summaryrefslogtreecommitdiffstats
path: root/arch/x86/mm/mmio-mod.c
blob: e7397e108beb2d55c192eee9413a28468ac66fb6 (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
/*
 * 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.
 *
 * Copyright (C) IBM Corporation, 2005
 *               Jeff Muizelaar, 2006, 2007
 *               Pekka Paalanen, 2008 <pq@iki.fi>
 *
 * Derived from the read-mod example from relay-examples by Tom Zanussi.
 */
#define DEBUG 1

#include <linux/module.h>
#include <linux/debugfs.h>
#include <linux/uaccess.h>
#include <linux/io.h>
#include <linux/version.h>
#include <linux/kallsyms.h>
#include <asm/pgtable.h>
#include <linux/mmiotrace.h>
#include <asm/e820.h> /* for ISA_START_ADDRESS */
#include <asm/atomic.h>
#include <linux/percpu.h>
#include <linux/cpu.h>

#include "pf_in.h"

#define NAME "mmiotrace: "

struct trap_reason {
	unsigned long addr;
	unsigned long ip;
	enum reason_type type;
	int active_traces;
};

struct remap_trace {
	struct list_head list;
	struct kmmio_probe probe;
	resource_size_t phys;
	unsigned long id;
};

/* Accessed per-cpu. */
static DEFINE_PER_CPU(struct trap_reason, pf_reason);
static DEFINE_PER_CPU(struct mmiotrace_rw, cpu_trace);

#if 0 /* XXX: no way gather this info anymore */
/* Access to this is not per-cpu. */
static DEFINE_PER_CPU(atomic_t, dropped);
#endif

static struct dentry *marker_file;

static DEFINE_MUTEX(mmiotrace_mutex);
static DEFINE_SPINLOCK(trace_lock);
static atomic_t mmiotrace_enabled;
static LIST_HEAD(trace_list);		/* struct remap_trace */

/*
 * Locking in this file:
 * - mmiotrace_mutex enforces enable/disable_mmiotrace() critical sections.
 * - mmiotrace_enabled may be modified only when holding mmiotrace_mutex
 *   and trace_lock.
 * - Routines depending on is_enabled() must take trace_lock.
 * - trace_list users must hold trace_lock.
 * - is_enabled() guarantees that mmio_trace_record is allowed.
 * - pre/post callbacks assume the effect of is_enabled() being true.
 */

/* module parameters */
static unsigned long	filter_offset;
static int		nommiotrace;
static int		trace_pc;

module_param(filter_offset, ulong, 0);
module_param(nommiotrace, bool, 0);
module_param(trace_pc, bool, 0);

MODULE_PARM_DESC(filter_offset, "Start address of traced mappings.");
MODULE_PARM_DESC(nommiotrace, "Disable actual MMIO tracing.");
MODULE_PARM_DESC(trace_pc, "Record address of faulting instructions.");

static bool is_enabled(void)
{
	return atomic_read(&mmiotrace_enabled);
}

#if 0 /* XXX: needs rewrite */
/*
 * Write callback for the debugfs entry:
 * Read a marker and write it to the mmio trace log
 */
static ssize_t write_marker(struct file *file, const char __user *buffer,
						size_t count, loff_t *ppos)
{
	char *event = NULL;
	struct mm_io_header *headp;
	ssize_t len = (count > 65535) ? 65535 : count;

	event = kzalloc(sizeof(*headp) + len, GFP_KERNEL);
	if (!event)
		return -ENOMEM;

	headp = (struct mm_io_header *)event;
	headp->type = MMIO_MAGIC | (MMIO_MARKER << MMIO_OPCODE_SHIFT);
	headp->data_len = len;

	if (copy_from_user(event + sizeof(*headp), buffer, len)) {
		kfree(event);
		return -EFAULT;
	}

	spin_lock_irq(&trace_lock);
#if 0 /* XXX: convert this to use tracing */
	if (is_enabled())
		relay_write(chan, event, sizeof(*headp) + len);
	else
#endif
		len = -EINVAL;
	spin_unlock_irq(&trace_lock);
	kfree(event);
	return len;
}
#endif

