1 | #ifdef CPPTRACE_GET_SYMBOLS_WITH_DBGHELP |
---|
2 | |
---|
3 | #include "cpptrace.hpp" |
---|
4 | #include "symbols.hpp" |
---|
5 | #include "program_name.hpp" |
---|
6 | |
---|
7 | #include <memory> |
---|
8 | #include <regex> |
---|
9 | #include <stdexcept> |
---|
10 | #include <system_error> |
---|
11 | #include <vector> |
---|
12 | |
---|
13 | #include <windows.h> |
---|
14 | #include <dbghelp.h> |
---|
15 | |
---|
16 | namespace cpptrace { |
---|
17 | namespace detail { |
---|
18 | namespace dbghelp { |
---|
19 | |
---|
20 | // SymFromAddr only returns the function's name. In order to get information about parameters, |
---|
21 | // important for C++ stack traces where functions may be overloaded, we have to manually use |
---|
22 | // Windows DIA to walk debug info structures. Resources: |
---|
23 | // https://web.archive.org/web/20201027025750/http://www.debuginfo.com/articles/dbghelptypeinfo.html |
---|
24 | // https://web.archive.org/web/20201203160805/http://www.debuginfo.com/articles/dbghelptypeinfofigures.html |
---|
25 | // https://github.com/DynamoRIO/dynamorio/blob/master/ext/drsyms/drsyms_windows.c#L1370-L1439 |
---|
26 | // TODO: Currently unable to detect rvalue references |
---|
27 | // TODO: Currently unable to detect const |
---|
28 | enum class SymTagEnum { |
---|
29 | SymTagNull, SymTagExe, SymTagCompiland, SymTagCompilandDetails, SymTagCompilandEnv, |
---|
30 | SymTagFunction, SymTagBlock, SymTagData, SymTagAnnotation, SymTagLabel, SymTagPublicSymbol, |
---|
31 | SymTagUDT, SymTagEnum, SymTagFunctionType, SymTagPointerType, SymTagArrayType, |
---|
32 | SymTagBaseType, SymTagTypedef, SymTagBaseClass, SymTagFriend, SymTagFunctionArgType, |
---|
33 | SymTagFuncDebugStart, SymTagFuncDebugEnd, SymTagUsingNamespace, SymTagVTableShape, |
---|
34 | SymTagVTable, SymTagCustom, SymTagThunk, SymTagCustomType, SymTagManagedType, |
---|
35 | SymTagDimension, SymTagCallSite, SymTagInlineSite, SymTagBaseInterface, SymTagVectorType, |
---|
36 | SymTagMatrixType, SymTagHLSLType, SymTagCaller, SymTagCallee, SymTagExport, |
---|
37 | SymTagHeapAllocationSite, SymTagCoffGroup, SymTagMax |
---|
38 | }; |
---|
39 | |
---|
40 | enum class IMAGEHLP_SYMBOL_TYPE_INFO { |
---|
41 | TI_GET_SYMTAG, TI_GET_SYMNAME, TI_GET_LENGTH, TI_GET_TYPE, TI_GET_TYPEID, TI_GET_BASETYPE, |
---|
42 | TI_GET_ARRAYINDEXTYPEID, TI_FINDCHILDREN, TI_GET_DATAKIND, TI_GET_ADDRESSOFFSET, |
---|
43 | TI_GET_OFFSET, TI_GET_VALUE, TI_GET_COUNT, TI_GET_CHILDRENCOUNT, TI_GET_BITPOSITION, |
---|
44 | TI_GET_VIRTUALBASECLASS, TI_GET_VIRTUALTABLESHAPEID, TI_GET_VIRTUALBASEPOINTEROFFSET, |
---|
45 | TI_GET_CLASSPARENTID, TI_GET_NESTED, TI_GET_SYMINDEX, TI_GET_LEXICALPARENT, TI_GET_ADDRESS, |
---|
46 | TI_GET_THISADJUST, TI_GET_UDTKIND, TI_IS_EQUIV_TO, TI_GET_CALLING_CONVENTION, |
---|
47 | TI_IS_CLOSE_EQUIV_TO, TI_GTIEX_REQS_VALID, TI_GET_VIRTUALBASEOFFSET, |
---|
48 | TI_GET_VIRTUALBASEDISPINDEX, TI_GET_IS_REFERENCE, TI_GET_INDIRECTVIRTUALBASECLASS, |
---|
49 | TI_GET_VIRTUALBASETABLETYPE, TI_GET_OBJECTPOINTERTYPE, IMAGEHLP_SYMBOL_TYPE_INFO_MAX |
