/*
 * 956_tuprecv_f64_run — runtime + byte-id regression net for #105: a
 * tuple-from-call receive corrupts the f64 word when the callee is
 * BRANCHED. Covers ALL THREE receive forms, which share the #83
 * tuple_rseq cursor and all carried the same defect:
 *   1. SINGLE-VAR   `let r = norm(); ...r.0`   (cglet 16B-tuple branch)
 *   2. DESTRUCTURE  `let (m,i) = norm()`        (N_MLET / cgmlet+tupstore)
 *   3. REASSIGN     `m,i = norm()`              (N_MASSIGN / cgmassign+tupstore)
 *
 * Root (#105, a #103-FACE-Z regression): a (f64,i64)/(i64,f64) tuple
 * returns its f64 word in X0 (the SSE return reg) and its integer word
 * in an integer reg (tuple_rseq AX/DX). All three receives spilled the
 * f64 word via MOVQ from the integer cursor — but that reg holds GARBAGE
 * (the f64 is in X0). #103-FACE-Z's field read (MOVSD slot,X0) then read
 * that garbage. The fix makes every receive spill CLASS-AWARE: an f64/f32
 * word spills `MOVSD/MOVSS X0, slot`, an integer word spills `MOVQ
 * <reg>, slot` (as before). cstage cgen.c (3 sites) + wwstage cgenstmt.ww
 * (cglet branch + the shared tupstore helper, which covers cgmlet and
 * cgmassign), identically.
 *
 * WHY BRANCHED CALLEES: a SINGLE-return callee masks the bug via
 * register coincidence — a float-literal return leaves the f64 bit
 * pattern in AX (literal materialise goes through AX), so MOVQ AX,slot
 * happens to store the right bits; and X0 stays live to the receive. A
 * BRANCHED / multi-statement callee whose f64 word is a non-literal
 * (e.g. an f64 param) has an inner CALL clobber AX, so the integer-reg
 * spill stores garbage. The bug rows below therefore all use branched
 * callees with an f64-param word — single-return rows are gate-blind to
 * #105 (this is the gate lesson the task pins).
 *
 * GATE-BLIND TO BYTE-ID ALONE: both stages are symmetric-WRONG on master
 * (both emit MOVQ AX,slot), so the cs==ww .s gate HOLDS on master for
 * the bug rows — they diverge only at RUNTIME. Each row carries BOTH
 * dimensions (modelled on 954_tuprecv_run):
 *   (a) cstage `ww build` + run, asserting the exit code — this is what
 *       catches #105 (master returns the wrong exit).
 *   (b) w6c vs w6c_ww `.s` cmp — guards rule-10 (both stages fixed
 *       identically).
 *
 * CONTROL rows pin that the fix touches nothing else: an all-integer
 * branched 2-tuple (correct pre- and post-fix; MOVQ path untouched), the
 * destructure form `let (a,b)=mk()` (cgmlet, a separate path the fix
 * does not touch), a single-return (f64,i64) field read (#103 FACE-Z
 * shape, correct pre- and post-fix), and a bare 0f64 compare (#103
 * FACE-X shape, no tuple involved).
 */
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>
#include <sys/stat.h>
#include <sys/wait.h>

static int
runwait(const char *cmd)
{
	int rc = system(cmd);
	if (rc == -1) return -1;
	if (WIFEXITED(rc)) return WEXITSTATUS(rc);
	return -1;
}

struct row { const char *label; const char *src; int want_exit; int chk_stamped; };

