IRQ-Routing for 5.3-BETA

[Matthias Schuendehuette]

Hello,

I have serious problems with the new IRQ-outing code since Aug 11, 16:00 
UTC.

For some reasons I don't understand, the new code uses IRQ10 instead of 
IRQ9 on my system, which breaks my internet connection thru my ISA 
ISDN-Card.

The old code report the following IRQs for PCI-Devices:

\_SB_.PCI0.LNKA irq   9: [  1  3  4  5  6  7 10 11 12 14 15] 
 low,level,sharable 0.1.0
\_SB_.PCI0.LNKB irq  11: [  1  3  4  5  6  7 10 11 12 14 15] 
 low,level,sharable 0.1.1
\_SB_.PCI0.LNKC irq  12: [  1  3  4  5  6  7 10 11 12 14 15] 
 low,level,sharable 0.1.2
\_SB_.PCI0.LNKD irq  11: [  1  3  4  5  6  7 10 11 12 14 15] 
 low,level,sharable 0.1.3

which leeds later on to:

pcib0: matched entry for 0.1.INTA (src \_SB_.PCI0.LNKA)
pcib0: slot 1 INTA is routed to irq 9
pcib1: slot 0 INTA is routed to irq 9
found-> vendor=0x10de, dev=0x0110, revid=0xa1
bus=1, slot=0, func=0
class=03-00-00, hdrtype=0x00, mfdev=0
cmdreg=0x0007, statreg=0x02b0, cachelnsz=0 (dwords)
lattimer=0x20 (960 ns), mingnt=0x05 (1250 ns), maxlat=0x01 (250 ns)
intpin=a, irq=9



The new code instead reports:

pcib0: matched entry for 0.1.INTA (src \_SB_.PCI0.LNKA)
pcib0: possible interrupts:  1  3  4  5  6  7 10 11 12 14 15
ACPI PCI link arbitrated settings:
\_SB_.PCI0.LNKA (references 8, priority 175018):
interrupts:    10     5    11     7     6     4     3    12    15    14\
 1
penalty: 1280  1330  2720  6280  6280  6280  6280  6360 51280\
 51280101280
atpic: Programming IRQ10 as level/low     <Oh, NO!!!>
pcib0: slot 1 INTA routed to irq 10 via \_SB_.PCI0.LNKA
pcib1: slot 0 INTA is routed to irq 10
found-> vendor=0x10de, dev=0x0110, revid=0xa1
bus=1, slot=0, func=0
class=03-00-00, hdrtype=0x00, mfdev=0
cmdreg=0x0007, statreg=0x02b0, cachelnsz=0 (dwords)
lattimer=0x20 (960 ns), mingnt=0x05 (1250 ns), maxlat=0x01 (250 ns)
intpin=a, irq=10


How come???!!! Why gets this PCI-bridge a different IRQ? IRQ10 is not 
available for PCI devices, it is marked as "Legacy ISA" in the BIOS 
setup and this was respected all the time. I tried "PnP-OS YES/NO" in 
the BIOS, but this didn't change anything.

I don't believe that the BIOS is buggy because this board (EPoX 8KTA2) 
had worked with the FreeBSD ACPI code for some time now without any 
problems.

Is there a way to tell the ACPI-Code *not* to use IRQxx via a 
'hint.acpi.xxx'?

Is there anything else I can do to avoid this new behaviour?
-- 
Ciao/BSD - Matthias

Matthias Schuendehuette <msch [at] snafu.de>, Berlin (Germany)
PGP-Key at <pgp.mit.edu> and <wwwkeys.de.pgp.net> ID: 0xDDFB0A5F