module.c 104.0 KB
Newer Older
1
/*
L
Linus Torvalds 已提交
2
   Copyright (C) 2002 Richard Henderson
3
   Copyright (C) 2001 Rusty Russell, 2002, 2010 Rusty Russell IBM.
L
Linus Torvalds 已提交
4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

    This program is free software; you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
    the Free Software Foundation; either version 2 of the License, or
    (at your option) any later version.

    This program is distributed in the hope that it will be useful,
    but WITHOUT ANY WARRANTY; without even the implied warranty of
    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
    GNU General Public License for more details.

    You should have received a copy of the GNU General Public License
    along with this program; if not, write to the Free Software
    Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
*/
19
#include <linux/export.h>
L
Linus Torvalds 已提交
20
#include <linux/moduleloader.h>
21
#include <linux/trace_events.h>
L
Linus Torvalds 已提交
22
#include <linux/init.h>
23
#include <linux/kallsyms.h>
24
#include <linux/file.h>
25
#include <linux/fs.h>
R
Roland McGrath 已提交
26
#include <linux/sysfs.h>
27
#include <linux/kernel.h>
L
Linus Torvalds 已提交
28 29 30
#include <linux/slab.h>
#include <linux/vmalloc.h>
#include <linux/elf.h>
31
#include <linux/proc_fs.h>
32
#include <linux/security.h>
L
Linus Torvalds 已提交
33 34 35 36
#include <linux/seq_file.h>
#include <linux/syscalls.h>
#include <linux/fcntl.h>
#include <linux/rcupdate.h>
37
#include <linux/capability.h>
L
Linus Torvalds 已提交
38 39 40 41 42 43
#include <linux/cpu.h>
#include <linux/moduleparam.h>
#include <linux/errno.h>
#include <linux/err.h>
#include <linux/vermagic.h>
#include <linux/notifier.h>
A
Al Viro 已提交
44
#include <linux/sched.h>
L
Linus Torvalds 已提交
45
#include <linux/device.h>
46
#include <linux/string.h>
A
Arjan van de Ven 已提交
47
#include <linux/mutex.h>
48
#include <linux/rculist.h>
L
Linus Torvalds 已提交
49 50
#include <asm/uaccess.h>
#include <asm/cacheflush.h>
51
#include <asm/mmu_context.h>
52
#include <linux/license.h>
53
#include <asm/sections.h>
M
Mathieu Desnoyers 已提交
54
#include <linux/tracepoint.h>
55
#include <linux/ftrace.h>
56
#include <linux/async.h>
57
#include <linux/percpu.h>
C
Catalin Marinas 已提交
58
#include <linux/kmemleak.h>
59
#include <linux/jump_label.h>
60
#include <linux/pfn.h>
61
#include <linux/bsearch.h>
62
#include <uapi/linux/module.h>
R
Rusty Russell 已提交
63
#include "module-internal.h"
L
Linus Torvalds 已提交
64

65 66 67
#define CREATE_TRACE_POINTS
#include <trace/events/module.h>

L
Linus Torvalds 已提交
68 69 70 71
#ifndef ARCH_SHF_SMALL
#define ARCH_SHF_SMALL 0
#endif

72 73 74 75 76 77 78 79 80 81 82
/*
 * Modules' sections will be aligned on page boundaries
 * to ensure complete separation of code and data, but
 * only when CONFIG_DEBUG_SET_MODULE_RONX=y
 */
#ifdef CONFIG_DEBUG_SET_MODULE_RONX
# define debug_align(X) ALIGN(X, PAGE_SIZE)
#else
# define debug_align(X) (X)
#endif

L
Linus Torvalds 已提交
83 84 85
/* If this is set, the section belongs in the init part of the module */
#define INIT_OFFSET_MASK (1UL << (BITS_PER_LONG-1))

86 87 88 89 90
/*
 * Mutex protects:
 * 1) List of modules (also safely readable with preempt_disable),
 * 2) module_use links,
 * 3) module_addr_min/module_addr_max.
91
 * (delete and add uses RCU list operations). */
92 93
DEFINE_MUTEX(module_mutex);
EXPORT_SYMBOL_GPL(module_mutex);
L
Linus Torvalds 已提交
94
static LIST_HEAD(modules);
95

96
#ifdef CONFIG_MODULES_TREE_LOOKUP
R
Rusty Russell 已提交
97

98 99 100 101
/*
 * Use a latched RB-tree for __module_address(); this allows us to use
 * RCU-sched lookups of the address from any context.
 *
102 103 104
 * This is conditional on PERF_EVENTS || TRACING because those can really hit
 * __module_address() hard by doing a lot of stack unwinding; potentially from
 * NMI context.
105 106 107
 */

static __always_inline unsigned long __mod_tree_val(struct latch_tree_node *n)
R
Rusty Russell 已提交
108
{
109
	struct module_layout *layout = container_of(n, struct module_layout, mtn.node);
R
Rusty Russell 已提交
110

111
	return (unsigned long)layout->base;
112 113 114 115
}

static __always_inline unsigned long __mod_tree_size(struct latch_tree_node *n)
{
116
	struct module_layout *layout = container_of(n, struct module_layout, mtn.node);
117

118
	return (unsigned long)layout->size;
119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139
}

static __always_inline bool
mod_tree_less(struct latch_tree_node *a, struct latch_tree_node *b)
{
	return __mod_tree_val(a) < __mod_tree_val(b);
}

static __always_inline int
mod_tree_comp(void *key, struct latch_tree_node *n)
{
	unsigned long val = (unsigned long)key;
	unsigned long start, end;

	start = __mod_tree_val(n);
	if (val < start)
		return -1;

	end = start + __mod_tree_size(n);
	if (val >= end)
		return 1;
R
Rusty Russell 已提交
140 141 142 143

	return 0;
}

144 145 146 147 148
static const struct latch_tree_ops mod_tree_ops = {
	.less = mod_tree_less,
	.comp = mod_tree_comp,
};

149 150 151 152 153 154
static struct mod_tree_root {
	struct latch_tree_root root;
	unsigned long addr_min;
	unsigned long addr_max;
} mod_tree __cacheline_aligned = {
	.addr_min = -1UL,
R
Rusty Russell 已提交
155 156
};

157 158 159 160 161 162 163 164 165 166 167 168
#define module_addr_min mod_tree.addr_min
#define module_addr_max mod_tree.addr_max

static noinline void __mod_tree_insert(struct mod_tree_node *node)
{
	latch_tree_insert(&node->node, &mod_tree.root, &mod_tree_ops);
}

static void __mod_tree_remove(struct mod_tree_node *node)
{
	latch_tree_erase(&node->node, &mod_tree.root, &mod_tree_ops);
}
169 170 171 172 173 174 175

/*
 * These modifications: insert, remove_init and remove; are serialized by the
 * module_mutex.
 */
static void mod_tree_insert(struct module *mod)
{
176 177
	mod->core_layout.mtn.mod = mod;
	mod->init_layout.mtn.mod = mod;
178

179 180 181
	__mod_tree_insert(&mod->core_layout.mtn);
	if (mod->init_layout.size)
		__mod_tree_insert(&mod->init_layout.mtn);
182 183 184 185
}

static void mod_tree_remove_init(struct module *mod)
{
186 187
	if (mod->init_layout.size)
		__mod_tree_remove(&mod->init_layout.mtn);
188 189 190 191
}

static void mod_tree_remove(struct module *mod)
{
192
	__mod_tree_remove(&mod->core_layout.mtn);
193 194 195
	mod_tree_remove_init(mod);
}

196
static struct module *mod_find(unsigned long addr)
197 198 199
{
	struct latch_tree_node *ltn;

200
	ltn = latch_tree_find((void *)addr, &mod_tree.root, &mod_tree_ops);
201 202 203 204 205 206
	if (!ltn)
		return NULL;

	return container_of(ltn, struct mod_tree_node, node)->mod;
}

207 208
#else /* MODULES_TREE_LOOKUP */

209 210
static unsigned long module_addr_min = -1UL, module_addr_max = 0;

211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228
static void mod_tree_insert(struct module *mod) { }
static void mod_tree_remove_init(struct module *mod) { }
static void mod_tree_remove(struct module *mod) { }

static struct module *mod_find(unsigned long addr)
{
	struct module *mod;

	list_for_each_entry_rcu(mod, &modules, list) {
		if (within_module(addr, mod))
			return mod;
	}

	return NULL;
}

#endif /* MODULES_TREE_LOOKUP */

229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245
/*
 * Bounds of module text, for speeding up __module_address.
 * Protected by module_mutex.
 */
static void __mod_update_bounds(void *base, unsigned int size)
{
	unsigned long min = (unsigned long)base;
	unsigned long max = min + size;

	if (min < module_addr_min)
		module_addr_min = min;
	if (max > module_addr_max)
		module_addr_max = max;
}

static void mod_update_bounds(struct module *mod)
{
246 247 248
	__mod_update_bounds(mod->core_layout.base, mod->core_layout.size);
	if (mod->init_layout.size)
		__mod_update_bounds(mod->init_layout.base, mod->init_layout.size);
249 250
}

251 252 253 254
#ifdef CONFIG_KGDB_KDB
struct list_head *kdb_modules = &modules; /* kdb needs the list of modules */
#endif /* CONFIG_KGDB_KDB */

255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270
static void module_assert_mutex(void)
{
	lockdep_assert_held(&module_mutex);
}

static void module_assert_mutex_or_preempt(void)
{
#ifdef CONFIG_LOCKDEP
	if (unlikely(!debug_locks))
		return;

	WARN_ON(!rcu_read_lock_sched_held() &&
		!lockdep_is_held(&module_mutex));
#endif
}

271 272
static bool sig_enforce = IS_ENABLED(CONFIG_MODULE_SIG_FORCE);
#ifndef CONFIG_MODULE_SIG_FORCE
R
Rusty Russell 已提交
273 274
module_param(sig_enforce, bool_enable_only, 0644);
#endif /* !CONFIG_MODULE_SIG_FORCE */
L
Linus Torvalds 已提交
275

276 277
/* Block module loading/unloading? */
int modules_disabled = 0;
278
core_param(nomodule, modules_disabled, bint, 0);
279

280 281 282
/* Waiting for a module to finish initializing? */
static DECLARE_WAIT_QUEUE_HEAD(module_wq);

283
static BLOCKING_NOTIFIER_HEAD(module_notify_list);
L
Linus Torvalds 已提交
284

285
int register_module_notifier(struct notifier_block *nb)
L
Linus Torvalds 已提交
286
{
287
	return blocking_notifier_chain_register(&module_notify_list, nb);
L
Linus Torvalds 已提交
288 289 290
}
EXPORT_SYMBOL(register_module_notifier);

291
int unregister_module_notifier(struct notifier_block *nb)
L
Linus Torvalds 已提交
292
{
293
	return blocking_notifier_chain_unregister(&module_notify_list, nb);
L
Linus Torvalds 已提交
294 295 296
}
EXPORT_SYMBOL(unregister_module_notifier);

297 298 299 300
struct load_info {
	Elf_Ehdr *hdr;
	unsigned long len;
	Elf_Shdr *sechdrs;
301
	char *secstrings, *strtab;
R
Rusty Russell 已提交
302
	unsigned long symoffs, stroffs;
303 304
	struct _ddebug *debug;
	unsigned int num_debug;
R
Rusty Russell 已提交
305
	bool sig_ok;
306 307 308
#ifdef CONFIG_KALLSYMS
	unsigned long mod_kallsyms_init_off;
#endif
309 310 311 312 313
	struct {
		unsigned int sym, str, mod, vers, info, pcpu;
	} index;
};

314 315
/* We require a truly strong try_module_get(): 0 means failure due to
   ongoing or failed initialization etc. */
L
Linus Torvalds 已提交
316 317
static inline int strong_try_module_get(struct module *mod)
{
318
	BUG_ON(mod && mod->state == MODULE_STATE_UNFORMED);
L
Linus Torvalds 已提交
319
	if (mod && mod->state == MODULE_STATE_COMING)
320 321
		return -EBUSY;
	if (try_module_get(mod))
L
Linus Torvalds 已提交
322
		return 0;
323 324
	else
		return -ENOENT;
L
Linus Torvalds 已提交
325 326
}

327 328
static inline void add_taint_module(struct module *mod, unsigned flag,
				    enum lockdep_ok lockdep_ok)
329
{
330
	add_taint(flag, lockdep_ok);
A
Andi Kleen 已提交
331
	mod->taints |= (1U << flag);
332 333
}

334 335 336
/*
 * A thread that wants to hold a reference to a module only while it
 * is running can call this to safely exit.  nfsd and lockd use this.
L
Linus Torvalds 已提交
337 338 339 340 341 342 343
 */
void __module_put_and_exit(struct module *mod, long code)
{
	module_put(mod);
	do_exit(code);
}
EXPORT_SYMBOL(__module_put_and_exit);
D
Daniel Walker 已提交
344

L
Linus Torvalds 已提交
345
/* Find a module section: 0 means not found. */
346
static unsigned int find_sec(const struct load_info *info, const char *name)
L
Linus Torvalds 已提交
347 348 349
{
	unsigned int i;

350 351
	for (i = 1; i < info->hdr->e_shnum; i++) {
		Elf_Shdr *shdr = &info->sechdrs[i];
L
Linus Torvalds 已提交
352
		/* Alloc bit cleared means "ignore it." */
353 354
		if ((shdr->sh_flags & SHF_ALLOC)
		    && strcmp(info->secstrings + shdr->sh_name, name) == 0)
L
Linus Torvalds 已提交
355
			return i;
356
	}
L
Linus Torvalds 已提交
357 358 359
	return 0;
}

R
Rusty Russell 已提交
360
/* Find a module section, or NULL. */
361
static void *section_addr(const struct load_info *info, const char *name)
R
Rusty Russell 已提交
362 363
{
	/* Section 0 has sh_addr 0. */
364
	return (void *)info->sechdrs[find_sec(info, name)].sh_addr;
R
Rusty Russell 已提交
365 366 367
}

/* Find a module section, or NULL.  Fill in number of "objects" in section. */
368
static void *section_objs(const struct load_info *info,
R
Rusty Russell 已提交
369 370 371 372
			  const char *name,
			  size_t object_size,
			  unsigned int *num)
{
373
	unsigned int sec = find_sec(info, name);
R
Rusty Russell 已提交
374 375

	/* Section 0 has sh_addr 0 and sh_size 0. */
376 377
	*num = info->sechdrs[sec].sh_size / object_size;
	return (void *)info->sechdrs[sec].sh_addr;
R
Rusty Russell 已提交
378 379
}

L
Linus Torvalds 已提交
380 381 382 383 384
/* Provided by the linker */
extern const struct kernel_symbol __start___ksymtab[];
extern const struct kernel_symbol __stop___ksymtab[];
extern const struct kernel_symbol __start___ksymtab_gpl[];
extern const struct kernel_symbol __stop___ksymtab_gpl[];
385 386
extern const struct kernel_symbol __start___ksymtab_gpl_future[];
extern const struct kernel_symbol __stop___ksymtab_gpl_future[];
L
Linus Torvalds 已提交
387 388
extern const unsigned long __start___kcrctab[];
extern const unsigned long __start___kcrctab_gpl[];
389
extern const unsigned long __start___kcrctab_gpl_future[];
390 391 392 393 394
#ifdef CONFIG_UNUSED_SYMBOLS
extern const struct kernel_symbol __start___ksymtab_unused[];
extern const struct kernel_symbol __stop___ksymtab_unused[];
extern const struct kernel_symbol __start___ksymtab_unused_gpl[];
extern const struct kernel_symbol __stop___ksymtab_unused_gpl[];
395 396
extern const unsigned long __start___kcrctab_unused[];
extern const unsigned long __start___kcrctab_unused_gpl[];
397
#endif
L
Linus Torvalds 已提交
398 399 400 401

#ifndef CONFIG_MODVERSIONS
#define symversion(base, idx) NULL
#else
A
Andrew Morton 已提交
402
#define symversion(base, idx) ((base != NULL) ? ((base) + (idx)) : NULL)
L
Linus Torvalds 已提交
403 404
#endif

405 406 407 408 409
static bool each_symbol_in_section(const struct symsearch *arr,
				   unsigned int arrsize,
				   struct module *owner,
				   bool (*fn)(const struct symsearch *syms,
					      struct module *owner,
410
					      void *data),
411
				   void *data)
412
{
413
	unsigned int j;
414

415
	for (j = 0; j < arrsize; j++) {
416 417
		if (fn(&arr[j], owner, data))
			return true;
418
	}
419 420

	return false;
421 422
}

423
/* Returns true as soon as fn returns true, otherwise false. */
424 425 426 427
bool each_symbol_section(bool (*fn)(const struct symsearch *arr,
				    struct module *owner,
				    void *data),
			 void *data)
428 429
{
	struct module *mod;
430
	static const struct symsearch arr[] = {
431
		{ __start___ksymtab, __stop___ksymtab, __start___kcrctab,
432
		  NOT_GPL_ONLY, false },
433
		{ __start___ksymtab_gpl, __stop___ksymtab_gpl,
434 435
		  __start___kcrctab_gpl,
		  GPL_ONLY, false },
436
		{ __start___ksymtab_gpl_future, __stop___ksymtab_gpl_future,
437 438
		  __start___kcrctab_gpl_future,
		  WILL_BE_GPL_ONLY, false },
439
#ifdef CONFIG_UNUSED_SYMBOLS
440
		{ __start___ksymtab_unused, __stop___ksymtab_unused,
441 442
		  __start___kcrctab_unused,
		  NOT_GPL_ONLY, true },
443
		{ __start___ksymtab_unused_gpl, __stop___ksymtab_unused_gpl,
444 445
		  __start___kcrctab_unused_gpl,
		  GPL_ONLY, true },
446
#endif
447
	};
448

449 450
	module_assert_mutex_or_preempt();

451 452
	if (each_symbol_in_section(arr, ARRAY_SIZE(arr), NULL, fn, data))
		return true;
453

454
	list_for_each_entry_rcu(mod, &modules, list) {
455 456
		struct symsearch arr[] = {
			{ mod->syms, mod->syms + mod->num_syms, mod->crcs,
457
			  NOT_GPL_ONLY, false },
458
			{ mod->gpl_syms, mod->gpl_syms + mod->num_gpl_syms,
459 460
			  mod->gpl_crcs,
			  GPL_ONLY, false },
461 462
			{ mod->gpl_future_syms,
			  mod->gpl_future_syms + mod->num_gpl_future_syms,
463 464
			  mod->gpl_future_crcs,
			  WILL_BE_GPL_ONLY, false },
465
#ifdef CONFIG_UNUSED_SYMBOLS
466 467
			{ mod->unused_syms,
			  mod->unused_syms + mod->num_unused_syms,
468 469
			  mod->unused_crcs,
			  NOT_GPL_ONLY, true },
470 471
			{ mod->unused_gpl_syms,
			  mod->unused_gpl_syms + mod->num_unused_gpl_syms,
472 473
			  mod->unused_gpl_crcs,
			  GPL_ONLY, true },
474
#endif
475 476
		};

477 478 479
		if (mod->state == MODULE_STATE_UNFORMED)
			continue;

480 481 482 483 484
		if (each_symbol_in_section(arr, ARRAY_SIZE(arr), mod, fn, data))
			return true;
	}
	return false;
}
485
EXPORT_SYMBOL_GPL(each_symbol_section);
486 487 488 489 490 491 492 493 494 495

struct find_symbol_arg {
	/* Input */
	const char *name;
	bool gplok;
	bool warn;

	/* Output */
	struct module *owner;
	const unsigned long *crc;
496
	const struct kernel_symbol *sym;
497 498
};

499 500 501
static bool check_symbol(const struct symsearch *syms,
				 struct module *owner,
				 unsigned int symnum, void *data)
502 503 504 505 506 507 508
{
	struct find_symbol_arg *fsa = data;

	if (!fsa->gplok) {
		if (syms->licence == GPL_ONLY)
			return false;
		if (syms->licence == WILL_BE_GPL_ONLY && fsa->warn) {
A
Andrew Morton 已提交
509 510 511
			pr_warn("Symbol %s is being used by a non-GPL module, "
				"which will not be allowed in the future\n",
				fsa->name);
512
		}
L
Linus Torvalds 已提交
513
	}
514

515
#ifdef CONFIG_UNUSED_SYMBOLS
516
	if (syms->unused && fsa->warn) {
A
Andrew Morton 已提交
517 518 519
		pr_warn("Symbol %s is marked as UNUSED, however this module is "
			"using it.\n", fsa->name);
		pr_warn("This symbol will go away in the future.\n");
520 521 522
		pr_warn("Please evaluate if this is the right api to use and "
			"if it really is, submit a report to the linux kernel "
			"mailing list together with submitting your code for "
A
Andrew Morton 已提交
523
			"inclusion.\n");
524
	}
525
#endif
526 527 528

	fsa->owner = owner;
	fsa->crc = symversion(syms->crcs, symnum);
529
	fsa->sym = &syms->start[symnum];
530 531 532
	return true;
}

