w6c+selfhost+lib: cgen quality batch + lib Hare-shape graduation
Six fixes across the toolchain, surfaced by lib/lisp porting work.
1. f64 compound assigns (`acc += d`, `-=`, `*=`, `/=`). Both stages
load slot → X1, OP X0 into X1, store back (ADDSD/SUBSD/MULSD/
DIVSD are reg-reg only). Previous MOVSD-overwrite dropped the
OP. Locals and top-level lets.
2. Top-level `[N]u8` arrays + `&arr[i]`. let_emit_size grows a
TY_ARRAY branch so zero-init DATAW lands; cgindex / N_INDEX
store / `&base[i]` all detect a global array base and use
LEAQ name(SB) instead of LEAQ (BP). TK_AMP no longer pre-
evaluates the operand as a value-load — `&base[i]` computes
base + i*esz directly. Unblocks Hare's static-buffer pattern:
strconv.{u64,i64,f64}tos graduate to module-level `*_buf`
arrays and return owned views.
3. Cross-module `pkg.Enum.MEMBER`. Nested N_DOT chains that
don't fold to a known shape now emit `MOVQ <leaf>(SB), AX`
(mirrors the bare-IDENT unresolved fallback), so isolation
probes — and the test 990 cgen-match floor — stay consistent
across stages. strconv exposes `base` as a real `enum i32`;
callers updated. The `main` exemption (linker entry-point
keeps bare name even when not exported) mirrors C-side
collectmods into selfhost cgendecl.
4. Sum-typed parameter ABI. lib/bytes.{index,rindex} take
`(u8 | []u8)` needle; lib/strings.byteindex / rbyteindex take
`(str | rune)` needle (Hare-shaped; the byte-wise misnomer
`index` is dropped). tagged_arg_size cap bumps to 48 (6 int
regs), with a new partial-fit branch on the callee: when an
N-word tagged arg overflows remaining regs, fill what fits and
stitch the rest from positive BP offsets. scanlocals MCASE
handles slice binds (24B) and walks each arm with a saved /
restored seenmark set so two arms naming the same local each
get their own slot — matches cstage's per-arm scope reset.
5. 4-reg tagged-return ABI (AX=tag, DX=word0, CX=word1, R8=word2),
up from 3 regs. Slice-payload variants (`([]T | E)`, slot 32B)
round-trip ptr/len/cap end-to-end. Every receive site updates:
let-init via cgwidentaggedstore, match scrutinee spill, cgindex
tagged-element load (both N_IDENT and fallback bases),
pushargsrev tagged-ident arg (reads word count from slot size),
cgreturn slice variant in the shuffle path.
6. `expr: TaggedAlias` is a widening, not a re-interpret. C cgen +
selfhost cgwidentaggedstore peel an N_CAST whose destination IS
the union — so cgexpr's natural shape (str: AX=ptr, BX=len;
slice: AX=ptr, BX=len, CX=cap) is consumed by the matching
concrete-variant branch instead of being misread as a tagged
AX/DX/CX triple. Inner casts to a concrete variant (`7: i32`)
keep their type for proper tag lookup. `[N]Alias` arrays
resolve element size via slotsize + aliaslookup, and aliaslookup
strips a `pkg.` prefix so cross-module references work.
lib/fmt grows `formattable = (i64 | str | bool | rune)` plus
`printv` / `printlnv` taking an explicit `[]formattable` slice (the
receive side of Hare's `args: formattable...`). Call-site variadic
gather isn't wired — callers either hand-build the slice or compose
strconv.i64tos + strings.concat.
700_e2e: 114 → 123 rows (f64 compound, top-level u8 arrays + `&buf[i]`,
pkg.Enum.MEMBER, sum-typed (str|rune) and (u8|[]u8) params, 4-reg
slice-return ABI, formattable array). 26/26 tests, bootstrap stable
through ww4.
