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

selfhost+test: wwstage type-info loss through struct-field N_INDEX (#37)

Browse Source 
The original #37 symptom (worker-34's `..findflag` mangle) cannot
reproduce on master — was a runtime miscompile misattributed to a
link-time issue. Investigation surfaced three real wwstage cgen
gaps in the N_INDEX-through-struct-field family, sister bugs to
#34 (N_INDEX N_IDENT-base) and #36 (primsize-default-to-8).

1. `indexbaseesz` slice-element stride defaulted to 8 for named-
   struct elements. `&opts.ptr[i]` for `opts: *[]option` computed
   MOVQ $8 instead of $24. Fix: route slice case through
   `elemsizeofc(c, innert)`; ptr-to-named-struct via structlookup.

2. `cgun TK_AMP N_INDEX` ignored N_DOT base. `&p.ptr[i]` left
   esz=8 because only N_IDENT base was handled. Mirror cgindex's
   existing N_DOT arm.

3. `nodeisstr` N_INDEX arm only walked N_IDENT bases. `cmd.argsptr[i]`
   for `argsptr: *str` returned false; pushargsrev dropped the
   .len half at call sites. Add N_DOT-base arm that walks the
   struct field's pointee.

Test 713 (struct_field_index): 3 rows × 2 stages = 6 fixtures, one
per fix shape. Runtime-only; bootstrap byte-id (995_self_rebuild)
covers cross-stage drift.

Residual: getopttest's errortable still fails through wwstage with
a slice-of-str-via-&arr[expr] miscompile. Filed as task #38.
This commit is contained in:
Hojun-Cho
2026-05-16 12:52:36 +09:00
parent 2fb594748c
commit de3bd5cc3b
6 changed files with 452 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
Makefile
View File

@@ -236,6 +236,7 @@ TESTS = $(BIN)/test_smoke $(BIN)/test_lex $(BIN)/test_parse $(BIN)/test_check \
$(BIN)/test_cgreturn_variant_zero \
$(BIN)/test_arrlit_str_full \
$(BIN)/test_redecl \
$(BIN)/test_struct_field_index \
$(BIN)/test_param_shadow_mod \
$(BIN)/test_localoff_scope \
$(BIN)/test_cast_enum_movl \
@@ -463,6 +464,12 @@ $(BIN)/test_redecl: test/wcc/712_redecl.c \
$(LIB)/libwwrt.a | $(BIN)
$(CC) $(CFLAGS) -o $@ $<
$(BIN)/test_struct_field_index: test/wcc/713_struct_field_index.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_use_promote_alias: test/wcc/699_use_promote_alias.c \
$(BIN)/ww $(BIN)/w6c $(BIN)/w6a $(BIN)/w6l \
$(LIB)/libwwrt.a | $(BIN)

