Re: [PATCH v2] bcm63xx_enet: fix internal phy IRQ assignment

On 26.02.2021 15:16, Andrew Lunn wrote:

quoted

quoted

OK, I see. Then there's no reason to complain upstream.
Either use the mainline B53 DSA driver of fix interrupt mode
downstream.

I agree.

This b53 driver has one PHY with the same BCM63XX phy_id, causing a
double probe. I'll send the original patch to the OpenWrt project.

Hi Daniel

There is a bit of a disconnect between OpenWRT and Mainline. They have
a lot of fixes that don't make it upstream. So it is good to see
somebody trying to fix mainline first, and then backport to
OpenWRT. But please do test mainline and confirm it is actually broken
before submitting patches.

When you do submit to OpenWRT, please make it clear this is an OpenWRT
problem so somebody does not try to push it to mainline again....

And if you have an itch to scratch, try adding mainline support for
this board. We can guide you.

Daniel has two conflicting PHY drivers for bcm63xx, the one from mainline,
and one in the OpenWRT downstream b53 driver. Removing the mainline
PHY driver would resolve the conflict, but the OpenWRT PHY driver has
no IRQ support so Daniel would gain nothing.
I think best would be to remove the duplicated PHY driver from the
OpenWRT b53 driver. Daniel could try to remove b53_phy_driver_id3 and
re-test.

	Andrew

Heiner