1 //===-- Disassembler.cpp ----------------------------------------*- C++ -*-===//
2 //
3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4 // See https://llvm.org/LICENSE.txt for license information.
5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6 //
7 //===----------------------------------------------------------------------===//
8 
9 #include "lldb/Core/Disassembler.h"
10 
11 #include "lldb/Core/AddressRange.h"
12 #include "lldb/Core/Debugger.h"
13 #include "lldb/Core/EmulateInstruction.h"
14 #include "lldb/Core/Mangled.h"
15 #include "lldb/Core/Module.h"
16 #include "lldb/Core/ModuleList.h"
17 #include "lldb/Core/PluginManager.h"
18 #include "lldb/Core/SourceManager.h"
19 #include "lldb/Host/FileSystem.h"
20 #include "lldb/Interpreter/OptionValue.h"
21 #include "lldb/Interpreter/OptionValueArray.h"
22 #include "lldb/Interpreter/OptionValueDictionary.h"
23 #include "lldb/Interpreter/OptionValueRegex.h"
24 #include "lldb/Interpreter/OptionValueString.h"
25 #include "lldb/Interpreter/OptionValueUInt64.h"
26 #include "lldb/Symbol/Function.h"
27 #include "lldb/Symbol/Symbol.h"
28 #include "lldb/Symbol/SymbolContext.h"
29 #include "lldb/Target/ExecutionContext.h"
30 #include "lldb/Target/SectionLoadList.h"
31 #include "lldb/Target/StackFrame.h"
32 #include "lldb/Target/Target.h"
33 #include "lldb/Target/Thread.h"
34 #include "lldb/Utility/DataBufferHeap.h"
35 #include "lldb/Utility/DataExtractor.h"
36 #include "lldb/Utility/RegularExpression.h"
37 #include "lldb/Utility/Status.h"
38 #include "lldb/Utility/Stream.h"
39 #include "lldb/Utility/StreamString.h"
40 #include "lldb/Utility/Timer.h"
41 #include "lldb/lldb-private-enumerations.h"
42 #include "lldb/lldb-private-interfaces.h"
43 #include "lldb/lldb-private-types.h"
44 #include "llvm/ADT/Triple.h"
45 #include "llvm/Support/Compiler.h"
46 
47 #include <cstdint>
48 #include <cstring>
49 #include <utility>
50 
51 #include <assert.h>
52 
53 #define DEFAULT_DISASM_BYTE_SIZE 32
54 
55 using namespace lldb;
56 using namespace lldb_private;
57 
58 DisassemblerSP Disassembler::FindPlugin(const ArchSpec &arch,
59                                         const char *flavor,
60                                         const char *plugin_name) {
61   static Timer::Category func_cat(LLVM_PRETTY_FUNCTION);
62   Timer scoped_timer(func_cat,
63                      "Disassembler::FindPlugin (arch = %s, plugin_name = %s)",
64                      arch.GetArchitectureName(), plugin_name);
65 
66   DisassemblerCreateInstance create_callback = nullptr;
67 
68   if (plugin_name) {
69     ConstString const_plugin_name(plugin_name);
70     create_callback = PluginManager::GetDisassemblerCreateCallbackForPluginName(
71         const_plugin_name);
72     if (create_callback) {
73       DisassemblerSP disassembler_sp(create_callback(arch, flavor));
74 
75       if (disassembler_sp)
76         return disassembler_sp;
77     }
78   } else {
79     for (uint32_t idx = 0;
80          (create_callback = PluginManager::GetDisassemblerCreateCallbackAtIndex(
81               idx)) != nullptr;
82          ++idx) {
83       DisassemblerSP disassembler_sp(create_callback(arch, flavor));
84 
85       if (disassembler_sp)
86         return disassembler_sp;
87     }
88   }
89   return DisassemblerSP();
90 }
91 
92 DisassemblerSP Disassembler::FindPluginForTarget(const TargetSP target_sp,
93                                                  const ArchSpec &arch,
94                                                  const char *flavor,
95                                                  const char *plugin_name) {
96   if (target_sp && flavor == nullptr) {
97     // FIXME - we don't have the mechanism in place to do per-architecture
98     // settings.  But since we know that for now we only support flavors on x86
99     // & x86_64,
100     if (arch.GetTriple().getArch() == llvm::Triple::x86 ||
101         arch.GetTriple().getArch() == llvm::Triple::x86_64)
102       flavor = target_sp->GetDisassemblyFlavor();
103   }
104   return FindPlugin(arch, flavor, plugin_name);
105 }
106 
107 static void ResolveAddress(const ExecutionContext &exe_ctx, const Address &addr,
108                            Address &resolved_addr) {
109   if (!addr.IsSectionOffset()) {
110     // If we weren't passed in a section offset address range, try and resolve
111     // it to something
112     Target *target = exe_ctx.GetTargetPtr();
113     if (target) {
114       bool is_resolved =
115           target->GetSectionLoadList().IsEmpty() ?
116               target->GetImages().ResolveFileAddress(addr.GetOffset(),
117                                                      resolved_addr) :
118               target->GetSectionLoadList().ResolveLoadAddress(addr.GetOffset(),
119                                                               resolved_addr);
120 
121       // We weren't able to resolve the address, just treat it as a raw address
122       if (is_resolved && resolved_addr.IsValid())
123         return;
124     }
125   }
126   resolved_addr = addr;
127 }
128 
129 size_t Disassembler::Disassemble(Debugger &debugger, const ArchSpec &arch,
130                                  const char *plugin_name, const char *flavor,
131                                  const ExecutionContext &exe_ctx,
132                                  SymbolContextList &sc_list,
133                                  uint32_t num_instructions,
134                                  bool mixed_source_and_assembly,
135                                  uint32_t num_mixed_context_lines,
136                                  uint32_t options, Stream &strm) {
137   size_t success_count = 0;
138   const size_t count = sc_list.GetSize();
139   SymbolContext sc;
140   AddressRange range;
141   const uint32_t scope =
142       eSymbolContextBlock | eSymbolContextFunction | eSymbolContextSymbol;
143   const bool use_inline_block_range = true;
144   for (size_t i = 0; i < count; ++i) {
145     if (!sc_list.GetContextAtIndex(i, sc))
146       break;
147     for (uint32_t range_idx = 0;
148          sc.GetAddressRange(scope, range_idx, use_inline_block_range, range);
149          ++range_idx) {
150       if (Disassemble(debugger, arch, plugin_name, flavor, exe_ctx, range,
151                       num_instructions, mixed_source_and_assembly,
152                       num_mixed_context_lines, options, strm)) {
153         ++success_count;
154         strm.EOL();
155       }
156     }
157   }
158   return success_count;
159 }
160 
161 bool Disassembler::Disassemble(
162     Debugger &debugger, const ArchSpec &arch, const char *plugin_name,
163     const char *flavor, const ExecutionContext &exe_ctx, ConstString name,
164     Module *module, uint32_t num_instructions, bool mixed_source_and_assembly,
165     uint32_t num_mixed_context_lines, uint32_t options, Stream &strm) {
166   // If no name is given there's nothing to disassemble.
167   if (!name)
168     return false;
169 
170   const bool include_symbols = true;
171   const bool include_inlines = true;
172 
173   // Find functions matching the given name.
