Failed to compile 128-bit tests using clang on mac

[Souvenal]

Platform:

Kernel: arm64-apple-darwin25.0.0
Chip: Apple M4
Clang: 20.1.5
Issue:
When running EASTL benchmarks and tests locally, I encountered compilation errors related to 128-bit integer support.

If I enable both benchmark and test targets, I get errors like:

error: no member named 'fetch_add' in 'eastl::atomic<unsigned __int128>'
__uint128_t val = atomic.fetch_add(1, eastl::memory_order_relaxed);

After investigation, I found that template specialization does not recognize __uint128_t as an integral type. This seems to be due to the following code in eacompilertraits.h:

#if (defined(__GNUC__) && defined(__x86_64__))
    #define EA_COMPILER_INTMAX_SIZE 16  // intmax_t is __int128_t (GCC extension) and is 16 bytes.
#else
    #define EA_COMPILER_INTMAX_SIZE 8   // intmax_t is int64_t and is 8 bytes.
#endif

which does not correctly define the size for my platform.

Even after manually setting the correct size, I encountered further issues with numeric_limits specialization. It seems that Clang and GCC use different names for 128-bit integer types, and I am not familiar with cross-compiler settings for this.

For now, I have disabled 128-bit tests to allow compilation.

Question:
Is there a recommended way to support 128-bit integer atomics and numeric limits across Clang and GCC? Can this be fixed in EASTL for better cross-compiler compatibility?