Hi Laszlo, Yes, I think it's enough to provide the user-friendly error message in EDK2. The command lines to start the VM/guest can be adjusted to have more than 128MB memory associated for NUMA node#0 when it's seen by users.
Thanks, Gavin Laszlo Ersek <ler...@redhat.com> 于2022年3月4日周五 22:24写道： > > Gerd, > > On 03/04/22 11:58, Peter Maydell wrote: > > On Fri, 4 Mar 2022 at 10:52, Igor Mammedov <imamm...@redhat.com> wrote: > >> if firmware is not an option, I wouldn't opposed to printing warning > >> message from QEMU if you can detect that you are running broken edk2 > >> and under condition that no new infa/hooks are invented for this. > >> (assuming it's worth all the effort at all) > > > > I am definitely not in favour of that. QEMU should provide the > > emulated hardware and let the firmware deal with detecting if > > it's missing important stuff. It should as far as is possible > > not have special-case detection-of-broken-guests handling. > > > > thanks > > -- PMM > > > > It's probably simplest if you replace the ASSERT()s in question; please > see the attachment. (Only smoke tested.) Gavin indicated up-thread he'd > be OK with an easier-to-understand error message. > > Laszlo
