summaryrefslogtreecommitdiff
path: root/drivers/misc/lkdtm/bugs.c
blob: f21854ac5cc2b74c647914ec9122da645eb9e988 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
// SPDX-License-Identifier: GPL-2.0
/*
 * This is for all the tests related to logic bugs (e.g. bad dereferences,
 * bad alignment, bad loops, bad locking, bad scheduling, deep stacks, and
 * lockups) along with other things that don't fit well into existing LKDTM
 * test source files.
 */
#include "lkdtm.h"
#include <linux/list.h>
#include <linux/sched.h>
#include <linux/sched/signal.h>
#include <linux/sched/task_stack.h>
#include <linux/uaccess.h>
#include <linux/slab.h>

#if IS_ENABLED(CONFIG_X86_32) && !IS_ENABLED(CONFIG_UML)
#include <asm/desc.h>
#endif

struct lkdtm_list {
	struct list_head node;
};

/*
 * Make sure our attempts to over run the kernel stack doesn't trigger
 * a compiler warning when CONFIG_FRAME_WARN is set. Then make sure we
 * recurse past the end of THREAD_SIZE by default.
 */
#if defined(CONFIG_FRAME_WARN) && (CONFIG_FRAME_WARN > 0)
#define REC_STACK_SIZE (_AC(CONFIG_FRAME_WARN, UL) / 2)
#else
#define REC_STACK_SIZE (THREAD_SIZE / 8)
#endif
#define REC_NUM_DEFAULT ((THREAD_SIZE / REC_STACK_SIZE) * 2)

static int recur_count = REC_NUM_DEFAULT;

static DEFINE_SPINLOCK(lock_me_up);

/*
 * Make sure compiler does not optimize this function or stack frame away:
 * - function marked noinline
 * - stack variables are marked volatile
 * - stack variables are written (memset()) and read (buf[..] passed as arg)
 * - function may have external effects (memzero_explicit())
 * - no tail recursion possible
 */
static int noinline recursive_loop(int remaining)
{
	volatile char buf[REC_STACK_SIZE];
	volatile int ret;

	memset((void *)buf, remaining & 0xFF, sizeof(buf));
	if (!remaining)
		ret = 0;
	else
		ret = recursive_loop((int)buf[remaining % sizeof(buf)] - 1);
	memzero_explicit((void *)buf, sizeof(buf));
	return ret;
}

/* If the depth is negative, use the default, otherwise keep parameter. */
void __init lkdtm_bugs_init(int *recur_param)
{
	if (*recur_param < 0)
		*recur_param = recur_count;
	else
		recur_count = *recur_param;
}

void lkdtm_PANIC(void)
{
	panic("dumptest");
}

void lkdtm_BUG(void)
{
	BUG();
}

static int warn_counter;

void lkdtm_WARNING(void)
{
	WARN_ON(++warn_counter);
}

void lkdtm_WARNING_MESSAGE(void)
{
	WARN(1, "Warning message trigger count: %d\n", ++warn_counter);
}

void lkdtm_EXCEPTION(void)
{
	*((volatile int *) 0) = 0;
}

void lkdtm_LOOP(void)
{
	for (;;)
		;
}

void lkdtm_EXHAUST_STACK(void)
{
	pr_info("Calling function with %lu frame size to depth %d ...\n",
		REC_STACK_SIZE, recur_count);
	recursive_loop(recur_count);
	pr_info("FAIL: survived without exhausting stack?!\n");
}

static noinline void __lkdtm_CORRUPT_STACK(void *stack)
{
	memset(stack, '\xff', 64);
}

/* This should trip the stack canary, not corrupt the return address. */
noinline void lkdtm_CORRUPT_STACK(void)
{
	/* Use default char array length that triggers stack protection. */
	char data[8] __aligned(sizeof(void *));

	pr_info("Corrupting stack containing char array ...\n");
	__lkdtm_CORRUPT_STACK((void *)&data);
}

/* Same as above but will only get a canary with -fstack-protector-strong */
noinline void lkdtm_CORRUPT_STACK_STRONG(void)
{
	union {
		unsigned short shorts[4];
		unsigned long *ptr;
	} data __aligned(sizeof(void *));

	pr_info("Corrupting stack containing union ...\n");
	__lkdtm_CORRUPT_STACK((void *)&data);
}

static pid_t stack_pid;
static unsigned long stack_addr;

void lkdtm_REPORT_STACK(void)
{
	volatile uintptr_t magic;
	pid_t pid = task_pid_nr(current);

	if (pid != stack_pid) {
		pr_info("Starting stack offset tracking for pid %d\n", pid);
		stack_pid = pid;
		stack_addr = (uintptr_t)&magic;
	}

	pr_info("Stack offset: %d\n", (int)(stack_addr - (uintptr_t)&magic));
}

static pid_t stack_canary_pid;
static unsigned long stack_canary;
static unsigned long stack_canary_offset;

static noinline void __lkdtm_REPORT_STACK_CANARY(void *stack)
{
	int i = 0;
	pid_t pid = task_pid_nr(current);
	unsigned long *canary = (unsigned long *)stack;
	unsigned long current_offset = 0, init_offset = 0;

