[Asterisk-Dev] Re: FXO PCI Master abort

[Tzafrir Cohen]

On Sat, Aug 13, 2005 at 03:24:28PM +0200, Mark Burton wrote:
> Dear Zaptel and wcfxo devellopers,
> 
> Hi, so far I have had no success moving this issue forward. Carl 
> Andersson has been kind enough to help build various kernels to try, 
> but with no success.
> 
> So, I have tried to debug the problem directly. So far I have applied 
> the patch below to wcfxo.c. (on the latest CVS head) This makes my 
> system stable again (in that I dont have to keep pressing the big red 
> button).
> In principle I can not see how this patch would harm a working system, 
> and seems to me to be the right thing to do anyway. I would love to 
> know why not.
> 
> However, of course, this does not fix the problem.
> 
> Firstly the watchdog never re-starts the DMA engine (even with the 
> watchdog enabled in zaptel). This means that the driver sits there 
> "impotent" and in order to recover you have to unload and reload the 
> module.
> 	Hardly ideal.
> 
> Second, of course, I am no nearer finding out why I am getting the 
> erroneous IRQ in the first place!
> 
> Any help much appreciated...
> 
> Cheers
> 
> Mark.
> 
> 
> 

> --- wcfxo.c     2005-08-13 14:16:44.690969912 +0100
> +++ wcfxo.c.old 2005-08-13 08:51:00.000000000 +0100
> @@ -726,18 +726,11 @@
>         /* Reset Master and TDM */
>         outb(0x01, wc->ioaddr + WC_CNTL);
>         outb(0x01, wc->ioaddr + WC_OPER);
> -
> -        printk("UnMasking IRQ\n");
> -       outb(0x3f, wc->ioaddr + WC_MASK0);
>  }
> 
> 
>  static void wcfxo_stop_dma(struct wcfxo *wc)
>  {
> -       /* Enable interrupts (we care about all of them, except the one 
> that gave us the abort) */
> -        printk("Masking IRQ's, waiting for watchdog to restart\n");
> -       outb(0x2f, wc->ioaddr + WC_MASK0);
> -
>         outb(0x00, wc->ioaddr + WC_OPER);
>  }

Have you posted a reveresed patch on purpose?

I also rather have patches attached, to avoid stupid white-space
issues. Sorry, nothing smarter to say.

But from the description of the problem it does sound like Zaptel fails
to handel fairly a slightly broken hardware, right?

-- 
Tzafrir Cohen     icq#16849755  +972-50-7952406
tzafrir.cohen at xorcom.com  http://www.xorcom.com