Re: tcpip shared port problem
From: Nom de Plume (soccer13player_at_yahoo.com)
Date: 02/24/04
- Next message: Robert Klute: "Re: It is almost certain now, INTEL will have 64bit x86 !!"
- Previous message: Didier Morandi: "VAX DECWindows Motif kit requested online"
- In reply to: Jeffrey Coffield: "tcpip shared port problem"
- Next in thread: Jeffrey Coffield: "Re: tcpip shared port problem"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: 24 Feb 2004 12:06:19 -0800
Jeffrey Coffield <jeffrey@digitalsynergyinc.com> wrote in message news:<cPB_b.16582$4U5.3606@newssvr29.news.prodigy.com>...
> OpenVMS 7.1, TCP/IP V5.0A
>
> I have been trying to use the UCX$C_SHARE option to have parent process
> accept a connection and then pass it off to a child process. I based my
> code on the QIO example from DSIN of PARENT.C and CHILD.C.
>
>
> Problem 1:
> When a connection exits, I immediately get another connection that
> returns a status of 1 but the IOSB(0) returns 1 and IOSB(1) returns 4.
> This appears to mean EINTR (Interrupted system call). This occurs in
> both my code and the example from DSIN. The DSIN example crashes at this
> point as the spawned child tries to read from the bad connection.
>
> Problem 2:
> I have the parent spawn 2 children to start. The first and second
> connections work fine. I then disconnect each connection, make the first
> connections again and so far so good. However when I make the second
> connection the second time I get a IOSB(0) of 20 (SYSTEM-F-BARPARAM, bad
> parameter value).
>
> Does anyone have an example in any language of the sequence of calls to
> share a socket between a parent and child?
>
> TIA
> Jeff Coffield
> Please respond in this newsgroup.
My guess is that since you are on the TCPIP stack the numeric value
represented by UCX$C_SHARE is incorrect.
Try TCPIP$C_SHARE symbol from the tcpip$inetdef.h file. Also, I'm
told (but have not yet validated) that you can use the TCPIP symbol in
the setsockopt instead of having to use the QIO call.
JMOD
- Next message: Robert Klute: "Re: It is almost certain now, INTEL will have 64bit x86 !!"
- Previous message: Didier Morandi: "VAX DECWindows Motif kit requested online"
- In reply to: Jeffrey Coffield: "tcpip shared port problem"
- Next in thread: Jeffrey Coffield: "Re: tcpip shared port problem"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|