ww/test/object/link_test.ww

package link_test;

// w6l LINK-level observers. Ports of the retired native carriers
// test/wcc/600_w6l.c and 620_data_link.c; every assertion preserved.
//
// staticexec (600) — w6c/w6a/w6l on a trivial program produce an
// ELF64 ET_EXEC. Strengthened over the carrier: the ldd host-tool
// proxy for staticness is replaced by a direct scan of every program
// header asserting no PT_INTERP.
//
// readglobal / writeglobal (620 test_read / test_write) — a DATAW
// global is readable (exit 42) and the R+W mapping truly writable
// (store 99, read back, exit 99).
//
// twoloads (620) — a DATAW input links to exactly 2 program headers,
// both PT_LOAD, exactly one R+X and one R+W (exact p_flags equality;
// any other combination fails).
//
// nodatasingleload (620) — an input without DATAW keeps the single
// PT_LOAD layout; test 992 (selfhost w6l diff, the terminal bootstrap
// gate) depends on that invariant.
//
// bssfilesz (620) — trailing .data zeros are trimmed from the file:
// the R+W PT_LOAD has p_filesz 1 < p_memsz 32 (the trim ignores
// symbol boundaries — nz's own zero tail goes too), and the loader
// zero-fills the gap (run exit 7).
//
// Dropped C machinery, not assertions: wwtestpkg.h package-clause
// injection and per-file unlink accounting (testenv.clean).

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

fn fail(label: str, why: str) void = {
	let m: str = strings.concat("link 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;
};

// External-ABI constants (System V gABI, ELF64 executables) — the
// layout contract w6l's exe writer encodes: e_ident magic + class at
// 0..4, Ehdr e_type at 16 (u16), e_phoff at 32 (u64), e_phnum at 56
// (u16); Phdr entries of 56 bytes with p_type +0 (u32), p_flags +4
// (u32), p_filesz +32 (u64), p_memsz +40 (u64). PT_LOAD == 1,
// PT_INTERP == 3; PF_X == 1, PF_W == 2, PF_R == 4.
def PHENT: i32 = 56;

fn phnum(o: str) i32 = { return testenv.leu16(o, 56): i32; };

fn phoff(o: str, i: i32) i32 = {
	return (testenv.leu64(o, 32): i32) + PHENT * i;
};

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;
};

fn build(td: str, tag: str, body: str) str = {
	let src: str = strings.concat(td, "/", tag, ".s");
	let obj: str = strings.concat(td, "/", tag, ".o");
	let exe: str = strings.concat(td, "/", tag, ".x");
	testenv.writefile(src, body);
	let aav: []str = [testenv.driver("w6a"), "-o", obj, src];
	if (runcode(td, strings.concat("a_", tag), aav) != 0) {
		fail(tag, "w6a failed");
	};
	let lav: []str = [testenv.driver("w6l"), "-o", exe, obj];
	if (runcode(td, strings.concat("l_", tag), lav) != 0) {
		fail(tag, "w6l failed");
	};
	return exe;
};

fn runexit(td: str, tag: str, exe: str) i32 = {
	let rav: []str = [exe];
	return runcode(td, strings.concat("run_", tag), rav);
};

@test fn staticexec() void = {
	let td: str = testenv.fresh();
	let src: str = strings.concat(td, "/m.ww");
	let asmf: str = strings.concat(td, "/m.s");
	let obj: str = strings.concat(td, "/m.o");
	let exe: str = strings.concat(td, "/m.x");
	testenv.writefile(src,
		"package main;\nfn main() i32 = { return 42; };\n");
	let cav: []str = [testenv.driver("w6c"), "-o", asmf, src];
	if (runcode(td, "c_m", cav) != 0) { fail("staticexec", "w6c failed"); };
	let aav: []str = [testenv.driver("w6a"), "-o", obj, asmf];
	if (runcode(td, "a_m", aav) != 0) { fail("staticexec", "w6a failed"); };
	let lav: []str = [testenv.driver("w6l"), "-o", exe, obj];
	if (runcode(td, "l_m", lav) != 0) { fail("staticexec", "w6l failed"); };

	let o: str = testenv.readfile(exe);
	if (!iself64(o)) { fail("staticexec", "not an ELF64"); };
	// ET_EXEC == 2
	if (testenv.leu16(o, 16) != 2u64) {
		fail("staticexec", "e_type != ET_EXEC");
	};
	// static executable: no PT_INTERP anywhere in the phdr table
	let i: i32 = 0;
	for (i < phnum(o)) {
		if (testenv.leu32(o, phoff(o, i)) == 3u64) {
			fail("staticexec", "PT_INTERP present, want static");
		};
		i += 1;
	};
	testenv.clean(td);
};

@test fn readglobal() void = {
	let td: str = testenv.fresh();
	let exe: str = build(td, "rd", strings.concat(
		"TEXT _start,$0\n",
		"\tMOVQ\tcounter(SB), DI\n",
		"\tMOVQ\t$60, AX\n",
		"\tSYSCALL\n",
		"DATAW counter(SB),\"\\x2a\\x00\\x00\\x00\\x00\\x00\\x00\\x00\"\n"));
	if (runexit(td, "rd", exe) != 42) {
		fail("readglobal", "exit != 42");
	};
	testenv.clean(td);
};

