pcic-patch.patch? / interrupt routing vs. ACPI

Was the most recent patch, pcic-patch.patch, (mentioned by Andrew, provided by 
Tony) posted to -mobile? I can't find it in the list archives.

I have a Sony SRX87 with a built-in Orinoco WaveLAN behind a TI 12XX bridge. 
Under 4.6-RELEASE, my machine won't boot properly with hw.pcic.intr_path=2; it 
complains of not being able to route INTA for the TI bridge, and then the 
console hangs after mounting the root fs.

With hw.pcic.intr_path=1 and hw.pcic.irq=0, it identifies the card and 
allocates it an IRQ (have tried all possible IRQs on the laptop), but no 
interrupts are ever successfully delivered: I get endless wi0 watchdog 
timeouts.

I've tried Warner's version of Tony's patch, but printf()s reveal the new code 
for changing the MUX register isn't even invoked when I boot with PCI 
interrupt routing for the pcic; it would only be called after the PCI IRQ is 
successfully assigned, and the box complains of PCI IRQ routing failing, and 
switches to ISA instead.

Is there any chance this pcic-patch.patch might make things work any better 
for me? If so, could someone forward it to me, please? Thanks.

I'm fairly desperate to make the pcics on this Sony SRX87 work. I can't get 
the Ricoh PCIC (for the machine's PC-Card slot) to work properly, either. I've 
tried -current with ACPI with no better luck. My deadline for returning the 
laptop to Fry's Electronics is Sunday. :-(

A naive question: is there laptop hardware for which it is truly *impossible* 
to make either PCI or ISA interrupt routing work without a correct and 
complete ACPI implementation? Or is it worth it for me to look into dirty, 
machine-specific hacks to get these bridges configured?

Is it possible that there's garbage in my $PIR table, and that I could hack 
something up that supplies fictitious hand-tweaked entries to the FreeBSD PCI 
IRQ routing code, to convince it to route PCI IRQs for my pcics?

Thanks,
-Brad, [email protected]



To Unsubscribe: send mail to [email protected]
with "unsubscribe freebsd-mobile" in the body of the message