174   SymbolContextList sc_list;
175   if (module) {
176     module->FindFunctions(name, nullptr, eFunctionNameTypeAuto, include_symbols,
177                           include_inlines, true, sc_list);
178   } else if (exe_ctx.GetTargetPtr()) {
179     exe_ctx.GetTargetPtr()->GetImages().FindFunctions(
180         name, eFunctionNameTypeAuto, include_symbols, include_inlines, false,
181         sc_list);
182   }
183 
184   // If no functions were found there's nothing to disassemble.
185   if (sc_list.IsEmpty())
186     return false;
187 
188   return Disassemble(debugger, arch, plugin_name, flavor, exe_ctx, sc_list,
189                      num_instructions, mixed_source_and_assembly,
190                      num_mixed_context_lines, options, strm);
191 }
192 
193 lldb::DisassemblerSP Disassembler::DisassembleRange(
194     const ArchSpec &arch, const char *plugin_name, const char *flavor,
195     const ExecutionContext &exe_ctx, const AddressRange &range,
196     bool prefer_file_cache) {
197   if (range.GetByteSize() <= 0)
198     return {};
199 
200   if (!range.GetBaseAddress().IsValid())
201     return {};
202 
203   lldb::DisassemblerSP disasm_sp = Disassembler::FindPluginForTarget(
204       exe_ctx.GetTargetSP(), arch, flavor, plugin_name);
205 
206   if (!disasm_sp)
207     return {};
208 
209   const size_t bytes_disassembled =
210       disasm_sp->ParseInstructions(&exe_ctx, range, nullptr, prefer_file_cache);
211   if (bytes_disassembled == 0)
212     return {};
213 
214   return disasm_sp;
215 }
216 
217 lldb::DisassemblerSP
218 Disassembler::DisassembleBytes(const ArchSpec &arch, const char *plugin_name,
219                                const char *flavor, const Address &start,
220                                const void *src, size_t src_len,
221                                uint32_t num_instructions, bool data_from_file) {
222   if (!src)
223     return {};
224 
225   lldb::DisassemblerSP disasm_sp =
226       Disassembler::FindPlugin(arch, flavor, plugin_name);
227 
228   if (!disasm_sp)
229     return {};
230 
231   DataExtractor data(src, src_len, arch.GetByteOrder(),
232                      arch.GetAddressByteSize());
233 
234   (void)disasm_sp->DecodeInstructions(start, data, 0, num_instructions, false,
235                                       data_from_file);
236   return disasm_sp;
237 }
238 
239 bool Disassembler::Disassemble(Debugger &debugger, const ArchSpec &arch,
240                                const char *plugin_name, const char *flavor,
241                                const ExecutionContext &exe_ctx,
242                                const AddressRange &disasm_range,
243                                uint32_t num_instructions,
244                                bool mixed_source_and_assembly,
245                                uint32_t num_mixed_context_lines,
246                                uint32_t options, Stream &strm) {
247   if (!disasm_range.GetByteSize())
248     return false;
249 
250   lldb::DisassemblerSP disasm_sp(Disassembler::FindPluginForTarget(
251       exe_ctx.GetTargetSP(), arch, flavor, plugin_name));
252 
253   if (!disasm_sp)
254     return false;
255 
256   AddressRange range;
257   ResolveAddress(exe_ctx, disasm_range.GetBaseAddress(),
258                  range.GetBaseAddress());
259   range.SetByteSize(disasm_range.GetByteSize());
260   const bool prefer_file_cache = false;
261   size_t bytes_disassembled =
262       disasm_sp->ParseInstructions(&exe_ctx, range, &strm, prefer_file_cache);
263   if (bytes_disassembled == 0)
264     return false;
265 
266   return PrintInstructions(disasm_sp.get(), debugger, arch, exe_ctx,
267                            num_instructions, mixed_source_and_assembly,
268                            num_mixed_context_lines, options, strm);
269 }
270 
271 bool Disassembler::Disassemble(Debugger &debugger, const ArchSpec &arch,
272                                const char *plugin_name, const char *flavor,
273                                const ExecutionContext &exe_ctx,
274                                const Address &start_address,
275                                uint32_t num_instructions,
276                                bool mixed_source_and_assembly,
277                                uint32_t num_mixed_context_lines,
278                                uint32_t options, Stream &strm) {
279   if (num_instructions == 0)
280     return false;
281 
282   lldb::DisassemblerSP disasm_sp(Disassembler::FindPluginForTarget(
283       exe_ctx.GetTargetSP(), arch, flavor, plugin_name));
284   if (!disasm_sp)
285     return false;
286 
287   Address addr;
288   ResolveAddress(exe_ctx, start_address, addr);
289 
290   const bool prefer_file_cache = false;
291   size_t bytes_disassembled = disasm_sp->ParseInstructions(
292       &exe_ctx, addr, num_instructions, prefer_file_cache);
293   if (bytes_disassembled == 0)
294     return false;
295 
296   return PrintInstructions(disasm_sp.get(), debugger, arch, exe_ctx,
297                            num_instructions, mixed_source_and_assembly,
298                            num_mixed_context_lines, options, strm);
299 }
300 
301 Disassembler::SourceLine
302 Disassembler::GetFunctionDeclLineEntry(const SymbolContext &sc) {
303   if (!sc.function)
304     return {};
305 
306   if (!sc.line_entry.IsValid())
307     return {};
308 
309   LineEntry prologue_end_line = sc.line_entry;
310   FileSpec func_decl_file;
311   uint32_t func_decl_line;
312   sc.function->GetStartLineSourceInfo(func_decl_file, func_decl_line);
313 
314   if (func_decl_file != prologue_end_line.file &&
315       func_decl_file != prologue_end_line.original_file)
316     return {};
317 
318   SourceLine decl_line;
319   decl_line.file = func_decl_file;
320   decl_line.line = func_decl_line;
321   // TODO: Do we care about column on these entries?  If so, we need to plumb
322   // that through GetStartLineSourceInfo.
323   decl_line.column = 0;
324   return decl_line;
325 }
326 
327 void Disassembler::AddLineToSourceLineTables(
328     SourceLine &line,
329     std::map<FileSpec, std::set<uint32_t>> &source_lines_seen) {
330   if (line.IsValid()) {
331     auto source_lines_seen_pos = source_lines_seen.find(line.file);
332     if (source_lines_seen_pos == source_lines_seen.end()) {
333       std::set<uint32_t> lines;
334       lines.insert(line.line);
335       source_lines_seen.emplace(line.file, lines);
336     } else {
337       source_lines_seen_pos->second.insert(line.line);
338     }
339   }
340 }
341 
342 bool Disassembler::ElideMixedSourceAndDisassemblyLine(
343     const ExecutionContext &exe_ctx, const SymbolContext &sc,
344     SourceLine &line) {
345 
346   // TODO: should we also check target.process.thread.step-avoid-libraries ?
