w6a: DATAW directive for writable .data section

First step toward top-level mutable `let`. Adds a sibling directive to DATA whose bytes land in a separate writable .data PROGBITS section (SHF_ALLOC|SHF_WRITE, STT_OBJECT) instead of .text. The section is emitted only when DATAW was used, so inputs without it produce a byte-identical .o — tests 991 (selfhost .o diff) and 995 (self-rebuild) keep passing unchanged. w6l still treats data-resident syms as undefined; that's the next step.
2026-05-12 11:35:50 +09:00
parent 922877309b
commit 1b0955c97b
7 changed files with 381 additions and 18 deletions
--- a/6
+++ b/6
@@ -210,7 +210,8 @@ $(BIN) $(LIB) $(OBJ)/wcc $(OBJ)/ww $(OBJ)/wwdump $(OBJ)/w6c $(OBJ)/w6a $(OBJ)/w6
 # ---- tests -------------------------------------------------------------
 # Each phase adds a $(BIN)/test_<name> target; the runner walks them.
 TESTS	= $(BIN)/test_smoke $(BIN)/test_lex $(BIN)/test_parse $(BIN)/test_check \
-	  $(BIN)/test_w6c $(BIN)/test_w6a $(BIN)/test_w6l $(BIN)/test_arch \
+	  $(BIN)/test_w6c $(BIN)/test_w6a $(BIN)/test_dataw $(BIN)/test_w6l \
+	  $(BIN)/test_arch \
 	  $(BIN)/test_e2e $(BIN)/test_ffi $(BIN)/test_dyn $(BIN)/test_stdlib \
 	  $(BIN)/test_at_test \
 	  $(BIN)/test_selfhost $(BIN)/test_w6a_ww $(BIN)/test_w6l_ww \
@@ -235,6 +236,9 @@ $(BIN)/test_w6c: test/wcc/400_w6c.c $(BIN)/w6c | $(BIN)
 $(BIN)/test_w6a: test/wcc/500_w6a.c $(BIN)/w6c $(BIN)/w6a | $(BIN)
 	$(CC) $(CFLAGS) -o $@ $<

+$(BIN)/test_dataw: test/wcc/510_dataw.c $(BIN)/w6a | $(BIN)
+	$(CC) $(CFLAGS) -o $@ $<
+
 $(BIN)/test_w6l: test/wcc/600_w6l.c $(BIN)/w6c $(BIN)/w6a $(BIN)/w6l | $(BIN)
 	$(CC) $(CFLAGS) -o $@ $<

--- a/cmd/w6a/a.h
+++ b/cmd/w6a/a.h
@@ -52,6 +52,7 @@ struct Asym {
 	const char *name;
 	int	defined;	/* 1 if we own its address */
 	int	is_text;	/* if 1, address is in .text */
+	int	is_data;	/* if 1, address is in .data (mutually exclusive with is_text) */
 	int	is_global;	/* exported (TEXT) */
 	u64	addr;		/* offset within section if defined */
 	int	idx;		/* ELF symtab index, filled at emit time */
@@ -81,6 +82,12 @@ struct Asm {
 	u8	*text;
 	u64	textcap, textlen;

+	/* output data section (writable). Empty unless any DATAW directive
+	 * was seen; in that case obj.c emits an extra .data PROGBITS
+	 * section with SHF_WRITE. */
+	u8	*data;
+	u64	datacap, datalen;
+
 	/* symbols */
 	Asym	*syms;
 	Areloc	*relocs;
@@ -102,6 +109,7 @@ int	a_emit_elf(Asm*, FILE *out);
 Asym	*a_intern(Asm*, const char *name);
 void	a_emit_byte(Asm*, u8);
 void	a_emit_u32(Asm*, u32);
+void	a_emit_data_byte(Asm*, u8);
 void	a_addreloc(Asm*, u64 off, int kind, Asym *s, i64 add);

 #endif
--- a/cmd/w6a/asm.c
+++ b/cmd/w6a/asm.c
@@ -54,6 +54,17 @@ a_emit_u32(Asm *a, u32 v)
 	a_emit_byte(a, (u8)((v >> 24) & 0xff));
 }

