ww/cmd/w6l/main.c

/*
 *   w6l -o out [-L<dir>...] [-l<name>...] file1.o file2.o ...
 *
 * The first symbol named "_start" defined among the inputs becomes
 * the entry point. If none is found, fall back to "main".
 */
#include "l.h"
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>

/* `path` is acceptable iff it's either an archive ("!<arch>\n") or
 * an ELF file ("\x7fELF"). Distros often ship lib<name>.so as a GNU
 * ld linker script (plain text); we skip those rather than parse the
 * GROUP/INPUT directives.
 */
static int
is_linkable(const char *path)
{
	FILE *f = fopen(path, "rb");
	if (f == NULL) return 0;
	u8 magic[8] = {0};
	size_t n = fread(magic, 1, 8, f);
	fclose(f);
	if (n >= 8 && memcmp(magic, "!<arch>\n", 8) == 0) return 1;
	if (n >= 4 && memcmp(magic, "\x7f""ELF", 4) == 0) return 1;
	return 0;
}

/* Resolve -l<name> to a filesystem path by walking the libdirs we
 * collected. Order: lib<name>.so → lib<name>.so.<N> globs → lib<name>.a.
 * Skip anything that isn't a real archive or ELF (e.g. ld scripts).
 */
static const char *
resolve_lib(const char *name, char **libdirs, int n_libdirs)
{
	static char buf[1024];
	for (int i = 0; i < n_libdirs; i++) {
		snprintf(buf, sizeof buf, "%s/lib%s.so", libdirs[i], name);
		if (access(buf, 0) == 0 && is_linkable(buf)) return strdup(buf);
		for (int v = 0; v <= 8; v++) {
			snprintf(buf, sizeof buf, "%s/lib%s.so.%d",
			    libdirs[i], name, v);
			if (access(buf, 0) == 0 && is_linkable(buf))
				return strdup(buf);
		}
		snprintf(buf, sizeof buf, "%s/lib%s.a", libdirs[i], name);
		if (access(buf, 0) == 0 && is_linkable(buf)) return strdup(buf);
	}
	return NULL;
}

int
main(int argc, char **argv)
{
	const char *out = NULL;
	const char **inputs = calloc(argc, sizeof *inputs);
	int ninputs = 0;
	char **libdirs = calloc(argc, sizeof *libdirs);
	int n_libdirs = 0;
	const char **lflags = calloc(argc, sizeof *lflags);
	int n_lflags = 0;
	u64 base = 0x400000;

	for (int i = 1; i < argc; i++) {
		if (strcmp(argv[i], "-o") == 0 && i + 1 < argc) {
			out = argv[++i];
		} else if (strncmp(argv[i], "-L", 2) == 0 && argv[i][2]) {
			libdirs[n_libdirs++] = strdup(argv[i] + 2);
		} else if (strcmp(argv[i], "-L") == 0 && i + 1 < argc) {
			libdirs[n_libdirs++] = strdup(argv[++i]);
		} else if (strncmp(argv[i], "-l", 2) == 0 && argv[i][2]) {
			lflags[n_lflags++] = argv[i] + 2;
		} else if (strcmp(argv[i], "-l") == 0 && i + 1 < argc) {
			lflags[n_lflags++] = argv[++i];
		} else if (argv[i][0] == '-') {
			fprintf(stderr, "w6l: unknown flag %s\n", argv[i]);
			return 2;
		} else {
			inputs[ninputs++] = argv[i];
		}
	}
	if (out == NULL || ninputs == 0) {
		fputs("usage: w6l -o exe [-L<dir>...] [-l<name>...] "
		    "file1.o [file2.o...]\n", stderr);
		return 2;
	}

	/* Append -l-resolved files to the input list, after the .o/.a the
	 * caller passed positionally. They obey the same archive-pull
	 * semantics as a positional .a. */
	for (int i = 0; i < n_lflags; i++) {
		const char *p = resolve_lib(lflags[i], libdirs, n_libdirs);
		if (p == NULL) {
			fprintf(stderr, "w6l: cannot find -l%s\n", lflags[i]);
			return 1;
		}
		inputs[ninputs++] = p;
	}

	Lnk l = {0};
	/* Seed the symbol table with the entry point so archive pulls
	 * include the .o that defines it. Without this, a libwwrt.a
	 * containing start.o is silently skipped if no user .o
	 * references _start, and the entry falls back to main — which
	 * has no proper exit path. */
	(void)l_intern(&l, "_start");
	for (int i = 0; i < ninputs; i++) {
		if (l_load(&l, inputs[i]) != 0) return 1;
	}
	if (l_resolve(&l) != 0) return 1;
	/* Relocation is deferred to the emit functions — each path knows
	 * its own layout (text_va, data_va), and the static and dynamic
	 * paths place .data at different VAs. */

	Lsym *entry = l_lookup(&l, "_start");
	if (entry == NULL || !entry->defined) entry = l_lookup(&l, "main");
	if (entry == NULL || !entry->defined) {
		fprintf(stderr, "w6l: no _start or main symbol defined\n");
		return 1;
	}

	FILE *f = fopen(out, "wb");
	if (f == NULL) {
		fprintf(stderr, "w6l: cannot open %s\n", out);
		return 1;
	}
	int rc = l_emit_elf(&l, f, base, base + 0x1000 + entry->val);
	fclose(f);
	if (rc == 0) {
		char cmd[1024];
		snprintf(cmd, sizeof cmd, "chmod +x %s", out);
		(void)system(cmd);
	}
	free(inputs);
	return rc;
}