[PATCH v2 0/3] aarch64: Relocation fixes and LTO

[Martin Storsjö]

On Mon, 18 Nov 2024, Evgeny Karpov wrote:

> Hello,
>
> Thank you for reviewing v1!
>
> Changes in v2:
> - Revert relocation processing for IMAGE_REL_ARM64_PAGEBASE_REL21. The required changes
>  to extend the limit for the offset in "symbol + offset" from 1MB to 16MB have been made
>  on the GCC side.
> - Change logic to prevent the reduction of a symbol name to a section name by using
>  aarch64_fix_adjustable.
> - Correct formatting.
>
> This patch series supports the GCC patch series
> "[PATCH v3 0/8] SMALL code model fixes, optimization fixes, LTO and minimal C++ enablement"
> which has been approved.
> https://gcc.gnu.org/pipermail/gcc-patches/2024-November/668484.html
>
> The "aarch64: Add AArch64 support to libiberty" patch is supported
> by that GCC patch series.
> "[PATCH v3 7/8] Add LTO support"
> https://gcc.gnu.org/pipermail/gcc-patches/2024-November/668493.html
>
> Regards,
> Evgeny
>
>
> Evgeny Karpov (3):
>  aarch64: Fix symbol reduction for relocations
>  aarch64: Add AArch64 support to libiberty
>  Support relocation for weak references
>
> bfd/coff-aarch64.c             | 52 ++++++++++++++++++++--------------
> bfd/cofflink.c                 |  3 ++
> gas/config/tc-aarch64.c        | 21 ++++++++++++++
> gas/config/tc-aarch64.h        |  6 ++++
> libiberty/simple-object-coff.c |  4 ++-
> 5 files changed, 64 insertions(+), 22 deletions(-)

No further objections from me, to this patchset.

I can't really say I know exactly what patch 1 does, though - it would be 
nice to have a concrete user level example of what kind of input was 
handled incorrectly before and which now should work though (and which 
related cases were handled correctly by the code before which now still 
should keep working as it did before).

// Martin