-
Notifications
You must be signed in to change notification settings - Fork 25
/
debug.c
2147 lines (1831 loc) · 66.2 KB
/
debug.c
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
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
/*
* Copyright (c) 2000-2020 Apple Inc. All rights reserved.
*
* @APPLE_OSREFERENCE_LICENSE_HEADER_START@
*
* This file contains Original Code and/or Modifications of Original Code
* as defined in and that are subject to the Apple Public Source License
* Version 2.0 (the 'License'). You may not use this file except in
* compliance with the License. The rights granted to you under the License
* may not be used to create, or enable the creation or redistribution of,
* unlawful or unlicensed copies of an Apple operating system, or to
* circumvent, violate, or enable the circumvention or violation of, any
* terms of an Apple operating system software license agreement.
*
* Please obtain a copy of the License at
* http://www.opensource.apple.com/apsl/ and read it before using this file.
*
* The Original Code and all software distributed under the License are
* distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
* EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
* INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
* Please see the License for the specific language governing rights and
* limitations under the License.
*
* @APPLE_OSREFERENCE_LICENSE_HEADER_END@
*/
/*
* @OSF_COPYRIGHT@
*/
/*
* Mach Operating System
* Copyright (c) 1991,1990,1989 Carnegie Mellon University
* All Rights Reserved.
*
* Permission to use, copy, modify and distribute this software and its
* documentation is hereby granted, provided that both the copyright
* notice and this permission notice appear in all copies of the
* software, derivative works or modified versions, and any portions
* thereof, and that both notices appear in supporting documentation.
*
* CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS"
* CONDITION. CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND FOR
* ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE.
*
* Carnegie Mellon requests users of this software to return to
*
* Software Distribution Coordinator or Software.Distribution@CS.CMU.EDU
* School of Computer Science
* Carnegie Mellon University
* Pittsburgh PA 15213-3890
*
* any improvements or extensions that they make and grant Carnegie Mellon
* the rights to redistribute these changes.
*/
#include <mach_assert.h>
#include <mach_kdp.h>
#include <kdp/kdp.h>
#include <kdp/kdp_core.h>
#include <kdp/kdp_internal.h>
#include <kdp/kdp_callout.h>
#include <kern/cpu_number.h>
#include <kern/kalloc.h>
#include <kern/percpu.h>
#include <kern/spl.h>
#include <kern/thread.h>
#include <kern/assert.h>
#include <kern/sched_prim.h>
#include <kern/socd_client.h>
#include <kern/misc_protos.h>
#include <kern/clock.h>
#include <kern/telemetry.h>
#include <kern/ecc.h>
#include <kern/kern_cdata.h>
#include <kern/zalloc_internal.h>
#include <kern/iotrace.h>
#include <pexpert/device_tree.h>
#include <vm/vm_kern.h>
#include <vm/vm_map.h>
#include <vm/pmap.h>
#include <vm/vm_compressor.h>
#include <stdarg.h>
#include <stdatomic.h>
#include <sys/pgo.h>
#include <console/serial_protos.h>
#include <IOKit/IOBSD.h>
#if !(MACH_KDP && CONFIG_KDP_INTERACTIVE_DEBUGGING)
#include <kdp/kdp_udp.h>
#endif
#include <kern/processor.h>
#if defined(__i386__) || defined(__x86_64__)
#include <IOKit/IOBSD.h>
#include <i386/cpu_threads.h>
#include <i386/pmCPU.h>
#include <i386/lbr.h>
#endif
#include <IOKit/IOPlatformExpert.h>
#include <machine/machine_cpu.h>
#include <machine/pal_routines.h>
#include <sys/kdebug.h>
#include <libkern/OSKextLibPrivate.h>
#include <libkern/OSAtomic.h>
#include <libkern/kernel_mach_header.h>
#include <libkern/section_keywords.h>
#include <uuid/uuid.h>
#include <mach_debug/zone_info.h>
#include <mach/resource_monitors.h>
#include <machine/machine_routines.h>
#include <sys/proc_require.h>
#include <os/log_private.h>
#include <kern/ext_paniclog.h>
#if defined(__arm64__)
#include <pexpert/pexpert.h> /* For gPanicBase */
#include <arm/caches_internal.h>
#include <arm/misc_protos.h>
extern volatile struct xnu_hw_shmem_dbg_command_info *hwsd_info;
#endif
#include <san/kcov.h>
#if CONFIG_XNUPOST
#include <tests/xnupost.h>
extern int vsnprintf(char *, size_t, const char *, va_list);
#endif
#if CONFIG_CSR
#include <sys/csr.h>
#endif
extern int IODTGetLoaderInfo( const char *key, void **infoAddr, int *infosize );
extern void IODTFreeLoaderInfo( const char *key, void *infoAddr, int infoSize );
unsigned int halt_in_debugger = 0;
unsigned int current_debugger = 0;
unsigned int active_debugger = 0;
SECURITY_READ_ONLY_LATE(unsigned int) panicDebugging = FALSE;
unsigned int kernel_debugger_entry_count = 0;
#if DEVELOPMENT || DEBUG
unsigned int panic_test_failure_mode = PANIC_TEST_FAILURE_MODE_BADPTR;
unsigned int panic_test_action_count = 1;
unsigned int panic_test_case = PANIC_TEST_CASE_DISABLED;
#endif
#if defined(__arm64__)
struct additional_panic_data_buffer *panic_data_buffers = NULL;
#endif
#if defined(__arm64__)
/*
* Magic number; this should be identical to the armv7 encoding for trap.