74
selfhost/cmd/w6c/main.combined.ww
View File

@@ -6798,6 +6798,60 @@ fn nodeisstr(c: *cgen, n: *node) bool = {
};
};
};
// N_INDEX through a struct field: e.g. cmd.argsptr[i]
// where argsptr: *str. cgindex correctly loads the
// (ptr, len) pair via indexbaseesz; without this arm
// pushargsrev would only push AX and lose the .len.
if (base.kind == nkind.N_DOT) {
let fld: str = base.str;
if (streq(fld, "ptr")) { return false; };
if (streq(fld, "len")) { return false; };
if (streq(fld, "cap")) { return false; };
let inner: *node = base.lhs;
if (inner != nil) {
if (inner.kind == nkind.N_IDENT) {
let lc: *local = localfindnode(c, inner.str);
if (lc != nil) {
let tn: *node = lc.tnode;
let sname: str;
sname.ptr = nil; sname.len = 0;
if (tn != nil) {
if (tn.kind == nkind.N_TNAME) { sname = tn.str; };
if (tn.kind == nkind.N_TPTR) {
let pinner: *node = tn.lhs;
if (pinner != nil) {
if (pinner.kind == nkind.N_TNAME) {
sname = pinner.str;
};
};
};
};
if (sname.len > 0) {
let si: *structinfo = structlookup(c, sname);
if (si != nil) {
let fi: *fieldinfo = si.fields;
for (fi != nil) {
if (streq(fi.fname, fld)) {
let ft: *node = fi.tnode;
if (ft != nil) {
let elem: *node = nil;
let fk: nkind = ft.kind;
if (fk == nkind.N_TPTR) { elem = ft.lhs; };
if (fk == nkind.N_TSLICE) { elem = ft.lhs; };
if (fk == nkind.N_TARRAY) { elem = ft.lhs; };
if (elem != nil) {
return isstrtype(c, elem);
};
};
};
fi = fi.finext;
};
};
};
};
};
};
};
};
return false;
};
@@ -7155,7 +7209,10 @@ fn indexbaseesz(c: *cgen, base: *node) i32 = {
if (innert.kind == nkind.N_TNAME) {
if (streq(innert.str, "str")) { return 1; };
};
if (innert.kind == nkind.N_TSLICE) { return elemsizeof(innert); };
// Slice element: resolve through elemsizeofc so a slice of a
// named struct (e.g. *[]option) returns the struct stride
// instead of falling through to elemsizeof's default 8.
if (innert.kind == nkind.N_TSLICE) { return elemsizeofc(c, innert); };
return 8;
};
@@ -7186,6 +7243,13 @@ fn indexbaseesz(c: *cgen, base: *node) i32 = {
if (streq(elem.str, "str")) { return 16; };
let ps: i32 = primsize(elem.str);
if (ps > 0) { return ps; };
// Pointer to named struct: indexing
// stride is the struct slot size.
// Without this, &p.ptr[i] for p.ptr:
// *S falls through to 8 and reads
// the wrong element.
let si: *structinfo = structlookup(c, elem.str);
if (si != nil) { return si.totsize; };
};
};
return 8;
@@ -11942,7 +12006,13 @@ fn cgun(c: *cgen, n: *node) void = {
};
};
};
};
} else { if (base.kind == nkind.N_DOT) {
// `&p.ptr[i]` shape: stride is the element
// of the slice/struct-pointer field, not
// the default 8. Mirrors cgindex's N_DOT
// arm so &p.ptr[i] and p.ptr[i] agree.
esz = indexbaseesz(c, base);
};};
};
cgexpr(c, idx);
if (esz > 1) {

8
selfhost/cmd/wcc/cgenexpr.ww
View File

@@ -2403,7 +2403,13 @@ fn cgun(c: *cgen, n: *node) void = {
};
};
};
};
} else { if (base.kind == nkind.N_DOT) {
// `&p.ptr[i]` shape: stride is the element
// of the slice/struct-pointer field, not
// the default 8. Mirrors cgindex's N_DOT
// arm so &p.ptr[i] and p.ptr[i] agree.
esz = indexbaseesz(c, base);
};};
};
cgexpr(c, idx);
if (esz > 1) {

66
selfhost/cmd/wcc/cgenutil.ww
View File

@@ -555,6 +555,60 @@ fn nodeisstr(c: *cgen, n: *node) bool = {
};
};
};
// N_INDEX through a struct field: e.g. cmd.argsptr[i]
// where argsptr: *str. cgindex correctly loads the
// (ptr, len) pair via indexbaseesz; without this arm
// pushargsrev would only push AX and lose the .len.
if (base.kind == nkind.N_DOT) {
let fld: str = base.str;
if (streq(fld, "ptr")) { return false; };
if (streq(fld, "len")) { return false; };
if (streq(fld, "cap")) { return false; };
let inner: *node = base.lhs;
if (inner != nil) {
if (inner.kind == nkind.N_IDENT) {
let lc: *local = localfindnode(c, inner.str);
if (lc != nil) {
let tn: *node = lc.tnode;
let sname: str;
sname.ptr = nil; sname.len = 0;
if (tn != nil) {
if (tn.kind == nkind.N_TNAME) { sname = tn.str; };
if (tn.kind == nkind.N_TPTR) {
let pinner: *node = tn.lhs;
if (pinner != nil) {
if (pinner.kind == nkind.N_TNAME) {
sname = pinner.str;
};
};
};
};
if (sname.len > 0) {
let si: *structinfo = structlookup(c, sname);
if (si != nil) {
let fi: *fieldinfo = si.fields;
for (fi != nil) {
if (streq(fi.fname, fld)) {
let ft: *node = fi.tnode;
if (ft != nil) {
let elem: *node = nil;
let fk: nkind = ft.kind;
if (fk == nkind.N_TPTR) { elem = ft.lhs; };
if (fk == nkind.N_TSLICE) { elem = ft.lhs; };
if (fk == nkind.N_TARRAY) { elem = ft.lhs; };
if (elem != nil) {
return isstrtype(c, elem);
};
};
};
fi = fi.finext;
};
};
};
};
};
};
};
};
return false;
};
@@ -912,7 +966,10 @@ fn indexbaseesz(c: *cgen, base: *node) i32 = {
if (innert.kind == nkind.N_TNAME) {
if (streq(innert.str, "str")) { return 1; };
};
if (innert.kind == nkind.N_TSLICE) { return elemsizeof(innert); };
// Slice element: resolve through elemsizeofc so a slice of a
// named struct (e.g. *[]option) returns the struct stride
// instead of falling through to elemsizeof's default 8.
if (innert.kind == nkind.N_TSLICE) { return elemsizeofc(c, innert); };
return 8;
};
@@ -943,6 +1000,13 @@ fn indexbaseesz(c: *cgen, base: *node) i32 = {
if (streq(elem.str, "str")) { return 16; };
let ps: i32 = primsize(elem.str);
if (ps > 0) { return ps; };
// Pointer to named struct: indexing
// stride is the struct slot size.
// Without this, &p.ptr[i] for p.ptr:
// *S falls through to 8 and reads
// the wrong element.
let si: *structinfo = structlookup(c, elem.str);
if (si != nil) { return si.totsize; };
};
};
return 8;

