Hojun-Cho/ww
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

wcc: reject transitive nested-tagged widen at type_assignable (#199 α)

Browse Source 
cgen has no wrapped-slot layout — the tagged-union slot is universally
[tag:8B][payload:up_to_24B], single level. The recursive walk admitted
let r: (size|io.eof|io.error) = u for u: io.underread (transitively
in io.error.params); cg_tag_for_variant + taggedvariantindext don't
recurse, returned -1, defaulted to tag=0, and the slot read back as
variant 0 = size at runtime.

Restores SSoT inside the checker pair: is / as / match variant
lookup is already non-recursive (#198 sibling), and the LET-init /
return / assign arms now agree. Aligns DOWN to the leaner side
(rule-10 stage symmetry). ww-stricter than Hare; harec keeps the
drill at ref/harec/src/types.c:702-739 (#199b is the deferred
wrapped-slot layout port).

Pre-flight audit (drew mandate): zero transitive-widen sites in
lib/ + selfhost/ + cmd/ + examples/. No wrapper-tagged variant
(io.error, strconv.error, fmt.field) is used as a variant of a
wider union anywhere in bootstrap. Mechanical fix.

Escape hatch for callers: spread (...wrapper) inlines the wrapper's
flat variants into the parent set at parse time. Wwstage's gate
additionally preserves the recursive drill on op == TK_ELLIPSIS
because wwstage stays AST-keyed (cstage flattens at resolve_type).

771_widen_transitive: 5 rows (reject_transitive_widen,
spread_alt_widen, direct_flat_variant, branched_callee_widen,
wrapper_typed_widen). Row 2 is CS-only — wwstage's is / match on
spread-expanded variants is open-bug #190/#198.
This commit is contained in:
Hojun-Cho
2026-05-29 03:04:18 +09:00
parent fb53b4798b
commit 4d44242363
6 changed files with 407 additions and 37 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
Makefile
View File

@@ -321,6 +321,7 @@ TESTS = $(BIN)/test_smoke $(BIN)/test_lex $(BIN)/test_parse $(BIN)/test_check \
$(BIN)/test_io_types_run \
$(BIN)/test_dot_aliased_ptr \
$(BIN)/test_return_tagged_forward \
$(BIN)/test_widen_transitive \
$(BIN)/test_use_promote_alias \
$(BIN)/test_field_signed $(BIN)/test_frame_argcount \
$(BIN)/test_selfhost $(BIN)/test_w6a_ww $(BIN)/test_w6l_ww \
@@ -653,6 +654,12 @@ $(BIN)/test_return_tagged_forward: test/wcc/770_return_tagged_forward.c \
$(LIB)/libwwrt.a | $(BIN)
$(CC) $(CFLAGS) -o $@ $<
$(BIN)/test_widen_transitive: test/wcc/771_widen_transitive.c \
$(BIN)/ww $(BIN)/w6c $(BIN)/w6a $(BIN)/w6l \
$(BIN)/ww_ww $(BIN)/w6c_ww $(BIN)/w6a_ww $(BIN)/w6l_ww \
$(LIB)/libwwrt.a | $(BIN)
$(CC) $(CFLAGS) -o $@ $<
$(BIN)/test_arrlit_str_full: test/wcc/711_arrlit_str_full.c \
$(BIN)/ww $(BIN)/w6c $(BIN)/w6a $(BIN)/w6l \
$(BIN)/ww_ww $(BIN)/w6c_ww $(BIN)/w6a_ww $(BIN)/w6l_ww \