*/
#define TRAP_DEBUGGER __asm__ volatile(".long 0xe7ffdeff")
#elif defined (__x86_64__)
#define TRAP_DEBUGGER __asm__("int3")
#else
#error No TRAP_DEBUGGER for this architecture
#endif
#if defined(__i386__) || defined(__x86_64__)
#define panic_stop() pmCPUHalt(PM_HALT_PANIC)
#else
#define panic_stop() panic_spin_forever()
#endif
#if defined(__arm64__) && (DEVELOPMENT || DEBUG)
/*
* More than enough for any typical format string passed to panic();
* anything longer will be truncated but that's better than nothing.
*/
#define EARLY_PANIC_BUFLEN 256
#endif
struct debugger_state {
uint64_t db_panic_options;
debugger_op db_current_op;
boolean_t db_proceed_on_sync_failure;
const char *db_message;
const char *db_panic_str;
va_list *db_panic_args;
void *db_panic_data_ptr;
unsigned long db_panic_caller;
/* incremented whenever we panic or call Debugger (current CPU panic level) */
uint32_t db_entry_count;
kern_return_t db_op_return;
};
static struct debugger_state PERCPU_DATA(debugger_state);
/* __pure2 is correct if this function is called with preemption disabled */
static inline __pure2 struct debugger_state *
current_debugger_state(void)
{
return PERCPU_GET(debugger_state);
}
#define CPUDEBUGGEROP current_debugger_state()->db_current_op
#define CPUDEBUGGERMSG current_debugger_state()->db_message
#define CPUPANICSTR current_debugger_state()->db_panic_str
#define CPUPANICARGS current_debugger_state()->db_panic_args
#define CPUPANICOPTS current_debugger_state()->db_panic_options
#define CPUPANICDATAPTR current_debugger_state()->db_panic_data_ptr
#define CPUDEBUGGERSYNC current_debugger_state()->db_proceed_on_sync_failure
#define CPUDEBUGGERCOUNT current_debugger_state()->db_entry_count
#define CPUDEBUGGERRET current_debugger_state()->db_op_return
#define CPUPANICCALLER current_debugger_state()->db_panic_caller
/*
* Usage:
* panic_test_action_count is in the context of other flags, e.g. for IO errors it is "succeed this many times then fail" and for nesting it is "panic this many times then succeed"
* panic_test_failure_mode is a bit map of things to do
* panic_test_case is what sort of test we are injecting
*
* For more details see definitions in debugger.h
*
* Note that not all combinations are sensible, but some actions can be combined, e.g.
