Shared IRQ

classic Classic list List threaded Threaded
7 messages Options
Reply | Threaded
Open this post in threaded view
|

Shared IRQ

João Salvatti
Hi,

I would like to know if a different hardware can shared the same IRQ
with another?

Eg:

inteldrm0 at vga1: apic 1 int 16 (irq 11)
ppb1 at pci0 dev 28 function 1 "Intel 82801GB PCIE" rev 0x02: apic 1
int 16 (irq 11)
uhci3 at pci0 dev 29 function 3 "Intel 82801GB USB" rev 0x02: apic 1
int 16 (irq 11)

Thanks in advance.

--
Joco Salvatti
Graduated in Computer Science
Federal University of Para - UFPA - Brazil
E-Mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: Shared IRQ

Henry Sieff
http://www.openbsd.org/faq/faq12.html

12.7.3

2009/5/14 Joco Salvatti <[hidden email]>:

> Hi,
>
> I would like to know if a different hardware can shared the same IRQ
> with another?
>
> Eg:
>
> inteldrm0 at vga1: apic 1 int 16 (irq 11)
> ppb1 at pci0 dev 28 function 1 "Intel 82801GB PCIE" rev 0x02: apic 1
> int 16 (irq 11)
> uhci3 at pci0 dev 29 function 3 "Intel 82801GB USB" rev 0x02: apic 1
> int 16 (irq 11)
>
> Thanks in advance.
>
> --
> Joco Salvatti
> Graduated in Computer Science
> Federal University of Para - UFPA - Brazil
> E-Mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: Shared IRQ

Peter Kay (Syllopsium)
>From: "Henry Sieff" <[hidden email]>
>To: "Joco Salvatti" <[hidden email]>


> http://www.openbsd.org/faq/faq12.html
>
> 12.7.3
>
> 2009/5/14 Joco Salvatti <[hidden email]>:
>> Hi,
>>
>> I would like to know if a different hardware can shared the same IRQ
>> with another?
12.7.3 is accurate, however there is a difference between 'can it' 'should
it' and 'will it'

'should it?' - yes, it should
'can it?' - yes, it can
'will it?' - that's the tricky one. Some devices just don't share interrupts
well. Perhaps it's shit hardware, a shit APIC, crappy BIOS, naff driver -
whatever.

PCI devices can theoretically share interrupts, but that doesn't necessarily
mean they will.

PK

Reply | Threaded
Open this post in threaded view
|

Re: Shared IRQ

Henry Sieff
On Thu, May 14, 2009 at 8:27 AM, Peter Kay - Syllopsium
<[hidden email]> wrote:

>> From: "Henry Sieff" <[hidden email]>
>> To: "Joco Salvatti" <[hidden email]>
>
>
>> http://www.openbsd.org/faq/faq12.html
>>
>> 12.7.3
>>
>> 2009/5/14 Joco Salvatti <[hidden email]>:
>>>
>>> Hi,
>>>
>>> I would like to know if a different hardware can shared the same IRQ
>>> with another?
>
> 12.7.3 is accurate, however there is a difference between 'can it' 'should
> it' and 'will it'
>
> 'should it?' - yes, it should
> 'can it?' - yes, it can
> 'will it?' - that's the tricky one. Some devices just don't share interrupts
> well. Perhaps it's shit hardware, a shit APIC, crappy BIOS, naff driver -
> whatever.
>
> PCI devices can theoretically share interrupts, but that doesn't necessarily
> mean they will.

I have only ever had an issue with off-brand NIC's, personally.

But you are of course correct - PCI devices are supposed to be able to
share IRQ's, but that doesn't mean all manufacturers do interop
testing to make sure that works.

Reply | Threaded
Open this post in threaded view
|

Re: Shared IRQ

Marco Peereboom
This makes no sense at all.

On Thu, May 14, 2009 at 09:07:56AM -0700, Henry Sieff wrote:

> On Thu, May 14, 2009 at 8:27 AM, Peter Kay - Syllopsium
> <[hidden email]> wrote:
> >> From: "Henry Sieff" <[hidden email]>
> >> To: "Joco Salvatti" <[hidden email]>
> >
> >
> >> http://www.openbsd.org/faq/faq12.html
> >>
> >> 12.7.3
> >>
> >> 2009/5/14 Joco Salvatti <[hidden email]>:
> >>>
> >>> Hi,
> >>>
> >>> I would like to know if a different hardware can shared the same IRQ
> >>> with another?
> >
> > 12.7.3 is accurate, however there is a difference between 'can it' 'should
> > it' and 'will it'
> >
> > 'should it?' - yes, it should
> > 'can it?' - yes, it can
> > 'will it?' - that's the tricky one. Some devices just don't share interrupts
> > well. Perhaps it's shit hardware, a shit APIC, crappy BIOS, naff driver -
> > whatever.
> >
> > PCI devices can theoretically share interrupts, but that doesn't necessarily
> > mean they will.
>
> I have only ever had an issue with off-brand NIC's, personally.
>
> But you are of course correct - PCI devices are supposed to be able to
> share IRQ's, but that doesn't mean all manufacturers do interop
> testing to make sure that works.

Reply | Threaded
Open this post in threaded view
|

Re: Shared IRQ

Henry Sieff
[cleaned up formatting, since I accidentally top-posed to begin with]

On Thu, May 14, 2009 at 9:24 AM, Marco Peereboom <[hidden email]> wrote:
> I worte:
>> I have only ever had an issue with off-brand NIC's, personally.

>> But you are of course correct - PCI devices are supposed to be able to
>> share IRQ's, but that doesn't mean all manufacturers do interop
>> testing to make sure that works.

> This makes no sense at all.

?

I have had occasional issues with PCI NIC's inexplicably refusing to
send traffic if they shared an IRQ - this was not an OpenBSD issue,
since in those cases the problem was not corrected by using a
different OS. NIC functioned fine when IRQ was no longer shared. Now,
I had always assumed it was because of a problem with the NIC itself.
Apparently, I am about to find out I was wrong :-).

Reply | Threaded
Open this post in threaded view
|

Re: Shared IRQ

Marco Peereboom
On Thu, May 14, 2009 at 09:35:45AM -0700, Henry Sieff wrote:

> [cleaned up formatting, since I accidentally top-posed to begin with]
>
> On Thu, May 14, 2009 at 9:24 AM, Marco Peereboom <[hidden email]> wrote:
> > I worte:
> >> I have only ever had an issue with off-brand NIC's, personally.
>
> >> But you are of course correct - PCI devices are supposed to be able to
> >> share IRQ's, but that doesn't mean all manufacturers do interop
> >> testing to make sure that works.
>
> > This makes no sense at all.
>
> ?
>
> I have had occasional issues with PCI NIC's inexplicably refusing to
> send traffic if they shared an IRQ - this was not an OpenBSD issue,
> since in those cases the problem was not corrected by using a
> different OS. NIC functioned fine when IRQ was no longer shared. Now,
> I had always assumed it was because of a problem with the NIC itself.
> Apparently, I am about to find out I was wrong :-).

This can not have been recent.  Sure in them olden days on garbage like
netware this was an issue but those days are long gone.