---|
50 | }; |
---|
51 | |
---|
52 | enum class BasicType { |
---|
53 | btNoType = 0, btVoid = 1, btChar = 2, btWChar = 3, btInt = 6, btUInt = 7, btFloat = 8, |
---|
54 | btBCD = 9, btBool = 10, btLong = 13, btULong = 14, btCurrency = 25, btDate = 26, |
---|
55 | btVariant = 27, btComplex = 28, btBit = 29, btBSTR = 30, btHresult = 31 |
---|
56 | }; |
---|
57 | |
---|
58 | // SymGetTypeInfo utility |
---|
59 | template<typename T, IMAGEHLP_SYMBOL_TYPE_INFO SymType, bool FAILABLE = false> |
---|
60 | T get_info(ULONG type_index, HANDLE proc, ULONG64 modbase) { |
---|
61 | T info; |
---|
62 | if( |
---|
63 | !SymGetTypeInfo( |
---|
64 | proc, |
---|
65 | modbase, |
---|
66 | type_index, |
---|
67 | static_cast<::IMAGEHLP_SYMBOL_TYPE_INFO>(SymType), |
---|
68 | &info |
---|
69 | ) |
---|
70 | ) { |
---|
71 | if(FAILABLE) { |
---|
72 | return (T)-1; |
---|
73 | } else { |
---|
74 | throw std::logic_error( |
---|
75 | std::string("SymGetTypeInfo failed: ") |
---|
76 | + std::system_error(GetLastError(), std::system_category()).what() |
---|
77 | ); |
---|
78 | } |
---|
79 | } |
---|
80 | return info; |
---|
81 | } |
---|
82 | |
---|
83 | template<IMAGEHLP_SYMBOL_TYPE_INFO SymType, bool FAILABLE = false> |
---|
84 | std::string get_info_wchar(ULONG type_index, HANDLE proc, ULONG64 modbase) { |
---|
85 | WCHAR* info; |
---|
86 | if( |
---|
87 | !SymGetTypeInfo(proc, modbase, type_index, static_cast<::IMAGEHLP_SYMBOL_TYPE_INFO>(SymType), &info) |
---|
88 | ) { |
---|
89 | throw std::logic_error( |
---|
90 | std::string("SymGetTypeInfo failed: ") |
---|
91 | + std::system_error(GetLastError(), std::system_category()).what() |
---|
92 | ); |
---|
93 | } |
---|
94 | // special case to properly free a buffer and convert string to narrow chars, only used for |
---|
95 | // TI_GET_SYMNAME |
---|
96 | static_assert( |
---|
97 | SymType == IMAGEHLP_SYMBOL_TYPE_INFO::TI_GET_SYMNAME, |
---|
98 | "get_info_wchar called with unexpected IMAGEHLP_SYMBOL_TYPE_INFO" |
---|
99 | ); |
---|
100 | std::wstring wstr(info); |
---|
101 | std::string str; |
---|
102 | str.reserve(wstr.size()); |
---|
103 | for(const auto c : wstr) { |
---|
104 | str.push_back(static_cast<char>(c)); |
---|
105 | } |
---|
106 | LocalFree(info); |
---|
107 | return str; |
---|
108 | } |
---|
109 | |
---|
110 | // Translate basic types to string |
---|
111 | static std::string get_basic_type(ULONG type_index, HANDLE proc, ULONG64 modbase) { |
---|
112 | auto basic_type = get_info<BasicType, IMAGEHLP_SYMBOL_TYPE_INFO::TI_GET_BASETYPE>( |
---|
113 | type_index, |
---|
114 | proc, |
---|
115 | modbase |
---|
116 | ); |
---|
117 | //auto length = get_info<ULONG64, IMAGEHLP_SYMBOL_TYPE_INFO::TI_GET_LENGTH>(type_index, proc, modbase); |
---|
118 | switch(basic_type) { |
---|
119 | case BasicType::btNoType: |
---|
120 | return "<no basic type>"; |
---|
121 | case BasicType::btVoid: |
---|
122 | return "void"; |
---|
123 | case BasicType::btChar: |
---|
124 | return "char"; |
---|
125 | case BasicType::btWChar: |