+void
+a_emit_data_byte(Asm *a, u8 b)
+{
+	if (a->datalen + 1 > a->datacap) {
+		u64 nc = a->datacap ? a->datacap * 2 : 256;
+		a->data = realloc(a->data, nc);
+		a->datacap = nc;
+	}
+	a->data[a->datalen++] = b;
+}
+
 void
 a_addreloc(Asm *a, u64 off, int kind, Asym *s, i64 add)
 {
@@ -310,6 +321,20 @@ a_encode(Asm *a)
 				a_emit_byte(a, p->bytes[i]);
 			break;
 		}
+		case A_DATAW: {
+			/* Writable variant: bytes go into .data (RW) instead
+			 * of .text. obj.c emits the extra section conditionally
+			 * on datalen > 0 so .o output stays byte-identical for
+			 * inputs that don't use DATAW. */
+			Asym *s = a_intern(a, p->to.sym);
+			s->defined = 1;
+			s->is_data = 1;
+			s->is_global = 1;
+			s->addr = a->datalen;
+			for (u64 i = 0; i < p->nbytes; i++)
+				a_emit_data_byte(a, p->bytes[i]);
+			break;
+		}
 		case A_RET:
 			a_emit_byte(a, 0xC3);
 			break;
--- a/cmd/w6a/obj.c
+++ b/cmd/w6a/obj.c
@@ -5,10 +5,15 @@
 *    [0] ELF header
 *    [1] Section .text (program bytes)
 *    [2] Section .rela.text (relocations)
- *    [3] Section .symtab
- *    [4] Section .strtab
- *    [5] Section .shstrtab
- *    [6] Section header table
+ *    [3] Section .data (writable; only present if datalen > 0)
+ *    [4] Section .symtab
+ *    [5] Section .strtab
+ *    [6] Section .shstrtab
+ *    [7] Section header table
+ *
+ * When no DATAW directive appears in the input, the .data section is
+ * omitted entirely so output stays byte-identical to the pre-DATAW
+ * format. Test 991 (selfhost .o byte-diff) depends on this.
 *
 * Symtab indices: 0 = STN_UNDEF, 1 = file (skipped), 2.. = our syms.
 * For simplicity we emit GLOBAL symbols only (no LOCAL ordering rules
@@ -33,6 +38,7 @@
 #define SHT_STRTAB	3
 #define SHT_RELA	4

+#define SHF_WRITE	0x1
 #define SHF_ALLOC	0x2
 #define SHF_EXECINSTR	0x4
 #define SHF_INFO_LINK	0x40
@@ -40,6 +46,7 @@
 #define STB_LOCAL	0
 #define STB_GLOBAL	1
 #define STT_NOTYPE	0
+#define STT_OBJECT	1
 #define STT_FUNC	2
 #define ELF64_ST_INFO(b,t)	(((b) << 4) + ((t) & 0xf))

@@ -108,9 +115,23 @@ a_emit_elf(Asm *a, FILE *f)
 	stput(&shstr, "");	/* idx 0 = empty */
 	stput(&str, "");

-	/* Section name offsets */
+	const int has_data = (a->datalen > 0);
+
+	/* Section indices.
+	 * Without data: 1=.text, 2=.rela.text, 3=.symtab, 4=.strtab, 5=.shstrtab
+	 * With data:    1=.text, 2=.rela.text, 3=.data, 4=.symtab, 5=.strtab, 6=.shstrtab
+	 */
+	const u16 SH_TEXT   = 1;
+	const u16 SH_DATA   = has_data ? 3 : 0;
+	const u16 SH_SYMTAB = has_data ? 4 : 3;
+	const u16 SH_STRTAB = has_data ? 5 : 4;
+	const u16 SH_SHSTR  = has_data ? 6 : 5;
+
+	/* Section name offsets. Append .data's name only when used so the
+	 * .shstrtab buffer stays byte-identical for the no-DATAW case. */
 	u32 shn_text	  = stput(&shstr, ".text");
 	u32 shn_rela	  = stput(&shstr, ".rela.text");
+	u32 shn_data	  = has_data ? stput(&shstr, ".data") : 0;
 	u32 shn_symtab	  = stput(&shstr, ".symtab");
 	u32 shn_strtab	  = stput(&shstr, ".strtab");
 	u32 shn_shstrtab  = stput(&shstr, ".shstrtab");
@@ -121,17 +142,22 @@ a_emit_elf(Asm *a, FILE *f)
 		bput(&sym, &z, sizeof z);
 	}