* - BADPTR+SPIN with action count = 3 will cause panic->panic->spin
* - BADPTR with action count = 2 will cause 2 nested panics (in addition to the initial panic)
* - IO_ERR with action 15 will cause 14 successful IOs, then fail on the next one
*/
#if DEVELOPMENT || DEBUG
#define INJECT_NESTED_PANIC_IF_REQUESTED(requested) \
MACRO_BEGIN \
if ((panic_test_case & requested) && panic_test_action_count) { \
panic_test_action_count--; \
volatile int *panic_test_badpointer = (int *)4; \
if ((panic_test_failure_mode & PANIC_TEST_FAILURE_MODE_SPIN) && (!panic_test_action_count)) { printf("inject spin...\n"); while(panic_test_badpointer); } \
if ((panic_test_failure_mode & PANIC_TEST_FAILURE_MODE_BADPTR) && (panic_test_action_count+1)) { printf("inject badptr...\n"); *panic_test_badpointer = 0; } \
if ((panic_test_failure_mode & PANIC_TEST_FAILURE_MODE_PANIC) && (panic_test_action_count+1)) { printf("inject panic...\n"); panic("nested panic level %d", panic_test_action_count); } \
} \
MACRO_END
#endif /* DEVELOPMENT || DEBUG */
debugger_op debugger_current_op = DBOP_NONE;
const char *debugger_panic_str = NULL;
va_list *debugger_panic_args = NULL;
void *debugger_panic_data = NULL;
uint64_t debugger_panic_options = 0;
const char *debugger_message = NULL;
unsigned long debugger_panic_caller = 0;
void panic_trap_to_debugger(const char *panic_format_str, va_list *panic_args,
unsigned int reason, void *ctx, uint64_t panic_options_mask, void *panic_data,
unsigned long panic_caller) __dead2 __printflike(1, 0);
static void kdp_machine_reboot_type(unsigned int type, uint64_t debugger_flags);
void panic_spin_forever(void) __dead2;
extern kern_return_t do_stackshot(void);
extern void PE_panic_hook(const char*);
extern int sync(proc_t p, void *, void *);
#define NESTEDDEBUGGERENTRYMAX 5
static TUNABLE(unsigned int, max_debugger_entry_count, "nested_panic_max",
NESTEDDEBUGGERENTRYMAX);
SECURITY_READ_ONLY_LATE(bool) awl_scratch_reg_supported = false;
static bool PERCPU_DATA(hv_entry_detected); // = false
static void awl_set_scratch_reg_hv_bit(void);
void awl_mark_hv_entry(void);
static bool awl_pm_state_change_cbk(void *param, enum cpu_event event, unsigned int cpu_or_cluster);
#if defined(__arm64__)
#define DEBUG_BUF_SIZE (4096)
/* debug_buf is directly linked with iBoot panic region for arm targets */
char *debug_buf_base = NULL;
char *debug_buf_ptr = NULL;
unsigned int debug_buf_size = 0;
SECURITY_READ_ONLY_LATE(boolean_t) kdp_explicitly_requested = FALSE;
#else /* defined(__arm64__) */
#define DEBUG_BUF_SIZE ((3 * PAGE_SIZE) + offsetof(struct macos_panic_header, mph_data))
/* EXTENDED_DEBUG_BUF_SIZE definition is now in debug.h */
static_assert(((EXTENDED_DEBUG_BUF_SIZE % PANIC_FLUSH_BOUNDARY) == 0), "Extended debug buf size must match SMC alignment requirements");
char debug_buf[DEBUG_BUF_SIZE];
struct macos_panic_header *panic_info = (struct macos_panic_header *)debug_buf;
char *debug_buf_base = (debug_buf + offsetof(struct macos_panic_header, mph_data));
char *debug_buf_ptr = (debug_buf + offsetof(struct macos_panic_header, mph_data));
/*
* We don't include the size of the panic header in the length of the data we actually write.
* On co-processor platforms, we lose sizeof(struct macos_panic_header) bytes from the end of
* the end of the log because we only support writing (3*PAGESIZE) bytes.
*/
unsigned int debug_buf_size = (DEBUG_BUF_SIZE - offsetof(struct macos_panic_header, mph_data));
boolean_t extended_debug_log_enabled = FALSE;
#endif /* defined(__arm64__) */
#if defined(XNU_TARGET_OS_OSX)
#define KDBG_TRACE_PANIC_FILENAME "/var/tmp/panic.trace"
#else
#define KDBG_TRACE_PANIC_FILENAME "/var/log/panic.trace"
#endif
/* Debugger state */
atomic_int debugger_cpu = DEBUGGER_NO_CPU;
boolean_t debugger_allcpus_halted = FALSE;
boolean_t debugger_safe_to_return = TRUE;
unsigned int debugger_context = 0;
static char model_name[64];
unsigned char *kernel_uuid;
boolean_t kernelcache_uuid_valid = FALSE;
uuid_t kernelcache_uuid;
uuid_string_t kernelcache_uuid_string;
boolean_t pageablekc_uuid_valid = FALSE;
uuid_t pageablekc_uuid;
uuid_string_t pageablekc_uuid_string;
boolean_t auxkc_uuid_valid = FALSE;
uuid_t auxkc_uuid;
uuid_string_t auxkc_uuid_string;
/*
* By default we treat Debugger() the same as calls to panic(), unless
* we have debug boot-args present and the DB_KERN_DUMP_ON_NMI *NOT* set.
* If DB_KERN_DUMP_ON_NMI is *NOT* set, return from Debugger() is supported.
*
* Return from Debugger() is currently only implemented on x86
*/
static boolean_t debugger_is_panic = TRUE;
TUNABLE(unsigned int, debug_boot_arg, "debug", 0);
TUNABLE(int, verbose_panic_flow_logging, "verbose_panic_flow_logging", 0);
char kernel_uuid_string[37]; /* uuid_string_t */
char kernelcache_uuid_string[37]; /* uuid_string_t */
char panic_disk_error_description[512];
size_t panic_disk_error_description_size = sizeof(panic_disk_error_description);
extern unsigned int write_trace_on_panic;
int kext_assertions_enable =
#if DEBUG || DEVELOPMENT
TRUE;
#else
FALSE;
#endif
#if (DEVELOPMENT || DEBUG)
uint64_t xnu_platform_stall_value = PLATFORM_STALL_XNU_DISABLE;
#endif
/*
* Maintain the physically-contiguous carveouts for the carveout bootargs.