18
cmd/wcc/type.c
View File

@@ -304,8 +304,24 @@ type_assignable(Type *dst, Type *src)
Type *su = (src->kind == TY_NAMED) ? src->under : src;
if (du && du->kind == TY_TAGGED &&
!(su && su->kind == TY_TAGGED)) {
for (Tparam *p = du->params; p; p = p->next)
/* #199(α): direct variant only — no transitive drill into
* a NAMED-tagged wrapper variant. Cgen has no wrapped-slot
* layout (single-level [tag][payload]), so admitting a
* nested widen silently miscompiled to tag=0 (#199 repro
* io.underread → (size|io.eof|io.error)). ww-stricter than
* Hare; harec types.c:702-739 keeps the drill (#199b is
* the deferred wrapped-slot port). Restores SSoT with
* `is`/`as`'s non-recursive variant lookup. Callers compose
* `let inner: Wrapper = sub; let r: parent = inner;`. */
for (Tparam *p = du->params; p; p = p->next) {
Type *pu = (p->type && p->type->kind == TY_NAMED)
? p->type->under : p->type;
if (pu && pu->kind == TY_TAGGED) {
if (type_eq(p->type, src)) return 1;
continue;
}
if (type_assignable(p->type, src)) return 1;
}
return 0;
}
if (du && du->kind == TY_TAGGED &&

35
selfhost/cmd/w6c/main.combined.ww
View File

@@ -13080,21 +13080,32 @@ fn isassignable(c: *checker, dst: *node, src: *node, confident: *bool) bool = {
return true;
};
// Tagged-union variant inclusion: src is one of dst's variants.
// Recursive isassignable mirrors cstage type_assignable
// (cmd/wcc/type.c:298-299) and harec tagged_select_subtype's
// recursive type_is_assignable call (ref/harec/src/types.c:702-739,
// :718; invoked from the TAGGED arm at :1110-1112). #39 cascade:
// the prior typeeqast-only walk rejected widenings that aren't
// strict surface-eq (NAMED-aliased variants, nested tagged inside
// a variant, concrete → variant after the wrap-induced exprtype
// reshape). #55 surface-nominal fast path is preserved by the
// recursive call's leading typeeqast (line 2257). #57 bare-vs-
// qualified TNAME residual unchanged.
// #199(α): direct variant only — no transitive drill into a
// NAMED-tagged wrapper variant. Cgen has no wrapped-slot layout
// (taggedvariantindext returns -1 → tag=0 silent miscompile on
// io.underread → (size|io.eof|io.error)). ww-stricter than Hare;
// harec keeps the drill at types.c:702-739 (#199b deferred port).
// Restores SSoT with `is`/`as` non-recursive lookup (#198 sibling).
// Callers compose `let inner: Wrapper = sub; let r: parent = inner;`.
if (du.kind == nkind.N_TTAGGED && su.kind != nkind.N_TTAGGED) {
let v: *node = du.list;
for (v != nil) {
let innerconf: bool = false;
if (isassignable(c, v, src, &innerconf)) { return true; };
// Spread `...wrapper` keeps the recursive drill: the
// wrapper's flat variants are intentionally inlined into
// the parent set, and AST-level params haven't been
// expanded yet (cstage flattens at resolve_type; wwstage
// stays AST-keyed). Plain wrapper variant gets the
// direct-only gate.
let vspread: bool = (v.op == tkind.TK_ELLIPSIS);
let vu: *node = resolvealias(c, unwrapbang(v));
let vtagged: bool = false;
if (vu != nil) { if (vu.kind == nkind.N_TTAGGED) { vtagged = true; }; };
if (vtagged && !vspread) {
if (typeeqast(v, src)) { return true; };
} else {
let innerconf: bool = false;
if (isassignable(c, v, src, &innerconf)) { return true; };
};
v = v.next;
};
return false;

