Login | Register For Free | Help
Search for: (Advanced)

Mailing List Archive: ivtv: devel

PVR-500 apic/irq troubles

 

 

ivtv devel RSS feed   Index | Next | Previous | View Threaded


mrvanes at gmail

Jul 7, 2008, 6:57 AM

Post #1 of 7 (3117 views)
Permalink
PVR-500 apic/irq troubles

Hi,

I haven't caught up with the mailinglist for some time, so this
problem may have been covered before, but the problem still exists in
2.6.25 so I thought it might be relevant nevertheless.

I recently upgraded my VIA EPIA ML6000 mediacenter MB to a newer
LN-1000. This sports a C7 proc and SATA RAID controller.

Since the upgrade I've had some trouble getting the Hauppauge PVR-500
to work. I got it to work eventually by disabling the APIC controller
(noapic) and using polling interrupts (irqpoll) in the kernel.

So far no problem, because I have a working system, but there are
still some questions I'd like to put on this list:

I read that irqpoll is not to be preferred, because this will ask all
interrupt handlers if they want to serve the interrupt each time one
occurs. Seems a waste of time to me. But without irqpoll I get the
'nobody cared (try booting with the "irqpoll" option)' error on
accessing /dev/video0.

Without the apic (noapic) my interrupts are configured as follows:
$ cat /proc/interrupts
CPU0
0: 270 XT-PIC-XT timer
1: 2 XT-PIC-XT i8042
2: 0 XT-PIC-XT cascade
5: 7085 XT-PIC-XT ehci_hcd:usb1, VIA8237, ivtv0, ivtv1
8: 2 XT-PIC-XT rtc
10: 96518 XT-PIC-XT uhci_hcd:usb2, uhci_hcd:usb3,
eth0, via [at] pc:0000:01:00.0
11: 66316 XT-PIC-XT uhci_hcd:usb4, uhci_hcd:usb5
14: 3885 XT-PIC-XT ide0
15: 83 XT-PIC-XT ide1
NMI: 0 Non-maskable interrupts
LOC: 1306477 Local timer interrupts
TRM: 0 Thermal event interrupts
SPU: 0 Spurious interrupts
ERR: 0

with the apic it looks like this:
CPU0
0: 63479 IO-APIC-edge timer
1: 8 IO-APIC-edge i8042
2: 0 XT-PIC-XT cascade
8: 2 IO-APIC-edge rtc
12: 0 IO-APIC-edge i8042
14: 1102 IO-APIC-edge ide0
15: 81 IO-APIC-edge ide1
18: 0 IO-APIC-fasteoi ivtv0, ivtv1
21: 20 IO-APIC-fasteoi ehci_hcd:usb1, uhci_hcd:usb2,
uhci_hcd:usb3, uhci_hcd:usb4, uhci_hcd:usb5
22: 0 IO-APIC-fasteoi VIA8237
23: 1452 IO-APIC-fasteoi eth0
NMI: 0
LOC: 63441
ERR: 0
MIS: 0

To me, this looks a lot more sane, but in this situation only the
second tuner (/dev/video1) works. If I do "cat /dev/video0" nothing
happens (if I do the same for /dev/video1 I see the usual garbage).
Depending on kernel and irqpoll I sometimes get the 'nobody cared'
error, depending on kernel and ivtv version.

I read some posts about the VIA RAID being a problem with ivtv, but
disabling the RAID in BIOS doesn't help.

So, is it sane to have ivtv0 and ivtv1 share the same interrupt? Why
does it work with noapic/irqpoll and why does only the second tuner
work if I use the apic, without irqpoll?

Regards,
Martin
--
if but was of any use, it would be a logic operator

_______________________________________________
ivtv-devel mailing list
ivtv-devel [at] ivtvdriver
http://ivtvdriver.org/mailman/listinfo/ivtv-devel


awalls at radix

Jul 7, 2008, 5:08 PM

Post #2 of 7 (2990 views)
Permalink
Re: PVR-500 apic/irq troubles [In reply to]

On Mon, 2008-07-07 at 15:57 +0200, Martin van Es wrote:
> Hi,
>
> I haven't caught up with the mailinglist for some time, so this
> problem may have been covered before, but the problem still exists in
> 2.6.25 so I thought it might be relevant nevertheless.


Here's a post the describes the same symptoms in a slightly different
setup with a VIA8237 chip as well:

http://ivtvdriver.org/pipermail/ivtv-users/2007-April/006332.html

It may give you more insight into your symptoms or at least provide some
comiseration.