static const struct row rows[] = {
	/* BUG — minimal repro. norm is BRANCHED (inner issub() CALL clobbers
	 * AX) and its f64 word is the param n, not a literal. Pre-fix the
	 * receive spills MOVQ AX,slot (garbage); r.0 != 16.0 -> return 1.
	 * Post-fix MOVSD X0,slot -> 16.0 -> return 0. */
	{ "f64_i64_br",
	  "package main;\n"
	  "fn issub(n: f64) bool = { return false; };\n"
	  "fn norm(n: f64) (f64, i64) = {\n"
	  "\tif (issub(n)) { return (n*2.0, -52); };\n"
	  "\treturn (n, 0);\n"
	  "};\n"
	  "export fn main() i32 = {\n"
	  "\tconst r = norm(16.0);\n"
	  "\tconst m: f64 = r.0;\n"
	  "\tif (m != 16.0) { return 1; };\n"
	  "\treturn 0;\n"
	  "};\n", 0 },
	/* BUG — deferred read. An intervening f64 CALL clobbers X0 AFTER the
	 * receive; the read of r.0 must come from the SPILLED slot, not a
	 * stale X0. Strongest catch: proves the spill happened at receive
	 * time and survives X0 clobber. */
	{ "f64_i64_deferred",
	  "package main;\n"
	  "fn issub(n: f64) bool = { return false; };\n"
	  "fn norm(n: f64) (f64, i64) = {\n"
	  "\tif (issub(n)) { return (n*2.0, -52); };\n"
	  "\treturn (n, 0);\n"
	  "};\n"
	  "fn clob(x: f64) f64 = { return x + 1.0; };\n"
	  "export fn main() i32 = {\n"
	  "\tconst r = norm(16.0);\n"
	  "\tconst junk: f64 = clob(3.0);\n"
	  "\tconst m: f64 = r.0;\n"
	  "\tif (m != 16.0) { return 1; };\n"
	  "\tif (junk != 4.0) { return 2; };\n"
	  "\treturn 0;\n"
	  "};\n", 0 },
	/* BUG — order-swap (i64, f64): f64 is word1, spilled from DX pre-fix
	 * (garbage; the f64 is in X0). Confirms the fix is position-aware:
	 * word1's f64 -> MOVSD X0, slot+8; word0's i64 -> MOVQ AX, slot+0. */
	{ "i64_f64_br",
	  "package main;\n"
	  "fn issub(n: f64) bool = { return false; };\n"
	  "fn norm(n: f64) (i64, f64) = {\n"
	  "\tif (issub(n)) { return (-52, n*2.0); };\n"
	  "\treturn (0, n);\n"
	  "};\n"
	  "export fn main() i32 = {\n"
	  "\tconst r = norm(16.0);\n"
	  "\tconst i: i64 = r.0;\n"
	  "\tconst m: f64 = r.1;\n"
	  "\tif (m != 16.0) { return 1; };\n"
	  "\tif (i != 0) { return 2; };\n"
	  "\treturn 0;\n"
	  "};\n", 0 },
	/* BUG — DESTRUCTURE form `let (m,i)=norm()` (N_MLET / cgmlet+tupstore).
	 * f64 binding m is element 0 (cursor AX); pre-fix MOVQ AX,slot stores
	 * garbage (issub clobbered AX). Post-fix MOVSD X0,slot. m=16.0, i=0. */
	{ "destr_f64_i64_br",
	  "package main;\n"
	  "fn issub(n: f64) bool = { return false; };\n"
	  "fn norm(n: f64) (f64, i64) = {\n"
	  "\tif (issub(n)) { return (n*2.0, -52); };\n"
	  "\treturn (n, 0);\n"
	  "};\n"
	  "export fn main() i32 = {\n"
	  "\tlet (m, i) = norm(16.0);\n"
	  "\tif (m != 16.0) { return 1; };\n"
	  "\tif (i != 0) { return 2; };\n"
	  "\treturn 0;\n"
	  "};\n", 0, 1 },
	/* BUG — DESTRUCTURE order-swap `let (i,m)=norm()`, (i64,f64). f64
	 * binding m is element 1 (cursor DX); pre-fix MOVQ DX,slot garbage,
	 * post-fix MOVSD X0,slot. i=0, m=16.0. */
	{ "destr_i64_f64_br",
	  "package main;\n"
	  "fn issub(n: f64) bool = { return false; };\n"
	  "fn norm(n: f64) (i64, f64) = {\n"
	  "\tif (issub(n)) { return (-52, n*2.0); };\n"
	  "\treturn (0, n);\n"
	  "};\n"
	  "export fn main() i32 = {\n"
	  "\tlet (i, m) = norm(16.0);\n"
	  "\tif (m != 16.0) { return 1; };\n"
	  "\tif (i != 0) { return 2; };\n"
