[Qemu-devel] Re: [PATCH 1/2] Use getaddrinfo for migration

[Juan Quintela]

Peter Maydell <address@hidden> wrote:
> On 16 March 2011 21:01, Juan Quintela <address@hidden> wrote:
>> +static int tcp_server_bind(int fd, struct addrinfo *rp)
>> +{
>> +    int val = 1;
>> +    int ret;
>> +
>> +    /* allow fast reuse */
>> +    setsockopt(fd, SOL_SOCKET, SO_REUSEADDR, (const char *)&val,
>> +               sizeof(val));
>> +
>> +    ret = bind(fd, rp->ai_addr, rp->ai_addrlen);
>> +
>> +    if (ret == -1) {
>> +        ret = -socket_error();
>> +    }
>> +    return ret;
>> +
>> +}
>
> I know this is just moved code, but does this do the right thing
> on Windows? Windows semantics for SO_REUSEADDR are completely
> different from the Unix ones:
> http://blogs.msdn.com/b/wndp/archive/2005/08/03/anthony-jones.aspx
> http://msdn.microsoft.com/en-us/library/ms740618%28VS.85%29.aspx
>
> Roughly, Unix default behaviour is like Windows SO_EXCLUSIVEADDRUSE;
> Unix SO_REUSEADDR is like Windows default behaviour; and Windows
> SO_REUSEADDR behaviour is really weird and you don't want it.
>
> (I think Cygwin has a workaround for this to give unix semantics,
> but Windows qemu builds as a mingw app, so we get the MS semantics,
> right?)

No clue really, but again, this is the code that existed there, we can
fix on top of it.


>> +static int tcp_start_common(const char *str, int *fd, bool server)
>> +{
> [...]
>> +    s = getaddrinfo(name, service, &hints, &result);
>> +    if (s != 0) {
>> +        fprintf(stderr, "qemu: getaddrinfo: %s\n", gai_strerror(s));
>> +        return -EINVAL;
>> +    }
>
> It seems a bit odd to have a low level utility routine printing
> diagnostics to stderr rather than just returning the error
> details to its caller somehow. Ditto the other fprintf later.

Previous code did it.  I am not clear what to do here.  The way to get
an error for getaddrinfo is to return one string with gai_strerror().

Suggestions appreciated.

Later, Jaun.