On Tue, Aug 01, 2000 at 02:07:41AM +0000, Andrew Morton wrote:
> I don't see that in the 2.4.0-test4 tree. Any idea who will be doing
> this, and when?
It is already done and works fine.
> AFAICS, the pcmcia-cs `make install' target will look to see if you've
> built your kernel with CONFIG_PCMCIA. If you have, the pcmcia_cs
> clients are not built. So one approach would be to massage the pcmcia
> config files at install time to refer to either 3c575_cb or to 3c59x.
> This doesn't feel right.
Agreed, since then the config files would be kernel-version-dependent,
which is nasty for people who flip between kernels.
> Another option is to use a new identifier altogether within the pcmcia
> package (3com_nic) and then rely upon modprobe resolving it
> appropriately, either via an if/then test or
> probe 3com_nic 3c575_cb 3c59x
> in modules.conf.
Something like this seems the best option; either that or (as I think
there is some intent) have a separate mechanism for handling loading
of hot plug PCI drivers that isn't tied with PCMCIA at all. That
would sidestep the compatibility issue.
Is it legal to have something like this:
probe 3c575_cb 3c575_cb 3c59x
or would modprobe try to recursively expand the 3c575_cb?
> Third option: globally replace 3c575_cb with 3c59x in the pcmcia_cs
This is unsatisfactory, since on a pre-2.4 system, there may already
be 3c59x modules hanging around that do not support CardBus, and
modprobe is not going to be able to select the right one to use.
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to firstname.lastname@example.org
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Mon Aug 07 2000 - 21:00:06 EST