This commit is contained in:
@@ -1,11 +1,10 @@
|
||||
// strconv — number↔string conversions.
|
||||
//
|
||||
// Mirrors Hare's strconv:: surface. The *tos functions return a fresh
|
||||
// owned `str`; release via os.free(r.ptr, r.len: u64) when done.
|
||||
// Hare returns `const str` into a static buffer; ww allocates per
|
||||
// call because the wwstage cgen doesn't currently support mutating a
|
||||
// module-level `*u8` (so a lazy-init shared buffer isn't expressible
|
||||
// today). Graduate to the static-buffer shape once that lands.
|
||||
// Mirrors Hare's strconv:: surface. The *tos functions return a
|
||||
// `const str` view into a module-level buffer that is overwritten on
|
||||
// the next call to the same function; callers must copy the bytes if
|
||||
// they need to outlive the next invocation. See [[strings.dup]] to
|
||||
// duplicate. Matches Hare's strconv::*tos semantics.
|
||||
|
||||
use os;
|
||||
use strings;
|
||||
@@ -22,43 +21,44 @@ export type overflow = !void;
|
||||
// error — any error from a strconv call. Mirrors Hare's strconv::error.
|
||||
export type error = !(invalid | overflow);
|
||||
|
||||
// base — numeric base for parsing/formatting. Plain i32 (not a named
|
||||
// enum) because cross-module `strconv.base.DEC` chains miscompile in
|
||||
// the cstage cgen — it emits a memory load through `base(SB)` rather
|
||||
// than inlining the enum value. Hare names them as `strconv::base`
|
||||
// enum values; we expose them as module-level `def`s so callers say
|
||||
// `strconv.DEC` and the cgen inlines the immediate.
|
||||
// base — numeric base for parsing/formatting. Mirrors Hare's
|
||||
// `strconv::base` (Hare uses `enum uint`; we pick `enum i32` since
|
||||
// the underlying parse/format loops index with i32).
|
||||
//
|
||||
// HEX is HEX_UPPER; HEX_LOWER is a separate pseudo-base that produces
|
||||
// lowercase a-f digits.
|
||||
export def DEFAULT: i32 = 0;
|
||||
export def BIN: i32 = 2;
|
||||
export def OCT: i32 = 8;
|
||||
export def DEC: i32 = 10;
|
||||
export def HEX_UPPER: i32 = 16;
|
||||
export def HEX: i32 = 16;
|
||||
export def HEX_LOWER: i32 = 17;
|
||||
// HEX is an alias for HEX_UPPER; HEX_LOWER is a pseudo-base that
|
||||
// produces lowercase a-f digits.
|
||||
export type base = enum i32 {
|
||||
DEFAULT = 0,
|
||||
BIN = 2,
|
||||
OCT = 8,
|
||||
DEC = 10,
|
||||
HEX_UPPER = 16,
|
||||
HEX = 16,
|
||||
HEX_LOWER = 17,
|
||||
};
|
||||
|
||||
fn basenum(b: i32) i64 = {
|
||||
if (b == BIN) { return 2; };
|
||||
if (b == OCT) { return 8; };
|
||||
if (b == HEX) { return 16; };
|
||||
if (b == HEX_UPPER) { return 16; };
|
||||
if (b == HEX_LOWER) { return 16; };
|
||||
fn basenum(b: base) i64 = {
|
||||
if (b == base.BIN) { return 2; };
|
||||
if (b == base.OCT) { return 8; };
|
||||
if (b == base.HEX) { return 16; };
|
||||
if (b == base.HEX_UPPER) { return 16; };
|
||||
if (b == base.HEX_LOWER) { return 16; };
|
||||
return 10; // DEC and DEFAULT
|
||||
};
|
||||
|
||||
fn basedigit(d: i64, b: i32) u8 = {
|
||||
fn basedigit(d: i64, b: base) u8 = {
|
||||
if (d < 10) { return (d + 48): u8; };
|
||||
let off: i64 = d - 10;
|
||||
if (b == HEX_LOWER) { return (off + 97): u8; };
|
||||
if (b == base.HEX_LOWER) { return (off + 97): u8; };
|
||||
return (off + 65): u8;
|
||||
};
|
||||
|
||||
// u64tos — convert v to a base-b numeric string. Returns owned str;
|
||||
// release via os.free(r.ptr, r.len: u64). Mirrors Hare's
|
||||
// strconv::u64tos (Hare returns const str into a static buffer).
|
||||
export fn u64tos(v: u64, b: i32) str = {
|
||||
// u64tos — convert v to a base-b numeric string. Returns a view into
|
||||
// `u64tos_buf` which is overwritten on the next call. Matches Hare's
|
||||
// strconv::u64tos.
