* Re: [PATCH net 1/2] dt-bindings: net: dsa: realtek: cleanup compatible strings
2022-04-16 6:25 [PATCH net 1/2] dt-bindings: net: dsa: realtek: cleanup compatible strings Luiz Angelo Daros de Luca
@ 2022-04-16 6:33 ` Arınç ÜNAL
2022-04-16 10:57 ` Alvin Šipraga
` (2 subsequent siblings)
3 siblings, 0 replies; 7+ messages in thread
From: Arınç ÜNAL @ 2022-04-16 6:33 UTC (permalink / raw)
To: Luiz Angelo Daros de Luca, netdev
Cc: linus.walleij, alsi, andrew, vivien.didelot, f.fainelli, olteanv,
davem, kuba, pabeni, robh+dt, krzk+dt, devicetree
On 16/04/2022 09:25, Luiz Angelo Daros de Luca wrote:
> Compatible strings are used to help the driver find the chip ID/version
> register for each chip family. After that, the driver can setup the
> switch accordingly. Keep only the first supported model for each family
> as a compatible string and reference other chip models in the
> description.
>
> CC: devicetree@vger.kernel.org
> Link: https://lore.kernel.org/netdev/20220414014055.m4wbmr7tdz6hsa3m@bang-olufsen.dk/
> Signed-off-by: Luiz Angelo Daros de Luca <luizluca@gmail.com>
> ---
> .../devicetree/bindings/net/dsa/realtek.yaml | 33 +++++++------------
> 1 file changed, 12 insertions(+), 21 deletions(-)
>
> diff --git a/Documentation/devicetree/bindings/net/dsa/realtek.yaml b/Documentation/devicetree/bindings/net/dsa/realtek.yaml
> index 8756060895a8..9bf862abb496 100644
> --- a/Documentation/devicetree/bindings/net/dsa/realtek.yaml
> +++ b/Documentation/devicetree/bindings/net/dsa/realtek.yaml
> @@ -27,32 +27,23 @@ description:
> The realtek-mdio driver is an MDIO driver and it must be inserted inside
> an MDIO node.
>
> + The compatibility string is used only to find an identification register,
> + (chip ID and version) which is at a different MDIO base address in different
> + switch families. The driver then uses the chip ID/version to device how to
> + drive the switch.
I think you meant to say "decide how to drive the switch"?
> +
> properties:
> compatible:
> enum:
> - realtek,rtl8365mb
> - - realtek,rtl8366
> - realtek,rtl8366rb
> - - realtek,rtl8366s
> - - realtek,rtl8367
> - - realtek,rtl8367b
> - - realtek,rtl8367rb
> - - realtek,rtl8367s
> - - realtek,rtl8368s
> - - realtek,rtl8369
> - - realtek,rtl8370
> description: |
> - realtek,rtl8365mb: 4+1 ports
> - realtek,rtl8366: 5+1 ports
> - realtek,rtl8366rb: 5+1 ports
> - realtek,rtl8366s: 5+1 ports
> - realtek,rtl8367:
> - realtek,rtl8367b:
> - realtek,rtl8367rb: 5+2 ports
> - realtek,rtl8367s: 5+2 ports
> - realtek,rtl8368s: 8 ports
> - realtek,rtl8369: 8+1 ports
> - realtek,rtl8370: 8+2 ports
> + realtek,rtl8365mb:
> + Use with models RTL8363NB, RTL8363NB-VB, RTL8363SC, RTL8363SC-VB,
> + RTL8364NB, RTL8364NB-VB, RTL8365MB, RTL8366SC, RTL8367RB-VB, RTL8367S,
> + RTL8367SB, RTL8370MB, RTL8310SR
> + realtek,rtl8367rb:
> + Use with models RTL8366RB, RTL8366S
You kept rtl8366rb string but defined description for rtl8367rb.
