ld: resolve: search all images for symbols, rather than just direct links

2026-03-22 19:05:47 +00:00
parent 44cd6c551a
commit 3e0fdb616c
5 changed files with 251 additions and 47 deletions
--- a/sys/ld/elf.c
+++ b/sys/ld/elf.c
@@ -1,5 +1,7 @@
 #include "elf.h"

+#include "image-list.h"
+#include "ld.h"
 #include "resolve.h"

 #include <errno.h>
@@ -181,6 +183,7 @@ static int map_image(struct elf_image *image)
 		}
 	}

+	kern_logf("ld: image %s -> %zx", image->e_leaf.l_name, image->e_base);
 	return SUCCESS;
 }

@@ -297,7 +300,7 @@ static int do_rela(struct elf_image *image, elf_rela_t *rela, bool lazy)
 			rela->r_addend);
 		break;
 	default:
-		kern_log("Unknown relocation type");
+		kern_trace("Unknown relocation type");
 		return ENOEXEC;
 	}

@@ -644,7 +647,7 @@ int elf_image_link(struct elf_image *img)
 	return SUCCESS;
 }

-extern int elf_image_collect_dependencies(
+int elf_image_collect_dependencies(
 	struct elf_image *img,
 	struct image_list *dest)
 {
@@ -652,6 +655,7 @@ extern int elf_image_collect_dependencies(
 		return SUCCESS;
 	}

+	kern_tracef("collecting dependencies for %s", img->e_leaf.l_name);
 	int nr_added = 0;
 	img->e_links = calloc(img->e_nr_links, sizeof(struct elf_image *));

@@ -663,24 +667,45 @@ extern int elf_image_collect_dependencies(
 		const char *name = (const char *)img->e_base + img->e_strtab
 		                 + img->e_dyn[i].d_un.d_val;

-		if (image_list_get(dest, name)) {
-			continue;
-		}
-
 		struct elf_image *dep = NULL;
-		int status = create_image_with_name(name, &dep);
-		if (status != SUCCESS) {
-			return -status;
+		struct image_list_leaf *leaf = NULL;
+		if ((leaf = image_list_get(dest, name))) {
+			dep = QUEUE_CONTAINER(struct elf_image, e_leaf, leaf);
+		} else {
+			int status = create_image_with_name(name, &dep);
+			if (status != SUCCESS) {
+				return -status;
+			}
+
+			image_list_put(dest, &dep->e_leaf);
 		}

-		image_list_put(dest, &dep->e_leaf);
 		img->e_links[nr_added] = dep;
+		kern_tracef(
+			"%s dependency %u = %s",
+			img->e_leaf.l_name,
+			nr_added,
+			dep->e_leaf.l_name);
 		nr_added++;
 	}

 	return nr_added;
 }

+int elf_image_add_dependency(struct elf_image *img, struct elf_image *dep)
+{
+	struct elf_image **deps = realloc(
+		img->e_links,
+		sizeof(struct elf_image *) * img->e_nr_links + 1);
+	if (!deps) {
+		return ENOMEM;
+	}
+
+	deps[img->e_nr_links++] = dep;
+	img->e_links = deps;
+	return SUCCESS;
+}
+
 void elf_image_close(struct elf_image *image)
 {
 	if (image->e_fd) {
@@ -714,7 +739,7 @@ static uint32_t gnu_hash(const char *name)
 	return h;
 }

-static virt_addr_t find_symbol_stdhash(
+static const elf_sym_t *find_symbol_stdhash(
 	struct elf_image *img,
 	const char *name,
 	uint32_t hash)
@@ -733,15 +758,22 @@ static virt_addr_t find_symbol_stdhash(
 	const uint32_t *chain = &bucket[nbucket];

 	for (uint32_t i = bucket[hash % nbucket]; i; i = chain[i]) {
-		if (strcmp(name, strtab + symtab[i].st_name) == 0) {
-			return img->e_base + symtab[i].st_value;
+		const elf_sym_t *sym = &symtab[i];
+		if (strcmp(name, strtab + sym->st_name) != 0) {
+			continue;
 		}
+
+		if (!sym->st_value) {
+			continue;
+		}
+
+		return sym;
 	}

 	return 0;
 }

-static virt_addr_t find_symbol_gnuhash(
+static const elf_sym_t *find_symbol_gnuhash(
 	struct elf_image *img,
 	const char *name,
 	uint32_t hash)
@@ -749,12 +781,14 @@ static virt_addr_t find_symbol_gnuhash(
 	return 0;
 }