-	/* Section indices: 1=.text, 2=.rela.text, 3=.symtab, 4=.strtab, 5=.shstrtab */
-	const u16 SH_TEXT = 1;
-
-	/* Build symbols (defined = global; undefined = global UND) */
+	/* 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. */
 	int idx = 1;
 	for (Asym *s = a->syms; s; s = s->next) {
 		Sym64 e = {0};
 		e.st_name = stput(&str, s->name);
 		if (s->defined) {
-			e.st_info = ELF64_ST_INFO(STB_GLOBAL, STT_FUNC);
-			e.st_shndx = SH_TEXT;
+			if (s->is_data) {
+				e.st_info = ELF64_ST_INFO(STB_GLOBAL, STT_OBJECT);
+				e.st_shndx = SH_DATA;
+			} else {
+				e.st_info = ELF64_ST_INFO(STB_GLOBAL, STT_FUNC);
+				e.st_shndx = SH_TEXT;
+			}
 			e.st_value = s->addr;
 			e.st_size = 0;
 		} else {
@@ -155,13 +181,14 @@ a_emit_elf(Asm *a, FILE *f)
 	u64 off = sizeof(Ehdr);
 	u64 off_text = off; off += a->textlen;
 	u64 off_rela = off; off += rela.n;
+	u64 off_data = off; if (has_data) off += a->datalen;
 	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 = 6;	/* null + 5 real */
+	const int NSECT = has_data ? 7 : 6;	/* null + reals */

 	Ehdr eh = {0};
 	memcpy(eh.e_ident, "\x7f""ELF", 4);
@@ -175,11 +202,12 @@ a_emit_elf(Asm *a, FILE *f)
 	eh.e_ehsize = sizeof(Ehdr);
 	eh.e_shentsize = sizeof(Shdr);
 	eh.e_shnum = NSECT;
-	eh.e_shstrndx = 5;
+	eh.e_shstrndx = SH_SHSTR;

 	fwrite(&eh, 1, sizeof eh, f);
 	if (a->textlen) fwrite(a->text, 1, a->textlen, f);
 	fwrite(rela.p, 1, rela.n, f);
+	if (has_data) fwrite(a->data, 1, a->datalen, f);
 	fwrite(sym.p, 1, sym.n, f);
 	fwrite(str.p, 1, str.n, f);
 	fwrite(shstr.p, 1, shstr.n, f);
@@ -204,18 +232,29 @@ a_emit_elf(Asm *a, FILE *f)
 	sh.sh_flags = SHF_INFO_LINK;
 	sh.sh_offset = off_rela;
 	sh.sh_size = rela.n;
-	sh.sh_link = 3;	/* symtab */
+	sh.sh_link = SH_SYMTAB;
 	sh.sh_info = 1;	/* applies to .text */
 	sh.sh_addralign = 8;
 	sh.sh_entsize = sizeof(Rela64);
 	fwrite(&sh, 1, sizeof sh, f);

