[asterisk-dev] M9645: provide zaptel master timing to spans

[Matthew Fredrickson]

On May 2, 2007, at 9:41 AM, Tzafrir Cohen wrote:

> On Wed, May 02, 2007 at 10:21:14AM -0400, Andrew Kohlsmith wrote:
>> On Wednesday 02 May 2007 10:01 am, Tzafrir Cohen wrote:
>>> What do you think about such an interface? Useful?
>>> Now that I think about it, that code is called near the end of the
>>> relevant part of zt_recieve . Calling it a bit earlier may have given
>>> the called spans a slightly cleaner spans (e.g: independent on the
>>> number of pseudo channels).
>>
>> I'd written about this a few years ago, wherein you had multiple 
>> cards in a
>> system and only the first detected card would have its physical 
>> interrupt
>> unmasked.  Essentially you could put four quadspan cards in and have 
>> a single
>> 1ms interrupt service all spans.
>>
>> What do you think of expanding the patch to disable interrupts on 
>> subsequent
>> cards?
>
> That approach helps for multiple cards that are driven by the same
> driver. What about a system with two different card types?

Yeah, and cards that the DMA data transfer is synchronized across cards 
as well.  That's a wrong assumption in just about every case.  The only 
time you can do that (at least with Digium cards, TE4xxp and TE2xxp) is 
if you share timing across cards using the external timing bus.

Matthew Fredrickson