At least in Alpha configuring the root bus wasn't required. This could be different in x86 since alpha just had a fixed mapping in memory space, and x86 might not (but since there is so much space in 64 bit linux it seems like it would). I can't look at the minute, but I'll poke around later today and see if I un-earth anything that might be helpful.
ali On Dec 16, 2008, at 2:55 AM, Gabe Black wrote: > Hi everybody. I'm currently trying to twist Linux's arm into > recognizing and configuring the PCI IDE controller, and the thing I'm > stuck on right now is I can't figure out how the IO resources of the > root bus are assigned. I found a function for child busses which is > bases off of the IO base and IO limit registers in the bridge, > something > I hoped would be true of the root bus as well. It looks like something > somewhere is supposed to extend the kernel's tree of "resource" > objects > to allocate the space the PCI bus responds to, but either that's never > happening or for some reason the kernel is losing track of it. One > thing > which may have something to do with it is that the kernel is trying to > configure the host bridges config registers as a device rather than a > bus. It might always do that, but I really don't know. There are a > number of tables that end up in memory that may have something to do > with it, but I've poked at one of those, the Intel MP table, with no > success. There's still the DMI table and the ACPI tables, but I'd > hesitate to assume that's the problem. Any help would be appreciated > since grepping for "resource" isn't getting me too far. > > Gabe > _______________________________________________ > m5-dev mailing list > [email protected] > http://m5sim.org/mailman/listinfo/m5-dev > _______________________________________________ m5-dev mailing list [email protected] http://m5sim.org/mailman/listinfo/m5-dev
