Re: SIGIO using UDP sockets just hangs (permantely blocks) socket I/O.

[Corinna Vinschen <corinna-cygwin@cygwin.com>]

[-- Attachment #1: Type: text/plain, Size: 1535 bytes --]

On Sep 30 11:50, tvonderhaar wrote:
>     Dear cygwin_list,
> 
>     I have tried to get signals working with UDP bound sockets. I have
> tried every combination and order of the included code, but once I
> ioctl() the socket with "FIOASYNC" I get one signal from the socket and
> the previously "FNDELAY" set socket blocks forever on a call to
> recvfrom(). I can detect the block with a select, but it never becomes
> ready. This code works flawlessly on Linux and in the past worked on
> SUN,HP- UX(I remove the HP specific code) and Alpha OSes. If I don't
> perform the ioctl() with "FIOASYNC" or perform a second call to ioctl()
> with the "set_state=0" the socket does not block as expected, but I of
> course I don't get my SIGIO signal. Note: I make the call to the
> function as follows:
> 
>    set_io_async(udp_socket_fd,call_back,SIGIO,FALSE);
>   /*FASLE defined as '0' and SIGIO as defined by CYGWIN.*/
> 
>  I only found one google search of an individual attempting non-blocking
> asynchronous() I/O and they indicated it worked using this code sequence
> under CYGWIN. "FASYNC" has no affect on CYGWIN using fcntl() as after

FASYNC/O_ASYNC is not supported at all by Cygwin and never was.
Signal-driven IO was never very high on the wishlist, so it wasn't fully
implemented.  What about using select from another thread instead?


Corinna

-- 
Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Maintainer                 cygwin AT cygwin DOT com
Red Hat

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 819 bytes --]