toolchain: banner purge + WHY-only comment sweep (rule 8)

selfhost/, cmd/, internal/ join the tree-wide sweep: every section banner dies (91 selfhost + the cmd C-style dividers -> 0); narration and stale contracts deleted (pre-#22 bundler notes, retired single-PT_LOAD and no-archive claims, superseded ABI tables); every ref/harec/qbe cite, task cite, encoding/ELF contract, and rule-10 twin pointer kept; lost lifetime/rationale lines restored where the sweep over-cut (elf_globals ownership, kwtab linear-scan). Comment- only proven: all five wwstage tool binaries byte-identical across the sweep; test-commit, test-byteid (161+1399, 0 pinned-divergent), and test-bootstrap (fixed point + 991-995 byte-id) all exit 0. The read-through banked 66 latent-bug leads (checkpoint).
2026-08-08 23:14:03 +09:00
parent 83f5956df2
commit 62b9d20383
60 changed files with 232 additions and 1045 deletions
--- a/cmd/w6a/a.h
+++ b/cmd/w6a/a.h
@@ -1,10 +1,7 @@
 /*
- * a.h — w6a-private header. Modelled on Plan 9 cmd/6a/a.h, trimmed
- * to the instruction subset that w6c emits.
- *
- * w6a is line-oriented and has no preprocessor: each non-blank, non-
- * label line is one instruction. We read the whole file into a list
- * of `Aprog`s, then encode and emit ELF64.
+ * Modelled on Plan 9 cmd/6a/a.h, trimmed to the instruction subset
+ * that w6c emits. Line-oriented, no preprocessor: each non-blank,
+ * non-label line is one instruction.
 */
 #ifndef SIX_A_H
 #define SIX_A_H
@@ -69,17 +66,14 @@ struct Areloc {
 };

 struct Asm {
-	/* parser state */
 	const char *file;
 	const char *src;
 	u64	srclen;
 	u64	pos;
 	int	line;

-	/* program list */
 	Aprog	*head, *tail;

-	/* output text section */
 	u8	*text;
 	u64	textcap, textlen;

@@ -89,7 +83,6 @@ struct Asm {
 	u8	*data;
 	u64	datacap, datalen;

-	/* symbols */
 	Asym	*syms;
 	Areloc	*relocs;

--- a/cmd/w6a/asm.c
+++ b/cmd/w6a/asm.c
@@ -1,8 +1,4 @@
 /*
- * asm.c — encode the parsed Aprog list into amd64 machine bytes,
- * appending to Asm.text. Relocations for CALL/branch targets that
- * resolve to externals are queued in Asm.relocs.
- *
 * Encoding subset: the instructions cgen emits today. Operand shapes
 * we accept:
 *   MOVQ $imm, reg           — C7 /0 imm32 (REX.W) [imm fits in i32]
@@ -95,8 +91,6 @@ a_addreloc_data(Asm *a, u64 off, int kind, Asym *s, i64 add)
 	a->relocs = r;
 }

-/* ------ register codes ------------------------------------------- */
-
 /* low 3 bits of register encoding */
 static int
 rcode(int r)
@@ -137,14 +131,12 @@ is_xmm(int r)
 	return r >= D_X0 && r <= D_X15;
 }

-/* ModR/M byte */
 static u8
 modrm(int mod, int reg, int rm)
 {
 	return (u8)(((mod & 3) << 6) | ((reg & 7) << 3) | (rm & 7));
 }

-/* emit REX with W=1 plus optional R/B for high regs */
 static void
 emit_rex(Asm *a, int regbit, int rmbit, int w)
 {
@@ -155,9 +147,7 @@ emit_rex(Asm *a, int regbit, int rmbit, int w)
 	if (b != 0x40 || w) a_emit_byte(a, b);
 }

-/* encode mod/disp for [base+disp]; returns 0 on ok.
- * Special-cases SP (needs SIB) and BP (forces disp).
- */
+/* Special-cases SP (needs SIB) and BP (forces disp). */
 static void
 emit_modrm_mem(Asm *a, int reg_field, int base, i64 disp)
 {
@@ -191,7 +181,6 @@ encode_rr(Asm *a, u8 opcode, int src, int dst)
 	a_emit_byte(a, modrm(3, rcode(src), rcode(dst)));
 }

-/* MOVQ src reg → mem(base, disp).  opcode = 0x89 */
 static void
 encode_rm(Asm *a, u8 opcode, int src_reg, int base, i64 disp)
 {
@@ -200,7 +189,6 @@ encode_rm(Asm *a, u8 opcode, int src_reg, int base, i64 disp)
 	emit_modrm_mem(a, rcode(src_reg), base, disp);
 }

