mono/mini/lldb.c

   1 /**
   2  * \file
   3  * Mono support for LLDB.
   4  *
   5  * Author:
   6  *   Zoltan Varga (vargaz@gmail.com)
   7  *
   8  * Copyright 2016 Xamarin, Inc (http://www.xamarin.com)
   9  */
  10
  11 #include "config.h"
  12 #include "mini.h"
  13 #include "lldb.h"
  14 #include "seq-points.h"
  15
  16 #include <mono/metadata/debug-internals.h>
  17 #include <mono/utils/mono-counters.h>
  18
  19 #if !defined(DISABLE_JIT) && !defined(DISABLE_LLDB)
  20
  21 typedef enum {
  22         ENTRY_CODE_REGION = 1,
  23         ENTRY_METHOD = 2,
  24         ENTRY_TRAMPOLINE = 3,
  25         ENTRY_UNLOAD_CODE_REGION = 4
  26 } EntryType;
  27
  28 /*
  29  * Need to make sure these structures have the same size and alignment on
  30  * all platforms.
  31  */
  32
  33 /* One data packet sent from the runtime to the debugger */
  34 typedef struct {
  35         /* Pointer to the next entry */
  36         guint64 next_addr;
  37         /* The type of data pointed to by ADDR */
  38         /* One of the ENTRY_ constants */
  39         guint32 type;
  40         /* Align */
  41         guint32 dummy;
  42         guint64 size;
  43         guint64 addr;
  44 } DebugEntry;
  45
  46 typedef struct
  47 {
  48         /* (MAJOR << 16) | MINOR */
  49         guint32 version;
  50         /* Align */
  51         guint32 dummy;
  52         DebugEntry *entry;
  53         /* List of all entries */
  54         /* Keep this as a pointer so accessing it is atomic */
  55         DebugEntry *all_entries;
  56         /* The current entry embedded here to reduce the amount of roundtrips */
  57         guint32 type;
  58         guint32 dummy2;
  59         guint64 size;
  60         guint64 addr;
  61 } JitDescriptor;
  62
  63 /*
  64  * Represents a memory region used for code.
  65  */
  66 typedef struct {
  67         /*
  68          * OBJFILE_MAGIC. This is needed to make it easier for lldb to
  69          * create object files from this packet.
  70          */
  71         char magic [32];
  72         guint64 start;
  73         guint32 size;
  74         int id;
  75 } CodeRegionEntry;
  76
  77 typedef struct {
  78         int id;
  79 } UnloadCodeRegionEntry;
  80
  81 /*
  82  * Represents a managed method
  83  */
  84 typedef struct {
  85         guint64 code;
  86         int id;
  87         /* The id of the codegen region which contains CODE */
  88         int region_id;
  89         int code_size;
  90         /* Align */
  91         guint32 dummy;
  92         /* Followed by variable size data */
  93 } MethodEntry;
  94
  95 /*
  96  * Represents a trampoline
  97  */
  98 typedef struct {
  99         guint64 code;
 100         int id;
 101         /* The id of the codegen region which contains CODE */
 102         int region_id;
 103         int code_size;
 104         /* Align */
 105         guint32 dummy;
 106         /* Followed by variable size data */
 107 } TrampolineEntry;
 108
 109 #define MAJOR_VERSION 1
 110 #define MINOR_VERSION 0
 111
 112 static const char* OBJFILE_MAGIC = { "MONO_JIT_OBJECT_FILE" };
 113
 114 JitDescriptor __mono_jit_debug_descriptor = { (MAJOR_VERSION << 16) | MINOR_VERSION };
 115
 116 static gboolean enabled;
 117 static int id_generator;
 118 static GHashTable *codegen_regions;
 119 static DebugEntry *last_entry;
 120 static mono_mutex_t mutex;
 121 static GHashTable *dyn_codegen_regions;
 122 static gint64 register_time;
 123 static int num_entries;
 124
 125 #define lldb_lock() mono_os_mutex_lock (&mutex)
 126 #define lldb_unlock() mono_os_mutex_unlock (&mutex)
 127
 128 G_BEGIN_DECLS
 129
 130 void MONO_NEVER_INLINE __mono_jit_debug_register_code (void);
 131
 132 /* The native debugger puts a breakpoint in this function. */
 133 void MONO_NEVER_INLINE
 134 __mono_jit_debug_register_code (void)
 135 {
 136         /* Make sure that even compilers that ignore __noinline__ don't inline this */
 137 #if defined(__GNUC__)
