Re: apr_socket_accept() interrupted on Win32 corrupts master socket?

["William A. Rowe, Jr." <wr...@rowe-clan.net>]

Vaclav, yes- this is a 'windoze thingy' :-)

I'd be happy to look at it, and probably others will be interested in looking
at the code.  Thank you for a very clean test case!

Bill

At 09:22 AM 2/3/2003, Vaclav Ovsik wrote:
>Hi,
>I have written some very simple test server using apr (src included).
>
>I met a problem on Windoze NT (among other things):
>When the receive timeout occures on a call apr_socket_recv() in the
>serving thread, a call to apr_socket_accept() is also interrupted in the
>main thread. Next accept on the same socket ends on "socket operation on
>non-socket".
>
>The code output on Windoze NT4.0 (service pack 6) follows.
>I simply ran the threadsock.exe and created one connection using the
>netcat (from other host) and waited for timeout.
>
>   apr_app_initialize(): done
>   apr_pool_create(): done
>   apr_socket_create(): done
>   apr_sockaddr_info_get(): done
>   apr_socket_bind(): done
>   apr_socket_listen(): done
>   apr_socket_accept(): done
>   apr_thread_create(): done
>   apr_socket_accept(): apr_thread_detach(): done
>   apr_socket_timeout_set(): done
>   apr_socket_send(): done
>    - sent 13 bytes
>   socket recv: apr_socket_recv() failed: Connection timed out
>   failed: Interrupted system call
>   apr_socket_accept(): failed: Socket operation on non-socket
>
>String "failed: Interrupted system call" belongs to preceding
>"apr_socket_accept():" of course. :-)
>
>The listening master socket is probably away. This is apr or windoze
>related behavior? (On Linux i386 this works like I expected.)
>
>Is it possible to write simple server like this (performance doesn't
>matter) without need a special code for Win32?
>
>Excuse my English, please.
>Thanks
>-- 
>Zito

Re: apr_socket_accept() interrupted on Win32 corrupts master socket?

[Bill Stoddard <bi...@wstoddard.com>]

William A. Rowe, Jr. wrote:

>Vaclav, yes- this is a 'windoze thingy' :-)
>  
>
Hummm perhaps... However, if the listener is really whacked after the 
timeout, then I would think the problem is in either apr_socket_accept 
for windows or more likely a bug in the application.

Bill