*/
TUNABLE_WRITEABLE(boolean_t, phys_carveout_core, "phys_carveout_core", 1);
TUNABLE(uint32_t, phys_carveout_mb, "phys_carveout_mb", 0);
SECURITY_READ_ONLY_LATE(vm_offset_t) phys_carveout = 0;
SECURITY_READ_ONLY_LATE(uintptr_t) phys_carveout_pa = 0;
SECURITY_READ_ONLY_LATE(size_t) phys_carveout_size = 0;
/*
* Returns whether kernel debugging is expected to be restricted
* on the device currently based on CSR or other platform restrictions.
*/
boolean_t
kernel_debugging_restricted(void)
{
#if XNU_TARGET_OS_OSX
#if CONFIG_CSR
if (csr_check(CSR_ALLOW_KERNEL_DEBUGGER) != 0) {
return TRUE;
}
#endif /* CONFIG_CSR */
return FALSE;
#else /* XNU_TARGET_OS_OSX */
return FALSE;
#endif /* XNU_TARGET_OS_OSX */
}
__startup_func
static void
panic_init(void)
{
unsigned long uuidlen = 0;
void *uuid;
uuid = getuuidfromheader(&_mh_execute_header, &uuidlen);
if ((uuid != NULL) && (uuidlen == sizeof(uuid_t))) {
kernel_uuid = uuid;
uuid_unparse_upper(*(uuid_t *)uuid, kernel_uuid_string);
}
/*
* Take the value of the debug boot-arg into account
*/
#if MACH_KDP
if (!kernel_debugging_restricted() && debug_boot_arg) {
if (debug_boot_arg & DB_HALT) {
halt_in_debugger = 1;
}
#if defined(__arm64__)
if (debug_boot_arg & DB_NMI) {
panicDebugging = TRUE;
}
#else
panicDebugging = TRUE;
#endif /* defined(__arm64__) */
}
#if defined(__arm64__)
char kdpname[80];
kdp_explicitly_requested = PE_parse_boot_argn("kdp_match_name", kdpname, sizeof(kdpname));
#endif /* defined(__arm64__) */
#endif /* MACH_KDP */
#if defined (__x86_64__)
/*
* By default we treat Debugger() the same as calls to panic(), unless
* we have debug boot-args present and the DB_KERN_DUMP_ON_NMI *NOT* set.
* If DB_KERN_DUMP_ON_NMI is *NOT* set, return from Debugger() is supported.
* This is because writing an on-device corefile is a destructive operation.
*
* Return from Debugger() is currently only implemented on x86
*/
if (PE_i_can_has_debugger(NULL) && !(debug_boot_arg & DB_KERN_DUMP_ON_NMI)) {
debugger_is_panic = FALSE;
}
#endif
}
STARTUP(TUNABLES, STARTUP_RANK_MIDDLE, panic_init);
#if defined (__x86_64__)
void
extended_debug_log_init(void)
{
assert(coprocessor_paniclog_flush);
/*
* Allocate an extended panic log buffer that has space for the panic
* stackshot at the end. Update the debug buf pointers appropriately
* to point at this new buffer.
*
* iBoot pre-initializes the panic region with the NULL character. We set this here
* so we can accurately calculate the CRC for the region without needing to flush the
* full region over SMC.
*/
char *new_debug_buf = kalloc_data(EXTENDED_DEBUG_BUF_SIZE, Z_WAITOK | Z_ZERO);
panic_info = (struct macos_panic_header *)new_debug_buf;
debug_buf_ptr = debug_buf_base = (new_debug_buf + offsetof(struct macos_panic_header, mph_data));
debug_buf_size = (EXTENDED_DEBUG_BUF_SIZE - offsetof(struct macos_panic_header, mph_data));
extended_debug_log_enabled = TRUE;
/*
* Insert a compiler barrier so we don't free the other panic stackshot buffer
* until after we've marked the new one as available
*/
__compiler_barrier();
kmem_free(kernel_map, panic_stackshot_buf, panic_stackshot_buf_len);
panic_stackshot_buf = 0;
panic_stackshot_buf_len = 0;
}
#endif /* defined (__x86_64__) */
void
debug_log_init(void)
{
#if defined(__arm64__)
if (!gPanicBase) {
printf("debug_log_init: Error!! gPanicBase is still not initialized\n");
return;
}
/* Shift debug buf start location and size by the length of the panic header */
debug_buf_base = (char *)gPanicBase + sizeof(struct embedded_panic_header);
debug_buf_ptr = debug_buf_base;
debug_buf_size = gPanicSize - sizeof(struct embedded_panic_header);
#if CONFIG_EXT_PANICLOG
ext_paniclog_init();
#endif
#else
kern_return_t kr = KERN_SUCCESS;
bzero(panic_info, DEBUG_BUF_SIZE);
assert(debug_buf_base != NULL);
assert(debug_buf_ptr != NULL);
assert(debug_buf_size != 0);
/*
* We allocate a buffer to store a panic time stackshot. If we later discover that this is a
* system that supports flushing a stackshot via an extended debug log (see above), we'll free this memory
* as it's not necessary on this platform. This information won't be available until the IOPlatform has come
* up.