	/* Do our best to find the canary in a 16 word window ... */
	for (i = 1; i < 16; i++) {
		canary = (unsigned long *)stack + i;
#ifdef CONFIG_STACKPROTECTOR
		if (*canary == current->stack_canary)
			current_offset = i;
		if (*canary == init_task.stack_canary)
			init_offset = i;
#endif
	}

	if (current_offset == 0) {
		/*
		 * If the canary doesn't match what's in the task_struct,
		 * we're either using a global canary or the stack frame
		 * layout changed.
		 */
		if (init_offset != 0) {
			pr_err("FAIL: global stack canary found at offset %ld (canary for pid %d matches init_task's)!\n",
			       init_offset, pid);
		} else {
			pr_warn("FAIL: did not correctly locate stack canary :(\n");
			pr_expected_config(CONFIG_STACKPROTECTOR);
		}

		return;
	} else if (init_offset != 0) {
		pr_warn("WARNING: found both current and init_task canaries nearby?!\n");
	}

	canary = (unsigned long *)stack + current_offset;
	if (stack_canary_pid == 0) {
		stack_canary = *canary;
		stack_canary_pid = pid;
		stack_canary_offset = current_offset;
		pr_info("Recorded stack canary for pid %d at offset %ld\n",
			stack_canary_pid, stack_canary_offset);
	} else if (pid == stack_canary_pid) {
		pr_warn("ERROR: saw pid %d again -- please use a new pid\n", pid);
	} else {
		if (current_offset != stack_canary_offset) {
			pr_warn("ERROR: canary offset changed from %ld to %ld!?\n",
				stack_canary_offset, current_offset);
			return;
		}

		if (*canary == stack_canary) {
			pr_warn("FAIL: canary identical for pid %d and pid %d at offset %ld!\n",
				stack_canary_pid, pid, current_offset);
		} else {
			pr_info("ok: stack canaries differ between pid %d and pid %d at offset %ld.\n",
				stack_canary_pid, pid, current_offset);
			/* Reset the test. */
			stack_canary_pid = 0;
		}
	}
}

void lkdtm_REPORT_STACK_CANARY(void)
{
	/* Use default char array length that triggers stack protection. */
	char data[8] __aligned(sizeof(void *)) = { };

	__lkdtm_REPORT_STACK_CANARY((void *)&data);
}

void lkdtm_UNALIGNED_LOAD_STORE_WRITE(void)
{
	static u8 data[5] __attribute__((aligned(4))) = {1, 2, 3, 4, 5};
	u32 *p;
	u32 val = 0x12345678;

	p = (u32 *)(data + 1);
	if (*p == 0)
		val = 0x87654321;
	*p = val;

	if (IS_ENABLED(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS))
		pr_err("XFAIL: arch has CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS\n");
}

void lkdtm_SOFTLOCKUP(void)
{
	preempt_disable();
	for (;;)
		cpu_relax();
}

void lkdtm_HARDLOCKUP(void)
{
	local_irq_disable();
	for (;;)
		cpu_relax();
}

void lkdtm_SPINLOCKUP(void)
{
	/* Must be called twice to trigger. */
	spin_lock(&lock_me_up);
	/* Let sparse know we intended to exit holding the lock. */
	__release(&lock_me_up);
}

void lkdtm_HUNG_TASK(void)
{
	set_current_state(TASK_UNINTERRUPTIBLE);
	schedule();
}

volatile unsigned int huge = INT_MAX - 2;
volatile unsigned int ignored;

void lkdtm_OVERFLOW_SIGNED(void)
{
	int value;

	value = huge;
	pr_info("Normal signed addition ...\n");
	value += 1;
	ignored = value;

	pr_info("Overflowing signed addition ...\n");
	value += 4;
	ignored = value;
}


void lkdtm_OVERFLOW_UNSIGNED(void)
{
	unsigned int value;

	value = huge;
	pr_info("Normal unsigned addition ...\n");
	value += 1;
	ignored = value;

	pr_info("Overflowing unsigned addition ...\n");
	value += 4;
	ignored = value;
}

/* Intentionally using old-style flex array definition of 1 byte. */
struct array_bounds_flex_array {
	int one;
	int two;
	char data[1];
};

struct array_bounds {
	int one;
	int two;
	char data[8];
	int three;
};

