On Mon, May 02, 2022 at 08:21:10PM +0200, Andrew Lunn wrote:

quoted

Could it be that the issue is writing the MSCR in fec_restart(),
`writel(fep->phy_speed, fep->hwp + FEC_MII_SPEED)`?

I do see the issue on link up/down event, when this function is actually
called.

quoted

From what I can understand from the previous history:

  1e6114f51f9d (net: fec: fix MDIO probing for some FEC hardware blocks, 2020-10-28) 
  f166f890c8f0 (net: ethernet: fec: Replace interrupt driven MDIO with polled IO, 2020-05-02)

writing to this register could trigger a FEC_ENET_MII interrupt actually
creating a race condition with fec_enet_mdio_read() that is called on
link change also.

You should read the discussion from when this code was added.

Of course, I did it.

Are you planning on adding:

       if (fep->quirks & FEC_QUIRK_CLEAR_SETUP_MII) {
                /* Clear MMFR to avoid to generate MII event by writing MSCR.
                 * MII event generation condition:
                 * - writing MSCR:
                 *      - mmfr[31:0]_not_zero & mscr[7:0]_is_zero &
                 *        mscr_reg_data_in[7:0] != 0
                 * - writing MMFR:
                 *      - mscr[7:0]_not_zero
                 */
                writel(0, fep->hwp + FEC_MII_DATA);
        }

To other locations which change FEC_MII_SPEED?

Correct, plus the required locking since both fec_enet_mdio_read() and
_write() do write into the same registers.

Francesco