*/
kr = kmem_alloc(kernel_map, &panic_stackshot_buf, PANIC_STACKSHOT_BUFSIZE,
KMA_DATA | KMA_ZERO, VM_KERN_MEMORY_DIAG);
assert(kr == KERN_SUCCESS);
if (kr == KERN_SUCCESS) {
panic_stackshot_buf_len = PANIC_STACKSHOT_BUFSIZE;
}
#endif
}
void
phys_carveout_init(void)
{
if (!PE_i_can_has_debugger(NULL)) {
return;
}
#if __arm__ || __arm64__
#if DEVELOPMENT || DEBUG
#endif /* DEVELOPMENT || DEBUG */
#endif /* __arm__ || __arm64__ */
struct carveout {
const char *name;
vm_offset_t *va;
uint32_t requested_size;
uintptr_t *pa;
size_t *allocated_size;
uint64_t present;
} carveouts[] = {
{
"phys_carveout",
&phys_carveout,
phys_carveout_mb,
&phys_carveout_pa,
&phys_carveout_size,
phys_carveout_mb != 0,
}
};
for (int i = 0; i < (sizeof(carveouts) / sizeof(struct carveout)); i++) {
if (carveouts[i].present) {
size_t temp_carveout_size = 0;
if (os_mul_overflow(carveouts[i].requested_size, 1024 * 1024, &temp_carveout_size)) {
panic("%s_mb size overflowed (%uMB)",
carveouts[i].name, carveouts[i].requested_size);
return;
}
kmem_alloc_contig(kernel_map, carveouts[i].va,
temp_carveout_size, PAGE_MASK, 0, 0,
KMA_NOFAIL | KMA_PERMANENT | KMA_NOPAGEWAIT | KMA_DATA,
VM_KERN_MEMORY_DIAG);
*carveouts[i].pa = kvtophys(*carveouts[i].va);
*carveouts[i].allocated_size = temp_carveout_size;
}
}
#if __arm64__ && (DEVELOPMENT || DEBUG)
/* likely panic_trace boot-arg is also set so check and enable tracing if necessary into new carveout */
PE_arm_debug_enable_trace(true);
#endif /* __arm64__ && (DEVELOPMENT || DEBUG) */
}
boolean_t
debug_is_in_phys_carveout(vm_map_offset_t va)
{
return phys_carveout_size && va >= phys_carveout &&
va < (phys_carveout + phys_carveout_size);
}
boolean_t
debug_can_coredump_phys_carveout(void)
{
return phys_carveout_core;
}
static void
DebuggerLock(void)
{
int my_cpu = cpu_number();
int debugger_exp_cpu = DEBUGGER_NO_CPU;
assert(ml_get_interrupts_enabled() == FALSE);
if (atomic_load(&debugger_cpu) == my_cpu) {
return;
}
while (!atomic_compare_exchange_strong(&debugger_cpu, &debugger_exp_cpu, my_cpu)) {
debugger_exp_cpu = DEBUGGER_NO_CPU;
}
return;
}
static void
DebuggerUnlock(void)
{
assert(atomic_load_explicit(&debugger_cpu, memory_order_relaxed) == cpu_number());
/*
* We don't do an atomic exchange here in case
* there's another CPU spinning to acquire the debugger_lock
* and we never get a chance to update it. We already have the
* lock so we can simply store DEBUGGER_NO_CPU and follow with
* a barrier.