void lkdtm_ARRAY_BOUNDS(void)
{
	struct array_bounds_flex_array *not_checked;
	struct array_bounds *checked;
	volatile int i;

	not_checked = kmalloc(sizeof(*not_checked) * 2, GFP_KERNEL);
	checked = kmalloc(sizeof(*checked) * 2, GFP_KERNEL);

	pr_info("Array access within bounds ...\n");
	/* For both, touch all bytes in the actual member size. */
	for (i = 0; i < sizeof(checked->data); i++)
		checked->data[i] = 'A';
	/*
	 * For the uninstrumented flex array member, also touch 1 byte
	 * beyond to verify it is correctly uninstrumented.
	 */
	for (i = 0; i < sizeof(not_checked->data) + 1; i++)
		not_checked->data[i] = 'A';

	pr_info("Array access beyond bounds ...\n");
	for (i = 0; i < sizeof(checked->data) + 1; i++)
		checked->data[i] = 'B';

	kfree(not_checked);
	kfree(checked);
	pr_err("FAIL: survived array bounds overflow!\n");
	pr_expected_config(CONFIG_UBSAN_BOUNDS);
}

void lkdtm_CORRUPT_LIST_ADD(void)
{
	/*
	 * Initially, an empty list via LIST_HEAD:
	 *	test_head.next = &test_head
	 *	test_head.prev = &test_head
	 */
	LIST_HEAD(test_head);
	struct lkdtm_list good, bad;
	void *target[2] = { };
	void *redirection = &target;

	pr_info("attempting good list addition\n");

	/*
	 * Adding to the list performs these actions:
	 *	test_head.next->prev = &good.node
	 *	good.node.next = test_head.next
	 *	good.node.prev = test_head
	 *	test_head.next = good.node
	 */
	list_add(&good.node, &test_head);

	pr_info("attempting corrupted list addition\n");
	/*
	 * In simulating this "write what where" primitive, the "what" is
	 * the address of &bad.node, and the "where" is the address held
	 * by "redirection".
	 */
	test_head.next = redirection;
	list_add(&bad.node, &test_head);

	if (target[0] == NULL && target[1] == NULL)
		pr_err("Overwrite did not happen, but no BUG?!\n");
	else {
		pr_err("list_add() corruption not detected!\n");
		pr_expected_config(CONFIG_DEBUG_LIST);
	}
}

void lkdtm_CORRUPT_LIST_DEL(void)
{
	LIST_HEAD(test_head);
	struct lkdtm_list item;
	void *target[2] = { };
	void *redirection = &target;

	list_add(&item.node, &test_head);

	pr_info("attempting good list removal\n");
	list_del(&item.node);

	pr_info("attempting corrupted list removal\n");
	list_add(&item.node, &test_head);

	/* As with the list_add() test above, this corrupts "next". */
	item.node.next = redirection;
	list_del(&item.node);

	if (target[0] == NULL && target[1] == NULL)
		pr_err("Overwrite did not happen, but no BUG?!\n");
	else {
		pr_err("list_del() corruption not detected!\n");
		pr_expected_config(CONFIG_DEBUG_LIST);
	}
}

/* Test that VMAP_STACK is actually allocating with a leading guard page */
void lkdtm_STACK_GUARD_PAGE_LEADING(void)
{
	const unsigned char *stack = task_stack_page(current);
	const unsigned char *ptr = stack - 1;
	volatile unsigned char byte;

	pr_info("attempting bad read from page below current stack\n");

	byte = *ptr;

	pr_err("FAIL: accessed page before stack! (byte: %x)\n", byte);
}

/* Test that VMAP_STACK is actually allocating with a trailing guard page */
void lkdtm_STACK_GUARD_PAGE_TRAILING(void)
{
	const unsigned char *stack = task_stack_page(current);
	const unsigned char *ptr = stack + THREAD_SIZE;
	volatile unsigned char byte;

	pr_info("attempting bad read from page above current stack\n");

	byte = *ptr;

	pr_err("FAIL: accessed page after stack! (byte: %x)\n", byte);
}

void lkdtm_UNSET_SMEP(void)
{
#if IS_ENABLED(CONFIG_X86_64) && !IS_ENABLED(CONFIG_UML)
#define MOV_CR4_DEPTH	64
	void (*direct_write_cr4)(unsigned long val);
	unsigned char *insn;
	unsigned long cr4;
	int i;

	cr4 = native_read_cr4();

	if ((cr4 & X86_CR4_SMEP) != X86_CR4_SMEP) {
		pr_err("FAIL: SMEP not in use\n");
		return;
	}
	cr4 &= ~(X86_CR4_SMEP);

	pr_info("trying to clear SMEP normally\n");
	native_write_cr4(cr4);
	if (cr4 == native_read_cr4()) {
		pr_err("FAIL: pinning SMEP failed!\n");
		cr4 |= X86_CR4_SMEP;
		pr_info("restoring SMEP\n");
		native_write_cr4(cr4);
		return;
	}
	pr_info("ok: SMEP did not get cleared\n");

