Re: ws2_32: negate return value of SIOCATMARK (try 2)

[Mike Kaplinskiy]

2010/4/28 André Hentschel <nerv(a)dawncrow.de>:
> confirmed by Mike Kaplinskiy
>
> try 1 was on 4th November 2009
>
> ---
>  dlls/ws2_32/socket.c     |   16 ++++++++++++++--
>  dlls/ws2_32/tests/sock.c |    2 +-
>  2 files changed, 15 insertions(+), 3 deletions(-)
>
> diff --git a/dlls/ws2_32/socket.c b/dlls/ws2_32/socket.c
> index e6a6973..62598d2 100644
> --- a/dlls/ws2_32/socket.c
> +++ b/dlls/ws2_32/socket.c
> @@ -2800,8 +2800,20 @@ int WINAPI WS_ioctlsocket(SOCKET s, LONG cmd, WS_u_long *argp)
>         return 0;
>
>     case WS_SIOCATMARK:
> -        newcmd=SIOCATMARK;
> -        break;
> +        fd = get_sock_fd( s, 0, NULL );
> +        if (fd != -1)
> +        {
> +            if( ioctl(fd, newcmd, (char*)argp ) == 0 )
> +            {
> +                release_sock_fd( s, fd );
> +                /* return value must be inverted here */
> +                *argp = !*argp;
> +                return 0;
> +            }
> +            SetLastError((errno == EBADF) ? WSAENOTSOCK : wsaErrno());
> +            release_sock_fd( s, fd );
> +        }
> +        return SOCKET_ERROR;
>
>     case WS_FIOASYNC:
>         WARN("Warning: WS1.1 shouldn't be using async I/O\n");
> diff --git a/dlls/ws2_32/tests/sock.c b/dlls/ws2_32/tests/sock.c
> index 88b62e3..7a7c688 100644
> --- a/dlls/ws2_32/tests/sock.c
> +++ b/dlls/ws2_32/tests/sock.c
> @@ -2486,7 +2486,7 @@ static void test_ioctlsocket(void)
>
>     ret = ioctlsocket(sock, SIOCATMARK, &arg);
>     if(ret != SOCKET_ERROR)
> -        todo_wine ok(arg, "expected a non-zero value\n");
> +        ok(arg, "expected a non-zero value\n");
>  }
>
>  static int drain_pause=0;
> --
>
> Best Regards, André Hentschel

Hey André,

I don't think we should rush this. The documentation says that the
return values are basically opposites, but I think we need more tests
with actual OOB data and different settings of SO_OOBINBLINE. MSDN has
far too little information about SIOCATMARK and some things they say
don't make a lot of sense.

Mike.

ws2_32: negate return value of SIOCATMARK (try 2)

[André Hentschel]

Am 29.04.2010 18:39, schrieb Mike Kaplinskiy:
> 2010/4/28 André Hentschel <nerv(a)dawncrow.de>:
> > confirmed by Mike Kaplinskiy
> >
> > try 1 was on 4th November 2009
> >
> > ---
> >  dlls/ws2_32/socket.c     |   16 ++++++++++++++--
> >  dlls/ws2_32/tests/sock.c |    2 +-
> >  2 files changed, 15 insertions(+), 3 deletions(-)
> >
> > diff --git a/dlls/ws2_32/socket.c b/dlls/ws2_32/socket.c
> > index e6a6973..62598d2 100644
> > --- a/dlls/ws2_32/socket.c
> > +++ b/dlls/ws2_32/socket.c
> > @@ -2800,8 +2800,20 @@ int WINAPI WS_ioctlsocket(SOCKET s, LONG cmd, WS_u_long *argp)
> >         return 0;
> >
> > case WS_SIOCATMARK:
> > -        newcmd=SIOCATMARK;
> > -        break;
> > +        fd = get_sock_fd( s, 0, NULL );
> > +        if (fd != -1)
> > +        {
> > +            if( ioctl(fd, newcmd, (char*)argp ) == 0 )
> > +            {
> > +                release_sock_fd( s, fd );
> > +                /* return value must be inverted here */
> > +                *argp = !*argp;
> > +                return 0;
> > +            }
> > +            SetLastError((errno == EBADF) ? WSAENOTSOCK : wsaErrno());
> > +            release_sock_fd( s, fd );
> > +        }
> > +        return SOCKET_ERROR;
> >
> > case WS_FIOASYNC:
> >         WARN("Warning: WS1.1 shouldn't be using async I/O\n");
> > diff --git a/dlls/ws2_32/tests/sock.c b/dlls/ws2_32/tests/sock.c
> > index 88b62e3..7a7c688 100644
> > --- a/dlls/ws2_32/tests/sock.c
> > +++ b/dlls/ws2_32/tests/sock.c
> > @@ -2486,7 +2486,7 @@ static void test_ioctlsocket(void)
> >
> > ret = ioctlsocket(sock, SIOCATMARK, &arg);
> >     if(ret != SOCKET_ERROR)
> > -        todo_wine ok(arg, "expected a non-zero value\n");
> > +        ok(arg, "expected a non-zero value\n");
> >  }
> >
> > static int drain_pause=0;
> > --
> >
> > Best Regards, André Hentschel
>
> Hey André,
>
> I don't think we should rush this. The documentation says that the
> return values are basically opposites, but I think we need more tests
> with actual OOB data and different settings of SO_OOBINBLINE. MSDN has
> far too little information about SIOCATMARK and some things they say
> don't make a lot of sense.
>
> Mike.

Ok, i just wanted to give it a second try.
So ill hand it out to you as i have no idea about OOB data, i know what it is generally but thats all.
Thx!

-- 

Best Regards, André Hentschel