Re: [PATCH v5 2/7] dt-bindings: net: amlogic,meson-dwmac: add amlogic,t7-dwmac
From: Chuan Liu
Date: Mon Nov 16 2026 - 07:03:27 EST
Hi Lucas,
On 11/13/2026 7:40 PM, Lucas Tanure wrote:
[ EXTERNAL EMAIL ]
On Tue, Nov 10, 2026 at 11:43 AM Chuan Liu <chuan.liu@xxxxxxxxxxx> wrote:
Hi Lucas,
On 11/9/2026 11:43 PM, Lucas Tanure wrote:
[ EXTERNAL EMAIL ]
On Fri, Nov 6, 2026 at 3:16 AM Chuan Liu <chuan.liu@xxxxxxxxxxx> wrote:
Hi Krzysztof, Lucas,
On 11/5/2026 7:57 PM, Krzysztof Kozlowski wrote:
[ EXTERNAL EMAIL ]
On Tue, Nov 03, 2026 at 04:37:10PM +0000, Lucas Tanure wrote:
Assisted-by: LLM
Drop this one.
Signed-off-by: Lucas Tanure <tanure@xxxxxxxxx>
---
.../bindings/net/amlogic,meson-dwmac.yaml | 59 +++++++++++++++++++
1 file changed, 59 insertions(+)
diff --git a/Documentation/devicetree/bindings/net/amlogic,meson-dwmac.yaml b/Documentation/devicetree/bindings/net/amlogic,meson-dwmac.yaml
index e769c9bb70d8..6de013e3e8f6 100644
--- a/Documentation/devicetree/bindings/net/amlogic,meson-dwmac.yaml
+++ b/Documentation/devicetree/bindings/net/amlogic,meson-dwmac.yaml
@@ -23,6 +23,7 @@ select:
- amlogic,meson-gxbb-dwmac
- amlogic,meson-axg-dwmac
- amlogic,meson-g12a-dwmac
+ - amlogic,t7-dwmac
required:
- compatible
@@ -57,6 +58,59 @@ allOf:
- const: clkin1
- const: timing-adjustment
+ - if:
+ properties:
+ compatible:
+ contains:
+ const: amlogic,t7-dwmac
+ then:
+ properties:
+ clocks:
+ items:
+ - description: GMAC main clock
+ - description: First parent clock of the internal mux
+ - description: Second parent clock of the internal mux
+ - description:
+ clock of the pipeline stage inserted in the bus path between
+ the controller and the DRAM. Without it, the controller cannot
+ complete DMA transfers.
+ - description:
+ clock of the register block holding PRG_ETH0 and PRG_ETH1.
+ Without it, the block reads as zero and ignores writes.
This sounds like you take clock from other block. That's wrong, that
device should take its own clock.
I see that the clock in the dts patch corresponds to <&clkc_periphs
CLKID_SYS_ETHPHY>. This clock provides the working clock for the APB Bus
of the ETH PHY controller, and this clock will affect the register
reading and writing of the ETH PHY (including the ETH_PLL). See section
15.3.4.2 RegisterDescription in A311D2 Datasheet for details.
Thanks. That section is the PHY block at 0xfe028000, the region of the
mdio multiplexer node. The registers in this description are PRG_ETH0
and PRG_ETH1 in eth_top at 0xfe024000. On the VIM4 the same gate feeds
them too: with it off they read as zero and ignore writes. My datasheet
revisions do not say which clock feeds eth_top. Can you confirm it is
this one?
The information I can confirm is:
- The base address of eth_phy is 0xfe028000, and its corresponding
APB clock is <&clkc_periphs CLKID_SYS_ETHPHY>.
- The base address of eth_top is 0xfe024000, and the base address of
eth_mac is 0xfdc00000; their corresponding APB clock is <&clkc_periphs
CLKID_SYS_ETH>.