347 
348   const RegularExpression *avoid_regex = nullptr;
349 
350   // Skip any line #0 entries - they are implementation details
351   if (line.line == 0)
352     return false;
353 
354   ThreadSP thread_sp = exe_ctx.GetThreadSP();
355   if (thread_sp) {
356     avoid_regex = thread_sp->GetSymbolsToAvoidRegexp();
357   } else {
358     TargetSP target_sp = exe_ctx.GetTargetSP();
359     if (target_sp) {
360       Status error;
361       OptionValueSP value_sp = target_sp->GetDebugger().GetPropertyValue(
362           &exe_ctx, "target.process.thread.step-avoid-regexp", false, error);
363       if (value_sp && value_sp->GetType() == OptionValue::eTypeRegex) {
364         OptionValueRegex *re = value_sp->GetAsRegex();
365         if (re) {
366           avoid_regex = re->GetCurrentValue();
367         }
368       }
369     }
370   }
371   if (avoid_regex && sc.symbol != nullptr) {
372     const char *function_name =
373         sc.GetFunctionName(Mangled::ePreferDemangledWithoutArguments)
374             .GetCString();
375     if (function_name && avoid_regex->Execute(function_name)) {
376       // skip this source line
377       return true;
378     }
379   }
380   // don't skip this source line
381   return false;
382 }
383 
384 bool Disassembler::PrintInstructions(Disassembler *disasm_ptr,
385                                      Debugger &debugger, const ArchSpec &arch,
386                                      const ExecutionContext &exe_ctx,
387                                      uint32_t num_instructions,
388                                      bool mixed_source_and_assembly,
389                                      uint32_t num_mixed_context_lines,
390                                      uint32_t options, Stream &strm) {
391   // We got some things disassembled...
392   size_t num_instructions_found = disasm_ptr->GetInstructionList().GetSize();
393 
394   if (num_instructions > 0 && num_instructions < num_instructions_found)
395     num_instructions_found = num_instructions;
396 
397   const uint32_t max_opcode_byte_size =
398       disasm_ptr->GetInstructionList().GetMaxOpcocdeByteSize();
399   SymbolContext sc;
400   SymbolContext prev_sc;
401   AddressRange current_source_line_range;
402   const Address *pc_addr_ptr = nullptr;
403   StackFrame *frame = exe_ctx.GetFramePtr();
404 
405   TargetSP target_sp(exe_ctx.GetTargetSP());
406   SourceManager &source_manager =
407       target_sp ? target_sp->GetSourceManager() : debugger.GetSourceManager();
408 
409   if (frame) {
410     pc_addr_ptr = &frame->GetFrameCodeAddress();
411   }
412   const uint32_t scope =
413       eSymbolContextLineEntry | eSymbolContextFunction | eSymbolContextSymbol;
414   const bool use_inline_block_range = false;
415 
416   const FormatEntity::Entry *disassembly_format = nullptr;
417   FormatEntity::Entry format;
418   if (exe_ctx.HasTargetScope()) {
419     disassembly_format =
420         exe_ctx.GetTargetRef().GetDebugger().GetDisassemblyFormat();
421   } else {
422     FormatEntity::Parse("${addr}: ", format);
423     disassembly_format = &format;
424   }
425 
426   // First pass: step through the list of instructions, find how long the
427   // initial addresses strings are, insert padding in the second pass so the
428   // opcodes all line up nicely.
429 
430   // Also build up the source line mapping if this is mixed source & assembly
431   // mode. Calculate the source line for each assembly instruction (eliding
432   // inlined functions which the user wants to skip).
433 
434   std::map<FileSpec, std::set<uint32_t>> source_lines_seen;
435   Symbol *previous_symbol = nullptr;
436 
437   size_t address_text_size = 0;
438   for (size_t i = 0; i < num_instructions_found; ++i) {
439     Instruction *inst =
440         disasm_ptr->GetInstructionList().GetInstructionAtIndex(i).get();
441     if (inst) {
442       const Address &addr = inst->GetAddress();
443       ModuleSP module_sp(addr.GetModule());
444       if (module_sp) {
445         const SymbolContextItem resolve_mask = eSymbolContextFunction |
446                                                eSymbolContextSymbol |
447                                                eSymbolContextLineEntry;
448         uint32_t resolved_mask =
449             module_sp->ResolveSymbolContextForAddress(addr, resolve_mask, sc);
450         if (resolved_mask) {
451           StreamString strmstr;
452           Debugger::FormatDisassemblerAddress(disassembly_format, &sc, nullptr,
453                                               &exe_ctx, &addr, strmstr);
454           size_t cur_line = strmstr.GetSizeOfLastLine();
455           if (cur_line > address_text_size)
456             address_text_size = cur_line;
457 
458           // Add entries to our "source_lines_seen" map+set which list which
459           // sources lines occur in this disassembly session.  We will print
460           // lines of context around a source line, but we don't want to print
461           // a source line that has a line table entry of its own - we'll leave
462           // that source line to be printed when it actually occurs in the
463           // disassembly.
464 
465           if (mixed_source_and_assembly && sc.line_entry.IsValid()) {
466             if (sc.symbol != previous_symbol) {
467               SourceLine decl_line = GetFunctionDeclLineEntry(sc);
468               if (!ElideMixedSourceAndDisassemblyLine(exe_ctx, sc, decl_line))
469                 AddLineToSourceLineTables(decl_line, source_lines_seen);
470             }
471             if (sc.line_entry.IsValid()) {
472               SourceLine this_line;
473               this_line.file = sc.line_entry.file;
474               this_line.line = sc.line_entry.line;
475               this_line.column = sc.line_entry.column;
476               if (!ElideMixedSourceAndDisassemblyLine(exe_ctx, sc, this_line))
477                 AddLineToSourceLineTables(this_line, source_lines_seen);
478             }
479           }
480         }
481         sc.Clear(false);
482       }
483     }
484   }
485 
486   previous_symbol = nullptr;
487   SourceLine previous_line;
488   for (size_t i = 0; i < num_instructions_found; ++i) {
489     Instruction *inst =
490         disasm_ptr->GetInstructionList().GetInstructionAtIndex(i).get();
491 
492     if (inst) {
493       const Address &addr = inst->GetAddress();
494       const bool inst_is_at_pc = pc_addr_ptr && addr == *pc_addr_ptr;
495       SourceLinesToDisplay source_lines_to_display;
496 
497       prev_sc = sc;
498 
499       ModuleSP module_sp(addr.GetModule());
500       if (module_sp) {
501         uint32_t resolved_mask = module_sp->ResolveSymbolContextForAddress(
502             addr, eSymbolContextEverything, sc);
503         if (resolved_mask) {
504           if (mixed_source_and_assembly) {
505 
506             // If we've started a new function (non-inlined), print all of the
507             // source lines from the function declaration until the first line
508             // table entry - typically the opening curly brace of the function.
509             if (previous_symbol != sc.symbol) {
510               // The default disassembly format puts an extra blank line
511               // between functions - so when we're displaying the source
512               // context for a function, we don't want to add a blank line
513               // after the source context or we'll end up with two of them.
514               if (previous_symbol != nullptr)
515                 source_lines_to_display.print_source_context_end_eol = false;
516 
517               previous_symbol = sc.symbol;
518               if (sc.function && sc.line_entry.IsValid()) {
519                 LineEntry prologue_end_line = sc.line_entry;
520                 if (!ElideMixedSourceAndDisassemblyLine(exe_ctx, sc,
521                                                         prologue_end_line)) {
522                   FileSpec func_decl_file;
523                   uint32_t func_decl_line;
524                   sc.function->GetStartLineSourceInfo(func_decl_file,
525                                                       func_decl_line);
526                   if (func_decl_file == prologue_end_line.file ||
527                       func_decl_file == prologue_end_line.original_file) {
528                     // Add all the lines between the function declaration and
529                     // the first non-prologue source line to the list of lines
530                     // to print.
