On Tue, 11 Sep 2007, Atsushi Nemoto wrote: > > Thanks a lot -- it makes things work for my SWARM as expected. But > > doesn't it break systems where there actually is a PCI-(E)ISA bridge and a > > legacy IDE interface? > > Yes, it would breaks such systems, but that is exactly we have been > doing for years, isn't it? Not quite so. The test for the PCI-(E)ISA bridge is there so that they are handled. Now I gather the use of no_pci_devices() in ide_probe_legacy() effectively disables the test entirely (thus making it a candidate for removal). Or am I missing something? > And if such a platform was really exists, mach-specific ide.h could be > used as final workaround. I think Malta qualifies -- it has an onboard PCI IDE interface (coupled in a single chip with a PCI-ISA bridge) which uses legacy I/O ports. No ISA slots though, if you are looking for a diehard example. ;-) Maciej