533 534 535 536 537 538 539 540
static int cmp_name(const void *va, const void *vb)
{
	const char *a;
	const struct kernel_symbol *b;
	a = va; b = vb;
	return strcmp(a, b->name);
}

541 542 543 544 545
static bool find_symbol_in_section(const struct symsearch *syms,
				   struct module *owner,
				   void *data)
{
	struct find_symbol_arg *fsa = data;
546 547 548 549 550 551 552
	struct kernel_symbol *sym;

	sym = bsearch(fsa->name, syms->start, syms->stop - syms->start,
			sizeof(struct kernel_symbol), cmp_name);

	if (sym != NULL && check_symbol(syms, owner, sym - syms->start, data))
		return true;
553 554 555 556

	return false;
}

557
/* Find a symbol and return it, along with, (optional) crc and
558
 * (optional) module which owns it.  Needs preempt disabled or module_mutex. */
559 560 561 562 563
const struct kernel_symbol *find_symbol(const char *name,
					struct module **owner,
					const unsigned long **crc,
					bool gplok,
					bool warn)
564 565 566 567 568 569 570
{
	struct find_symbol_arg fsa;

	fsa.name = name;
	fsa.gplok = gplok;
	fsa.warn = warn;

571
	if (each_symbol_section(find_symbol_in_section, &fsa)) {
572 573 574 575
		if (owner)
			*owner = fsa.owner;
		if (crc)
			*crc = fsa.crc;
576
		return fsa.sym;
577 578
	}

579
	pr_debug("Failed to find symbol %s\n", name);
580
	return NULL;
L
Linus Torvalds 已提交
581
}
582
EXPORT_SYMBOL_GPL(find_symbol);
L
Linus Torvalds 已提交
583

584 585 586 587
/*
 * Search for module by name: must hold module_mutex (or preempt disabled
 * for read-only access).
 */
588
static struct module *find_module_all(const char *name, size_t len,
589
				      bool even_unformed)
L
Linus Torvalds 已提交
590 591 592
{
	struct module *mod;

593
	module_assert_mutex_or_preempt();
594

L
Linus Torvalds 已提交
595
	list_for_each_entry(mod, &modules, list) {
596 597
		if (!even_unformed && mod->state == MODULE_STATE_UNFORMED)
			continue;
598
		if (strlen(mod->name) == len && !memcmp(mod->name, name, len))
L
Linus Torvalds 已提交
599 600 601 602
			return mod;
	}
	return NULL;
}
603 604 605

struct module *find_module(const char *name)
{
606
	module_assert_mutex();
607
	return find_module_all(name, strlen(name), false);
608
}
609
EXPORT_SYMBOL_GPL(find_module);
L
Linus Torvalds 已提交
610 611

#ifdef CONFIG_SMP
612

613
static inline void __percpu *mod_percpu(struct module *mod)
614
{
615 616
	return mod->percpu;
}
617

R
Rusty Russell 已提交
618
static int percpu_modalloc(struct module *mod, struct load_info *info)
619
{
R
Rusty Russell 已提交
620 621 622 623 624 625
	Elf_Shdr *pcpusec = &info->sechdrs[info->index.pcpu];
	unsigned long align = pcpusec->sh_addralign;

	if (!pcpusec->sh_size)
		return 0;

626
	if (align > PAGE_SIZE) {
A
Andrew Morton 已提交
627 628
		pr_warn("%s: per-cpu alignment %li > %li\n",
			mod->name, align, PAGE_SIZE);
629 630 631
		align = PAGE_SIZE;
	}

R
Rusty Russell 已提交
632
	mod->percpu = __alloc_reserved_percpu(pcpusec->sh_size, align);
633
	if (!mod->percpu) {
A
Andrew Morton 已提交
634 635
		pr_warn("%s: Could not allocate %lu bytes percpu data\n",
			mod->name, (unsigned long)pcpusec->sh_size);
636 637
		return -ENOMEM;
	}
R
Rusty Russell 已提交
638
	mod->percpu_size = pcpusec->sh_size;
639
	return 0;
640 641
}

642
static void percpu_modfree(struct module *mod)
643
{
644
	free_percpu(mod->percpu);
645 646
}

647
static unsigned int find_pcpusec(struct load_info *info)
648
{
649
	return find_sec(info, ".data..percpu");
650 651
}

652 653
static void percpu_modcopy(struct module *mod,
			   const void *from, unsigned long size)
654 655 656 657
{
	int cpu;

	for_each_possible_cpu(cpu)
658
		memcpy(per_cpu_ptr(mod->percpu, cpu), from, size);
659 660
}

661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677
/**
 * is_module_percpu_address - test whether address is from module static percpu
 * @addr: address to test
 *
 * Test whether @addr belongs to module static percpu area.
 *
 * RETURNS:
 * %true if @addr is from module static percpu area
 */
bool is_module_percpu_address(unsigned long addr)
{
	struct module *mod;
	unsigned int cpu;

	preempt_disable();

	list_for_each_entry_rcu(mod, &modules, list) {
678 679
		if (mod->state == MODULE_STATE_UNFORMED)
			continue;
680 681 682 683 684 685 686 687 688 689 690 691 692 693 694
		if (!mod->percpu_size)
			continue;
		for_each_possible_cpu(cpu) {
			void *start = per_cpu_ptr(mod->percpu, cpu);

			if ((void *)addr >= start &&
			    (void *)addr < start + mod->percpu_size) {
				preempt_enable();
				return true;
			}
		}
	}

	preempt_enable();
	return false;
695 696
}

L
Linus Torvalds 已提交
697
#else /* ... !CONFIG_SMP */
698

699
static inline void __percpu *mod_percpu(struct module *mod)
L
Linus Torvalds 已提交
700 701 702
{
	return NULL;
}
R
Rusty Russell 已提交
703
static int percpu_modalloc(struct module *mod, struct load_info *info)
704
{
R
Rusty Russell 已提交
705 706 707 708
	/* UP modules shouldn't have this section: ENOMEM isn't quite right */
	if (info->sechdrs[info->index.pcpu].sh_size != 0)
		return -ENOMEM;
	return 0;
709 710
}
static inline void percpu_modfree(struct module *mod)
L
Linus Torvalds 已提交
711 712
{
}
713
static unsigned int find_pcpusec(struct load_info *info)
L
Linus Torvalds 已提交
714 715 716
{
	return 0;
}
717 718
static inline void percpu_modcopy(struct module *mod,
				  const void *from, unsigned long size)
L
Linus Torvalds 已提交
719 720 721 722
{
	/* pcpusec should be 0, and size of that section should be 0. */
	BUG_ON(size != 0);
}
723 724 725 726
bool is_module_percpu_address(unsigned long addr)
{
	return false;
}
727

L
Linus Torvalds 已提交
728 729
#endif /* CONFIG_SMP */

730 731 732 733 734 735
#define MODINFO_ATTR(field)	\
static void setup_modinfo_##field(struct module *mod, const char *s)  \
{                                                                     \
	mod->field = kstrdup(s, GFP_KERNEL);                          \
}                                                                     \
static ssize_t show_modinfo_##field(struct module_attribute *mattr,   \
736
			struct module_kobject *mk, char *buffer)      \
737
{                                                                     \
738
	return scnprintf(buffer, PAGE_SIZE, "%s\n", mk->mod->field);  \
739 740 741 742 743 744 745
}                                                                     \
static int modinfo_##field##_exists(struct module *mod)               \
{                                                                     \
	return mod->field != NULL;                                    \
}                                                                     \
static void free_modinfo_##field(struct module *mod)                  \
{                                                                     \
D
Daniel Walker 已提交
746 747
	kfree(mod->field);                                            \
	mod->field = NULL;                                            \
748 749
}                                                                     \
static struct module_attribute modinfo_##field = {                    \
750
	.attr = { .name = __stringify(field), .mode = 0444 },         \
751 752 753 754 755 756 757 758 759
	.show = show_modinfo_##field,                                 \
	.setup = setup_modinfo_##field,                               \
	.test = modinfo_##field##_exists,                             \
	.free = free_modinfo_##field,                                 \
};

MODINFO_ATTR(version);
MODINFO_ATTR(srcversion);

760 761
static char last_unloaded_module[MODULE_NAME_LEN+1];

762
#ifdef CONFIG_MODULE_UNLOAD
763 764 765

EXPORT_TRACEPOINT_SYMBOL(module_get);

766 767 768
/* MODULE_REF_BASE is the base reference count by kmodule loader. */
#define MODULE_REF_BASE	1

L
Linus Torvalds 已提交
769
/* Init the unload section of the module. */
770
static int module_unload_init(struct module *mod)
L
Linus Torvalds 已提交
771
{
772 773 774 775 776
	/*
	 * Initialize reference counter to MODULE_REF_BASE.
	 * refcnt == 0 means module is going.
	 */
	atomic_set(&mod->refcnt, MODULE_REF_BASE);
777

778 779
	INIT_LIST_HEAD(&mod->source_list);
	INIT_LIST_HEAD(&mod->target_list);
780

L
Linus Torvalds 已提交
781
	/* Hold reference count during initialization. */
782
	atomic_inc(&mod->refcnt);
783 784

	return 0;
L
Linus Torvalds 已提交
785 786 787 788 789 790 791
}

/* Does a already use b? */
static int already_uses(struct module *a, struct module *b)
{
	struct module_use *use;

792 793
	list_for_each_entry(use, &b->source_list, source_list) {
		if (use->source == a) {
794
			pr_debug("%s uses %s!\n", a->name, b->name);
L
Linus Torvalds 已提交
795 796 797
			return 1;
		}
	}
798
	pr_debug("%s does not use %s!\n", a->name, b->name);
L
Linus Torvalds 已提交
799 800 801
	return 0;
}

802 803 804 805 806 807 808 809 810 811 812
/*
 * Module a uses b
 *  - we add 'a' as a "source", 'b' as a "target" of module use
 *  - the module_use is added to the list of 'b' sources (so
 *    'b' can walk the list to see who sourced them), and of 'a'
 *    targets (so 'a' can see what modules it targets).
 */
static int add_module_usage(struct module *a, struct module *b)
{
	struct module_use *use;

813
	pr_debug("Allocating new usage for %s.\n", a->name);
814 815
	use = kmalloc(sizeof(*use), GFP_ATOMIC);
	if (!use) {
A
Andrew Morton 已提交
816
		pr_warn("%s: out of memory loading\n", a->name);
817 818 819 820 821 822 823 824 825 826
		return -ENOMEM;
	}

	use->source = a;
	use->target = b;
	list_add(&use->source_list, &b->source_list);
	list_add(&use->target_list, &a->target_list);
	return 0;
}

827
/* Module a uses b: caller needs module_mutex() */
828
int ref_module(struct module *a, struct module *b)
L
Linus Torvalds 已提交
829
{
830
	int err;
K
Kay Sievers 已提交
831

832
	if (b == NULL || already_uses(a, b))
833 834
		return 0;

835 836
	/* If module isn't available, we fail. */
	err = strong_try_module_get(b);
837
	if (err)
838
		return err;
L
Linus Torvalds 已提交
839

840 841
	err = add_module_usage(a, b);
	if (err) {
L
Linus Torvalds 已提交
842
		module_put(b);
843
		return err;
L
Linus Torvalds 已提交
844
	}
845
	return 0;
L
Linus Torvalds 已提交
846
}
847
EXPORT_SYMBOL_GPL(ref_module);
L
Linus Torvalds 已提交
848 849 850 851

/* Clear the unload stuff of the module. */
static void module_unload_free(struct module *mod)
{
852
	struct module_use *use, *tmp;
L
Linus Torvalds 已提交
853

854
	mutex_lock(&module_mutex);
855 856
	list_for_each_entry_safe(use, tmp, &mod->target_list, target_list) {
		struct module *i = use->target;
857
		pr_debug("%s unusing %s\n", mod->name, i->name);
858 859 860 861
		module_put(i);
		list_del(&use->source_list);
		list_del(&use->target_list);
		kfree(use);
L
Linus Torvalds 已提交
862
	}
863
	mutex_unlock(&module_mutex);
L
Linus Torvalds 已提交
864 865 866
}

#ifdef CONFIG_MODULE_FORCE_UNLOAD
867
static inline int try_force_unload(unsigned int flags)
L
Linus Torvalds 已提交
868 869 870
{
	int ret = (flags & O_TRUNC);
	if (ret)
871
		add_taint(TAINT_FORCED_RMMOD, LOCKDEP_NOW_UNRELIABLE);
L
Linus Torvalds 已提交
872 873 874
	return ret;
}
#else
875
static inline int try_force_unload(unsigned int flags)
L
Linus Torvalds 已提交
876 877 878 879 880
{
	return 0;
}
#endif /* CONFIG_MODULE_FORCE_UNLOAD */

881 882
/* Try to release refcount of module, 0 means success. */
static int try_release_module_ref(struct module *mod)
L
Linus Torvalds 已提交
883
{
884
	int ret;
L
Linus Torvalds 已提交
885

886 887 888 889 890 891
	/* Try to decrement refcnt which we set at loading */
	ret = atomic_sub_return(MODULE_REF_BASE, &mod->refcnt);
	BUG_ON(ret < 0);
	if (ret)
		/* Someone can put this right now, recover with checking */
		ret = atomic_add_unless(&mod->refcnt, MODULE_REF_BASE, 0);
L
Linus Torvalds 已提交
892

893 894
	return ret;
}
L
Linus Torvalds 已提交
895

896 897
static int try_stop_module(struct module *mod, int flags, int *forced)
{
898
	/* If it's not unused, quit unless we're forcing. */
899 900 901
	if (try_release_module_ref(mod) != 0) {
		*forced = try_force_unload(flags);
		if (!(*forced))
L
Linus Torvalds 已提交
902 903 904 905
			return -EWOULDBLOCK;
	}

	/* Mark it as dying. */
906
	mod->state = MODULE_STATE_GOING;
L
Linus Torvalds 已提交
907

908
	return 0;
L
Linus Torvalds 已提交
909 910
}

911 912 913 914 915 916 917 918 919 920
/**
 * module_refcount - return the refcount or -1 if unloading
 *
 * @mod:	the module we're checking
 *
 * Returns:
 *	-1 if the module is in the process of unloading
 *	otherwise the number of references in the kernel to the module
 */
int module_refcount(struct module *mod)
L
Linus Torvalds 已提交
921
{
922
	return atomic_read(&mod->refcnt) - MODULE_REF_BASE;
L
Linus Torvalds 已提交
923 924 925 926 927 928
}
EXPORT_SYMBOL(module_refcount);

/* This exists whether we can unload or not */
static void free_module(struct module *mod);

929 930
SYSCALL_DEFINE2(delete_module, const char __user *, name_user,
		unsigned int, flags)
L
Linus Torvalds 已提交
931 932
{
	struct module *mod;
933
	char name[MODULE_NAME_LEN];
L
Linus Torvalds 已提交
934 935
	int ret, forced = 0;

936
	if (!capable(CAP_SYS_MODULE) || modules_disabled)
937 938 939 940 941 942
		return -EPERM;

	if (strncpy_from_user(name, name_user, MODULE_NAME_LEN-1) < 0)
		return -EFAULT;
	name[MODULE_NAME_LEN-1] = '\0';

943 944
	if (mutex_lock_interruptible(&module_mutex) != 0)
		return -EINTR;
L
Linus Torvalds 已提交
945 946 947 948 949 950 951

	mod = find_module(name);
	if (!mod) {
		ret = -ENOENT;
		goto out;
	}

952
	if (!list_empty(&mod->source_list)) {
L
Linus Torvalds 已提交
953 954 955 956 957 958 959
		/* Other modules depend on us: get rid of them first. */
		ret = -EWOULDBLOCK;
		goto out;
	}

	/* Doing init or already dying? */
	if (mod->state != MODULE_STATE_LIVE) {
960
		/* FIXME: if (force), slam module count damn the torpedoes */
961
		pr_debug("%s already dying\n", mod->name);
L
Linus Torvalds 已提交
962 963 964 965 966
		ret = -EBUSY;
		goto out;
	}

	/* If it has an init func, it must have an exit func to unload */
R
Rusty Russell 已提交
967
	if (mod->init && !mod->exit) {
968
		forced = try_force_unload(flags);
L
Linus Torvalds 已提交
969 970 971 972 973 974 975 976 977 978 979 980
		if (!forced) {
			/* This module can't be removed */
			ret = -EBUSY;
			goto out;
		}
	}

	/* Stop the machine so refcounts can't move and disable module. */
	ret = try_stop_module(mod, flags, &forced);
	if (ret != 0)
		goto out;

981
	mutex_unlock(&module_mutex);
L
Lucas De Marchi 已提交
982
	/* Final destruction now no one is using it. */
983
	if (mod->exit != NULL)
L
Linus Torvalds 已提交
984
		mod->exit();
985 986
	blocking_notifier_call_chain(&module_notify_list,
				     MODULE_STATE_GOING, mod);
987
	async_synchronize_full();
988

989
	/* Store the name of the last unloaded module for diagnostic purposes */
990
	strlcpy(last_unloaded_module, mod->name, sizeof(last_unloaded_module));
L
Linus Torvalds 已提交
991

992 993 994
	free_module(mod);
	return 0;
out:
995
	mutex_unlock(&module_mutex);
L
Linus Torvalds 已提交
996 997 998
	return ret;
}

999
static inline void print_unload_info(struct seq_file *m, struct module *mod)
L
Linus Torvalds 已提交
1000 1001 1002 1003
{
	struct module_use *use;
	int printed_something = 0;

1004
	seq_printf(m, " %i ", module_refcount(mod));
L
Linus Torvalds 已提交
1005

1006 1007 1008 1009
	/*
	 * Always include a trailing , so userspace can differentiate
	 * between this and the old multi-field proc format.
	 */
1010
	list_for_each_entry(use, &mod->source_list, source_list) {
L
Linus Torvalds 已提交
1011
		printed_something = 1;
1012
		seq_printf(m, "%s,", use->source->name);
L
Linus Torvalds 已提交
1013 1014 1015 1016
	}

	if (mod->init != NULL && mod->exit == NULL) {
		printed_something = 1;
1017
		seq_puts(m, "[permanent],");
L
Linus Torvalds 已提交
1018 1019 1020
	}

	if (!printed_something)
1021
		seq_puts(m, "-");
L
Linus Torvalds 已提交
1022 1023 1024 1025 1026 1027
}

void __symbol_put(const char *symbol)
{
	struct module *owner;

R
Rusty Russell 已提交
1028
	preempt_disable();
1029
	if (!find_symbol(symbol, &owner, NULL, true, false))
L
Linus Torvalds 已提交
1030 1031
		BUG();
	module_put(owner);
R
Rusty Russell 已提交
1032
	preempt_enable();
L
Linus Torvalds 已提交
1033 1034 1035
}
EXPORT_SYMBOL(__symbol_put);

1036
/* Note this assumes addr is a function, which it currently always is. */
L
Linus Torvalds 已提交
1037 1038
void symbol_put_addr(void *addr)
{
1039
	struct module *modaddr;
1040
	unsigned long a = (unsigned long)dereference_function_descriptor(addr);
L
Linus Torvalds 已提交
1041

1042
	if (core_kernel_text(a))
1043
		return;
L
Linus Torvalds 已提交
1044

1045 1046 1047 1048 1049
	/*
	 * Even though we hold a reference on the module; we still need to
	 * disable preemption in order to safely traverse the data structure.
	 */
	preempt_disable();
1050
	modaddr = __module_text_address(a);
1051
	BUG_ON(!modaddr);
1052
	module_put(modaddr);
1053
	preempt_enable();
L
Linus Torvalds 已提交
1054 1055 1056 1057
}
EXPORT_SYMBOL_GPL(symbol_put_addr);

static ssize_t show_refcnt(struct module_attribute *mattr,
1058
			   struct module_kobject *mk, char *buffer)
L
Linus Torvalds 已提交
1059
{
1060
	return sprintf(buffer, "%i\n", module_refcount(mk->mod));
L
Linus Torvalds 已提交
1061 1062
}

1063 1064
static struct module_attribute modinfo_refcnt =
	__ATTR(refcnt, 0444, show_refcnt, NULL);
L
Linus Torvalds 已提交
1065

1066 1067 1068 1069
void __module_get(struct module *module)
{
	if (module) {
		preempt_disable();
1070
		atomic_inc(&module->refcnt);
1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082
		trace_module_get(module, _RET_IP_);
		preempt_enable();
	}
}
EXPORT_SYMBOL(__module_get);

bool try_module_get(struct module *module)
{
	bool ret = true;

	if (module) {
		preempt_disable();
1083 1084 1085
		/* Note: here, we can fail to get a reference */
		if (likely(module_is_live(module) &&
			   atomic_inc_not_zero(&module->refcnt) != 0))
1086
			trace_module_get(module, _RET_IP_);
1087
		else
1088 1089 1090 1091 1092 1093 1094 1095
			ret = false;

		preempt_enable();
	}
	return ret;
}
EXPORT_SYMBOL(try_module_get);