35
selfhost/cmd/wcc/check.ww
View File

@@ -3050,21 +3050,32 @@ fn isassignable(c: *checker, dst: *node, src: *node, confident: *bool) bool = {
return true;
};
// Tagged-union variant inclusion: src is one of dst's variants.
// Recursive isassignable mirrors cstage type_assignable
// (cmd/wcc/type.c:298-299) and harec tagged_select_subtype's
// recursive type_is_assignable call (ref/harec/src/types.c:702-739,
// :718; invoked from the TAGGED arm at :1110-1112). #39 cascade:
// the prior typeeqast-only walk rejected widenings that aren't
// strict surface-eq (NAMED-aliased variants, nested tagged inside
// a variant, concrete → variant after the wrap-induced exprtype
// reshape). #55 surface-nominal fast path is preserved by the
// recursive call's leading typeeqast (line 2257). #57 bare-vs-
// qualified TNAME residual unchanged.
// #199(α): direct variant only — no transitive drill into a
// NAMED-tagged wrapper variant. Cgen has no wrapped-slot layout
// (taggedvariantindext returns -1 → tag=0 silent miscompile on
// io.underread → (size|io.eof|io.error)). ww-stricter than Hare;
// harec keeps the drill at types.c:702-739 (#199b deferred port).
// Restores SSoT with `is`/`as` non-recursive lookup (#198 sibling).
// Callers compose `let inner: Wrapper = sub; let r: parent = inner;`.
if (du.kind == nkind.N_TTAGGED && su.kind != nkind.N_TTAGGED) {
let v: *node = du.list;
for (v != nil) {
let innerconf: bool = false;
if (isassignable(c, v, src, &innerconf)) { return true; };
// Spread `...wrapper` keeps the recursive drill: the
// wrapper's flat variants are intentionally inlined into
// the parent set, and AST-level params haven't been
// expanded yet (cstage flattens at resolve_type; wwstage
// stays AST-keyed). Plain wrapper variant gets the
// direct-only gate.
let vspread: bool = (v.op == tkind.TK_ELLIPSIS);
let vu: *node = resolvealias(c, unwrapbang(v));
let vtagged: bool = false;
if (vu != nil) { if (vu.kind == nkind.N_TTAGGED) { vtagged = true; }; };
if (vtagged && !vspread) {
if (typeeqast(v, src)) { return true; };
} else {
let innerconf: bool = false;
if (isassignable(c, v, src, &innerconf)) { return true; };
};
v = v.next;
};
return false;

35
selfhost/cmd/wwdump/main.combined.ww
View File

@@ -13080,21 +13080,32 @@ fn isassignable(c: *checker, dst: *node, src: *node, confident: *bool) bool = {
return true;
};
// Tagged-union variant inclusion: src is one of dst's variants.
// Recursive isassignable mirrors cstage type_assignable
// (cmd/wcc/type.c:298-299) and harec tagged_select_subtype's
// recursive type_is_assignable call (ref/harec/src/types.c:702-739,
// :718; invoked from the TAGGED arm at :1110-1112). #39 cascade:
// the prior typeeqast-only walk rejected widenings that aren't
// strict surface-eq (NAMED-aliased variants, nested tagged inside
// a variant, concrete → variant after the wrap-induced exprtype
// reshape). #55 surface-nominal fast path is preserved by the
// recursive call's leading typeeqast (line 2257). #57 bare-vs-
// qualified TNAME residual unchanged.
// #199(α): direct variant only — no transitive drill into a
// NAMED-tagged wrapper variant. Cgen has no wrapped-slot layout
// (taggedvariantindext returns -1 → tag=0 silent miscompile on
// io.underread → (size|io.eof|io.error)). ww-stricter than Hare;
// harec keeps the drill at types.c:702-739 (#199b deferred port).
// Restores SSoT with `is`/`as` non-recursive lookup (#198 sibling).
// Callers compose `let inner: Wrapper = sub; let r: parent = inner;`.
if (du.kind == nkind.N_TTAGGED && su.kind != nkind.N_TTAGGED) {
let v: *node = du.list;
for (v != nil) {
let innerconf: bool = false;
if (isassignable(c, v, src, &innerconf)) { return true; };
// Spread `...wrapper` keeps the recursive drill: the
// wrapper's flat variants are intentionally inlined into
// the parent set, and AST-level params haven't been
// expanded yet (cstage flattens at resolve_type; wwstage
// stays AST-keyed). Plain wrapper variant gets the
// direct-only gate.
let vspread: bool = (v.op == tkind.TK_ELLIPSIS);
let vu: *node = resolvealias(c, unwrapbang(v));
let vtagged: bool = false;
if (vu != nil) { if (vu.kind == nkind.N_TTAGGED) { vtagged = true; }; };
if (vtagged && !vspread) {
if (typeeqast(v, src)) { return true; };
} else {
let innerconf: bool = false;
if (isassignable(c, v, src, &innerconf)) { return true; };
};
v = v.next;
};
return false;