|
||||
let u64tos_buf: [65]u8;
|
||||
|
||||
export fn u64tos(v: u64, b: base) str = {
|
||||
let nb: u64 = basenum(b): u64;
|
||||
let tmp: [65]u8;
|
||||
let i: i32 = 0;
|
||||
@@ -70,22 +70,25 @@ export fn u64tos(v: u64, b: i32) str = {
|
||||
n = n / nb;
|
||||
i += 1;
|
||||
};
|
||||
let buf: *u8 = os.alloc(i: u64): *u8;
|
||||
let out: i32 = 0;
|
||||
for (i > 0) {
|
||||
i -= 1;
|
||||
buf[out] = tmp[i];
|
||||
u64tos_buf[out] = tmp[i];
|
||||
out += 1;
|
||||
};
|
||||
let r: str;
|
||||
r.ptr = buf;
|
||||
r.ptr = &u64tos_buf[0];
|
||||
r.len = out;
|
||||
return r;
|
||||
};
|
||||
|
||||
// i64tos — convert v to a base-b numeric string. Returns owned str;
|
||||
// release via os.free. Mirrors Hare's strconv::i64tos.
|
||||
export fn i64tos(v: i64, b: i32) str = {
|
||||
// i64tos — convert v to a base-b numeric string. Returns a view into
|
||||
// `i64tos_buf` which is overwritten on the next call. Independent
|
||||
// buffer from u64tos so i64tos's own call to u64tos doesn't clobber
|
||||
// the in-flight result. Matches Hare's strconv::i64tos.
|
||||
let i64tos_buf: [66]u8;
|
||||
|
||||
export fn i64tos(v: i64, b: base) str = {
|
||||
let neg: bool = false;
|
||||
let n: i64 = v;
|
||||
if (n < 0) { neg = true; n = -n; };
|
||||
@@ -99,37 +102,33 @@ export fn i64tos(v: i64, b: i32) str = {
|
||||
n = n / nb;
|
||||
i += 1;
|
||||
};
|
||||
let extra: i32 = 0;
|
||||
if (neg) { extra = 1; };
|
||||
let total: i32 = i + extra;
|
||||
let buf: *u8 = os.alloc(total: u64): *u8;
|
||||
let out: i32 = 0;
|
||||
if (neg) { buf[out] = 45u8; out += 1; }; // '-'
|
||||
if (neg) { i64tos_buf[out] = 45u8; out += 1; }; // '-'
|
||||
for (i > 0) {
|
||||
i -= 1;
|
||||
buf[out] = tmp[i];
|
||||
i64tos_buf[out] = tmp[i];
|
||||
out += 1;
|
||||
};
|
||||
let r: str;
|
||||
r.ptr = buf;
|
||||
r.ptr = &i64tos_buf[0];
|
||||
r.len = out;
|
||||
return r;
|
||||
};
|
||||
|
||||
export fn i32tos(v: i32, b: i32) str = { return i64tos(v: i64, b); };
|
||||
export fn i16tos(v: i16, b: i32) str = { return i64tos(v: i64, b); };
|
||||
export fn i8tos(v: i8, b: i32) str = { return i64tos(v: i64, b); };
|
||||
export fn i32tos(v: i32, b: base) str = { return i64tos(v: i64, b); };
|
||||
export fn i16tos(v: i16, b: base) str = { return i64tos(v: i64, b); };
|
||||
export fn i8tos(v: i8, b: base) str = { return i64tos(v: i64, b); };
|
||||
|
||||
export fn u32tos(v: u32, b: i32) str = { return u64tos(v: u64, b); };
|
||||
export fn u16tos(v: u16, b: i32) str = { return u64tos(v: u64, b); };
|
||||
export fn u8tos(v: u8, b: i32) str = { return u64tos(v: u64, b); };
|
||||
export fn u32tos(v: u32, b: base) str = { return u64tos(v: u64, b); };
|
||||
export fn u16tos(v: u16, b: base) str = { return u64tos(v: u64, b); };
|
||||
export fn u8tos(v: u8, b: base) str = { return u64tos(v: u64, b); };
|
||||
|
||||
// digval — value of digit byte `c` under base `b`, or -1 if not a
|
||||
// valid digit. Letters are accepted case-insensitively under HEX /
|
||||
// HEX_UPPER; only lowercase under HEX_LOWER.