A
Al Viro 已提交
1096 1097
void module_put(struct module *module)
{
1098 1099
	int ret;

A
Al Viro 已提交
1100
	if (module) {
1101
		preempt_disable();
1102 1103
		ret = atomic_dec_if_positive(&module->refcnt);
		WARN_ON(ret < 0);	/* Failed to put refcount */
1104
		trace_module_put(module, _RET_IP_);
1105
		preempt_enable();
A
Al Viro 已提交
1106 1107 1108 1109
	}
}
EXPORT_SYMBOL(module_put);

L
Linus Torvalds 已提交
1110
#else /* !CONFIG_MODULE_UNLOAD */
1111
static inline void print_unload_info(struct seq_file *m, struct module *mod)
L
Linus Torvalds 已提交
1112 1113
{
	/* We don't know the usage count, or what modules are using. */
1114
	seq_puts(m, " - -");
L
Linus Torvalds 已提交
1115 1116 1117 1118 1119 1120
}

static inline void module_unload_free(struct module *mod)
{
}

1121
int ref_module(struct module *a, struct module *b)
L
Linus Torvalds 已提交
1122
{
1123
	return strong_try_module_get(b);
L
Linus Torvalds 已提交
1124
}
1125
EXPORT_SYMBOL_GPL(ref_module);
L
Linus Torvalds 已提交
1126

1127
static inline int module_unload_init(struct module *mod)
L
Linus Torvalds 已提交
1128
{
1129
	return 0;
L
Linus Torvalds 已提交
1130 1131 1132
}
#endif /* CONFIG_MODULE_UNLOAD */

1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144
static size_t module_flags_taint(struct module *mod, char *buf)
{
	size_t l = 0;

	if (mod->taints & (1 << TAINT_PROPRIETARY_MODULE))
		buf[l++] = 'P';
	if (mod->taints & (1 << TAINT_OOT_MODULE))
		buf[l++] = 'O';
	if (mod->taints & (1 << TAINT_FORCED_MODULE))
		buf[l++] = 'F';
	if (mod->taints & (1 << TAINT_CRAP))
		buf[l++] = 'C';
1145
	if (mod->taints & (1 << TAINT_UNSIGNED_MODULE))
1146
		buf[l++] = 'E';
1147 1148
	/*
	 * TAINT_FORCED_RMMOD: could be added.
1149
	 * TAINT_CPU_OUT_OF_SPEC, TAINT_MACHINE_CHECK, TAINT_BAD_PAGE don't
1150 1151 1152 1153 1154
	 * apply to modules.
	 */
	return l;
}

1155
static ssize_t show_initstate(struct module_attribute *mattr,
1156
			      struct module_kobject *mk, char *buffer)
1157 1158 1159
{
	const char *state = "unknown";

1160
	switch (mk->mod->state) {
1161 1162 1163 1164 1165 1166 1167 1168 1169
	case MODULE_STATE_LIVE:
		state = "live";
		break;
	case MODULE_STATE_COMING:
		state = "coming";
		break;
	case MODULE_STATE_GOING:
		state = "going";
		break;
1170 1171
	default:
		BUG();
1172 1173 1174 1175
	}
	return sprintf(buffer, "%s\n", state);
}

1176 1177
static struct module_attribute modinfo_initstate =
	__ATTR(initstate, 0444, show_initstate, NULL);
1178

1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189
static ssize_t store_uevent(struct module_attribute *mattr,
			    struct module_kobject *mk,
			    const char *buffer, size_t count)
{
	enum kobject_action action;

	if (kobject_action_type(buffer, count, &action) == 0)
		kobject_uevent(&mk->kobj, action);
	return count;
}

1190 1191 1192 1193 1194 1195
struct module_attribute module_uevent =
	__ATTR(uevent, 0200, NULL, store_uevent);

static ssize_t show_coresize(struct module_attribute *mattr,
			     struct module_kobject *mk, char *buffer)
{
1196
	return sprintf(buffer, "%u\n", mk->mod->core_layout.size);
1197 1198 1199 1200 1201 1202 1203 1204
}

static struct module_attribute modinfo_coresize =
	__ATTR(coresize, 0444, show_coresize, NULL);

static ssize_t show_initsize(struct module_attribute *mattr,
			     struct module_kobject *mk, char *buffer)
{
1205
	return sprintf(buffer, "%u\n", mk->mod->init_layout.size);
1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222
}

static struct module_attribute modinfo_initsize =
	__ATTR(initsize, 0444, show_initsize, NULL);

static ssize_t show_taint(struct module_attribute *mattr,
			  struct module_kobject *mk, char *buffer)
{
	size_t l;

	l = module_flags_taint(mk->mod, buffer);
	buffer[l++] = '\n';
	return l;
}

static struct module_attribute modinfo_taint =
	__ATTR(taint, 0444, show_taint, NULL);
1223

1224
static struct module_attribute *modinfo_attrs[] = {
1225
	&module_uevent,
1226 1227
	&modinfo_version,
	&modinfo_srcversion,
1228 1229 1230 1231
	&modinfo_initstate,
	&modinfo_coresize,
	&modinfo_initsize,
	&modinfo_taint,
1232
#ifdef CONFIG_MODULE_UNLOAD
1233
	&modinfo_refcnt,
1234 1235 1236 1237
#endif
	NULL,
};

L
Linus Torvalds 已提交
1238 1239
static const char vermagic[] = VERMAGIC_STRING;

1240
static int try_to_force_load(struct module *mod, const char *reason)
1241 1242
{
#ifdef CONFIG_MODULE_FORCE_LOAD
A
Andi Kleen 已提交
1243
	if (!test_taint(TAINT_FORCED_MODULE))
A
Andrew Morton 已提交
1244
		pr_warn("%s: %s: kernel tainted.\n", mod->name, reason);
1245
	add_taint_module(mod, TAINT_FORCED_MODULE, LOCKDEP_NOW_UNRELIABLE);
1246 1247 1248 1249 1250 1251
	return 0;
#else
	return -ENOEXEC;
#endif
}

L
Linus Torvalds 已提交
1252
#ifdef CONFIG_MODVERSIONS
1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263
/* If the arch applies (non-zero) relocations to kernel kcrctab, unapply it. */
static unsigned long maybe_relocated(unsigned long crc,
				     const struct module *crc_owner)
{
#ifdef ARCH_RELOCATES_KCRCTAB
	if (crc_owner == NULL)
		return crc - (unsigned long)reloc_start;
#endif
	return crc;
}

L
Linus Torvalds 已提交
1264 1265 1266
static int check_version(Elf_Shdr *sechdrs,
			 unsigned int versindex,
			 const char *symname,
1267
			 struct module *mod,
1268 1269
			 const unsigned long *crc,
			 const struct module *crc_owner)
L
Linus Torvalds 已提交
1270 1271 1272 1273 1274 1275 1276 1277
{
	unsigned int i, num_versions;
	struct modversion_info *versions;

	/* Exporting module didn't supply crcs?  OK, we're already tainted. */
	if (!crc)
		return 1;

1278 1279 1280 1281
	/* No versions at all?  modprobe --force does this. */
	if (versindex == 0)
		return try_to_force_load(mod, symname) == 0;

L
Linus Torvalds 已提交
1282 1283 1284 1285 1286 1287 1288 1289
	versions = (void *) sechdrs[versindex].sh_addr;
	num_versions = sechdrs[versindex].sh_size
		/ sizeof(struct modversion_info);

	for (i = 0; i < num_versions; i++) {
		if (strcmp(versions[i].name, symname) != 0)
			continue;

1290
		if (versions[i].crc == maybe_relocated(*crc, crc_owner))
L
Linus Torvalds 已提交
1291
			return 1;
1292
		pr_debug("Found checksum %lX vs module %lX\n",
1293
		       maybe_relocated(*crc, crc_owner), versions[i].crc);
1294
		goto bad_version;
L
Linus Torvalds 已提交
1295
	}
1296

A
Andrew Morton 已提交
1297
	pr_warn("%s: no symbol version for %s\n", mod->name, symname);
1298
	return 0;
1299 1300

bad_version:
1301
	pr_warn("%s: disagrees about version of symbol %s\n",
1302 1303
	       mod->name, symname);
	return 0;
L
Linus Torvalds 已提交
1304 1305 1306 1307 1308 1309 1310 1311
}

static inline int check_modstruct_version(Elf_Shdr *sechdrs,
					  unsigned int versindex,
					  struct module *mod)
{
	const unsigned long *crc;

1312 1313 1314 1315 1316
	/*
	 * Since this should be found in kernel (which can't be removed), no
	 * locking is necessary -- use preempt_disable() to placate lockdep.
	 */
	preempt_disable();
R
Rusty Russell 已提交
1317
	if (!find_symbol(VMLINUX_SYMBOL_STR(module_layout), NULL,
1318 1319
			 &crc, true, false)) {
		preempt_enable();
L
Linus Torvalds 已提交
1320
		BUG();
1321 1322
	}
	preempt_enable();
1323 1324
	return check_version(sechdrs, versindex,
			     VMLINUX_SYMBOL_STR(module_layout), mod, crc,
1325
			     NULL);
L
Linus Torvalds 已提交
1326 1327
}

1328 1329 1330
/* First part is kernel version, which we ignore if module has crcs. */
static inline int same_magic(const char *amagic, const char *bmagic,
			     bool has_crcs)
L
Linus Torvalds 已提交
1331
{
1332 1333 1334 1335
	if (has_crcs) {
		amagic += strcspn(amagic, " ");
		bmagic += strcspn(bmagic, " ");
	}
L
Linus Torvalds 已提交
1336 1337 1338 1339 1340 1341
	return strcmp(amagic, bmagic) == 0;
}
#else
static inline int check_version(Elf_Shdr *sechdrs,
				unsigned int versindex,
				const char *symname,
1342
				struct module *mod,
1343 1344
				const unsigned long *crc,
				const struct module *crc_owner)
L
Linus Torvalds 已提交
1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355
{
	return 1;
}

static inline int check_modstruct_version(Elf_Shdr *sechdrs,
					  unsigned int versindex,
					  struct module *mod)
{
	return 1;
}

1356 1357
static inline int same_magic(const char *amagic, const char *bmagic,
			     bool has_crcs)
L
Linus Torvalds 已提交
1358 1359 1360 1361 1362
{
	return strcmp(amagic, bmagic) == 0;
}
#endif /* CONFIG_MODVERSIONS */

1363
/* Resolve a symbol for this module.  I.e. if we find one, record usage. */
1364 1365
static const struct kernel_symbol *resolve_symbol(struct module *mod,
						  const struct load_info *info,
1366
						  const char *name,
1367
						  char ownername[])
L
Linus Torvalds 已提交
1368 1369
{
	struct module *owner;
1370
	const struct kernel_symbol *sym;
L
Linus Torvalds 已提交
1371
	const unsigned long *crc;
1372
	int err;
L
Linus Torvalds 已提交
1373

1374 1375 1376 1377 1378 1379
	/*
	 * The module_mutex should not be a heavily contended lock;
	 * if we get the occasional sleep here, we'll go an extra iteration
	 * in the wait_event_interruptible(), which is harmless.
	 */
	sched_annotate_sleep();
1380
	mutex_lock(&module_mutex);
1381
	sym = find_symbol(name, &owner, &crc,
A
Andi Kleen 已提交
1382
			  !(mod->taints & (1 << TAINT_PROPRIETARY_MODULE)), true);
1383 1384 1385
	if (!sym)
		goto unlock;

1386 1387
	if (!check_version(info->sechdrs, info->index.vers, name, mod, crc,
			   owner)) {
1388 1389
		sym = ERR_PTR(-EINVAL);
		goto getname;
L
Linus Torvalds 已提交
1390
	}
1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401

	err = ref_module(mod, owner);
	if (err) {
		sym = ERR_PTR(err);
		goto getname;
	}

getname:
	/* We must make copy under the lock if we failed to get ref. */
	strncpy(ownername, module_name(owner), MODULE_NAME_LEN);
unlock:
1402
	mutex_unlock(&module_mutex);
1403
	return sym;
L
Linus Torvalds 已提交
1404 1405
}

1406 1407 1408 1409
static const struct kernel_symbol *
resolve_symbol_wait(struct module *mod,
		    const struct load_info *info,
		    const char *name)
1410 1411
{
	const struct kernel_symbol *ksym;
1412
	char owner[MODULE_NAME_LEN];
1413 1414

	if (wait_event_interruptible_timeout(module_wq,
1415 1416
			!IS_ERR(ksym = resolve_symbol(mod, info, name, owner))
			|| PTR_ERR(ksym) != -EBUSY,
1417
					     30 * HZ) <= 0) {
A
Andrew Morton 已提交
1418 1419
		pr_warn("%s: gave up waiting for init of module %s.\n",
			mod->name, owner);
1420 1421 1422 1423
	}
	return ksym;
}

L
Linus Torvalds 已提交
1424 1425 1426 1427
/*
 * /sys/module/foo/sections stuff
 * J. Corbet <corbet@lwn.net>
 */
R
Rusty Russell 已提交
1428
#ifdef CONFIG_SYSFS
1429

R
Rusty Russell 已提交
1430
#ifdef CONFIG_KALLSYMS
1431 1432 1433 1434 1435
static inline bool sect_empty(const Elf_Shdr *sect)
{
	return !(sect->sh_flags & SHF_ALLOC) || sect->sh_size == 0;
}

1436
struct module_sect_attr {
1437 1438 1439 1440 1441
	struct module_attribute mattr;
	char *name;
	unsigned long address;
};

1442
struct module_sect_attrs {
1443 1444 1445 1446 1447
	struct attribute_group grp;
	unsigned int nsections;
	struct module_sect_attr attrs[0];
};

L
Linus Torvalds 已提交
1448
static ssize_t module_sect_show(struct module_attribute *mattr,
1449
				struct module_kobject *mk, char *buf)
L
Linus Torvalds 已提交
1450 1451 1452
{
	struct module_sect_attr *sattr =
		container_of(mattr, struct module_sect_attr, mattr);
1453
	return sprintf(buf, "0x%pK\n", (void *)sattr->address);
L
Linus Torvalds 已提交
1454 1455
}

1456 1457
static void free_sect_attrs(struct module_sect_attrs *sect_attrs)
{
1458
	unsigned int section;
1459 1460 1461 1462 1463 1464

	for (section = 0; section < sect_attrs->nsections; section++)
		kfree(sect_attrs->attrs[section].name);
	kfree(sect_attrs);
}

R
Rusty Russell 已提交
1465
static void add_sect_attrs(struct module *mod, const struct load_info *info)
L
Linus Torvalds 已提交
1466 1467 1468 1469 1470
{
	unsigned int nloaded = 0, i, size[2];
	struct module_sect_attrs *sect_attrs;
	struct module_sect_attr *sattr;
	struct attribute **gattr;
D
Daniel Walker 已提交
1471

L
Linus Torvalds 已提交
1472
	/* Count loaded sections and allocate structures */
R
Rusty Russell 已提交
1473 1474
	for (i = 0; i < info->hdr->e_shnum; i++)
		if (!sect_empty(&info->sechdrs[i]))
L
Linus Torvalds 已提交
1475 1476 1477 1478 1479
			nloaded++;
	size[0] = ALIGN(sizeof(*sect_attrs)
			+ nloaded * sizeof(sect_attrs->attrs[0]),
			sizeof(sect_attrs->grp.attrs[0]));
	size[1] = (nloaded + 1) * sizeof(sect_attrs->grp.attrs[0]);
1480 1481
	sect_attrs = kzalloc(size[0] + size[1], GFP_KERNEL);
	if (sect_attrs == NULL)
L
Linus Torvalds 已提交
1482 1483 1484 1485 1486 1487
		return;

	/* Setup section attributes. */
	sect_attrs->grp.name = "sections";
	sect_attrs->grp.attrs = (void *)sect_attrs + size[0];

1488
	sect_attrs->nsections = 0;
L
Linus Torvalds 已提交
1489 1490
	sattr = &sect_attrs->attrs[0];
	gattr = &sect_attrs->grp.attrs[0];
R
Rusty Russell 已提交
1491 1492 1493
	for (i = 0; i < info->hdr->e_shnum; i++) {
		Elf_Shdr *sec = &info->sechdrs[i];
		if (sect_empty(sec))
1494
			continue;
R
Rusty Russell 已提交
1495 1496
		sattr->address = sec->sh_addr;
		sattr->name = kstrdup(info->secstrings + sec->sh_name,
1497 1498 1499 1500
					GFP_KERNEL);
		if (sattr->name == NULL)
			goto out;
		sect_attrs->nsections++;
1501
		sysfs_attr_init(&sattr->mattr.attr);
L
Linus Torvalds 已提交
1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515
		sattr->mattr.show = module_sect_show;
		sattr->mattr.store = NULL;
		sattr->mattr.attr.name = sattr->name;
		sattr->mattr.attr.mode = S_IRUGO;
		*(gattr++) = &(sattr++)->mattr.attr;
	}
	*gattr = NULL;

	if (sysfs_create_group(&mod->mkobj.kobj, &sect_attrs->grp))
		goto out;

	mod->sect_attrs = sect_attrs;
	return;
  out:
1516
	free_sect_attrs(sect_attrs);
L
Linus Torvalds 已提交
1517 1518 1519 1520 1521 1522 1523 1524 1525
}

static void remove_sect_attrs(struct module *mod)
{
	if (mod->sect_attrs) {
		sysfs_remove_group(&mod->mkobj.kobj,
				   &mod->sect_attrs->grp);
		/* We are positive that no one is using any sect attrs
		 * at this point.  Deallocate immediately. */
1526
		free_sect_attrs(mod->sect_attrs);
L
Linus Torvalds 已提交
1527 1528 1529 1530
		mod->sect_attrs = NULL;
	}
}

R
Roland McGrath 已提交
1531 1532 1533 1534 1535 1536 1537 1538 1539 1540
/*
 * /sys/module/foo/notes/.section.name gives contents of SHT_NOTE sections.
 */

struct module_notes_attrs {
	struct kobject *dir;
	unsigned int notes;
	struct bin_attribute attrs[0];
};

1541
static ssize_t module_notes_read(struct file *filp, struct kobject *kobj,
R
Roland McGrath 已提交
1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558
				 struct bin_attribute *bin_attr,
				 char *buf, loff_t pos, size_t count)
{
	/*
	 * The caller checked the pos and count against our size.
	 */
	memcpy(buf, bin_attr->private + pos, count);
	return count;
}

static void free_notes_attrs(struct module_notes_attrs *notes_attrs,
			     unsigned int i)
{
	if (notes_attrs->dir) {
		while (i-- > 0)
			sysfs_remove_bin_file(notes_attrs->dir,
					      &notes_attrs->attrs[i]);
1559
		kobject_put(notes_attrs->dir);
R
Roland McGrath 已提交
1560 1561 1562 1563
	}
	kfree(notes_attrs);
}

R
Rusty Russell 已提交
1564
static void add_notes_attrs(struct module *mod, const struct load_info *info)
R
Roland McGrath 已提交
1565 1566 1567 1568 1569
{
	unsigned int notes, loaded, i;
	struct module_notes_attrs *notes_attrs;
	struct bin_attribute *nattr;

1570 1571 1572 1573
	/* failed to create section attributes, so can't create notes */
	if (!mod->sect_attrs)
		return;

R
Roland McGrath 已提交
1574 1575
	/* Count notes sections and allocate structures.  */
	notes = 0;
R
Rusty Russell 已提交
1576 1577 1578
	for (i = 0; i < info->hdr->e_shnum; i++)
		if (!sect_empty(&info->sechdrs[i]) &&
		    (info->sechdrs[i].sh_type == SHT_NOTE))
R
Roland McGrath 已提交
1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591
			++notes;

	if (notes == 0)
		return;

	notes_attrs = kzalloc(sizeof(*notes_attrs)
			      + notes * sizeof(notes_attrs->attrs[0]),
			      GFP_KERNEL);
	if (notes_attrs == NULL)
		return;

	notes_attrs->notes = notes;
	nattr = &notes_attrs->attrs[0];
R
Rusty Russell 已提交
1592 1593
	for (loaded = i = 0; i < info->hdr->e_shnum; ++i) {
		if (sect_empty(&info->sechdrs[i]))
R
Roland McGrath 已提交
1594
			continue;
R
Rusty Russell 已提交
1595
		if (info->sechdrs[i].sh_type == SHT_NOTE) {
1596
			sysfs_bin_attr_init(nattr);
R
Roland McGrath 已提交
1597 1598
			nattr->attr.name = mod->sect_attrs->attrs[loaded].name;
			nattr->attr.mode = S_IRUGO;
R
Rusty Russell 已提交
1599 1600
			nattr->size = info->sechdrs[i].sh_size;
			nattr->private = (void *) info->sechdrs[i].sh_addr;
R
Roland McGrath 已提交
1601 1602 1603 1604 1605 1606
			nattr->read = module_notes_read;
			++nattr;
		}
		++loaded;
	}