-static virt_addr_t find_symbol_slow(struct elf_image *img, const char *name)
+static const elf_sym_t *find_symbol_slow(
+	struct elf_image *img,
+	const char *name)
 {
 	return 0;
 }

-static virt_addr_t find_symbol(
+static const elf_sym_t *find_symbol(
 	struct elf_image *img,
 	const char *name,
 	uint32_t std_hash,
@@ -770,31 +804,144 @@ static virt_addr_t find_symbol(
 	}
 }

-virt_addr_t elf_image_find_symbol(struct elf_image *img, const char *name)
+int elf_image_find_symbol(
+	struct elf_image *img,
+	const char *name,
+	struct elf_symbol *out)
 {
 	uint32_t std_hash_val = std_hash(name);
 	uint32_t gnu_hash_val = gnu_hash(name);
-	return find_symbol(img, name, std_hash_val, gnu_hash_val);
+	const elf_sym_t *sym
+		= find_symbol(img, name, std_hash_val, gnu_hash_val);
+
+	if (sym) {
+		out->sym_container = img;
+		out->sym_info = sym;
+		return SUCCESS;
+	}
+
+	return ENOENT;
 }

-virt_addr_t elf_image_find_linked_symbol(
+int elf_image_find_linked_symbol(
 	struct elf_image *img,
-	const char *name)
+	const char *name,
+	struct elf_symbol *out)
 {
 	uint32_t std_hash_val = std_hash(name);
 	uint32_t gnu_hash_val = gnu_hash(name);

-	virt_addr_t sym = 0;
+	struct elf_image *candidate_container = NULL;
+	const elf_sym_t *candidate = NULL;
+	const elf_sym_t *sym = NULL;
+
+	kern_tracef("searching %p (%zx)", img, img->e_base);
+	sym = find_symbol(img, name, std_hash_val, gnu_hash_val);
+	if (sym) {
+		if (ELF64_ST_BIND(sym->st_info) == STB_WEAK) {
+			kern_tracef(
+				"found %s:%s (weak): %zx",
+				img->e_leaf.l_name,
+				name,
+				sym->st_value);
+			candidate = sym;
+			candidate_container = img;
+		} else {
+			kern_tracef(
+				"found %s:%s (strong): %zx",
+				img->e_leaf.l_name,
+				name,
+				sym->st_value);
+			out->sym_container = img;
+			out->sym_info = sym;
+			return SUCCESS;
+		}
+	}
+
 	for (size_t i = 0; i < img->e_nr_links; i++) {
+		kern_tracef(
+			"searching %p (%zx)",
+			img->e_links[i],
+			img->e_links[i]->e_base);
 		sym = find_symbol(
 			img->e_links[i],
 			name,
 			std_hash_val,
 			gnu_hash_val);
-		if (sym) {
-			break;
+		if (!sym) {
+			continue;
+		}
+
+		if (ELF64_ST_BIND(sym->st_info) == STB_WEAK) {
+			kern_tracef(
+				"found %s:%s (weak): %zx",
+				img->e_links[i]->e_leaf.l_name,
+				name,
+				sym->st_value);
+			candidate = sym;
+			candidate_container = img->e_links[i];
+		} else {
+			kern_tracef(
+				"found %s:%s (strong): %zx",
+				img->e_links[i]->e_leaf.l_name,
+				name,
+				sym->st_value);
+			out->sym_container = img;
+			out->sym_info = sym;
+			return SUCCESS;
 		}
 	}

