Re: [RFC net-next 2/2] net: dsa: add Realtek RTL8366S switch driver

From: Linus Walleij <linus.walleij@linaro.org>
To: DENG Qingfang <dqfext@gmail.com>
Cc: Andrew Lunn <andrew@lunn.ch>,
	Vivien Didelot <vivien.didelot@gmail.com>,
	Florian Fainelli <f.fainelli@gmail.com>,
	Vladimir Oltean <olteanv@gmail.com>,
	"David S. Miller" <davem@davemloft.net>,
	Jakub Kicinski <kuba@kernel.org>, netdev <netdev@vger.kernel.org>,
	Mauri Sandberg <sandberg@mailfence.com>
Subject: Re: [RFC net-next 2/2] net: dsa: add Realtek RTL8366S switch driver
Date: Wed, 17 Feb 2021 12:12:17 +0100	[thread overview]
Message-ID: <CACRpkdZEQYahteQ3GdftkS82O2rz_ZZ88AUN0HGMhNQDHaFWRw@mail.gmail.com> (raw)
In-Reply-To: <20210217062139.7893-3-dqfext@gmail.com>

Hi Qingfang,

thanks for your patch! Overall it looks good and I will not
nitpick details since this is RFC.

On Wed, Feb 17, 2021 at 7:21 AM DENG Qingfang <dqfext@gmail.com> wrote:
>
> Support Realtek RTL8366S/SR switch
>
> Signed-off-by: DENG Qingfang <dqfext@gmail.com>

I would mention that the DT bindings for the switch are already in
Documentation/devicetree/bindings/net/dsa/realtek-smi.txt
(Hmmm we should convert these bindings to YAML too.)

>         select NET_DSA_TAG_RTL4_A
> +       select NET_DSA_TAG_RTL8366S

Hopefully we can use NET_DSA_TAG_RTL4_A for this
switch as mentioned.

> +/* bits 0..7 = port 0, bits 8..15 = port 1 */
> +#define RTL8366S_PAACR0                        0x0011
> +/* bits 0..7 = port 2, bits 8..15 = port 3 */
> +#define RTL8366S_PAACR1                        0x0012
> +/* bits 0..7 = port 4, bits 8..15 = port 5 */
> +#define RTL8366S_PAACR2                        0x0013

Is this all? I was under the impression that RTL8366S
supports up to 8 ports (7+1).

> +#define RTL8366S_CHIP_ID_REG                   0x0105
> +#define RTL8366S_CHIP_ID_8366                  0x8366

Curiously RTL8366RB presents ID 0x5937. Oh well. Interesting
engineering process I suppose.

> +/* LED control registers */
> +#define RTL8366S_LED_BLINKRATE_REG             0x0420
> +#define RTL8366S_LED_BLINKRATE_BIT             0
> +#define RTL8366S_LED_BLINKRATE_MASK            0x0007
> +
> +#define RTL8366S_LED_CTRL_REG                  0x0421
> +#define RTL8366S_LED_0_1_CTRL_REG              0x0422
> +#define RTL8366S_LED_2_3_CTRL_REG              0x0423

Again I wonder if there are more registers here for 8 ports?

> +/* PHY registers control */
> +#define RTL8366S_PHY_ACCESS_CTRL_REG           0x8028
> +#define RTL8366S_PHY_ACCESS_DATA_REG           0x8029
> +
> +#define RTL8366S_PHY_CTRL_READ                 1
> +#define RTL8366S_PHY_CTRL_WRITE                        0
> +
> +#define RTL8366S_PORT_NUM_CPU          5
> +#define RTL8366S_NUM_PORTS             6

Hm 5+1? Isn't RTL8366S 7+1?

> +#define RTL8366S_PORT_1                        BIT(0) /* In userspace port 0 */
> +#define RTL8366S_PORT_2                        BIT(1) /* In userspace port 1 */
> +#define RTL8366S_PORT_3                        BIT(2) /* In userspace port 2 */
> +#define RTL8366S_PORT_4                        BIT(3) /* In userspace port 3 */
> +#define RTL8366S_PORT_5                        BIT(4) /* In userspace port 4 */
> +#define RTL8366S_PORT_CPU              BIT(5) /* CPU port */

Same here.

Overall the question about whether the switch is 5+1 or 7+1 is my
big design remark.

Maybe it is only 5+1 who knows...

Yours,
Linus Walleij