314
test/wcc/771_widen_transitive.c Normal file
View File

@@ -0,0 +1,314 @@
/*
* 771_widen_transitive — project #199 (α): cstage type_assignable +
* wwstage isassignable must REJECT transitive nested-tagged widen.
* Cgen has no wrapped-slot layout (#199b deferred future-work); the
* pre-α recursive `type_assignable(p->type, src)` walk admitted
* `let r: (size|io.eof|io.error) = u` for u: io.underread, then cgen
* silently miscompiled (cg_tag_for_variant returned -1 → tag=0 → r
* reads as variant 0 = size at runtime).
*
* Fix (cmd/wcc/type.c:308 + selfhost/cmd/wcc/check.ww:3063): the
* tagged-non-tagged arm now matches direct variants ONLY. NAMED-tagged
* wrapper variants (vu.kind == TY_TAGGED) compare nominally (type_eq /
* typeeqast); other variants keep the leaf recursion for primitive
* compat / untyped resolution. wwstage additionally preserves the
* spread-recursion path: `...wrapper` variants (op == TK_ELLIPSIS) drill
* through because the AST list hasn't pre-flattened (cstage flattens at
* resolve_type, wwstage stays AST-keyed).
*
* Restores SSoT inside the checker pair: `is` / `as` / `match` arm
* lookup is already non-recursive (#198 sibling), and the LET-init /
* return / assign arms now agree. Aligns DOWN to the leaner side
* (rule-10 stage symmetry). ww-stricter than Hare; harec keeps the
* drill at ref/harec/src/types.c:702-739 (#199b — wrapped-slot layout
* port deferred per drew).
*
* Coverage (5 rows):
* 1. reject_transitive_widen — `let r: (size|wrapper) = u` with
* u: underread, wrapper = !(underread | nomem). BOTH stages
* MUST reject at the checker. The repro.
* 2. spread_alt_widen — `let r: (size|...wrapper) = u;`. The
* `...` spread inlines wrapper's variants into the parent flat
* set, so underread becomes a DIRECT variant; widen succeeds.
* CS-only runtime: wwstage's `is`/`match` cross-spread variant
* lookup is open-bug #190/#198 (wwstage walks original AST list,
* doesn't see expanded variants); byte-id still gates the asm.
* 3. direct_flat_variant — `let r: (size|underread|nomem) = u`
* with the underlying variants flat in the parent (no wrapper).
* Regression gate: direct widen unchanged by the fix.
* 4. branched_callee_widen — fn returns (size|underread|nomem),
* runtime picks between two flat-variant returns. Verifies the
* cg_widen_tagged_store scalar branch's tag synthesis is
* unchanged for direct widens.
* 5. wrapper_typed_widen — `let r: wrapper = u` with u:
* underread (DIRECT variant of wrapper). Regression gate: the
* single-level wrapper widen is the supported escape hatch and
* MUST continue working.
*
* Per-row gates: cstage compile + runtime, wwstage compile + runtime,
* cs.s == ww.s byte-identical (rule-10). Row 1 uses
* expected_exit = -2 to indicate "build must fail" — the driver
* invocation is expected to return non-zero from the checker.
*/
#include <stdio.h>
#include <stdlib.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;
}
#define STAGE_CS 1
#define STAGE_WW 2
#define EXPECT_REJECT (-2)
struct row {
const char *label;
