[PATCH v4 08/12] device core: Introduce multiple dma pfn offsets

Jim Quinlan james.quinlan at broadcom.com
Tue Jun 9 13:13:57 UTC 2020


Hi Andy,

On Tue, Jun 9, 2020 at 7:18 AM Andy Shevchenko
<andriy.shevchenko at linux.intel.com> wrote:
>
> On Mon, Jun 08, 2020 at 11:48:51AM -0400, Jim Quinlan wrote:
> > On Sun, Jun 7, 2020 at 12:500f9bfe0fb8840b268af1bbcc51f1cd440514e PM
> > Andy Shevchenko <andriy.shevchenko at linux.intel.com> wrote:
> > > On Fri, Jun 05, 2020 at 05:26:48PM -0400, Jim Quinlan wrote:
>
> ...
>
> > > > +     *map_size = (num_ranges + 1) * sizeof(**map);
> > > > +     r = kzalloc(*map_size, GFP_KERNEL);
> > >
> > > kcalloc()
> > Since I have to calculate the size anyway I thought kzalloc was fine.
> > I'll switch.
>
> The point is to check multiplication overflow. See overflow.h for helpers.

I am aware of this check and didn't think of it as applicable here, as
the most dma-ranges I can envision is six. I suppose that it is
possible that this may change in the future to some big number.  At
any rate, the next version has kcalloc().

Regards,
Jim
>
>
> > > > +     if (!r)
> > > > +             return -ENOMEM;
>
> --
> With Best Regards,
> Andy Shevchenko
>
>


More information about the devel mailing list