1// SPDX-License-Identifier: GPL-2.0
2/*
3 * KCSAN reporting.
4 *
5 * Copyright (C) 2019, Google LLC.
6 */
7
8#include <linux/debug_locks.h>
9#include <linux/delay.h>
10#include <linux/jiffies.h>
11#include <linux/kallsyms.h>
12#include <linux/kernel.h>
13#include <linux/lockdep.h>
14#include <linux/preempt.h>
15#include <linux/printk.h>
16#include <linux/sched.h>
17#include <linux/spinlock.h>
18#include <linux/stacktrace.h>
19
20#include "kcsan.h"
21#include "encoding.h"
22
23/*
24 * Max. number of stack entries to show in the report.
25 */
26#define NUM_STACK_ENTRIES 64
27
28/* Common access info. */
29struct access_info {
30 const volatile void *ptr;
31 size_t size;
32 int access_type;
33 int task_pid;
34 int cpu_id;
35 unsigned long ip;
36};
37
38/*
39 * Other thread info: communicated from other racing thread to thread that set
40 * up the watchpoint, which then prints the complete report atomically.
41 */
42struct other_info {
43 struct access_info ai;
44 unsigned long stack_entries[NUM_STACK_ENTRIES];
45 int num_stack_entries;
46
47 /*
48 * Optionally pass @current. Typically we do not need to pass @current
49 * via @other_info since just @task_pid is sufficient. Passing @current
50 * has additional overhead.
51 *
52 * To safely pass @current, we must either use get_task_struct/
53 * put_task_struct, or stall the thread that populated @other_info.
54 *
55 * We cannot rely on get_task_struct/put_task_struct in case
56 * release_report() races with a task being released, and would have to
57 * free it in release_report(). This may result in deadlock if we want
58 * to use KCSAN on the allocators.
59 *
60 * Since we also want to reliably print held locks for
61 * CONFIG_KCSAN_VERBOSE, the current implementation stalls the thread
62 * that populated @other_info until it has been consumed.
63 */
64 struct task_struct *task;
65};
66
67/*
68 * To never block any producers of struct other_info, we need as many elements
69 * as we have watchpoints (upper bound on concurrent races to report).
70 */
71static struct other_info other_infos[CONFIG_KCSAN_NUM_WATCHPOINTS + NUM_SLOTS-1];
72
73/*
74 * Information about reported races; used to rate limit reporting.
75 */
76struct report_time {
77 /*
78 * The last time the race was reported.
79 */
80 unsigned long time;
81
82 /*
83 * The frames of the 2 threads; if only 1 thread is known, one frame
84 * will be 0.
85 */
86 unsigned long frame1;
87 unsigned long frame2;
88};
89
90/*
91 * Since we also want to be able to debug allocators with KCSAN, to avoid
92 * deadlock, report_times cannot be dynamically resized with krealloc in
93 * rate_limit_report.
94 *
95 * Therefore, we use a fixed-size array, which at most will occupy a page. This
96 * still adequately rate limits reports, assuming that a) number of unique data
97 * races is not excessive, and b) occurrence of unique races within the
98 * same time window is limited.
99 */
100#define REPORT_TIMES_MAX (PAGE_SIZE / sizeof(struct report_time))
101#define REPORT_TIMES_SIZE \
102 (CONFIG_KCSAN_REPORT_ONCE_IN_MS > REPORT_TIMES_MAX ? \
103 REPORT_TIMES_MAX : \
104 CONFIG_KCSAN_REPORT_ONCE_IN_MS)
105static struct report_time report_times[REPORT_TIMES_SIZE];
106
107/*
108 * Spinlock serializing report generation, and access to @other_infos. Although
109 * it could make sense to have a finer-grained locking story for @other_infos,
110 * report generation needs to be serialized either way, so not much is gained.
111 */
112static DEFINE_RAW_SPINLOCK(report_lock);
113
114/*
115 * Checks if the race identified by thread frames frame1 and frame2 has
116 * been reported since (now - KCSAN_REPORT_ONCE_IN_MS).
