[PATCH] kallsyms in proc

 This adds a /proc/kallsyms if you have CONFIG_KALLSYMS in your
kernel.  The output is nm-like, with symbols in caps (global) if
exported using EXPORT_SYMBOL, rather than the normal static
vs. non-static differentiation.

This is useful for things like performance monitoring tools (profiling
etc) that want to match addresses to names in user space.

include/linux/module.h

@@ -266,6 +266,13 @@ static inline int module_is_live(struct module *mod)
 /* Is this address in a module? */
 struct module *module_text_address(unsigned long addr);
 
+/* Returns module and fills in value, defined and namebuf, or NULL if
+   symnum out of range. */
+struct module *module_get_kallsym(unsigned int symnum,
+				  unsigned long *value,
+				  char *type,
+				  char namebuf[128]);
+int is_exported(const char *name, const struct module *mod);
 #ifdef CONFIG_MODULE_UNLOAD
 
 unsigned int module_refcount(struct module *mod);
@@ -411,6 +418,19 @@ static inline const char *module_address_lookup(unsigned long addr,
 	return NULL;
 }
 
+static inline struct module *module_get_kallsym(unsigned int symnum,
+						unsigned long *value,
+						char *type,
+						char namebuf[128])
+{
+	return NULL;
+}
+
+static inline int is_exported(const char *name, const struct module *mod)
+{
+	return 0;
+}
+
 static inline int register_module_notifier(struct notifier_block * nb)
 {
 	/* no events will happen anyway, so this can always succeed */

kernel/kallsyms.c

@@ -8,6 +8,11 @@
  */
 #include <linux/kallsyms.h>
 #include <linux/module.h>
+#include <linux/init.h>
+#include <linux/seq_file.h>
+#include <linux/fs.h>
+#include <linux/err.h>
+#include <linux/proc_fs.h>
 
 /* These will be re-linked against their real values during the second link stage */
 extern unsigned long kallsyms_addresses[] __attribute__((weak));
@@ -117,5 +122,170 @@ void __print_symbol(const char *fmt, unsigned long address)
 	}
 }
 
