BOG BOG wrote: > I have followed your suggestions regarding -current. > > amd64 -current (GENERIC.MP) does no longer hangs when trying to detach re0, > but now, \ > re0 is sometimes detected correctly and sometimes not:
no. main controller itself works in both shown cases. there is something wrong with PHY and you need to find out a way to make this issue reproducible. > > I give the messages in both cases: > Failure: > re0 an pci2 dev0 function 0 "Realtek 8168" rev 0x02: RTL8168C/8111C (0x3c00), > apic2 \ > int 16 (irq 11), address 00:19:99:54:31:7f > re0: PHY write failed > re0: PHY write failed > re0: no PHY found! > re0: reset never completed! > re0 detached > > Success: > re0 an pci2 dev0 function 0 "Realtek 8168" rev 0x02: RTL8168C/8111C (0x3c00), > apic2 \ > int 16 (irq 11), address 00:19:99:54:31:7f rgephy0 at re0 phy 7: > RTL8169S/8110S PHY, \ > rev. 2 > > the controller is: 10/100/1000 MBit/s Realtek RTL8111C, so i think the kernel > \ > sometimes initializes the controller with a driver and sometimes with a > 'close enough \ > driver', and in the later case it works. Alexey