---|
126 | return "wchar_t"; |
---|
127 | case BasicType::btInt: |
---|
128 | return "int"; |
---|
129 | case BasicType::btUInt: |
---|
130 | return "unsigned int"; |
---|
131 | case BasicType::btFloat: |
---|
132 | return "float"; |
---|
133 | case BasicType::btBool: |
---|
134 | return "bool"; |
---|
135 | case BasicType::btLong: |
---|
136 | return "long"; |
---|
137 | case BasicType::btULong: |
---|
138 | return "unsigned long"; |
---|
139 | default: |
---|
140 | return "<unknown basic type>"; |
---|
141 | } |
---|
142 | } |
---|
143 | |
---|
144 | static std::string resolve_type(ULONG type_index, HANDLE proc, ULONG64 modbase); |
---|
145 | |
---|
146 | struct class_name_result { |
---|
147 | bool has_class_name; |
---|
148 | std::string name; |
---|
149 | }; |
---|
150 | // Helper for member pointers |
---|
151 | static class_name_result lookup_class_name(ULONG type_index, HANDLE proc, ULONG64 modbase) { |
---|
152 | DWORD class_parent_id = get_info<DWORD, IMAGEHLP_SYMBOL_TYPE_INFO::TI_GET_CLASSPARENTID, true>( |
---|
153 | type_index, |
---|
154 | proc, |
---|
155 | modbase |
---|
156 | ); |
---|
157 | if(class_parent_id == (DWORD)-1) { |
---|
158 | return {false, ""}; |
---|
159 | } else { |
---|
160 | return {true, resolve_type(class_parent_id, proc, modbase)}; |
---|
161 | } |
---|
162 | } |
---|
163 | |
---|
164 | struct type_result { |
---|
165 | std::string base; |
---|
166 | std::string extent; |
---|
167 | }; |
---|
168 | // Resolve more complex types |
---|
169 | // returns [base, extent] |
---|
170 | static type_result lookup_type(ULONG type_index, HANDLE proc, ULONG64 modbase) { |
---|
171 | auto tag = get_info<SymTagEnum, IMAGEHLP_SYMBOL_TYPE_INFO::TI_GET_SYMTAG>(type_index, proc, modbase); |
---|
172 | switch(tag) { |
---|
173 | case SymTagEnum::SymTagBaseType: |
---|
174 | return {get_basic_type(type_index, proc, modbase), ""}; |
---|
175 | case SymTagEnum::SymTagPointerType: { |
---|
176 | DWORD underlying_type_id = get_info<DWORD, IMAGEHLP_SYMBOL_TYPE_INFO::TI_GET_TYPEID>( |
---|
177 | type_index, |
---|
178 | proc, |
---|
179 | modbase |
---|
180 | ); |
---|
181 | bool is_ref = get_info<BOOL, IMAGEHLP_SYMBOL_TYPE_INFO::TI_GET_IS_REFERENCE>( |
---|
182 | type_index, |
---|
183 | proc, |
---|
184 | modbase |
---|
185 | ); |
---|
186 | std::string pp = is_ref ? "&" : "*"; // pointer punctuator |
---|
187 | auto class_name_res = lookup_class_name(type_index, proc, modbase); |
---|
188 | if(class_name_res.has_class_name) { |
---|
189 | pp = class_name_res.name + "::" + pp; |
---|
190 | } |
---|
191 | const auto type = lookup_type(underlying_type_id, proc, modbase); |
---|
192 | if(type.extent.empty()) { |
---|
193 | return {type.base + (pp.size() > 1 ? " " : "") + pp, ""}; |
---|
194 | } else { |
---|
195 | return {type.base + "(" + pp, ")" + type.extent}; |
---|
196 | } |
---|
197 | } |
---|
198 | case SymTagEnum::SymTagArrayType: { |
---|
199 | DWORD underlying_type_id = get_info<DWORD, IMAGEHLP_SYMBOL_TYPE_INFO::TI_GET_TYPEID>( |
---|
