Re: [PATCH net] net: stmmac: fix system hang caused by eee_ctrl_timer during suspend/resume

On Fri, Nov 12, 2021 at 01:21:57AM +0000, Joakim Zhang wrote:

Hi Vladimir,

quoted

Hi Joakim,

This patch looks strange to me. You are basically saying that the LPI timer for
MAC-level EEE is clocked from the clk_ptp_ref clock?! Are you sure this is
correct? I thought this clock is only used for the PTP timestamping counter.
Maybe the clock definitions in imx8mp.dtsi are not correct?

No, MAC-level EEE is not clocked from the clk_ptp_ref clock, this clock is for PTP. To fix this issue,
I can only move pm_runtime_force_suspend() into noirq suspend stage. As commit message said,
"postpone clocks management from suspend stage to noirq suspend stage", that means I move all
the clocks management into noirq suspend stage, including PTP clock, it's should be harmless for PTP
function. Do you find any regression with this patch?

No, I didn't find any regression, I was just frustrated yesterday that a
bugfix patch I wanted to send to "stable" on the handling of clk_ptp_ref
is conflicting with your (apparently unnecessary) movement of this clock
to the noirq suspend stage. I'm better now.