mbox series

[RFC,0/3] net: stmmac: dwmac-rk: use stmmac helper functions and clean up

Message ID 20210915170255.30561-1-michael.riesch@wolfvision.net (mailing list archive)
Headers show
Series net: stmmac: dwmac-rk: use stmmac helper functions and clean up | expand

Message

Michael Riesch Sept. 15, 2021, 5:02 p.m. UTC 
Hi all,

This series aims to clean up the dwmac-rk glue driver by
making the stmmac core responsible for clock and power management.
The dwmac-rk specific code is passed to the core via callbacks.
The patches have been tested successfully on a RK3568 EVB1.

With any luck, the patches fix the Ethernet regression on different
RK3399 boards, which has been introduced recently in v5.14 -- but
don't be disappointed if that is not the case :-)

The series is marked as RFC since the handling of the (internal) phy
is not quite clear yet. Should rk_gmac_{init,exit} still consider
device_may_wakeup at some point? Should the glue driver power on/off
the external phy in the first place? Or will the stmmac core take
care of it?

Also, I am sending this RFC to linux-rockchip exclusively on purpose
as I would like to see some test results on other boards before
proceeding.

As an addition, the usage of clk_bulk_* functions is envisaged but
not yet implemented due to time constraints.

Looking forward to your comments!

Best regards,
Michael

Michael Riesch (3):
  net: stmmac: dwmac-rk: use stmmac helper functions for pm ops and
    remove
  net: stmmac: dwmac-rk: clean up includes
  net: stmmac: dwmac-rk: use stmmac helper functions for clock
    management

 .../net/ethernet/stmicro/stmmac/dwmac-rk.c    | 117 +++++-------------
 1 file changed, 32 insertions(+), 85 deletions(-)

Comments

Punit Agrawal Sept. 15, 2021, 11:29 p.m. UTC | #1 
Hi Michael,

Michael Riesch <michael.riesch@wolfvision.net> writes:

> Hi all,
>
> This series aims to clean up the dwmac-rk glue driver by
> making the stmmac core responsible for clock and power management.
> The dwmac-rk specific code is passed to the core via callbacks.
> The patches have been tested successfully on a RK3568 EVB1.
>
> With any luck, the patches fix the Ethernet regression on different
> RK3399 boards, which has been introduced recently in v5.14 -- but
> don't be disappointed if that is not the case :-)

I did a quick test run - the ethernet is still buggered on RockPro64 for
me.

Thanks,
Punit

[...]