1607
	notes_attrs->dir = kobject_create_and_add("notes", &mod->mkobj.kobj);
R
Roland McGrath 已提交
1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628
	if (!notes_attrs->dir)
		goto out;

	for (i = 0; i < notes; ++i)
		if (sysfs_create_bin_file(notes_attrs->dir,
					  &notes_attrs->attrs[i]))
			goto out;

	mod->notes_attrs = notes_attrs;
	return;

  out:
	free_notes_attrs(notes_attrs, i);
}

static void remove_notes_attrs(struct module *mod)
{
	if (mod->notes_attrs)
		free_notes_attrs(mod->notes_attrs, mod->notes_attrs->notes);
}

L
Linus Torvalds 已提交
1629
#else
1630

R
Rusty Russell 已提交
1631 1632
static inline void add_sect_attrs(struct module *mod,
				  const struct load_info *info)
L
Linus Torvalds 已提交
1633 1634 1635 1636 1637 1638
{
}

static inline void remove_sect_attrs(struct module *mod)
{
}
R
Roland McGrath 已提交
1639

R
Rusty Russell 已提交
1640 1641
static inline void add_notes_attrs(struct module *mod,
				   const struct load_info *info)
R
Roland McGrath 已提交
1642 1643 1644 1645 1646 1647
{
}

static inline void remove_notes_attrs(struct module *mod)
{
}
R
Rusty Russell 已提交
1648
#endif /* CONFIG_KALLSYMS */
L
Linus Torvalds 已提交
1649

1650 1651 1652 1653 1654 1655
static void add_usage_links(struct module *mod)
{
#ifdef CONFIG_MODULE_UNLOAD
	struct module_use *use;
	int nowarn;

1656
	mutex_lock(&module_mutex);
1657 1658 1659 1660
	list_for_each_entry(use, &mod->target_list, target_list) {
		nowarn = sysfs_create_link(use->target->holders_dir,
					   &mod->mkobj.kobj, mod->name);
	}
1661
	mutex_unlock(&module_mutex);
1662 1663 1664 1665 1666 1667 1668 1669
#endif
}

static void del_usage_links(struct module *mod)
{
#ifdef CONFIG_MODULE_UNLOAD
	struct module_use *use;

1670
	mutex_lock(&module_mutex);
1671 1672
	list_for_each_entry(use, &mod->target_list, target_list)
		sysfs_remove_link(use->target->holders_dir, mod->name);
1673
	mutex_unlock(&module_mutex);
1674 1675 1676
#endif
}

1677
static int module_add_modinfo_attrs(struct module *mod)
1678 1679
{
	struct module_attribute *attr;
1680
	struct module_attribute *temp_attr;
1681 1682 1683
	int error = 0;
	int i;

1684 1685 1686 1687 1688 1689 1690
	mod->modinfo_attrs = kzalloc((sizeof(struct module_attribute) *
					(ARRAY_SIZE(modinfo_attrs) + 1)),
					GFP_KERNEL);
	if (!mod->modinfo_attrs)
		return -ENOMEM;

	temp_attr = mod->modinfo_attrs;
1691 1692
	for (i = 0; (attr = modinfo_attrs[i]) && !error; i++) {
		if (!attr->test ||
1693 1694
		    (attr->test && attr->test(mod))) {
			memcpy(temp_attr, attr, sizeof(*temp_attr));
1695
			sysfs_attr_init(&temp_attr->attr);
1696 1697
			error = sysfs_create_file(&mod->mkobj.kobj,
					&temp_attr->attr);
1698 1699
			++temp_attr;
		}
1700 1701 1702 1703
	}
	return error;
}

1704
static void module_remove_modinfo_attrs(struct module *mod)
1705 1706 1707 1708
{
	struct module_attribute *attr;
	int i;

1709 1710 1711 1712
	for (i = 0; (attr = &mod->modinfo_attrs[i]); i++) {
		/* pick a field to test for end of list */
		if (!attr->attr.name)
			break;
1713
		sysfs_remove_file(&mod->mkobj.kobj, &attr->attr);
1714 1715
		if (attr->free)
			attr->free(mod);
1716
	}
1717
	kfree(mod->modinfo_attrs);
1718
}
L
Linus Torvalds 已提交
1719

1720 1721 1722 1723 1724 1725 1726 1727
static void mod_kobject_put(struct module *mod)
{
	DECLARE_COMPLETION_ONSTACK(c);
	mod->mkobj.kobj_completion = &c;
	kobject_put(&mod->mkobj.kobj);
	wait_for_completion(&c);
}

1728
static int mod_sysfs_init(struct module *mod)
L
Linus Torvalds 已提交
1729 1730
{
	int err;
1731
	struct kobject *kobj;
L
Linus Torvalds 已提交
1732

1733
	if (!module_sysfs_initialized) {
A
Andrew Morton 已提交
1734
		pr_err("%s: module sysfs not initialized\n", mod->name);
1735 1736 1737
		err = -EINVAL;
		goto out;
	}
1738 1739 1740

	kobj = kset_find_obj(module_kset, mod->name);
	if (kobj) {
A
Andrew Morton 已提交
1741
		pr_err("%s: module is already loaded\n", mod->name);
1742 1743 1744 1745 1746
		kobject_put(kobj);
		err = -EINVAL;
		goto out;
	}

L
Linus Torvalds 已提交
1747
	mod->mkobj.mod = mod;
1748

1749 1750 1751 1752 1753
	memset(&mod->mkobj.kobj, 0, sizeof(mod->mkobj.kobj));
	mod->mkobj.kobj.kset = module_kset;
	err = kobject_init_and_add(&mod->mkobj.kobj, &module_ktype, NULL,
				   "%s", mod->name);
	if (err)
1754
		mod_kobject_put(mod);
K
Kay Sievers 已提交
1755

1756
	/* delay uevent until full sysfs population */
K
Kay Sievers 已提交
1757 1758 1759 1760
out:
	return err;
}

1761
static int mod_sysfs_setup(struct module *mod,
R
Rusty Russell 已提交
1762
			   const struct load_info *info,
K
Kay Sievers 已提交
1763 1764 1765 1766 1767
			   struct kernel_param *kparam,
			   unsigned int num_params)
{
	int err;

1768 1769 1770 1771
	err = mod_sysfs_init(mod);
	if (err)
		goto out;

1772
	mod->holders_dir = kobject_create_and_add("holders", &mod->mkobj.kobj);
1773 1774
	if (!mod->holders_dir) {
		err = -ENOMEM;
K
Kay Sievers 已提交
1775
		goto out_unreg;
1776
	}
K
Kay Sievers 已提交
1777

L
Linus Torvalds 已提交
1778 1779
	err = module_param_sysfs_setup(mod, kparam, num_params);
	if (err)
K
Kay Sievers 已提交
1780
		goto out_unreg_holders;
L
Linus Torvalds 已提交
1781

1782 1783
	err = module_add_modinfo_attrs(mod);
	if (err)
1784
		goto out_unreg_param;
1785

1786
	add_usage_links(mod);
R
Rusty Russell 已提交
1787 1788
	add_sect_attrs(mod, info);
	add_notes_attrs(mod, info);
1789

1790
	kobject_uevent(&mod->mkobj.kobj, KOBJ_ADD);
L
Linus Torvalds 已提交
1791 1792
	return 0;

1793 1794
out_unreg_param:
	module_param_sysfs_remove(mod);
K
Kay Sievers 已提交
1795
out_unreg_holders:
1796
	kobject_put(mod->holders_dir);
K
Kay Sievers 已提交
1797
out_unreg:
1798
	mod_kobject_put(mod);
1799
out:
L
Linus Torvalds 已提交
1800 1801
	return err;
}
1802 1803 1804

static void mod_sysfs_fini(struct module *mod)
{
R
Rusty Russell 已提交
1805 1806
	remove_notes_attrs(mod);
	remove_sect_attrs(mod);
1807
	mod_kobject_put(mod);
1808 1809
}

1810 1811 1812 1813
static void init_param_lock(struct module *mod)
{
	mutex_init(&mod->param_lock);
}
R
Rusty Russell 已提交
1814
#else /* !CONFIG_SYSFS */
1815

R
Rusty Russell 已提交
1816 1817
static int mod_sysfs_setup(struct module *mod,
			   const struct load_info *info,
1818 1819 1820 1821 1822 1823
			   struct kernel_param *kparam,
			   unsigned int num_params)
{
	return 0;
}

1824 1825 1826 1827
static void mod_sysfs_fini(struct module *mod)
{
}

1828 1829 1830 1831
static void module_remove_modinfo_attrs(struct module *mod)
{
}

1832 1833 1834 1835
static void del_usage_links(struct module *mod)
{
}

1836 1837 1838
static void init_param_lock(struct module *mod)
{
}
1839
#endif /* CONFIG_SYSFS */
L
Linus Torvalds 已提交
1840

1841
static void mod_sysfs_teardown(struct module *mod)
L
Linus Torvalds 已提交
1842
{
1843
	del_usage_links(mod);
1844
	module_remove_modinfo_attrs(mod);
L
Linus Torvalds 已提交
1845
	module_param_sysfs_remove(mod);
1846 1847
	kobject_put(mod->mkobj.drivers_dir);
	kobject_put(mod->holders_dir);
1848
	mod_sysfs_fini(mod);
L
Linus Torvalds 已提交
1849 1850
}

1851 1852 1853 1854
#ifdef CONFIG_DEBUG_SET_MODULE_RONX
/*
 * LKM RO/NX protection: protect module's text/ro-data
 * from modification and any data from execution.
R
Rusty Russell 已提交
1855 1856 1857 1858 1859 1860 1861 1862
 *
 * General layout of module is:
 *          [text] [read-only-data] [writable data]
 * text_size -----^                ^               ^
 * ro_size ------------------------|               |
 * size -------------------------------------------|
 *
 * These values are always page-aligned (as is base)
1863
 */
R
Rusty Russell 已提交
1864 1865
static void frob_text(const struct module_layout *layout,
		      int (*set_memory)(unsigned long start, int num_pages))
1866
{
R
Rusty Russell 已提交
1867 1868 1869 1870
	BUG_ON((unsigned long)layout->base & (PAGE_SIZE-1));
	BUG_ON((unsigned long)layout->text_size & (PAGE_SIZE-1));
	set_memory((unsigned long)layout->base,
		   layout->text_size >> PAGE_SHIFT);
1871 1872
}

R
Rusty Russell 已提交
1873 1874
static void frob_rodata(const struct module_layout *layout,
			int (*set_memory)(unsigned long start, int num_pages))
1875
{
R
Rusty Russell 已提交
1876 1877 1878 1879 1880
	BUG_ON((unsigned long)layout->base & (PAGE_SIZE-1));
	BUG_ON((unsigned long)layout->text_size & (PAGE_SIZE-1));
	BUG_ON((unsigned long)layout->ro_size & (PAGE_SIZE-1));
	set_memory((unsigned long)layout->base + layout->text_size,
		   (layout->ro_size - layout->text_size) >> PAGE_SHIFT);
1881 1882
}

R
Rusty Russell 已提交
1883 1884
static void frob_writable_data(const struct module_layout *layout,
			       int (*set_memory)(unsigned long start, int num_pages))
1885
{
R
Rusty Russell 已提交
1886 1887 1888 1889 1890
	BUG_ON((unsigned long)layout->base & (PAGE_SIZE-1));
	BUG_ON((unsigned long)layout->ro_size & (PAGE_SIZE-1));
	BUG_ON((unsigned long)layout->size & (PAGE_SIZE-1));
	set_memory((unsigned long)layout->base + layout->ro_size,
		   (layout->size - layout->ro_size) >> PAGE_SHIFT);
1891 1892
}

R
Rusty Russell 已提交
1893 1894
/* livepatching wants to disable read-only so it can frob module. */
void module_disable_ro(const struct module *mod)
1895
{
R
Rusty Russell 已提交
1896 1897 1898 1899
	frob_text(&mod->core_layout, set_memory_rw);
	frob_rodata(&mod->core_layout, set_memory_rw);
	frob_text(&mod->init_layout, set_memory_rw);
	frob_rodata(&mod->init_layout, set_memory_rw);
1900
}
1901

R
Rusty Russell 已提交
1902
void module_enable_ro(const struct module *mod)
1903
{
R
Rusty Russell 已提交
1904 1905 1906 1907
	frob_text(&mod->core_layout, set_memory_ro);
	frob_rodata(&mod->core_layout, set_memory_ro);
	frob_text(&mod->init_layout, set_memory_ro);
	frob_rodata(&mod->init_layout, set_memory_ro);
1908 1909
}

R
Rusty Russell 已提交
1910
static void module_enable_nx(const struct module *mod)
1911
{
R
Rusty Russell 已提交
1912 1913 1914 1915
	frob_rodata(&mod->core_layout, set_memory_nx);
	frob_writable_data(&mod->core_layout, set_memory_nx);
	frob_rodata(&mod->init_layout, set_memory_nx);
	frob_writable_data(&mod->init_layout, set_memory_nx);
1916 1917
}

R
Rusty Russell 已提交
1918
static void module_disable_nx(const struct module *mod)
1919
{
R
Rusty Russell 已提交
1920 1921 1922 1923
	frob_rodata(&mod->core_layout, set_memory_x);
	frob_writable_data(&mod->core_layout, set_memory_x);
	frob_rodata(&mod->init_layout, set_memory_x);
	frob_writable_data(&mod->init_layout, set_memory_x);
1924 1925 1926
}

/* Iterate through all modules and set each module's text as RW */
1927
void set_all_modules_text_rw(void)
1928 1929 1930 1931 1932
{
	struct module *mod;

	mutex_lock(&module_mutex);
	list_for_each_entry_rcu(mod, &modules, list) {
1933 1934
		if (mod->state == MODULE_STATE_UNFORMED)
			continue;
R
Rusty Russell 已提交
1935 1936 1937

		frob_text(&mod->core_layout, set_memory_rw);
		frob_text(&mod->init_layout, set_memory_rw);
1938 1939 1940 1941 1942
	}
	mutex_unlock(&module_mutex);
}

/* Iterate through all modules and set each module's text as RO */
1943
void set_all_modules_text_ro(void)
1944 1945 1946 1947 1948
{
	struct module *mod;

	mutex_lock(&module_mutex);
	list_for_each_entry_rcu(mod, &modules, list) {
1949 1950
		if (mod->state == MODULE_STATE_UNFORMED)
			continue;
R
Rusty Russell 已提交
1951 1952 1953

		frob_text(&mod->core_layout, set_memory_ro);
		frob_text(&mod->init_layout, set_memory_ro);
1954 1955 1956
	}
	mutex_unlock(&module_mutex);
}
R
Rusty Russell 已提交
1957 1958 1959 1960 1961 1962 1963 1964 1965

static void disable_ro_nx(const struct module_layout *layout)
{
	frob_text(layout, set_memory_rw);
	frob_rodata(layout, set_memory_rw);
	frob_rodata(layout, set_memory_x);
	frob_writable_data(layout, set_memory_x);
}

1966
#else
R
Rusty Russell 已提交
1967 1968 1969
static void disable_ro_nx(const struct module_layout *layout) { }
static void module_enable_nx(const struct module *mod) { }
static void module_disable_nx(const struct module *mod) { }
1970 1971
#endif

1972
void __weak module_memfree(void *module_region)
1973 1974 1975 1976 1977 1978 1979 1980
{
	vfree(module_region);
}

void __weak module_arch_cleanup(struct module *mod)
{
}

1981 1982 1983 1984
void __weak module_arch_freeing_init(struct module *mod)
{
}

1985
/* Free a module, remove from lists, etc. */
L
Linus Torvalds 已提交
1986 1987
static void free_module(struct module *mod)
{
1988 1989
	trace_module_free(mod);

1990
	mod_sysfs_teardown(mod);
L
Linus Torvalds 已提交
1991

1992 1993
	/* We leave it in list to prevent duplicate loads, but make sure
	 * that noone uses it while it's being deconstructed. */
1994
	mutex_lock(&module_mutex);
1995
	mod->state = MODULE_STATE_UNFORMED;
1996
	mutex_unlock(&module_mutex);
1997

1998 1999 2000
	/* Remove dynamic debug info */
	ddebug_remove_module(mod->name);

L
Linus Torvalds 已提交
2001 2002 2003 2004 2005 2006
	/* Arch-specific cleanup. */
	module_arch_cleanup(mod);

	/* Module unload stuff */
	module_unload_free(mod);

2007 2008 2009
	/* Free any allocated parameters. */
	destroy_params(mod->kp, mod->num_kp);

2010 2011
	/* Now we can delete it from the lists */
	mutex_lock(&module_mutex);
2012 2013
	/* Unlink carefully: kallsyms could be walking list. */
	list_del_rcu(&mod->list);
2014
	mod_tree_remove(mod);
2015
	/* Remove this module from bug list, this uses list_del_rcu */
2016
	module_bug_cleanup(mod);
2017 2018
	/* Wait for RCU-sched synchronizing before releasing mod->list and buglist. */
	synchronize_sched();
2019 2020
	mutex_unlock(&module_mutex);

R
Rusty Russell 已提交
2021 2022
	/* This may be empty, but that's OK */
	disable_ro_nx(&mod->init_layout);
2023
	module_arch_freeing_init(mod);
2024
	module_memfree(mod->init_layout.base);
L
Linus Torvalds 已提交
2025
	kfree(mod->args);
2026
	percpu_modfree(mod);
2027

2028
	/* Free lock-classes; relies on the preceding sync_rcu(). */
2029
	lockdep_free_key_range(mod->core_layout.base, mod->core_layout.size);
I
Ingo Molnar 已提交
2030

L
Linus Torvalds 已提交
2031
	/* Finally, free the core (containing the module structure) */
R
Rusty Russell 已提交
2032
	disable_ro_nx(&mod->core_layout);
2033
	module_memfree(mod->core_layout.base);
2034 2035 2036 2037

#ifdef CONFIG_MPU
	update_protections(current->mm);
#endif
L
Linus Torvalds 已提交
2038 2039 2040 2041 2042
}

void *__symbol_get(const char *symbol)
{
	struct module *owner;
2043
	const struct kernel_symbol *sym;
L
Linus Torvalds 已提交
2044

R
Rusty Russell 已提交
2045
	preempt_disable();
2046 2047 2048
	sym = find_symbol(symbol, &owner, NULL, true, true);
	if (sym && strong_try_module_get(owner))
		sym = NULL;
R
Rusty Russell 已提交
2049
	preempt_enable();
L
Linus Torvalds 已提交
2050

2051
	return sym ? (void *)sym->value : NULL;
L
Linus Torvalds 已提交
2052 2053 2054
}
EXPORT_SYMBOL_GPL(__symbol_get);

2055 2056
/*
 * Ensure that an exported symbol [global namespace] does not already exist
2057
 * in the kernel or in some other module's exported symbol table.
2058 2059
 *
 * You must hold the module_mutex.
2060 2061 2062
 */
static int verify_export_symbols(struct module *mod)
{
2063
	unsigned int i;
2064
	struct module *owner;
2065 2066 2067 2068 2069 2070 2071 2072
	const struct kernel_symbol *s;
	struct {
		const struct kernel_symbol *sym;
		unsigned int num;
	} arr[] = {
		{ mod->syms, mod->num_syms },
		{ mod->gpl_syms, mod->num_gpl_syms },
		{ mod->gpl_future_syms, mod->num_gpl_future_syms },
2073
#ifdef CONFIG_UNUSED_SYMBOLS
2074 2075
		{ mod->unused_syms, mod->num_unused_syms },
		{ mod->unused_gpl_syms, mod->num_unused_gpl_syms },
2076
#endif
2077
	};
2078

2079 2080
	for (i = 0; i < ARRAY_SIZE(arr); i++) {
		for (s = arr[i].sym; s < arr[i].sym + arr[i].num; s++) {
2081
			if (find_symbol(s->name, &owner, NULL, true, false)) {
A
Andrew Morton 已提交
2082
				pr_err("%s: exports duplicate symbol %s"
2083 2084 2085 2086
				       " (owned by %s)\n",
				       mod->name, s->name, module_name(owner));
				return -ENOEXEC;
			}
2087
		}
2088 2089
	}
	return 0;
2090 2091
}

2092
/* Change all symbols so that st_value encodes the pointer directly. */
2093 2094 2095 2096
static int simplify_symbols(struct module *mod, const struct load_info *info)
{
	Elf_Shdr *symsec = &info->sechdrs[info->index.sym];
	Elf_Sym *sym = (void *)symsec->sh_addr;
L
Linus Torvalds 已提交
2097
	unsigned long secbase;
2098
	unsigned int i;
L
Linus Torvalds 已提交
2099
	int ret = 0;
2100
	const struct kernel_symbol *ksym;
L
Linus Torvalds 已提交
2101

