Merge branch 'normalize-enum-values-in-libbpf-and-bpftool'

Hungyu Lin says:

====================
Normalize enum values in libbpf and bpftool

This series fixes enum signedness in libbpf and bpftool data dumps
using the shared btf_enum_value() helper suggested by Alexei. It also
uses the helper for enum-backed map fields, where sign extension can
change the requested map_extra value.

Ordinary enums are interpreted according to kflag. When kflag is
clear, values are treated as unsigned, including BTF from producers
that did not record signedness.

The helper returns __u64, consistent with btf_enum64_value(). bpftool
casts to __s64 at signed numeric output. CO-RE relocation handling is
left for a separate change.

Changes in v5:
- Keep patches 1 and 2 and the map_extra test unchanged.
- Adopt the bpftool and libbpf tests posted by Alexei in the v4 review.

Tested on arm64 (Linux 6.12.68, GCC 13.3, Clang 18.1). libbpf and
bpftool build with -Werror under both compilers. With patch 3 alone,
the new bpftool_map_dump/enum and map_extra/enum32_high subtests and
the expanded enum_data subtest in btf_dump fail. With the full series,
all 25 subtests pass. The 22 unchanged existing subtests pass in both
cases.

v4: https://lore.kernel.org/bpf/20261107020725.36675-1-dennylin0707@gmail.com/
v3: https://lore.kernel.org/bpf/20261101103221.51534-1-dennylin0707@gmail.com/

Thanks,
Hungyu
====================

Link: https://patch.msgid.link/20261108025632.10799-1-dennylin0707@gmail.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>