74
selfhost/cmd/wwdump/main.combined.ww
View File

@@ -6798,6 +6798,60 @@ fn nodeisstr(c: *cgen, n: *node) bool = {
};
};
};
// N_INDEX through a struct field: e.g. cmd.argsptr[i]
// where argsptr: *str. cgindex correctly loads the
// (ptr, len) pair via indexbaseesz; without this arm
// pushargsrev would only push AX and lose the .len.
if (base.kind == nkind.N_DOT) {
let fld: str = base.str;
if (streq(fld, "ptr")) { return false; };
if (streq(fld, "len")) { return false; };
if (streq(fld, "cap")) { return false; };
let inner: *node = base.lhs;
if (inner != nil) {
if (inner.kind == nkind.N_IDENT) {
let lc: *local = localfindnode(c, inner.str);
if (lc != nil) {
let tn: *node = lc.tnode;
let sname: str;
sname.ptr = nil; sname.len = 0;
if (tn != nil) {
if (tn.kind == nkind.N_TNAME) { sname = tn.str; };
if (tn.kind == nkind.N_TPTR) {
let pinner: *node = tn.lhs;
if (pinner != nil) {
if (pinner.kind == nkind.N_TNAME) {
sname = pinner.str;
};
};
};
};
if (sname.len > 0) {
let si: *structinfo = structlookup(c, sname);
if (si != nil) {
let fi: *fieldinfo = si.fields;
for (fi != nil) {
if (streq(fi.fname, fld)) {
let ft: *node = fi.tnode;
if (ft != nil) {
let elem: *node = nil;
let fk: nkind = ft.kind;
if (fk == nkind.N_TPTR) { elem = ft.lhs; };
if (fk == nkind.N_TSLICE) { elem = ft.lhs; };
if (fk == nkind.N_TARRAY) { elem = ft.lhs; };
if (elem != nil) {
return isstrtype(c, elem);
};
};
};
fi = fi.finext;
};
};
};
};
};
};
};
};
return false;
};
@@ -7155,7 +7209,10 @@ fn indexbaseesz(c: *cgen, base: *node) i32 = {
if (innert.kind == nkind.N_TNAME) {
if (streq(innert.str, "str")) { return 1; };
};
if (innert.kind == nkind.N_TSLICE) { return elemsizeof(innert); };
// Slice element: resolve through elemsizeofc so a slice of a
// named struct (e.g. *[]option) returns the struct stride
// instead of falling through to elemsizeof's default 8.
if (innert.kind == nkind.N_TSLICE) { return elemsizeofc(c, innert); };
return 8;
};
@@ -7186,6 +7243,13 @@ fn indexbaseesz(c: *cgen, base: *node) i32 = {
if (streq(elem.str, "str")) { return 16; };
let ps: i32 = primsize(elem.str);
if (ps > 0) { return ps; };
// Pointer to named struct: indexing
// stride is the struct slot size.
// Without this, &p.ptr[i] for p.ptr:
// *S falls through to 8 and reads
// the wrong element.
let si: *structinfo = structlookup(c, elem.str);
if (si != nil) { return si.totsize; };
};
};
return 8;
@@ -11942,7 +12006,13 @@ fn cgun(c: *cgen, n: *node) void = {
};
};
};
};
} else { if (base.kind == nkind.N_DOT) {
// `&p.ptr[i]` shape: stride is the element
// of the slice/struct-pointer field, not
// the default 8. Mirrors cgindex's N_DOT
// arm so &p.ptr[i] and p.ptr[i] agree.
esz = indexbaseesz(c, base);
};};
};
cgexpr(c, idx);
if (esz > 1) {