2102 2103 2104
	for (i = 1; i < symsec->sh_size / sizeof(Elf_Sym); i++) {
		const char *name = info->strtab + sym[i].st_name;

L
Linus Torvalds 已提交
2105 2106
		switch (sym[i].st_shndx) {
		case SHN_COMMON:
2107 2108 2109 2110
			/* Ignore common symbols */
			if (!strncmp(name, "__gnu_lto", 9))
				break;

L
Linus Torvalds 已提交
2111 2112
			/* We compiled with -fno-common.  These are not
			   supposed to happen.  */
2113
			pr_debug("Common symbol: %s\n", name);
2114
			pr_warn("%s: please compile with -fno-common\n",
L
Linus Torvalds 已提交
2115 2116 2117 2118 2119 2120
			       mod->name);
			ret = -ENOEXEC;
			break;

		case SHN_ABS:
			/* Don't need to do anything */
2121
			pr_debug("Absolute symbol: 0x%08lx\n",
L
Linus Torvalds 已提交
2122 2123 2124 2125
			       (long)sym[i].st_value);
			break;

		case SHN_UNDEF:
2126
			ksym = resolve_symbol_wait(mod, info, name);
L
Linus Torvalds 已提交
2127
			/* Ok if resolved.  */
2128
			if (ksym && !IS_ERR(ksym)) {
2129
				sym[i].st_value = ksym->value;
L
Linus Torvalds 已提交
2130
				break;
2131 2132
			}

L
Linus Torvalds 已提交
2133
			/* Ok if weak.  */
2134
			if (!ksym && ELF_ST_BIND(sym[i].st_info) == STB_WEAK)
L
Linus Torvalds 已提交
2135 2136
				break;

A
Andrew Morton 已提交
2137 2138
			pr_warn("%s: Unknown symbol %s (err %li)\n",
				mod->name, name, PTR_ERR(ksym));
2139
			ret = PTR_ERR(ksym) ?: -ENOENT;
L
Linus Torvalds 已提交
2140 2141 2142 2143
			break;

		default:
			/* Divert to percpu allocation if a percpu var. */
2144
			if (sym[i].st_shndx == info->index.pcpu)
2145
				secbase = (unsigned long)mod_percpu(mod);
L
Linus Torvalds 已提交
2146
			else
2147
				secbase = info->sechdrs[sym[i].st_shndx].sh_addr;
L
Linus Torvalds 已提交
2148 2149 2150 2151 2152 2153 2154 2155
			sym[i].st_value += secbase;
			break;
		}
	}

	return ret;
}

2156
static int apply_relocations(struct module *mod, const struct load_info *info)
2157 2158 2159 2160 2161
{
	unsigned int i;
	int err = 0;

	/* Now do relocations. */
2162 2163
	for (i = 1; i < info->hdr->e_shnum; i++) {
		unsigned int infosec = info->sechdrs[i].sh_info;
2164 2165

		/* Not a valid relocation section? */
2166
		if (infosec >= info->hdr->e_shnum)
2167 2168 2169
			continue;

		/* Don't bother with non-allocated sections */
2170
		if (!(info->sechdrs[infosec].sh_flags & SHF_ALLOC))
2171 2172
			continue;

2173 2174 2175 2176 2177 2178
		if (info->sechdrs[i].sh_type == SHT_REL)
			err = apply_relocate(info->sechdrs, info->strtab,
					     info->index.sym, i, mod);
		else if (info->sechdrs[i].sh_type == SHT_RELA)
			err = apply_relocate_add(info->sechdrs, info->strtab,
						 info->index.sym, i, mod);
2179 2180 2181 2182 2183 2184
		if (err < 0)
			break;
	}
	return err;
}

2185 2186 2187 2188 2189 2190 2191 2192
/* Additional bytes needed by arch in front of individual sections */
unsigned int __weak arch_mod_section_prepend(struct module *mod,
					     unsigned int section)
{
	/* default implementation just returns zero */
	return 0;
}

L
Linus Torvalds 已提交
2193
/* Update size with this section: return offset. */
2194 2195
static long get_offset(struct module *mod, unsigned int *size,
		       Elf_Shdr *sechdr, unsigned int section)
L
Linus Torvalds 已提交
2196 2197 2198
{
	long ret;

2199
	*size += arch_mod_section_prepend(mod, section);
L
Linus Torvalds 已提交
2200 2201 2202 2203 2204 2205 2206 2207 2208
	ret = ALIGN(*size, sechdr->sh_addralign ?: 1);
	*size = ret + sechdr->sh_size;
	return ret;
}

/* Lay out the SHF_ALLOC sections in a way not dissimilar to how ld
   might -- code, read-only data, read-write data, small data.  Tally
   sizes, and place the offsets into sh_entsize fields: high bit means it
   belongs in init. */
2209
static void layout_sections(struct module *mod, struct load_info *info)
L
Linus Torvalds 已提交
2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221
{
	static unsigned long const masks[][2] = {
		/* NOTE: all executable code must be the first section
		 * in this array; otherwise modify the text_size
		 * finder in the two loops below */
		{ SHF_EXECINSTR | SHF_ALLOC, ARCH_SHF_SMALL },
		{ SHF_ALLOC, SHF_WRITE | ARCH_SHF_SMALL },
		{ SHF_WRITE | SHF_ALLOC, ARCH_SHF_SMALL },
		{ ARCH_SHF_SMALL | SHF_ALLOC, 0 }
	};
	unsigned int m, i;

2222 2223
	for (i = 0; i < info->hdr->e_shnum; i++)
		info->sechdrs[i].sh_entsize = ~0UL;
L
Linus Torvalds 已提交
2224

2225
	pr_debug("Core section allocation order:\n");
L
Linus Torvalds 已提交
2226
	for (m = 0; m < ARRAY_SIZE(masks); ++m) {
2227 2228 2229
		for (i = 0; i < info->hdr->e_shnum; ++i) {
			Elf_Shdr *s = &info->sechdrs[i];
			const char *sname = info->secstrings + s->sh_name;
L
Linus Torvalds 已提交
2230 2231 2232 2233

			if ((s->sh_flags & masks[m][0]) != masks[m][0]
			    || (s->sh_flags & masks[m][1])
			    || s->sh_entsize != ~0UL
2234
			    || strstarts(sname, ".init"))
L
Linus Torvalds 已提交
2235
				continue;
2236
			s->sh_entsize = get_offset(mod, &mod->core_layout.size, s, i);
2237
			pr_debug("\t%s\n", sname);
L
Linus Torvalds 已提交
2238
		}
2239 2240
		switch (m) {
		case 0: /* executable */
2241 2242
			mod->core_layout.size = debug_align(mod->core_layout.size);
			mod->core_layout.text_size = mod->core_layout.size;
2243 2244
			break;
		case 1: /* RO: text and ro-data */
2245 2246
			mod->core_layout.size = debug_align(mod->core_layout.size);
			mod->core_layout.ro_size = mod->core_layout.size;
2247 2248
			break;
		case 3: /* whole core */
2249
			mod->core_layout.size = debug_align(mod->core_layout.size);
2250 2251
			break;
		}
L
Linus Torvalds 已提交
2252 2253
	}

2254
	pr_debug("Init section allocation order:\n");
L
Linus Torvalds 已提交
2255
	for (m = 0; m < ARRAY_SIZE(masks); ++m) {
2256 2257 2258
		for (i = 0; i < info->hdr->e_shnum; ++i) {
			Elf_Shdr *s = &info->sechdrs[i];
			const char *sname = info->secstrings + s->sh_name;
L
Linus Torvalds 已提交
2259 2260 2261 2262

			if ((s->sh_flags & masks[m][0]) != masks[m][0]
			    || (s->sh_flags & masks[m][1])
			    || s->sh_entsize != ~0UL
2263
			    || !strstarts(sname, ".init"))
L
Linus Torvalds 已提交
2264
				continue;
2265
			s->sh_entsize = (get_offset(mod, &mod->init_layout.size, s, i)
L
Linus Torvalds 已提交
2266
					 | INIT_OFFSET_MASK);
2267
			pr_debug("\t%s\n", sname);
L
Linus Torvalds 已提交
2268
		}
2269 2270
		switch (m) {
		case 0: /* executable */
2271 2272
			mod->init_layout.size = debug_align(mod->init_layout.size);
			mod->init_layout.text_size = mod->init_layout.size;
2273 2274
			break;
		case 1: /* RO: text and ro-data */
2275 2276
			mod->init_layout.size = debug_align(mod->init_layout.size);
			mod->init_layout.ro_size = mod->init_layout.size;
2277 2278
			break;
		case 3: /* whole init */
2279
			mod->init_layout.size = debug_align(mod->init_layout.size);
2280 2281
			break;
		}
L
Linus Torvalds 已提交
2282 2283 2284 2285 2286 2287 2288 2289
	}
}

static void set_license(struct module *mod, const char *license)
{
	if (!license)
		license = "unspecified";

2290
	if (!license_is_gpl_compatible(license)) {
A
Andi Kleen 已提交
2291
		if (!test_taint(TAINT_PROPRIETARY_MODULE))
A
Andrew Morton 已提交
2292 2293
			pr_warn("%s: module license '%s' taints kernel.\n",
				mod->name, license);
2294 2295
		add_taint_module(mod, TAINT_PROPRIETARY_MODULE,
				 LOCKDEP_NOW_UNRELIABLE);
L
Linus Torvalds 已提交
2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317
	}
}

/* Parse tag=value strings from .modinfo section */
static char *next_string(char *string, unsigned long *secsize)
{
	/* Skip non-zero chars */
	while (string[0]) {
		string++;
		if ((*secsize)-- <= 1)
			return NULL;
	}

	/* Skip any zero padding. */
	while (!string[0]) {
		string++;
		if ((*secsize)-- <= 1)
			return NULL;
	}
	return string;
}

2318
static char *get_modinfo(struct load_info *info, const char *tag)
L
Linus Torvalds 已提交
2319 2320 2321
{
	char *p;
	unsigned int taglen = strlen(tag);
2322 2323
	Elf_Shdr *infosec = &info->sechdrs[info->index.info];
	unsigned long size = infosec->sh_size;
L
Linus Torvalds 已提交
2324

2325
	for (p = (char *)infosec->sh_addr; p; p = next_string(p, &size)) {
L
Linus Torvalds 已提交
2326 2327 2328 2329 2330 2331
		if (strncmp(p, tag, taglen) == 0 && p[taglen] == '=')
			return p + taglen + 1;
	}
	return NULL;
}

2332
static void setup_modinfo(struct module *mod, struct load_info *info)
2333 2334 2335 2336 2337 2338
{
	struct module_attribute *attr;
	int i;

	for (i = 0; (attr = modinfo_attrs[i]); i++) {
		if (attr->setup)
2339
			attr->setup(mod, get_modinfo(info, attr->attr.name));
2340 2341 2342
	}
}

2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353
static void free_modinfo(struct module *mod)
{
	struct module_attribute *attr;
	int i;

	for (i = 0; (attr = modinfo_attrs[i]); i++) {
		if (attr->free)
			attr->free(mod);
	}
}

L
Linus Torvalds 已提交
2354
#ifdef CONFIG_KALLSYMS
2355 2356 2357 2358 2359 2360

/* lookup symbol in given range of kernel_symbols */
static const struct kernel_symbol *lookup_symbol(const char *name,
	const struct kernel_symbol *start,
	const struct kernel_symbol *stop)
{
2361 2362
	return bsearch(name, start, stop - start,
			sizeof(struct kernel_symbol), cmp_name);
2363 2364
}

2365 2366
static int is_exported(const char *name, unsigned long value,
		       const struct module *mod)
L
Linus Torvalds 已提交
2367
{
2368 2369 2370
	const struct kernel_symbol *ks;
	if (!mod)
		ks = lookup_symbol(name, __start___ksymtab, __stop___ksymtab);
2371
	else
2372 2373
		ks = lookup_symbol(name, mod->syms, mod->syms + mod->num_syms);
	return ks != NULL && ks->value == value;
L
Linus Torvalds 已提交
2374 2375 2376
}

/* As per nm */
2377
static char elf_type(const Elf_Sym *sym, const struct load_info *info)
L
Linus Torvalds 已提交
2378
{
2379 2380
	const Elf_Shdr *sechdrs = info->sechdrs;

L
Linus Torvalds 已提交
2381 2382 2383 2384 2385 2386 2387 2388
	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';
2389
	if (sym->st_shndx == SHN_ABS || sym->st_shndx == info->index.pcpu)
L
Linus Torvalds 已提交
2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409
		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';
	}
2410 2411
	if (strstarts(info->secstrings + sechdrs[sym->st_shndx].sh_name,
		      ".debug")) {
L
Linus Torvalds 已提交
2412
		return 'n';
2413
	}
L
Linus Torvalds 已提交
2414 2415 2416
	return '?';
}

2417
static bool is_core_symbol(const Elf_Sym *src, const Elf_Shdr *sechdrs,
2418
			unsigned int shnum, unsigned int pcpundx)
2419 2420 2421 2422 2423 2424 2425 2426
{
	const Elf_Shdr *sec;

	if (src->st_shndx == SHN_UNDEF
	    || src->st_shndx >= shnum
	    || !src->st_name)
		return false;

2427 2428 2429 2430 2431
#ifdef CONFIG_KALLSYMS_ALL
	if (src->st_shndx == pcpundx)
		return true;
#endif

2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442
	sec = sechdrs + src->st_shndx;
	if (!(sec->sh_flags & SHF_ALLOC)
#ifndef CONFIG_KALLSYMS_ALL
	    || !(sec->sh_flags & SHF_EXECINSTR)
#endif
	    || (sec->sh_entsize & INIT_OFFSET_MASK))
		return false;

	return true;
}

2443 2444 2445 2446 2447 2448 2449
/*
 * We only allocate and copy the strings needed by the parts of symtab
 * we keep.  This is simple, but has the effect of making multiple
 * copies of duplicates.  We could be more sophisticated, see
 * linux-kernel thread starting with
 * <73defb5e4bca04a6431392cc341112b1@localhost>.
 */
2450
static void layout_symtab(struct module *mod, struct load_info *info)
2451
{
2452 2453
	Elf_Shdr *symsect = info->sechdrs + info->index.sym;
	Elf_Shdr *strsect = info->sechdrs + info->index.str;
2454
	const Elf_Sym *src;
2455
	unsigned int i, nsrc, ndst, strtab_size = 0;
2456 2457 2458

	/* Put symbol section at end of init part of module. */
	symsect->sh_flags |= SHF_ALLOC;
2459
	symsect->sh_entsize = get_offset(mod, &mod->init_layout.size, symsect,
2460
					 info->index.sym) | INIT_OFFSET_MASK;
2461
	pr_debug("\t%s\n", info->secstrings + symsect->sh_name);
2462

2463
	src = (void *)info->hdr + symsect->sh_offset;
2464
	nsrc = symsect->sh_size / sizeof(*src);
2465

2466
	/* Compute total space required for the core symbols' strtab. */
2467 2468
	for (ndst = i = 0; i < nsrc; i++) {
		if (i == 0 ||
2469 2470
		    is_core_symbol(src+i, info->sechdrs, info->hdr->e_shnum,
				   info->index.pcpu)) {
2471
			strtab_size += strlen(&info->strtab[src[i].st_name])+1;
2472
			ndst++;
2473
		}
2474
	}
2475 2476

	/* Append room for core symbols at end of core part. */
2477 2478 2479 2480
	info->symoffs = ALIGN(mod->core_layout.size, symsect->sh_addralign ?: 1);
	info->stroffs = mod->core_layout.size = info->symoffs + ndst * sizeof(Elf_Sym);
	mod->core_layout.size += strtab_size;
	mod->core_layout.size = debug_align(mod->core_layout.size);
2481

2482 2483
	/* Put string table section at end of init part of module. */
	strsect->sh_flags |= SHF_ALLOC;
2484
	strsect->sh_entsize = get_offset(mod, &mod->init_layout.size, strsect,
2485
					 info->index.str) | INIT_OFFSET_MASK;
2486
	pr_debug("\t%s\n", info->secstrings + strsect->sh_name);
2487 2488 2489 2490 2491 2492 2493

	/* We'll tack temporary mod_kallsyms on the end. */
	mod->init_layout.size = ALIGN(mod->init_layout.size,
				      __alignof__(struct mod_kallsyms));
	info->mod_kallsyms_init_off = mod->init_layout.size;
	mod->init_layout.size += sizeof(struct mod_kallsyms);
	mod->init_layout.size = debug_align(mod->init_layout.size);
2494 2495
}

2496 2497 2498 2499 2500
/*
 * We use the full symtab and strtab which layout_symtab arranged to
 * be appended to the init section.  Later we switch to the cut-down
 * core-only ones.
 */
2501
static void add_kallsyms(struct module *mod, const struct load_info *info)
L
Linus Torvalds 已提交
2502
{
2503 2504 2505
	unsigned int i, ndst;
	const Elf_Sym *src;
	Elf_Sym *dst;
2506
	char *s;
2507
	Elf_Shdr *symsec = &info->sechdrs[info->index.sym];
L
Linus Torvalds 已提交
2508

2509 2510 2511 2512 2513
	/* Set up to point into init section. */
	mod->kallsyms = mod->init_layout.base + info->mod_kallsyms_init_off;

	mod->kallsyms->symtab = (void *)symsec->sh_addr;
	mod->kallsyms->num_symtab = symsec->sh_size / sizeof(Elf_Sym);
2514
	/* Make sure we get permanent strtab: don't use info->strtab. */
2515
	mod->kallsyms->strtab = (void *)info->sechdrs[info->index.str].sh_addr;
L
Linus Torvalds 已提交
2516 2517

	/* Set types up while we still have access to sections. */
2518 2519 2520 2521 2522 2523 2524 2525 2526
	for (i = 0; i < mod->kallsyms->num_symtab; i++)
		mod->kallsyms->symtab[i].st_info
			= elf_type(&mod->kallsyms->symtab[i], info);

	/* Now populate the cut down core kallsyms for after init. */
	mod->core_kallsyms.symtab = dst = mod->core_layout.base + info->symoffs;
	mod->core_kallsyms.strtab = s = mod->core_layout.base + info->stroffs;
	src = mod->kallsyms->symtab;
	for (ndst = i = 0; i < mod->kallsyms->num_symtab; i++) {
2527
		if (i == 0 ||
2528 2529
		    is_core_symbol(src+i, info->sechdrs, info->hdr->e_shnum,
				   info->index.pcpu)) {
2530
			dst[ndst] = src[i];
2531 2532
			dst[ndst++].st_name = s - mod->core_kallsyms.strtab;
			s += strlcpy(s, &mod->kallsyms->strtab[src[i].st_name],
2533 2534
				     KSYM_NAME_LEN) + 1;
		}
2535
	}
2536
	mod->core_kallsyms.num_symtab = ndst;
L
Linus Torvalds 已提交
2537 2538
}
#else
2539
static inline void layout_symtab(struct module *mod, struct load_info *info)
2540 2541
{
}
2542

2543
static void add_kallsyms(struct module *mod, const struct load_info *info)
L
Linus Torvalds 已提交
2544 2545 2546 2547
{
}
#endif /* CONFIG_KALLSYMS */

2548
static void dynamic_debug_setup(struct _ddebug *debug, unsigned int num)
2549
{
2550 2551
	if (!debug)
		return;
2552 2553
#ifdef CONFIG_DYNAMIC_DEBUG
	if (ddebug_add_module(debug, num, debug->modname))
A
Andrew Morton 已提交
2554 2555
		pr_err("dynamic debug error adding module: %s\n",
			debug->modname);
2556
#endif
R
Rusty Russell 已提交
2557
}
2558

2559 2560 2561 2562 2563 2564
static void dynamic_debug_remove(struct _ddebug *debug)
{
	if (debug)
		ddebug_remove_module(debug->modname);
}

2565 2566
void * __weak module_alloc(unsigned long size)
{
R
Rusty Russell 已提交
2567
	return vmalloc_exec(size);
2568 2569
}

C
Catalin Marinas 已提交
2570
#ifdef CONFIG_DEBUG_KMEMLEAK
2571 2572
static void kmemleak_load_module(const struct module *mod,
				 const struct load_info *info)
C
Catalin Marinas 已提交
2573 2574 2575 2576
{
	unsigned int i;

	/* only scan the sections containing data */
2577
	kmemleak_scan_area(mod, sizeof(struct module), GFP_KERNEL);
C
Catalin Marinas 已提交
2578

2579
	for (i = 1; i < info->hdr->e_shnum; i++) {
2580 2581 2582 2583
		/* Scan all writable sections that's not executable */
		if (!(info->sechdrs[i].sh_flags & SHF_ALLOC) ||
		    !(info->sechdrs[i].sh_flags & SHF_WRITE) ||
		    (info->sechdrs[i].sh_flags & SHF_EXECINSTR))
C
Catalin Marinas 已提交
2584 2585
			continue;