117 */
118static bool rate_limit_report(unsigned long frame1, unsigned long frame2)
119{
120 struct report_time *use_entry = &report_times[0];
121 unsigned long invalid_before;
122 int i;
123
124 BUILD_BUG_ON(CONFIG_KCSAN_REPORT_ONCE_IN_MS != 0 && REPORT_TIMES_SIZE == 0);
125
126 if (CONFIG_KCSAN_REPORT_ONCE_IN_MS == 0)
127 return false;
128
129 invalid_before = jiffies - msecs_to_jiffies(CONFIG_KCSAN_REPORT_ONCE_IN_MS);
130
131 /* Check if a matching race report exists. */
132 for (i = 0; i < REPORT_TIMES_SIZE; ++i) {
133 struct report_time *rt = &report_times[i];
134
135 /*
136 * Must always select an entry for use to store info as we
137 * cannot resize report_times; at the end of the scan, use_entry
138 * will be the oldest entry, which ideally also happened before
139 * KCSAN_REPORT_ONCE_IN_MS ago.
140 */
141 if (time_before(rt->time, use_entry->time))
142 use_entry = rt;
143
144 /*
145 * Initially, no need to check any further as this entry as well
146 * as following entries have never been used.
147 */
148 if (rt->time == 0)
149 break;
150
151 /* Check if entry expired. */
152 if (time_before(rt->time, invalid_before))
153 continue; /* before KCSAN_REPORT_ONCE_IN_MS ago */
154
155 /* Reported recently, check if race matches. */
156 if ((rt->frame1 == frame1 && rt->frame2 == frame2) ||
157 (rt->frame1 == frame2 && rt->frame2 == frame1))
158 return true;
159 }
160
161 use_entry->time = jiffies;
162 use_entry->frame1 = frame1;
163 use_entry->frame2 = frame2;
164 return false;
165}
166
167/*
168 * Special rules to skip reporting.
169 */
170static bool
171skip_report(enum kcsan_value_change value_change, unsigned long top_frame)
172{
173 /* Should never get here if value_change==FALSE. */
174 WARN_ON_ONCE(value_change == KCSAN_VALUE_CHANGE_FALSE);
175
176 /*
177 * The first call to skip_report always has value_change==TRUE, since we
178 * cannot know the value written of an instrumented access. For the 2nd
179 * call there are 6 cases with CONFIG_KCSAN_REPORT_VALUE_CHANGE_ONLY:
180 *
181 * 1. read watchpoint, conflicting write (value_change==TRUE): report;
182 * 2. read watchpoint, conflicting write (value_change==MAYBE): skip;
183 * 3. write watchpoint, conflicting write (value_change==TRUE): report;
184 * 4. write watchpoint, conflicting write (value_change==MAYBE): skip;
185 * 5. write watchpoint, conflicting read (value_change==MAYBE): skip;
186 * 6. write watchpoint, conflicting read (value_change==TRUE): report;
187 *
188 * Cases 1-4 are intuitive and expected; case 5 ensures we do not report
189 * data races where the write may have rewritten the same value; case 6
190 * is possible either if the size is larger than what we check value
191 * changes for or the access type is KCSAN_ACCESS_ASSERT.
192 */
193 if (IS_ENABLED(CONFIG_KCSAN_REPORT_VALUE_CHANGE_ONLY) &&
194 value_change == KCSAN_VALUE_CHANGE_MAYBE) {
195 /*
196 * The access is a write, but the data value did not change.
197 *
198 * We opt-out of this filter for certain functions at request of
199 * maintainers.