-	return sym;
+	if (candidate) {
+		out->sym_container = candidate_container;
+		out->sym_info = candidate;
+		return SUCCESS;
+	}
+
+	return ENOENT;
+}
+
+virt_addr_t find_global_symbol(const char *name)
+{
+	struct image_list *images = global_image_list();
+	if (!images) {
+		return 0;
+	}
+
+	struct image_list_iterator it;
+	image_list_iterator_begin(&it, images);
+
+	virt_addr_t candidate = 0;
+	struct elf_symbol sym = {0};
+	while (it.it_leaf) {
+		struct elf_image *image
+			= QUEUE_CONTAINER(struct elf_image, e_leaf, it.it_leaf);
+		int status = elf_image_find_symbol(image, name, &sym);
+		if (status == ENOENT) {
+			image_list_iterator_move_next(&it);
+			continue;
+		}
+
+		if (ELF64_ST_BIND(sym.sym_info->st_info) == STB_WEAK) {
+			kern_tracef(
+				"found %s:%s (weak): %zx",
+				sym.sym_container->e_leaf.l_name,
+				name,
+				sym.sym_info->st_value);
+			candidate = (virt_addr_t)sym.sym_container->e_base
+			          + sym.sym_info->st_value;
+		} else {
+			kern_tracef(
+				"found %s:%s (strong): %zx",
+				sym.sym_container->e_leaf.l_name,
+				name,
+				sym.sym_info->st_value);
+			return (virt_addr_t)sym.sym_container->e_base
+			     + sym.sym_info->st_value;
+		}
+
+		image_list_iterator_move_next(&it);
+	}
+
+	return candidate;
 }
--- a/sys/ld/elf.h
+++ b/sys/ld/elf.h
@@ -344,6 +344,11 @@ struct elf_image {
 	size_t e_nr_links;
 };

+struct elf_symbol {
+	struct elf_image *sym_container;
+	const elf_sym_t *sym_info;
+};
+
 extern const char *elf_image_status_to_string(enum elf_image_status status);

 extern int elf_image_open(const char *path, struct elf_image **out);
@@ -352,14 +357,25 @@ extern int elf_image_load(struct elf_image *img);
 extern int elf_image_collect_dependencies(
 	struct elf_image *img,
 	struct image_list *dest);
+extern int elf_image_add_dependency(
+	struct elf_image *img,
+	struct elf_image *dep);
 extern int elf_image_link(struct elf_image *img);
 extern void elf_image_close(struct elf_image *img);

-extern virt_addr_t elf_image_find_symbol(
+extern int elf_image_find_symbol(
 	struct elf_image *img,
-	const char *name);
-extern virt_addr_t elf_image_find_linked_symbol(
+	const char *name,
+	struct elf_symbol *out);
+extern int elf_image_find_symbol(
 	struct elf_image *img,
-	const char *name);
+	const char *name,
+	struct elf_symbol *out);
+extern int elf_image_find_linked_symbol(
+	struct elf_image *img,
+	const char *name,
+	struct elf_symbol *out);
+
+extern virt_addr_t find_global_symbol(const char *name);

 #endif
--- a/sys/ld/ld.h
+++ b/sys/ld/ld.h
@@ -0,0 +1,8 @@
+#ifndef LD_H_
+#define LD_H_
+
+struct image_list;
+
+extern struct image_list *global_image_list(void);
+
+#endif
--- a/sys/ld/main.c
+++ b/sys/ld/main.c
@@ -27,6 +27,13 @@ static const char *search_paths[] = {
 static const size_t nr_search_paths
 	= sizeof search_paths / sizeof search_paths[0];

+static struct image_list images = {0};
+
+struct image_list *global_image_list(void)
+{
+	return &images;
+}
+
 static void report_error(const char *name, int err, const char *msg, ...)
 {
 	char buf[1024];
@@ -35,7 +42,7 @@ static void report_error(const char *name, int err, const char *msg, ...)
 	vsnprintf(buf, sizeof buf, msg, arg);
 	va_end(arg);

-	kern_tracef("%s: %s: %s", name, buf, strerror(err));
+	kern_logf("%s: %s: %s", name, buf, strerror(err));
 }

 static const char *get_image_name(const char *path)
@@ -71,7 +78,7 @@ static int find_image(struct elf_image *img)
 	return ENOENT;
 }

