| // -*- mode: C++ -*- |
| |
| // Copyright (c) 2010, Google Inc. |
| // All rights reserved. |
| // |
| // Redistribution and use in source and binary forms, with or without |
| // modification, are permitted provided that the following conditions are |
| // met: |
| // |
| // * Redistributions of source code must retain the above copyright |
| // notice, this list of conditions and the following disclaimer. |
| // * Redistributions in binary form must reproduce the above |
| // copyright notice, this list of conditions and the following disclaimer |
| // in the documentation and/or other materials provided with the |
| // distribution. |
| // * Neither the name of Google Inc. nor the names of its |
| // contributors may be used to endorse or promote products derived from |
| // this software without specific prior written permission. |
| // |
| // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS |
| // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT |
| // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR |
| // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT |
| // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, |
| // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT |
| // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, |
| // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY |
| // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT |
| // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE |
| // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. |
| |
| // Original author: Jim Blandy <jimb@mozilla.com> <jimb@red-bean.com> |
| |
| // synth_minidump.h: Interface to SynthMinidump: fake minidump generator. |
| // |
| // We treat a minidump file as the concatenation of a bunch of |
| // test_assembler::Sections. The file header, stream directory, |
| // streams, memory regions, strings, and so on --- each is a Section |
| // that eventually gets appended to the minidump. Dump, Memory, |
| // Context, Thread, and so on all inherit from test_assembler::Section. |
| // For example: |
| // |
| // using google_breakpad::test_assembler::kLittleEndian; |
| // using google_breakpad::SynthMinidump::Context; |
| // using google_breakpad::SynthMinidump::Dump; |
| // using google_breakpad::SynthMinidump::Memory; |
| // using google_breakpad::SynthMinidump::Thread; |
| // |
| // Dump minidump(MD_NORMAL, kLittleEndian); |
| // |
| // Memory stack1(minidump, 0x569eb0a9); |
| // ... build contents of stack1 with test_assembler::Section functions ... |
| // |
| // MDRawContextX86 x86_context1; |
| // x86_context1.context_flags = MD_CONTEXT_X86; |
| // x86_context1.eip = 0x7c90eb94; |
| // x86_context1.esp = 0x569eb0a9; |
| // x86_context1.ebp = x86_context1.esp + something appropriate; |
| // Context context1(minidump, x86_context1); |
| // |
| // Thread thread1(minidump, 0xe4a4821d, stack1, context1); |
| // |
| // minidump.Add(&stack1); |
| // minidump.Add(&context1); |
| // minidump.Add(&thread1); |
| // minidump.Finish(); |
| // |
| // string contents; |
| // EXPECT_TRUE(minidump.GetContents(&contents)); |
| // // contents now holds the bytes of a minidump file |
| // |
| // Because the test_assembler classes let us write Label references to |
| // sections before the Labels' values are known, this gives us |
| // flexibility in how we put the dump together: minidump pieces can |
| // hold the file offsets of other minidump pieces before the |
| // referents' positions have been decided. As long as everything has |
| // been placed by the time we call dump.GetContents to obtain the |
| // bytes, all the Labels' values will be known, and everything will |
| // get patched up appropriately. |
| // |
| // The dump.Add(thing) functions append THINGS's contents to the |
| // minidump, but they also do two other things: |
| // |
| // - dump.Add(thing) invokes thing->Finish, which tells *thing the |
| // offset within the file at which it was placed, and allows *thing |
| // to do any final content generation. |
| // |
| // - If THING is something which should receive an entry in some sort |
| // of list or directory, then dump.Add(THING) automatically creates |
| // the appropriate directory or list entry. Streams must appear in |
| // the stream directory; memory ranges should be listed in the |
