ww/test/object/elfobj_test.ww

package elfobj_test;

// w6a OBJECT-level observers. Ports of the retired native carriers
// test/wcc/500_w6a.c, 510_dataw.c, and 520_datar.c; every assertion
// preserved.
//
// elf64header (500) — all 4 inline programs: w6c exit 0, w6a exit 0,
// and the .o opens with the ELF magic + ELFCLASS64 ident byte.
//
// dataw (510) — hand-written DATAW .s: e_shnum == 7 (NULL, .text,
// .rela.text, .data, .symtab, .strtab, .shstrtab); .data is
// SHT_PROGBITS, ALLOC|WRITE, non-EXECINSTR, 8 bytes equal to the LE
// i64 42; symbol `counter` sits in .data at st_value 0.
//
// nodataw (510) — a .s without DATAW keeps the legacy 6-section
// layout with no .data; test 991 (selfhost w6a .o byte-diff, the
// terminal bootstrap gate) depends on that invariant.
//
// datar (520) — DATAR emits exactly one R_X86_64_64 entry at offset 0
// in .rela.data; w6l links the .o and the binary prints the 5 bytes
// `hello` through the relocated pointer and exits 0. Strengthened
// over the carrier: the reloc's symbol index is resolved through
// .symtab and must name `greeting` (the C code declared but never
// implemented that check).
//
// Dropped C machinery, not assertions: wwtestpkg.h package-clause
// injection (the sources here spell `package main;` out) and per-file
// unlink accounting (testenv.clean asserts the removal).

import os;
import os.exec;
import strconv;
import strings;
import testenv;
import time;

fn fail(label: str, why: str) void = {
	let m: str = strings.concat("elfobj FAIL: ", label, " -- ", why,
		"\n");
	os.write(2, m.ptr, m.len: u64);
	assert(false);
};

fn tmo() time.duration = {
	return (240i64 * (time.second: i64)): time.duration;
};

// -1 encodes an abnormal (non-EXIT) termination, never a valid code.
fn runcode(dir: str, name: str, argv: []str) i32 = {
	let co: testenv.commandout;
	testenv.runcommand(dir, dir, name, argv, tmo(), &co);
	if (co.termination != exec.termination.EXIT) { return -1; };
	return co.code;
};

// itos returns a view into a static buffer; dup materializes it.
fn dec(n: i32) str = {
	return strings.dup(strconv.itos(n: int, strconv.base.DEC));
};

// External-ABI constants (System V gABI, ELF64 relocatable objects) —
// the layout contract w6a's obj writer encodes: Ehdr e_shoff at 40
// (u64), e_shnum at 60 (u16), e_shstrndx at 62 (u16); Shdr entries of
// 64 bytes with sh_name +0 (u32), sh_type +4 (u32), sh_flags +8
// (u64), sh_offset +24 (u64), sh_size +32 (u64), sh_link +40 (u32);
// Sym64 entries of 24 bytes with st_name +0 (u32), st_shndx +6 (u16),
// st_value +8 (u64); Rela64 entries of 24 bytes with r_offset +0
// (u64), r_info +8 (u64; low u32 = type, next u32 = symbol index).
def SHENT: i32 = 64;
def SYMENT: i32 = 24;
def RELAENT: i32 = 24;

fn shnum(o: str) i32 = { return testenv.leu16(o, 60): i32; };

fn shoff(o: str, i: i32) i32 = {
	return (testenv.leu64(o, 40): i32) + SHENT * i;
};

fn cstr(o: str, off: i32) str = {
	let nb: []u8 = alloc([], 64u64)!;
	let k: i32 = 0;
	for (o[off + k] != 0u8) {
		append(nb, o[off + k]);
		k += 1;
	};
	return strings.frombytes(nb);
};

fn shname(o: str, i: i32) str = {
	let strsh: i32 = shoff(o, testenv.leu16(o, 62): i32);
	let stroff: i32 = testenv.leu64(o, strsh + 24): i32;
	return cstr(o, stroff + (testenv.leu32(o, shoff(o, i)): i32));
};

fn findsection(o: str, name: str) i32 = {
	let i: i32 = 0;
	for (i < shnum(o)) {
		if (testenv.same(shname(o, i), name)) { return i; };
		i += 1;
	};
	return -1;
};

fn findsectiontype(o: str, ty: u64) i32 = {
	let i: i32 = 0;
	for (i < shnum(o)) {
		if (testenv.leu32(o, shoff(o, i) + 4) == ty) { return i; };
		i += 1;
	};
	return -1;
};