200 | type_index, |
---|
201 | proc, |
---|
202 | modbase |
---|
203 | ); |
---|
204 | DWORD length = get_info<DWORD, IMAGEHLP_SYMBOL_TYPE_INFO::TI_GET_COUNT>( |
---|
205 | type_index, |
---|
206 | proc, |
---|
207 | modbase |
---|
208 | ); |
---|
209 | const auto type = lookup_type(underlying_type_id, proc, modbase); |
---|
210 | return {type.base, "[" + std::to_string(length) + "]" + type.extent}; |
---|
211 | } |
---|
212 | case SymTagEnum::SymTagFunctionType: { |
---|
213 | DWORD return_type_id = get_info<DWORD, IMAGEHLP_SYMBOL_TYPE_INFO::TI_GET_TYPEID>( |
---|
214 | type_index, |
---|
215 | proc, |
---|
216 | modbase |
---|
217 | ); |
---|
218 | DWORD n_children = get_info<DWORD, IMAGEHLP_SYMBOL_TYPE_INFO::TI_GET_COUNT, true>( |
---|
219 | type_index, |
---|
220 | proc, |
---|
221 | modbase |
---|
222 | ); |
---|
223 | DWORD class_parent_id = get_info<DWORD, IMAGEHLP_SYMBOL_TYPE_INFO::TI_GET_CLASSPARENTID, true>( |
---|
224 | type_index, |
---|
225 | proc, |
---|
226 | modbase |
---|
227 | ); |
---|
228 | int n_ignore = class_parent_id != (DWORD)-1; // ignore this param |
---|
229 | // this must be ignored before TI_FINDCHILDREN_PARAMS::Count is set, else error |
---|
230 | n_children -= n_ignore; |
---|
231 | // return type |
---|
232 | const auto return_type = lookup_type(return_type_id, proc, modbase); |
---|
233 | if(n_children == 0) { |
---|
234 | return {return_type.base, "()" + return_type.extent}; |
---|
235 | } else { |
---|
236 | // alignment should be fine |
---|
237 | size_t sz = sizeof(TI_FINDCHILDREN_PARAMS) + |
---|
238 | (n_children) * sizeof(TI_FINDCHILDREN_PARAMS::ChildId[0]); |
---|
239 | TI_FINDCHILDREN_PARAMS* children = (TI_FINDCHILDREN_PARAMS*) new char[sz]; |
---|
240 | children->Start = 0; |
---|
241 | children->Count = n_children; |
---|
242 | if( |
---|
243 | !SymGetTypeInfo( |
---|
244 | proc, modbase, type_index, |
---|
245 | static_cast<::IMAGEHLP_SYMBOL_TYPE_INFO>( |
---|
246 | IMAGEHLP_SYMBOL_TYPE_INFO::TI_FINDCHILDREN |
---|
247 | ), |
---|
248 | children |
---|
249 | ) |
---|
250 | ) { |
---|
251 | throw std::logic_error( |
---|
252 | std::string("SymGetTypeInfo failed: ") |
---|
253 | + std::system_error(GetLastError(), std::system_category()).what() |
---|
254 | ); |
---|
255 | } |
---|
256 | // get children type |
---|
257 | std::string extent = "("; |
---|
258 | if(children->Start != 0) { |
---|
259 | throw std::logic_error("Error: children->Start == 0"); |
---|
260 | } |
---|
261 | for(std::size_t i = 0; i < n_children; i++) { |
---|
262 | extent += (i == 0 ? "" : ", ") + resolve_type(children->ChildId[i], proc, modbase); |
---|
263 | } |
---|
264 | extent += ")"; |
---|
265 | delete[] (char*) children; |
---|
266 | return {return_type.base, extent + return_type.extent}; |
---|
267 | } |
---|
268 | } |
---|
269 | case SymTagEnum::SymTagFunctionArgType: { |
---|
270 | DWORD underlying_type_id = |
---|
271 | get_info<DWORD, IMAGEHLP_SYMBOL_TYPE_INFO::TI_GET_TYPEID>(type_index, proc, modbase); |
---|
272 | return {resolve_type(underlying_type_id, proc, modbase), ""}; |
