nAt Sat, 1 Nov 2008 19:08:59 +0200, Karsten Wiese wrote:
Hi Takashi,
Am Sonntag, 26. Oktober 2008 schrieb Takashi Iwai:
At Sun, 26 Oct 2008 17:31:02 +0100, =?UTF-8?Q?Vedran_Mileti=C4=87?= wrote:
Yeap, that one does it. No need to revert it fully, though. I just uncommented the unsigned char mask line and removed that #if 0 and #endif. Now it works perfectly. Thanks! Can you fix it in your tree?
Well, before doing that, I'd like to understand the problem more.
If the problem is about the initial IRQ mask value, setting 0 should work as well. Could you change the code just to do like below, and check whether it works? outb(0, ICEREG1724(ice, IRQMASK));
Makes a pc hang sometimes here, while reverting the entire patch is stable. no midi tested though, only audio.
Hmm, that's a bad news. Basically my change is just to disable MPU_TX irq when unresolved irqs occur. The detection was already there, but it was enabled only when CONFIG_SND_DEBUG.
So, the question is, whether does the hang-up happens even with the old code when you enable CONFIG_SND_DEBUG?
Also, a bit more detail would be appreciated -- what kind of hang-up, whether you get the error message, or so.
thanks,
Takashi