On Tue, Jul 23, 2013 at 4:38 AM, Tanu Kaskinen
<tanu.kaskinen at linux.intel.com> wrote:
> On Fri, 2013-07-12 at 15:07 -0300, jprvita at gmail.com wrote:
>> +/* Run from main thread */
>> +static void transport_config(struct userdata *u) {
>> + sbc_info_t *sbc_info = &u->sbc_info;
>> + a2dp_sbc_t *config;
>> +
>> + pa_assert(u->transport);
>> +
>> + u->sample_spec.format = PA_SAMPLE_S16LE;
>> + config = (a2dp_sbc_t *) u->transport->config;
>> +
>> + if (sbc_info->sbc_initialized)
>> + sbc_reinit(&sbc_info->sbc, 0);
>> + else
>> + sbc_init(&sbc_info->sbc, 0);
>> + sbc_info->sbc_initialized = true;
>> +
>> + switch (config->frequency) {
>> + case SBC_SAMPLING_FREQ_16000:
>> + sbc_info->sbc.frequency = SBC_FREQ_16000;
>> + u->sample_spec.rate = 16000U;
>> + break;
>> + case SBC_SAMPLING_FREQ_32000:
>> + sbc_info->sbc.frequency = SBC_FREQ_32000;
>> + u->sample_spec.rate = 32000U;
>> + break;
>> + case SBC_SAMPLING_FREQ_44100:
>> + sbc_info->sbc.frequency = SBC_FREQ_44100;
>> + u->sample_spec.rate = 44100U;
>> + break;
>> + case SBC_SAMPLING_FREQ_48000:
>> + sbc_info->sbc.frequency = SBC_FREQ_48000;
>> + u->sample_spec.rate = 48000U;
>> + break;
>> + default:
>> + pa_assert_not_reached();
>> + }
>
> If broken config is handled with assertions (which I totally agree is a
> good way to do it), then the config definitely must be checked at the
> time it's received.
>
Ok, I'm adding that to the patch that implements the SetConfiguration() method.
>> @@ -517,6 +722,8 @@ void pa__done(pa_module *m) {
>> if (u->device_connection_changed_slot)
>> pa_hook_slot_free(u->device_connection_changed_slot);
>>
>> + sbc_finish(&u->sbc_info.sbc);
>
> This should be done only if sbc_initialized is true. The code may work
> just fine, since the sbc struct is initialized to all zero when userdata
> is allocated, but that's relying on undocumented internal details of
> sbc_finish().
>
Ok.
--
Jo?o Paulo Rechi Vita
http://about.me/jprvita