// Name of .symtab entry `symidx` through the sh_link strtab; "" when
// there is no symtab (SHT_SYMTAB == 2).
fn symname(o: str, symidx: i32) str = {
	let ist: i32 = findsectiontype(o, 2u64);
	if (ist < 0) { return ""; };
	let sh: i32 = shoff(o, ist);
	let symoff: i32 = testenv.leu64(o, sh + 24): i32;
	let strsh: i32 = shoff(o, testenv.leu32(o, sh + 40): i32);
	let stroff: i32 = testenv.leu64(o, strsh + 24): i32;
	let e: i32 = symoff + SYMENT * symidx;
	return cstr(o, stroff + (testenv.leu32(o, e): i32));
};

// Byte offset of the named Sym64 entry, or -1. Entry 0 is the
// reserved null symbol.
fn findsym(o: str, name: str) i32 = {
	let ist: i32 = findsectiontype(o, 2u64);
	if (ist < 0) { return -1; };
	let sh: i32 = shoff(o, ist);
	let symoff: i32 = testenv.leu64(o, sh + 24): i32;
	let n: i32 = (testenv.leu64(o, sh + 32): i32) / SYMENT;
	let i: i32 = 1;
	for (i < n) {
		if (testenv.same(symname(o, i), name)) {
			return symoff + SYMENT * i;
		};
		i += 1;
	};
	return -1;
};

fn iself64(o: str) bool = {
	if (o.len < 64) { return false; };
	return o[0] == 0x7fu8 && o[1] == 0x45u8 && o[2] == 0x4cu8
		&& o[3] == 0x46u8 && o[4] == 2u8;
};

// ---- elf64header (500) -------------------------------------------------

@test fn elf64header() void = {
	let td: str = testenv.fresh();
	let programs: []str = [
		"fn main() i32 = { return 42; };",
		"fn add(a: i32, b: i32) i32 = { return a + b; };",
		"fn loop() i32 = { let i: i32 = 0; for (i < 10) { i += 1; }; return i; };",
		"fn cmp(a: i32, b: i32) bool = { return a < b; };",
	];
	let i: i32 = 0;
	for (i < 4) {
		let stem: str = strings.concat(td, "/p", dec(i));
		let src: str = strings.concat(stem, ".ww");
		let asmf: str = strings.concat(stem, ".s");
		let obj: str = strings.concat(stem, ".o");
		testenv.writefile(src, strings.concat("package main;\n",
			programs[i], "\n"));
		let cav: []str = [testenv.driver("w6c"), "-o", asmf, src];
		if (runcode(td, strings.concat("c", dec(i)), cav) != 0) {
			fail("elf64header", strings.concat("w6c failed on ",
				programs[i]));
		};
		let aav: []str = [testenv.driver("w6a"), "-o", obj, asmf];
		if (runcode(td, strings.concat("a", dec(i)), aav) != 0) {
			fail("elf64header", strings.concat("w6a failed on ",
				programs[i]));
		};
		if (!iself64(testenv.readfile(obj))) {
			fail("elf64header", strings.concat("not an ELF64: ",
				programs[i]));
		};
		i += 1;
	};
	testenv.clean(td);
};

// ---- dataw (510) -------------------------------------------------------

@test fn dataw() void = {
	let td: str = testenv.fresh();
	let src: str = strings.concat(td, "/dataw.s");
	let obj: str = strings.concat(td, "/dataw.o");
	testenv.writefile(src, strings.concat(
		"TEXT _start(SB),$0\n",
		"\tMOVQ\t$0, AX\n",
		"\tRET\n",
		"DATAW counter(SB),\"\\x2a\\x00\\x00\\x00\\x00\\x00\\x00\\x00\"\n"));
	let av: []str = [testenv.driver("w6a"), "-o", obj, src];
	if (runcode(td, "dataw_a", av) != 0) {
		fail("dataw", "w6a failed");
	};
	let o: str = testenv.readfile(obj);

	// NULL, .text, .rela.text, .data, .symtab, .strtab, .shstrtab
	if (shnum(o) != 7) {
		fail("dataw", strings.concat("e_shnum=", dec(shnum(o)),
			", want 7"));
	};
	let idx: i32 = findsection(o, ".data");
	if (idx < 0) { fail("dataw", ".data section not found"); };
	let sh: i32 = shoff(o, idx);
	// SHT_PROGBITS == 1
	if (testenv.leu32(o, sh + 4) != 1u64) {
		fail("dataw", ".data sh_type != PROGBITS");
	};
	// SHF_WRITE|SHF_ALLOC == 0x3, SHF_EXECINSTR == 0x4
	let fl: u64 = testenv.leu64(o, sh + 8);
	if ((fl & 3u64) != 3u64) {
		fail("dataw", ".data flags miss ALLOC|WRITE");
	};
	if ((fl & 4u64) != 0u64) {
		fail("dataw", ".data is executable, want non-X");
	};
	if (testenv.leu64(o, sh + 32) != 8u64) {
		fail("dataw", ".data size != 8");
	};
	let doff: i32 = testenv.leu64(o, sh + 24): i32;
	if (o[doff] != 0x2au8) {
		fail("dataw", ".data contents mismatch (LE i64 42)");
	};
	let k: i32 = 1;
	for (k < 8) {
		if (o[doff + k] != 0u8) {
			fail("dataw", ".data contents mismatch (LE i64 42)");
		};
		k += 1;
	};
	let sym: i32 = findsym(o, "counter");
	if (sym < 0) { fail("dataw", "symbol 'counter' not in .symtab"); };
	if ((testenv.leu16(o, sym + 6): i32) != idx) {
		fail("dataw", "'counter'.st_shndx is not the .data index");
	};
	if (testenv.leu64(o, sym + 8) != 0u64) {
		fail("dataw", "'counter'.st_value != 0");
	};
	testenv.clean(td);
};

