On Tue, 15 Apr 2025 20:52:30 -0700 Jakub Kicinski wrote:
On Tue, 15 Apr 2025 03:35:07 +0200 Andrew Lunn wrote:
@@ -182,7 +182,7 @@ static int ch9200_mdio_read(struct net_device *netdev, int phy_id, int loc) __func__, phy_id, loc); if (phy_id != 0)
return -ENODEV;
return 0;An actually MDIO bus would return 0xffff is asked to read from a PHY which is not on the bus. But i've no idea how the ancient mii code handles this.
If this code every gets updated to using phylib, many of the changes you are making will need reverting because phylib actually wants to see the errors. So i'm somewhat reluctant to make changes like this.
Right.
I mean most of the patches seem to be adding error checking, unlike this one, but since Qasim doesn't have access to this HW they are more likely to break stuff than fix. I'm going to apply the first patch, Qasim if you'd like to clean up the rest I think it should be done separately without the Fixes tags, if at all.
Ah, no, patch 1 also does return 0. Hm. Maybe let's propagate the real error to silence the syzbot error and if someone with access to the HW comes along they can try to move this driver to more modern infra?