+	if (has_data) {
+		memset(&sh, 0, sizeof sh);
+		sh.sh_name = shn_data;
+		sh.sh_type = SHT_PROGBITS;
+		sh.sh_flags = SHF_ALLOC | SHF_WRITE;
+		sh.sh_offset = off_data;
+		sh.sh_size = a->datalen;
+		sh.sh_addralign = 8;
+		fwrite(&sh, 1, sizeof sh, f);
+	}
+
 	memset(&sh, 0, sizeof sh);
 	sh.sh_name = shn_symtab;
 	sh.sh_type = SHT_SYMTAB;
 	sh.sh_offset = off_sym;
 	sh.sh_size = sym.n;
-	sh.sh_link = 4;	/* strtab */
+	sh.sh_link = SH_STRTAB;
 	sh.sh_info = 1;	/* one local: STN_UNDEF */
 	sh.sh_addralign = 8;
 	sh.sh_entsize = sizeof(Sym64);
--- a/cmd/w6a/parse.c
+++ b/cmd/w6a/parse.c
@@ -120,6 +120,7 @@ opcode_lookup(const char *m)
 		{ "SYSCALL", A_SYSCALL },
 		{ "TEXT", A_TEXT },
 		{ "DATA", A_DATA },
+		{ "DATAW", A_DATAW },
 		{ NULL, 0 }
 	};
 	for (int i = 0; tab[i].m; i++)
@@ -324,8 +325,9 @@ a_parse(Asm *a)
 				prg->from.type = D_CONST;
 				prg->from.offset = a_parsenum(m, NULL);
 			}
-		} else if (op == A_DATA) {
-			/* DATA name(SB),"escaped bytes" */
+		} else if (op == A_DATA || op == A_DATAW) {
+			/* DATA  name(SB),"escaped bytes" — read-only in .text
+			 * DATAW name(SB),"escaped bytes" — writable in .data */
 			char nbuf[128] = {0};
 			int nn = 0;
 			while (*m && *m != '(' && nn < 127) nbuf[nn++] = *m++;
--- a/cmd/w6c/6.out.h
+++ b/cmd/w6c/6.out.h
@@ -40,6 +40,7 @@ enum {
 	A_NOP = 0,
 	A_TEXT,
 	A_DATA,
+	A_DATAW,	/* writable DATA: lands in .data (RW) instead of .text */
 	A_GLOBL,
 	A_END,