*/
atomic_store(&debugger_cpu, DEBUGGER_NO_CPU);
OSMemoryBarrier();
return;
}
static kern_return_t
DebuggerHaltOtherCores(boolean_t proceed_on_failure, bool is_stackshot)
{
#if defined(__arm64__)
return DebuggerXCallEnter(proceed_on_failure, is_stackshot);
#else /* defined(__arm64__) */
#pragma unused(proceed_on_failure)
#pragma unused(is_stackshot)
mp_kdp_enter(proceed_on_failure);
return KERN_SUCCESS;
#endif
}
static void
DebuggerResumeOtherCores(void)
{
#if defined(__arm64__)
DebuggerXCallReturn();
#else /* defined(__arm64__) */
mp_kdp_exit();
#endif
}
__printflike(3, 0)
static void
DebuggerSaveState(debugger_op db_op, const char *db_message, const char *db_panic_str,
va_list *db_panic_args, uint64_t db_panic_options, void *db_panic_data_ptr,
boolean_t db_proceed_on_sync_failure, unsigned long db_panic_caller)
{
CPUDEBUGGEROP = db_op;
/*
* Note:
* if CPUDEBUGGERCOUNT == 1 then we are in the normal case - record the panic data
* if CPUDEBUGGERCOUNT > 1 and CPUPANICSTR == NULL then we are in a nested panic that happened before DebuggerSaveState was called, so store the nested panic data
* if CPUDEBUGGERCOUNT > 1 and CPUPANICSTR != NULL then we are in a nested panic that happened after DebuggerSaveState was called, so leave the original panic data
*
* TODO: is it safe to flatten this to if (CPUPANICSTR == NULL)?
*/
if (CPUDEBUGGERCOUNT == 1 || CPUPANICSTR == NULL) {
CPUDEBUGGERMSG = db_message;
CPUPANICSTR = db_panic_str;
CPUPANICARGS = db_panic_args;
CPUPANICDATAPTR = db_panic_data_ptr;
CPUPANICCALLER = db_panic_caller;
}
CPUDEBUGGERSYNC = db_proceed_on_sync_failure;
CPUDEBUGGERRET = KERN_SUCCESS;
/* Reset these on any nested panics */
// follow up in rdar://88497308 (nested panics should not clobber panic flags)
CPUPANICOPTS = db_panic_options;
return;
}
/*
* Save the requested debugger state/action into the current processor's
* percu state and trap to the debugger.
*/
kern_return_t
DebuggerTrapWithState(debugger_op db_op, const char *db_message, const char *db_panic_str,
va_list *db_panic_args, uint64_t db_panic_options, void *db_panic_data_ptr,
boolean_t db_proceed_on_sync_failure, unsigned long db_panic_caller)
{
kern_return_t ret;
#if defined(__arm64__) && (DEVELOPMENT || DEBUG)
if (!PE_arm_debug_and_trace_initialized()) {
/*
* In practice this can only happen if we panicked very early,
* when only the boot CPU is online and before it has finished
* initializing the debug and trace infrastructure. We're going
* to hang soon, so let's at least make sure the message passed
* to panic() is actually logged.
*/
char buf[EARLY_PANIC_BUFLEN];
vsnprintf(buf, EARLY_PANIC_BUFLEN, db_panic_str, *db_panic_args);
paniclog_append_noflush("%s\n", buf);
}
#endif
assert(ml_get_interrupts_enabled() == FALSE);
DebuggerSaveState(db_op, db_message, db_panic_str, db_panic_args,
db_panic_options, db_panic_data_ptr,
db_proceed_on_sync_failure, db_panic_caller);
/*
* On ARM this generates an uncategorized exception -> sleh code ->
* DebuggerCall -> kdp_trap -> handle_debugger_trap
* So that is how XNU ensures that only one core can panic.
* The rest of the cores are halted by IPI if possible; if that
* fails it will fall back to dbgwrap.
*/
TRAP_DEBUGGER;
ret = CPUDEBUGGERRET;
DebuggerSaveState(DBOP_NONE, NULL, NULL, NULL, 0, NULL, FALSE, 0);
return ret;
}
void __attribute__((noinline))
Assert(
const char *file,
int line,
const char *expression
)
{
#if CONFIG_NONFATAL_ASSERTS
static TUNABLE(bool, mach_assert, "assertions", true);
if (!mach_assert) {
kprintf("%s:%d non-fatal Assertion: %s", file, line, expression);
return;
}
#endif
panic_plain("%s:%d Assertion failed: %s", file, line, expression);
}
boolean_t
debug_is_current_cpu_in_panic_state(void)
{
return current_debugger_state()->db_entry_count > 0;
}
/*
* check if we are in a nested panic, report findings, take evasive action where necessary
*
* see also PE_update_panicheader_nestedpanic
*/
static void
check_and_handle_nested_panic(uint64_t panic_options_mask, unsigned long panic_caller, const char *db_panic_str, va_list *db_panic_args)
{
if ((CPUDEBUGGERCOUNT > 1) && (CPUDEBUGGERCOUNT < max_debugger_entry_count)) {
// Note: this is the first indication in the panic log or serial that we are off the rails...