Arınç
^ permalink raw reply [flat|nested] 7+ messages in thread* Re: [PATCH net 1/2] dt-bindings: net: dsa: realtek: cleanup compatible strings
2022-04-16 6:25 [PATCH net 1/2] dt-bindings: net: dsa: realtek: cleanup compatible strings Luiz Angelo Daros de Luca
2022-04-16 6:33 ` Arınç ÜNAL
@ 2022-04-16 10:57 ` Alvin Šipraga
2022-04-16 14:13 ` Andrew Lunn
2022-04-17 13:58 ` Linus Walleij
3 siblings, 0 replies; 7+ messages in thread
From: Alvin Šipraga @ 2022-04-16 10:57 UTC (permalink / raw)
To: Luiz Angelo Daros de Luca
Cc: netdev@vger.kernel.org, linus.walleij@linaro.org, andrew@lunn.ch,
vivien.didelot@gmail.com, f.fainelli@gmail.com, olteanv@gmail.com,
davem@davemloft.net, kuba@kernel.org, pabeni@redhat.com,
robh+dt@kernel.org, krzk+dt@kernel.org, arinc.unal@arinc9.com,
devicetree@vger.kernel.org
On Sat, Apr 16, 2022 at 03:25:03AM -0300, Luiz Angelo Daros de Luca wrote:
> Compatible strings are used to help the driver find the chip ID/version
> register for each chip family. After that, the driver can setup the
> switch accordingly. Keep only the first supported model for each family
> as a compatible string and reference other chip models in the
> description.
>
> CC: devicetree@vger.kernel.org
> Link: https://lore.kernel.org/netdev/20220414014055.m4wbmr7tdz6hsa3m@bang-olufsen.dk/
> Signed-off-by: Luiz Angelo Daros de Luca <luizluca@gmail.com>
> ---
> .../devicetree/bindings/net/dsa/realtek.yaml | 33 +++++++------------
> 1 file changed, 12 insertions(+), 21 deletions(-)
>
> diff --git a/Documentation/devicetree/bindings/net/dsa/realtek.yaml b/Documentation/devicetree/bindings/net/dsa/realtek.yaml
> index 8756060895a8..9bf862abb496 100644
> --- a/Documentation/devicetree/bindings/net/dsa/realtek.yaml
> +++ b/Documentation/devicetree/bindings/net/dsa/realtek.yaml
> @@ -27,32 +27,23 @@ description:
> The realtek-mdio driver is an MDIO driver and it must be inserted inside
> an MDIO node.
>
> + The compatibility string is used only to find an identification register,
s/compatibility/compatible/
> + (chip ID and version) which is at a different MDIO base address in different
> + switch families. The driver then uses the chip ID/version to device how to
> + drive the switch.
This applies to Marvell nicely but it is not accurate for Realtek, so
either make it less precise (don't mention MDIO base address), or be
more precise. Here is my suggestion:
| The compatible string is only used to identify which (silicon) family
| the switch belongs to. Roughly speaking, a family is any set of Realtek
| switches whose chip identification register(s) have a common location
| and semantics. The different models in a given family can be
| automatically disambiguated by parsing the chip identification
| register(s) according to the given family, obviating the need for a
| unique compatible string for each model.
This also makes it clear when and why a new compatible string should be
added.
> +
> properties:
> compatible:
> enum:
> - realtek,rtl8365mb
> - - realtek,rtl8366
> - realtek,rtl8366rb
> - - realtek,rtl8366s
> - - realtek,rtl8367
> - - realtek,rtl8367b
> - - realtek,rtl8367rb
> - - realtek,rtl8367s
> - - realtek,rtl8368s
> - - realtek,rtl8369
> - - realtek,rtl8370
> description: |
> - realtek,rtl8365mb: 4+1 ports
> - realtek,rtl8366: 5+1 ports
> - realtek,rtl8366rb: 5+1 ports
> - realtek,rtl8366s: 5+1 ports
> - realtek,rtl8367:
> - realtek,rtl8367b:
> - realtek,rtl8367rb: 5+2 ports
> - realtek,rtl8367s: 5+2 ports
> - realtek,rtl8368s: 8 ports
> - realtek,rtl8369: 8+1 ports
> - realtek,rtl8370: 8+2 ports
> + realtek,rtl8365mb:
> + Use with models RTL8363NB, RTL8363NB-VB, RTL8363SC, RTL8363SC-VB,
> + RTL8364NB, RTL8364NB-VB, RTL8365MB, RTL8366SC, RTL8367RB-VB, RTL8367S,
> + RTL8367SB, RTL8370MB, RTL8310SR
> + realtek,rtl8367rb:
s/67rb/66rb/?