 138         asm ("");
 139 #endif
 140 }
 141
 142 G_END_DECLS
 143
 144 /*
 145  * Functions to encode protocol data
 146  */
 147
 148 typedef struct {
 149         guint8 *buf, *p, *end;
 150 } Buffer;
 151
 152 static inline void
 153 buffer_init (Buffer *buf, int size)
 154 {
 155         buf->buf = (guint8 *)g_malloc (size);
 156         buf->p = buf->buf;
 157         buf->end = buf->buf + size;
 158 }
 159
 160 static inline int
 161 buffer_len (Buffer *buf)
 162 {
 163         return buf->p - buf->buf;
 164 }
 165
 166 static inline void
 167 buffer_make_room (Buffer *buf, int size)
 168 {
 169         if (buf->end - buf->p < size) {
 170                 int new_size = buf->end - buf->buf + size + 32;
 171                 guint8 *p = (guint8 *)g_realloc (buf->buf, new_size);
 172                 size = buf->p - buf->buf;
 173                 buf->buf = p;
 174                 buf->p = p + size;
 175                 buf->end = buf->buf + new_size;
 176         }
 177 }
 178
 179 static inline void
 180 buffer_add_byte (Buffer *buf, guint8 val)
 181 {
 182         buffer_make_room (buf, 1);
 183         buf->p [0] = val;
 184         buf->p++;
 185 }
 186
 187 static inline void
 188 buffer_add_short (Buffer *buf, guint32 val)
 189 {
 190         buffer_make_room (buf, 2);
 191         buf->p [0] = (val >> 8) & 0xff;
 192         buf->p [1] = (val >> 0) & 0xff;
 193         buf->p += 2;
 194 }
 195
 196 static inline void
 197 buffer_add_int (Buffer *buf, guint32 val)
 198 {
 199         buffer_make_room (buf, 4);
 200         buf->p [0] = (val >> 24) & 0xff;
 201         buf->p [1] = (val >> 16) & 0xff;
 202         buf->p [2] = (val >> 8) & 0xff;
 203         buf->p [3] = (val >> 0) & 0xff;
 204         buf->p += 4;
 205 }
 206
 207 static inline void
 208 buffer_add_long (Buffer *buf, guint64 l)
 209 {
 210         buffer_add_int (buf, (l >> 32) & 0xffffffff);
 211         buffer_add_int (buf, (l >> 0) & 0xffffffff);
 212 }
 213
 214 static inline void
 215 buffer_add_id (Buffer *buf, int id)
 216 {
 217         buffer_add_int (buf, (guint64)id);
 218 }
 219
 220 static inline void
 221 buffer_add_data (Buffer *buf, guint8 *data, int len)
 222 {
 223         buffer_make_room (buf, len);
 224         memcpy (buf->p, data, len);
 225         buf->p += len;
 226 }
 227
 228 static inline void
 229 buffer_add_string (Buffer *buf, const char *str)
 230 {
 231         int len;
 232
 233         if (str == NULL) {
 234                 buffer_add_int (buf, 0);
 235         } else {
 236                 len = strlen (str);
 237                 buffer_add_int (buf, len);
 238                 buffer_add_data (buf, (guint8*)str, len);
 239         }
 240 }
 241
 242 static inline void
 243 buffer_add_buffer (Buffer *buf, Buffer *data)
 244 {
 245         buffer_add_data (buf, data->buf, buffer_len (data));
 246 }
 247
 248 static inline void
 249 buffer_free (Buffer *buf)
 250 {
 251         g_free (buf->buf);
 252 }
 253
 254 typedef struct {
 255         gpointer code;
 256         gpointer region_start;
 257         guint32 region_size;
 258         gboolean found;
 259 } UserData;
 260
 261 static int
 262 find_code_region (void *data, int csize, int size, void *user_data)
 263 {
 264         UserData *ud = (UserData*)user_data;
 265
 266         if ((char*)ud->code >= (char*)data && (char*)ud->code < (char*)data + csize) {
 267                 ud->region_start = data;
 268                 ud->region_size = csize;
 269                 ud->found = TRUE;
 270                 return 1;
 271         }
 272         return 0;
 273 }
 274
 275 static void
 276 add_entry (EntryType type, Buffer *buf)
 277 {
 278         DebugEntry *entry;
 279         guint8 *data;
 280         int size = buffer_len (buf);
 281
 282         data = g_malloc (size);
 283         memcpy (data, buf->buf, size);
 284
 285         entry = g_malloc0 (sizeof (DebugEntry));
 286         entry->type = type;
 287         entry->addr = (guint64)(gsize)data;
