On 2014-05-06 11:34, Tanu Kaskinen wrote:
> On Tue, 2014-04-29 at 15:22 +0200, David Henningsson wrote:
>> The srchannel is enabled if protocol version >= 30 and
>> SHM is available.
>>
>> Signed-off-by: David Henningsson <david.henningsson at canonical.com>
>> ---
>> src/pulsecore/protocol-native.c | 44 +++++++++++++++++++++++++++++++++++++++++
>> 1 file changed, 44 insertions(+)
>>
>> diff --git a/src/pulsecore/protocol-native.c b/src/pulsecore/protocol-native.c
>> index 65b2636..c2ec644 100644
>> --- a/src/pulsecore/protocol-native.c
>> +++ b/src/pulsecore/protocol-native.c
>> @@ -2568,6 +2568,45 @@ static void command_exit(pa_pdispatch *pd, uint32_t command, uint32_t tag, pa_ta
>> pa_pstream_send_simple_ack(c->pstream, tag); /* nonsense */
>> }
>>
>> +static void setup_srchannel(pa_native_connection *c) {
>> + pa_srchannel_template srt;
>> + pa_srchannel *sr;
>> + pa_memchunk mc;
>> + pa_tagstruct *t;
>> + int fdlist[2];
>> +
>> + if (c->version < 30) {
>> + pa_log_debug("Disabling srchannel, reason: Protocol too old");
>> + return;
>> + }
>> +
>> + if (!c->protocol->core->rw_mempool) {
>> + pa_log_debug("Disabling srchannel, reason: No rw memory pool");
>> + return;
>> + }
>> +
>> + pa_log_debug("Enabling srchannel...");
>> + sr = pa_srchannel_new(c->protocol->core->mainloop, c->protocol->core->rw_mempool);
>> + pa_srchannel_export(sr, &srt);
>> +
>> + /* Allow client to access the memblock */
>> + mc.memblock = srt.memblock;
>> + mc.index = 0;
>> + mc.length = pa_memblock_get_length(srt.memblock);
>> + pa_pstream_send_memblock(c->pstream, 0, 0, 0, &mc);
>> +
>> + /* Send enable command to client */
>> + t = pa_tagstruct_new(NULL, 0);
>> + pa_tagstruct_putu32(t, PA_COMMAND_ENABLE_RINGBUFFER);
>> + pa_tagstruct_putu32(t, (uint32_t) -1); /* tag */
>> + fdlist[0] = srt.readfd;
>> + fdlist[1] = srt.writefd;
>> + pa_pstream_send_tagstruct_with_fds(c->pstream, t, 2, fdlist);
>> +
>> + /* ...and switch over */
>> + pa_pstream_set_srchannel(c->pstream, sr);
>
> There seems to be no way for the client to refuse this, so we can never
> stop supporting srchannel without breaking protocol compatibility.
Fixed in next revision.
--
David Henningsson, Canonical Ltd.
https://launchpad.net/~diwic
