Re: [PATCH u-boot v3 1/2] clk: ast2600: set default clk-delay for all ethernet macs to 0

On Mon, 2026-05-25 at 21:03 -0500, Anirudh Srinivasan wrote:
> u-boot currently setups non zero values for the ethernet mac clk delays.
> This is not always needed (depending on the board design), and prevents
> us from using rgmii-id as the phy mode in u-boot and linux. u-boot
> already has the funationality to read and use values for the clk delays
> from the DT, so based on the discussion here [1], update the default
> values for these clk delays to be zero instead.
> 
> Also update the DTs for 3 existing boards to add scu nodes with the
> existing default value for the clk delay. ast2600-fpga and msx4-bmc use
> mac0 for networking, so that particular clk-delay value was added.
> ast2600-slt uses all 4 macs, so all 4 clk-delay values were added.
> 
> Newly added boards that require no clk delays can use rgmii-id mode
> directly, and boards that actually require a clk delay can set it via
> the scu DT node.
> 
> [1] https://lore.kernel.org/openbmc/a7d364d72bbfa4ed3d41e9f7282f8e87a783cf48.camel@codeconstruct.com.au/
> 
> Signed-off-by: Anirudh Srinivasan <[email protected]>
> ---
> The fpga and slt DTs don't seem to have the DT nodes sorted in any
> order, so I added the scu nodes at the end of those files.
> 
> Setting a non-zero clk delay for the msx4 BMC device with
> rgmii-id as the phy mode feels wrong. But that's the current state of
> the tree (DT + default value of clk delay), so I've kept it as is.
> 
> Ender, Marc, would it be possible for someone from nvidia to test this
> on the mx4 platform?

Ender, Marc: Have you had an opportunity to test this?

> 
> Should we test this on the aspeed fpga and slt platforms too? Would
> someone from aspeed be able to do this?

Aspeed people: same query here.

Andrew