static void print_pte(unsigned long address)
{
	unsigned int level;
	pte_t *pte = lookup_address(address, &level);

	if (!pte) {
		pr_err(NAME "Error in %s: no pte for page 0x%08lx\n",
							__func__, address);
		return;
	}

	if (level == PG_LEVEL_2M) {
		pr_emerg(NAME "4MB pages are not currently supported: "
							"0x%08lx\n", address);
		BUG();
	}
	pr_info(NAME "pte for 0x%lx: 0x%llx 0x%llx\n", address,
		(unsigned long long)pte_val(*pte),
		(unsigned long long)pte_val(*pte) & _PAGE_PRESENT);
}

/*
 * For some reason the pre/post pairs have been called in an
 * unmatched order. Report and die.
 */
static void die_kmmio_nesting_error(struct pt_regs *regs, unsigned long addr)
{
	const struct trap_reason *my_reason = &get_cpu_var(pf_reason);
	pr_emerg(NAME "unexpected fault for address: 0x%08lx, "
					"last fault for address: 0x%08lx\n",
					addr, my_reason->addr);
	print_pte(addr);
	print_symbol(KERN_EMERG "faulting IP is at %s\n", regs->ip);
	print_symbol(KERN_EMERG "last faulting IP was at %s\n", my_reason->ip);
#ifdef __i386__
	pr_emerg("eax: %08lx   ebx: %08lx   ecx: %08lx   edx: %08lx\n",
			regs->ax, regs->bx, regs->cx, regs->dx);
	pr_emerg("esi: %08lx   edi: %08lx   ebp: %08lx   esp: %08lx\n",
			regs->si, regs->di, regs->bp, regs->sp);
#else
	pr_emerg("rax: %016lx   rcx: %016lx   rdx: %016lx\n",
					regs->ax, regs->cx, regs->dx);
	pr_emerg("rsi: %016lx   rdi: %016lx   rbp: %016lx   rsp: %016lx\n",
				regs->si, regs->di, regs->bp, regs->sp);
#endif
	put_cpu_var(pf_reason);
	BUG();
}

static void pre(struct kmmio_probe *p, struct pt_regs *regs,
						unsigned long addr)
{
	struct trap_reason *my_reason = &get_cpu_var(pf_reason);
	struct mmiotrace_rw *my_trace = &get_cpu_var(cpu_trace);
	const unsigned long instptr = instruction_pointer(regs);
	const enum reason_type type = get_ins_type(instptr);
	struct remap_trace *trace = p->private;

	/* it doesn't make sense to have more than one active trace per cpu */
	if (my_reason->active_traces)
		die_kmmio_nesting_error(regs, addr);
	else
		my_reason->active_traces++;

	my_reason->type = type;
	my_reason->addr = addr;
	my_reason->ip = instptr;

	my_trace->phys = addr - trace->probe.addr + trace->phys;
	my_trace->map_id = trace->id;

	/*
	 * Only record the program counter when requested.
	 * It may taint clean-room reverse engineering.
	 */
	if (trace_pc)
		my_trace->pc = instptr;
	else
		my_trace->pc = 0;

	/*
	 * XXX: the timestamp recorded will be *after* the tracing has been
	 * done, not at the time we hit the instruction. SMP implications
	 * on event ordering?
	 */

	switch (type) {
	case REG_READ:
		my_trace->opcode = MMIO_READ;
		my_trace->width = get_ins_mem_width(instptr);
		break;
	case REG_WRITE:
		my_trace->opcode = MMIO_WRITE;
		my_trace->width = get_ins_mem_width(instptr);
		my_trace->value = get_ins_reg_val(instptr, regs);
		break;
	case IMM_WRITE:
		my_trace->opcode = MMIO_WRITE;
		my_trace->width = get_ins_mem_width(instptr);
		my_trace->value = get_ins_imm_val(instptr);
		break;
	default:
		{
			unsigned char *ip = (unsigned char *)instptr;
			my_trace->opcode = MMIO_UNKNOWN_OP;
			my_trace->width = 0;
			my_trace->value = (*ip) << 16 | *(ip + 1) << 8 |
								*(ip + 2);
		}
	}
	put_cpu_var(cpu_trace);
	put_cpu_var(pf_reason);
}

static void post(struct kmmio_probe *p, unsigned long condition,
							struct pt_regs *regs)
{
	struct trap_reason *my_reason = &get_cpu_var(pf_reason);
	struct mmiotrace_rw *my_trace = &get_cpu_var(cpu_trace);

