I cannot say for sure, but e1000 does not have this behavior, vmxnet3 in vmware also does not behave this way. In addition, the address gets there only after a reboot, it does not get there during normal operation, as you can see from logs i attached. It seems to me that this is not quite the expected behavior, isn't it?
вт, 31 янв. 2023 г. в 21:48, Peter Maydell <peter.mayd...@linaro.org>: > > On Tue, 31 Jan 2023 at 18:22, Vlao Mao <vlao...@gmail.com> wrote: > > > > A bug was found with the vmxnet3 driver, after changing the MAC address > > and rebooting, it falls into the category of a permanent address until > > the VM is turned off. > > For some network device models (those where the hardware has an > EEPROM and we're modelling "guest programmed a new MAC address > into the EEPROM") that is the correct behaviour. Is the vmxnet3 > definitely one of the kind which is not supposed to have a > permanent MAC address backing storage ? (It's paravirtual, so > this would hopefully be indicated in the specs for it.) > > thanks > -- PMM