-/* MOVQ mem(base, disp) → reg.  opcode = 0x8B */
 static void
 encode_mr(Asm *a, u8 opcode, int dst_reg, int base, i64 disp)
 {
@@ -250,7 +238,7 @@ sse_mr_load(Asm *a, u8 prefix, u8 op2, int reg_op, int base, i64 disp)
 	emit_modrm_mem(a, rcode(reg_op), base, disp);
 }

-/* like sse_mr_load but encoded with REX.W (used by CVTTSD2SI / CVTSI2SD
+/* like sse_rr but encoded with REX.W (used by CVTTSD2SI / CVTSI2SD
 * which target/source 64-bit integer regs) */
 static void
 sse_rr_w(Asm *a, u8 prefix, u8 op2, int reg_op, int rm_op)
@@ -262,8 +250,6 @@ sse_rr_w(Asm *a, u8 prefix, u8 op2, int reg_op, int rm_op)
 	a_emit_byte(a, modrm(3, rcode(reg_op), rcode(rm_op)));
 }

-/* ------ second-pass helper: resolve labels to addresses ---------- */
-
 static u64
 resolve_label(Asm *a, const char *name)
 {
@@ -281,8 +267,6 @@ label_defined(Asm *a, const char *name)
 	return 0;
 }

-/* ------ first pass: encode ---------------------------------------- */
-
 /* For local labels, we record a "fixup" — an offset in .text that
 * needs to be patched once the label is resolved at end of pass. */
 typedef struct Fixup Fixup;
@@ -307,10 +291,9 @@ int
 a_encode(Asm *a)
 {
 	fixups = NULL;
-	const char *cur_text = NULL;	/* current TEXT name */
+	const char *cur_text = NULL;
 	(void)cur_text;
 	for (Aprog *p = a->head; p; p = p->link) {
-		/* Define any pending label at the current PC */
 		if (p->label) {
 			Asym *s = a_intern(a, p->label);
 			s->defined = 1;
@@ -800,7 +783,6 @@ a_encode(Asm *a)
 				/* R_X86_64_PLT32 (4); addend -4 */
 				a_addreloc(a, reloff, 4, s, -4);
 			} else if (p->to.type == D_BRANCH) {
-				/* local call to a label */
 				a_emit_byte(a, 0xE8);
 				add_fixup(a->textlen, p->to.sym);
 				a_emit_u32(a, 0);
@@ -846,7 +828,6 @@ a_encode(Asm *a)
 		}
 	}

-	/* second pass: patch fixups */
 	for (Fixup *f = fixups; f; f = f->next) {
 		if (!label_defined(a, f->label)) {
 			fprintf(stderr, "w6a: undefined label '%s'\n", f->label);
--- a/cmd/w6a/lex.c
+++ b/cmd/w6a/lex.c
@@ -1,8 +1,3 @@
-/*
- * lex.c — character-level helpers for w6a's line-oriented parser.
- * The parser itself lives in parse.c; here we keep the tokenisers
- * for identifiers and numbers so parse.c stays focused on syntax.
- */
 #include "a.h"
 #include <ctype.h>
 #include <stdlib.h>
--- a/cmd/w6a/main.c
+++ b/cmd/w6a/main.c
@@ -1,6 +1,3 @@
-/*
- * w6a — amd64 assembler driver. Read .s, parse, encode, emit ELF .o.
- */
 #include "a.h"
 #include <stdio.h>
 #include <stdlib.h>
--- a/cmd/w6a/obj.c
+++ b/cmd/w6a/obj.c
@@ -1,6 +1,4 @@
 /*
- * obj.c — emit a tiny ELF64 relocatable object.
- *
 * Layout (in file order):
 *    [0] ELF header
 *    [1] Section .text (program bytes)
@@ -24,7 +22,6 @@
 #include <string.h>
 #include <stdio.h>

-/* ELF constants */
 #define ELFMAG		"\x7f""ELF"
 #define ELFCLASS64	2
 #define ELFDATA2LSB	1
@@ -54,7 +51,6 @@
 #define R_X86_64_PLT32	4
 #define ELF64_R_INFO(s,t)	(((u64)(s) << 32) | ((u64)(t) & 0xffffffff))

-/* growable byte buffer */
 typedef struct Buf Buf;
 struct Buf { u8 *p; size_t n, cap; };

@@ -157,10 +153,9 @@ a_emit_elf(Asm *a, FILE *f)
 		bput(&sym, &z, sizeof z);
 	}