	/* this should always return the active_trace count to 0 */
	my_reason->active_traces--;
	if (my_reason->active_traces) {
		pr_emerg(NAME "unexpected post handler");
		BUG();
	}

	switch (my_reason->type) {
	case REG_READ:
		my_trace->value = get_ins_reg_val(my_reason->ip, regs);
		break;
	default:
		break;
	}

	mmio_trace_rw(my_trace);
	put_cpu_var(cpu_trace);
	put_cpu_var(pf_reason);
}

static void ioremap_trace_core(resource_size_t offset, unsigned long size,
							void __iomem *addr)
{
	static atomic_t next_id;
	struct remap_trace *trace = kmalloc(sizeof(*trace), GFP_KERNEL);
	/* These are page-unaligned. */
	struct mmiotrace_map map = {
		.phys = offset,
		.virt = (unsigned long)addr,
		.len = size,
		.opcode = MMIO_PROBE
	};

	if (!trace) {
		pr_err(NAME "kmalloc failed in ioremap\n");
		return;
	}

	*trace = (struct remap_trace) {
		.probe = {
			.addr = (unsigned long)addr,
			.len = size,
			.pre_handler = pre,
			.post_handler = post,
			.private = trace
		},
		.phys = offset,
		.id = atomic_inc_return(&next_id)
	};
	map.map_id = trace->id;

	spin_lock_irq(&trace_lock);
	if (!is_enabled())
		goto not_enabled;

	mmio_trace_mapping(&map);
	list_add_tail(&trace->list, &trace_list);
	if (!nommiotrace)
		register_kmmio_probe(&trace->probe);

not_enabled:
	spin_unlock_irq(&trace_lock);
}

void mmiotrace_ioremap(resource_size_t offset, unsigned long size,
						void __iomem *addr)
{
	if (!is_enabled()) /* recheck and proper locking in *_core() */
		return;

	pr_debug(NAME "ioremap_*(0x%llx, 0x%lx) = %p\n",
				(unsigned long long)offset, size, addr);
	if ((filter_offset) && (offset != filter_offset))
		return;
	ioremap_trace_core(offset, size, addr);
}

static void iounmap_trace_core(volatile void __iomem *addr)
{
	struct mmiotrace_map map = {
		.phys = 0,
		.virt = (unsigned long)addr,
		.len = 0,
		.opcode = MMIO_UNPROBE
	};
	struct remap_trace *trace;
	struct remap_trace *tmp;
	struct remap_trace *found_trace = NULL;

	pr_debug(NAME "Unmapping %p.\n", addr);

	spin_lock_irq(&trace_lock);
	if (!is_enabled())
		goto not_enabled;

	list_for_each_entry_safe(trace, tmp, &trace_list, list) {
		if ((unsigned long)addr == trace->probe.addr) {
			if (!nommiotrace)
				unregister_kmmio_probe(&trace->probe);
			list_del(&trace->list);
			found_trace = trace;
			break;
		}
	}
	map.map_id = (found_trace) ? found_trace->id : -1;
	mmio_trace_mapping(&map);

not_enabled:
	spin_unlock_irq(&trace_lock);
	if (found_trace) {
		synchronize_rcu(); /* unregister_kmmio_probe() requirement */
		kfree(found_trace);
	}
}

void mmiotrace_iounmap(volatile void __iomem *addr)
{
	might_sleep();
	if (is_enabled()) /* recheck and proper locking in *_core() */
		iounmap_trace_core(addr);
}

static void clear_trace_list(void)
{
	struct remap_trace *trace;
	struct remap_trace *tmp;

