[Xenomai-help] Shared interrupt :(

[Xenomai-help] Shared interrupt :(

[bite]

[-- Attachment #1: Type: text/plain, Size: 1081 bytes --]

Hi, this is my first post, I'm using Xenomai since a few days, I beg your
pardon if my question has already been answered.

My problem is not directly related to Xenomai, but it prevents me from using
Xenomai (as well as any other kind of realtime extension, I think).

The point is: I have a PCI axis control board and with Xenomai I can use it
well on some kind of hardware.

On a different machine, the irq of this board happens to be shared with an
usb irq, and I get lots of problems there. Erratic behaviour, pc hanging and
killed cats.

Things do not change if I include I_PROPAGATE in the interrupt mode flags.

I recompiled the kernel with IO_APIC and LOCAL_APIC set, but what happens
then is just that the conflicting interrupts both move together, hand in
hand, from the same irq 9 to the same irq 18 :) Shame!

Is there any known solution for this embarrassing problem? Somebody
suggested me to put the board in a different PCI slot, but unluckily I
cannot do so because of a geometric constraint.

Thank you in advance for any help or suggestion & best regards,

Bite

[-- Attachment #2: Type: text/html, Size: 1151 bytes --]

Re: [Xenomai-help] Shared interrupt :(

[Jan Kiszka]

[-- Attachment #1: Type: text/plain, Size: 1752 bytes --]

bite wrote:
> Hi, this is my first post, I'm using Xenomai since a few days, I beg your
> pardon if my question has already been answered.
> 
> My problem is not directly related to Xenomai, but it prevents me from using
> Xenomai (as well as any other kind of realtime extension, I think).
> 
> The point is: I have a PCI axis control board and with Xenomai I can use it
> well on some kind of hardware.
> 
> On a different machine, the irq of this board happens to be shared with an
> usb irq, and I get lots of problems there. Erratic behaviour, pc hanging and
> killed cats.
> 
> Things do not change if I include I_PROPAGATE in the interrupt mode flags.
> 
> I recompiled the kernel with IO_APIC and LOCAL_APIC set, but what happens
> then is just that the conflicting interrupts both move together, hand in
> hand, from the same irq 9 to the same irq 18 :) Shame!
> 
> Is there any known solution for this embarrassing problem? Somebody
> suggested me to put the board in a different PCI slot, but unluckily I
> cannot do so because of a geometric constraint.

This is really a FAQ in this list, and if I weren't too busy, I would
now take the chance and hack up a wiki page on this. Please browse the
mailing list archives for the shared irq topic. And if you (anyone else
is invited as well) would like to do us a favour, put what you find into
a new wiki FAQ page on xenomai.org so that we can simply cite it on
future postings like this. :)

Jan

PS: If you stumble over the RT-IRQ-stub-for-Linux-devices approach in
the archive: I have some idea to ease the development, but it is stalled
due the limited time to work it out. If you feel like you need such a
stub, I can tell you more, just ask again.


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 249 bytes --]