2586 2587
		kmemleak_scan_area((void *)info->sechdrs[i].sh_addr,
				   info->sechdrs[i].sh_size, GFP_KERNEL);
C
Catalin Marinas 已提交
2588 2589 2590
	}
}
#else
2591 2592
static inline void kmemleak_load_module(const struct module *mod,
					const struct load_info *info)
C
Catalin Marinas 已提交
2593 2594 2595 2596
{
}
#endif

R
Rusty Russell 已提交
2597
#ifdef CONFIG_MODULE_SIG
2598
static int module_sig_check(struct load_info *info)
R
Rusty Russell 已提交
2599 2600
{
	int err = -ENOKEY;
2601 2602
	const unsigned long markerlen = sizeof(MODULE_SIG_STRING) - 1;
	const void *mod = info->hdr;
2603

2604 2605
	if (info->len > markerlen &&
	    memcmp(mod + info->len - markerlen, MODULE_SIG_STRING, markerlen) == 0) {
2606
		/* We truncate the module to discard the signature */
2607 2608
		info->len -= markerlen;
		err = mod_verify_sig(mod, &info->len);
R
Rusty Russell 已提交
2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622
	}

	if (!err) {
		info->sig_ok = true;
		return 0;
	}

	/* Not having a signature is only an error if we're strict. */
	if (err == -ENOKEY && !sig_enforce)
		err = 0;

	return err;
}
#else /* !CONFIG_MODULE_SIG */
2623
static int module_sig_check(struct load_info *info)
R
Rusty Russell 已提交
2624 2625 2626 2627 2628
{
	return 0;
}
#endif /* !CONFIG_MODULE_SIG */

2629 2630
/* Sanity checks against invalid binaries, wrong arch, weird elf version. */
static int elf_header_check(struct load_info *info)
2631
{
2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644
	if (info->len < sizeof(*(info->hdr)))
		return -ENOEXEC;

	if (memcmp(info->hdr->e_ident, ELFMAG, SELFMAG) != 0
	    || info->hdr->e_type != ET_REL
	    || !elf_check_arch(info->hdr)
	    || info->hdr->e_shentsize != sizeof(Elf_Shdr))
		return -ENOEXEC;

	if (info->hdr->e_shoff >= info->len
	    || (info->hdr->e_shnum * sizeof(Elf_Shdr) >
		info->len - info->hdr->e_shoff))
		return -ENOEXEC;
2645

2646 2647 2648
	return 0;
}

2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665
#define COPY_CHUNK_SIZE (16*PAGE_SIZE)

static int copy_chunked_from_user(void *dst, const void __user *usrc, unsigned long len)
{
	do {
		unsigned long n = min(len, COPY_CHUNK_SIZE);

		if (copy_from_user(dst, usrc, n) != 0)
			return -EFAULT;
		cond_resched();
		dst += n;
		usrc += n;
		len -= n;
	} while (len);
	return 0;
}

2666 2667 2668
/* Sets info->hdr and info->len. */
static int copy_module_from_user(const void __user *umod, unsigned long len,
				  struct load_info *info)
2669 2670 2671
{
	int err;

2672 2673
	info->len = len;
	if (info->len < sizeof(*(info->hdr)))
2674 2675
		return -ENOEXEC;

2676 2677 2678 2679
	err = security_kernel_module_from_file(NULL);
	if (err)
		return err;

2680
	/* Suck in entire file: we'll want most of it. */
2681 2682
	info->hdr = __vmalloc(info->len,
			GFP_KERNEL | __GFP_HIGHMEM | __GFP_NOWARN, PAGE_KERNEL);
2683
	if (!info->hdr)
2684 2685
		return -ENOMEM;

2686
	if (copy_chunked_from_user(info->hdr, umod, info->len) != 0) {
2687 2688
		vfree(info->hdr);
		return -EFAULT;
2689 2690
	}

2691 2692 2693 2694 2695 2696
	return 0;
}

/* Sets info->hdr and info->len. */
static int copy_module_from_fd(int fd, struct load_info *info)
{
A
Al Viro 已提交
2697
	struct fd f = fdget(fd);
2698 2699 2700 2701 2702
	int err;
	struct kstat stat;
	loff_t pos;
	ssize_t bytes = 0;

A
Al Viro 已提交
2703
	if (!f.file)
2704 2705
		return -ENOEXEC;

A
Al Viro 已提交
2706
	err = security_kernel_module_from_file(f.file);
R
Rusty Russell 已提交
2707
	if (err)
2708
		goto out;
R
Rusty Russell 已提交
2709

A
Al Viro 已提交
2710
	err = vfs_getattr(&f.file->f_path, &stat);
R
Rusty Russell 已提交
2711
	if (err)
2712
		goto out;
2713

2714 2715 2716
	if (stat.size > INT_MAX) {
		err = -EFBIG;
		goto out;
2717
	}
2718 2719 2720 2721 2722 2723 2724

	/* Don't hand 0 to vmalloc, it whines. */
	if (stat.size == 0) {
		err = -EINVAL;
		goto out;
	}

2725 2726 2727 2728
	info->hdr = vmalloc(stat.size);
	if (!info->hdr) {
		err = -ENOMEM;
		goto out;
2729
	}
R
Rusty Russell 已提交
2730

2731 2732
	pos = 0;
	while (pos < stat.size) {
A
Al Viro 已提交
2733
		bytes = kernel_read(f.file, pos, (char *)(info->hdr) + pos,
2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744
				    stat.size - pos);
		if (bytes < 0) {
			vfree(info->hdr);
			err = bytes;
			goto out;
		}
		if (bytes == 0)
			break;
		pos += bytes;
	}
	info->len = pos;
2745

2746
out:
A
Al Viro 已提交
2747
	fdput(f);
2748 2749 2750
	return err;
}

R
Rusty Russell 已提交
2751 2752 2753 2754 2755
static void free_copy(struct load_info *info)
{
	vfree(info->hdr);
}

2756
static int rewrite_section_headers(struct load_info *info, int flags)
2757 2758 2759 2760 2761 2762 2763 2764 2765 2766
{
	unsigned int i;

	/* This should always be true, but let's be sure. */
	info->sechdrs[0].sh_addr = 0;

	for (i = 1; i < info->hdr->e_shnum; i++) {
		Elf_Shdr *shdr = &info->sechdrs[i];
		if (shdr->sh_type != SHT_NOBITS
		    && info->len < shdr->sh_offset + shdr->sh_size) {
A
Andrew Morton 已提交
2767
			pr_err("Module len %lu truncated\n", info->len);
2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780
			return -ENOEXEC;
		}

		/* Mark all sections sh_addr with their address in the
		   temporary image. */
		shdr->sh_addr = (size_t)info->hdr + shdr->sh_offset;

#ifndef CONFIG_MODULE_UNLOAD
		/* Don't load .exit sections */
		if (strstarts(info->secstrings+shdr->sh_name, ".exit"))
			shdr->sh_flags &= ~(unsigned long)SHF_ALLOC;
#endif
	}
2781 2782

	/* Track but don't keep modinfo and version sections. */
2783 2784 2785 2786
	if (flags & MODULE_INIT_IGNORE_MODVERSIONS)
		info->index.vers = 0; /* Pretend no __versions section! */
	else
		info->index.vers = find_sec(info, "__versions");
2787
	info->index.info = find_sec(info, ".modinfo");
2788 2789
	info->sechdrs[info->index.info].sh_flags &= ~(unsigned long)SHF_ALLOC;
	info->sechdrs[info->index.vers].sh_flags &= ~(unsigned long)SHF_ALLOC;
2790 2791 2792
	return 0;
}

L
Linus Torvalds 已提交
2793 2794 2795 2796 2797 2798 2799 2800
/*
 * Set up our basic convenience variables (pointers to section headers,
 * search for module section index etc), and do some basic section
 * verification.
 *
 * Return the temporary module pointer (we'll replace it with the final
 * one when we move the module sections around).
 */
2801
static struct module *setup_load_info(struct load_info *info, int flags)
L
Linus Torvalds 已提交
2802 2803
{
	unsigned int i;
2804
	int err;
L
Linus Torvalds 已提交
2805 2806 2807 2808
	struct module *mod;

	/* Set up the convenience variables */
	info->sechdrs = (void *)info->hdr + info->hdr->e_shoff;
2809 2810
	info->secstrings = (void *)info->hdr
		+ info->sechdrs[info->hdr->e_shstrndx].sh_offset;
L
Linus Torvalds 已提交
2811

2812
	err = rewrite_section_headers(info, flags);
2813 2814
	if (err)
		return ERR_PTR(err);
L
Linus Torvalds 已提交
2815

2816 2817
	/* Find internal symbols and strings. */
	for (i = 1; i < info->hdr->e_shnum; i++) {
L
Linus Torvalds 已提交
2818 2819 2820
		if (info->sechdrs[i].sh_type == SHT_SYMTAB) {
			info->index.sym = i;
			info->index.str = info->sechdrs[i].sh_link;
2821 2822 2823
			info->strtab = (char *)info->hdr
				+ info->sechdrs[info->index.str].sh_offset;
			break;
L
Linus Torvalds 已提交
2824 2825 2826
		}
	}

2827
	info->index.mod = find_sec(info, ".gnu.linkonce.this_module");
L
Linus Torvalds 已提交
2828
	if (!info->index.mod) {
A
Andrew Morton 已提交
2829
		pr_warn("No module found in object\n");
L
Linus Torvalds 已提交
2830 2831 2832 2833 2834 2835
		return ERR_PTR(-ENOEXEC);
	}
	/* This is temporary: point mod into copy of data. */
	mod = (void *)info->sechdrs[info->index.mod].sh_addr;

	if (info->index.sym == 0) {
A
Andrew Morton 已提交
2836
		pr_warn("%s: module has no symbols (stripped?)\n", mod->name);
L
Linus Torvalds 已提交
2837 2838 2839
		return ERR_PTR(-ENOEXEC);
	}

2840
	info->index.pcpu = find_pcpusec(info);
L
Linus Torvalds 已提交
2841 2842 2843 2844 2845 2846 2847 2848

	/* Check module struct version now, before we try to use module. */
	if (!check_modstruct_version(info->sechdrs, info->index.vers, mod))
		return ERR_PTR(-ENOEXEC);

	return mod;
}

2849
static int check_modinfo(struct module *mod, struct load_info *info, int flags)
2850
{
2851
	const char *modmagic = get_modinfo(info, "vermagic");
2852 2853
	int err;

2854 2855 2856
	if (flags & MODULE_INIT_IGNORE_VERMAGIC)
		modmagic = NULL;

2857 2858 2859 2860 2861
	/* This is allowed: modprobe --force will invalidate it. */
	if (!modmagic) {
		err = try_to_force_load(mod, "bad vermagic");
		if (err)
			return err;
2862
	} else if (!same_magic(modmagic, vermagic, info->index.vers)) {
A
Andrew Morton 已提交
2863
		pr_err("%s: version magic '%s' should be '%s'\n",
2864 2865 2866 2867
		       mod->name, modmagic, vermagic);
		return -ENOEXEC;
	}

2868
	if (!get_modinfo(info, "intree"))
2869
		add_taint_module(mod, TAINT_OOT_MODULE, LOCKDEP_STILL_OK);
2870

2871
	if (get_modinfo(info, "staging")) {
2872
		add_taint_module(mod, TAINT_CRAP, LOCKDEP_STILL_OK);
A
Andrew Morton 已提交
2873 2874
		pr_warn("%s: module is from the staging directory, the quality "
			"is unknown, you have been warned.\n", mod->name);
2875
	}
2876 2877

	/* Set up license info based on the info section */
2878
	set_license(mod, get_modinfo(info, "license"));
2879

2880 2881 2882
	return 0;
}

2883
static int find_module_sections(struct module *mod, struct load_info *info)
L
Linus Torvalds 已提交
2884
{
2885
	mod->kp = section_objs(info, "__param",
L
Linus Torvalds 已提交
2886
			       sizeof(*mod->kp), &mod->num_kp);
2887
	mod->syms = section_objs(info, "__ksymtab",
L
Linus Torvalds 已提交
2888
				 sizeof(*mod->syms), &mod->num_syms);
2889 2890
	mod->crcs = section_addr(info, "__kcrctab");
	mod->gpl_syms = section_objs(info, "__ksymtab_gpl",
L
Linus Torvalds 已提交
2891 2892
				     sizeof(*mod->gpl_syms),
				     &mod->num_gpl_syms);
2893 2894
	mod->gpl_crcs = section_addr(info, "__kcrctab_gpl");
	mod->gpl_future_syms = section_objs(info,
L
Linus Torvalds 已提交
2895 2896 2897
					    "__ksymtab_gpl_future",
					    sizeof(*mod->gpl_future_syms),
					    &mod->num_gpl_future_syms);
2898
	mod->gpl_future_crcs = section_addr(info, "__kcrctab_gpl_future");
L
Linus Torvalds 已提交
2899 2900

#ifdef CONFIG_UNUSED_SYMBOLS
2901
	mod->unused_syms = section_objs(info, "__ksymtab_unused",
L
Linus Torvalds 已提交
2902 2903
					sizeof(*mod->unused_syms),
					&mod->num_unused_syms);
2904 2905
	mod->unused_crcs = section_addr(info, "__kcrctab_unused");
	mod->unused_gpl_syms = section_objs(info, "__ksymtab_unused_gpl",
L
Linus Torvalds 已提交
2906 2907
					    sizeof(*mod->unused_gpl_syms),
					    &mod->num_unused_gpl_syms);
2908
	mod->unused_gpl_crcs = section_addr(info, "__kcrctab_unused_gpl");
L
Linus Torvalds 已提交
2909 2910
#endif
#ifdef CONFIG_CONSTRUCTORS
2911
	mod->ctors = section_objs(info, ".ctors",
L
Linus Torvalds 已提交
2912
				  sizeof(*mod->ctors), &mod->num_ctors);
2913 2914 2915 2916 2917 2918 2919 2920
	if (!mod->ctors)
		mod->ctors = section_objs(info, ".init_array",
				sizeof(*mod->ctors), &mod->num_ctors);
	else if (find_sec(info, ".init_array")) {
		/*
		 * This shouldn't happen with same compiler and binutils
		 * building all parts of the module.
		 */
2921
		pr_warn("%s: has both .ctors and .init_array.\n",
2922 2923 2924
		       mod->name);
		return -EINVAL;
	}
L
Linus Torvalds 已提交
2925 2926 2927
#endif

#ifdef CONFIG_TRACEPOINTS
2928 2929 2930
	mod->tracepoints_ptrs = section_objs(info, "__tracepoints_ptrs",
					     sizeof(*mod->tracepoints_ptrs),
					     &mod->num_tracepoints);
L
Linus Torvalds 已提交
2931
#endif
2932 2933 2934 2935 2936
#ifdef HAVE_JUMP_LABEL
	mod->jump_entries = section_objs(info, "__jump_table",
					sizeof(*mod->jump_entries),
					&mod->num_jump_entries);
#endif
L
Linus Torvalds 已提交
2937
#ifdef CONFIG_EVENT_TRACING
2938
	mod->trace_events = section_objs(info, "_ftrace_events",
L
Linus Torvalds 已提交
2939 2940
					 sizeof(*mod->trace_events),
					 &mod->num_trace_events);
2941 2942 2943
	mod->trace_enums = section_objs(info, "_ftrace_enum_map",
					sizeof(*mod->trace_enums),
					&mod->num_trace_enums);
L
Linus Torvalds 已提交
2944
#endif
2945 2946 2947 2948 2949
#ifdef CONFIG_TRACING
	mod->trace_bprintk_fmt_start = section_objs(info, "__trace_printk_fmt",
					 sizeof(*mod->trace_bprintk_fmt_start),
					 &mod->num_trace_bprintk_fmt);
#endif
L
Linus Torvalds 已提交
2950 2951
#ifdef CONFIG_FTRACE_MCOUNT_RECORD
	/* sechdrs[0].sh_size is always zero */
2952
	mod->ftrace_callsites = section_objs(info, "__mcount_loc",
L
Linus Torvalds 已提交
2953 2954 2955
					     sizeof(*mod->ftrace_callsites),
					     &mod->num_ftrace_callsites);
#endif
2956

2957 2958 2959
	mod->extable = section_objs(info, "__ex_table",
				    sizeof(*mod->extable), &mod->num_exentries);

2960
	if (section_addr(info, "__obsparm"))
A
Andrew Morton 已提交
2961
		pr_warn("%s: Ignoring obsolete parameters\n", mod->name);
2962 2963 2964

	info->debug = section_objs(info, "__verbose",
				   sizeof(*info->debug), &info->num_debug);
2965 2966

	return 0;
L
Linus Torvalds 已提交
2967 2968
}

2969
static int move_module(struct module *mod, struct load_info *info)
2970 2971 2972 2973 2974
{
	int i;
	void *ptr;

	/* Do the allocs. */
2975
	ptr = module_alloc(mod->core_layout.size);
2976 2977 2978 2979 2980 2981 2982
	/*
	 * The pointer to this block is stored in the module structure
	 * which is inside the block. Just mark it as not being a
	 * leak.
	 */
	kmemleak_not_leak(ptr);
	if (!ptr)
R
Rusty Russell 已提交
2983
		return -ENOMEM;
2984

2985 2986
	memset(ptr, 0, mod->core_layout.size);
	mod->core_layout.base = ptr;
2987

2988 2989
	if (mod->init_layout.size) {
		ptr = module_alloc(mod->init_layout.size);
R
Rusty Russell 已提交
2990 2991 2992 2993 2994 2995 2996 2997
		/*
		 * The pointer to this block is stored in the module structure
		 * which is inside the block. This block doesn't need to be
		 * scanned as it contains data and code that will be freed
		 * after the module is initialized.
		 */
		kmemleak_ignore(ptr);
		if (!ptr) {
2998
			module_memfree(mod->core_layout.base);
R
Rusty Russell 已提交
2999 3000
			return -ENOMEM;
		}
3001 3002
		memset(ptr, 0, mod->init_layout.size);
		mod->init_layout.base = ptr;
R
Rusty Russell 已提交
3003
	} else
3004
		mod->init_layout.base = NULL;
3005 3006

	/* Transfer each section which specifies SHF_ALLOC */
3007
	pr_debug("final section addresses:\n");
3008
	for (i = 0; i < info->hdr->e_shnum; i++) {
3009
		void *dest;
3010
		Elf_Shdr *shdr = &info->sechdrs[i];
3011

3012
		if (!(shdr->sh_flags & SHF_ALLOC))
3013 3014
			continue;

3015
		if (shdr->sh_entsize & INIT_OFFSET_MASK)
3016
			dest = mod->init_layout.base
3017
				+ (shdr->sh_entsize & ~INIT_OFFSET_MASK);
3018
		else
3019
			dest = mod->core_layout.base + shdr->sh_entsize;
3020

3021 3022
		if (shdr->sh_type != SHT_NOBITS)
			memcpy(dest, (void *)shdr->sh_addr, shdr->sh_size);
3023
		/* Update sh_addr to point to copy in image. */
3024
		shdr->sh_addr = (unsigned long)dest;
3025 3026
		pr_debug("\t0x%lx %s\n",
			 (long)shdr->sh_addr, info->secstrings + shdr->sh_name);
3027
	}
R
Rusty Russell 已提交
3028 3029

	return 0;
3030 3031
}

3032
static int check_module_license_and_versions(struct module *mod)
3033 3034 3035 3036 3037 3038 3039
{
	/*
	 * ndiswrapper is under GPL by itself, but loads proprietary modules.
	 * Don't use add_taint_module(), as it would prevent ndiswrapper from
	 * using GPL-only symbols it needs.
	 */
	if (strcmp(mod->name, "ndiswrapper") == 0)
3040
		add_taint(TAINT_PROPRIETARY_MODULE, LOCKDEP_NOW_UNRELIABLE);
3041 3042 3043

	/* driverloader was caught wrongly pretending to be under GPL */
	if (strcmp(mod->name, "driverloader") == 0)
3044 3045
		add_taint_module(mod, TAINT_PROPRIETARY_MODULE,
				 LOCKDEP_NOW_UNRELIABLE);
3046

3047 3048
	/* lve claims to be GPL but upstream won't provide source */
	if (strcmp(mod->name, "lve") == 0)
3049 3050
		add_taint_module(mod, TAINT_PROPRIETARY_MODULE,
				 LOCKDEP_NOW_UNRELIABLE);
3051

3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065 3066 3067 3068 3069 3070 3071 3072 3073 3074 3075 3076 3077 3078 3079 3080
#ifdef CONFIG_MODVERSIONS
	if ((mod->num_syms && !mod->crcs)
	    || (mod->num_gpl_syms && !mod->gpl_crcs)
	    || (mod->num_gpl_future_syms && !mod->gpl_future_crcs)
#ifdef CONFIG_UNUSED_SYMBOLS
	    || (mod->num_unused_syms && !mod->unused_crcs)
	    || (mod->num_unused_gpl_syms && !mod->unused_gpl_crcs)
#endif
		) {
		return try_to_force_load(mod,
					 "no versions for exported symbols");
	}
#endif
	return 0;
}

static void flush_module_icache(const struct module *mod)
{
	mm_segment_t old_fs;

