At Mon, 31 Oct 2011 13:38:20 +0100, Daniel Mack wrote:
On 10/31/2011 01:10 PM, Daniel Mack wrote:
I didn't sign-off the patches on purpose, as I would really like to get them reviewed before they go in. Can people have a look and state whether the whole idea is at all sane?
To briefly state the idea as such: the new implementation defines a model (snd_usb_endpoint) that handles everything that is related to an USB endpoint and its streaming. There are functions to activate and deactivate an endpoint (which call usb_set_interface()), and to start and stop its URBs. It also has function pointers to be called when data was received or is about to be sent, and pointer to a sync slave (another snd_usb_endpoint) that is informed when data has been received. A snd_usb_endpoint knows about its state and implements a refcounting, so only the first user will actually start the URBs and only the last one to stop it will tear them down again.
With this sort of abstraction, the actual streaming is decoupled from the pcm handling, which makes the "implicit feedback" mechanisms easy to implement. All the code that actually handles the payload of a stream's packets is now implemented in pcm.c, which is were it belongs to.
But I'm sure there are some unresolved corner cases which need attention.
The overall design looks good to me. Nice work! A few nitpicking:
- No need for check of activated flags at starting streams? - Better to clear subs->data_endpoint & co at closing. - There might be unbalances when deactivate_endpoints() isn't called in prepare callback before activate_endpoints(). The apps may call like open -> hw_params -> prepare -> hw_params -> prepare -> trigger and trigger (stop) -> prepare -> trigger (start) So, be careful about refcounting and active flags.
And, yes, the protection for snd_usb_add_endpoint() would be really needed for concurrent access. The card-global mutex would be enough.
thanks,
Takashi