> + Use with models RTL8366RB, RTL8366S
>
> mdc-gpios:
> description: GPIO line for the MDC clock line.
> @@ -335,7 +326,7 @@ examples:
> #size-cells = <0>;
>
> switch@29 {
> - compatible = "realtek,rtl8367s";
> + compatible = "realtek,rtl8365mb";
> reg = <29>;
>
> reset-gpios = <&gpio2 20 GPIO_ACTIVE_LOW>;
> --
> 2.35.1
>
^ permalink raw reply [flat|nested] 7+ messages in thread* Re: [PATCH net 1/2] dt-bindings: net: dsa: realtek: cleanup compatible strings
2022-04-16 6:25 [PATCH net 1/2] dt-bindings: net: dsa: realtek: cleanup compatible strings Luiz Angelo Daros de Luca
2022-04-16 6:33 ` Arınç ÜNAL
2022-04-16 10:57 ` Alvin Šipraga
@ 2022-04-16 14:13 ` Andrew Lunn
2022-04-17 13:58 ` Linus Walleij
3 siblings, 0 replies; 7+ messages in thread
From: Andrew Lunn @ 2022-04-16 14:13 UTC (permalink / raw)
To: Luiz Angelo Daros de Luca
Cc: netdev, linus.walleij, alsi, vivien.didelot, f.fainelli, olteanv,
davem, kuba, pabeni, robh+dt, krzk+dt, arinc.unal, devicetree
On Sat, Apr 16, 2022 at 03:25:03AM -0300, Luiz Angelo Daros de Luca wrote:
> Compatible strings are used to help the driver find the chip ID/version
> register for each chip family. After that, the driver can setup the
> switch accordingly. Keep only the first supported model for each family
> as a compatible string and reference other chip models in the
> description.
It would be good to mention here that these compatible strings have
never been used in a released kernel, so it is safe to remove them.
That is the sort of information which makes the job of reviewing
patches simpler. It is great to have the answers to questions you
cannot see directly from the code in the commit message.
Andrew
^ permalink raw reply [flat|nested] 7+ messages in thread* Re: [PATCH net 1/2] dt-bindings: net: dsa: realtek: cleanup compatible strings
2022-04-16 6:25 [PATCH net 1/2] dt-bindings: net: dsa: realtek: cleanup compatible strings Luiz Angelo Daros de Luca
` (2 preceding siblings ...)
2022-04-16 14:13 ` Andrew Lunn
@ 2022-04-17 13:58 ` Linus Walleij
2022-04-18 23:05 ` Luiz Angelo Daros de Luca
3 siblings, 1 reply; 7+ messages in thread
From: Linus Walleij @ 2022-04-17 13:58 UTC (permalink / raw)
To: Luiz Angelo Daros de Luca
Cc: netdev, ALSI, andrew, vivien.didelot, f.fainelli, olteanv, davem,
kuba, pabeni, robh+dt, krzk+dt, arinc.unal, devicetree
On Sat, Apr 16, 2022 at 8:25 AM Luiz Angelo Daros de Luca
<luizluca@gmail.com> wrote:
> Compatible strings are used to help the driver find the chip ID/version
> register for each chip family. After that, the driver can setup the
> switch accordingly. Keep only the first supported model for each family
> as a compatible string and reference other chip models in the
> description.
