Open MPI logo

Open MPI Development Mailing List Archives

  |   Home   |   Support   |   FAQ   |   all Development mailing list

From: Tim S. Woodall (twoodall_at_[hidden])
Date: 2006-01-05 09:38:47


Yo Jeff,

jsquyres_at_[hidden] wrote:
> Author: jsquyres
> Date: 2006-01-04 17:29:09 -0500 (Wed, 04 Jan 2006)
> New Revision: 8643
>
> Modified:
> trunk/orte/mca/oob/tcp/oob_tcp.c
> Log:
> Avoid a false positive in bcheck
>
>
> Modified: trunk/orte/mca/oob/tcp/oob_tcp.c
> ===================================================================
> --- trunk/orte/mca/oob/tcp/oob_tcp.c 2006-01-04 22:18:30 UTC (rev 8642)
> +++ trunk/orte/mca/oob/tcp/oob_tcp.c 2006-01-04 22:29:09 UTC (rev 8643)
> @@ -463,7 +463,12 @@
> }
> OBJ_RELEASE(event);
>
> + /* Some mem checkers don't realize that hdr will guarantee to be
> + fully filled in during the read(), below :-( */
> + OMPI_DEBUG_ZERO(hdr);
> +
> /* recv the process identifier */
> + hdr.msg_src.cellid = 78;

Do we really want the above line?

> while((rc = recv(sd, (char *)&hdr, sizeof(hdr), 0)) != sizeof(hdr)) {
> if(rc >= 0) {
> if(mca_oob_tcp_component.tcp_debug > 1) {
> @@ -480,6 +485,10 @@
> return;
> }
> }
> + assert(rc == sizeof(hdr));
> + if (rc != sizeof(hdr)) {
> + abort();
> + }
> MCA_OOB_TCP_HDR_NTOH(&hdr);
>
> /* dispatch based on message type */
>
> _______________________________________________
> svn mailing list
> svn_at_[hidden]
> http://www.open-mpi.org/mailman/listinfo.cgi/svn
>