Contradicting `SOL_TCP` and `SOL_UDP` definition for p1

[ChihweiLHBird]

They are not defined within the p1 scope in libc-bottom-half/headers/public/__header_sys_socket.h:

wasi-libc/libc-bottom-half/headers/public/__header_sys_socket.h

Lines 27 to 28 in 0d5332a

	#define SOL_TCP 6
	#define SOL_UDP 17

But they are included in expected/wasm32-wasip1/predefined-macros.txt:
https://github.com/WebAssembly/wasi-libc/blob/0d5332aceea8e72698b6abc270550e666ba97aea/expected/wasm32-wasip1/predefined-macros.txt

Which one should be the source of truth?

[Hokila]

Thanks for filing this. I traced the current p1/p2/p3 header split at commit e1655e8:

• In libc-bottom-half/headers/public/__header_sys_socket.h, SOL_TCP 6 and SOL_UDP 17 are defined only in the __wasip2__ || __wasip3__ branch; the __wasip1__ branch includes <wasi/api.h> and defines the MSG constants but not these two socket-level constants.
• libc-top-half/musl/include/netinet/tcp.h and udp.h define SOL_TCP 6 and SOL_UDP 17, respectively.
• The generated p1 predefined-macro snapshot nevertheless contains both values, while the p2/p3 snapshots get them from the public socket header path.

Before I write the regression test, could you confirm the intended source of truth for p1: should SOL_TCP/SOL_UDP be added to the p1 branch of __header_sys_socket.h, or should the generated p1 expectation remain derived through the netinet headers? I will keep the patch scoped to p1 and preserve the p2/p3 behavior.