Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
References: <20200817163152.13468-1-liambeguin@gmail.com>
In-Reply-To: <20200817163152.13468-1-liambeguin@gmail.com>
From:   Geert Uytterhoeven <geert@linux-m68k.org>
Date:   Wed, 19 Aug 2020 15:57:34 +0200
Message-ID: <CAMuHMdUGu+pqZs5YoiBnZ2j7Yx86VDU9oKr4C15c1Jc9GjQuNA@mail.gmail.com>
Subject: Re: [PATCH v3 1/1] phy: tusb1210: use bitmasks to set VENDOR_SPECIFIC2
To:     Liam Beguin <liambeguin@gmail.com>
Cc:     Kishon Vijay Abraham I <kishon@ti.com>, Vinod <vkoul@kernel.org>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

Hi Liam,


On Mon, Aug 17, 2020 at 7:38 PM Liam Beguin <liambeguin@gmail.com> wrote:
> From: Liam Beguin <lvb@xiphos.com>
>
> Start by reading the content of the VENDOR_SPECIFIC2 register and update
> each bit field based on device properties when defined.
>
> The use of bit masks prevents fields from overriding each other and
> enables users to clear bits which are set by default, like datapolarity
> in this instance.
>
> Signed-off-by: Liam Beguin <lvb@xiphos.com>
> ---
> Changes since v1:
> - use set_mask_bits
>
> Changes since v2:
> - fix missing bit shift dropped in v2
> - rebase on 5.9-rc1
>
>  drivers/phy/ti/phy-tusb1210.c | 27 +++++++++++++++++----------
>  1 file changed, 17 insertions(+), 10 deletions(-)
>
> diff --git a/drivers/phy/ti/phy-tusb1210.c b/drivers/phy/ti/phy-tusb1210.c
> index d8d0cc11d187..358842b5790f 100644
> --- a/drivers/phy/ti/phy-tusb1210.c
> +++ b/drivers/phy/ti/phy-tusb1210.c
> @@ -14,8 +14,11 @@
>
>  #define TUSB1210_VENDOR_SPECIFIC2              0x80
>  #define TUSB1210_VENDOR_SPECIFIC2_IHSTX_SHIFT  0
> +#define TUSB1210_VENDOR_SPECIFIC2_IHSTX_MASK   GENMASK(3, 0)
>  #define TUSB1210_VENDOR_SPECIFIC2_ZHSDRV_SHIFT 4
> +#define TUSB1210_VENDOR_SPECIFIC2_ZHSDRV_MASK  GENMASK(5, 4)
>  #define TUSB1210_VENDOR_SPECIFIC2_DP_SHIFT     6
> +#define TUSB1210_VENDOR_SPECIFIC2_DP_MASK      BIT(6)
>
>  struct tusb1210 {
>         struct ulpi *ulpi;
> @@ -118,23 +121,27 @@ static int tusb1210_probe(struct ulpi *ulpi)
>          * diagram optimization and DP/DM swap.
>          */
>
> +       reg = ulpi_read(ulpi, TUSB1210_VENDOR_SPECIFIC2);
> +
>         /* High speed output drive strength configuration */
> -       device_property_read_u8(&ulpi->dev, "ihstx", &val);
> -       reg = val << TUSB1210_VENDOR_SPECIFIC2_IHSTX_SHIFT;
> +       if (!device_property_read_u8(&ulpi->dev, "ihstx", &val))
> +               reg = set_mask_bits(&reg, TUSB1210_VENDOR_SPECIFIC2_IHSTX_MASK,
> +                                   val << TUSB1210_VENDOR_SPECIFIC2_IHSTX_SHIFT);

Triggered by your patches to add support for cmpxchg on u8 pointers to
various architectures (which is a valuable goal in itself ;-), I decided
to have a look at the underlying problem you were facing.

IMHO, using set_mask_bits() is overkill here.  That helper is meant to
update individual bits in a variable that may be accessed concurrently,
hence its use of cmpxchg().

In this driver, you just want to modify a local variable, by clearing a
field, and setting some bits. No concurrency is involved.
Perhaps using FIELD_PREP() from include/linux/bitfield.h is more
appropriate?

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds