[pxc] smp 2.4.x oops

[M Bianchi]

For what it is worth, I had a problem when I put 2 PCI PXC200 cards and
a US Robotics PCI Modem on the same uniprocessor system.  The modem and
one PXC200 card wound up sharing an interrupt.

I tried to understand the  setpci(8)  command, but was unsuccessful in
moving interrupts around.

The simpler thing to do (which I should have tried first) was to just
reorder the cards in the PCI bus slots, and then the modem got one
interrupt (9, I think) and the 2 PXC200s both got another which the
shared (10, I think).  In that way both PXC200s work just fine.


While I'm here, I'll suggest a change to  pxc_live .
The line
	set imgfile /tmp/bla.pgm

sets the name of an intermediate file to a constant value.  That means
that if you run two instances of pxc_live on different frame grabbers,
the images get intermixed.  Changing the line to
	set imgfile /tmp/pxc_live.[pid].pgm

identifies the file as being from pxc_live,
associated with a particular process id,
and containing pgm image data.

It also keeps the images separate.


On Sun, Sep 09, 2001 at 11:14:10AM +1000, Chanop Silpa-Anan wrote:
> recent attemp to get pxc200-F on a smp-box piii using via apollo
> chipset up and running with smp kernel 2.4.10-pre4.
>
> I've got the following opps from serial console duing "insmod ./pxc.o"
>
>
> found PXC200F: mem=0xdf101000, irq=10
> NMI Watchdog detected LOCKUP on CPU1, registers:
> CPU:    1
> EIP:    0010:[<e5b1401a>]
> EFLAGS: 00000086
> eax: 00000000   ebx: 00000000   ecx: 0000002e   edx: c9156000
> esi: cbacc3a0   edi: 00000246   ebp: c9157e44   esp: c9157e38
> ds: 0018   es: 0018   ss: 0018
> Process insmod (pid: 5163, stackpage=c9157000)
> Stack: cbacc3a0 cbacc3a0 00000000 c9157e74 e5b110d8 cbacc3a0 00000000 00000282
>        00000001 cbacc3a0 e5b148b7 00000000 e5b148b7 00000000 c04ffa69 c9157eb4
>        e5b114f5 cbacc3a0 00000026 c9157eb4 e5b114ec e5b148cd e5b148b7 df101000
> Call Trace: [<e5b110d8>] [<e5b148b7>] [<e5b148b7>] [<e5b114f5>] [<e5b114ec>]
>    [<e5b148cd>] [<e5b148b7>] [<e5b13bf9>] [<e5b13be4>] [<e5b13aba>] [<e5b10064>
>    [<e5b10064>] [<e5b15dc0>] [<c0117a1b>] [<e5b10060>] [<e5b10060>] [<c0106feb>
>
> Code: 7e f8 e9 e8 cf ff ff 80 7b 14 00 f3 90 7e f8 e9 19 d5 ff ff
> console shuts up ...
>
>
> I still have (presumably) interrupt sharing problem when using up-kernel
> 2.4.x with this box which has to be investigated furthur. The symtomp is
> that reading from the device has never return.
>
>
> I cannot downgrade to 2.2.x for this box unless I reinstall every
> partition with other file system than xfs.
>
> Has anyone tried smp / 2.4.x. 0.3.0-beta has some supports.
> Unfortunately, I don't have other box handy at the moment.
>
>
>
> cheers,
>
> Chanop
> _______________________________________________
> Pxc mailing list
> Pxc@lists.linux.it
> http://www.linux.it/mailman/listinfo/pxc

--
 Mike Bianchi
 Foveal Systems
 190 Loantaka Way
 Madison NJ  07940-1910

 +1 973 822-2085	Voice and Fax

 MBianchi@Foveal.com

 Create videos of auditorium presentations without a crew?
 Yes!
 http://www.AutoAuditorium.com