Re: [PATCH 2/5] iio: adc: xilinx: use devm_krealloc() instead of kfree() + kcalloc()
From: Bartosz Golaszewski
Date: Tue Oct 27 2020 - 06:41:28 EST
On Tue, Oct 27, 2020 at 11:29 AM Andy Shevchenko
<andy.shevchenko@xxxxxxxxx> wrote:
>
> On Tue, Oct 27, 2020 at 12:04 PM Bartosz Golaszewski <brgl@xxxxxxxx> wrote:
> > On Tue, Oct 27, 2020 at 10:33 AM Andy Shevchenko
> > <andy.shevchenko@xxxxxxxxx> wrote:
> > >
> > > On Mon, Oct 26, 2020 at 4:03 PM Bartosz Golaszewski <brgl@xxxxxxxx> wrote:
> > > >
> > > > From: Bartosz Golaszewski <bgolaszewski@xxxxxxxxxxxx>
> > > >
> > > > We now have devm_krealloc() in the kernel Use it indstead of calling
> > > > kfree() and kcalloc() separately.
> > >
> > > Which is completely lawful when size > previous_size (I mean, the
> > > additional patch you sent previously seems not related to this).
> > >
> >
> > Sure but devm_krealloc() is cleaner and adds the benefit of resource management.
>
> I meant devm_krealloc(). It should work in this case without your
> additional "fix" patch.
>
I know, this is why I sent the fix separately. The fix is still
correct on its own.
> > > > - kfree(xadc->data);
> > > > - xadc->data = kcalloc(n, sizeof(*xadc->data), GFP_KERNEL);
> > > > + xadc->data = devm_krealloc(indio_dev->dev.parent, xadc->data,
> > > > + n * sizeof(*xadc->data),
> > >
> > > I think you need to use something from overflow.h instead of explicit
> > > multiplication here.
> > >
> >
> > Or maybe add devm_krealloc_array() which would perform the checks
> > behind the scenes?
>
> Maybe. But what to do in the cases when you have struct with flexible
> arrays, like
> struct foo {
> ...
> type bar[];
> };
>
> ?
Just use regular devm_krealloc() with struct_size()?
>
> And you do kzalloc(sizeof(foo)) followed by krealloc(). The above name
> (krealloc_array) may be a bit ambiguous.
But devm_krealloc_array() would only be useful for memory allocated by
kmalloc_array() or kcalloc(). I don't see what's your point.
Bartosz