BCM63268 requires special RGMII configuration to work.
Signed-off-by: Álvaro Fernández Rojas <[email protected]>
---
drivers/net/dsa/b53/b53_common.c | 6 +++++-
drivers/net/dsa/b53/b53_regs.h | 1 +
2 files changed, 6 insertions(+), 1 deletion(-)
diff --git a/drivers/net/dsa/b53/b53_common.c b/drivers/net/dsa/b53/b53_common.c
index 6e212f6f1cb9..d0a22c8a55c9 100644
--- a/drivers/net/dsa/b53/b53_common.c
+++ b/drivers/net/dsa/b53/b53_common.c
@@ -1240,8 +1240,12 @@ static void b53_adjust_63xx_rgmii(struct dsa_switch *ds, int port,
break;
}
- if (port != dev->imp_port)
+ if (port != dev->imp_port) {
+ if (is63268(dev))
+ rgmii_ctrl |= RGMII_CTRL_MII_OVERRIDE;
+
rgmii_ctrl |= RGMII_CTRL_ENABLE_GMII;
+ }
b53_write8(dev, B53_CTRL_PAGE, off, rgmii_ctrl);
diff --git a/drivers/net/dsa/b53/b53_regs.h b/drivers/net/dsa/b53/b53_regs.h
index b2c539a42154..bfbcb66bef66 100644
--- a/drivers/net/dsa/b53/b53_regs.h
+++ b/drivers/net/dsa/b53/b53_regs.h
@@ -138,6 +138,7 @@
#define B53_RGMII_CTRL_IMP 0x60
#define RGMII_CTRL_ENABLE_GMII BIT(7)
+#define RGMII_CTRL_MII_OVERRIDE BIT(6)
#define RGMII_CTRL_TIMING_SEL BIT(2)
#define RGMII_CTRL_DLL_RXC BIT(1)
#define RGMII_CTRL_DLL_TXC BIT(0)
--
2.30.2
On 3/20/23 08:50, Álvaro Fernández Rojas wrote:
> BCM63268 requires special RGMII configuration to work.
>
> Signed-off-by: Álvaro Fernández Rojas <[email protected]>
> ---
> drivers/net/dsa/b53/b53_common.c | 6 +++++-
> drivers/net/dsa/b53/b53_regs.h | 1 +
> 2 files changed, 6 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/net/dsa/b53/b53_common.c b/drivers/net/dsa/b53/b53_common.c
> index 6e212f6f1cb9..d0a22c8a55c9 100644
> --- a/drivers/net/dsa/b53/b53_common.c
> +++ b/drivers/net/dsa/b53/b53_common.c
> @@ -1240,8 +1240,12 @@ static void b53_adjust_63xx_rgmii(struct dsa_switch *ds, int port,
> break;
> }
>
> - if (port != dev->imp_port)
> + if (port != dev->imp_port) {
> + if (is63268(dev))
> + rgmii_ctrl |= RGMII_CTRL_MII_OVERRIDE;
AFAICT the override bit is defined and valid for both 63268 and 6318,
essentially whenever more than one RGMII control register for port 4,
but also for other ports, it seems like the bit becomes valid. The
comment I have says that the override bit ensures that what is populated
in bits 5:4 which is the actual RGMII interface mode is applied. That
mode can be one of:
0b00: RGMII mode
0b01: MII mode
0b10: RVMII mode
0b11: GMII mode
even though this is not documented as such, I suspect that the override
bit does not only set the mode, but also ensures that the delays are
also applied.
Once you update patch 3, this LGTM and you may add:
Reviewed-by: Florian Fainelli <[email protected]>
For your v2. Thanks!
--
Florian
On Mon, Mar 20, 2023 at 04:50:24PM +0100, Álvaro Fernández Rojas wrote:
> BCM63268 requires special RGMII configuration to work.
>
> Signed-off-by: Álvaro Fernández Rojas <[email protected]>
Reviewed-by: Simon Horman <[email protected]>
(Excuse me for my previous email in HTML, but I forgot Android Gmail
app uses HTML)
2023-03-20 21:00 GMT+01:00, Florian Fainelli <[email protected]>:
> On 3/20/23 08:50, Álvaro Fernández Rojas wrote:
>> BCM63268 requires special RGMII configuration to work.
>>
>> Signed-off-by: Álvaro Fernández Rojas <[email protected]>
>> ---
>> drivers/net/dsa/b53/b53_common.c | 6 +++++-
>> drivers/net/dsa/b53/b53_regs.h | 1 +
>> 2 files changed, 6 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/net/dsa/b53/b53_common.c
>> b/drivers/net/dsa/b53/b53_common.c
>> index 6e212f6f1cb9..d0a22c8a55c9 100644
>> --- a/drivers/net/dsa/b53/b53_common.c
>> +++ b/drivers/net/dsa/b53/b53_common.c
>> @@ -1240,8 +1240,12 @@ static void b53_adjust_63xx_rgmii(struct dsa_switch
>> *ds, int port,
>> break;
>> }
>>
>> - if (port != dev->imp_port)
>> + if (port != dev->imp_port) {
>> + if (is63268(dev))
>> + rgmii_ctrl |= RGMII_CTRL_MII_OVERRIDE;
>
> AFAICT the override bit is defined and valid for both 63268 and 6318,
Should we add a specific ID for the 6318?
I don't know if it's different enough from the 63268 to need a special
treatment such as is6318()...
> essentially whenever more than one RGMII control register for port 4,
> but also for other ports, it seems like the bit becomes valid. The
> comment I have says that the override bit ensures that what is populated
> in bits 5:4 which is the actual RGMII interface mode is applied. That
> mode can be one of:
>
> 0b00: RGMII mode
> 0b01: MII mode
> 0b10: RVMII mode
> 0b11: GMII mode
This is interesting since we never set those bits in the bcm63xx
enetsw controller...
Should we add configuration for those bits in future patches?
Which SoCs hace those bits? Only 6318 and 63268 or every 63xx has them?
>
> even though this is not documented as such, I suspect that the override
> bit does not only set the mode, but also ensures that the delays are
> also applied.
>
> Once you update patch 3, this LGTM and you may add:
>
> Reviewed-by: Florian Fainelli <[email protected]>
>
> For your v2. Thanks!
> --
> Florian
>
>
--
Álvaro