DragonFly BSD
DragonFly commits List (threaded) for 2005-10
[Date Prev][Date Next]  [Thread Prev][Thread Next]  [Date Index][Thread Index]

Re: cvs commit: src/sys/bus/firewire firewire.c fwohci_pci.c src/sys/bus/isa isa_common.c isahint.c pnp.c src/sys/bus/pci pci.c pci_pcib.c pcivar.h src/sys/bus/pci/i386 pcibus.c src/sys/bus/ppbus ppbconf.c src/sys/dev/acpica5 acpi.c acpi_ec.c ...


From: Matthew Dillon <dillon@xxxxxxxxxxxxxxxxxxxx>
Date: Sat, 29 Oct 2005 22:38:09 -0700 (PDT)

    Hopefully I didn't break anything with this commit.  I'm working myself
    up to some major work on the interrupt routing code.  I am first going
    to attempt to see if I can route interrupts at least on the host-bridge
    (at the moment my test box can't even do that!), after that I am going
    to explore the PCI BIOS a bit (but I expect I'll have to write it off
    like everyone else has).  Finally I will try to fix the bus scan ordering
    and get the ACPI interrupt routing code reintegrated into the system.
    I expect ACPI is the only way we're gonna be able to properly route
    interrupts across bridges (the conclusion that FreeBSD came to), but I'm
    wondering if a hybrid approach will work for PCI devices that are
    directly on the host-pci bridge.

    In anycase, its messy stuff.  The existing code is virtually undocumented
    and contains a huge number of assumptions, special cases, and hacks,
    mainly because I think nobody *ELSE* understood it when it was modified
    in the past.  Makes me want to cry, actually, because it took two days
    of hair pulling just to understand what the current nexus/pci code does
    and how the device architecture winds up probing and attaching it all.

    In anycase, please report any new breakages due to this work.

						-Matt




[Date Prev][Date Next]  [Thread Prev][Thread Next]  [Date Index][Thread Index]