> Hi Hans,
>
> On Thursday 07 April 2011 09:50:13 Hans Verkuil wrote:
>> > On Thu, 7 Apr 2011, Hans Verkuil wrote:
>
> [snip]
>
>> >> Regarding DESTROY_BUFS: perhaps we should just skip this for now and
>> wait
>> >> for the first use-case. That way we don't need to care about holes. I
>> >> don't like artificial restrictions like 'no holes'. If someone has a
>> good
>> >> use-case for selectively destroying buffers, then we need to look at
>> this
>> >> again.
>> >
>> > Sorry, skip what? skip the ioctl completely and rely on REQBUFS(0) /
>> > close()?
>>
>> Yes.
>
> I don't really like that as it would mix CREATE and REQBUFS calls.
> Applications should either use the old API (REQBUFS) or the new one, but
> not
> mix both.
That's a completely unnecessary limitation. And from the point of view of
vb2 it shouldn't even matter.
> The fact that freeing arbitrary spans of buffers gives us uneasy feelings
> might be a sign that the CREATE/DESTROY API is not mature enough. I'd
> rather
> try to solve the issue now instead of postponing it for later and discover
> that our CREATE API should have been different.
What gives me an uneasy feeling is prohibiting freeing arbitrary spans of
buffers. I rather choose not to implement the DESTROY ioctl instead of
implementing a limited version of it, also because we do not have proper
use cases yet. But I have no problems with the CREATE/DESTROY API as such.
Regards,
Hans
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