+/* To avoid O(n^2) iteration, we carry prefix along. */
+struct kallsym_iter
+{
+	loff_t pos;
+	struct module *owner;
+	unsigned long value;
+	unsigned int nameoff; /* If iterating in core kernel symbols */
+	char type;
+	char name[128];
+};
+
+/* Only label it "global" if it is exported. */
+static void upcase_if_global(struct kallsym_iter *iter)
+{
+	if (is_exported(iter->name, iter->owner))
+		iter->type += 'A' - 'a';
+}
+
+static int get_ksymbol_mod(struct kallsym_iter *iter)
+{
+	iter->owner = module_get_kallsym(iter->pos - kallsyms_num_syms,
+					 &iter->value,
+					 &iter->type, iter->name);
+	if (iter->owner == NULL)
+		return 0;
+
+	upcase_if_global(iter);
+	return 1;
+}
+
+static void get_ksymbol_core(struct kallsym_iter *iter)
+{
+	unsigned stemlen;
+
+	/* First char of each symbol name indicates prefix length
+	   shared with previous name (stem compresion). */
+	stemlen = kallsyms_names[iter->nameoff++];
+
+	strlcpy(iter->name+stemlen, kallsyms_names+iter->nameoff, 128-stemlen);
+	iter->nameoff += strlen(kallsyms_names + iter->nameoff) + 1;
+	iter->owner = NULL;
+	iter->value = kallsyms_addresses[iter->pos];
+	iter->type = 't';
+
+	upcase_if_global(iter);
+}
+
+static void reset_iter(struct kallsym_iter *iter)
+{
+	iter->name[0] = '\0';
+	iter->nameoff = 0;
+	iter->pos = 0;
+}
+
+/* Returns false if pos at or past end of file. */
+static int update_iter(struct kallsym_iter *iter, loff_t pos)
+{
+	/* Module symbols can be accessed randomly. */
+	if (pos >= kallsyms_num_syms) {
+		iter->pos = pos;
+		return get_ksymbol_mod(iter);
+	}
+	
+	/* If we're past the desired position, reset to start. */
+	if (pos < iter->pos)
+		reset_iter(iter);
+
+	/* We need to iterate through the previous symbols. */
+	for (; iter->pos <= pos; iter->pos++)
+		get_ksymbol_core(iter);
+	return 1;
+}
+
+static void *s_next(struct seq_file *m, void *p, loff_t *pos)
+{
+	(*pos)++;
+
+	if (!update_iter(m->private, *pos))
+		return NULL;
+	return p;
+}
+
+static void *s_start(struct seq_file *m, loff_t *pos)
+{
+	if (!update_iter(m->private, *pos))
+		return NULL;
+	return m->private;
+}
+
+static void s_stop(struct seq_file *m, void *p)
+{
+}
+
+static int s_show(struct seq_file *m, void *p)
+{
+	struct kallsym_iter *iter = m->private;
+
+	/* Some debugging symbols have no name.  Ignore them. */ 
+	if (!iter->name[0])
+		return 0;
+
+	if (iter->owner)
+		seq_printf(m, "%0*lx %c %s\t[%s]\n",
+			   (int)(2*sizeof(void*)),
+			   iter->value, iter->type, iter->name,
+			   module_name(iter->owner));
+	else
+		seq_printf(m, "%0*lx %c %s\n",
+			   (int)(2*sizeof(void*)),
+			   iter->value, iter->type, iter->name);
+	return 0;
+}
+
+struct seq_operations kallsyms_op = {
+	.start = s_start,
+	.next = s_next,
+	.stop = s_stop,
+	.show = s_show
+};
+
+static int kallsyms_open(struct inode *inode, struct file *file)
+{
+	/* We keep iterator in m->private, since normal case is to
+	 * s_start from where we left off, so we avoid O(N^2). */
+	struct kallsym_iter *iter;
+	int ret;
+
+	iter = kmalloc(sizeof(*iter), GFP_KERNEL);
+	if (!iter)
+		return -ENOMEM;
+
+	ret = seq_open(file, &kallsyms_op);
+	if (ret == 0)
+		((struct seq_file *)file->private_data)->private = iter;
+	else
+		kfree(iter);
+	return ret;
+}
+
+static int kallsyms_release(struct inode *inode, struct file *file)
+{
+	struct seq_file *m = (struct seq_file *)file->private_data;
+	kfree(m->private);
+	return seq_release(inode, file);
+}
+
+static struct file_operations kallsyms_operations = {
+	.open = kallsyms_open,
+	.read = seq_read,
+	.llseek = seq_lseek,
+	.release = kallsyms_release,
+};
+
+int __init kallsyms_init(void)
+{
+	struct proc_dir_entry *entry;
+
+	/* root-only: could chew up lots of cpu by read, seek back, read... */
+	entry = create_proc_entry("kallsyms", 0400, NULL);
+	if (entry)
+		entry->proc_fops = &kallsyms_operations;
+	return 0;
+}
+__initcall(kallsyms_init);
+
 EXPORT_SYMBOL(kallsyms_lookup);
 EXPORT_SYMBOL(__print_symbol);

kernel/module.c

@@ -1274,6 +1274,83 @@ static char *get_modinfo(Elf_Shdr *sechdrs,
 	return NULL;
 }
 