	  "\treturn 0;\n"
	  "};\n", 0, 1 },
	/* BUG — REASSIGN form `m,i = norm()` (N_MASSIGN / cgmassign+tupstore)
	 * into pre-declared slots. Same f64-element-from-X0 defect. m=16.0. */
	{ "massign_f64_i64_br",
	  "package main;\n"
	  "fn issub(n: f64) bool = { return false; };\n"
	  "fn norm(n: f64) (f64, i64) = {\n"
	  "\tif (issub(n)) { return (n*2.0, -52); };\n"
	  "\treturn (n, 0);\n"
	  "};\n"
	  "export fn main() i32 = {\n"
	  "\tlet m: f64 = 0.0;\n"
	  "\tlet i: i64 = 0;\n"
	  "\tm, i = norm(16.0);\n"
	  "\tif (m != 16.0) { return 1; };\n"
	  "\tif (i != 0) { return 2; };\n"
	  "\treturn 0;\n"
	  "};\n", 0 },
	/* BUG — REASSIGN order-swap `i,m = norm()`, (i64,f64). f64 reassign
	 * target is element 1 (cursor DX). i=0, m=16.0. */
	{ "massign_i64_f64_br",
	  "package main;\n"
	  "fn issub(n: f64) bool = { return false; };\n"
	  "fn norm(n: f64) (i64, f64) = {\n"
	  "\tif (issub(n)) { return (-52, n*2.0); };\n"
	  "\treturn (0, n);\n"
	  "};\n"
	  "export fn main() i32 = {\n"
	  "\tlet i: i64 = 0;\n"
	  "\tlet m: f64 = 0.0;\n"
	  "\ti, m = norm(16.0);\n"
	  "\tif (m != 16.0) { return 1; };\n"
	  "\tif (i != 0) { return 2; };\n"
	  "\treturn 0;\n"
	  "};\n", 0 },
	/* CONTROL — all-integer branched 2-tuple. The integer-cursor MOVQ
	 * path is untouched by the fix (e0/e1 not float), so this is correct
	 * pre- and post-fix and byte-id both ways. a=5, b=7 -> 12. */
	{ "ctl_int_br",
	  "package main;\n"
	  "fn issub(n: i64) bool = { return false; };\n"
	  "fn mk(n: i64) (i64, i64) = {\n"
	  "\tif (issub(n)) { return (n*2, -1); };\n"
	  "\treturn (n, 7);\n"
	  "};\n"
	  "export fn main() i32 = {\n"
	  "\tconst r = mk(5);\n"
	  "\tconst a: i64 = r.0;\n"
	  "\tconst b: i64 = r.1;\n"
	  "\treturn (a: i32) + (b: i32);\n"
	  "};\n", 12 },
	/* CONTROL — destructure-with-NO-f64 `let (a,b)=mk()`. The fix now
	 * touches cgmlet/tupstore, but an all-integer element takes the
	 * unchanged MOVQ path, so this stays byte-id and correct pre- and
	 * post-fix — guards that the class check doesn't perturb integers.
	 * a=5, b=7 -> 12. */
	{ "ctl_destr",
	  "package main;\n"
	  "fn issub(n: i64) bool = { return false; };\n"
	  "fn mk(n: i64) (i64, i64) = {\n"
	  "\tif (issub(n)) { return (n*2, -1); };\n"
	  "\treturn (n, 7);\n"
	  "};\n"
	  "export fn main() i32 = {\n"
	  "\tlet (a, b) = mk(5);\n"
	  "\treturn (a: i32) + (b: i32);\n"
	  "};\n", 12 },
	/* CONTROL — #103 FACE-Z single-return (f64,i64) field read. The f64
	 * word is a literal, so AX coincidentally holds its bits on master;
	 * correct pre- and post-fix (the fix changes MOVQ AX,slot -> MOVSD
	 * X0,slot but the run result is unchanged). f=2.5->2, i=7 -> 9. */
	{ "ctl_facez_single",
	  "package main;\n"
	  "fn mk() (f64, i64) = { return (2.5, 7); };\n"
	  "export fn main() i32 = {\n"
	  "\tconst t = mk();\n"
	  "\tconst f: f64 = t.0;\n"
	  "\tconst i: i64 = t.1;\n"
	  "\treturn (f: i32) + (i: i32);\n"
	  "};\n", 9 },
	/* CONTROL — #103 FACE-X bare 0f64 compare (no tuple). The fix does
	 * not touch the compare path; pure no-op guard. -> 0. */
	{ "ctl_facex_0f64",
	  "package main;\n"
	  "export fn main() i32 = {\n"
	  "\tconst z: f64 = 0.0;\n"
	  "\tif (z != 0.0) { return 1; };\n"
	  "\treturn 0;\n"
	  "};\n", 0 },
	{ NULL, NULL, 0 }
};

