Re: [PATCH 2/5] ALSA: snd-usb: implement new endpoint streaming model

[Daniel Mack <zonque@gmail.com>]

On 12/20/2011 04:42 PM, Takashi Iwai wrote:
> At Tue, 20 Dec 2011 10:48:37 +0100,
> Daniel Mack wrote:
>>
>> @@ -99,7 +104,7 @@ static int deactivate_urbs(struct snd_usb_substream *subs, int force, int can_sl
>>   */
>>  static void release_urb_ctx(struct snd_urb_ctx *u)
>>  {
>> -	if (u->urb) {
>> +	if (!u || u->urb) {
>>  		if (u->buffer_size)
> 
> This will lead to Oops when u == NULL, no?

Yes, that's right. That said, it seems this hunk can be removed
entirely. I'll fix this in the next version.

>> +/* determine the number of frames in the next packet */
>> +static int next_packet_size(struct snd_usb_endpoint *ep)
>> +{
>> +	unsigned long flags;
>> +
>> +	if (ep->fill_max)
>> +		return ep->maxframesize;
>> +
>> +	spin_lock_irqsave(&ep->lock, flags);
>> +	ep->phase = (ep->phase & 0xffff)
>> +		+ (ep->freqm << ep->datainterval);
>> +	spin_unlock_irqrestore(&ep->lock, flags);
>> +
>> +	return min(ep->phase >> 16, ep->maxframesize);
> 
> If you need a protection here, safer to cover until min(), I guess.

Ok, will do.

I'll wait with a new version until there's feedback from people who
tested the code on their hardware :)

Thanks for the review.

Daniel