[ofw] [PATCH] Update libibmad and infiniband-diags

Hefty, Sean sean.hefty at intel.com
Fri Aug 19 10:04:58 PDT 2011


I committed the patch below to svn.

> -----Original Message-----
> From: Leonid Keller [mailto:leonid at mellanox.co.il]
> Sent: Thursday, August 18, 2011 1:11 AM
> To: Hefty, Sean; Uri Habusha; ofw at lists.openfabrics.org
> Cc: Ira Weiny (weiny2 at llnl.gov)
> Subject: RE: [PATCH] Update libibmad and infiniband-diags
> 
> OK with me
> 
> > -----Original Message-----
> > From: Hefty, Sean [mailto:sean.hefty at intel.com]
> > Sent: Monday, August 15, 2011 8:20 PM
> > To: Leonid Keller; Uri Habusha; ofw at lists.openfabrics.org
> > Cc: Ira Weiny (weiny2 at llnl.gov)
> > Subject: RE: [PATCH] Update libibmad and infiniband-diags
> >
> > > After applying your patch we got a build break on free version of Win8.
> > > It was a warning, something like "possible usage of uninitialized variable
> > > 'mad' " in rpc.c#213
> > > I made locally the following patch.
> > > What do you think?
> >
> > I think the compiler sounds brain dead.  :P
> >
> > > (I understand that as far as 'redirect' is coded to be 1, 'mad' is to be
> > > initialized. But compiler doesn't understand. :( )
> >
> > I think I would go a step further and remove the redirect variable completely.
> > If we replace the while loop with a do-while, I'm hoping that the
> > compiler can at least figure that out.  Can you see if the patch below works
> > for win8 (I don't have a build system for win8 myself)?  If it looks okay,
> > I'll submit this to Ira (who's copied).
> >
> > Fix build on win8
> >
> > From: Sean Hefty <sean.hefty at intel.com>
> >
> > The win8 compiler appears to be even more brain dead than its
> > predecessor.  It complains of an uninitialized variable mad in
> > rpc.c around line 213.  Fix this by removing the redirect
> > variable and using do - while instead.
> >
> > Signed-off-by: Sean Hefty <sean.hefty at intel.com>
> > ---
> >  src/rpc.c |    9 ++++-----
> >  1 files changed, 4 insertions(+), 5 deletions(-)
> >
> > diff --git a/src/rpc.c b/src/rpc.c
> > index d20c321..19a5a70 100644
> > --- a/src/rpc.c
> > +++ b/src/rpc.c
> > @@ -213,13 +213,12 @@ void *mad_rpc(const struct ibmad_port *port, ib_rpc_t *
> > rpc,
> >  {
> >  	int status, len;
> >  	uint8_t sndbuf[1024], rcvbuf[1024], *mad;
> > -	int redirect = 1;
> >  	ib_rpc_v1_t *rpcv1 = (ib_rpc_v1_t *)rpc;
> >  	int error = 0;
> >
> >  	if ((rpc->mgtclass & IB_MAD_RPC_VERSION_MASK) == IB_MAD_RPC_VERSION1)
> >  		rpcv1->error = 0;
> > -	while (redirect) {
> > +	do {
> >  		len = 0;
> >  		memset(sndbuf, 0, umad_size() + IB_MAD_SIZE);
> >
> > @@ -247,10 +246,10 @@ void *mad_rpc(const struct ibmad_port *port, ib_rpc_t *
> > rpc,
> >  			/* update dport for next request and retry */
> >  			/* bail if redirection fails */
> >  			if (redirect_port(dport, mad))
> > -				redirect = 0;
> > +				break;
> >  		} else
> > -			redirect = 0;
> > -	}
> > +			break;
> > +	} while (1);
> >
> >  	if ((rpc->mgtclass & IB_MAD_RPC_VERSION_MASK) == IB_MAD_RPC_VERSION1)
> >  		rpcv1->error = error;
> >




More information about the ofw mailing list