|
||||
fn digval(c: u8, b: i32) i32 = {
|
||||
fn digval(c: u8, b: base) i32 = {
|
||||
if (c >= 48u8) { if (c <= 57u8) { return (c - 48u8): i32; }; };
|
||||
if (b == HEX_LOWER) {
|
||||
if (b == base.HEX_LOWER) {
|
||||
if (c >= 97u8) { if (c <= 102u8) { return ((c - 97u8) + 10u8): i32; }; };
|
||||
return -1;
|
||||
};
|
||||
@@ -142,7 +141,7 @@ fn digval(c: u8, b: i32) i32 = {
|
||||
// No locale, no whitespace, no underscores: optional leading '-' then
|
||||
// digits. Returns invalid with the offending index or overflow on
|
||||
// out-of-range.
|
||||
export fn stoi64(s: str, b: i32) (i64 | invalid | overflow) = {
|
||||
export fn stoi64(s: str, b: base) (i64 | invalid | overflow) = {
|
||||
if (s.len == 0) { return 0: invalid; };
|
||||
let i: i32 = 0;
|
||||
let neg: bool = false;
|
||||
@@ -163,7 +162,7 @@ export fn stoi64(s: str, b: i32) (i64 | invalid | overflow) = {
|
||||
};
|
||||
|
||||
// stou64 — parse unsigned base-b number. Mirrors Hare's strconv::stou64.
|
||||
export fn stou64(s: str, b: i32) (u64 | invalid | overflow) = {
|
||||
export fn stou64(s: str, b: base) (u64 | invalid | overflow) = {
|
||||
if (s.len == 0) { return 0: invalid; };
|
||||
let nb: u64 = basenum(b): u64;
|
||||
let v: u64 = 0u64;
|
||||
@@ -179,7 +178,7 @@ export fn stou64(s: str, b: i32) (u64 | invalid | overflow) = {
|
||||
return v;
|
||||
};
|
||||
|
||||
export fn stoi32(s: str, b: i32) (i32 | invalid | overflow) = {
|
||||
export fn stoi32(s: str, b: base) (i32 | invalid | overflow) = {
|
||||
let r = stoi64(s, b);
|
||||
match (r) {
|
||||
case let v: i64 => {
|
||||
@@ -193,7 +192,7 @@ export fn stoi32(s: str, b: i32) (i32 | invalid | overflow) = {
|
||||
return 0: invalid; // unreachable; appeases the path-cov checker
|
||||
};
|
||||
|
||||
export fn stoi16(s: str, b: i32) (i16 | invalid | overflow) = {
|
||||
export fn stoi16(s: str, b: base) (i16 | invalid | overflow) = {
|
||||
let r = stoi64(s, b);
|
||||
match (r) {
|
||||
case let v: i64 => {
|
||||
@@ -207,7 +206,7 @@ export fn stoi16(s: str, b: i32) (i16 | invalid | overflow) = {
|
||||
return 0: invalid;
|
||||
};
|
||||
|
||||
export fn stoi8(s: str, b: i32) (i8 | invalid | overflow) = {
|
||||
export fn stoi8(s: str, b: base) (i8 | invalid | overflow) = {
|
||||
let r = stoi64(s, b);
|
||||
match (r) {
|
||||
case let v: i64 => {
|
||||
@@ -221,7 +220,7 @@ export fn stoi8(s: str, b: i32) (i8 | invalid | overflow) = {
|
||||
return 0: invalid;
|
||||
};
|
||||
|
||||
export fn stou32(s: str, b: i32) (u32 | invalid | overflow) = {
|
||||
export fn stou32(s: str, b: base) (u32 | invalid | overflow) = {
|
||||
let r = stou64(s, b);
|
||||
match (r) {
|
||||
case let v: u64 => {
|
||||
@@ -234,7 +233,7 @@ export fn stou32(s: str, b: i32) (u32 | invalid | overflow) = {
|
||||
return 0: invalid;
|
||||
};
|
||||
|
||||
export fn stou16(s: str, b: i32) (u16 | invalid | overflow) = {
|
||||
export fn stou16(s: str, b: base) (u16 | invalid | overflow) = {
|
||||
let r = stou64(s, b);
|
||||
match (r) {
|
||||
case let v: u64 => {
|
||||
@@ -247,7 +246,7 @@ export fn stou16(s: str, b: i32) (u16 | invalid | overflow) = {
|
||||
return 0: invalid;
|
||||
};
|
||||
|
||||
export fn stou8(s: str, b: i32) (u8 | invalid | overflow) = {
|
||||
export fn stou8(s: str, b: base) (u8 | invalid | overflow) = {
|
||||
let r = stou64(s, b);
|
||||
match (r) {
|
||||
case let v: u64 => {
|
||||
@@ -283,13 +282,14 @@ export fn stou8(s: str, b: i32) (u8 | invalid | overflow) = {
|
||||
// the ww-side wwdump currently skips TK_FLOAT.fval while the C side
|
||||
// %g-formats it. Same trick lib/ww/lex/lex.ww's parsef64 uses:
|
||||
// build f64 constants via int-to-f64 casts.