-static int load_images(struct image_list *list)
+static int load_images(const char *task_name, struct image_list *list)
 {
 	int status = SUCCESS;
 	struct image_list_iterator it;
@@ -91,18 +98,33 @@ static int load_images(struct image_list *list)
 			/* Find the image using its name */
 			status = find_image(image);
 			if (status != SUCCESS) {
+				report_error(
+					task_name,
+					status,
+					"error while loading %s",
+					image->e_leaf.l_name);
 				return status;
 			}
 		case ELF_IMAGE_OPEN:
 			/* parse the image */
 			status = elf_image_parse(image);
 			if (status != SUCCESS) {
+				report_error(
+					task_name,
+					status,
+					"error while loading %s",
+					image->e_leaf.l_name);
 				return status;
 			}
 		case ELF_IMAGE_PARSED:
 			/* load the image */
 			status = elf_image_load(image);
 			if (status != SUCCESS) {
+				report_error(
+					task_name,
+					status,
+					"error while loading %s",
+					image->e_leaf.l_name);
 				return status;
 			}
 		case ELF_IMAGE_LOADED:
@@ -114,6 +136,11 @@ static int load_images(struct image_list *list)
 		}

 		if (new_dependencies < 0) {
+			report_error(
+				task_name,
+				-new_dependencies,
+				"error while loading %s",
+				image->e_leaf.l_name);
 			return -new_dependencies;
 		}

@@ -127,7 +154,7 @@ static int load_images(struct image_list *list)
 	return SUCCESS;
 }

-static int link_images(struct image_list *list)
+static int link_images(const char *task_name, struct image_list *list)
 {
 	int status = SUCCESS;
 	struct image_list_iterator it;
@@ -143,6 +170,14 @@ static int link_images(struct image_list *list)
 			elf_image_status_to_string(image->e_status));

 		status = elf_image_link(image);
+		if (status != SUCCESS) {
+			report_error(
+				task_name,
+				status,
+				"error while loading %s",
+				image->e_leaf.l_name);
+			return status;
+		}

 		image_list_iterator_move_next(&it);
 	}
@@ -162,7 +197,6 @@ int main(const struct rosetta_bootstrap *bs)
 	const char *task_name = bs->bs_argv[2];
 	const char *image_name = get_image_name(exec_path);

-	struct image_list images;
 	image_list_init(&images);

 	struct elf_image *exec = NULL;
@@ -183,23 +217,13 @@ int main(const struct rosetta_bootstrap *bs)
 		image_name);

 	image_list_put(&images, &exec->e_leaf);
-	err = load_images(&images);
+	err = load_images(task_name, &images);
 	if (err != SUCCESS) {
-		report_error(
-			task_name,
-			err,
-			"error while loading %s",
-			exec_path);
 		return -1;
 	}

-	err = link_images(&images);
+	err = link_images(task_name, &images);
 	if (err != SUCCESS) {
-		report_error(
-			task_name,
-			err,
-			"error while loading %s",
-			exec_path);
 		return -1;
 	}

@@ -218,7 +242,12 @@ int main(const struct rosetta_bootstrap *bs)
 	}

 	kern_tracef("ld finished");
-	int (*entry)(int, const char **)
-		= (int (*)(int, const char **))exec->e_entry;
-	return entry(bs->bs_argc - 2, bs->bs_argv + 2);
+	struct rosetta_bootstrap exec_bsinfo;
+	memcpy(&exec_bsinfo, bs, sizeof exec_bsinfo);
+
+	exec_bsinfo.bs_argc -= 2;
+	exec_bsinfo.bs_argv += 2;
+	int (*entry)(const struct rosetta_bootstrap *)
+		= (int (*)(const struct rosetta_bootstrap *))exec->e_entry;
+	return entry(&exec_bsinfo);
 }
--- a/sys/ld/resolve.c
+++ b/sys/ld/resolve.c
@@ -39,7 +39,11 @@ uintptr_t dl_runtime_resolve(struct elf_image *img, unsigned long sym_id)
 		img->e_leaf.l_name,
 		sym_id,
 		sym_name);
+#if 0
 	virt_addr_t sym_addr = elf_image_find_linked_symbol(img, sym_name);
+#else
+	virt_addr_t sym_addr = find_global_symbol(sym_name);
+#endif
 	kern_tracef("symbol %s = %zx", sym_name, sym_addr);
 	kern_tracef("slot %s = %zx", sym_name, &got[slot_id]);
 	got[slot_id] = sym_addr;