Re: [PATCH v6 3/6] interconnect: icc-clk: Add devm_icc_clk_register
From: Dmitry Baryshkov
Date: Tue Apr 02 2024 - 07:17:28 EST
On Tue, 2 Apr 2024 at 14:02, Varadarajan Narayanan
<quic_varada@xxxxxxxxxxx> wrote:
>
> On Tue, Apr 02, 2024 at 01:48:08PM +0300, Dmitry Baryshkov wrote:
> > On Tue, 2 Apr 2024 at 13:40, Dmitry Baryshkov
> > <dmitry.baryshkov@xxxxxxxxxx> wrote:
> > >
> > > On Tue, 2 Apr 2024 at 13:34, Varadarajan Narayanan
> > > <quic_varada@xxxxxxxxxxx> wrote:
> > > >
> > > > Wrap icc_clk_register to create devm_icc_clk_register to be
> > > > able to release the resources properly.
> > > >
> > > > Signed-off-by: Varadarajan Narayanan <quic_varada@xxxxxxxxxxx>
> > > > ---
> > > > v5: Introduced devm_icc_clk_register
> > > > ---
> > > > drivers/interconnect/icc-clk.c | 29 +++++++++++++++++++++++++++++
> > > > include/linux/interconnect-clk.h | 4 ++++
> > > > 2 files changed, 33 insertions(+)
> > >
> > > Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@xxxxxxxxxx>
> >
> > Wait. Actually,
> >
> > Unreviewed-by: me
> >
> > Please return int from devm_icc_clk_register instead of returning the pointer.
>
> Wouldn't returning int break the general assumption that
> devm_foo(), returns the same type as foo(). For example
> devm_clk_hw_get_clk and clk_hw_get_clk return struct clk *?
Not always. The only reason to return icc_provider was to make it
possible to destroy it. With devres-managed function you don't have to
do anything.
--
With best wishes
Dmitry