	/*
	 * No locking required, because the caller ensures we are in a
	 * critical section via mutex, and is_enabled() is false,
	 * i.e. nothing can traverse or modify this list.
	 * Caller also ensures is_enabled() cannot change.
	 */
	list_for_each_entry(trace, &trace_list, list) {
		pr_notice(NAME "purging non-iounmapped "
					"trace @0x%08lx, size 0x%lx.\n",
					trace->probe.addr, trace->probe.len);
		if (!nommiotrace)
			unregister_kmmio_probe(&trace->probe);
	}
	synchronize_rcu(); /* unregister_kmmio_probe() requirement */

	list_for_each_entry_safe(trace, tmp, &trace_list, list) {
		list_del(&trace->list);
		kfree(trace);
	}
}

#ifdef CONFIG_HOTPLUG_CPU
static cpumask_t downed_cpus;

static void enter_uniprocessor(void)
{
	int cpu;
	int err;

	get_online_cpus();
	downed_cpus = cpu_online_map;
	cpu_clear(first_cpu(cpu_online_map), downed_cpus);
	if (num_online_cpus() > 1)
		pr_notice(NAME "Disabling non-boot CPUs...\n");
	put_online_cpus();

	for_each_cpu_mask(cpu, downed_cpus) {
		err = cpu_down(cpu);
		if (!err)
			pr_info(NAME "CPU%d is down.\n", cpu);
		else
			pr_err(NAME "Error taking CPU%d down: %d\n", cpu, err);
	}
	if (num_online_cpus() > 1)
		pr_warning(NAME "multiple CPUs still online, "
						"may miss events.\n");
}

static void leave_uniprocessor(void)
{
	int cpu;
	int err;

	if (cpus_weight(downed_cpus) == 0)
		return;
	pr_notice(NAME "Re-enabling CPUs...\n");
	for_each_cpu_mask(cpu, downed_cpus) {
		err = cpu_up(cpu);
		if (!err)
			pr_info(NAME "enabled CPU%d.\n", cpu);
		else
			pr_err(NAME "cannot re-enable CPU%d: %d\n", cpu, err);
	}
}

#else /* !CONFIG_HOTPLUG_CPU */
static void enter_uniprocessor(void)
{
	if (num_online_cpus() > 1)
		pr_warning(NAME "multiple CPUs are online, may miss events. "
			"Suggest booting with maxcpus=1 kernel argument.\n");
}

static void leave_uniprocessor(void)
{
}
#endif

#if 0 /* XXX: out of order */
static struct file_operations fops_marker = {
	.owner =	THIS_MODULE,
	.write =	write_marker
};
#endif

void enable_mmiotrace(void)
{
	mutex_lock(&mmiotrace_mutex);
	if (is_enabled())
		goto out;

#if 0 /* XXX: tracing does not support text entries */
	marker_file = debugfs_create_file("marker", 0660, dir, NULL,
								&fops_marker);
	if (!marker_file)
		pr_err(NAME "marker file creation failed.\n");
#endif

	if (nommiotrace)
		pr_info(NAME "MMIO tracing disabled.\n");
	enter_uniprocessor();
	spin_lock_irq(&trace_lock);
	atomic_inc(&mmiotrace_enabled);
	spin_unlock_irq(&trace_lock);
	pr_info(NAME "enabled.\n");
out:
	mutex_unlock(&mmiotrace_mutex);
}

void disable_mmiotrace(void)
{
	mutex_lock(&mmiotrace_mutex);
	if (!is_enabled())
		goto out;

	spin_lock_irq(&trace_lock);
	atomic_dec(&mmiotrace_enabled);
	BUG_ON(is_enabled());
	spin_unlock_irq(&trace_lock);

	clear_trace_list(); /* guarantees: no more kmmio callbacks */
	leave_uniprocessor();
	if (marker_file) {
		debugfs_remove(marker_file);
		marker_file = NULL;
	}

	pr_info(NAME "disabled.\n");
out:
	mutex_unlock(&mmiotrace_mutex);
}
OpenPOWER on IntegriCloud