	/* flush the icache in correct context */
	old_fs = get_fs();
	set_fs(KERNEL_DS);

	/*
	 * Flush the instruction cache, since we've played with text.
	 * Do it before processing of module parameters, so the module
	 * can provide parameter accessor functions of its own.
	 */
3081 3082 3083 3084 3085 3086
	if (mod->init_layout.base)
		flush_icache_range((unsigned long)mod->init_layout.base,
				   (unsigned long)mod->init_layout.base
				   + mod->init_layout.size);
	flush_icache_range((unsigned long)mod->core_layout.base,
			   (unsigned long)mod->core_layout.base + mod->core_layout.size);
3087 3088 3089 3090

	set_fs(old_fs);
}

3091 3092 3093 3094 3095 3096 3097 3098
int __weak module_frob_arch_sections(Elf_Ehdr *hdr,
				     Elf_Shdr *sechdrs,
				     char *secstrings,
				     struct module *mod)
{
	return 0;
}

3099
static struct module *layout_and_allocate(struct load_info *info, int flags)
L
Linus Torvalds 已提交
3100
{
R
Rusty Russell 已提交
3101
	/* Module within temporary copy. */
L
Linus Torvalds 已提交
3102
	struct module *mod;
R
Rusty Russell 已提交
3103
	int err;
3104

3105
	mod = setup_load_info(info, flags);
R
Rusty Russell 已提交
3106 3107
	if (IS_ERR(mod))
		return mod;
L
Linus Torvalds 已提交
3108

3109
	err = check_modinfo(mod, info, flags);
3110 3111
	if (err)
		return ERR_PTR(err);
L
Linus Torvalds 已提交
3112 3113

	/* Allow arches to frob section contents and sizes.  */
3114 3115
	err = module_frob_arch_sections(info->hdr, info->sechdrs,
					info->secstrings, mod);
L
Linus Torvalds 已提交
3116
	if (err < 0)
3117
		return ERR_PTR(err);
L
Linus Torvalds 已提交
3118

3119 3120
	/* We will do a special allocation for per-cpu sections later. */
	info->sechdrs[info->index.pcpu].sh_flags &= ~(unsigned long)SHF_ALLOC;
L
Linus Torvalds 已提交
3121 3122 3123 3124

	/* Determine total sizes, and put offsets in sh_entsize.  For now
	   this is done generically; there doesn't appear to be any
	   special cases for the architectures. */
3125 3126
	layout_sections(mod, info);
	layout_symtab(mod, info);
L
Linus Torvalds 已提交
3127

3128
	/* Allocate and move to the final place */
3129
	err = move_module(mod, info);
R
Rusty Russell 已提交
3130
	if (err)
3131
		return ERR_PTR(err);
R
Rusty Russell 已提交
3132 3133 3134

	/* Module has been copied to its final place now: return it. */
	mod = (void *)info->sechdrs[info->index.mod].sh_addr;
3135
	kmemleak_load_module(mod, info);
R
Rusty Russell 已提交
3136 3137 3138 3139 3140 3141 3142
	return mod;
}

/* mod is no longer valid after this! */
static void module_deallocate(struct module *mod, struct load_info *info)
{
	percpu_modfree(mod);
3143
	module_arch_freeing_init(mod);
3144 3145
	module_memfree(mod->init_layout.base);
	module_memfree(mod->core_layout.base);
R
Rusty Russell 已提交
3146 3147
}

3148 3149 3150 3151 3152 3153 3154
int __weak module_finalize(const Elf_Ehdr *hdr,
			   const Elf_Shdr *sechdrs,
			   struct module *me)
{
	return 0;
}

3155 3156
static int post_relocation(struct module *mod, const struct load_info *info)
{
3157
	/* Sort exception table now relocations are done. */
3158 3159 3160 3161 3162 3163
	sort_extable(mod->extable, mod->extable + mod->num_exentries);

	/* Copy relocated percpu area over. */
	percpu_modcopy(mod, (void *)info->sechdrs[info->index.pcpu].sh_addr,
		       info->sechdrs[info->index.pcpu].sh_size);

3164
	/* Setup kallsyms-specific fields. */
3165 3166 3167 3168 3169 3170
	add_kallsyms(mod, info);

	/* Arch-specific module finalizing. */
	return module_finalize(info->hdr, info->sechdrs, mod);
}

3171 3172 3173 3174 3175 3176
/* Is this module of this name done loading?  No locks held. */
static bool finished_loading(const char *name)
{
	struct module *mod;
	bool ret;

3177 3178 3179 3180 3181 3182
	/*
	 * The module_mutex should not be a heavily contended lock;
	 * if we get the occasional sleep here, we'll go an extra iteration
	 * in the wait_event_interruptible(), which is harmless.
	 */
	sched_annotate_sleep();
3183
	mutex_lock(&module_mutex);
3184
	mod = find_module_all(name, strlen(name), true);
3185 3186
	ret = !mod || mod->state == MODULE_STATE_LIVE
		|| mod->state == MODULE_STATE_GOING;
3187 3188 3189 3190 3191
	mutex_unlock(&module_mutex);

	return ret;
}

3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202
/* Call module constructors. */
static void do_mod_ctors(struct module *mod)
{
#ifdef CONFIG_CONSTRUCTORS
	unsigned long i;

	for (i = 0; i < mod->num_ctors; i++)
		mod->ctors[i]();
#endif
}

3203 3204 3205 3206 3207 3208 3209 3210 3211 3212 3213 3214 3215
/* For freeing module_init on success, in case kallsyms traversing */
struct mod_initfree {
	struct rcu_head rcu;
	void *module_init;
};

static void do_free_init(struct rcu_head *head)
{
	struct mod_initfree *m = container_of(head, struct mod_initfree, rcu);
	module_memfree(m->module_init);
	kfree(m);
}

3216 3217 3218 3219 3220 3221 3222
/*
 * This is where the real work happens.
 *
 * Keep it uninlined to provide a reliable breakpoint target, e.g. for the gdb
 * helper command 'lx-symbols'.
 */
static noinline int do_init_module(struct module *mod)
3223 3224
{
	int ret = 0;
3225 3226 3227 3228 3229 3230 3231
	struct mod_initfree *freeinit;

	freeinit = kmalloc(sizeof(*freeinit), GFP_KERNEL);
	if (!freeinit) {
		ret = -ENOMEM;
		goto fail;
	}
3232
	freeinit->module_init = mod->init_layout.base;
3233

3234 3235 3236 3237 3238 3239
	/*
	 * We want to find out whether @mod uses async during init.  Clear
	 * PF_USED_ASYNC.  async_schedule*() will set it.
	 */
	current->flags &= ~PF_USED_ASYNC;

3240 3241 3242 3243 3244
	do_mod_ctors(mod);
	/* Start the module */
	if (mod->init != NULL)
		ret = do_one_initcall(mod->init);
	if (ret < 0) {
3245
		goto fail_free_freeinit;
3246 3247
	}
	if (ret > 0) {
A
Andrew Morton 已提交
3248 3249 3250 3251
		pr_warn("%s: '%s'->init suspiciously returned %d, it should "
			"follow 0/-E convention\n"
			"%s: loading module anyway...\n",
			__func__, mod->name, ret, __func__);
3252 3253 3254 3255 3256 3257 3258 3259
		dump_stack();
	}

	/* Now it's a first class citizen! */
	mod->state = MODULE_STATE_LIVE;
	blocking_notifier_call_chain(&module_notify_list,
				     MODULE_STATE_LIVE, mod);

3260 3261 3262 3263 3264 3265 3266 3267 3268 3269 3270 3271 3272 3273 3274 3275 3276
	/*
	 * We need to finish all async code before the module init sequence
	 * is done.  This has potential to deadlock.  For example, a newly
	 * detected block device can trigger request_module() of the
	 * default iosched from async probing task.  Once userland helper
	 * reaches here, async_synchronize_full() will wait on the async
	 * task waiting on request_module() and deadlock.
	 *
	 * This deadlock is avoided by perfomring async_synchronize_full()
	 * iff module init queued any async jobs.  This isn't a full
	 * solution as it will deadlock the same if module loading from
	 * async jobs nests more than once; however, due to the various
	 * constraints, this hack seems to be the best option for now.
	 * Please refer to the following thread for details.
	 *
	 * http://thread.gmane.org/gmane.linux.kernel/1420814
	 */
3277
	if (!mod->async_probe_requested && (current->flags & PF_USED_ASYNC))
3278
		async_synchronize_full();
3279 3280 3281 3282 3283 3284

	mutex_lock(&module_mutex);
	/* Drop initial reference. */
	module_put(mod);
	trim_init_extable(mod);
#ifdef CONFIG_KALLSYMS
3285 3286
	/* Switch to core kallsyms now init is done: kallsyms may be walking! */
	rcu_assign_pointer(mod->kallsyms, &mod->core_kallsyms);
3287
#endif
3288
	mod_tree_remove_init(mod);
R
Rusty Russell 已提交
3289
	disable_ro_nx(&mod->init_layout);
3290
	module_arch_freeing_init(mod);
3291 3292 3293 3294
	mod->init_layout.base = NULL;
	mod->init_layout.size = 0;
	mod->init_layout.ro_size = 0;
	mod->init_layout.text_size = 0;
3295 3296
	/*
	 * We want to free module_init, but be aware that kallsyms may be
3297 3298 3299
	 * walking this with preempt disabled.  In all the failure paths, we
	 * call synchronize_sched(), but we don't want to slow down the success
	 * path, so use actual RCU here.
3300
	 */
3301
	call_rcu_sched(&freeinit->rcu, do_free_init);
3302 3303 3304 3305
	mutex_unlock(&module_mutex);
	wake_up_all(&module_wq);

	return 0;
3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318

fail_free_freeinit:
	kfree(freeinit);
fail:
	/* Try to protect us from buggy refcounters. */
	mod->state = MODULE_STATE_GOING;
	synchronize_sched();
	module_put(mod);
	blocking_notifier_call_chain(&module_notify_list,
				     MODULE_STATE_GOING, mod);
	free_module(mod);
	wake_up_all(&module_wq);
	return ret;
3319 3320 3321 3322 3323 3324 3325 3326 3327 3328
}

static int may_init_module(void)
{
	if (!capable(CAP_SYS_MODULE) || modules_disabled)
		return -EPERM;

	return 0;
}

3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342
/*
 * We try to place it in the list now to make sure it's unique before
 * we dedicate too many resources.  In particular, temporary percpu
 * memory exhaustion.
 */
static int add_unformed_module(struct module *mod)
{
	int err;
	struct module *old;

	mod->state = MODULE_STATE_UNFORMED;

again:
	mutex_lock(&module_mutex);
3343 3344
	old = find_module_all(mod->name, strlen(mod->name), true);
	if (old != NULL) {
3345 3346 3347 3348
		if (old->state == MODULE_STATE_COMING
		    || old->state == MODULE_STATE_UNFORMED) {
			/* Wait in case it fails to load. */
			mutex_unlock(&module_mutex);
3349 3350
			err = wait_event_interruptible(module_wq,
					       finished_loading(mod->name));
3351 3352 3353 3354 3355 3356 3357
			if (err)
				goto out_unlocked;
			goto again;
		}
		err = -EEXIST;
		goto out;
	}
3358
	mod_update_bounds(mod);
3359
	list_add_rcu(&mod->list, &modules);
3360
	mod_tree_insert(mod);
3361 3362 3363 3364 3365 3366 3367 3368 3369 3370 3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382
	err = 0;

out:
	mutex_unlock(&module_mutex);
out_unlocked:
	return err;
}

static int complete_formation(struct module *mod, struct load_info *info)
{
	int err;

	mutex_lock(&module_mutex);

	/* Find duplicate symbols (must be called under lock). */
	err = verify_export_symbols(mod);
	if (err < 0)
		goto out;

	/* This relies on module_mutex for list integrity. */
	module_bug_finalize(info->hdr, info->sechdrs, mod);

3383
	/* Set RO and NX regions */
R
Rusty Russell 已提交
3384 3385
	module_enable_ro(mod);
	module_enable_nx(mod);
3386

3387 3388 3389
	/* Mark state as coming so strong_try_module_get() ignores us,
	 * but kallsyms etc. can see us. */
	mod->state = MODULE_STATE_COMING;
3390 3391 3392 3393 3394
	mutex_unlock(&module_mutex);

	blocking_notifier_call_chain(&module_notify_list,
				     MODULE_STATE_COMING, mod);
	return 0;
3395 3396 3397 3398 3399 3400

out:
	mutex_unlock(&module_mutex);
	return err;
}

3401 3402
static int unknown_module_param_cb(char *param, char *val, const char *modname,
				   void *arg)
3403
{
3404 3405 3406 3407 3408 3409 3410 3411
	struct module *mod = arg;
	int ret;

	if (strcmp(param, "async_probe") == 0) {
		mod->async_probe_requested = true;
		return 0;
	}

3412
	/* Check for magic 'dyndbg' arg */
3413
	ret = ddebug_dyndbg_module_param_cb(param, val, modname);
A
Andrew Morton 已提交
3414 3415
	if (ret != 0)
		pr_warn("%s: unknown parameter '%s' ignored\n", modname, param);
3416 3417 3418
	return 0;
}

R
Rusty Russell 已提交
3419 3420
/* Allocate and load the module: note that size of section 0 is always
   zero, and we rely on this for optional sections. */
3421 3422
static int load_module(struct load_info *info, const char __user *uargs,
		       int flags)
R
Rusty Russell 已提交
3423
{
3424
	struct module *mod;
R
Rusty Russell 已提交
3425
	long err;
3426
	char *after_dashes;
R
Rusty Russell 已提交
3427

3428 3429 3430
	err = module_sig_check(info);
	if (err)
		goto free_copy;
R
Rusty Russell 已提交
3431

3432
	err = elf_header_check(info);
R
Rusty Russell 已提交
3433
	if (err)
3434
		goto free_copy;
R
Rusty Russell 已提交
3435 3436

	/* Figure out module layout, and allocate all the memory. */
3437
	mod = layout_and_allocate(info, flags);
3438 3439
	if (IS_ERR(mod)) {
		err = PTR_ERR(mod);
R
Rusty Russell 已提交
3440
		goto free_copy;
L
Linus Torvalds 已提交
3441 3442
	}

3443 3444 3445
	/* Reserve our place in the list. */
	err = add_unformed_module(mod);
	if (err)
3446 3447
		goto free_module;

R
Rusty Russell 已提交
3448
#ifdef CONFIG_MODULE_SIG
3449
	mod->sig_ok = info->sig_ok;
3450
	if (!mod->sig_ok) {
A
Andrew Morton 已提交
3451
		pr_notice_once("%s: module verification failed: signature "
3452
			       "and/or required key missing - tainting "
A
Andrew Morton 已提交
3453
			       "kernel\n", mod->name);
3454
		add_taint_module(mod, TAINT_UNSIGNED_MODULE, LOCKDEP_STILL_OK);
3455
	}
R
Rusty Russell 已提交
3456 3457
#endif

3458
	/* To avoid stressing percpu allocator, do this once we're unique. */
R
Rusty Russell 已提交
3459
	err = percpu_modalloc(mod, info);
3460 3461 3462
	if (err)
		goto unlink_mod;

3463
	/* Now module is in final location, initialize linked lists, etc. */
3464 3465
	err = module_unload_init(mod);
	if (err)
3466
		goto unlink_mod;
L
Linus Torvalds 已提交
3467

3468
	init_param_lock(mod);
3469

3470 3471
	/* Now we've got everything in the final locations, we can
	 * find optional sections. */
3472 3473 3474
	err = find_module_sections(mod, info);
	if (err)
		goto free_unload;
3475

3476
	err = check_module_license_and_versions(mod);
3477 3478
	if (err)
		goto free_unload;
3479

3480
	/* Set up MODINFO_ATTR fields */
3481
	setup_modinfo(mod, info);
3482

L
Linus Torvalds 已提交
3483
	/* Fix up syms, so that st_value is a pointer to location. */
3484
	err = simplify_symbols(mod, info);
L
Linus Torvalds 已提交
3485
	if (err < 0)
R
Rusty Russell 已提交
3486
		goto free_modinfo;
L
Linus Torvalds 已提交
3487

3488
	err = apply_relocations(mod, info);
3489
	if (err < 0)
R
Rusty Russell 已提交
3490
		goto free_modinfo;
L
Linus Torvalds 已提交
3491

3492
	err = post_relocation(mod, info);
L
Linus Torvalds 已提交
3493
	if (err < 0)
R
Rusty Russell 已提交
3494
		goto free_modinfo;
L
Linus Torvalds 已提交
3495

3496
	flush_module_icache(mod);
3497

3498 3499 3500 3501 3502 3503
	/* Now copy in args */
	mod->args = strndup_user(uargs, ~0UL >> 1);
	if (IS_ERR(mod->args)) {
		err = PTR_ERR(mod->args);
		goto free_arch_cleanup;
	}
R
Rusty Russell 已提交
3504

3505
	dynamic_debug_setup(info->debug, info->num_debug);
3506

3507 3508 3509
	/* Ftrace init must be called in the MODULE_STATE_UNFORMED state */
	ftrace_module_init(mod);

3510 3511 3512
	/* Finally it's fully formed, ready to start executing. */
	err = complete_formation(mod, info);
	if (err)
3513
		goto ddebug_cleanup;
3514

3515
	/* Module is ready to execute: parsing args may do that. */
3516
	after_dashes = parse_args(mod->name, mod->args, mod->kp, mod->num_kp,
3517
				  -32768, 32767, mod,
3518
				  unknown_module_param_cb);
3519 3520
	if (IS_ERR(after_dashes)) {
		err = PTR_ERR(after_dashes);
3521
		goto bug_cleanup;
3522 3523 3524 3525
	} else if (after_dashes) {
		pr_warn("%s: parameters '%s' after `--' ignored\n",
		       mod->name, after_dashes);
	}
L
Linus Torvalds 已提交
3526

3527
	/* Link in to syfs. */
3528
	err = mod_sysfs_setup(mod, info, mod->kp, mod->num_kp);
L
Linus Torvalds 已提交
3529
	if (err < 0)
3530
		goto bug_cleanup;
3531

3532
	/* Get rid of temporary copy. */
3533
	free_copy(info);
L
Linus Torvalds 已提交
3534 3535

	/* Done! */
3536
	trace_module_load(mod);
3537 3538

	return do_init_module(mod);
L
Linus Torvalds 已提交
3539

3540 3541
 bug_cleanup:
	/* module_bug_cleanup needs module_mutex protection */
3542
	mutex_lock(&module_mutex);
3543
	module_bug_cleanup(mod);
3544
	mutex_unlock(&module_mutex);
3545

3546 3547 3548
	blocking_notifier_call_chain(&module_notify_list,
				     MODULE_STATE_GOING, mod);

3549
	/* we can't deallocate the module until we clear memory protection */
R
Rusty Russell 已提交
3550 3551
	module_disable_ro(mod);
	module_disable_nx(mod);
3552

3553
 ddebug_cleanup:
3554
	dynamic_debug_remove(info->debug);
3555
	synchronize_sched();
3556 3557
	kfree(mod->args);
 free_arch_cleanup:
L
Linus Torvalds 已提交
3558
	module_arch_cleanup(mod);
R
Rusty Russell 已提交
3559
 free_modinfo:
3560
	free_modinfo(mod);
3561
 free_unload:
L
Linus Torvalds 已提交
3562
	module_unload_free(mod);
3563 3564 3565 3566
 unlink_mod:
	mutex_lock(&module_mutex);
	/* Unlink carefully: kallsyms could be walking list. */
	list_del_rcu(&mod->list);
3567
	mod_tree_remove(mod);
3568
	wake_up_all(&module_wq);
3569 3570
	/* Wait for RCU-sched synchronizing before releasing mod->list. */
	synchronize_sched();
3571
	mutex_unlock(&module_mutex);
R
Rusty Russell 已提交
3572
 free_module:
3573 3574 3575 3576 3577 3578
	/*
	 * Ftrace needs to clean up what it initialized.
	 * This does nothing if ftrace_module_init() wasn't called,
	 * but it must be called outside of module_mutex.
	 */
	ftrace_release_mod(mod);
3579
	/* Free lock-classes; relies on the preceding sync_rcu() */
3580
	lockdep_free_key_range(mod->core_layout.base, mod->core_layout.size);
3581

3582
	module_deallocate(mod, info);
R
Rusty Russell 已提交
3583
 free_copy:
3584 3585
	free_copy(info);
	return err;
3586 3587
}

3588 3589
SYSCALL_DEFINE3(init_module, void __user *, umod,
		unsigned long, len, const char __user *, uargs)
L
Linus Torvalds 已提交
3590
{
3591 3592
	int err;
	struct load_info info = { };
L
Linus Torvalds 已提交
3593