+#ifdef CONFIG_KALLSYMS
+int is_exported(const char *name, const struct module *mod)
+{
+	unsigned int i;
+
+	if (!mod) {
+		for (i = 0; __start___ksymtab+i < __stop___ksymtab; i++)
+			if (strcmp(__start___ksymtab[i].name, name) == 0)
+				return 1;
+		return 0;
+	}
+	for (i = 0; i < mod->num_syms; i++)
+		if (strcmp(mod->syms[i].name, name) == 0)
+			return 1;
+	return 0;
+}
+
+/* As per nm */
+static char elf_type(const Elf_Sym *sym,
+		     Elf_Shdr *sechdrs,
+		     const char *secstrings,
+		     struct module *mod)
+{
+	if (ELF_ST_BIND(sym->st_info) == STB_WEAK) {
+		if (ELF_ST_TYPE(sym->st_info) == STT_OBJECT)
+			return 'v';
+		else
+			return 'w';
+	}
+	if (sym->st_shndx == SHN_UNDEF)
+		return 'U';
+	if (sym->st_shndx == SHN_ABS)
+		return 'a';
+	if (sym->st_shndx >= SHN_LORESERVE)
+		return '?';
+	if (sechdrs[sym->st_shndx].sh_flags & SHF_EXECINSTR)
+		return 't';
+	if (sechdrs[sym->st_shndx].sh_flags & SHF_ALLOC
+	    && sechdrs[sym->st_shndx].sh_type != SHT_NOBITS) {
+		if (!(sechdrs[sym->st_shndx].sh_flags & SHF_WRITE))
+			return 'r';
+		else if (sechdrs[sym->st_shndx].sh_flags & ARCH_SHF_SMALL)
+			return 'g';
+		else
+			return 'd';
+	}
+	if (sechdrs[sym->st_shndx].sh_type == SHT_NOBITS) {
+		if (sechdrs[sym->st_shndx].sh_flags & ARCH_SHF_SMALL)
+			return 's';
+		else
+			return 'b';
+	}
+	if (strncmp(secstrings + sechdrs[sym->st_shndx].sh_name,
+		    ".debug", strlen(".debug")) == 0)
+		return 'n';
+	return '?';
+}
+
+static void add_kallsyms(struct module *mod,
+			 Elf_Shdr *sechdrs,
+			 unsigned int symindex,
+			 unsigned int strindex,
+			 const char *secstrings)
+{
+	unsigned int i;
+
+	mod->symtab = (void *)sechdrs[symindex].sh_addr;
+	mod->num_symtab = sechdrs[symindex].sh_size / sizeof(Elf_Sym);
+	mod->strtab = (void *)sechdrs[strindex].sh_addr;
+
+	/* Set types up while we still have access to sections. */
+	for (i = 0; i < mod->num_symtab; i++)
+		mod->symtab[i].st_info
+			= elf_type(&mod->symtab[i], sechdrs, secstrings, mod);
+}
+#endif
+
 /* Allocate and load the module: note that size of section 0 is always
    zero, and we rely on this for optional sections. */
 static struct module *load_module(void __user *umod,
@@ -1525,15 +1602,12 @@ static struct module *load_module(void __user *umod,
 	percpu_modcopy(mod->percpu, (void *)sechdrs[pcpuindex].sh_addr,
 		       sechdrs[pcpuindex].sh_size);
 
-#ifdef CONFIG_KALLSYMS
-	mod->symtab = (void *)sechdrs[symindex].sh_addr;
-	mod->num_symtab = sechdrs[symindex].sh_size / sizeof(Elf_Sym);
-	mod->strtab = (void *)sechdrs[strindex].sh_addr;
-#endif
 	err = module_finalize(hdr, sechdrs, mod);
 	if (err < 0)
 		goto cleanup;
 
+	add_kallsyms(mod, sechdrs, symindex, strindex, secstrings);
+
 	mod->args = args;
 	if (obsparmindex) {
 		err = obsolete_params(mod->name, mod->args,
@@ -1715,6 +1789,30 @@ const char *module_address_lookup(unsigned long addr,
 	}
 	return NULL;
 }
+
+struct module *module_get_kallsym(unsigned int symnum,
+				  unsigned long *value,
+				  char *type,
+				  char namebuf[128])
+{
+	struct module *mod;
+
+	down(&module_mutex);
+	list_for_each_entry(mod, &modules, list) {
+		if (symnum < mod->num_symtab) {
+			*value = mod->symtab[symnum].st_value;
+			*type = mod->symtab[symnum].st_info;
+			strncpy(namebuf,
+				mod->strtab + mod->symtab[symnum].st_name,
+				127);
+			up(&module_mutex);
+			return mod;
+		}
+		symnum -= mod->num_symtab;
+	}
+	up(&module_mutex);
+	return NULL;
+}
 #endif /* CONFIG_KALLSYMS */
 
 /* Called by the /proc file system to return a list of modules. */