 288         entry->size = size;
 289
 290         mono_memory_barrier ();
 291
 292         lldb_lock ();
 293
 294         /* The debugger can read the list of entries asynchronously, so this has to be async safe */
 295         // FIXME: Make sure this is async safe
 296         if (last_entry) {
 297                 last_entry->next_addr = (guint64)(gsize) (entry);
 298                 last_entry = entry;
 299         } else {
 300                 last_entry = entry;
 301                 __mono_jit_debug_descriptor.all_entries = entry;
 302         }
 303
 304         __mono_jit_debug_descriptor.entry = entry;
 305
 306         __mono_jit_debug_descriptor.type = entry->type;
 307         __mono_jit_debug_descriptor.size = entry->size;
 308         __mono_jit_debug_descriptor.addr = entry->addr;
 309         mono_memory_barrier ();
 310
 311         gint64 start = mono_time_track_start ();
 312         __mono_jit_debug_register_code ();
 313         mono_time_track_end (&register_time, start);
 314         num_entries ++;
 315         //printf ("%lf %d %d\n", register_time, num_entries, entry->type);
 316
 317         lldb_unlock ();
 318 }
 319
 320 /*
 321  * register_codegen_region:
 322  *
 323  * Register a codegen region with the debugger if needed.
 324  * Return a region id.
 325  */
 326 static int
 327 register_codegen_region (gpointer region_start, int region_size, gboolean dynamic)
 328 {
 329         CodeRegionEntry *region_entry;
 330         int id;
 331         Buffer tmp_buf;
 332         Buffer *buf = &tmp_buf;
 333
 334         if (dynamic) {
 335                 lldb_lock ();
 336                 id = ++id_generator;
 337                 lldb_unlock ();
 338         } else {
 339                 lldb_lock ();
 340                 if (!codegen_regions)
 341                         codegen_regions = g_hash_table_new (NULL, NULL);
 342                 id = GPOINTER_TO_INT (g_hash_table_lookup (codegen_regions, region_start));
 343                 if (id) {
 344                         lldb_unlock ();
 345                         return id;
 346                 }
 347                 id = ++id_generator;
 348                 g_hash_table_insert (codegen_regions, region_start, GINT_TO_POINTER (id));
 349                 lldb_unlock ();
 350         }
 351
 352         buffer_init (buf, 128);
 353
 354         region_entry = (CodeRegionEntry*)buf->p;
 355         buf->p += sizeof (CodeRegionEntry);
 356         memset (region_entry, 0, sizeof (CodeRegionEntry));
 357         strcpy (region_entry->magic, OBJFILE_MAGIC);
 358         region_entry->id = id;
 359         region_entry->start = (gsize)region_start;
 360         region_entry->size = (gsize)region_size;
 361
 362         add_entry (ENTRY_CODE_REGION, buf);
 363         buffer_free (buf);
 364         return id;
 365 }
 366
 367 static void
 368 emit_unwind_info (GSList *unwind_ops, Buffer *buf)
 369 {
 370         int ret_reg;
 371         int nunwind_ops;
 372         GSList *l;
 373
 374         ret_reg = mono_unwind_get_dwarf_pc_reg ();
 375         g_assert (ret_reg < 256);
 376
 377         /* We use the unencoded version of the unwind info to make it easier to decode */
 378         nunwind_ops = 0;
 379         for (l = unwind_ops; l; l = l->next) {
 380                 MonoUnwindOp *op = (MonoUnwindOp*)l->data;
 381
 382                 /* lldb can't handle these */
 383                 if (op->op == DW_CFA_mono_advance_loc)
 384                         break;
 385                 nunwind_ops ++;
 386         }
 387
 388         buffer_add_byte (buf, ret_reg);
 389         buffer_add_int (buf, nunwind_ops);
 390         for (l = unwind_ops; l; l = l->next) {
 391                 MonoUnwindOp *op = (MonoUnwindOp*)l->data;
 392
 393                 if (op->op == DW_CFA_mono_advance_loc)
 394                         break;
 395                 buffer_add_int (buf, op->op);