3594 3595 3596
	err = may_init_module();
	if (err)
		return err;
L
Linus Torvalds 已提交
3597

3598 3599
	pr_debug("init_module: umod=%p, len=%lu, uargs=%p\n",
	       umod, len, uargs);
L
Linus Torvalds 已提交
3600

3601 3602 3603
	err = copy_module_from_user(umod, len, &info);
	if (err)
		return err;
L
Linus Torvalds 已提交
3604

3605
	return load_module(&info, uargs, 0);
3606
}
3607

3608
SYSCALL_DEFINE3(finit_module, int, fd, const char __user *, uargs, int, flags)
3609 3610 3611
{
	int err;
	struct load_info info = { };
3612

3613 3614 3615
	err = may_init_module();
	if (err)
		return err;
L
Linus Torvalds 已提交
3616

3617
	pr_debug("finit_module: fd=%d, uargs=%p, flags=%i\n", fd, uargs, flags);
3618

3619 3620 3621
	if (flags & ~(MODULE_INIT_IGNORE_MODVERSIONS
		      |MODULE_INIT_IGNORE_VERMAGIC))
		return -EINVAL;
3622

3623 3624 3625
	err = copy_module_from_fd(fd, &info);
	if (err)
		return err;
L
Linus Torvalds 已提交
3626

3627
	return load_module(&info, uargs, flags);
L
Linus Torvalds 已提交
3628 3629 3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641
}

static inline int within(unsigned long addr, void *start, unsigned long size)
{
	return ((void *)addr >= start && (void *)addr < start + size);
}

#ifdef CONFIG_KALLSYMS
/*
 * This ignores the intensely annoying "mapping symbols" found
 * in ARM ELF files: $a, $t and $d.
 */
static inline int is_arm_mapping_symbol(const char *str)
{
3642 3643
	if (str[0] == '.' && str[1] == 'L')
		return true;
K
Kyle McMartin 已提交
3644
	return str[0] == '$' && strchr("axtd", str[1])
L
Linus Torvalds 已提交
3645 3646 3647
	       && (str[2] == '\0' || str[2] == '.');
}

3648
static const char *symname(struct mod_kallsyms *kallsyms, unsigned int symnum)
R
Rusty Russell 已提交
3649
{
3650
	return kallsyms->strtab + kallsyms->symtab[symnum].st_name;
R
Rusty Russell 已提交
3651 3652
}

L
Linus Torvalds 已提交
3653 3654 3655 3656 3657 3658 3659
static const char *get_ksymbol(struct module *mod,
			       unsigned long addr,
			       unsigned long *size,
			       unsigned long *offset)
{
	unsigned int i, best = 0;
	unsigned long nextval;
3660
	struct mod_kallsyms *kallsyms = rcu_dereference_sched(mod->kallsyms);
L
Linus Torvalds 已提交
3661 3662

	/* At worse, next value is at end of module */
3663
	if (within_module_init(addr, mod))
3664
		nextval = (unsigned long)mod->init_layout.base+mod->init_layout.text_size;
D
Daniel Walker 已提交
3665
	else
3666
		nextval = (unsigned long)mod->core_layout.base+mod->core_layout.text_size;
L
Linus Torvalds 已提交
3667

L
Lucas De Marchi 已提交
3668
	/* Scan for closest preceding symbol, and next symbol. (ELF
D
Daniel Walker 已提交
3669
	   starts real symbols at 1). */
3670 3671
	for (i = 1; i < kallsyms->num_symtab; i++) {
		if (kallsyms->symtab[i].st_shndx == SHN_UNDEF)
L
Linus Torvalds 已提交
3672 3673 3674 3675
			continue;

		/* We ignore unnamed symbols: they're uninformative
		 * and inserted at a whim. */
3676 3677
		if (*symname(kallsyms, i) == '\0'
		    || is_arm_mapping_symbol(symname(kallsyms, i)))
R
Rusty Russell 已提交
3678 3679
			continue;

3680 3681
		if (kallsyms->symtab[i].st_value <= addr
		    && kallsyms->symtab[i].st_value > kallsyms->symtab[best].st_value)
L
Linus Torvalds 已提交
3682
			best = i;
3683 3684 3685
		if (kallsyms->symtab[i].st_value > addr
		    && kallsyms->symtab[i].st_value < nextval)
			nextval = kallsyms->symtab[i].st_value;
L
Linus Torvalds 已提交
3686 3687 3688 3689 3690
	}

	if (!best)
		return NULL;

A
Alexey Dobriyan 已提交
3691
	if (size)
3692
		*size = nextval - kallsyms->symtab[best].st_value;
A
Alexey Dobriyan 已提交
3693
	if (offset)
3694 3695
		*offset = addr - kallsyms->symtab[best].st_value;
	return symname(kallsyms, best);
L
Linus Torvalds 已提交
3696 3697
}

3698 3699
/* For kallsyms to ask for address resolution.  NULL means not found.  Careful
 * not to lock to avoid deadlock on oopses, simply disable preemption. */
3700
const char *module_address_lookup(unsigned long addr,
3701 3702 3703 3704
			    unsigned long *size,
			    unsigned long *offset,
			    char **modname,
			    char *namebuf)
L
Linus Torvalds 已提交
3705
{
3706
	const char *ret = NULL;
3707
	struct module *mod;
L
Linus Torvalds 已提交
3708

3709
	preempt_disable();
3710 3711 3712 3713 3714
	mod = __module_address(addr);
	if (mod) {
		if (modname)
			*modname = mod->name;
		ret = get_ksymbol(mod, addr, size, offset);
L
Linus Torvalds 已提交
3715
	}
3716 3717 3718 3719 3720
	/* Make a copy in here where it's safe */
	if (ret) {
		strncpy(namebuf, ret, KSYM_NAME_LEN - 1);
		ret = namebuf;
	}
3721
	preempt_enable();
3722

3723
	return ret;
L
Linus Torvalds 已提交
3724 3725
}

3726 3727 3728 3729
int lookup_module_symbol_name(unsigned long addr, char *symname)
{
	struct module *mod;

3730
	preempt_disable();
3731
	list_for_each_entry_rcu(mod, &modules, list) {
3732 3733
		if (mod->state == MODULE_STATE_UNFORMED)
			continue;
3734
		if (within_module(addr, mod)) {
3735 3736 3737 3738 3739
			const char *sym;

			sym = get_ksymbol(mod, addr, NULL, NULL);
			if (!sym)
				goto out;
3740
			strlcpy(symname, sym, KSYM_NAME_LEN);
3741
			preempt_enable();
3742 3743 3744 3745
			return 0;
		}
	}
out:
3746
	preempt_enable();
3747 3748 3749
	return -ERANGE;
}

3750 3751 3752 3753 3754
int lookup_module_symbol_attrs(unsigned long addr, unsigned long *size,
			unsigned long *offset, char *modname, char *name)
{
	struct module *mod;

3755
	preempt_disable();
3756
	list_for_each_entry_rcu(mod, &modules, list) {
3757 3758
		if (mod->state == MODULE_STATE_UNFORMED)
			continue;
3759
		if (within_module(addr, mod)) {
3760 3761 3762 3763 3764 3765
			const char *sym;

			sym = get_ksymbol(mod, addr, size, offset);
			if (!sym)
				goto out;
			if (modname)
3766
				strlcpy(modname, mod->name, MODULE_NAME_LEN);
3767
			if (name)
3768
				strlcpy(name, sym, KSYM_NAME_LEN);
3769
			preempt_enable();
3770 3771 3772 3773
			return 0;
		}
	}
out:
3774
	preempt_enable();
3775 3776 3777
	return -ERANGE;
}

3778 3779
int module_get_kallsym(unsigned int symnum, unsigned long *value, char *type,
			char *name, char *module_name, int *exported)
L
Linus Torvalds 已提交
3780 3781 3782
{
	struct module *mod;

3783
	preempt_disable();
3784
	list_for_each_entry_rcu(mod, &modules, list) {
3785 3786
		struct mod_kallsyms *kallsyms;

3787 3788
		if (mod->state == MODULE_STATE_UNFORMED)
			continue;
3789 3790 3791 3792 3793
		kallsyms = rcu_dereference_sched(mod->kallsyms);
		if (symnum < kallsyms->num_symtab) {
			*value = kallsyms->symtab[symnum].st_value;
			*type = kallsyms->symtab[symnum].st_info;
			strlcpy(name, symname(kallsyms, symnum), KSYM_NAME_LEN);
3794
			strlcpy(module_name, mod->name, MODULE_NAME_LEN);
3795
			*exported = is_exported(name, *value, mod);
3796
			preempt_enable();
3797
			return 0;
L
Linus Torvalds 已提交
3798
		}
3799
		symnum -= kallsyms->num_symtab;
L
Linus Torvalds 已提交
3800
	}
3801
	preempt_enable();
3802
	return -ERANGE;
L
Linus Torvalds 已提交
3803 3804 3805 3806 3807
}

static unsigned long mod_find_symname(struct module *mod, const char *name)
{
	unsigned int i;
3808
	struct mod_kallsyms *kallsyms = rcu_dereference_sched(mod->kallsyms);
L
Linus Torvalds 已提交
3809

3810 3811 3812 3813
	for (i = 0; i < kallsyms->num_symtab; i++)
		if (strcmp(name, symname(kallsyms, i)) == 0 &&
		    kallsyms->symtab[i].st_info != 'U')
			return kallsyms->symtab[i].st_value;
L
Linus Torvalds 已提交
3814 3815 3816 3817 3818 3819 3820 3821 3822 3823 3824
	return 0;
}

/* Look for this name: can be of form module:name. */
unsigned long module_kallsyms_lookup_name(const char *name)
{
	struct module *mod;
	char *colon;
	unsigned long ret = 0;

	/* Don't lock: we're in enough trouble already. */
3825
	preempt_disable();
L
Linus Torvalds 已提交
3826
	if ((colon = strchr(name, ':')) != NULL) {
3827
		if ((mod = find_module_all(name, colon - name, false)) != NULL)
L
Linus Torvalds 已提交
3828 3829
			ret = mod_find_symname(mod, colon+1);
	} else {
3830 3831 3832
		list_for_each_entry_rcu(mod, &modules, list) {
			if (mod->state == MODULE_STATE_UNFORMED)
				continue;
L
Linus Torvalds 已提交
3833 3834
			if ((ret = mod_find_symname(mod, name)) != 0)
				break;
3835
		}
L
Linus Torvalds 已提交
3836
	}
3837
	preempt_enable();
L
Linus Torvalds 已提交
3838 3839
	return ret;
}
3840 3841 3842 3843 3844 3845 3846 3847 3848

int module_kallsyms_on_each_symbol(int (*fn)(void *, const char *,
					     struct module *, unsigned long),
				   void *data)
{
	struct module *mod;
	unsigned int i;
	int ret;

3849 3850
	module_assert_mutex();

3851
	list_for_each_entry(mod, &modules, list) {
3852 3853 3854
		/* We hold module_mutex: no need for rcu_dereference_sched */
		struct mod_kallsyms *kallsyms = mod->kallsyms;

3855 3856
		if (mod->state == MODULE_STATE_UNFORMED)
			continue;
3857 3858 3859
		for (i = 0; i < kallsyms->num_symtab; i++) {
			ret = fn(data, symname(kallsyms, i),
				 mod, kallsyms->symtab[i].st_value);
3860 3861 3862 3863 3864 3865
			if (ret != 0)
				return ret;
		}
	}
	return 0;
}
L
Linus Torvalds 已提交
3866 3867
#endif /* CONFIG_KALLSYMS */

3868
static char *module_flags(struct module *mod, char *buf)
3869 3870 3871
{
	int bx = 0;

3872
	BUG_ON(mod->state == MODULE_STATE_UNFORMED);
3873 3874 3875
	if (mod->taints ||
	    mod->state == MODULE_STATE_GOING ||
	    mod->state == MODULE_STATE_COMING) {
3876
		buf[bx++] = '(';
3877
		bx += module_flags_taint(mod, buf + bx);
3878 3879 3880 3881 3882 3883
		/* Show a - for module-is-being-unloaded */
		if (mod->state == MODULE_STATE_GOING)
			buf[bx++] = '-';
		/* Show a + for module-is-being-loaded */
		if (mod->state == MODULE_STATE_COMING)
			buf[bx++] = '+';
3884 3885 3886 3887 3888 3889 3890
		buf[bx++] = ')';
	}
	buf[bx] = '\0';

	return buf;
}

3891 3892 3893 3894 3895 3896 3897 3898 3899 3900 3901 3902 3903 3904 3905 3906 3907 3908
#ifdef CONFIG_PROC_FS
/* Called by the /proc file system to return a list of modules. */
static void *m_start(struct seq_file *m, loff_t *pos)
{
	mutex_lock(&module_mutex);
	return seq_list_start(&modules, *pos);
}

static void *m_next(struct seq_file *m, void *p, loff_t *pos)
{
	return seq_list_next(p, &modules, pos);
}

static void m_stop(struct seq_file *m, void *p)
{
	mutex_unlock(&module_mutex);
}

L
Linus Torvalds 已提交
3909 3910 3911
static int m_show(struct seq_file *m, void *p)
{
	struct module *mod = list_entry(p, struct module, list);
3912 3913
	char buf[8];

3914 3915 3916 3917
	/* We always ignore unformed modules. */
	if (mod->state == MODULE_STATE_UNFORMED)
		return 0;

3918
	seq_printf(m, "%s %u",
3919
		   mod->name, mod->init_layout.size + mod->core_layout.size);
L
Linus Torvalds 已提交
3920 3921 3922 3923
	print_unload_info(m, mod);

	/* Informative for users. */
	seq_printf(m, " %s",
3924 3925
		   mod->state == MODULE_STATE_GOING ? "Unloading" :
		   mod->state == MODULE_STATE_COMING ? "Loading" :
L
Linus Torvalds 已提交
3926 3927
		   "Live");
	/* Used by oprofile and other similar tools. */
3928
	seq_printf(m, " 0x%pK", mod->core_layout.base);
L
Linus Torvalds 已提交
3929

3930 3931
	/* Taints info */
	if (mod->taints)
3932
		seq_printf(m, " %s", module_flags(mod, buf));
3933

3934
	seq_puts(m, "\n");
L
Linus Torvalds 已提交
3935 3936 3937 3938 3939 3940 3941 3942
	return 0;
}

/* Format: modulename size refcount deps address

   Where refcount is a number or -, and deps is a comma-separated list
   of depends or -.
*/
3943
static const struct seq_operations modules_op = {
L
Linus Torvalds 已提交
3944 3945 3946 3947 3948 3949
	.start	= m_start,
	.next	= m_next,
	.stop	= m_stop,
	.show	= m_show
};

3950 3951 3952 3953 3954 3955 3956 3957 3958 3959 3960 3961 3962 3963 3964 3965 3966 3967 3968 3969
static int modules_open(struct inode *inode, struct file *file)
{
	return seq_open(file, &modules_op);
}

static const struct file_operations proc_modules_operations = {
	.open		= modules_open,
	.read		= seq_read,
	.llseek		= seq_lseek,
	.release	= seq_release,
};

static int __init proc_modules_init(void)
{
	proc_create("modules", 0, NULL, &proc_modules_operations);
	return 0;
}
module_init(proc_modules_init);
#endif

L
Linus Torvalds 已提交
3970 3971 3972 3973 3974 3975
/* Given an address, look for it in the module exception tables. */
const struct exception_table_entry *search_module_extables(unsigned long addr)
{
	const struct exception_table_entry *e = NULL;
	struct module *mod;

R
Rusty Russell 已提交
3976
	preempt_disable();
3977
	list_for_each_entry_rcu(mod, &modules, list) {
3978 3979
		if (mod->state == MODULE_STATE_UNFORMED)
			continue;
L
Linus Torvalds 已提交
3980 3981
		if (mod->num_exentries == 0)
			continue;
D
Daniel Walker 已提交
3982

L
Linus Torvalds 已提交
3983 3984 3985 3986 3987 3988
		e = search_extable(mod->extable,
				   mod->extable + mod->num_exentries - 1,
				   addr);
		if (e)
			break;
	}
R
Rusty Russell 已提交
3989
	preempt_enable();
L
Linus Torvalds 已提交
3990 3991

	/* Now, if we found one, we are running inside it now, hence
D
Daniel Walker 已提交
3992
	   we cannot unload the module, hence no refcnt needed. */
L
Linus Torvalds 已提交
3993 3994 3995
	return e;
}

3996
/*
R
Rusty Russell 已提交
3997 3998 3999 4000 4001
 * is_module_address - is this address inside a module?
 * @addr: the address to check.
 *
 * See is_module_text_address() if you simply want to see if the address
 * is code (not data).
4002
 */
R
Rusty Russell 已提交
4003
bool is_module_address(unsigned long addr)
4004
{
R
Rusty Russell 已提交
4005
	bool ret;
4006

R
Rusty Russell 已提交
4007
	preempt_disable();
R
Rusty Russell 已提交
4008
	ret = __module_address(addr) != NULL;
R
Rusty Russell 已提交
4009
	preempt_enable();
4010

R
Rusty Russell 已提交
4011
	return ret;
4012 4013
}

R
Rusty Russell 已提交
4014 4015 4016 4017 4018 4019 4020
/*
 * __module_address - get the module which contains an address.
 * @addr: the address.
 *
 * Must be called with preempt disabled or module mutex held so that
 * module doesn't get freed during this.
 */
4021
struct module *__module_address(unsigned long addr)
L
Linus Torvalds 已提交
4022 4023 4024
{
	struct module *mod;

4025 4026 4027
	if (addr < module_addr_min || addr > module_addr_max)
		return NULL;

4028 4029
	module_assert_mutex_or_preempt();

4030
	mod = mod_find(addr);
4031 4032
	if (mod) {
		BUG_ON(!within_module(addr, mod));
4033
		if (mod->state == MODULE_STATE_UNFORMED)
4034
			mod = NULL;
4035
	}
4036
	return mod;
L
Linus Torvalds 已提交
4037
}
4038
EXPORT_SYMBOL_GPL(__module_address);
L
Linus Torvalds 已提交
4039

R
Rusty Russell 已提交
4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050 4051 4052 4053 4054 4055 4056 4057 4058 4059 4060 4061 4062 4063 4064 4065 4066 4067 4068 4069 4070
/*
 * is_module_text_address - is this address inside module code?
 * @addr: the address to check.
 *
 * See is_module_address() if you simply want to see if the address is
 * anywhere in a module.  See kernel_text_address() for testing if an
 * address corresponds to kernel or module code.
 */
bool is_module_text_address(unsigned long addr)
{
	bool ret;

	preempt_disable();
	ret = __module_text_address(addr) != NULL;
	preempt_enable();

	return ret;
}

/*
 * __module_text_address - get the module whose code contains an address.
 * @addr: the address.
 *
 * Must be called with preempt disabled or module mutex held so that
 * module doesn't get freed during this.
 */
struct module *__module_text_address(unsigned long addr)
{
	struct module *mod = __module_address(addr);
	if (mod) {
		/* Make sure it's within the text section. */
4071 4072
		if (!within(addr, mod->init_layout.base, mod->init_layout.text_size)
		    && !within(addr, mod->core_layout.base, mod->core_layout.text_size))
R
Rusty Russell 已提交
4073 4074 4075 4076
			mod = NULL;
	}
	return mod;
}
4077
EXPORT_SYMBOL_GPL(__module_text_address);
R
Rusty Russell 已提交
4078

L
Linus Torvalds 已提交
4079 4080 4081 4082
/* Don't grab lock, we're oopsing. */
void print_modules(void)
{
	struct module *mod;
4083
	char buf[8];
L
Linus Torvalds 已提交
4084

4085
	printk(KERN_DEFAULT "Modules linked in:");
4086 4087
	/* Most callers should already have preempt disabled, but make sure */
	preempt_disable();
4088 4089 4090
	list_for_each_entry_rcu(mod, &modules, list) {
		if (mod->state == MODULE_STATE_UNFORMED)
			continue;
J
Jiri Slaby 已提交
4091
		pr_cont(" %s%s", mod->name, module_flags(mod, buf));
4092
	}
4093
	preempt_enable();
4094
	if (last_unloaded_module[0])
J
Jiri Slaby 已提交
4095 4096
		pr_cont(" [last unloaded: %s]", last_unloaded_module);
	pr_cont("\n");
L
Linus Torvalds 已提交
4097 4098 4099
}

#ifdef CONFIG_MODVERSIONS
4100 4101 4102 4103 4104 4105
/* Generate the signature for all relevant module structures here.
 * If these change, we don't want to try to parse the module. */
void module_layout(struct module *mod,
		   struct modversion_info *ver,
		   struct kernel_param *kp,
		   struct kernel_symbol *ks,
4106
		   struct tracepoint * const *tp)
4107 4108 4109
{
}
EXPORT_SYMBOL(module_layout);
L
Linus Torvalds 已提交
4110
#endif