---|
273 | } |
---|
274 | case SymTagEnum::SymTagTypedef: |
---|
275 | case SymTagEnum::SymTagEnum: |
---|
276 | case SymTagEnum::SymTagUDT: |
---|
277 | case SymTagEnum::SymTagBaseClass: |
---|
278 | return { |
---|
279 | get_info_wchar<IMAGEHLP_SYMBOL_TYPE_INFO::TI_GET_SYMNAME>(type_index, proc, modbase), "" |
---|
280 | }; |
---|
281 | default: |
---|
282 | return { |
---|
283 | "<unknown type " + |
---|
284 | std::to_string(static_cast<std::underlying_type<SymTagEnum>::type>(tag)) + |
---|
285 | ">", |
---|
286 | "" |
---|
287 | }; |
---|
288 | }; |
---|
289 | } |
---|
290 | |
---|
291 | static std::string resolve_type(ULONG type_index, HANDLE proc, ULONG64 modbase) { |
---|
292 | const auto type = lookup_type(type_index, proc, modbase); |
---|
293 | return type.base + type.extent; |
---|
294 | } |
---|
295 | |
---|
296 | struct function_info { |
---|
297 | HANDLE proc; |
---|
298 | ULONG64 modbase; |
---|
299 | int counter; |
---|
300 | int n_children; |
---|
301 | int n_ignore; |
---|
302 | std::string str; |
---|
303 | }; |
---|
304 | |
---|
305 | // Enumerates function parameters |
---|
306 | static BOOL __stdcall enumerator_callback( |
---|
307 | PSYMBOL_INFO symbol_info, |
---|
308 | ULONG, |
---|
309 | PVOID data |
---|
310 | ) { |
---|
311 | function_info* ctx = (function_info*)data; |
---|
312 | if(ctx->counter++ >= ctx->n_children) { |
---|
313 | return false; |
---|
314 | } |
---|
315 | if(ctx->n_ignore-- > 0) { |
---|
316 | return true; // just skip |
---|
317 | } |
---|
318 | ctx->str += resolve_type(symbol_info->TypeIndex, ctx->proc, ctx->modbase); |
---|
319 | if(ctx->counter < ctx->n_children) { |
---|
320 | ctx->str += ", "; |
---|
321 | } |
---|
322 | return true; |
---|
323 | } |
---|
324 | |
---|
325 | std::mutex dbghelp_lock; |
---|
326 | |
---|
327 | // TODO: Handle backtrace_pcinfo calling the callback multiple times on inlined functions |
---|
328 | stacktrace_frame resolve_frame(HANDLE proc, void* addr) { |
---|
329 | const std::lock_guard<std::mutex> lock(dbghelp_lock); // all dbghelp functions are not thread safe |
---|
330 | alignas(SYMBOL_INFO) char buffer[sizeof(SYMBOL_INFO) + MAX_SYM_NAME * sizeof(TCHAR)]; |
---|
331 | SYMBOL_INFO* symbol = (SYMBOL_INFO*)buffer; |
---|
332 | symbol->SizeOfStruct = sizeof(SYMBOL_INFO); |
---|
333 | symbol->MaxNameLen = MAX_SYM_NAME; |
---|
334 | union { DWORD64 a; DWORD b; } displacement; |
---|
335 | IMAGEHLP_LINE64 line; |
---|
336 | bool got_line = SymGetLineFromAddr64(proc, (DWORD64)addr, &displacement.b, &line); |
---|
337 | if(SymFromAddr(proc, (DWORD64)addr, &displacement.a, symbol)) { |
---|
338 | if(got_line) { |
---|
339 | IMAGEHLP_STACK_FRAME frame; |
---|
340 | frame.InstructionOffset = symbol->Address; |
---|
341 | // https://docs.microsoft.com/en-us/windows/win32/api/dbghelp/nf-dbghelp-symsetcontext |
---|
342 | // "If you call SymSetContext to set the context to its current value, the |
---|
343 | // function fails but GetLastError returns ERROR_SUCCESS." |
---|