>
> CC: devicetree@vger.kernel.org
> Link: https://lore.kernel.org/netdev/20220414014055.m4wbmr7tdz6hsa3m@bang-olufsen.dk/
> Signed-off-by: Luiz Angelo Daros de Luca <luizluca@gmail.com>
OK, I suppose we know that Realtek has always maintained the
ID numbers in the hardware? Otherwise we will end up where
bindings/arm/primecell.yaml is: hardware ID numbers that were
supposed to be updated but weren't, so now both DT and the
kernel has to go through all kinds of loops and hoops to make it
work by encoding the number that should have been in the
hardware is instead in the device tree...
Yours,
Linus Walleij
^ permalink raw reply [flat|nested] 7+ messages in thread* Re: [PATCH net 1/2] dt-bindings: net: dsa: realtek: cleanup compatible strings
2022-04-17 13:58 ` Linus Walleij
@ 2022-04-18 23:05 ` Luiz Angelo Daros de Luca
2022-04-18 23:36 ` Luiz Angelo Daros de Luca
0 siblings, 1 reply; 7+ messages in thread
From: Luiz Angelo Daros de Luca @ 2022-04-18 23:05 UTC (permalink / raw)
To: Linus Walleij
Cc: open list:NETWORKING DRIVERS, Alvin Šipraga, Andrew Lunn,
Vivien Didelot, Florian Fainelli, Vladimir Oltean,
David S. Miller, Jakub Kicinski, Paolo Abeni, Rob Herring,
krzk+dt, Arınç ÜNAL, devicetree
> On Sat, Apr 16, 2022 at 8:25 AM Luiz Angelo Daros de Luca
> <luizluca@gmail.com> wrote:
>
> > Compatible strings are used to help the driver find the chip ID/version
> > register for each chip family. After that, the driver can setup the
> > switch accordingly. Keep only the first supported model for each family
> > as a compatible string and reference other chip models in the
> > description.
> >
> > CC: devicetree@vger.kernel.org
> > Link: https://lore.kernel.org/netdev/20220414014055.m4wbmr7tdz6hsa3m@bang-olufsen.dk/
> > Signed-off-by: Luiz Angelo Daros de Luca <luizluca@gmail.com>
>
> OK, I suppose we know that Realtek has always maintained the
> ID numbers in the hardware? Otherwise we will end up where
> bindings/arm/primecell.yaml is: hardware ID numbers that were
> supposed to be updated but weren't, so now both DT and the
> kernel has to go through all kinds of loops and hoops to make it
> work by encoding the number that should have been in the
> hardware is instead in the device tree...
Thanks, Linus. The rtl8367c driver seems to depend on information
retrieved from registers, mainly chip id/ver. If they forget to update
a chip id/version, it might be the case that it does not really matter
from the driver's point of view.
Anyway, if deemed to be necessary, adding a compatible string is much
easier than removing one after a kernel is released.
Regards,
Luiz
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH net 1/2] dt-bindings: net: dsa: realtek: cleanup compatible strings
2022-04-18 23:05 ` Luiz Angelo Daros de Luca
@ 2022-04-18 23:36 ` Luiz Angelo Daros de Luca
0 siblings, 0 replies; 7+ messages in thread
From: Luiz Angelo Daros de Luca @ 2022-04-18 23:36 UTC (permalink / raw)
To: Linus Walleij
Cc: open list:NETWORKING DRIVERS, Alvin Šipraga, Andrew Lunn,
Vivien Didelot, Florian Fainelli, Vladimir Oltean,
David S. Miller, Jakub Kicinski, Paolo Abeni, Rob Herring,
krzk+dt, Arınç ÜNAL, devicetree
Thanks Arınç, Alvin, Andrew and Linus for the reviews. I just sent v2.
Regards,
Luiz
^ permalink raw reply [flat|nested] 7+ messages in thread