https://bugzilla.tianocore.org/show_bug.cgi?id=3987

--- Comment #17 from Dionna Glaze <dionnaglaze@...> ---
The E/F options were speculative as to whether the OS community (Linux in my
case) would accept baking in this kind of configuration information directly
into the binary. Ard was firmly against this
https://lore.kernel.org/linux-mm/CAMj1kXHEc=vEt=CtfdiPEsUe2i8QogAi+jvtY6h1awo7GZ-nRg@mail.gmail.com/

My original idea about E/F would be to inspect the binary before running
anything, then set up a QemuFwCfg file opt/ovmf/AcceptAllMemory if the binary
was missing the appropriate information. We'd use that file to determine in PEI
whether to accept all memory or allow creation of unaccepted memory.

With inspection, we couldn't handle something as complicated (impossible?) as
finding and analyzing a grub.cfg file to determine in the kernel that will be
booted will have the appropriate support information included. Boot loaders
make this hard, since they can call ExitBootServices before the guest kernel
has a chance to start and indicate the right support bit. That is, if we use
the OsIndications solution over the GetMemoryMapEx solution.

--
You are receiving this mail because:
You are on the CC list for the bug.