344 | // This is the stupidest fucking api I've ever worked with. |
---|
345 | if(SymSetContext(proc, &frame, nullptr) == FALSE && GetLastError() != ERROR_SUCCESS) { |
---|
346 | fprintf(stderr, "Stack trace: Internal error while calling SymSetContext\n"); |
---|
347 | return { |
---|
348 | reinterpret_cast<uintptr_t>(addr), |
---|
349 | static_cast<std::uint_least32_t>(line.LineNumber), |
---|
350 | 0, |
---|
351 | line.FileName, |
---|
352 | symbol->Name |
---|
353 | }; |
---|
354 | } |
---|
355 | DWORD n_children = get_info<DWORD, IMAGEHLP_SYMBOL_TYPE_INFO::TI_GET_COUNT, true>( |
---|
356 | symbol->TypeIndex, |
---|
357 | proc, |
---|
358 | symbol->ModBase |
---|
359 | ); |
---|
360 | DWORD class_parent_id = get_info<DWORD, IMAGEHLP_SYMBOL_TYPE_INFO::TI_GET_CLASSPARENTID, true>( |
---|
361 | symbol->TypeIndex, |
---|
362 | proc, |
---|
363 | symbol->ModBase |
---|
364 | ); |
---|
365 | function_info fi { |
---|
366 | proc, |
---|
367 | symbol->ModBase, |
---|
368 | 0, |
---|
369 | int(n_children), |
---|
370 | class_parent_id != (DWORD)-1, |
---|
371 | "" |
---|
372 | }; |
---|
373 | SymEnumSymbols(proc, 0, nullptr, enumerator_callback, &fi); |
---|
374 | std::string signature = symbol->Name + std::string("(") + fi.str + ")"; |
---|
375 | // There's a phenomina with DIA not inserting commas after template parameters. Fix them here. |
---|
376 | static std::regex comma_re(R"(,(?=\S))"); |
---|
377 | signature = std::regex_replace(signature, comma_re, ", "); |
---|
378 | return { |
---|
379 | reinterpret_cast<uintptr_t>(addr), |
---|
380 | static_cast<std::uint_least32_t>(line.LineNumber), |
---|
381 | 0, |
---|
382 | line.FileName, |
---|
383 | signature |
---|
384 | }; |
---|
385 | } else { |
---|
386 | return { |
---|
387 | reinterpret_cast<uintptr_t>(addr), |
---|
388 | 0, |
---|
389 | 0, |
---|
390 | "", |
---|
391 | symbol->Name |
---|
392 | }; |
---|
393 | } |
---|
394 | } else { |
---|
395 | return { |
---|
396 | reinterpret_cast<uintptr_t>(addr), |
---|
397 | 0, |
---|
398 | 0, |
---|
399 | "", |
---|
400 | "" |
---|
401 | }; |
---|
402 | } |
---|
403 | } |
---|
404 | |
---|
405 | std::vector<stacktrace_frame> resolve_frames(const std::vector<void*>& frames) { |
---|
406 | std::vector<stacktrace_frame> trace; |
---|
407 | trace.reserve(frames.size()); |
---|
408 | |
---|
409 | // TODO: When does this need to be called? Can it be moved to the symbolizer? |
---|
410 | SymSetOptions(SYMOPT_ALLOW_ABSOLUTE_SYMBOLS); |
---|
411 | HANDLE proc = GetCurrentProcess(); |
---|
412 | if(!SymInitialize(proc, NULL, TRUE)) { |
---|
413 | //TODO? |
---|
414 | throw std::logic_error("SymInitialize failed"); |
---|
415 | } |
---|
416 | for(const auto frame : frames) { |
---|
417 | trace.push_back(resolve_frame(proc, frame)); |
---|
418 | } |
---|
419 | if(!SymCleanup(proc)) { |
---|
420 | //throw std::logic_error("SymCleanup failed"); |
---|
421 | } |
---|
422 | |
---|
423 | return trace; |
---|
424 | } |
---|
425 | } |
---|
426 | } |
---|
427 | } |
---|
428 | |
---|
429 | #endif |
---|