James Bottomley <James.Bottomley@xxxxxxxxxxxx> wrote:
>
> On Sun, 2005-10-16 at 14:08 -0700, Andrew Morton wrote:
> > drivers/scsi/ips.c: In function `ips_scmd_buf_write':
> > drivers/scsi/ips.c:3665: warning: null argument where non-null required (arg 1)
> > drivers/scsi/ips.c: In function `ips_scmd_buf_read':
> > drivers/scsi/ips.c:3697: warning: null argument where non-null required (arg 2)
> > drivers/scsi/ips.c: In function `ips_register_scsi':
> >
> > Due to
> >
> > memcpy(IPS_SG_ADDRESS(&sg[i]), &cdata[xfer_cnt],
> > min_cnt);
> >
> > I guess the compiler is saying that if IPS_SG_ADDRESS indeed evaluates to
> > NULL (as it is designed to do), we have an oops.
>
> That's weird ... the compiler can't possibly be in a position to make
> that judgement call. We have lots of places where we return null if
> something goes wrong and the kernel oopses. The compiler certainly
> isn't warning about all of them.
Well I guess the difference is that the compiler can _see_ that
IPS_SG_ADDRESS() might return NULL:
#define IPS_SG_ADDRESS(sg) (page_address((sg)->page) ? \
page_address((sg)->page)+(sg)->offset : NULL)
What's the point in this expression anyway? Why not just assume that
page_address() returns non-NULL?
-
: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