static int
slurp_eq(const char *a, const char *b)
{
	FILE *fa = fopen(a, "rb");
	FILE *fb = fopen(b, "rb");
	if (!fa || !fb) { if (fa) fclose(fa); if (fb) fclose(fb); return -1; }
	int rc = 0;
	for (;;) {
		int ca = fgetc(fa);
		int cb = fgetc(fb);
		if (ca != cb) { rc = -1; break; }
		if (ca == EOF) break;
	}
	fclose(fa); fclose(fb);
	return rc;
}

int
main(void)
{
	const char *bin = getenv("BIN");
	if (!bin) bin = "out/bin";
	char absbin[1024];
	if (bin[0] != '/') {
		char cwd[1024];
		if (getcwd(cwd, sizeof cwd) == NULL) return 1;
		snprintf(absbin, sizeof absbin, "%s/%s", cwd, bin);
		bin = absbin;
	}

	char w6c[1100], w6c_ww[1100];
	snprintf(w6c, sizeof w6c, "%s/w6c", bin);
	snprintf(w6c_ww, sizeof w6c_ww, "%s/w6c_ww", bin);
	if (access(w6c_ww, X_OK) != 0) {
		fprintf(stderr, "tuprecv_f64: w6c_ww missing — cannot run the "
		    "cs==ww byte-id gate\n");
		return 1;
	}

	int n = 0, fail = 0;
	for (int i = 0; rows[i].src; i++, n++) {
		char src[64];
		snprintf(src, sizeof src, "/tmp/wwtupf_%d_%d.ww", getpid(), i);
		FILE *f = fopen(src, "wb");
		if (f == NULL) { fail++; continue; }
		fputs(rows[i].src, f);
		fclose(f);

		/* (a) cstage build + run in a scratch dir. */
		char tmpdir[64];
		snprintf(tmpdir, sizeof tmpdir, "/tmp/wwtupf_%d_d_%d",
		    getpid(), i);
		mkdir(tmpdir, 0755);

		char cmd[2048];
		snprintf(cmd, sizeof cmd, "cd %s && %s/ww build %s",
		    tmpdir, bin, src);
		if (runwait(cmd) != 0) {
			fprintf(stderr, "row[%s]: cstage build failed\n",
			    rows[i].label);
			fail++;
			unlink(src); rmdir(tmpdir);
			continue;
		}

		char outbin[128];
		const char *base = strrchr(src, '/');
		base = base ? base + 1 : src;
		snprintf(outbin, sizeof outbin, "%s/%s", tmpdir, base);
		char *dot = strrchr(outbin, '.');
		if (dot && strcmp(dot, ".ww") == 0) *dot = '\0';

		int got = runwait(outbin);
		if (got != rows[i].want_exit) {
			fprintf(stderr, "row[%s]: cstage exit %d, want %d\n",
			    rows[i].label, got, rows[i].want_exit);
			fail++;
		}
		unlink(outbin); rmdir(tmpdir);

		/* (b) cs==ww byte-id gate: emit .s from both stages, cmp. */
		char cs_s[64], ws_s[64];
		snprintf(cs_s, sizeof cs_s, "/tmp/wwtupf_%d_%d_cs.s",
		    getpid(), i);
		snprintf(ws_s, sizeof ws_s, "/tmp/wwtupf_%d_%d_ww.s",
		    getpid(), i);

		snprintf(cmd, sizeof cmd, "%s -o %s %s 2>/dev/null",
		    w6c, cs_s, src);
		if (runwait(cmd) != 0) {
			fprintf(stderr, "row[%s]: w6c failed\n", rows[i].label);
			fail++; unlink(src); continue;
		}
		snprintf(cmd, sizeof cmd, "%s -o %s %s 2>/dev/null",
		    w6c_ww, ws_s, src);
		if (runwait(cmd) != 0) {
			fprintf(stderr, "row[%s]: w6c_ww failed\n",
			    rows[i].label);
			fail++; unlink(src); unlink(cs_s); continue;
		}
		if (slurp_eq(cs_s, ws_s) != 0) {
			fprintf(stderr,
			    "row[%s]: cstage/wwstage .s DIFFER (rule-10 "
			    "byte-id violation)\n", rows[i].label);
			fail++;
		}

		/* (c) #121 A-narrow stamp gate: the un-annotated float-
		 * destructure binding must now carry a checker type stamp, so
		 * w6c_ww emits no `asserttyped:` diagnostic. Non-vacuous —
		 * pre-stamp (HEAD) w6c_ww fires asserttyped on the f64 binding
		 * ident (nil n.type_). */
		if (rows[i].chk_stamped) {
			char errf[80];
			snprintf(errf, sizeof errf,
			    "/tmp/wwtupf_%d_%d_err.txt", getpid(), i);
			snprintf(cmd, sizeof cmd,
			    "%s -o /dev/null %s 2>%s", w6c_ww, src, errf);
			runwait(cmd);
			snprintf(cmd, sizeof cmd,
			    "grep -q asserttyped %s", errf);
			if (runwait(cmd) == 0) {
				fprintf(stderr, "row[%s]: w6c_ww emitted "
				    "asserttyped (destructure binding "
				    "unstamped)\n", rows[i].label);
				fail++;
			}
			unlink(errf);
		}
		unlink(src); unlink(cs_s); unlink(ws_s);
	}

	if (fail) {
		fprintf(stderr, "%d/%d tuple-receive-f64 tests failed\n",
		    fail, n);
		return 1;
	}
	printf("tuprecv_f64: %d/%d ok (cstage run + cs==ww byte-id)\n", n, n);
	return 0;
}