const char *src;
int expected_exit; /* EXPECT_REJECT means build must fail */
int stage_mask;
int byte_id; /* run cs.s == ww.s diff (works rows only) */
};
static const struct row rows[] = {
{ "reject_transitive_widen",
"package main;\n"
"type underread = !i32;\n"
"type nomem = !void;\n"
"type wrapper = !(underread | nomem);\n"
"export fn main() i32 = {\n"
" let u: underread;\n"
" let r: (size | wrapper) = u;\n"
" return 0;\n"
"};\n",
EXPECT_REJECT,
STAGE_CS | STAGE_WW, 0 },
{ "spread_alt_widen",
"package main;\n"
"type underread = !i32;\n"
"type nomem = !void;\n"
"type wrapper = !(underread | nomem);\n"
"export fn main() i32 = {\n"
" let u: underread;\n"
" let r: (size | ...wrapper) = u;\n"
" if (r is underread) { return 70; };\n"
" return 99;\n"
"};\n",
70,
STAGE_CS, 0 },
{ "direct_flat_variant",
"package main;\n"
"type underread = !i32;\n"
"type nomem = !void;\n"
"export fn main() i32 = {\n"
" let u: underread;\n"
" let r: (size | underread | nomem) = u;\n"
" if (r is underread) { return 71; };\n"
" return 99;\n"
"};\n",
71,
STAGE_CS | STAGE_WW, 1 },
{ "branched_callee_widen",
"package main;\n"
"type underread = !i32;\n"
"type nomem = !void;\n"
"fn pick(b: bool) (size | underread | nomem) = {\n"
" if (b) { let u: underread; return u; };\n"
" return 42: size;\n"
"};\n"
"export fn main() i32 = {\n"
" let r: (size | underread | nomem) = pick(true);\n"
" if (r is underread) { return 72; };\n"
" return 99;\n"
"};\n",
72,
STAGE_CS | STAGE_WW, 1 },
{ "wrapper_typed_widen",
"package main;\n"
"type underread = !i32;\n"
"type nomem = !void;\n"
"type wrapper = !(underread | nomem);\n"
"export fn main() i32 = {\n"
" let u: underread;\n"
" let r: wrapper = u;\n"
" if (r is underread) { return 73; };\n"
" return 99;\n"
"};\n",
73,
STAGE_CS | STAGE_WW, 1 },
};
static int
write_source(const char *path, const char *src)
{
FILE *f = fopen(path, "wb");
if (!f) return -1;
fputs(src, f);
fclose(f);
return 0;
}
static void
cleanup_tmp(const char *tmpdir, const char *base)
{
char p[512];
snprintf(p, sizeof p, "%s/%s.ww", tmpdir, base); unlink(p);
snprintf(p, sizeof p, "%s/%s.s", tmpdir, base); unlink(p);
snprintf(p, sizeof p, "%s/%s.o", tmpdir, base); unlink(p);
snprintf(p, sizeof p, "%s/%s.combined.ww", tmpdir, base); unlink(p);
snprintf(p, sizeof p, "%s/%s", tmpdir, base); unlink(p);
rmdir(tmpdir);
}
static int
build_via_driver(const char *driver, const char *tmpdir, const char *src)
{
char cmd[1024];
snprintf(cmd, sizeof cmd, "cd %s && timeout 180 %s build %s 2>/dev/null",
tmpdir, driver, src);
return runwait(cmd);
}
static int
run_row(const char *driver, const struct row *r, int seq)
{
char tmpdir[256], src[512], base[64], outbin[768];
snprintf(tmpdir, sizeof tmpdir, "/tmp/wt_%d_d_%d", getpid(), seq);
snprintf(src, sizeof src, "%s/main771.ww", tmpdir);
snprintf(base, sizeof base, "main771");
mkdir(tmpdir, 0755);
if (write_source(src, r->src) != 0) {
cleanup_tmp(tmpdir, base);
return -1;
}
int rc;
int br = build_via_driver(driver, tmpdir, src);
if (r->expected_exit == EXPECT_REJECT) {
