[Spice-devel] [PATCH spice-server] reds: Check we don't register a channel twice in reds_register_channel

[Snir Sheriber]

On 3/18/19 12:12 PM, Frediano Ziglio wrote:
>> Hi,
>>
>> On 3/15/19 11:27 AM, Frediano Ziglio wrote:
>>> To avoid possibly regression check it only if extra checks are
>>> enabled.
>>
>> Currently there's already a "channel duplication" warning upon connection
>> but won't hurt to have this extra check and emit an error.
>>
> I cannot find it. This patch is for spice-server.


Oh, sorry, "duplicate channel" on red_client_add_channel


>
>>> This allowed to check previous "Move channel registration to constructed
>>
>> This allow/s? i don't get this sentence, it just avoids duplicate call.
>> I think
>> i'd not even mention the previous commit
>>
> No, it will also exit the program if this condition is met.


Yes, i got it, i meant that such patch would have make sense also before 
the "Move channel reg.." commit,
so not necessarily need to mention it.



>
>> Snir.
>>
>>> vfunc" commit.
>>>
>>> Signed-off-by: Frediano Ziglio <fziglio at redhat.com>
>>> ---
>>>    server/reds.c | 5 +++++
>>>    1 file changed, 5 insertions(+)
>>>
>>> diff --git a/server/reds.c b/server/reds.c
>>> index 429f8142..e182eba7 100644
>>> --- a/server/reds.c
>>> +++ b/server/reds.c
>>> @@ -380,6 +380,11 @@ void stat_remove_counter(SpiceServer *reds,
>>> RedStatCounter *counter)
>>>    void reds_register_channel(RedsState *reds, RedChannel *channel)
>>>    {
>>>        spice_assert(reds);
>>> +    if (spice_extra_checks) {
>>> +        uint32_t this_type, this_id;
>>> +        g_object_get(channel, "channel-type", &this_type, "id", &this_id,
>>> NULL);
>>> +        spice_assert(reds_find_channel(reds, this_type, this_id) == NULL);
>>> +    }
>>>        reds->channels = g_list_prepend(reds->channels, channel);
>>>        // create new channel in the client if possible
>>>        main_channel_registered_new_channel(reds->main_channel, channel);
> Frediano