@test fn writeglobal() void = {
	let td: str = testenv.fresh();
	let exe: str = build(td, "wr", strings.concat(
		"TEXT _start,$0\n",
		"\tMOVQ\t$99, AX\n",
		"\tMOVQ\tAX, counter(SB)\n",
		"\tMOVQ\tcounter(SB), DI\n",
		"\tMOVQ\t$60, AX\n",
		"\tSYSCALL\n",
		"DATAW counter(SB),\"\\x00\\x00\\x00\\x00\\x00\\x00\\x00\\x00\"\n"));
	if (runexit(td, "wr", exe) != 99) {
		fail("writeglobal", "exit != 99 (R+W mapping not writable?)");
	};
	testenv.clean(td);
};

@test fn twoloads() void = {
	let td: str = testenv.fresh();
	let exe: str = build(td, "tl", strings.concat(
		"TEXT _start,$0\n",
		"\tMOVQ\tcounter(SB), AX\n",
		"\tMOVQ\t$60, AX\n",
		"\tMOVQ\t$0, DI\n",
		"\tSYSCALL\n",
		"DATAW counter(SB),\"\\x01\\x00\\x00\\x00\\x00\\x00\\x00\\x00\"\n"));
	let o: str = testenv.readfile(exe);
	if (phnum(o) != 2) { fail("twoloads", "e_phnum != 2"); };
	let sawrx: bool = false;
	let sawrw: bool = false;
	let i: i32 = 0;
	for (i < 2) {
		let ph: i32 = phoff(o, i);
		if (testenv.leu32(o, ph) != 1u64) {
			fail("twoloads", "phdr not PT_LOAD");
		};
		let flags: u64 = testenv.leu32(o, ph + 4);
		if (flags == 5u64) { sawrx = true; }
		else if (flags == 6u64) { sawrw = true; }
		else {
			fail("twoloads", "phdr flags not exactly R+X or R+W");
		};
		i += 1;
	};
	if (!sawrx) { fail("twoloads", "missing R+X PT_LOAD"); };
	if (!sawrw) { fail("twoloads", "missing R+W PT_LOAD"); };
	testenv.clean(td);
};

@test fn nodatasingleload() void = {
	let td: str = testenv.fresh();
	let exe: str = build(td, "nd", strings.concat(
		"TEXT _start,$0\n",
		"\tMOVQ\t$60, AX\n",
		"\tMOVQ\t$0, DI\n",
		"\tSYSCALL\n"));
	let o: str = testenv.readfile(exe);
	// 992 (selfhost w6l diff) depends on the single-PT_LOAD layout.
	if (phnum(o) != 1) {
		fail("nodatasingleload", "e_phnum != 1 (legacy layout lost)");
	};
	testenv.clean(td);
};

@test fn bssfilesz() void = {
	let td: str = testenv.fresh();
	// nz = \x07 + 7 zeros, zlong = 24 zeros: memsz covers all 32B,
	// the file stops after the single non-zero byte.
	let exe: str = build(td, "bss", strings.concat(
		"TEXT _start,$0\n",
		"\tMOVQ\tnz(SB), DI\n",
		"\tMOVQ\t$60, AX\n",
		"\tSYSCALL\n",
		"DATAW nz(SB),\"\\x07\\x00\\x00\\x00\\x00\\x00\\x00\\x00\"\n",
		"DATAW zlong(SB),\"\\x00\\x00\\x00\\x00\\x00\\x00\\x00\\x00",
		"\\x00\\x00\\x00\\x00\\x00\\x00\\x00\\x00",
		"\\x00\\x00\\x00\\x00\\x00\\x00\\x00\\x00\"\n"));
	let o: str = testenv.readfile(exe);
	let sawrw: bool = false;
	let i: i32 = 0;
	for (i < phnum(o)) {
		let ph: i32 = phoff(o, i);
		if (testenv.leu32(o, ph) != 1u64) { i += 1; continue; };
		if (testenv.leu32(o, ph + 4) != 6u64) { i += 1; continue; };
		sawrw = true;
		let filesz: u64 = testenv.leu64(o, ph + 32);
		let memsz: u64 = testenv.leu64(o, ph + 40);
		if (filesz >= memsz) {
			fail("bssfilesz", "want filesz < memsz");
		};
		// the trim ignores symbol boundaries: nz's zero tail goes too
		if (filesz != 1u64) { fail("bssfilesz", "filesz != 1"); };
		if (memsz != 32u64) { fail("bssfilesz", "memsz != 32"); };
		i += 1;
	};
	if (!sawrw) { fail("bssfilesz", "no R+W PT_LOAD found"); };
	// the loader must zero-fill the BSS gap
	if (runexit(td, "bss", exe) != 7) {
		fail("bssfilesz", "exit != 7");
	};
	testenv.clean(td);
};