200 */
201 char buf[64];
202 int len = scnprintf(buf, size: sizeof(buf), fmt: "%ps", (void *)top_frame);
203
204 if (!strnstr(buf, "rcu_", len) &&
205 !strnstr(buf, "_rcu", len) &&
206 !strnstr(buf, "_srcu", len))
207 return true;
208 }
209
210 return kcsan_skip_report_debugfs(func_addr: top_frame);
211}
212
213static const char *get_access_type(int type)
214{
215 if (type & KCSAN_ACCESS_ASSERT) {
216 if (type & KCSAN_ACCESS_SCOPED) {
217 if (type & KCSAN_ACCESS_WRITE)
218 return "assert no accesses (reordered)";
219 else
220 return "assert no writes (reordered)";
221 } else {
222 if (type & KCSAN_ACCESS_WRITE)
223 return "assert no accesses";
224 else
225 return "assert no writes";
226 }
227 }
228
229 switch (type) {
230 case 0:
231 return "read";
232 case KCSAN_ACCESS_ATOMIC:
233 return "read (marked)";
234 case KCSAN_ACCESS_WRITE:
235 return "write";
236 case KCSAN_ACCESS_WRITE | KCSAN_ACCESS_ATOMIC:
237 return "write (marked)";
238 case KCSAN_ACCESS_COMPOUND | KCSAN_ACCESS_WRITE:
239 return "read-write";
240 case KCSAN_ACCESS_COMPOUND | KCSAN_ACCESS_WRITE | KCSAN_ACCESS_ATOMIC:
241 return "read-write (marked)";
242 case KCSAN_ACCESS_SCOPED:
243 return "read (reordered)";
244 case KCSAN_ACCESS_SCOPED | KCSAN_ACCESS_ATOMIC:
245 return "read (marked, reordered)";
246 case KCSAN_ACCESS_SCOPED | KCSAN_ACCESS_WRITE:
247 return "write (reordered)";
248 case KCSAN_ACCESS_SCOPED | KCSAN_ACCESS_WRITE | KCSAN_ACCESS_ATOMIC:
249 return "write (marked, reordered)";
250 case KCSAN_ACCESS_SCOPED | KCSAN_ACCESS_COMPOUND | KCSAN_ACCESS_WRITE:
251 return "read-write (reordered)";
252 case KCSAN_ACCESS_SCOPED | KCSAN_ACCESS_COMPOUND | KCSAN_ACCESS_WRITE | KCSAN_ACCESS_ATOMIC:
253 return "read-write (marked, reordered)";
254 default:
255 BUG();
256 }
257}
258
259static const char *get_bug_type(int type)
260{
261 return (type & KCSAN_ACCESS_ASSERT) != 0 ? "assert: race" : "data-race";
262}
263
264/* Return thread description: in task or interrupt. */
265static const char *get_thread_desc(int task_id)
266{
267 if (task_id != -1) {
268 static char buf[32]; /* safe: protected by report_lock */
269
270 snprintf(buf, size: sizeof(buf), fmt: "task %i", task_id);
271 return buf;
272 }
273 return "interrupt";
274}
275
276/* Helper to skip KCSAN-related functions in stack-trace. */
277static int get_stack_skipnr(const unsigned long stack_entries[], int num_entries)
278{
279 char buf[64];
280 char *cur;
281 int len, skip;
282
283 for (skip = 0; skip < num_entries; ++skip) {
284 len = scnprintf(buf, size: sizeof(buf), fmt: "%ps", (void *)stack_entries[skip]);
285
286 /* Never show tsan_* or {read,write}_once_size. */
287 if (strnstr(buf, "tsan_", len) ||
288 strnstr(buf, "_once_size", len))
289 continue;
290
291 cur = strnstr(buf, "kcsan_", len);
292 if (cur) {
293 cur += strlen("kcsan_");
294 if (!str_has_prefix(str: cur, prefix: "test"))
295 continue; /* KCSAN runtime function. */
296 /* KCSAN related test. */
297 }
298
299 /*
300 * No match for runtime functions -- @skip entries to skip to
301 * get to first frame of interest.
302 */
303 break;
304 }
305
306 return skip;
307}
308
309/*
310 * Skips to the first entry that matches the function of @ip, and then replaces
311 * that entry with @ip, returning the entries to skip with @replaced containing
312 * the replaced entry.