229
test/wcc/713_struct_field_index.c Normal file
View File

@@ -0,0 +1,229 @@
/*
* 713_struct_field_index — wwstage N_INDEX through a struct-field base
* carries the element type. Three sister gaps in wwstage cgen's untyped
* AST defaulted to esz=8 / scalar-push when the base was N_DOT with the
* field a *T or []T. Surfaced by getopttest wwstage build (#37
* investigation, originally misattributed as a symbol-mangling bug).
*
* Pre-fix sites (selfhost/cmd/wcc/):
*
* 1. cgenutil.ww indexbaseesz `.ptr` pseudo-field on a slice field:
* `&p.ptr[i]` where p: *[]S fell through `elemsizeof(innert)`
* which returned 8 (named-struct element), not sizeof(S). So
* `IMULQ $8, AX` instead of `IMULQ $24, AX` — every cross-element
* addressing landed inside element 0. Fix: route slice case
* through elemsizeofc so structlookup resolves the stride.
*
* 2. cgenexpr.ww cgun TK_AMP N_INDEX with N_DOT base: only
* base.kind == N_IDENT was handled; N_DOT base left esz=8 from
* the function-top default. `&p.ptr[i]` for any p.ptr (slice ptr
* or *T field) hit this. Fix: add the N_DOT arm calling
* indexbaseesz.
*
* 3. cgenutil.ww nodeisstr N_INDEX with N_DOT base: only N_IDENT
* base was handled; `cmd.argsptr[i]` (argsptr: *str) returned
* false, so pushargsrev pushed only AX and the call's str arg
* lost its .len half to stack residue. Sister to #34's
* [N]str-through-N_IDENT fix. Fix: walk the struct field's
* element type and return isstrtype.
*
* All three are the same root pattern as #34 / #36: the untyped AST
* loses the type through a struct field, so the per-shape recognizer
* has to walk N_DOT → struct → field → element manually. A typed AST
* (#11 wwstage check pass) would replace all three with one query.
*
* Rows are runtime-only (no asm byte-id): wwstage's local layout
* legitimately differs from cstage's (slot offsets shift by 8), and
* 995_self_rebuild already covers cross-stage drift on driver corpora.
* The semantic invariant — values round-trip correctly — is what we
* pin here.
*
* row | what it pins
* -----------------------------+----------------------------------
* amp_ptr_index_struct_stride | &p.ptr[i] esz = sizeof(S)=24,
* | not 8 (fix 1+2)
* dot_ptr_index_str_arg | cmd.argsptr[i] as str arg pushes
* | both halves (fix 3)
* amp_ptr_index_str_pair | &p.ptr[i] esz=16 for *[]str —
* | sister regression-pin to fix 1
*/
#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; };
static const struct row rows[] = {
/* 1. `&p.ptr[i]` where p: *[]S, S size 24 (i32 + str). Pre-fix
* wwstage: IMULQ $8 → store landed inside element 0. Post-fix:
* IMULQ $24 → store lands in element i. Reads back element 1's
* `a` = 22 (mod 256). */
{ "amp_ptr_index_struct_stride",
"type S = struct { a: i32, b: str };\n"
"fn main() i32 = {\n"
" let buf: [3]S;\n"
" buf[0].a = 10i32; buf[0].b = \"x\";\n"
" buf[1].a = 22i32; buf[1].b = \"yy\";\n"
" buf[2].a = 30i32; buf[2].b = \"zzz\";\n"
" let xs: []S;\n"
" xs.ptr = &buf[0];\n"
" xs.len = 3;\n"
" xs.cap = 3;\n"
" let p: *[]S = &xs;\n"
" let q: *S = &p.ptr[1];\n"
" return q.a;\n"
"};\n",
22 },
/* 2. `cmd.argsptr[i]` where argsptr: *str passed as str arg.
* Pre-fix wwstage: nodeisstr returned false → call-arg push
* dropped .len → streq received .len from stack residue → no