 396                 buffer_add_int (buf, op->when);
 397                 int dreg;
 398 #if TARGET_X86
 399                 // LLDB doesn't see to use the switched esp/ebp
 400                 if (op->reg == X86_ESP)
 401                         dreg = X86_ESP;
 402                 else if (op->reg == X86_EBP)
 403                         dreg = X86_EBP;
 404                 else
 405                         dreg = mono_hw_reg_to_dwarf_reg (op->reg);
 406 #else
 407                 dreg = mono_hw_reg_to_dwarf_reg (op->reg);
 408 #endif
 409                 buffer_add_int (buf, dreg);
 410                 buffer_add_int (buf, op->val);
 411         }
 412 }
 413
 414 void
 415 mono_lldb_init (const char *options)
 416 {
 417         enabled = TRUE;
 418         mono_os_mutex_init_recursive (&mutex);
 419
 420         mono_counters_register ("Time spent in LLDB", MONO_COUNTER_JIT | MONO_COUNTER_LONG | MONO_COUNTER_TIME, &register_time);
 421 }
 422
 423 typedef struct
 424 {
 425         MonoSymSeqPoint sp;
 426         int native_offset;
 427 } FullSeqPoint;
 428
 429 static int
 430 compare_by_addr (const void *arg1, const void *arg2)
 431 {
 432         const FullSeqPoint *sp1 = (const FullSeqPoint *)arg1;
 433         const FullSeqPoint *sp2 = (const FullSeqPoint *)arg2;
 434
 435         return sp1->native_offset - sp2->native_offset;
 436 }
 437
 438 void
 439 mono_lldb_save_method_info (MonoCompile *cfg)
 440 {
 441         MethodEntry *entry;
 442         UserData udata;
 443         int region_id;
 444         Buffer tmpbuf;
 445         Buffer *buf = &tmpbuf;
 446         MonoDebugMethodInfo *minfo;
 447         int i, j, n_il_offsets;
 448         int *source_files;
 449         GPtrArray *source_file_list;
 450         MonoSymSeqPoint *sym_seq_points;
 451         FullSeqPoint *locs;
 452
 453         if (!enabled)
 454                 return;
 455
 456         /* Find the codegen region which contains the code */
 457         memset (&udata, 0, sizeof (udata));
 458         udata.code = cfg->native_code;
 459         if (cfg->method->dynamic) {
 460                 mono_code_manager_foreach (cfg->dynamic_info->code_mp, find_code_region, &udata);
 461                 g_assert (udata.found);
 462
 463                 region_id = register_codegen_region (udata.region_start, udata.region_size, TRUE);
 464
 465                 lldb_lock ();
 466                 if (!dyn_codegen_regions)
 467                         dyn_codegen_regions = g_hash_table_new (NULL, NULL);
 468                 g_hash_table_insert (dyn_codegen_regions, cfg->method, GINT_TO_POINTER (region_id));
 469                 lldb_unlock ();
 470         } else {
 471                 mono_domain_code_foreach (cfg->domain, find_code_region, &udata);
 472                 g_assert (udata.found);
 473
 474                 region_id = register_codegen_region (udata.region_start, udata.region_size, FALSE);
 475         }
 476
 477         buffer_init (buf, 256);
 478
 479         entry = (MethodEntry*)buf->p;
 480         buf->p += sizeof (MethodEntry);
 481         entry->id = ++id_generator;
 482         entry->region_id = region_id;
 483         entry->code = (gsize)cfg->native_code;
 484         entry->code_size = cfg->code_size;
 485
 486         emit_unwind_info (cfg->unwind_ops, buf);
 487
 488         char *s = mono_method_full_name (cfg->method, TRUE);
 489         buffer_add_string (buf, s);
 490         g_free (s);
 491
 492         minfo = mono_debug_lookup_method (cfg->method);
 493         MonoSeqPointInfo *seq_points = cfg->seq_point_info;
 494         if (minfo && seq_points) {
 495                 mono_debug_get_seq_points (minfo, NULL, &source_file_list, &source_files, &sym_seq_points, &n_il_offsets);
 496                 buffer_add_int (buf, source_file_list->len);
 497                 for (i = 0; i < source_file_list->len; ++i) {
 498                         MonoDebugSourceInfo *sinfo = (MonoDebugSourceInfo *)g_ptr_array_index (source_file_list, i);
