> > The efi specific page fault handler offers us two advantages: > > 1. Avoid panics/hangs caused by buggy firmware. > > 2. Shout loud that the firmware is buggy and hence is not a kernel bug. > > > > Finally, this new mapping will not impact a reboot from kexec, as > > kexec is only concerned about runtime memory regions. > > No. This is just a horrible hack to make completely bogus firmware work and > never fixed. >
Yes, that's true. > The proper thing to do is to have a minimal page fault handler which does: > > 1) Yell loudly if that ever happens > > 2) Handles the reboot request gracefully > > 3) Freeze and disable the EFI mess for all other cases > > That does not require any hackery to make these mappings work from atomic > context and keeps the mess confined to the EFI code where it belongs. > > Ideally we just blacklist the offending system and be done with it. This makes sense to me. I will implement the above suggested and as said should avoid the need for making mappings work from atomic context. Regards, Sai
