Tested together with D41857 on Zybo Z7 board, no problem with this change.

Tested on Zybo Z7 with sources from June 1st, 2023 and patch applied (partially) manually, system boots and everything seems to work as before.
Test on riscv64 (Microchip PolarFire SoC) should not be problem, just more time is necessary to prepare and conduct the test :(

Fix probe issue pointed to by skibo

Inline comments made by mhorne addressed...

In D34764#787943, @mhorne wrote:

The quirk handling changes ... should be committed separately from the phy_contype + typo changes.

This diff was done with -U99999 and module directory was renamed to sys/modules/if_cgem. Hopefully there is no space/tabs issue, now it should apply correctly.

In D34687#786708, @mhorne wrote:

Hi, the patch does not apply correctly to the main branch. Can you rebase and re-upload the diff with context?

One final suggestion: the new subfolder should be named sys/modules/if_cgem, to match the module name.

In D34687#786108, @mhorne wrote:

Minor notes, but this looks good!

In the future, please upload diffs with full context -- git show -U99999.

Additional note: Makefile originally provided by Thomas Skibo, I'd like to attribute it properly.

New revision is just being created, https://reviews.freebsd.org/D34687, for building if_cgem.ko kernel module, used to test changes proposed here for miibus.ko kernel module

Testing with 'make buildenv' revealed opt_platform.h is necessary as well. As part of normal buildkernel, this file is already created.

Minimal diff with new files added at the end.

In D34256#775528, @mindal_semihalf.com wrote:

I think it might be better to keep the file sorted the way it was before this change.
This will make this patch much easier to read.

Original patch missed inclusion of VSCPHY driver, corrected here.

After some mail exchange regarding problem elsewhere, applied this review minus maxspeed handling for brevity, DP83867 PHY looks like working for me (I need to fix something else unrelated to this review to be sure) - at least media status sense and autonegotiation works as expected.

Looking a bit more into it, I think maxspeed variable and mii_maxspeed struct member is just read from DTB, but not used anywhere.

I'd like to test this review, but two errors occured when building. First one, unknown function 'device_get_property' means I need fresh sources (I see it in recently fetched git repository).
Second one, no member named 'mii_maxspeed' in 'struct mii_softc' I can't resolve recently - looking into file sys/dev/mii/miivar.h, where struct mii_softc is defined, does not show such member of structure.