 499                         buffer_add_string (buf, sinfo->source_file);
 500                         for (j = 0; j < 16; ++j)
 501                                 buffer_add_byte (buf, sinfo->hash [j]);
 502                 }
 503
 504                 // The sym seq points are ordered by il offset, need to order them by address
 505                 int skipped = 0;
 506                 locs = g_new0 (FullSeqPoint, n_il_offsets);
 507                 for (i = 0; i < n_il_offsets; ++i) {
 508                         locs [i].sp = sym_seq_points [i];
 509
 510                         // FIXME: O(n^2)
 511                         SeqPoint seq_point;
 512                         if (mono_seq_point_find_by_il_offset (seq_points, sym_seq_points [i].il_offset, &seq_point)) {
 513                                 locs [i].native_offset = seq_point.native_offset;
 514                         } else {
 515                                 locs [i].native_offset = 0xffffff;
 516                                 skipped ++;
 517                         }
 518                 }
 519                 qsort (locs, n_il_offsets, sizeof (FullSeqPoint), compare_by_addr);
 520
 521                 n_il_offsets -= skipped;
 522                 buffer_add_int (buf, n_il_offsets);
 523                 for (i = 0; i < n_il_offsets; ++i) {
 524                         MonoSymSeqPoint *sp = &locs [i].sp;
 525
 526                         //printf ("%s %x %d %d\n", cfg->method->name, locs [i].native_offset, sp->il_offset, sp->line);
 527                         buffer_add_int (buf, locs [i].native_offset);
 528                         buffer_add_int (buf, sp->il_offset);
 529                         buffer_add_int (buf, sp->line);
 530                         buffer_add_int (buf, source_files [i]);
 531                         buffer_add_int (buf, sp->column);
 532                         buffer_add_int (buf, sp->end_line);
 533                         buffer_add_int (buf, sp->end_column);
 534                 }
 535                 g_free (locs);
 536                 g_free (source_files);
 537                 g_free (sym_seq_points);
 538                 g_ptr_array_free (source_file_list, TRUE);
 539         } else {
 540                 buffer_add_int (buf, 0);
 541                 buffer_add_int (buf, 0);
 542         }
 543
 544         add_entry (ENTRY_METHOD, buf);
 545         buffer_free (buf);
 546 }
 547
 548 void
 549 mono_lldb_remove_method (MonoDomain *domain, MonoMethod *method, MonoJitDynamicMethodInfo *info)
 550 {
 551         int region_id;
 552         UnloadCodeRegionEntry *entry;
 553         Buffer tmpbuf;
 554         Buffer *buf = &tmpbuf;
 555
 556         if (!enabled)
 557                 return;
 558
 559         g_assert (method->dynamic);
 560
 561         lldb_lock ();
 562         region_id = GPOINTER_TO_INT (g_hash_table_lookup (dyn_codegen_regions, method));
 563         g_hash_table_remove (dyn_codegen_regions, method);
 564         lldb_unlock ();
 565
 566         buffer_init (buf, 256);
 567
 568         entry = (UnloadCodeRegionEntry*)buf->p;
 569         buf->p += sizeof (UnloadCodeRegionEntry);
 570         entry->id = region_id;
 571
 572         add_entry (ENTRY_UNLOAD_CODE_REGION, buf);
 573         buffer_free (buf);
 574
 575         /* The method is associated with the code region, so it doesn't have to be unloaded */
 576 }
 577
 578 void
 579 mono_lldb_save_trampoline_info (MonoTrampInfo *info)
 580 {
 581         TrampolineEntry *entry;
 582         UserData udata;
 583         int region_id;
 584         Buffer tmpbuf;
 585         Buffer *buf = &tmpbuf;
 586
 587         if (!enabled)
 588                 return;
 589
 590         /* Find the codegen region which contains the code */
 591         memset (&udata, 0, sizeof (udata));
 592         udata.code = info->code;
 593         mono_global_codeman_foreach (find_code_region, &udata);
 594         if (!udata.found)
