[gentoo-user] Preventing a shared interrupt

[gentoo-user] Preventing a shared interrupt

[Grant]

Does anyone know how to prevent my "nouveau" video card and "ath" wifi
card from sharing interrupt 19?  I can't move the video card slot and
I'd rather not move the wifi card slot if possible.

BTW, does anyone see any other opportunity for interrupt optimization
here?  I'm trying to optimize for video and especially audio, and I'm
using a USB 1.0 sound card.

# cat /proc/interrupts
           CPU0
  0:     473139   IO-APIC-edge      timer
  1:          2   IO-APIC-edge      i8042
  7:          1   IO-APIC-edge
  8:        127   IO-APIC-edge      rtc0
  9:          0   IO-APIC-fasteoi   acpi
 12:          3   IO-APIC-edge      i8042
 14:          0   IO-APIC-edge      ide0
 19:     135597   IO-APIC-fasteoi   nouveau, ath
 20:          2   IO-APIC-fasteoi   ehci_hcd:usb2
 21:      13860   IO-APIC-fasteoi   ehci_hcd:usb1
 22:     971970   IO-APIC-fasteoi   ohci_hcd:usb4
 23:       3446   IO-APIC-fasteoi   ohci_hcd:usb3
 28:      21524   PCI-MSI-edge      ahci
NMI:          0   Non-maskable interrupts
LOC:    2234210   Local timer interrupts
SPU:          0   Spurious interrupts
PMI:          0   Performance monitoring interrupts
PND:          0   Performance pending work
THR:          0   Threshold APIC interrupts
MCE:          0   Machine check exceptions
MCP:          6   Machine check polls
ERR:          1
MIS:          0

- Grant

Re: [gentoo-user] Preventing a shared interrupt

[Volker Armin Hemmann]

if your mobo does not supporting re-assigning interrupts: you can't.

Re: [gentoo-user] Preventing a shared interrupt

[Grant]

> if your mobo does not supporting re-assigning interrupts: you can't.

Would that be done in the BIOS?  There is a list of interrupts in the
BIOS and I can select either "Unassigned" or "Reserved" I think.  What
affect will reserving an interrupt have?

- Grant

Re: [gentoo-user] Preventing a shared interrupt

[Volker Armin Hemmann]

On Freitag 21 Mai 2010, Grant wrote:
> > if your mobo does not supporting re-assigning interrupts: you can't.
> 
> Would that be done in the BIOS?  There is a list of interrupts in the
> BIOS and I can select either "Unassigned" or "Reserved" I think.  What
> affect will reserving an interrupt have?
> 
> - Grant

I really don't know. Ask the support of your vendor ;)

Re: [gentoo-user] Preventing a shared interrupt

[Grant]

>> > if your mobo does not supporting re-assigning interrupts: you can't.
>>
>> Would that be done in the BIOS?  There is a list of interrupts in the
>> BIOS and I can select either "Unassigned" or "Reserved" I think.  What
>> affect will reserving an interrupt have?
>>
>> - Grant
>
> I really don't know. Ask the support of your vendor ;)

Fair enough.  Can you tell me why a motherboard would assign 2 devices
to the same IRQ when there are obviously many free ones?  Is it just
bad implementation?

- Grant

Re: [gentoo-user] Preventing a shared interrupt

[Volker Armin Hemmann]

On Freitag 21 Mai 2010, Grant wrote:
> >> > if your mobo does not supporting re-assigning interrupts: you can't.
> >> 
> >> Would that be done in the BIOS?  There is a list of interrupts in the
> >> BIOS and I can select either "Unassigned" or "Reserved" I think.  What
> >> affect will reserving an interrupt have?
> >> 
> >> - Grant
> > 
> > I really don't know. Ask the support of your vendor ;)
> 
> Fair enough.  Can you tell me why a motherboard would assign 2 devices
> to the same IRQ when there are obviously many free ones?  Is it just
> bad implementation?
> 
> - Grant

because interrupt sharing is not a problem anymore? 

Re: [gentoo-user] Preventing a shared interrupt

[Paul Hartman]

On Fri, May 21, 2010 at 3:54 PM, Grant <emailgrant@gmail.com> wrote:
>>> > if your mobo does not supporting re-assigning interrupts: you can't.
>>>
>>> Would that be done in the BIOS?  There is a list of interrupts in the
>>> BIOS and I can select either "Unassigned" or "Reserved" I think.  What
>>> affect will reserving an interrupt have?
>>>
>>> - Grant
>>
>> I really don't know. Ask the support of your vendor ;)
>
> Fair enough.  Can you tell me why a motherboard would assign 2 devices
> to the same IRQ when there are obviously many free ones?  Is it just
> bad implementation?

Looks like you have MSI support, so probably your hardware isn't
ancient enough for interrupt sharing to be a problem in the first
place.

Re: [gentoo-user] Preventing a shared interrupt

[Grant]

>>>> > if your mobo does not supporting re-assigning interrupts: you can't.
>>>>
>>>> Would that be done in the BIOS?  There is a list of interrupts in the
>>>> BIOS and I can select either "Unassigned" or "Reserved" I think.  What
>>>> affect will reserving an interrupt have?
>>>>
>>>> - Grant
>>>
>>> I really don't know. Ask the support of your vendor ;)
>>
>> Fair enough.  Can you tell me why a motherboard would assign 2 devices
>> to the same IRQ when there are obviously many free ones?  Is it just
>> bad implementation?
>
> Looks like you have MSI support, so probably your hardware isn't
> ancient enough for interrupt sharing to be a problem in the first
> place.

