At Mon, 15 Dec 2008 11:10:12 +0100 (CET), Jean-Pierre ANDRE wrote:
Hi Takashi,
Than you for your help.
Switching to polling mode is OK and mostly harmless. But single_cmd is not. There is a VERY big difference between them.
Ok.
Probably related is the fact that I have never seen an IRQ 22.
from proc/interrupts :
22: 0 0 IO-APIC-fasteoi uhci_hcd:usb6, HDA Intel
Hum, then it must be an interrupt thing. Ask rather ACPI guys.
Well, you must be right. Using the irqpoll boot option, the sound is correct, and I see four IRQ 20 per second (and none when there is no sound).
- Can you confirm an expected IRQ every 250ms ?
This depends on the application set up, i.e. period size. The value appears reasonable, though.
The APIC routes PCI INT B to IRQ 22, and alsa is expecting IRQ 22. Now I have to get alsa to expect IRQ 20 or get APIC to route an appropriate PCI INT to IRQ 22....
- Now I get a warning "hda-intel: IRQ timing workaround
is activated for card #0. Suggest a bigger bdl_pos_adj."
What is the range of values I should try ?
Usually bdl_pos_adj=32 should suffice. You can give more, but it's rather the system problem in such a case.
Takashi