On Thu, Dec 14, 2023 at 02:00:16PM +0200, Vladimir Oltean wrote: > On Thu, Dec 14, 2023 at 02:54:00PM +0300, Serge Semin wrote: > > > > > The pcs-rzn1-miic.c driver puts a device_link to the MAC to at least > > > > > tear down the whole thing when the PCS is unbound, which is saner than > > > > > crashing the kernel. I don't see the equivalent protection mechanism here? > > > > > > > > You are right. I don't have any equivalent protection here. Thanks for > > > > suggesting a solution. > > > > > > I think that a device link between the "ethernet" device and the "mdio" > > > device (controller, parent of the PHY or PCS), if the Ethernet is not a > > > parent of the MDIO controller, could also solve that. But it would also > > > require ACK from PHY maintainers, who may have grander plans to address > > > this snag. > > > > Ok. I'll add it in v2. Let's see what the maintainers think about > > that. > > Are you not following the parallel discussion on the topic of PCS > devices having bound drivers? > https://lore.kernel.org/netdev/ZXnV%2FPk1PYxAm%2FjS@xxxxxxxxxxxxxxxxxxxxx/ > > Sadly I don't have much spare time to join that discussion, but it looks > like you could. Ok. Thanks for sharing the link. At least I'll follow up the discussion in order to pick up/wait for a solution they'll come up with. -Serge(y)