--- a/test/wcc/510_dataw.c
+++ b/test/wcc/510_dataw.c
@@ -0,0 +1,284 @@
+/*
+ * 510_dataw — DATAW directive smoke. Assemble a tiny .s that uses
+ * DATAW and verify the resulting .o has a writable .data section
+ * containing the right bytes, with a symbol that points to it.
+ *
+ * Also confirms the no-DATAW path stays byte-identical to the
+ * pre-DATAW format (6 sections, no .data); the selfhost .o diff in
+ * 991 depends on that invariant.
+ */
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <stdint.h>
+#include <unistd.h>
+
+#define SHT_PROGBITS	1
+#define SHT_SYMTAB	2
+#define SHF_WRITE	0x1
+#define SHF_ALLOC	0x2
+#define SHF_EXECINSTR	0x4
+
+#pragma pack(push, 1)
+typedef struct {
+	uint8_t  e_ident[16];
+	uint16_t e_type, e_machine;
+	uint32_t e_version;
+	uint64_t e_entry, e_phoff, e_shoff;
+	uint32_t e_flags;
+	uint16_t e_ehsize, e_phentsize, e_phnum, e_shentsize, e_shnum, e_shstrndx;
+} Ehdr;
+
+typedef struct {
+	uint32_t sh_name, sh_type;
+	uint64_t sh_flags, sh_addr, sh_offset, sh_size;
+	uint32_t sh_link, sh_info;
+	uint64_t sh_addralign, sh_entsize;
+} Shdr;
+
+typedef struct {
+	uint32_t st_name;
+	uint8_t  st_info, st_other;
+	uint16_t st_shndx;
+	uint64_t st_value, st_size;
+} Sym64;
+#pragma pack(pop)
+
+static int
+write_file(const char *path, const char *body)
+{
+	FILE *f = fopen(path, "wb");
+	if (!f) return -1;
+	fputs(body, f);
+	fclose(f);
+	return 0;
+}
+
+static int
+slurp(const char *path, uint8_t **out, size_t *outn)
+{
+	FILE *f = fopen(path, "rb");
+	if (!f) return -1;
+	fseek(f, 0, SEEK_END);
+	long n = ftell(f);
+	fseek(f, 0, SEEK_SET);
+	uint8_t *b = malloc((size_t)n);
+	if (fread(b, 1, (size_t)n, f) != (size_t)n) {
+		free(b);
+		fclose(f);
+		return -1;
+	}
+	fclose(f);
+	*out = b;
+	*outn = (size_t)n;
+	return 0;
+}
+
+/* Find the section with the given name. Returns index, or -1. */
+static int
+find_section(const uint8_t *buf, const char *name)
+{
+	const Ehdr *eh = (const Ehdr *)buf;
+	const Shdr *sh = (const Shdr *)(buf + eh->e_shoff);
+	const char *shstr = (const char *)(buf + sh[eh->e_shstrndx].sh_offset);
+	for (uint16_t i = 0; i < eh->e_shnum; i++) {
+		if (strcmp(shstr + sh[i].sh_name, name) == 0)
+			return (int)i;
+	}
+	return -1;
+}
+
+/* Find a symbol by name in the .symtab. Returns pointer, or NULL. */
+static const Sym64 *
+find_symbol(const uint8_t *buf, const char *name)
+{
+	const Ehdr *eh = (const Ehdr *)buf;
+	const Shdr *sh = (const Shdr *)(buf + eh->e_shoff);
+	int idx_symtab = -1;
+	for (uint16_t i = 0; i < eh->e_shnum; i++)
+		if (sh[i].sh_type == SHT_SYMTAB) { idx_symtab = i; break; }
+	if (idx_symtab < 0) return NULL;
+	int idx_strtab = (int)sh[idx_symtab].sh_link;
+	const Sym64 *syms = (const Sym64 *)(buf + sh[idx_symtab].sh_offset);
+	uint64_t nsyms = sh[idx_symtab].sh_size / sizeof(Sym64);
+	const char *str = (const char *)(buf + sh[idx_strtab].sh_offset);
+	for (uint64_t i = 1; i < nsyms; i++) {
+		if (strcmp(str + syms[i].st_name, name) == 0)
+			return &syms[i];
+	}
+	return NULL;
+}
+
+static int
+test_dataw_emits_writable_section(const char *bin)
+{
+	char src[64], obj[64];
+	snprintf(src, sizeof src, "/tmp/wwt_dataw_%d.s", getpid());
+	snprintf(obj, sizeof obj, "/tmp/wwt_dataw_%d.o", getpid());
+
+	/* Hand-crafted .s: one writable data symbol holding 8 bytes (the
+	 * little-endian encoding of the i64 42) plus a trivial TEXT so
+	 * the assembler has something in .text to anchor. */
+	const char *body =
+	    "TEXT _start(SB),$0\n"
+	    "\tMOVQ\t$0, AX\n"
+	    "\tRET\n"
+	    "DATAW counter(SB),\"\\x2a\\x00\\x00\\x00\\x00\\x00\\x00\\x00\"\n";
+	if (write_file(src, body) < 0) {
+		fprintf(stderr, "dataw: cannot write %s\n", src);
+		return -1;
+	}
+
+	char cmd[512];
+	snprintf(cmd, sizeof cmd, "%s/w6a -o %s %s", bin, obj, src);