//
// if we panic *before* the paniclog is finalized then this will end up in the ips report with a panic_caller addr that gives us a clue
// if we panic *after* the log is finalized then we will only see it in the serial log
//
paniclog_append_noflush("Nested panic detected - entry count: %d panic_caller: 0x%016lx\n", CPUDEBUGGERCOUNT, panic_caller);
paniclog_flush();
// print the *new* panic string to the console, we might not get it by other means...
// TODO: I tried to write this stuff to the paniclog, but the serial output gets corrupted and the panicstring in the ips file is <mysterious>
// rdar://87846117 (NestedPanic: output panic string to paniclog)
if (db_panic_str) {
printf("Nested panic string:\n");
#pragma clang diagnostic push
#pragma clang diagnostic ignored "-Wformat-nonliteral"
_doprnt(db_panic_str, db_panic_args, PE_kputc, 0);
#pragma clang diagnostic pop
printf("\n<end nested panic string>\n");
}
}
// Stage 1 bailout
//
// Try to complete the normal panic flow, i.e. try to make sure the callouts happen and we flush the paniclog. If this fails with another nested
// panic then we will land in Stage 2 below...
//
if (CPUDEBUGGERCOUNT == max_debugger_entry_count) {
uint32_t panic_details = 0;
// if this is a force-reset panic then capture a log and reboot immediately.
if (panic_options_mask & DEBUGGER_OPTION_PANICLOGANDREBOOT) {
panic_details |= kPanicDetailsForcePowerOff;
}
// normally the kPEPanicBegin is sent from debugger_collect_diagnostics(), but we might nested-panic before we get
// there. To be safe send another notification, the function called below will only send kPEPanicBegin if it has not yet been sent.
//
PEHaltRestartInternal(kPEPanicBegin, panic_details);
paniclog_append_noflush("Nested panic count exceeds limit %d, machine will reset or spin\n", max_debugger_entry_count);
PE_update_panicheader_nestedpanic();
paniclog_flush();
if (!panicDebugging) {
// note that this will also send kPEPanicEnd
kdp_machine_reboot_type(kPEPanicRestartCPU, panic_options_mask);
}
// prints to console
paniclog_append_noflush("\nNested panic stall. Stage 1 bailout. Please go to https://panic.apple.com to report this panic\n");
panic_spin_forever();
}
// Stage 2 bailout
//
// Things are severely hosed, we have nested to the point of bailout and then nested again during the bailout path. Try to issue
// a chipreset as quickly as possible, hopefully something in the panic log is salvageable, since we flushed it during Stage 1.
//
if (CPUDEBUGGERCOUNT == max_debugger_entry_count + 1) {
if (!panicDebugging) {
// note that:
// - this code path should be audited for prints, as that is a common cause of nested panics
// - this code path should take the fastest route to the actual reset, and not call any un-necessary code
kdp_machine_reboot_type(kPEPanicRestartCPU, panic_options_mask & DEBUGGER_OPTION_SKIP_PANICEND_CALLOUTS);
}
// prints to console, but another nested panic will land in Stage 3 where we simply spin, so that is sort of ok...
paniclog_append_noflush("\nIn Nested panic stall. Stage 2 bailout. Please go to https://panic.apple.com to report this panic\n");
panic_spin_forever();
}
// Stage 3 bailout
//
// We are done here, we were unable to reset the platform without another nested panic. Spin until the watchdog kicks in.
