Implement some string functions for libzigc

[RcCreeperTech]

Progress towards #2879.

• Add memchr, memrchr, strchr, strchrnul, strrchr, index, rindex, memset, memccpy, memmem, mempcpy, stpcpy, stpncpy, strcat, strcpy, strncpy, strnlen
• Remove memcmp and depend on compiler_rt implementation

[rpkak]

I thought about creating a PR with functions like this a few weeks ago but decided not to start working on it until #23538 is resolved because I don't want to make a nearly unnoticeable diversion between libzigc and libc. Since this PR exists now, I tested it against "libc-test/src/functional/string*.c" and found the following error:

src/functional/string_strchr.c:73: strchr(s,128) with align=0 returned 0, wanted str+127
src/functional/string_strchr.c:73: strchr(s,128) with align=1 returned 0, wanted str+127
src/functional/string_strchr.c:73: strchr(s,128) with align=2 returned 0, wanted str+127
src/functional/string_strchr.c:73: strchr(s,128) with align=3 returned 0, wanted str+127
src/functional/string_strchr.c:73: strchr(s,128) with align=4 returned 0, wanted str+127
src/functional/string_strchr.c:73: strchr(s,128) with align=5 returned 0, wanted str+127
src/functional/string_strchr.c:73: strchr(s,128) with align=6 returned 0, wanted str+127
src/functional/string_strchr.c:73: strchr(s,128) with align=7 returned 0, wanted str+127
src/functional/string_strchr.c:74: strchr(s,255) with align=0 returned 0, wanted str+254
src/functional/string_strchr.c:74: strchr(s,255) with align=1 returned 0, wanted str+254
src/functional/string_strchr.c:74: strchr(s,255) with align=2 returned 0, wanted str+254
src/functional/string_strchr.c:74: strchr(s,255) with align=3 returned 0, wanted str+254
src/functional/string_strchr.c:74: strchr(s,255) with align=4 returned 0, wanted str+254
src/functional/string_strchr.c:74: strchr(s,255) with align=5 returned 0, wanted str+254
src/functional/string_strchr.c:74: strchr(s,255) with align=6 returned 0, wanted str+254
src/functional/string_strchr.c:74: strchr(s,255) with align=7 returned 0, wanted str+254

It is caused by @as(c_char, @bitCast(@as(u8, 128))) == -128 != 128.

[alexrp]

Has conflicts after #23835 due to files being moved; a rebase should be enough to deal with it automatically.

[alexrp]

Ah, I did not realize that these are hidden symbols. In that case, we can just remove them altogether.

For future reference, lib/libc/musl/libc.S is a good reference for which symbols are actually part of the musl ABI and thus should be provided by libzigc.

[RcCreeperTech]

That is fine with me. I only tried to export these symbols because I was getting linker errors when they were removed if I am remembering correctly.

[alexrp]

Oh? Do you recall the linker errors? Maybe there's something I'm missing here.

[RcCreeperTech]

I can't remember exactly since it was a few days ago, I just removed them again and I am re-running the test suite locally so we will see if the errors re-appear.

[alexrp]

In hindsight, the way I set things up in lib/c.zig didn't really make sense:

zig/lib/c.zig

Lines 16 to 34 in 833d4c9

	comptime {
	if (builtin.target.isMuslLibC() or builtin.target.isWasiLibC()) {
	// Files specific to musl and wasi-libc.
	_ = @import("c/string.zig");
	_ = @import("c/strings.zig");
	}
	if (builtin.target.isMuslLibC()) {
	// Files specific to musl.
	}
	if (builtin.target.isWasiLibC()) {
	// Files specific to wasi-libc.
	}
	if (builtin.target.isMinGW()) {
	// Files specific to MinGW-w64.
	}
	}

I suggest you move those two @imports above any of the libc-specific blocks, and instead cover the @exports with appropriate builtin.target.is*LibC() checks in lib/c/string[s].zig. In particular, most of the exports aren't needed for MinGW - only mempcpy and strnlen at this time, AFAICT.

[alexrp]

index and rindex are part of strings.h and so should go into strings.zig (you'll need to rebase on upstream master for that).