acpi at irq9 gets ints for pci device at irq20
Andrew Gallatin
gallatin at cs.duke.edu
Mon Dec 22 13:02:53 PST 2003
I have a Tyan Trinity GC-SL (S2707). After the recent (mid or late
November) improvements to ioapic interrupt handling, irq 20 seems to
be broadcast to ACPI.
Ie:
% vmstat -i
interrupt total rate
irq4: sio0 173 6
irq6: fdc0 4 0
irq8: rtc 3118 124
irq13: npx0 1 0
stray irq13 1 0
irq14: ata0 1154 46
irq15: ata1 30 1
irq31: fxp0 261 10
irq0: clk 2436 97
Total 7178 287
# load driver which uses irq20
% sudo kldload gm.ko
# generate some interrupts on irq20
% gm_allsize -g --blocking-no-spin >& /dev/null
% vmstat -i
interrupt total rate
irq4: sio0 1531 22
irq6: fdc0 4 0
irq8: rtc 8780 127
irq9: acpi0 6430 93
irq13: npx0 1 0
stray irq13 1 0
irq14: ata0 1300 18
irq15: ata1 30 0
irq20: gm1 6430 93
irq24: gm2 1 0
irq31: fxp0 678 9
irq0: clk 6858 99
Total 32044 464
Note how irq20 has the same number of interrupts as the acpi device
at irq9.
I've tried http://www.freebsd.org/~jhb/patches/acpi_irq.patch, but it
didn't help.
I've left verbose dmesg, mptable, and acpi info at
http://people.freebsd.org/~gallatin/trinity. This boot was with the
acpi_irq patch appied.
FWIW, this motherboard has always caused acpi to squeal like stuck
pig...
Drew
More information about the freebsd-current
mailing list