/* Build must fail with a non-zero exit (checker error). */
rc = (br != 0) ? r->expected_exit : 0;
} else if (br == 0) {
snprintf(outbin, sizeof outbin, "%s/%s", tmpdir, base);
rc = runwait(outbin);
} else {
rc = -1;
}
cleanup_tmp(tmpdir, base);
return rc;
}
/* asm_byte_identical — diff cstage vs wwstage .s. Parallel trees so
* ww_ww writing intermediates next to the source doesn't clobber the
* cstage .s (CLAUDE.md rule 14 phase split). */
static int
asm_byte_identical(const char *cdrv, const char *wdrv,
const struct row *r, int seq)
{
char src[512], tdc[256], tdw[256], base[64], cs[512], ws[512];
snprintf(tdc, sizeof tdc, "/tmp/wt_%d_c_%d", getpid(), seq);
snprintf(tdw, sizeof tdw, "/tmp/wt_%d_w_%d", getpid(), seq);
snprintf(base, sizeof base, "main771");
mkdir(tdc, 0755);
mkdir(tdw, 0755);
snprintf(src, sizeof src, "%s/main771.ww", tdc);
if (write_source(src, r->src) != 0) { cleanup_tmp(tdc, base); cleanup_tmp(tdw, base); return -1; }
int rc = -1;
if (build_via_driver(cdrv, tdc, src) != 0) goto out;
snprintf(cs, sizeof cs, "%s/%s.s", tdc, base);
snprintf(src, sizeof src, "%s/main771.ww", tdw);
if (write_source(src, r->src) != 0) goto out;
if (build_via_driver(wdrv, tdw, src) != 0) goto out;
snprintf(ws, sizeof ws, "%s/%s.s", tdw, base);
FILE *fc = fopen(cs, "rb");
FILE *fw = fopen(ws, "rb");
if (fc && fw) {
rc = 0;
for (;;) {
int a = fgetc(fc);
int b = fgetc(fw);
if (a != b) { rc = -1; break; }
if (a == EOF) break;
}
}
if (fc) fclose(fc);
if (fw) fclose(fw);
out:
cleanup_tmp(tdc, base);
cleanup_tmp(tdw, base);
return rc;
}
int
main(void)
{
const char *bin = getenv("BIN");
if (!bin) bin = "out/bin";
char absbin[512];
if (bin[0] != '/') {
char cwd[256];
if (getcwd(cwd, sizeof cwd) == NULL) return 1;
snprintf(absbin, sizeof absbin, "%s/%s", cwd, bin);
bin = absbin;
}
char cdrv[640], wdrv[640];
snprintf(cdrv, sizeof cdrv, "%s/ww", bin);
snprintf(wdrv, sizeof wdrv, "%s/ww_ww", bin);
int n = (int)(sizeof rows / sizeof rows[0]);
int total = 0, fail = 0;
int wwpresent = (access(wdrv, X_OK) == 0);
int seq = 0;
for (int i = 0; i < n; i++) {
if (rows[i].stage_mask & STAGE_CS) {
total++;
int got = run_row(cdrv, &rows[i], seq++);
if (got != rows[i].expected_exit) {
fprintf(stderr,
"widen_transitive[cstage run][%s]: exit=%d want=%d\n",
rows[i].label, got, rows[i].expected_exit);
fail++;
}
}
if (wwpresent && (rows[i].stage_mask & STAGE_WW)) {
total++;
int got = run_row(wdrv, &rows[i], seq++);
if (got != rows[i].expected_exit) {
fprintf(stderr,
"widen_transitive[wwstage run][%s]: exit=%d want=%d\n",
rows[i].label, got, rows[i].expected_exit);
fail++;
}
if (rows[i].byte_id) {
total++;
if (asm_byte_identical(cdrv, wdrv, &rows[i], seq++) != 0) {
fprintf(stderr,
"widen_transitive[byte-id][%s]: cstage vs wwstage asm differs\n",
rows[i].label);
fail++;
}
}
}
}
if (!wwpresent)
fprintf(stderr, "widen_transitive: skip wwstage (no %s)\n", wdrv);
if (fail) {
fprintf(stderr, "widen_transitive: %d/%d fixtures failed\n",
fail, total);
return 1;
}
printf("widen_transitive: %d/%d ok\n", total, total);
return 0;
}