Re: [PATCH] x86_64/efi: Mapping Boot and Runtime EFI memory regions to different starting virtual address

[joeyli <jlee-IBi9RG/b67k@public.gmane.org>]

On Thu, Jul 30, 2015 at 03:05:42PM +0100, Matt Fleming wrote:
> On Thu, 30 Jul, at 09:39:47PM, joeyli wrote:
> > 
> > OK, understood! Thanks for your suggestion!
> > 
> > But, I have a question about mapping Boot Code/Data to -4G area. I understand
> > we need Runtime regions, and 1:1 mapping is the workaround of some buggy BIOS.
> > But why should kernel mapping Boot regions to -4G area?
> 
> We map the Boot Service regions in the -4G area for simplicity, i.e.
> it's easier to iterate through the EFI memory map in order, and map
> things starting at -4G and working down to lower addresses, than it
> would be to make different mapping decisions based on the region type.
> 
> As for why we map the Boot Service regions *at all*, that's also a bug
> workaround, see commit 916f676f8dc0 ("x86, efi: Retain boot service code
> until after switching to virtual mode"). We setup the mappings before we
> call efi_free_boot_services().
> 
> -- 
> Matt Fleming, Intel Open Source Technology Center

Thanks for your explanation.

For my issue, I will check if rewriting the VA of runtime services can fix issue.
If not, then I think need find a way to sync the mapping in EFI page table between
boot kernel and image kernel.


Thanks a lot!
Joey Lee