This commit fixes the build process for the current limitations of
objcopy. Objcopy still has the problem of not fully understanding
relocations so while this compiles, it will not be loadable by uefi.

Signed-off-by: Bart Kuivenhoven <bemk@redhat.com>

Work has been underway to add support for pe-coff on aarch64 to the
binutils suite. This work has produced a framework which as of yet
requires these settings to be used.

Also there is something going on with a memset call that doesn't happen
on intel. There supposedly is a flag disabling it there, but I heavily
suspect this too hasn't been ported to aarch64 yet.

Signed-off-by: Bart Kuivenhoven <bemk@redhat.com>

gnu/efi/reloc_aarch64.c: should mostly work as it operates on image
metadata level, rather than architecture. No guarantees though, as I
haven't been able to test this.

elf_aarch64_efi.lds: should also work, going from the comments I got from
the original developers. No tests have been done as of yet, as binutils
hasn't been ported to enable the tests yet.

inc/aarch64/efi.h: Ehrm, yep, this is more or less, because I couldn't
find a good cross compilable library, so yanked this file out from the C
library for x86 and tried to make it fit the aarch64 compiler and stuff.
It should work (right?) but no tests have been done at this time to
confirm this.

Basically, this code should not be trusted, at least not yet. It's a
stepping stone.

Signed-off-by: Bart Kuivenhoven <bemk@redhat.com>

These functions have been slightly modified, in that they don't include
the right information for GDB anymore.

I haven't taken the time to investigate in what GDB requires these
functions to do, but assume. If you want to debug this using GDB, good
luck.

Signed-off-by: Bart Kuivenhoven <bemk@redhat.com>

After some tests using the aarch64 GNU compiler, I've found that these
are the best settings to make the type sizes match.

Signed-off-by: Bart Kuivenhoven <bemk@redhat.com>

Signed-off-by: Bart Kuivenhoven <bemk@redhat.com>

After making sure the code compiles for the architecture the next
logical step is to got through all the warnings, and fix those.

This commit takes away all the low hanging fruit.

Signed-off-by: Bart Kuivenhoven <bemk@redhat.com>

To make sure we can find the things that have to be ported more easily,
I've added todo comments to the stub files.

Grep and you shal find.

Signed-off-by: Bart Kuivenhoven <bemk@redhat.com>

This patch makes the code compile using the linaro tools for aarch64.

http://www.linaro.org/engineering/engineering-projects/armv8

The code in this patch has not yet been completed. On the todo list are
amongst others:

- Constants to be verified
- Functions to be ported
- Support to be added for other compilers

Signed-off-by: Bart Kuivenhoven <bemk@redhat.com>

There were some warnings on compiling, which have now been fixed.

The first set of warnings, comes from variables that only get used in an
assert statement. In order to make sure that those aren't broken in case
this is used for debugging, I've put pragma's around them, to prevent
warnings (in case of GCC).

The next set of warnings came from unused variables, that weren't used
in anything else.

Those have been commented out, but still left in the codebase so that if
it was something somebody was working on, can still continue, while this
batch of code is also free of warnings.

The last set of warnings I fixed was because of integers being cast to
pointers of different size. Since I didn't feel like digging in too
deep, this too has been put on the ignore list using some pragmas.

Signed-off-by: Bart Kuivenhoven <bemk@redhat.com>