531                     for (uint32_t lineno = func_decl_line;
532                          lineno <= prologue_end_line.line; lineno++) {
533                       SourceLine this_line;
534                       this_line.file = func_decl_file;
535                       this_line.line = lineno;
536                       source_lines_to_display.lines.push_back(this_line);
537                     }
538                     // Mark the last line as the "current" one.  Usually this
539                     // is the open curly brace.
540                     if (source_lines_to_display.lines.size() > 0)
541                       source_lines_to_display.current_source_line =
542                           source_lines_to_display.lines.size() - 1;
543                   }
544                 }
545               }
546               sc.GetAddressRange(scope, 0, use_inline_block_range,
547                                  current_source_line_range);
548             }
549 
550             // If we've left a previous source line's address range, print a
551             // new source line
552             if (!current_source_line_range.ContainsFileAddress(addr)) {
553               sc.GetAddressRange(scope, 0, use_inline_block_range,
554                                  current_source_line_range);
555 
556               if (sc != prev_sc && sc.comp_unit && sc.line_entry.IsValid()) {
557                 SourceLine this_line;
558                 this_line.file = sc.line_entry.file;
559                 this_line.line = sc.line_entry.line;
560 
561                 if (!ElideMixedSourceAndDisassemblyLine(exe_ctx, sc,
562                                                         this_line)) {
563                   // Only print this source line if it is different from the
564                   // last source line we printed.  There may have been inlined
565                   // functions between these lines that we elided, resulting in
566                   // the same line being printed twice in a row for a
567                   // contiguous block of assembly instructions.
568                   if (this_line != previous_line) {
569 
570                     std::vector<uint32_t> previous_lines;
571                     for (uint32_t i = 0;
572                          i < num_mixed_context_lines &&
573                          (this_line.line - num_mixed_context_lines) > 0;
574                          i++) {
575                       uint32_t line =
576                           this_line.line - num_mixed_context_lines + i;
577                       auto pos = source_lines_seen.find(this_line.file);
578                       if (pos != source_lines_seen.end()) {
579                         if (pos->second.count(line) == 1) {
580                           previous_lines.clear();
581                         } else {
582                           previous_lines.push_back(line);
583                         }
584                       }
585                     }
586                     for (size_t i = 0; i < previous_lines.size(); i++) {
587                       SourceLine previous_line;
588                       previous_line.file = this_line.file;
589                       previous_line.line = previous_lines[i];
590                       auto pos = source_lines_seen.find(previous_line.file);
591                       if (pos != source_lines_seen.end()) {
592                         pos->second.insert(previous_line.line);
593                       }
594                       source_lines_to_display.lines.push_back(previous_line);
595                     }
596 
597                     source_lines_to_display.lines.push_back(this_line);
598                     source_lines_to_display.current_source_line =
599                         source_lines_to_display.lines.size() - 1;
600 
601                     for (uint32_t i = 0; i < num_mixed_context_lines; i++) {
602                       SourceLine next_line;
603                       next_line.file = this_line.file;
604                       next_line.line = this_line.line + i + 1;
605                       auto pos = source_lines_seen.find(next_line.file);
606                       if (pos != source_lines_seen.end()) {
607                         if (pos->second.count(next_line.line) == 1)
608                           break;
609                         pos->second.insert(next_line.line);
610                       }
611                       source_lines_to_display.lines.push_back(next_line);
612                     }
613                   }
614                   previous_line = this_line;
615                 }
616               }
617             }
618           }
619         } else {
620           sc.Clear(true);
621         }
622       }
623 
624       if (source_lines_to_display.lines.size() > 0) {
625         strm.EOL();
626         for (size_t idx = 0; idx < source_lines_to_display.lines.size();
627              idx++) {
628           SourceLine ln = source_lines_to_display.lines[idx];
629           const char *line_highlight = "";
630           if (inst_is_at_pc && (options & eOptionMarkPCSourceLine)) {
631             line_highlight = "->";
632           } else if (idx == source_lines_to_display.current_source_line) {
633             line_highlight = "**";
634           }
635           source_manager.DisplaySourceLinesWithLineNumbers(
636               ln.file, ln.line, ln.column, 0, 0, line_highlight, &strm);
637         }
638         if (source_lines_to_display.print_source_context_end_eol)
639           strm.EOL();
640       }
641 
642       const bool show_bytes = (options & eOptionShowBytes) != 0;
643       inst->Dump(&strm, max_opcode_byte_size, true, show_bytes, &exe_ctx, &sc,
644                  &prev_sc, nullptr, address_text_size);
645       strm.EOL();
646     } else {
647       break;
648     }
649   }
650 
651   return true;
652 }
653 
654 bool Disassembler::Disassemble(Debugger &debugger, const ArchSpec &arch,
655                                const char *plugin_name, const char *flavor,
656                                const ExecutionContext &exe_ctx,
657                                uint32_t num_instructions,
658                                bool mixed_source_and_assembly,
659                                uint32_t num_mixed_context_lines,
660                                uint32_t options, Stream &strm) {
661   AddressRange range;
662   StackFrame *frame = exe_ctx.GetFramePtr();
663   if (frame) {
664     SymbolContext sc(
665         frame->GetSymbolContext(eSymbolContextFunction | eSymbolContextSymbol));
666     if (sc.function) {
667       range = sc.function->GetAddressRange();
668     } else if (sc.symbol && sc.symbol->ValueIsAddress()) {
669       range.GetBaseAddress() = sc.symbol->GetAddressRef();
670       range.SetByteSize(sc.symbol->GetByteSize());
671     } else {
672       range.GetBaseAddress() = frame->GetFrameCodeAddress();
673     }
674 
675     if (range.GetBaseAddress().IsValid() && range.GetByteSize() == 0)
676       range.SetByteSize(DEFAULT_DISASM_BYTE_SIZE);
677   }
678 
679   return Disassemble(debugger, arch, plugin_name, flavor, exe_ctx, range,
680                      num_instructions, mixed_source_and_assembly,
681                      num_mixed_context_lines, options, strm);
682 }
683 
684 Instruction::Instruction(const Address &address, AddressClass addr_class)
685     : m_address(address), m_address_class(addr_class), m_opcode(),
686       m_calculated_strings(false) {}
687 
688 Instruction::~Instruction() = default;
689 
690 AddressClass Instruction::GetAddressClass() {
691   if (m_address_class == AddressClass::eInvalid)
692     m_address_class = m_address.GetAddressClass();
693   return m_address_class;
694 }
695 
696 void Instruction::Dump(lldb_private::Stream *s, uint32_t max_opcode_byte_size,
697                        bool show_address, bool show_bytes,
698                        const ExecutionContext *exe_ctx,
699                        const SymbolContext *sym_ctx,
700                        const SymbolContext *prev_sym_ctx,
701                        const FormatEntity::Entry *disassembly_addr_format,
702                        size_t max_address_text_size) {
703   size_t opcode_column_width = 7;
704   const size_t operand_column_width = 25;
705 
706   CalculateMnemonicOperandsAndCommentIfNeeded(exe_ctx);
707 
708   StreamString ss;
709 
710   if (show_address) {
711     Debugger::FormatDisassemblerAddress(disassembly_addr_format, sym_ctx,
712                                         prev_sym_ctx, exe_ctx, &m_address, ss);
713     ss.FillLastLineToColumn(max_address_text_size, ' ');
714   }
715 
716   if (show_bytes) {
717     if (m_opcode.GetType() == Opcode::eTypeBytes) {
718       // x86_64 and i386 are the only ones that use bytes right now so pad out
719       // the byte dump to be able to always show 15 bytes (3 chars each) plus a
720       // space
721       if (max_opcode_byte_size > 0)
722         m_opcode.Dump(&ss, max_opcode_byte_size * 3 + 1);
723       else
724         m_opcode.Dump(&ss, 15 * 3 + 1);
725     } else {
726       // Else, we have ARM or MIPS which can show up to a uint32_t 0x00000000
727       // (10 spaces) plus two for padding...
