Re: [Qemu-devel] [PATCH v2 2/5] net: fix -netdev socket, fd= for UDP sockets

[Jens Freimann <jfreimann@redhat.com>]

On Fri, Nov 03, 2017 at 06:46:57PM +0000, Peter Maydell wrote:
>On 8 August 2017 at 21:38, Jens Freimann <jfreimann@redhat.com> wrote:
>> @@ -333,8 +333,13 @@ static NetSocketState *net_socket_fd_init_dgram(NetClientState *peer,
>>       * by ONLY ONE process: we must "clone" this dgram socket --jjo
>>       */
>>
>> -    if (is_connected) {
>> -        if (getsockname(fd, (struct sockaddr *) &saddr, &saddr_len) == 0) {
>> +    if (is_connected && mcast != NULL) {
>
>This changes the condition() under which we fill in the struct sockaddr_in saddr
>from "if (is_connected)" to "if (is_connected && mcast != NULL)"...
>
>> +            if (parse_host_port(&saddr, mcast) < 0) {
>> +                fprintf(stderr,
>> +                        "qemu: error: init_dgram: fd=%d failed parse_host_port()\n",
>> +                        fd);
>> +                goto err;
>> +            }
>>              /* must be bound */
>>              if (saddr.sin_addr.s_addr == 0) {
>>                  fprintf(stderr, "qemu: error: init_dgram: fd=%d unbound, "
>
>...but later in the function we do:
>
>    /* mcast: save bound address as dst */
>    if (is_connected) {

This should be changed to "if (is_connected && mcast != NULL)" because
it is only necessary to do this if there is a multicast address specified. 

>        s->dgram_dst = saddr;
>        snprintf(nc->info_str, sizeof(nc->info_str),
>                 "socket: fd=%d (cloned mcast=%s:%d)",
>                 fd, inet_ntoa(saddr.sin_addr), ntohs(saddr.sin_port));
>    } else {
>        snprintf(nc->info_str, sizeof(nc->info_str),
>                 "socket: fd=%d", fd);
>    }
>
>and coverity correctly points out that if is_connected is true
>but mcast is NULL then we use 'saddr' without having initialized
>it properly.
>
>Any suggestions for the correct fix for this?

I think we should initialize saddr to 0 and do the above change. I'll send a
patch.

Thanks!

regards,
Jens