 595                 mono_domain_code_foreach (mono_get_root_domain (), find_code_region, &udata);
 596         if (!udata.found)
 597                 /* Can happen with AOT */
 598                 return;
 599
 600         region_id = register_codegen_region (udata.region_start, udata.region_size, FALSE);
 601
 602         buffer_init (buf, 1024);
 603
 604         entry = (TrampolineEntry*)buf->p;
 605         buf->p += sizeof (TrampolineEntry);
 606         entry->id = ++id_generator;
 607         entry->region_id = region_id;
 608         entry->code = (gsize)info->code;
 609         entry->code_size = info->code_size;
 610
 611         emit_unwind_info (info->unwind_ops, buf);
 612
 613         buffer_add_string (buf, info->name);
 614
 615         add_entry (ENTRY_TRAMPOLINE, buf);
 616         buffer_free (buf);
 617 }
 618
 619 void
 620 mono_lldb_save_specific_trampoline_info (gpointer arg1, MonoTrampolineType tramp_type, MonoDomain *domain, gpointer code, guint32 code_len)
 621 {
 622         /*
 623          * Avoid emitting these for now,
 624          * they slow down execution too much, and they are
 625          * only needed during single stepping which doesn't
 626          * work anyway.
 627          */
 628 #if 0
 629         TrampolineEntry *entry;
 630         UserData udata;
 631         int region_id;
 632         Buffer tmpbuf;
 633         Buffer *buf = &tmpbuf;
 634
 635         if (!enabled)
 636                 return;
 637
 638         /* Find the codegen region which contains the code */
 639         memset (&udata, 0, sizeof (udata));
 640         udata.code = code;
 641         mono_global_codeman_foreach (find_code_region, &udata);
 642         if (!udata.found)
 643                 mono_domain_code_foreach (mono_get_root_domain (), find_code_region, &udata);
 644         g_assert (udata.found);
 645
 646         region_id = register_codegen_region (udata.region_start, udata.region_size, FALSE);
 647
 648         buffer_init (buf, 1024);
 649
 650         entry = (TrampolineEntry*)buf->p;
 651         buf->p += sizeof (TrampolineEntry);
 652         entry->id = ++id_generator;
 653         entry->region_id = region_id;
 654         entry->code = (gsize)code;
 655         entry->code_size = code_len;
 656
 657         GSList *unwind_ops = mono_unwind_get_cie_program ();
 658         emit_unwind_info (unwind_ops, buf);
 659
 660         buffer_add_string (buf, "");
 661
 662         add_entry (ENTRY_TRAMPOLINE, buf);
 663         buffer_free (buf);
 664 #endif
 665 }
 666
 667 /*
 668 DESIGN:
 669
 670 Communication:
 671 Similar to the gdb jit interface. The runtime communicates with a plugin running inside lldb.
 672 - The runtime allocates a data packet, points a symbol with a well known name at it.
 673 - It calls a dummy function with a well known name.
 674 - The plugin sets a breakpoint at this function, causing the runtime to be suspended.
 675 - The plugin reads the data pointed to by the other symbol and processes it.
 676
 677 The data packets are kept in a list, so lldb can read all of them after attaching.
 678 Lldb will associate an object file with each mono codegen region.
 679
 680 Packet design:
 681 - use a flat byte array so the whole data can be read in one operation.
 682 - use 64 bit ints for pointers.
 683 */
 684
 685 #else
 686
 687 void
 688 mono_lldb_init (const char *options)
 689 {
 690         g_error ("lldb support has been disabled at configure time.");
 691 }
 692
 693 void
 694 mono_lldb_save_method_info (MonoCompile *cfg)
 695 {
 696 }
 697
 698 void
 699 mono_lldb_save_trampoline_info (MonoTrampInfo *info)
 700 {
 701 }
 702
 703 void
 704 mono_lldb_remove_method (MonoDomain *domain, MonoMethod *method, MonoJitDynamicMethodInfo *info)
 705 {
 706 }
 707
 708 void
 709 mono_lldb_save_specific_trampoline_info (gpointer arg1, MonoTrampolineType tramp_type, MonoDomain *domain, gpointer code, guint32 code_len)
 710 {
 711 }
 712
 713 #endif