313 */
314static int
315replace_stack_entry(unsigned long stack_entries[], int num_entries, unsigned long ip,
316 unsigned long *replaced)
317{
318 unsigned long symbolsize, offset;
319 unsigned long target_func;
320 int skip;
321
322 if (kallsyms_lookup_size_offset(addr: ip, symbolsize: &symbolsize, offset: &offset))
323 target_func = ip - offset;
324 else
325 goto fallback;
326
327 for (skip = 0; skip < num_entries; ++skip) {
328 unsigned long func = stack_entries[skip];
329
330 if (!kallsyms_lookup_size_offset(addr: func, symbolsize: &symbolsize, offset: &offset))
331 goto fallback;
332 func -= offset;
333
334 if (func == target_func) {
335 *replaced = stack_entries[skip];
336 stack_entries[skip] = ip;
337 return skip;
338 }
339 }
340
341fallback:
342 /* Should not happen; the resulting stack trace is likely misleading. */
343 WARN_ONCE(1, "Cannot find frame for %pS in stack trace", (void *)ip);
344 return get_stack_skipnr(stack_entries, num_entries);
345}
346
347static int
348sanitize_stack_entries(unsigned long stack_entries[], int num_entries, unsigned long ip,
349 unsigned long *replaced)
350{
351 return ip ? replace_stack_entry(stack_entries, num_entries, ip, replaced) :
352 get_stack_skipnr(stack_entries, num_entries);
353}
354
355/* Compares symbolized strings of addr1 and addr2. */
356static int sym_strcmp(void *addr1, void *addr2)
357{
358 char buf1[64];
359 char buf2[64];
360
361 snprintf(buf: buf1, size: sizeof(buf1), fmt: "%pS", addr1);
362 snprintf(buf: buf2, size: sizeof(buf2), fmt: "%pS", addr2);
363
364 return strncmp(buf1, buf2, sizeof(buf1));
365}
366
367static void
368print_stack_trace(unsigned long stack_entries[], int num_entries, unsigned long reordered_to)
369{
370 stack_trace_print(trace: stack_entries, nr_entries: num_entries, spaces: 0);
371 if (reordered_to)
372 pr_err(" |\n +-> reordered to: %pS\n", (void *)reordered_to);
373}
374
375static void print_verbose_info(struct task_struct *task)
376{
377 if (!task)
378 return;
379
380 /* Restore IRQ state trace for printing. */
381 kcsan_restore_irqtrace(task);
382
383 pr_err("\n");
384 debug_show_held_locks(task);
385 print_irqtrace_events(curr: task);
386}
387
388static void print_report(enum kcsan_value_change value_change,
389 const struct access_info *ai,
390 struct other_info *other_info,
391 u64 old, u64 new, u64 mask)
392{
393 unsigned long reordered_to = 0;
394 unsigned long stack_entries[NUM_STACK_ENTRIES] = { 0 };
395 int num_stack_entries = stack_trace_save(store: stack_entries, NUM_STACK_ENTRIES, skipnr: 1);
396 int skipnr = sanitize_stack_entries(stack_entries, num_entries: num_stack_entries, ip: ai->ip, replaced: &reordered_to);
397 unsigned long this_frame = stack_entries[skipnr];
398 unsigned long other_reordered_to = 0;
399 unsigned long other_frame = 0;
400 int other_skipnr = 0; /* silence uninit warnings */
401
402 /*
403 * Must check report filter rules before starting to print.
404 */
405 if (skip_report(value_change: KCSAN_VALUE_CHANGE_TRUE, top_frame: stack_entries[skipnr]))
406 return;
407
408 if (other_info) {
409 other_skipnr = sanitize_stack_entries(stack_entries: other_info->stack_entries,
410 num_entries: other_info->num_stack_entries,
411 ip: other_info->ai.ip, replaced: &other_reordered_to);
412 other_frame = other_info->stack_entries[other_skipnr];
413
414 /* @value_change is only known for the other thread */
415 if (skip_report(value_change, top_frame: other_frame))
416 return;
417 }
418
419 if (rate_limit_report(frame1: this_frame, frame2: other_frame))
420 return;
421
422 /* Print report header. */
423 pr_err("==================================================================\n");
424 if (other_info) {
425 int cmp;
426
427 /*
428 * Order functions lexographically for consistent bug titles.