That is not what the VIM4 does. The stmmac core enables
CLKID_SYS_ETH before any glue code runs. With that clock on and
CLKID_SYS_ETHPHY off, PRG_ETH0 and PRG_ETH1 read as zero and the
driver's writes are lost: U-Boot's 0xa0000 was still in PRG_ETH1
after the driver had written 0. Enabling CLKID_SYS_ETHPHY right
before the write makes it land. That was the v3 bug, every frame a
CRC error. I repeated it today on the running kernel, toggling only
bit 4 of CLKCTRL_SYS_CLK_EN0_REG0 from user space, every other gate
left on:
both gates on EN0=ffffffff EN1=ffffffff PRG_ETH0=00001601
ethphy off, eth on EN0=ffffffef EN1=ffffffff PRG_ETH0=00000000
ethphy restored EN0=ffffffff EN1=ffffffff PRG_ETH0=00001601
Sorry that the information I gave before is not completely accurate. The conclusion you drew from the test is correct. The relationship between them should be:
- eth_mac: works with CLKID_SYS_ETH only
- eth_phy: works with CLKID_SYS_ETHPHY only
- eth_top: needs both CLKID_SYS_ETH and CLKID_SYS_ETHPHY
So on this chip CLKID_SYS_ETH alone does not give access to eth_top.
Could the register interface of eth_top sit behind the eth_phy APB
slave?
eth_top is the glue between the eth_mac and eth_phy domains, its registers are consumed on both sides of that boundary, so a register access has to complete in both clock domains - with either gate off, reads return zero and writes are lost.
Either way the MAC node has to enable CLKID_SYS_ETHPHY, and
the binding will describe it as the clock of the glue registers.
Thanks,
The CPU reads and writes the registers of these peripheral controllers
(or Processing Units) via the APB. Consequently, the registers of these
controllers cannot be read and written properly if the corresponding APB
clock (sys_clk/clk81) is not enabled.
Thanks,
Lucas
+
+ clock-names:
+ items:
+ - const: stmmaceth
+ - const: clkin0
+ - const: clkin1
+ - const: pipeline
+ - const: pclk
+
+ # The PHY adds the 2ns RGMII delays, see ethernet-controller.yaml.
+ # The RX delay line of this glue is fine tuning in 200ps steps,
+ # added on top of what phy-mode describes. The TX side has no fine
+ # tuning: its only step is a quarter of the clock period, the full
+ # 2ns at 1Gbit/s, so the TX property is not accepted.
+ tx-internal-delay-ps: false
+
+ # Deprecated: new SoCs use the standard *-internal-delay-ps
+ # properties instead.
+ amlogic,tx-delay-ns: false
+ amlogic,rx-delay-ns: false
+
+ # Delay definitions for Meson8b and newer
+ - if:
+ properties:
+ compatible:
+ contains:
+ enum:
+ - amlogic,meson8b-dwmac
+ - amlogic,meson8m2-dwmac
+ - amlogic,meson-gxbb-dwmac
+ - amlogic,meson-axg-dwmac
+ - amlogic,meson-g12a-dwmac
+ then:
+ properties:
amlogic,tx-delay-ns:
enum: [0, 2, 4, 6]
default: 2
@@ -106,6 +160,7 @@ allOf:
contains:
enum:
- amlogic,meson-g12a-dwmac
+ - amlogic,t7-dwmac
then:
properties:
rx-internal-delay-ps:
@@ -142,6 +197,10 @@ properties:
- items:
- const: amlogic,meson6-dwmac
- const: snps,dwmac
+ - items:
+ - const: amlogic,t7-dwmac
+ - const: snps,dwmac-5.10a
+ - const: snps,dwmac
I see no such hunk in the code, so not sure how to review this.
Best regards,
Krzysztof
_______________________________________________
linux-amlogic mailing list
linux-amlogic@xxxxxxxxxxxxxxxxxxx
http://lists.infradead.org/mailman/listinfo/linux-amlogic
--
Best regards,
Chuan
--
Best regards,
Chuan
Lucas
--
Best regards,
Chuan