[Stgt-devel] ibmvio - rdma error

[FUJITA Tomonori]

From: Bastian Blank <waldi at berlios.de>
Subject: Re: [Stgt-devel] ibmvio - rdma error
Date: Fri, 6 Apr 2007 18:13:05 +0200

> On Sat, Apr 07, 2007 at 12:36:04AM +0900, FUJITA Tomonori wrote:
> > diff --git a/drivers/scsi/ibmvscsi/ibmvstgt.c b/drivers/scsi/ibmvscsi/ibmvstgt.c
> > index a7fbd10..46f9ca1 100644
> > --- a/drivers/scsi/ibmvscsi/ibmvstgt.c
> > +++ b/drivers/scsi/ibmvscsi/ibmvstgt.c
> > @@ -35,7 +35,7 @@
> >  #include "ibmvscsi.h"
> >  
> >  #define	INITIAL_SRP_LIMIT	16
> > -#define	DEFAULT_MAX_SECTORS	512
> > +#define	DEFAULT_MAX_SECTORS	128
> 
> 256, not 128.

Are you sure 256 should work though it seems to work? We know 128
works. If you definitely sure (e.g. you read the IBM document), I'm
happy to use 256.


> > -	if (sc->result != SAM_STAT_GOOD) {
> > +	if (err|| sc->result != SAM_STAT_GOOD) {
> 
> coding style?
> 
> > @@ -508,7 +509,6 @@ static void process_iu(struct viosrp_crq *crq, struct srp_target *target)
> >  
> >  	if (err != H_SUCCESS) {
> >  		eprintk("%ld transferring data error %p\n", err, iue);
> > -		done = 1;
> >  		goto out;
> >  	}
> 
> Replace that with:
>         if (err != H_SUCCESS) 
>                 eprintk("%ld transferring data error %p\n", err, iue);
>         else if (crq->format == VIOSRP_MAD_FORMAT)
>                 done = process_mad_iu(iue);
>         else
>                 done = process_srp_iu(iue);
> 
>         if (done)
>                 srp_iu_put(iue);
> 
> No more goto.

I think that using goto for an error path is common in Linux kernel
(and simple).


> Revised patch attached.
> 
> Signed-off-by: Bastian Blank <bastian at waldi.eu.org>

OK, thanks.