| // memory list; threads should be placed in the thread list; and so |
| // on. |
| // |
| // By convention, Section subclass constructors that take references |
| // to other Sections do not take care of 'Add'ing their arguments to |
| // the dump. For example, although the Thread constructor takes |
| // references to a Memory and a Context, it does not add them to the |
| // dump on the caller's behalf. Rather, the caller is responsible for |
| // 'Add'ing every section they create. This allows Sections to be |
| // cited from more than one place; for example, Memory ranges are |
| // cited both from Thread objects (as their stack contents) and by the |
| // memory list stream. |
| // |
| // If you forget to Add some Section, the Dump::GetContents call will |
| // fail, as the test_assembler::Labels used to cite the Section's |
| // contents from elsewhere will still be undefined. |
| #ifndef PROCESSOR_SYNTH_MINIDUMP_H_ |
| #define PROCESSOR_SYNTH_MINIDUMP_H_ |
| |
| #include <assert.h> |
| |
| #include <iostream> |
| #include <string> |
| |
| #include "common/test_assembler.h" |
| #include "common/using_std_string.h" |
| #include "google_breakpad/common/breakpad_types.h" |
| #include "google_breakpad/common/minidump_format.h" |
| |
| namespace google_breakpad { |
| |
| namespace SynthMinidump { |
| |
| using test_assembler::Endianness; |
| using test_assembler::kBigEndian; |
| using test_assembler::kLittleEndian; |
| using test_assembler::kUnsetEndian; |
| using test_assembler::Label; |
| |
| class Dump; |
| class Memory; |
| class String; |
| |
| // A test_assembler::Section which will be appended to a minidump. |
| class Section: public test_assembler::Section { |
| public: |
| explicit Section(const Dump &dump); |
| |
| // Append an MDLocationDescriptor referring to this section to SECTION. |
| // If 'this' is NULL, append a descriptor with a zero length and MDRVA. |
| // |
| // (I couldn't find the language in the C++ standard that says that |
| // invoking member functions of a NULL pointer to a class type is |
| // bad, if such language exists. Having this function handle NULL |
| // 'this' is convenient, but if it causes trouble, it's not hard to |
| // do differently.) |
| void CiteLocationIn(test_assembler::Section *section) const; |
| |
| // Note that this section's contents are complete, and that it has |
| // been placed in the minidump file at OFFSET. The 'Add' member |
| // functions call the Finish member function of the object being |
| // added for you; if you are 'Add'ing this section, you needn't Finish it. |
| virtual void Finish(const Label &offset) { |
| file_offset_ = offset; size_ = Size(); |
| } |
| |
| protected: |
| // This section's size and offset within the minidump file. |
| Label file_offset_, size_; |
| }; |
| |
| // A stream within a minidump file. 'Add'ing a stream to a minidump |
| // creates an entry for it in the minidump's stream directory. |
| class Stream: public Section { |
| public: |
| // Create a stream of type TYPE. You can append whatever contents |
| // you like to this stream using the test_assembler::Section methods. |
| Stream(const Dump &dump, uint32_t type) : Section(dump), type_(type) { } |
| |
| // Append an MDRawDirectory referring to this stream to SECTION. |
| void CiteStreamIn(test_assembler::Section *section) const; |
| |
| private: |
| // The type of this stream. |
| uint32_t type_; |
| }; |
| |
| class SystemInfo: public Stream { |
| public: |
| // Create an MD_SYSTEM_INFO_STREAM stream belonging to DUMP holding |
| // an MDRawSystem info structure initialized with the values from |
| // SYSTEM_INFO, except that the csd_version field is replaced with |
| // the file offset of the string CSD_VERSION, which can be 'Add'ed |
| // to the dump at the desired location. |
| // |
| // Remember that you are still responsible for 'Add'ing CSD_VERSION |
| // to the dump yourself. |
| SystemInfo(const Dump &dump, |
| const MDRawSystemInfo &system_info, |
| const String &csd_version); |
| |
| // Stock MDRawSystemInfo information and associated strings, for |
| // writing tests. |
| static const MDRawSystemInfo windows_x86; |
| static const string windows_x86_csd_version; |
| }; |
| |
| // An MDString: a string preceded by a 32-bit length. |
| class String: public Section { |
| public: |
