Andre Hedrick wrote:
>
> On Tue, 7 Mar 2000, Alan Cox wrote:
>
> > > ide is much worst than 'net' but fixable......
> > >
> > > ./drivers/net/Config.in
> > >
> > > - bool ' EISA, VLB, PCI and on board controllers' CONFIG_NET_PCI
> > > + if [ "$CONFIG_PCI" = "y" ]; then
> > > + bool ' EISA, VLB, PCI and on board controllers' CONFIG_NET_PCI
> > > + fi
> > >
> >
> > That is wrong. Many of those controllers are on non PCI machines. Whoever
> > made that diff fucked up
>
> Alan,
>
> Nobady made the diff yet, I suggested something of this nature because of
> catching several PCI device build options without PCI_CONIFG set...
> If someone is building a kernel that has not PCI devices but guess at a
> driver to use because of name, they non-build after several hours would
> make people pissed.
Look at Tim Coleman's latest patch. It appears to address mentioned
issues, and looks good to me for the 4-5 drivers I manually checked
against his patch.
As long as CONFIG_PCI testing is done at the individual driver level,
after that individual driver has been checked to make sure it is 100%
PCI, that seems the correct way to go.
(as I mentioned in private mail to Andre) Note that pci.h needs to be
fixed in 2.3.current so that it correctly and fully handles the
!CONFIG_PCI case. Ideally pci.h should define inline no-op functions
when !CONFIG_PCI, so that drivers need not test CONFIG_PCI at all
themselves.
-- Jeff Garzik | My to-do list is a function Building 1024 | which approaches infinity. MandrakeSoft, Inc. |- 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/
This archive was generated by hypermail 2b29 : Tue Mar 07 2000 - 21:00:23 EST