728       if (max_opcode_byte_size > 0)
729         m_opcode.Dump(&ss, max_opcode_byte_size * 3 + 1);
730       else
731         m_opcode.Dump(&ss, 12);
732     }
733   }
734 
735   const size_t opcode_pos = ss.GetSizeOfLastLine();
736 
737   // The default opcode size of 7 characters is plenty for most architectures
738   // but some like arm can pull out the occasional vqrshrun.s16.  We won't get
739   // consistent column spacing in these cases, unfortunately.
740   if (m_opcode_name.length() >= opcode_column_width) {
741     opcode_column_width = m_opcode_name.length() + 1;
742   }
743 
744   ss.PutCString(m_opcode_name);
745   ss.FillLastLineToColumn(opcode_pos + opcode_column_width, ' ');
746   ss.PutCString(m_mnemonics);
747 
748   if (!m_comment.empty()) {
749     ss.FillLastLineToColumn(
750         opcode_pos + opcode_column_width + operand_column_width, ' ');
751     ss.PutCString(" ; ");
752     ss.PutCString(m_comment);
753   }
754   s->PutCString(ss.GetString());
755 }
756 
757 bool Instruction::DumpEmulation(const ArchSpec &arch) {
758   std::unique_ptr<EmulateInstruction> insn_emulator_up(
759       EmulateInstruction::FindPlugin(arch, eInstructionTypeAny, nullptr));
760   if (insn_emulator_up) {
761     insn_emulator_up->SetInstruction(GetOpcode(), GetAddress(), nullptr);
762     return insn_emulator_up->EvaluateInstruction(0);
763   }
764 
765   return false;
766 }
767 
768 bool Instruction::CanSetBreakpoint () {
769   return !HasDelaySlot();
770 }
771 
772 bool Instruction::HasDelaySlot() {
773   // Default is false.
774   return false;
775 }
776 
777 OptionValueSP Instruction::ReadArray(FILE *in_file, Stream *out_stream,
778                                      OptionValue::Type data_type) {
779   bool done = false;
780   char buffer[1024];
781 
782   auto option_value_sp = std::make_shared<OptionValueArray>(1u << data_type);
783 
784   int idx = 0;
785   while (!done) {
786     if (!fgets(buffer, 1023, in_file)) {
787       out_stream->Printf(
788           "Instruction::ReadArray:  Error reading file (fgets).\n");
789       option_value_sp.reset();
790       return option_value_sp;
791     }
792 
793     std::string line(buffer);
794 
795     size_t len = line.size();
796     if (line[len - 1] == '\n') {
797       line[len - 1] = '\0';
798       line.resize(len - 1);
799     }
800 
801     if ((line.size() == 1) && line[0] == ']') {
802       done = true;
803       line.clear();
804     }
805 
806     if (!line.empty()) {
807       std::string value;
808       static RegularExpression g_reg_exp(
809           llvm::StringRef("^[ \t]*([^ \t]+)[ \t]*$"));
810       llvm::SmallVector<llvm::StringRef, 2> matches;
811       if (g_reg_exp.Execute(line, &matches))
812         value = matches[1].str();
813       else
814         value = line;
815 
816       OptionValueSP data_value_sp;
817       switch (data_type) {
818       case OptionValue::eTypeUInt64:
819         data_value_sp = std::make_shared<OptionValueUInt64>(0, 0);
820         data_value_sp->SetValueFromString(value);
821         break;
822       // Other types can be added later as needed.
823       default:
824         data_value_sp = std::make_shared<OptionValueString>(value.c_str(), "");
825         break;
826       }
827 
828       option_value_sp->GetAsArray()->InsertValue(idx, data_value_sp);
829       ++idx;
830     }
831   }
832 
833   return option_value_sp;
834 }
835 
836 OptionValueSP Instruction::ReadDictionary(FILE *in_file, Stream *out_stream) {
837   bool done = false;
838   char buffer[1024];
839 
840   auto option_value_sp = std::make_shared<OptionValueDictionary>();
841   static ConstString encoding_key("data_encoding");
842   OptionValue::Type data_type = OptionValue::eTypeInvalid;
843 
844   while (!done) {
845     // Read the next line in the file
846     if (!fgets(buffer, 1023, in_file)) {
847       out_stream->Printf(
848           "Instruction::ReadDictionary: Error reading file (fgets).\n");
849       option_value_sp.reset();
850       return option_value_sp;
851     }
852 
853     // Check to see if the line contains the end-of-dictionary marker ("}")
854     std::string line(buffer);
855 
856     size_t len = line.size();
857     if (line[len - 1] == '\n') {
858       line[len - 1] = '\0';
859       line.resize(len - 1);
860     }
861 
862     if ((line.size() == 1) && (line[0] == '}')) {
863       done = true;
864       line.clear();
865     }
866 
867     // Try to find a key-value pair in the current line and add it to the
868     // dictionary.
869     if (!line.empty()) {
870       static RegularExpression g_reg_exp(llvm::StringRef(
871           "^[ \t]*([a-zA-Z_][a-zA-Z0-9_]*)[ \t]*=[ \t]*(.*)[ \t]*$"));
872 
873       llvm::SmallVector<llvm::StringRef, 3> matches;
874 
875       bool reg_exp_success = g_reg_exp.Execute(line, &matches);
876       std::string key;
877       std::string value;
878       if (reg_exp_success) {
879         key = matches[1].str();
880         value = matches[2].str();
881       } else {
882         out_stream->Printf("Instruction::ReadDictionary: Failure executing "
883                            "regular expression.\n");
884         option_value_sp.reset();
885         return option_value_sp;
886       }
887 
888       ConstString const_key(key.c_str());
889       // Check value to see if it's the start of an array or dictionary.
890 
891       lldb::OptionValueSP value_sp;
892       assert(value.empty() == false);
893       assert(key.empty() == false);
894 
895       if (value[0] == '{') {
896         assert(value.size() == 1);
897         // value is a dictionary
898         value_sp = ReadDictionary(in_file, out_stream);
899         if (!value_sp) {
900           option_value_sp.reset();
901           return option_value_sp;
902         }
903       } else if (value[0] == '[') {
904         assert(value.size() == 1);
905         // value is an array
906         value_sp = ReadArray(in_file, out_stream, data_type);
907         if (!value_sp) {
908           option_value_sp.reset();
909           return option_value_sp;
910         }
911         // We've used the data_type to read an array; re-set the type to
912         // Invalid
913         data_type = OptionValue::eTypeInvalid;
914       } else if ((value[0] == '0') && (value[1] == 'x')) {
915         value_sp = std::make_shared<OptionValueUInt64>(0, 0);
916         value_sp->SetValueFromString(value);
917       } else {
918         size_t len = value.size();
919         if ((value[0] == '"') && (value[len - 1] == '"'))
920           value = value.substr(1, len - 2);
921         value_sp = std::make_shared<OptionValueString>(value.c_str(), "");
922       }
923 
924       if (const_key == encoding_key) {
925         // A 'data_encoding=..." is NOT a normal key-value pair; it is meta-data
926         // indicating the
927         // data type of an upcoming array (usually the next bit of data to be
928         // read in).
