Wireless issues...
John
dingo at coco2.arach.net.au
Sun Sep 19 12:13:59 UTC 2004
Matthew Garrett wrote:
>It seems to be a problem with newer kernels. The parallel port drivers
>seem to grab IRQ 7 and refuse to let anything else use it. If the
>parallel port driver is loaded after everything else, it will realise
>that it can't have the interrupt and use polling instead.
>
>
>
The parallel port is an ISA port, and IRQ is dedicated to that port: ISA
interrupt lines are edge-sensitive (they trip on change of voltage) and
so are not sharable.
In most BIOSes _can_ configure that IRQ as reserved, or you can
configure the printer port to another standard location (and maybe only
move the problem).
Once the printer driver is loaded and grabbed its IRQ, nothing else
should touch it.
Pretty much all the IRQs below 8 have standard assignments; 2==9 because
it's used to cascade a second programmable interrupt controller (PIC).
I don't recall now what all the standard assignments are, but devices
using them include
Timer
NMI
Two serial ports
Two printer ports
That doesn't leave many out of the eight in the original PC.
IRQ 12 is hardwired to the PS/2 mouse. Available if you don't use one.
IRQ 14 used by IDE0.
IRQ 15 was reserved then used by IDE1.
Soundblaster cards, NICs, SCSI cards & such typically pinched a printer
IRQ, or maybe a serial port IRQ, and that lead to configuration
difficulties.
Modern PCs don't have ISA slots, but they stll have a few ISA devices on
the motherboard.
However, it seems the end of the problem is nigh: here are the interrupt
assignments from my Athlon:
summer at Glider:~$ cat /proc/interrupts
CPU0
0: 35483261 IO-APIC-edge timer
1: 10 IO-APIC-edge i8042
7: 2 IO-APIC-edge parport0
8: 4 IO-APIC-edge rtc
9: 0 IO-APIC-level acpi
14: 93651 IO-APIC-edge ide0
169: 843583 IO-APIC-level ide2, ide3
177: 0 IO-APIC-level SiS 7012
185: 45 IO-APIC-level ohci_hcd
193: 183 IO-APIC-level ohci_hcd
201: 833638 IO-APIC-level ehci_hcd
209: 9668866 IO-APIC-level eth0
NMI: 0
LOC: 35484318
ERR: 0
MIS: 0
summer at Glider:~$
"level" is good. The signal remains active while any device asserts it,
so all the devices using it can be handled.
More information about the ubuntu-users
mailing list