|
||||
let f64tos_buf: [64]u8;
|
||||
|
||||
export fn f64tos(v: f64) str = {
|
||||
let tmp: [64]u8;
|
||||
let out: i32 = 0;
|
||||
let f: f64 = v;
|
||||
let zero: f64 = 0: f64;
|
||||
if (f < zero) {
|
||||
tmp[out] = 45u8; // '-'
|
||||
f64tos_buf[out] = 45u8; // '-'
|
||||
out += 1;
|
||||
f = -f;
|
||||
};
|
||||
@@ -299,12 +299,9 @@ export fn f64tos(v: f64) str = {
|
||||
if (f >= cap) {
|
||||
let s: str = "huge";
|
||||
let k: i32 = 0;
|
||||
for (k < s.len) { tmp[out] = s[k]; out += 1; k += 1; };
|
||||
let buf: *u8 = os.alloc(out: u64): *u8;
|
||||
let q: i32 = 0;
|
||||
for (q < out) { buf[q] = tmp[q]; q += 1; };
|
||||
for (k < s.len) { f64tos_buf[out] = s[k]; out += 1; k += 1; };
|
||||
let r: str;
|
||||
r.ptr = buf;
|
||||
r.ptr = &f64tos_buf[0];
|
||||
r.len = out;
|
||||
return r;
|
||||
};
|
||||
@@ -323,32 +320,27 @@ export fn f64tos(v: f64) str = {
|
||||
ip += 1;
|
||||
fp = 0;
|
||||
};
|
||||
let intstr: str = i64tos(ip, DEC);
|
||||
let intstr: str = i64tos(ip, base.DEC);
|
||||
let k: i32 = 0;
|
||||
for (k < intstr.len) { tmp[out] = intstr.ptr[k]; out += 1; k += 1; };
|
||||
os.free(intstr.ptr: *void, intstr.len: u64);
|
||||
for (k < intstr.len) { f64tos_buf[out] = intstr.ptr[k]; out += 1; k += 1; };
|
||||
if (fp != 0) {
|
||||
tmp[out] = 46u8; // '.'
|
||||
f64tos_buf[out] = 46u8; // '.'
|
||||
out += 1;
|
||||
let fracstr: str = u64tos(fp: u64, DEC);
|
||||
let fracstr: str = u64tos(fp: u64, base.DEC);
|
||||
// Pad fractional to 6 digits with leading zeros (e.g. 0.05 →
|
||||
// fp=50000, fracstr="50000", pad one '0' before).
|
||||
let z: i32 = 6 - fracstr.len;
|
||||
for (z > 0) { tmp[out] = 48u8; out += 1; z -= 1; };
|
||||
for (z > 0) { f64tos_buf[out] = 48u8; out += 1; z -= 1; };
|
||||
k = 0;
|
||||
for (k < fracstr.len) { tmp[out] = fracstr.ptr[k]; out += 1; k += 1; };
|
||||
os.free(fracstr.ptr: *void, fracstr.len: u64);
|
||||
for (k < fracstr.len) { f64tos_buf[out] = fracstr.ptr[k]; out += 1; k += 1; };
|
||||
// Trim trailing zeros in the fractional part.
|
||||
for (out > 0) {
|
||||
if (tmp[out - 1] != 48u8) { break; };
|
||||
if (f64tos_buf[out - 1] != 48u8) { break; };
|
||||
out -= 1;
|
||||
};
|
||||
};
|
||||
let buf: *u8 = os.alloc(out: u64): *u8;
|
||||
let q: i32 = 0;
|
||||
for (q < out) { buf[q] = tmp[q]; q += 1; };
|
||||
let r: str;
|
||||
r.ptr = buf;
|
||||
r.ptr = &f64tos_buf[0];
|
||||
r.len = out;
|
||||
return r;
|
||||
};
|
||||
|
||||
Reference in New Issue
Block a user