-Andy

> I recently upgraded my VIA EPIA ML6000 mediacenter MB to a newer
> LN-1000. This sports a C7 proc and SATA RAID controller.
>
> Since the upgrade I've had some trouble getting the Hauppauge PVR-500
> to work. I got it to work eventually by disabling the APIC controller
> (noapic) and using polling interrupts (irqpoll) in the kernel.
>
> So far no problem, because I have a working system, but there are
> still some questions I'd like to put on this list:
>
> I read that irqpoll is not to be preferred, because this will ask all
> interrupt handlers if they want to serve the interrupt each time one
> occurs. Seems a waste of time to me. But without irqpoll I get the
> 'nobody cared (try booting with the "irqpoll" option)' error on
> accessing /dev/video0.
>
> Without the apic (noapic) my interrupts are configured as follows:
> $ cat /proc/interrupts
> CPU0
> 0: 270 XT-PIC-XT timer
> 1: 2 XT-PIC-XT i8042
> 2: 0 XT-PIC-XT cascade
> 5: 7085 XT-PIC-XT ehci_hcd:usb1, VIA8237, ivtv0, ivtv1
> 8: 2 XT-PIC-XT rtc
> 10: 96518 XT-PIC-XT uhci_hcd:usb2, uhci_hcd:usb3,
> eth0, via [at] pc:0000:01:00.0
> 11: 66316 XT-PIC-XT uhci_hcd:usb4, uhci_hcd:usb5
> 14: 3885 XT-PIC-XT ide0
> 15: 83 XT-PIC-XT ide1
> NMI: 0 Non-maskable interrupts
> LOC: 1306477 Local timer interrupts
> TRM: 0 Thermal event interrupts
> SPU: 0 Spurious interrupts
> ERR: 0
>
> with the apic it looks like this:
> CPU0
> 0: 63479 IO-APIC-edge timer
> 1: 8 IO-APIC-edge i8042
> 2: 0 XT-PIC-XT cascade
> 8: 2 IO-APIC-edge rtc
> 12: 0 IO-APIC-edge i8042
> 14: 1102 IO-APIC-edge ide0
> 15: 81 IO-APIC-edge ide1
> 18: 0 IO-APIC-fasteoi ivtv0, ivtv1
> 21: 20 IO-APIC-fasteoi ehci_hcd:usb1, uhci_hcd:usb2,
> uhci_hcd:usb3, uhci_hcd:usb4, uhci_hcd:usb5
> 22: 0 IO-APIC-fasteoi VIA8237
> 23: 1452 IO-APIC-fasteoi eth0
> NMI: 0
> LOC: 63441
> ERR: 0
> MIS: 0
>
> To me, this looks a lot more sane, but in this situation only the
> second tuner (/dev/video1) works. If I do "cat /dev/video0" nothing
> happens (if I do the same for /dev/video1 I see the usual garbage).
> Depending on kernel and irqpoll I sometimes get the 'nobody cared'
> error, depending on kernel and ivtv version.
>
> I read some posts about the VIA RAID being a problem with ivtv, but
> disabling the RAID in BIOS doesn't help.
>
> So, is it sane to have ivtv0 and ivtv1 share the same interrupt? Why
> does it work with noapic/irqpoll and why does only the second tuner
> work if I use the apic, without irqpoll?
>
> Regards,
> Martin


_______________________________________________
ivtv-devel mailing list
ivtv-devel [at] ivtvdriver
http://ivtvdriver.org/mailman/listinfo/ivtv-devel


mrvanes at gmail

Jul 8, 2008, 1:28 AM

Post #3 of 7 (2994 views)
Permalink
Re: PVR-500 apic/irq troubles [In reply to]

On 7/8/08, Andy Walls <awalls [at] radix> wrote:
> On Mon, 2008-07-07 at 15:57 +0200, Martin van Es wrote:
> > Hi,
> >
> > I haven't caught up with the mailinglist for some time, so this
> > problem may have been covered before, but the problem still exists in
> > 2.6.25 so I thought it might be relevant nevertheless.
>
> Here's a post the describes the same symptoms in a slightly different
> setup with a VIA8237 chip as well:
>
> http://ivtvdriver.org/pipermail/ivtv-users/2007-April/006332.html
>
> It may give you more insight into your symptoms or at least provide some
> comiseration.
>

Thx for the link. Yes, it looks like Ronald and I can shake hands.

Leaves me wondering what is bothering us and what the consequences of
irqpoll are? I must admit MythTV seem quite stable and recording is
good using noapic and irqpoll (on both tuners) it just feel 'wrong'
not to use the apic (and have the interrupts assigned better) and use
of irqpoll.

Can anyone shed some light on that?

Grtz
Martin

_______________________________________________
ivtv-devel mailing list
ivtv-devel [at] ivtvdriver
http://ivtvdriver.org/mailman/listinfo/ivtv-devel


harri.jarvi at ajatus

Jul 8, 2008, 5:37 AM

Post #4 of 7 (2989 views)
Permalink
Re: PVR-500 apic/irq troubles [In reply to]

On Tue, Jul 08, 2008 at 10:28:10 +0200, Martin van Es wrote:
> Leaves me wondering what is bothering us and what the consequences of
> irqpoll are? I must admit MythTV seem quite stable and recording is
> good using noapic and irqpoll (on both tuners) it just feel 'wrong'
> not to use the apic (and have the interrupts assigned better) and use
> of irqpoll.
>
> Can anyone shed some light on that?

I don't have experience in this matter, but your problem just prompted
me to ask if you have tried if it helps to force the two chips to have
separate irq's assigned. Maybe there is a bug in the driver that makes
irq sharing fail.

