsewhere.

And there are still decisions to be made. Like, drivers/net/pcmcia
seems sensible, but do we also want to mirror that in the module tree
with /lib/modules/.../net/pcmcia? Or do they go in modules/.../net,
or modules/.../pcmcia? This impacts the user-mode PCMCIA stuff, so
I'd want to be involved in that decision, rather than having people
dropping PCMCIA drivers into the kernel tree as they see fit.

-- Dave Hinds

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/