429 * Do not print offset of functions to keep title short.
430 */
431 cmp = sym_strcmp(addr1: (void *)other_frame, addr2: (void *)this_frame);
432 pr_err("BUG: KCSAN: %s in %ps / %ps\n",
433 get_bug_type(ai->access_type | other_info->ai.access_type),
434 (void *)(cmp < 0 ? other_frame : this_frame),
435 (void *)(cmp < 0 ? this_frame : other_frame));
436 } else {
437 pr_err("BUG: KCSAN: %s in %pS\n", get_bug_type(ai->access_type),
438 (void *)this_frame);
439 }
440
441 pr_err("\n");
442
443 /* Print information about the racing accesses. */
444 if (other_info) {
445 pr_err("%s to 0x%px of %zu bytes by %s on cpu %i:\n",
446 get_access_type(other_info->ai.access_type), other_info->ai.ptr,
447 other_info->ai.size, get_thread_desc(other_info->ai.task_pid),
448 other_info->ai.cpu_id);
449
450 /* Print the other thread's stack trace. */
451 print_stack_trace(stack_entries: other_info->stack_entries + other_skipnr,
452 num_entries: other_info->num_stack_entries - other_skipnr,
453 reordered_to: other_reordered_to);
454 if (IS_ENABLED(CONFIG_KCSAN_VERBOSE))
455 print_verbose_info(task: other_info->task);
456
457 pr_err("\n");
458 pr_err("%s to 0x%px of %zu bytes by %s on cpu %i:\n",
459 get_access_type(ai->access_type), ai->ptr, ai->size,
460 get_thread_desc(ai->task_pid), ai->cpu_id);
461 } else {
462 pr_err("race at unknown origin, with %s to 0x%px of %zu bytes by %s on cpu %i:\n",
463 get_access_type(ai->access_type), ai->ptr, ai->size,
464 get_thread_desc(ai->task_pid), ai->cpu_id);
465 }
466 /* Print stack trace of this thread. */
467 print_stack_trace(stack_entries: stack_entries + skipnr, num_entries: num_stack_entries - skipnr, reordered_to);
468 if (IS_ENABLED(CONFIG_KCSAN_VERBOSE))
469 print_verbose_info(current);
470
471 /* Print observed value change. */
472 if (ai->size <= 8) {
473 int hex_len = ai->size * 2;
474 u64 diff = old ^ new;
475
476 if (mask)
477 diff &= mask;
478 if (diff) {
479 pr_err("\n");
480 pr_err("value changed: 0x%0*llx -> 0x%0*llx\n",
481 hex_len, old, hex_len, new);
482 if (mask) {
483 pr_err(" bits changed: 0x%0*llx with mask 0x%0*llx\n",
484 hex_len, diff, hex_len, mask);
485 }
486 }
487 }
488
489 /* Print report footer. */
490 pr_err("\n");
491 pr_err("Reported by Kernel Concurrency Sanitizer on:\n");
492 dump_stack_print_info(KERN_DEFAULT);
493 pr_err("==================================================================\n");
494
495 check_panic_on_warn(origin: "KCSAN");
496}
497
498static void release_report(unsigned long *flags, struct other_info *other_info)
499{
500 /*
501 * Use size to denote valid/invalid, since KCSAN entirely ignores
502 * 0-sized accesses.
503 */
504 other_info->ai.size = 0;
505 raw_spin_unlock_irqrestore(&report_lock, *flags);
506}
507
508/*
509 * Sets @other_info->task and awaits consumption of @other_info.
510 *
511 * Precondition: report_lock is held.
512 * Postcondition: report_lock is held.