* match. Post-fix: full str round-trip. */
{ "dot_ptr_index_str_arg",
"fn streq(a: str, b: str) i32 = {\n"
" if (a.len != b.len) { return 0i32; };\n"
" let i: i32 = 0;\n"
" for (i < a.len) {\n"
" if (a[i] != b[i]) { return 0i32; };\n"
" i += 1;\n"
" };\n"
" return 1i32;\n"
"};\n"
"type cmd = struct { argsptr: *str, argslen: i32 };\n"
"fn main() i32 = {\n"
" let xs: [2]str;\n"
" xs[0] = \"first\";\n"
" xs[1] = \"second\";\n"
" let c: cmd;\n"
" c.argsptr = &xs[0];\n"
" c.argslen = 2;\n"
" if (streq(c.argsptr[0], \"first\") == 0i32) { return 1; };\n"
" if (streq(c.argsptr[1], \"second\") == 0i32) { return 2; };\n"
" return 0;\n"
"};\n",
0 },
/* 3. `&p.ptr[i]` for *[]str — sister to row 1 with str (16B)
* elements. Reads the .len half of element 1 to assert the
* stride lands on the right slot. Pre-fix cgun N_DOT-base
* (fix 2) left esz=8 even though elemsizeof would have returned
* 16 for str directly, so `&p.ptr[1]` pointed at element 0's
* `.len` half instead of element 1's `.ptr`. element 1's .len
* = 2 ("yy"). */
{ "amp_ptr_index_str_pair",
"fn main() i32 = {\n"
" let buf: [3]str;\n"
" buf[0] = \"a\";\n"
" buf[1] = \"yy\";\n"
" buf[2] = \"zzz\";\n"
" let xs: []str;\n"
" xs.ptr = &buf[0];\n"
" xs.len = 3;\n"
" xs.cap = 3;\n"
" let p: *[]str = &xs;\n"
" let q: *str = &p.ptr[1];\n"
" return q.len: i32;\n"
"};\n",
2 },
};
static int
run_driver(const char *driver, const struct row *r, int i)
{
char src[64], tmpdir[64], cmd[1024];
snprintf(src, sizeof src, "/tmp/wcrsfi_%d_%d.ww", getpid(), i);
snprintf(tmpdir, sizeof tmpdir, "/tmp/wcrsfi_%d_d_%d", getpid(), i);
FILE *f = fopen(src, "wb");
if (!f) return -1;
fputs(r->src, f);
fclose(f);
mkdir(tmpdir, 0755);
snprintf(cmd, sizeof cmd, "cd %s && %s build %s",
tmpdir, driver, src);
if (runwait(cmd) != 0) {
fprintf(stderr, "row[%s]: build via %s failed\n",
r->label, driver);
unlink(src); rmdir(tmpdir);
return -1;
}
const char *base = strrchr(src, '/');
base = base ? base + 1 : src;
char outbin[128];
snprintf(outbin, sizeof outbin, "%s/%s", tmpdir, base);
char *dot = strrchr(outbin, '.');
if (dot && strcmp(dot, ".ww") == 0) *dot = '\0';
int got = runwait(outbin);
unlink(src); unlink(outbin); rmdir(tmpdir);
return got;
}
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];
snprintf(cdrv, sizeof cdrv, "%s/ww", bin);
char wdrv[640];
snprintf(wdrv, sizeof wdrv, "%s/ww_ww", bin);
struct { const char *name; const char *path; int gated_on_existence; }
drivers[] = {
{ "cstage", cdrv, 0 },
{ "wwstage", wdrv, 1 },
{ NULL, NULL, 0 },
};
int n = (int)(sizeof rows / sizeof rows[0]);
int total = 0, fail = 0;
for (int d = 0; drivers[d].name; d++) {
if (drivers[d].gated_on_existence
&& access(drivers[d].path, X_OK) != 0) {
fprintf(stderr, "struct_field_index: skip %s (no %s)\n",
drivers[d].name, drivers[d].path);
continue;
}
for (int i = 0; i < n; i++) {
int got = run_driver(drivers[d].path, &rows[i], i);
total++;
if (got != rows[i].want) {
fprintf(stderr,
"struct_field_index[%s] row[%s]: exit=%d want=%d\n",
drivers[d].name, rows[i].label, got,
rows[i].want);
fail++;
}
}
}
if (fail) {
fprintf(stderr,
"struct_field_index: %d/%d row(s) failed\n", fail, total);
return 1;
}
printf("struct_field_index: %d/%d ok\n", total, total);
return 0;
}