+	if (system(cmd) != 0) {
+		fprintf(stderr, "dataw: w6a failed\n");
+		unlink(src);
+		return -1;
+	}
+
+	uint8_t *buf = NULL;
+	size_t n = 0;
+	if (slurp(obj, &buf, &n) < 0) {
+		fprintf(stderr, "dataw: cannot read .o\n");
+		unlink(src); unlink(obj);
+		return -1;
+	}
+
+	const Ehdr *eh = (const Ehdr *)buf;
+	int rc = 0;
+
+	/* Expect 7 sections: NULL, .text, .rela.text, .data, .symtab, .strtab, .shstrtab. */
+	if (eh->e_shnum != 7) {
+		fprintf(stderr, "dataw: e_shnum=%u, want 7\n", eh->e_shnum);
+		rc = -1;
+	}
+
+	int idx_data = find_section(buf, ".data");
+	if (idx_data < 0) {
+		fprintf(stderr, "dataw: .data section not found\n");
+		rc = -1;
+		goto out;
+	}
+	const Shdr *sh = (const Shdr *)(buf + eh->e_shoff);
+	const Shdr *sd = &sh[idx_data];
+	if (sd->sh_type != SHT_PROGBITS) {
+		fprintf(stderr, "dataw: .data sh_type=%u, want PROGBITS\n", sd->sh_type);
+		rc = -1;
+	}
+	if ((sd->sh_flags & (SHF_ALLOC | SHF_WRITE)) != (SHF_ALLOC | SHF_WRITE)) {
+		fprintf(stderr, "dataw: .data flags=0x%lx, want ALLOC|WRITE\n",
+		    (unsigned long)sd->sh_flags);
+		rc = -1;
+	}
+	if ((sd->sh_flags & SHF_EXECINSTR) != 0) {
+		fprintf(stderr, "dataw: .data is executable, want non-X\n");
+		rc = -1;
+	}
+	if (sd->sh_size != 8) {
+		fprintf(stderr, "dataw: .data size=%lu, want 8\n",
+		    (unsigned long)sd->sh_size);
+		rc = -1;
+	}
+
+	/* Verify the bytes themselves match the LE i64 42 we wrote. */
+	const uint8_t *db = buf + sd->sh_offset;
+	uint8_t want[8] = { 0x2a, 0, 0, 0, 0, 0, 0, 0 };
+	if (memcmp(db, want, 8) != 0) {
+		fprintf(stderr, "dataw: .data contents mismatch\n");
+		rc = -1;
+	}
+
+	const Sym64 *s = find_symbol(buf, "counter");
+	if (s == NULL) {
+		fprintf(stderr, "dataw: symbol 'counter' not in .symtab\n");
+		rc = -1;
+		goto out;
+	}
+	if ((int)s->st_shndx != idx_data) {
+		fprintf(stderr, "dataw: 'counter'.st_shndx=%u, want %d (.data)\n",
+		    s->st_shndx, idx_data);
+		rc = -1;
+	}
+	if (s->st_value != 0) {
+		fprintf(stderr, "dataw: 'counter'.st_value=%lu, want 0\n",
+		    (unsigned long)s->st_value);
+		rc = -1;
+	}
+
+out:
+	free(buf);
+	unlink(src);
+	unlink(obj);
+	return rc;
+}
+
+static int
+test_no_dataw_keeps_legacy_layout(const char *bin)
+{
+	char src[64], obj[64];
+	snprintf(src, sizeof src, "/tmp/wwt_nodataw_%d.s", getpid());
+	snprintf(obj, sizeof obj, "/tmp/wwt_nodataw_%d.o", getpid());
+
+	/* A .s without any DATAW must still produce the legacy 6-section
+	 * layout. Test 991 (selfhost .o byte-diff) relies on this. */
+	const char *body =
+	    "TEXT _start(SB),$0\n"
+	    "\tMOVQ\t$0, AX\n"
+	    "\tRET\n";
+	if (write_file(src, body) < 0) {
+		fprintf(stderr, "nodataw: cannot write %s\n", src);
+		return -1;
+	}
+
+	char cmd[512];
+	snprintf(cmd, sizeof cmd, "%s/w6a -o %s %s", bin, obj, src);
+	if (system(cmd) != 0) {
+		fprintf(stderr, "nodataw: w6a failed\n");
+		unlink(src);
+		return -1;
+	}
+
+	uint8_t *buf = NULL;
+	size_t n = 0;
+	if (slurp(obj, &buf, &n) < 0) {
+		fprintf(stderr, "nodataw: cannot read .o\n");
+		unlink(src); unlink(obj);
+		return -1;
+	}
+
+	const Ehdr *eh = (const Ehdr *)buf;
+	int rc = 0;
+	if (eh->e_shnum != 6) {
+		fprintf(stderr, "nodataw: e_shnum=%u, want 6 (legacy)\n",
+		    eh->e_shnum);
+		rc = -1;
+	}
+	if (find_section(buf, ".data") >= 0) {
+		fprintf(stderr, "nodataw: .data section present, want absent\n");
+		rc = -1;
+	}
+
+	free(buf);
+	unlink(src);
+	unlink(obj);
+	return rc;
+}
+
+int
+main(void)
+{
+	const char *bin = getenv("BIN");
+	if (!bin) bin = "out/bin";
+
+	int fail = 0;
+	if (test_dataw_emits_writable_section(bin) != 0) fail++;
+	if (test_no_dataw_keeps_legacy_layout(bin)   != 0) fail++;
+
+	if (fail) {
+		fprintf(stderr, "dataw: %d/2 subtests failed\n", fail);
+		return 1;
+	}
+	printf("dataw: 2/2 ok\n");
+	return 0;
+}