513 */
514static void set_other_info_task_blocking(unsigned long *flags,
515 const struct access_info *ai,
516 struct other_info *other_info)
517{
518 /*
519 * We may be instrumenting a code-path where current->state is already
520 * something other than TASK_RUNNING.
521 */
522 const bool is_running = task_is_running(current);
523 /*
524 * To avoid deadlock in case we are in an interrupt here and this is a
525 * race with a task on the same CPU (KCSAN_INTERRUPT_WATCHER), provide a
526 * timeout to ensure this works in all contexts.
527 *
528 * Await approximately the worst case delay of the reporting thread (if
529 * we are not interrupted).
530 */
531 int timeout = max(kcsan_udelay_task, kcsan_udelay_interrupt);
532
533 other_info->task = current;
534 do {
535 if (is_running) {
536 /*
537 * Let lockdep know the real task is sleeping, to print
538 * the held locks (recall we turned lockdep off, so
539 * locking/unlocking @report_lock won't be recorded).
540 */
541 set_current_state(TASK_UNINTERRUPTIBLE);
542 }
543 raw_spin_unlock_irqrestore(&report_lock, *flags);
544 /*
545 * We cannot call schedule() since we also cannot reliably
546 * determine if sleeping here is permitted -- see in_atomic().
547 */
548
549 udelay(1);
550 raw_spin_lock_irqsave(&report_lock, *flags);
551 if (timeout-- < 0) {
552 /*
553 * Abort. Reset @other_info->task to NULL, since it
554 * appears the other thread is still going to consume
555 * it. It will result in no verbose info printed for
556 * this task.
557 */
558 other_info->task = NULL;
559 break;
560 }
561 /*
562 * If invalid, or @ptr nor @current matches, then @other_info
563 * has been consumed and we may continue. If not, retry.
564 */
565 } while (other_info->ai.size && other_info->ai.ptr == ai->ptr &&
566 other_info->task == current);
567 if (is_running)
568 set_current_state(TASK_RUNNING);
569}
570
571/* Populate @other_info; requires that the provided @other_info not in use. */
572static void prepare_report_producer(unsigned long *flags,
573 const struct access_info *ai,
574 struct other_info *other_info)
575{
576 raw_spin_lock_irqsave(&report_lock, *flags);
577
578 /*
579 * The same @other_infos entry cannot be used concurrently, because
580 * there is a one-to-one mapping to watchpoint slots (@watchpoints in
581 * core.c), and a watchpoint is only released for reuse after reporting
582 * is done by the consumer of @other_info. Therefore, it is impossible
583 * for another concurrent prepare_report_producer() to set the same
584 * @other_info, and are guaranteed exclusivity for the @other_infos
585 * entry pointed to by @other_info.
586 *
587 * To check this property holds, size should never be non-zero here,
588 * because every consumer of struct other_info resets size to 0 in
589 * release_report().
590 */
591 WARN_ON(other_info->ai.size);
592
593 other_info->ai = *ai;
594 other_info->num_stack_entries = stack_trace_save(store: other_info->stack_entries, NUM_STACK_ENTRIES, skipnr: 2);
595
596 if (IS_ENABLED(CONFIG_KCSAN_VERBOSE))
597 set_other_info_task_blocking(flags, ai, other_info);
598
599 raw_spin_unlock_irqrestore(&report_lock, *flags);
600}
601
602/* Awaits producer to fill @other_info and then returns. */
603static bool prepare_report_consumer(unsigned long *flags,
604 const struct access_info *ai,
605 struct other_info *other_info)
606{
607
608 raw_spin_lock_irqsave(&report_lock, *flags);
609 while (!other_info->ai.size) { /* Await valid @other_info. */
610 raw_spin_unlock_irqrestore(&report_lock, *flags);
611 cpu_relax();
612 raw_spin_lock_irqsave(&report_lock, *flags);
613 }
614
615 /* Should always have a matching access based on watchpoint encoding. */
616 if (WARN_ON(!matching_access((unsigned long)other_info->ai.ptr & WATCHPOINT_ADDR_MASK, other_info->ai.size,
617 (unsigned long)ai->ptr & WATCHPOINT_ADDR_MASK, ai->size)))
618 goto discard;
619
620 if (!matching_access(addr1: (unsigned long)other_info->ai.ptr, size1: other_info->ai.size,
621 addr2: (unsigned long)ai->ptr, size2: ai->size)) {
622 /*
623 * If the actual accesses to not match, this was a false
624 * positive due to watchpoint encoding.