929         if (strcmp(value.c_str(), "uint32_t") == 0)
930           data_type = OptionValue::eTypeUInt64;
931       } else
932         option_value_sp->GetAsDictionary()->SetValueForKey(const_key, value_sp,
933                                                            false);
934     }
935   }
936 
937   return option_value_sp;
938 }
939 
940 bool Instruction::TestEmulation(Stream *out_stream, const char *file_name) {
941   if (!out_stream)
942     return false;
943 
944   if (!file_name) {
945     out_stream->Printf("Instruction::TestEmulation:  Missing file_name.");
946     return false;
947   }
948   FILE *test_file = FileSystem::Instance().Fopen(file_name, "r");
949   if (!test_file) {
950     out_stream->Printf(
951         "Instruction::TestEmulation: Attempt to open test file failed.");
952     return false;
953   }
954 
955   char buffer[256];
956   if (!fgets(buffer, 255, test_file)) {
957     out_stream->Printf(
958         "Instruction::TestEmulation: Error reading first line of test file.\n");
959     fclose(test_file);
960     return false;
961   }
962 
963   if (strncmp(buffer, "InstructionEmulationState={", 27) != 0) {
964     out_stream->Printf("Instructin::TestEmulation: Test file does not contain "
965                        "emulation state dictionary\n");
966     fclose(test_file);
967     return false;
968   }
969 
970   // Read all the test information from the test file into an
971   // OptionValueDictionary.
972 
973   OptionValueSP data_dictionary_sp(ReadDictionary(test_file, out_stream));
974   if (!data_dictionary_sp) {
975     out_stream->Printf(
976         "Instruction::TestEmulation:  Error reading Dictionary Object.\n");
977     fclose(test_file);
978     return false;
979   }
980 
981   fclose(test_file);
982 
983   OptionValueDictionary *data_dictionary =
984       data_dictionary_sp->GetAsDictionary();
985   static ConstString description_key("assembly_string");
986   static ConstString triple_key("triple");
987 
988   OptionValueSP value_sp = data_dictionary->GetValueForKey(description_key);
989 
990   if (!value_sp) {
991     out_stream->Printf("Instruction::TestEmulation:  Test file does not "
992                        "contain description string.\n");
993     return false;
994   }
995 
996   SetDescription(value_sp->GetStringValue());
997 
998   value_sp = data_dictionary->GetValueForKey(triple_key);
999   if (!value_sp) {
1000     out_stream->Printf(
1001         "Instruction::TestEmulation: Test file does not contain triple.\n");
1002     return false;
1003   }
1004 
1005   ArchSpec arch;
1006   arch.SetTriple(llvm::Triple(value_sp->GetStringValue()));
1007 
1008   bool success = false;
1009   std::unique_ptr<EmulateInstruction> insn_emulator_up(
1010       EmulateInstruction::FindPlugin(arch, eInstructionTypeAny, nullptr));
1011   if (insn_emulator_up)
1012     success =
1013         insn_emulator_up->TestEmulation(out_stream, arch, data_dictionary);
1014 
1015   if (success)
1016     out_stream->Printf("Emulation test succeeded.");
1017   else
1018     out_stream->Printf("Emulation test failed.");
1019 
1020   return success;
1021 }
1022 
1023 bool Instruction::Emulate(
1024     const ArchSpec &arch, uint32_t evaluate_options, void *baton,
1025     EmulateInstruction::ReadMemoryCallback read_mem_callback,
1026     EmulateInstruction::WriteMemoryCallback write_mem_callback,
1027     EmulateInstruction::ReadRegisterCallback read_reg_callback,
1028     EmulateInstruction::WriteRegisterCallback write_reg_callback) {
1029   std::unique_ptr<EmulateInstruction> insn_emulator_up(
1030       EmulateInstruction::FindPlugin(arch, eInstructionTypeAny, nullptr));
1031   if (insn_emulator_up) {
1032     insn_emulator_up->SetBaton(baton);
1033     insn_emulator_up->SetCallbacks(read_mem_callback, write_mem_callback,
1034                                    read_reg_callback, write_reg_callback);
1035     insn_emulator_up->SetInstruction(GetOpcode(), GetAddress(), nullptr);
1036     return insn_emulator_up->EvaluateInstruction(evaluate_options);
1037   }
1038 
1039   return false;
1040 }
1041 
1042 uint32_t Instruction::GetData(DataExtractor &data) {
1043   return m_opcode.GetData(data);
1044 }
1045 
1046 InstructionList::InstructionList() : m_instructions() {}
1047 
1048 InstructionList::~InstructionList() = default;
1049 
1050 size_t InstructionList::GetSize() const { return m_instructions.size(); }
1051 
1052 uint32_t InstructionList::GetMaxOpcocdeByteSize() const {
1053   uint32_t max_inst_size = 0;
1054   collection::const_iterator pos, end;
1055   for (pos = m_instructions.begin(), end = m_instructions.end(); pos != end;
1056        ++pos) {
1057     uint32_t inst_size = (*pos)->GetOpcode().GetByteSize();
1058     if (max_inst_size < inst_size)
1059       max_inst_size = inst_size;
1060   }
1061   return max_inst_size;
1062 }
1063 
1064 InstructionSP InstructionList::GetInstructionAtIndex(size_t idx) const {
1065   InstructionSP inst_sp;
1066   if (idx < m_instructions.size())
1067     inst_sp = m_instructions[idx];
1068   return inst_sp;
1069 }
1070 
1071 void InstructionList::Dump(Stream *s, bool show_address, bool show_bytes,
1072                            const ExecutionContext *exe_ctx) {
1073   const uint32_t max_opcode_byte_size = GetMaxOpcocdeByteSize();
1074   collection::const_iterator pos, begin, end;
1075 
1076   const FormatEntity::Entry *disassembly_format = nullptr;
1077   FormatEntity::Entry format;
1078   if (exe_ctx && exe_ctx->HasTargetScope()) {
1079     disassembly_format =
1080         exe_ctx->GetTargetRef().GetDebugger().GetDisassemblyFormat();
1081   } else {
1082     FormatEntity::Parse("${addr}: ", format);
1083     disassembly_format = &format;
1084   }
1085 
1086   for (begin = m_instructions.begin(), end = m_instructions.end(), pos = begin;
1087        pos != end; ++pos) {
1088     if (pos != begin)
1089       s->EOL();
1090     (*pos)->Dump(s, max_opcode_byte_size, show_address, show_bytes, exe_ctx,
1091                  nullptr, nullptr, disassembly_format, 0);
1092   }
1093 }
1094 
1095 void InstructionList::Clear() { m_instructions.clear(); }
1096 
1097 void InstructionList::Append(lldb::InstructionSP &inst_sp) {
1098   if (inst_sp)
1099     m_instructions.push_back(inst_sp);
1100 }
1101 
1102 uint32_t
1103 InstructionList::GetIndexOfNextBranchInstruction(uint32_t start,
1104                                                  Target &target,
1105                                                  bool ignore_calls) const {
1106   size_t num_instructions = m_instructions.size();
1107 
1108   uint32_t next_branch = UINT32_MAX;
1109   size_t i;
1110   for (i = start; i < num_instructions; i++) {
1111     if (m_instructions[i]->DoesBranch()) {
1112       if (ignore_calls && m_instructions[i]->IsCall())
1113         continue;
1114       next_branch = i;
1115       break;
1116     }
1117   }
1118 
1119   // Hexagon needs the first instruction of the packet with the branch. Go
1120   // backwards until we find an instruction marked end-of-packet, or until we
1121   // hit start.
