[linux-dvb] Re: V4 API Implementation Q's

[Johannes Stezenbach <js@convergence.de>]

Rob.McConnell@Zarlink.Com wrote:
> 
> Just got around to having a look @ the V4 API code in CVS and was well
> impressed at how far we have got since I spoke last : )
> I see that we have decided on using the original V4 API as intended by
> Johannes/Holger rather than Michael's with different output devices.
> Having had a quick look at the code, I cannot see where we perform checks
> on the allocation of feeds/filters.  What I mean is there doesn't seem to
> be any code that checks to ensure that there are enough underlying h/w
> filters left to allocate when someone open's up an fd on the demux device.
> If it is there, then could you point me in the right direction - thanks : )
> 
> This leads me into the next point/question regarding output devices.
> Wouldn't it be better to have output demux devices such as PID filter,
> recording filter, section filter, decoder feed with each one allocated a
> different minor number.  Then when the user app came to open up an fd on
> this device, we would know immediately if we had enough
> resources/filters/feeds left to allow the "open" syscall to succeed.  At
> the moment we open up an fd, but we don't know what the underlying usage of
> it is going to be until we perform one of the demux IOCTLs to setup a
> feed/filter.  Only here, is when we know whether we can actually get a
> handle to a feed/filter.  Not really that elegant is it?

Well, this is the way we designed it. It fits our application's
use of resources -- we don't grab them in advance to "reserve" them,
we allocate resources when we need them and release them when
we are done with them. So it does not make a difference if open()
fails or ioctl(SET_FILTER) fails.

Johannes


-- 
Info:
To unsubscribe send a mail to ecartis@linuxtv.org with "unsubscribe linux-dvb" as subject.