Forcing separate irq's to the two chips could require changes in the
source.

Yours,
Harri

_______________________________________________
ivtv-devel mailing list
ivtv-devel [at] ivtvdriver
http://ivtvdriver.org/mailman/listinfo/ivtv-devel


mrvanes at gmail

Jul 8, 2008, 6:07 AM

Post #5 of 7 (2964 views)
Permalink
Re: PVR-500 apic/irq troubles [In reply to]

On 7/8/08, Harri Järvi <harri.jarvi [at] ajatus> wrote:
> On Tue, Jul 08, 2008 at 10:28:10 +0200, Martin van Es wrote:
> > Leaves me wondering what is bothering us and what the consequences of
> > irqpoll are? I must admit MythTV seem quite stable and recording is
> > good using noapic and irqpoll (on both tuners) it just feel 'wrong'
> > not to use the apic (and have the interrupts assigned better) and use
> > of irqpoll.
> >
> > Can anyone shed some light on that?
>
>
> I don't have experience in this matter, but your problem just prompted
> me to ask if you have tried if it helps to force the two chips to have
> separate irq's assigned.

I would if I knew how?

> Forcing separate irq's to the two chips could require changes in the
> source.

Hmmm... then I'd definitely would need a little help to point me to
the right place in the code.

Regards,
Martin

_______________________________________________
ivtv-devel mailing list
ivtv-devel [at] ivtvdriver
http://ivtvdriver.org/mailman/listinfo/ivtv-devel


harri.jarvi at ajatus

Jul 8, 2008, 6:45 AM

Post #6 of 7 (2974 views)
Permalink
Re: PVR-500 apic/irq troubles [In reply to]

On Tue, Jul 08, 2008 at 15:07:54 +0200, Martin van Es wrote:
> On 7/8/08, Harri Järvi <harri.jarvi [at] ajatus> wrote:
> > I don't have experience in this matter, but your problem just prompted
> > me to ask if you have tried if it helps to force the two chips to have
> > separate irq's assigned.
>
> I would if I knew how?
>
> > Forcing separate irq's to the two chips could require changes in the
> > source.
>
> Hmmm... then I'd definitely would need a little help to point me to
> the right place in the code.

Sorry, forgot to mention I don't know where to look. Maybe someone
more familiar with irq's could help.

Yours,
Harri

_______________________________________________
ivtv-devel mailing list
ivtv-devel [at] ivtvdriver
http://ivtvdriver.org/mailman/listinfo/ivtv-devel


mrvanes at gmail

Jul 9, 2008, 12:02 PM

Post #7 of 7 (2943 views)
Permalink
Re: PVR-500 apic/irq troubles [In reply to]

On Tue, Jul 8, 2008 at 15:45, Harri Järvi <harri.jarvi [at] ajatus> wrote:
> On Tue, Jul 08, 2008 at 15:07:54 +0200, Martin van Es wrote:
>> On 7/8/08, Harri Järvi <harri.jarvi [at] ajatus> wrote:
>> > I don't have experience in this matter, but your problem just prompted
>> > me to ask if you have tried if it helps to force the two chips to have
>> > separate irq's assigned.
>>
>> I would if I knew how?
>>
>> > Forcing separate irq's to the two chips could require changes in the
>> > source.

Or enabling acpi=force as a kernel boot parameter. It turns out (after
close inspection of the boot messages) that the ACPI table of the
motherboard is crappy and is not used by default and acpi=force is
hinted at in the output. After forcing the use of ACPI I have this
/proc/interrupts:

CPU0
0: 67 IO-APIC-edge timer
1: 2 IO-APIC-edge i8042
8: 2 IO-APIC-edge rtc
9: 0 IO-APIC-fasteoi acpi
14: 12445 IO-APIC-edge ide0
15: 81 IO-APIC-edge ide1
16: 71613 IO-APIC-fasteoi via [at] pc:0000:01:00.0
17: 49536 IO-APIC-fasteoi ivtv0
18: 37 IO-APIC-fasteoi ivtv1
20: 38 IO-APIC-fasteoi sata_via
21: 127 IO-APIC-fasteoi ehci_hcd:usb1, uhci_hcd:usb2,
uhci_hcd:usb3, uhci_hcd:usb4, uhci_hcd:usb5
22: 5443 IO-APIC-fasteoi VIA8237
23: 608 IO-APIC-fasteoi eth0
NMI: 0 Non-maskable interrupts
LOC: 2179623 Local timer interrupts
TRM: 0 Thermal event interrupts
SPU: 0 Spurious interrupts
ERR: 0
MIS: 0

Which looks even more sane than it already did ;)
Hope this helps other users of VIA MB's in the future and if Ronald is
listening: try it. Curious if it works for you as well.


Regards,
Martin

_______________________________________________
ivtv-devel mailing list
ivtv-devel [at] ivtvdriver
http://ivtvdriver.org/mailman/listinfo/ivtv-devel

ivtv devel RSS feed   Index | Next | Previous | View Threaded
 
 


Interested in having your list archived? Contact Gossamer Threads
 
  Web Applications & Managed Hosting Powered by Gossamer Threads Inc.