The Kconfig logic to select HAVE_DYNAMIC_FTRACE_WITH_ARGS is incorrect,
and HAVE_DYNAMIC_FTRACE_WITH_ARGS may be selected when it is not
supported by the combination of clang and GNU LD, resulting in link-time
errors:

  aarch64-linux-gnu-ld: .init.data has both ordered [`__patchable_function_entries' in init/main.o] and unordered [`.meminit.data' in mm/sparse.o] sections
  aarch64-linux-gnu-ld: final link failed: bad value

... which can be seen when building with CC=clang using a binutils
version older than 2.36.

We originally fixed that in commit:

  45bd8951 ("arm64: Improve HAVE_DYNAMIC_FTRACE_WITH_REGS selection for clang")

... by splitting the "select HAVE_DYNAMIC_FTRACE_WITH_ARGS" statement
into separete CLANG_SUPPORTS_DYNAMIC_FTRACE_WITH_ARGS and
GCC_SUPPORTS_DYNAMIC_FTRACE_WITH_ARGS options which individually select
HAVE_DYNAMIC_FTRACE_WITH_ARGS.

Subsequently we accidentally re-introduced the common "select
HAVE_DYNAMIC_FTRACE_WITH_ARGS" statement in commit:

  26299b3f ("ftrace: arm64: move from REGS to ARGS")

... then we removed it again in commit:

  68a63a41 ("arm64: Fix build with CC=clang, CONFIG_FTRACE=y and CONFIG_STACK_TRACER=y")

... then we accidentally re-introduced it again in commit:

  2aa6ac03 ("arm64: ftrace: Add direct call support")

Fix this for the third time by keeping the unified select statement and
making this depend onf either GCC_SUPPORTS_DYNAMIC_FTRACE_WITH_ARGS or
CLANG_SUPPORTS_DYNAMIC_FTRACE_WITH_ARGS. This is more consistent with
usual style and less likely to go wrong in future.

Fixes: 2aa6ac03 ("arm64: ftrace: Add direct call support")
Cc: <stable@vger.kernel.org> # 6.4.x
Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Cc: Will Deacon <will@kernel.org>
Link: https://lore.kernel.org/r/20240930120448.3352564-1-mark.rutland@arm.comSigned-off-by: Catalin Marinas <catalin.marinas@arm.com>