-	/* Build symbols (defined = global; undefined = global UND). Data
-	 * symbols carry STT_OBJECT and st_shndx=SH_DATA; everything else
-	 * keeps the legacy STT_FUNC/SH_TEXT shape so non-DATAW outputs
-	 * stay byte-identical. */
+	/* Data symbols carry STT_OBJECT and st_shndx=SH_DATA; everything
+	 * else keeps the legacy STT_FUNC/SH_TEXT shape so non-DATAW
+	 * outputs stay byte-identical. */
 	int idx = 1;
 	for (Asym *s = a->syms; s; s = s->next) {
 		Sym64 e = {0};
@@ -193,7 +188,6 @@ a_emit_elf(Asm *a, FILE *f)
 		bput(r->section == 1 ? &relad : &rela, &re, sizeof re);
 	}

-	/* Layout offsets in the file */
 	u64 off = sizeof(Ehdr);
 	u64 off_text = off; off += a->textlen;
 	u64 off_rela = off; off += rela.n;
@@ -202,7 +196,6 @@ a_emit_elf(Asm *a, FILE *f)
 	u64 off_sym  = off; off += sym.n;
 	u64 off_str  = off; off += str.n;
 	u64 off_shstr= off; off += shstr.n;
-	/* align to 8 */
 	while (off % 8) off++;
 	u64 off_shdr = off;
 	const int NSECT = has_data
--- a/cmd/w6a/parse.c
+++ b/cmd/w6a/parse.c
@@ -1,6 +1,4 @@
 /*
- * parse.c — line-oriented parser for the asm subset emitted by w6c.
- *
 * Grammar:
 *   line := blank | comment | label | text | instr
 *   blank := /^\s*$/
@@ -52,7 +50,6 @@ a_intern(Asm *a, const char *name)
 	return s;
 }

-/* ------------------------------------------------------------------ */
 /* line iterator: returns the next line as a NUL-terminated buffer in
 *   line/llen pointers, advances pos. Returns 0 on EOF.
 */
@@ -70,7 +67,6 @@ nextline(Asm *a, char **line, size_t *llen, char *buf, size_t bufsz)
 	return 1;
 }

-/* skip leading whitespace */
 static const char *
 skipws(const char *p)
 {
@@ -168,7 +164,6 @@ parse_operand(Asm *a, const char *s, Aoperand *out)
 		return 0;
 	}

-	/* (REG) form */
 	if (*s == '(') {
 		s++;
 		char rbuf[8] = {0};
@@ -183,7 +178,6 @@ parse_operand(Asm *a, const char *s, Aoperand *out)
 		return 0;
 	}

-	/* number(REG) form, or label form, or REG */
 	const char *p = s;
 	int sign = 1;
 	if (*p == '-') { sign = -1; p++; }
@@ -208,7 +202,6 @@ parse_operand(Asm *a, const char *s, Aoperand *out)
 		return 0;
 	}

-	/* IDENT — register or symbol-or-label */
 	if (a_isidstart((unsigned char)*s)) {
 		char buf[256] = {0};
 		int n = 0;
@@ -226,7 +219,6 @@ parse_operand(Asm *a, const char *s, Aoperand *out)
 			s = end;
 		}

-		/* ID(SB) means external symbol */
 		if (*s == '(') {
 			char rbuf[8] = {0};
 			int rn = 0;
@@ -253,7 +245,6 @@ parse_operand(Asm *a, const char *s, Aoperand *out)
 			out->type = r;
 			return 0;
 		}
-		/* otherwise it's a branch target */
 		out->type = D_BRANCH;
 		out->sym = strdup(buf);
 		return 0;
@@ -281,7 +272,6 @@ a_parse(Asm *a)
 			continue;
 		}

-		/* label? */
 		if (a_isidstart((unsigned char)*p) && line[0] != '\t') {
 			const char *q = p;
 			while (a_isidcont((unsigned char)*q)) q++;
@@ -308,7 +298,6 @@ a_parse(Asm *a)
 			}
 		}

-		/* TEXT or instruction */
 		const char *m = p;
 		char mnem[16] = {0};
 		int n = 0;
@@ -332,7 +321,6 @@ a_parse(Asm *a)
 		const char *rest = m;

 		if (op == A_TEXT) {
-			/* TEXT name,$framesize */
 			char nbuf[256] = {0};
 			int nn = 0;
 			while (*m && *m != ',' && nn < 255) nbuf[nn++] = *m++;
@@ -363,7 +351,6 @@ a_parse(Asm *a)
 				prg->nbytes = 0;
 			} else {
 				m++;
-				/* parse escapes into a fresh buffer */
 				size_t cap = 32, len = 0;
 				u8 *buf = malloc(cap);
 				while (*m && *m != '"') {
@@ -398,7 +385,6 @@ a_parse(Asm *a)
 				prg->nbytes = len;
 			}
 		} else {
-			/* split rest at top-level comma */
 			const char *comma = NULL;
 			for (const char *q = rest; *q; q++)
 				if (*q == ',' && comma == NULL) comma = q;