	/*
	 * To test the post-write pinning verification we need to call
	 * directly into the middle of native_write_cr4() where the
	 * cr4 write happens, skipping any pinning. This searches for
	 * the cr4 writing instruction.
	 */
	insn = (unsigned char *)native_write_cr4;
	for (i = 0; i < MOV_CR4_DEPTH; i++) {
		/* mov %rdi, %cr4 */
		if (insn[i] == 0x0f && insn[i+1] == 0x22 && insn[i+2] == 0xe7)
			break;
		/* mov %rdi,%rax; mov %rax, %cr4 */
		if (insn[i]   == 0x48 && insn[i+1] == 0x89 &&
		    insn[i+2] == 0xf8 && insn[i+3] == 0x0f &&
		    insn[i+4] == 0x22 && insn[i+5] == 0xe0)
			break;
	}
	if (i >= MOV_CR4_DEPTH) {
		pr_info("ok: cannot locate cr4 writing call gadget\n");
		return;
	}
	direct_write_cr4 = (void *)(insn + i);

	pr_info("trying to clear SMEP with call gadget\n");
	direct_write_cr4(cr4);
	if (native_read_cr4() & X86_CR4_SMEP) {
		pr_info("ok: SMEP removal was reverted\n");
	} else {
		pr_err("FAIL: cleared SMEP not detected!\n");
		cr4 |= X86_CR4_SMEP;
		pr_info("restoring SMEP\n");
		native_write_cr4(cr4);
	}
#else
	pr_err("XFAIL: this test is x86_64-only\n");
#endif
}

void lkdtm_DOUBLE_FAULT(void)
{
#if IS_ENABLED(CONFIG_X86_32) && !IS_ENABLED(CONFIG_UML)
	/*
	 * Trigger #DF by setting the stack limit to zero.  This clobbers
	 * a GDT TLS slot, which is okay because the current task will die
	 * anyway due to the double fault.
	 */
	struct desc_struct d = {
		.type = 3,	/* expand-up, writable, accessed data */
		.p = 1,		/* present */
		.d = 1,		/* 32-bit */
		.g = 0,		/* limit in bytes */
		.s = 1,		/* not system */
	};

	local_irq_disable();
	write_gdt_entry(get_cpu_gdt_rw(smp_processor_id()),
			GDT_ENTRY_TLS_MIN, &d, DESCTYPE_S);

	/*
	 * Put our zero-limit segment in SS and then trigger a fault.  The
	 * 4-byte access to (%esp) will fault with #SS, and the attempt to
	 * deliver the fault will recursively cause #SS and result in #DF.
	 * This whole process happens while NMIs and MCEs are blocked by the
	 * MOV SS window.  This is nice because an NMI with an invalid SS
	 * would also double-fault, resulting in the NMI or MCE being lost.
	 */
	asm volatile ("movw %0, %%ss; addl $0, (%%esp)" ::
		      "r" ((unsigned short)(GDT_ENTRY_TLS_MIN << 3)));

	pr_err("FAIL: tried to double fault but didn't die\n");
#else
	pr_err("XFAIL: this test is ia32-only\n");
#endif
}

#ifdef CONFIG_ARM64
static noinline void change_pac_parameters(void)
{
	if (IS_ENABLED(CONFIG_ARM64_PTR_AUTH_KERNEL)) {
		/* Reset the keys of current task */
		ptrauth_thread_init_kernel(current);
		ptrauth_thread_switch_kernel(current);
	}
}
#endif

noinline void lkdtm_CORRUPT_PAC(void)
{
#ifdef CONFIG_ARM64
#define CORRUPT_PAC_ITERATE	10
	int i;

	if (!IS_ENABLED(CONFIG_ARM64_PTR_AUTH_KERNEL))
		pr_err("FAIL: kernel not built with CONFIG_ARM64_PTR_AUTH_KERNEL\n");

	if (!system_supports_address_auth()) {
		pr_err("FAIL: CPU lacks pointer authentication feature\n");
		return;
	}

	pr_info("changing PAC parameters to force function return failure...\n");
	/*
	 * PAC is a hash value computed from input keys, return address and
	 * stack pointer. As pac has fewer bits so there is a chance of
	 * collision, so iterate few times to reduce the collision probability.
	 */
	for (i = 0; i < CORRUPT_PAC_ITERATE; i++)
		change_pac_parameters();

	pr_err("FAIL: survived PAC changes! Kernel may be unstable from here\n");
#else
	pr_err("XFAIL: this test is arm64-only\n");
#endif
}