//
if (CPUDEBUGGERCOUNT > max_debugger_entry_count + 1) {
kdp_machine_reboot_type(kPEHangCPU, 0);
}
}
void
Debugger(const char *message)
{
DebuggerWithContext(0, NULL, message, DEBUGGER_OPTION_NONE, (unsigned long)(char *)__builtin_return_address(0));
}
/*
* Enter the Debugger
*
* This is similar to, but not the same as a panic
*
* Key differences:
* - we get here from a debugger entry action (e.g. NMI)
* - the system is resumable on x86 (in theory, however it is not clear if this is tested)
* - rdar://57738811 (xnu: support resume from debugger via KDP on arm devices)
*
*/
void
DebuggerWithContext(unsigned int reason, void *ctx, const char *message,
uint64_t debugger_options_mask, unsigned long debugger_caller)
{
spl_t previous_interrupts_state;
boolean_t old_doprnt_hide_pointers = doprnt_hide_pointers;
#if defined(__x86_64__) && (DEVELOPMENT || DEBUG)
read_lbr();
#endif
previous_interrupts_state = ml_set_interrupts_enabled(FALSE);
disable_preemption();
/* track depth of debugger/panic entry */
CPUDEBUGGERCOUNT++;
/* emit a tracepoint as early as possible in case of hang */
SOCD_TRACE_XNU(PANIC, PACK_2X32(VALUE(cpu_number()), VALUE(CPUDEBUGGERCOUNT)), VALUE(debugger_options_mask), ADDR(message), ADDR(debugger_caller));
/* do max nested panic/debugger check, this will report nesting to the console and spin forever if we exceed a limit */
check_and_handle_nested_panic(debugger_options_mask, debugger_caller, message, NULL);
/* Handle any necessary platform specific actions before we proceed */
PEInitiatePanic();
#if DEVELOPMENT || DEBUG
INJECT_NESTED_PANIC_IF_REQUESTED(PANIC_TEST_CASE_RECURPANIC_ENTRY);
#endif
PE_panic_hook(message);
doprnt_hide_pointers = FALSE;
if (ctx != NULL) {
DebuggerSaveState(DBOP_DEBUGGER, message,
NULL, NULL, debugger_options_mask, NULL, TRUE, 0);
handle_debugger_trap(reason, 0, 0, ctx);
DebuggerSaveState(DBOP_NONE, NULL, NULL,
NULL, 0, NULL, FALSE, 0);
} else {
DebuggerTrapWithState(DBOP_DEBUGGER, message,
NULL, NULL, debugger_options_mask, NULL, TRUE, 0);
}
/* resume from the debugger */
CPUDEBUGGERCOUNT--;
doprnt_hide_pointers = old_doprnt_hide_pointers;
enable_preemption();
ml_set_interrupts_enabled(previous_interrupts_state);
}
static struct kdp_callout {
struct kdp_callout * callout_next;
kdp_callout_fn_t callout_fn;
boolean_t callout_in_progress;
void * callout_arg;
} * kdp_callout_list = NULL;
/*
* Called from kernel context to register a kdp event callout.
*/
void
kdp_register_callout(kdp_callout_fn_t fn, void * arg)
{
struct kdp_callout * kcp;
struct kdp_callout * list_head;
kcp = zalloc_permanent_type(struct kdp_callout);
kcp->callout_fn = fn;
kcp->callout_arg = arg;
kcp->callout_in_progress = FALSE;
/* Lock-less list insertion using compare and exchange. */
do {
list_head = kdp_callout_list;
kcp->callout_next = list_head;
} while (!OSCompareAndSwapPtr(list_head, kcp, &kdp_callout_list));
}
static void
kdp_callouts(kdp_event_t event)
{
struct kdp_callout *kcp = kdp_callout_list;
while (kcp) {
if (!kcp->callout_in_progress) {
kcp->callout_in_progress = TRUE;
kcp->callout_fn(kcp->callout_arg, event);
kcp->callout_in_progress = FALSE;
}
kcp = kcp->callout_next;
}
}
#if defined(__arm64__)
/*
* Register an additional buffer with data to include in the panic log
*
* <rdar://problem/50137705> tracks supporting more than one buffer
*
* Note that producer_name and buf should never be de-allocated as we reference these during panic.
*/
void
register_additional_panic_data_buffer(const char *producer_name, void *buf, int len)
{
if (panic_data_buffers != NULL) {
panic("register_additional_panic_data_buffer called with buffer already registered");
}
if (producer_name == NULL || (strlen(producer_name) == 0)) {
panic("register_additional_panic_data_buffer called with invalid producer_name");
}
if (buf == NULL) {
panic("register_additional_panic_data_buffer called with invalid buffer pointer");
}
if ((len <= 0) || (len > ADDITIONAL_PANIC_DATA_BUFFER_MAX_LEN)) {
panic("register_additional_panic_data_buffer called with invalid length");
}
struct additional_panic_data_buffer *new_panic_data_buffer = zalloc_permanent_type(struct additional_panic_data_buffer);
new_panic_data_buffer->producer_name = producer_name;
new_panic_data_buffer->buf = buf;
new_panic_data_buffer->len = len;
if (!OSCompareAndSwapPtr(NULL, new_panic_data_buffer, &panic_data_buffers)) {
panic("register_additional_panic_data_buffer called with buffer already registered");
}
return;
}
#endif /* defined(__arm64__) */
/*
* An overview of the xnu panic path:
*
* Several panic wrappers (panic(), panic_with_options(), etc.) all funnel into panic_trap_to_debugger().
* panic_trap_to_debugger() sets the panic state in the current processor's debugger_state prior
* to trapping into the debugger. Once we trap to the debugger, we end up in handle_debugger_trap()
* which tries to acquire the panic lock by atomically swapping the current CPU number into debugger_cpu.
* debugger_cpu acts as a synchronization point, from which the winning CPU can halt the other cores and
* continue to debugger_collect_diagnostics() where we write the paniclog, corefile (if appropriate) and proceed
* according to the device's boot-args.