OK.  Do higher numbered IRQ's have priority over lower numbered ones?
I'm starting a new project where I'll build a super-minimal Gentoo
system and I'm trying to learn as much as I can about optimizing for
performance.

- Grant

Re: [gentoo-user] Preventing a shared interrupt

[Adam]

> Does anyone know how to prevent my "nouveau" video card and "ath" wifi
> card from sharing interrupt 19?  I can't move the video card slot and
> I'd rather not move the wifi card slot if possible.
> 
> BTW, does anyone see any other opportunity for interrupt optimization
> here?  I'm trying to optimize for video and especially audio, and I'm
> using a USB 1.0 sound card.
> 
> # cat /proc/interrupts
>            CPU0
>   0:     473139   IO-APIC-edge      timer
>   1:          2   IO-APIC-edge      i8042
>   7:          1   IO-APIC-edge
>   8:        127   IO-APIC-edge      rtc0
>   9:          0   IO-APIC-fasteoi   acpi
>  12:          3   IO-APIC-edge      i8042
>  14:          0   IO-APIC-edge      ide0
>  19:     135597   IO-APIC-fasteoi   nouveau, ath
>  20:          2   IO-APIC-fasteoi   ehci_hcd:usb2
>  21:      13860   IO-APIC-fasteoi   ehci_hcd:usb1
>  22:     971970   IO-APIC-fasteoi   ohci_hcd:usb4
>  23:       3446   IO-APIC-fasteoi   ohci_hcd:usb3
>  28:      21524   PCI-MSI-edge      ahci

FWIW a grep of the nouveau source doesnt bring up any hits on MSI, so it
looks like the driver doesnt support it yet. Likewise no hits for ath5k,
however ath9k does support it.

Re: [gentoo-user] Preventing a shared interrupt

[Grant]

>> Does anyone know how to prevent my "nouveau" video card and "ath" wifi
>> card from sharing interrupt 19?  I can't move the video card slot and
>> I'd rather not move the wifi card slot if possible.
>>
>> BTW, does anyone see any other opportunity for interrupt optimization
>> here?  I'm trying to optimize for video and especially audio, and I'm
>> using a USB 1.0 sound card.
>>
>> # cat /proc/interrupts
>>            CPU0
>>   0:     473139   IO-APIC-edge      timer
>>   1:          2   IO-APIC-edge      i8042
>>   7:          1   IO-APIC-edge
>>   8:        127   IO-APIC-edge      rtc0
>>   9:          0   IO-APIC-fasteoi   acpi
>>  12:          3   IO-APIC-edge      i8042
>>  14:          0   IO-APIC-edge      ide0
>>  19:     135597   IO-APIC-fasteoi   nouveau, ath
>>  20:          2   IO-APIC-fasteoi   ehci_hcd:usb2
>>  21:      13860   IO-APIC-fasteoi   ehci_hcd:usb1
>>  22:     971970   IO-APIC-fasteoi   ohci_hcd:usb4
>>  23:       3446   IO-APIC-fasteoi   ohci_hcd:usb3
>>  28:      21524   PCI-MSI-edge      ahci
>
> FWIW a grep of the nouveau source doesnt bring up any hits on MSI, so it
> looks like the driver doesnt support it yet. Likewise no hits for ath5k,
> however ath9k does support it.

Nice, thanks Adam.  Sounds like ath9k is the way out.

- Grant

Re: [gentoo-user] Preventing a shared interrupt

[Grant]

>>> Does anyone know how to prevent my "nouveau" video card and "ath" wifi
>>> card from sharing interrupt 19?  I can't move the video card slot and
>>> I'd rather not move the wifi card slot if possible.
>>>
>>> BTW, does anyone see any other opportunity for interrupt optimization
>>> here?  I'm trying to optimize for video and especially audio, and I'm
>>> using a USB 1.0 sound card.
>>>
>>> # cat /proc/interrupts
>>>            CPU0
>>>   0:     473139   IO-APIC-edge      timer
>>>   1:          2   IO-APIC-edge      i8042
>>>   7:          1   IO-APIC-edge
>>>   8:        127   IO-APIC-edge      rtc0
>>>   9:          0   IO-APIC-fasteoi   acpi
>>>  12:          3   IO-APIC-edge      i8042
>>>  14:          0   IO-APIC-edge      ide0
>>>  19:     135597   IO-APIC-fasteoi   nouveau, ath
>>>  20:          2   IO-APIC-fasteoi   ehci_hcd:usb2
>>>  21:      13860   IO-APIC-fasteoi   ehci_hcd:usb1
>>>  22:     971970   IO-APIC-fasteoi   ohci_hcd:usb4
>>>  23:       3446   IO-APIC-fasteoi   ohci_hcd:usb3
>>>  28:      21524   PCI-MSI-edge      ahci
>>
>> FWIW a grep of the nouveau source doesnt bring up any hits on MSI, so it
>> looks like the driver doesnt support it yet. Likewise no hits for ath5k,
>> however ath9k does support it.
>
> Nice, thanks Adam.  Sounds like ath9k is the way out.
>
> - Grant

Ah nevermind, I guess ath9k is 802.11n only.  Maybe they'll add MSI
support to ath5k.

- Grant