1122   if (target.GetArchitecture().GetTriple().getArch() == llvm::Triple::hexagon) {
1123     // If we didn't find a branch, find the last packet start.
1124     if (next_branch == UINT32_MAX) {
1125       i = num_instructions - 1;
1126     }
1127 
1128     while (i > start) {
1129       --i;
1130 
1131       Status error;
1132       uint32_t inst_bytes;
1133       bool prefer_file_cache = false; // Read from process if process is running
1134       lldb::addr_t load_addr = LLDB_INVALID_ADDRESS;
1135       target.ReadMemory(m_instructions[i]->GetAddress(), prefer_file_cache,
1136                         &inst_bytes, sizeof(inst_bytes), error, &load_addr);
1137       // If we have an error reading memory, return start
1138       if (!error.Success())
1139         return start;
1140       // check if this is the last instruction in a packet bits 15:14 will be
1141       // 11b or 00b for a duplex
1142       if (((inst_bytes & 0xC000) == 0xC000) ||
1143           ((inst_bytes & 0xC000) == 0x0000)) {
1144         // instruction after this should be the start of next packet
1145         next_branch = i + 1;
1146         break;
1147       }
1148     }
1149 
1150     if (next_branch == UINT32_MAX) {
1151       // We couldn't find the previous packet, so return start
1152       next_branch = start;
1153     }
1154   }
1155   return next_branch;
1156 }
1157 
1158 uint32_t
1159 InstructionList::GetIndexOfInstructionAtAddress(const Address &address) {
1160   size_t num_instructions = m_instructions.size();
1161   uint32_t index = UINT32_MAX;
1162   for (size_t i = 0; i < num_instructions; i++) {
1163     if (m_instructions[i]->GetAddress() == address) {
1164       index = i;
1165       break;
1166     }
1167   }
1168   return index;
1169 }
1170 
1171 uint32_t
1172 InstructionList::GetIndexOfInstructionAtLoadAddress(lldb::addr_t load_addr,
1173                                                     Target &target) {
1174   Address address;
1175   address.SetLoadAddress(load_addr, &target);
1176   return GetIndexOfInstructionAtAddress(address);
1177 }
1178 
1179 size_t Disassembler::ParseInstructions(const ExecutionContext *exe_ctx,
1180                                        const AddressRange &range,
1181                                        Stream *error_strm_ptr,
1182                                        bool prefer_file_cache) {
1183   if (exe_ctx) {
1184     Target *target = exe_ctx->GetTargetPtr();
1185     const addr_t byte_size = range.GetByteSize();
1186     if (target == nullptr || byte_size == 0 ||
1187         !range.GetBaseAddress().IsValid())
1188       return 0;
1189 
1190     auto data_sp = std::make_shared<DataBufferHeap>(byte_size, '\0');
1191 
1192     Status error;
1193     lldb::addr_t load_addr = LLDB_INVALID_ADDRESS;
1194     const size_t bytes_read = target->ReadMemory(
1195         range.GetBaseAddress(), prefer_file_cache, data_sp->GetBytes(),
1196         data_sp->GetByteSize(), error, &load_addr);
1197 
1198     if (bytes_read > 0) {
1199       if (bytes_read != data_sp->GetByteSize())
1200         data_sp->SetByteSize(bytes_read);
1201       DataExtractor data(data_sp, m_arch.GetByteOrder(),
1202                          m_arch.GetAddressByteSize());
1203       const bool data_from_file = load_addr == LLDB_INVALID_ADDRESS;
1204       return DecodeInstructions(range.GetBaseAddress(), data, 0, UINT32_MAX,
1205                                 false, data_from_file);
1206     } else if (error_strm_ptr) {
1207       const char *error_cstr = error.AsCString();
1208       if (error_cstr) {
1209         error_strm_ptr->Printf("error: %s\n", error_cstr);
1210       }
1211     }
1212   } else if (error_strm_ptr) {
1213     error_strm_ptr->PutCString("error: invalid execution context\n");
1214   }
1215   return 0;
1216 }
1217 
1218 size_t Disassembler::ParseInstructions(const ExecutionContext *exe_ctx,
1219                                        const Address &start,
1220                                        uint32_t num_instructions,
1221                                        bool prefer_file_cache) {
1222   m_instruction_list.Clear();
1223 
1224   if (exe_ctx == nullptr || num_instructions == 0 || !start.IsValid())
1225     return 0;
1226 
1227   Target *target = exe_ctx->GetTargetPtr();
1228   // Calculate the max buffer size we will need in order to disassemble
1229   const addr_t byte_size = num_instructions * m_arch.GetMaximumOpcodeByteSize();
1230 
1231   if (target == nullptr || byte_size == 0)
1232     return 0;
1233 
1234   DataBufferHeap *heap_buffer = new DataBufferHeap(byte_size, '\0');
1235   DataBufferSP data_sp(heap_buffer);
1236 
1237   Status error;
1238   lldb::addr_t load_addr = LLDB_INVALID_ADDRESS;
1239   const size_t bytes_read =
1240       target->ReadMemory(start, prefer_file_cache, heap_buffer->GetBytes(),
1241                          byte_size, error, &load_addr);
1242 
1243   const bool data_from_file = load_addr == LLDB_INVALID_ADDRESS;
1244 
1245   if (bytes_read == 0)
1246     return 0;
1247   DataExtractor data(data_sp, m_arch.GetByteOrder(),
1248                      m_arch.GetAddressByteSize());
1249 
1250   const bool append_instructions = true;
1251   DecodeInstructions(start, data, 0, num_instructions, append_instructions,
1252                      data_from_file);
1253 
1254   return m_instruction_list.GetSize();
1255 }
1256 
1257 // Disassembler copy constructor
1258 Disassembler::Disassembler(const ArchSpec &arch, const char *flavor)
1259     : m_arch(arch), m_instruction_list(), m_base_addr(LLDB_INVALID_ADDRESS),
1260       m_flavor() {
1261   if (flavor == nullptr)
1262     m_flavor.assign("default");
1263   else
1264     m_flavor.assign(flavor);
1265 
1266   // If this is an arm variant that can only include thumb (T16, T32)
1267   // instructions, force the arch triple to be "thumbv.." instead of "armv..."