625 */
626 atomic_long_inc(v: &kcsan_counters[KCSAN_COUNTER_ENCODING_FALSE_POSITIVES]);
627 goto discard;
628 }
629
630 return true;
631
632discard:
633 release_report(flags, other_info);
634 return false;
635}
636
637static struct access_info prepare_access_info(const volatile void *ptr, size_t size,
638 int access_type, unsigned long ip)
639{
640 return (struct access_info) {
641 .ptr = ptr,
642 .size = size,
643 .access_type = access_type,
644 .task_pid = in_task() ? task_pid_nr(current) : -1,
645 .cpu_id = raw_smp_processor_id(),
646 /* Only replace stack entry with @ip if scoped access. */
647 .ip = (access_type & KCSAN_ACCESS_SCOPED) ? ip : 0,
648 };
649}
650
651void kcsan_report_set_info(const volatile void *ptr, size_t size, int access_type,
652 unsigned long ip, int watchpoint_idx)
653{
654 const struct access_info ai = prepare_access_info(ptr, size, access_type, ip);
655 unsigned long flags;
656
657 kcsan_disable_current();
658 lockdep_off(); /* See kcsan_report_known_origin(). */
659
660 prepare_report_producer(flags: &flags, ai: &ai, other_info: &other_infos[watchpoint_idx]);
661
662 lockdep_on();
663 kcsan_enable_current();
664}
665
666void kcsan_report_known_origin(const volatile void *ptr, size_t size, int access_type,
667 unsigned long ip, enum kcsan_value_change value_change,
668 int watchpoint_idx, u64 old, u64 new, u64 mask)
669{
670 const struct access_info ai = prepare_access_info(ptr, size, access_type, ip);
671 struct other_info *other_info = &other_infos[watchpoint_idx];
672 unsigned long flags = 0;
673
674 kcsan_disable_current();
675 /*
676 * Because we may generate reports when we're in scheduler code, the use
677 * of printk() could deadlock. Until such time that all printing code
678 * called in print_report() is scheduler-safe, accept the risk, and just
679 * get our message out. As such, also disable lockdep to hide the
680 * warning, and avoid disabling lockdep for the rest of the kernel.
681 */
682 lockdep_off();
683
684 if (!prepare_report_consumer(flags: &flags, ai: &ai, other_info))
685 goto out;
686 /*
687 * Never report if value_change is FALSE, only when it is
688 * either TRUE or MAYBE. In case of MAYBE, further filtering may
689 * be done once we know the full stack trace in print_report().
690 */
691 if (value_change != KCSAN_VALUE_CHANGE_FALSE)
692 print_report(value_change, ai: &ai, other_info, old, new, mask);
693
694 release_report(flags: &flags, other_info);
695out:
696 lockdep_on();
697 kcsan_enable_current();
698}
699
700void kcsan_report_unknown_origin(const volatile void *ptr, size_t size, int access_type,
701 unsigned long ip, u64 old, u64 new, u64 mask)
702{
703 const struct access_info ai = prepare_access_info(ptr, size, access_type, ip);
704 unsigned long flags;
705
706 kcsan_disable_current();
707 lockdep_off(); /* See kcsan_report_known_origin(). */
708
709 raw_spin_lock_irqsave(&report_lock, flags);
710 print_report(value_change: KCSAN_VALUE_CHANGE_TRUE, ai: &ai, NULL, old, new, mask);
711 raw_spin_unlock_irqrestore(&report_lock, flags);
712
713 lockdep_on();
714 kcsan_enable_current();
715}
716

source code of linux/kernel/kcsan/report.c