// ---- nodataw (510) -----------------------------------------------------

@test fn nodataw() void = {
	let td: str = testenv.fresh();
	let src: str = strings.concat(td, "/nodataw.s");
	let obj: str = strings.concat(td, "/nodataw.o");
	testenv.writefile(src, strings.concat(
		"TEXT _start(SB),$0\n",
		"\tMOVQ\t$0, AX\n",
		"\tRET\n"));
	let av: []str = [testenv.driver("w6a"), "-o", obj, src];
	if (runcode(td, "nodataw_a", av) != 0) {
		fail("nodataw", "w6a failed");
	};
	let o: str = testenv.readfile(obj);
	// 991 (selfhost w6a .o byte-diff) relies on the legacy layout.
	if (shnum(o) != 6) {
		fail("nodataw", strings.concat("e_shnum=", dec(shnum(o)),
			", want 6 (legacy)"));
	};
	if (findsection(o, ".data") >= 0) {
		fail("nodataw", ".data section present, want absent");
	};
	testenv.clean(td);
};

// ---- datar (520) -------------------------------------------------------

@test fn datar() void = {
	let td: str = testenv.fresh();
	let src: str = strings.concat(td, "/datar.s");
	let obj: str = strings.concat(td, "/datar.o");
	let exe: str = strings.concat(td, "/datar.x");
	testenv.writefile(src, strings.concat(
		"TEXT _start,$0\n",
		"\tMOVQ\t$1, AX\n",
		"\tMOVQ\t$1, DI\n",
		"\tMOVQ\tpair(SB), SI\n",
		"\tMOVQ\t$5, DX\n",
		"\tSYSCALL\n",
		"\tMOVQ\t$60, AX\n",
		"\tMOVQ\t$0, DI\n",
		"\tSYSCALL\n",
		"\n",
		"DATAW pair(SB),\"\\x00\\x00\\x00\\x00\\x00\\x00\\x00\\x00",
		"\\x05\\x00\\x00\\x00\\x00\\x00\\x00\\x00\"\n",
		"DATAR pair+0(SB),greeting(SB)\n",
		"DATA greeting(SB),\"hello\"\n"));
	let aav: []str = [testenv.driver("w6a"), "-o", obj, src];
	if (runcode(td, "datar_a", aav) != 0) {
		fail("datar", "w6a failed");
	};
	let o: str = testenv.readfile(obj);
	let idx: i32 = findsection(o, ".rela.data");
	if (idx < 0) { fail("datar", ".rela.data section missing"); };
	let sh: i32 = shoff(o, idx);
	// SHT_RELA == 4
	if (testenv.leu32(o, sh + 4) != 4u64) {
		fail("datar", ".rela.data sh_type != SHT_RELA");
	};
	// exactly one Rela64 entry
	if (testenv.leu64(o, sh + 32) != RELAENT: u64) {
		fail("datar", ".rela.data entry count != 1");
	};
	let ro: i32 = testenv.leu64(o, sh + 24): i32;
	if (testenv.leu64(o, ro) != 0u64) {
		fail("datar", "reloc offset != 0");
	};
	// R_X86_64_64 == 1
	if (testenv.leu32(o, ro + 8) != 1u64) {
		fail("datar", "reloc kind != R_X86_64_64");
	};
	if (!testenv.same(symname(o, testenv.leu32(o, ro + 12): i32),
		"greeting")) {
		fail("datar", "reloc symbol != greeting");
	};

	let lav: []str = [testenv.driver("w6l"), "-o", exe, obj];
	if (runcode(td, "datar_l", lav) != 0) {
		fail("datar", "w6l failed");
	};
	let rav: []str = [exe];
	let co: testenv.commandout;
	testenv.runcommand(td, td, "datar_run", rav, tmo(), &co);
	if (co.termination != exec.termination.EXIT || co.code != 0) {
		fail("datar", "exe exit != 0");
	};
	// the linker patched the .data slot with greeting's runtime VA
	if (!testenv.same(co.stdout, "hello")) {
		fail("datar", "stdout != 'hello'");
	};
	testenv.clean(td);
};