1268   if (arch.IsAlwaysThumbInstructions()) {
1269     std::string thumb_arch_name(arch.GetTriple().getArchName().str());
1270     // Replace "arm" with "thumb" so we get all thumb variants correct
1271     if (thumb_arch_name.size() > 3) {
1272       thumb_arch_name.erase(0, 3);
1273       thumb_arch_name.insert(0, "thumb");
1274     }
1275     m_arch.SetTriple(thumb_arch_name.c_str());
1276   }
1277 }
1278 
1279 Disassembler::~Disassembler() = default;
1280 
1281 InstructionList &Disassembler::GetInstructionList() {
1282   return m_instruction_list;
1283 }
1284 
1285 const InstructionList &Disassembler::GetInstructionList() const {
1286   return m_instruction_list;
1287 }
1288 
1289 // Class PseudoInstruction
1290 
1291 PseudoInstruction::PseudoInstruction()
1292     : Instruction(Address(), AddressClass::eUnknown), m_description() {}
1293 
1294 PseudoInstruction::~PseudoInstruction() = default;
1295 
1296 bool PseudoInstruction::DoesBranch() {
1297   // This is NOT a valid question for a pseudo instruction.
1298   return false;
1299 }
1300 
1301 bool PseudoInstruction::HasDelaySlot() {
1302   // This is NOT a valid question for a pseudo instruction.
1303   return false;
1304 }
1305 
1306 size_t PseudoInstruction::Decode(const lldb_private::Disassembler &disassembler,
1307                                  const lldb_private::DataExtractor &data,
1308                                  lldb::offset_t data_offset) {
1309   return m_opcode.GetByteSize();
1310 }
1311 
1312 void PseudoInstruction::SetOpcode(size_t opcode_size, void *opcode_data) {
1313   if (!opcode_data)
1314     return;
1315 
1316   switch (opcode_size) {
1317   case 8: {
1318     uint8_t value8 = *((uint8_t *)opcode_data);
1319     m_opcode.SetOpcode8(value8, eByteOrderInvalid);
1320     break;
1321   }
1322   case 16: {
1323     uint16_t value16 = *((uint16_t *)opcode_data);
1324     m_opcode.SetOpcode16(value16, eByteOrderInvalid);
1325     break;
1326   }
1327   case 32: {
1328     uint32_t value32 = *((uint32_t *)opcode_data);
1329     m_opcode.SetOpcode32(value32, eByteOrderInvalid);
1330     break;
1331   }
1332   case 64: {
1333     uint64_t value64 = *((uint64_t *)opcode_data);
1334     m_opcode.SetOpcode64(value64, eByteOrderInvalid);
1335     break;
1336   }
1337   default:
1338     break;
1339   }
1340 }
1341 
1342 void PseudoInstruction::SetDescription(llvm::StringRef description) {
1343   m_description = description;
1344 }
1345 
1346 Instruction::Operand Instruction::Operand::BuildRegister(ConstString &r) {
1347   Operand ret;
1348   ret.m_type = Type::Register;
1349   ret.m_register = r;
1350   return ret;
1351 }
1352 
1353 Instruction::Operand Instruction::Operand::BuildImmediate(lldb::addr_t imm,
1354                                                           bool neg) {
1355   Operand ret;
1356   ret.m_type = Type::Immediate;
1357   ret.m_immediate = imm;
1358   ret.m_negative = neg;
1359   return ret;
1360 }
1361 
1362 Instruction::Operand Instruction::Operand::BuildImmediate(int64_t imm) {
1363   Operand ret;
1364   ret.m_type = Type::Immediate;
1365   if (imm < 0) {
1366     ret.m_immediate = -imm;
1367     ret.m_negative = true;
1368   } else {
1369     ret.m_immediate = imm;
1370     ret.m_negative = false;
1371   }
1372   return ret;
1373 }
1374 
1375 Instruction::Operand
1376 Instruction::Operand::BuildDereference(const Operand &ref) {
1377   Operand ret;
1378   ret.m_type = Type::Dereference;
1379   ret.m_children = {ref};
1380   return ret;
1381 }
1382 
1383 Instruction::Operand Instruction::Operand::BuildSum(const Operand &lhs,
1384                                                     const Operand &rhs) {
1385   Operand ret;
1386   ret.m_type = Type::Sum;
1387   ret.m_children = {lhs, rhs};
1388   return ret;
1389 }
1390 
1391 Instruction::Operand Instruction::Operand::BuildProduct(const Operand &lhs,
1392                                                         const Operand &rhs) {
1393   Operand ret;
1394   ret.m_type = Type::Product;
1395   ret.m_children = {lhs, rhs};
1396   return ret;
1397 }
1398 
1399 std::function<bool(const Instruction::Operand &)>
1400 lldb_private::OperandMatchers::MatchBinaryOp(
1401     std::function<bool(const Instruction::Operand &)> base,
1402     std::function<bool(const Instruction::Operand &)> left,
1403     std::function<bool(const Instruction::Operand &)> right) {
1404   return [base, left, right](const Instruction::Operand &op) -> bool {
1405     return (base(op) && op.m_children.size() == 2 &&
1406             ((left(op.m_children[0]) && right(op.m_children[1])) ||
1407              (left(op.m_children[1]) && right(op.m_children[0]))));
1408   };
1409 }
1410 
1411 std::function<bool(const Instruction::Operand &)>
1412 lldb_private::OperandMatchers::MatchUnaryOp(
1413     std::function<bool(const Instruction::Operand &)> base,
1414     std::function<bool(const Instruction::Operand &)> child) {
1415   return [base, child](const Instruction::Operand &op) -> bool {
1416     return (base(op) && op.m_children.size() == 1 && child(op.m_children[0]));
1417   };
1418 }
1419 
1420 std::function<bool(const Instruction::Operand &)>
1421 lldb_private::OperandMatchers::MatchRegOp(const RegisterInfo &info) {
1422   return [&info](const Instruction::Operand &op) {
1423     return (op.m_type == Instruction::Operand::Type::Register &&
1424             (op.m_register == ConstString(info.name) ||
1425              op.m_register == ConstString(info.alt_name)));
1426   };
1427 }
1428 
1429 std::function<bool(const Instruction::Operand &)>
1430 lldb_private::OperandMatchers::FetchRegOp(ConstString &reg) {
1431   return [&reg](const Instruction::Operand &op) {
1432     if (op.m_type != Instruction::Operand::Type::Register) {
1433       return false;
1434     }
1435     reg = op.m_register;
1436     return true;
1437   };
1438 }
1439 
1440 std::function<bool(const Instruction::Operand &)>
1441 lldb_private::OperandMatchers::MatchImmOp(int64_t imm) {
1442   return [imm](const Instruction::Operand &op) {
1443     return (op.m_type == Instruction::Operand::Type::Immediate &&
1444             ((op.m_negative && op.m_immediate == (uint64_t)-imm) ||
1445              (!op.m_negative && op.m_immediate == (uint64_t)imm)));
1446   };
1447 }
1448 
1449 std::function<bool(const Instruction::Operand &)>
1450 lldb_private::OperandMatchers::FetchImmOp(int64_t &imm) {
1451   return [&imm](const Instruction::Operand &op) {
1452     if (op.m_type != Instruction::Operand::Type::Immediate) {
1453       return false;
1454     }
1455     if (op.m_negative) {
1456       imm = -((int64_t)op.m_immediate);
1457     } else {
1458       imm = ((int64_t)op.m_immediate);
1459     }
1460     return true;
1461   };
1462 }
1463 
1464 std::function<bool(const Instruction::Operand &)>
1465 lldb_private::OperandMatchers::MatchOpType(Instruction::Operand::Type type) {
1466   return [type](const Instruction::Operand &op) { return op.m_type == type; };
1467 }
1468