Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.34 as permitted sender) client-ip=23.128.96.34;
MIME-Version: 1.0
References: <20231120222832.4063882-1-masahiroy@kernel.org>
 <20231120222832.4063882-2-masahiroy@kernel.org> <CAK7LNATK3BVDZ88v7MWjzOk3Kv8CR4Tz_k77yi1OO=zc+71j=g@mail.gmail.com>
In-Reply-To: <CAK7LNATK3BVDZ88v7MWjzOk3Kv8CR4Tz_k77yi1OO=zc+71j=g@mail.gmail.com>
From:   Linus Walleij <linus.walleij@linaro.org>
Date:   Fri, 24 Nov 2023 11:06:50 +0100
Message-ID: <CACRpkdaMBrWUR8OYLPiQNNJ64mfenFyQa9f4iB+w5kZ44WjSeQ@mail.gmail.com>
Subject: Re: [PATCH 1/2] pinctrl: pinconf-generic: resize the pin config array directly
To:     Masahiro Yamada <masahiroy@kernel.org>,
        Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Cc:     Bartosz Golaszewski <brgl@bgdev.pl>, linux-gpio@vger.kernel.org,
        linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

Hi Masahiro,

On Tue, Nov 21, 2023 at 11:21=E2=80=AFAM Masahiro Yamada <masahiroy@kernel.=
org> wrote:
> On Tue, Nov 21, 2023 at 7:28=E2=80=AFAM Masahiro Yamada <masahiroy@kernel=
.org> wrote:
> >
> > pinconf_generic_parse_dt_config() allocates memory that is large enough
> > to contain all the config parameters. Then, kmemdup() copies the found
> > configs to the memory with the exact size.
> >
> > There is no need to allocate memory twice; you can directly resize the
> > initial memory using krealloc_array().
> >
> > I also changed kcalloc() to kmalloc_array() to keep the consistency wit=
h
> > krealloc_array(). This change has no impact because you do not need to
> > zero out the 'cfg' array.
> >
> > Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
>
> Sorry, I retract this patch set.
>
> krealloc() does not save any memory
> when the new_size is smaller than the current size.

But the first part where you switch to kmalloc_array() is still a nice chan=
ge.

The fact that we use kmemdup to be able to also shrink the allocation is a
bit of an oddity I guess, but let's run this patch by Andy Shevchenko, and
ask what he thinks about simply introducing kmemdup_array() or if he
has other ideas for this.

Yours,
Linus Walleij