| String(const Dump &dump, const string &value); |
| |
| // Append an MDRVA referring to this string to SECTION. |
| void CiteStringIn(test_assembler::Section *section) const; |
| }; |
| |
| // A range of memory contents. 'Add'ing a memory range to a minidump |
| // creates n entry for it in the minidump's memory list. By |
| // convention, the 'start', 'Here', and 'Mark' member functions refer |
| // to memory addresses. |
| class Memory: public Section { |
| public: |
| Memory(const Dump &dump, uint64_t address) |
| : Section(dump), address_(address) { start() = address; } |
| |
| // Append an MDMemoryDescriptor referring to this memory range to SECTION. |
| void CiteMemoryIn(test_assembler::Section *section) const; |
| |
| private: |
| // The process address from which these memory contents were taken. |
| // Shouldn't this be a Label? |
| uint64_t address_; |
| }; |
| |
| class Context: public Section { |
| public: |
| // Create a context belonging to DUMP whose contents are a copy of CONTEXT. |
| Context(const Dump &dump, const MDRawContextX86 &context); |
| Context(const Dump &dump, const MDRawContextARM &context); |
| Context(const Dump &dump, const MDRawContextMIPS &context); |
| // Add an empty context to the dump. |
| Context(const Dump &dump) : Section(dump) {} |
| // Add constructors for other architectures here. Remember to byteswap. |
| }; |
| |
| class Thread: public Section { |
| public: |
| // Create a thread belonging to DUMP with the given values, citing |
| // STACK and CONTEXT (which you must Add to the dump separately). |
| Thread(const Dump &dump, |
| uint32_t thread_id, |
| const Memory &stack, |
| const Context &context, |
| uint32_t suspend_count = 0, |
| uint32_t priority_class = 0, |
| uint32_t priority = 0, |
| uint64_t teb = 0); |
| }; |
| |
| class Module: public Section { |
| public: |
| // Create a module with the given values. Note that CV_RECORD and |
| // MISC_RECORD can be NULL, in which case the corresponding location |
| // descriptior in the minidump will have a length of zero. |
| Module(const Dump &dump, |
| uint64_t base_of_image, |
| uint32_t size_of_image, |
| const String &name, |
| uint32_t time_date_stamp = 1262805309, |
| uint32_t checksum = 0, |
| const MDVSFixedFileInfo &version_info = Module::stock_version_info, |
| const Section *cv_record = NULL, |
| const Section *misc_record = NULL); |
| |
| private: |
| // A standard MDVSFixedFileInfo structure to use as a default for |
| // minidumps. There's no reason to make users write out all this crap |
| // over and over. |
| static const MDVSFixedFileInfo stock_version_info; |
| }; |
| |
| class UnloadedModule: public Section { |
| public: |
| UnloadedModule(const Dump &dump, |
| uint64_t base_of_image, |
| uint32_t size_of_image, |
| const String &name, |
| uint32_t checksum = 0, |
| uint32_t time_date_stamp = 1262805309); |
| }; |
| |
| class Exception : public Stream { |
| public: |
| Exception(const Dump &dump, |
| const Context &context, |
| uint32_t thread_id = 0, |
| uint32_t exception_code = 0, |
| uint32_t exception_flags = 0, |
| uint64_t exception_address = 0); |
| }; |
| |
| // A list of entries starting with a 32-bit count, like a memory list |
| // or a thread list. |
| template<typename Element> |
| class List: public Stream { |
| public: |
| List(const Dump &dump, uint32_t type) : Stream(dump, type), count_(0) { |
| D32(count_label_); |
| } |
| |
| // Add ELEMENT to this list. |
| void Add(Element *element) { |
| element->Finish(file_offset_ + Size()); |
| Append(*element); |
| count_++; |
| } |
| |
| // Return true if this List is empty, false otherwise. |
| bool Empty() { return count_ == 0; } |
| |
| // Finish up the contents of this section, mark it as having been |
| // placed at OFFSET. |
| virtual void Finish(const Label &offset) { |
| Stream::Finish(offset); |
| count_label_ = count_; |
| } |
| |
| private: |
| size_t count_; |
| |
| protected: |
| // This constructor allows derived lists to specify their own layout |
| // rather than starting with count as specified in the public constructor. |
| List(const Dump &dump, uint32_t type, bool) : Stream(dump, type), count_(0) {} |
| |
| Label count_label_; |
| }; |
| |
| class UnloadedModuleList : public List<UnloadedModule> { |
| public: |
| UnloadedModuleList(const Dump &dump, uint32_t type); |
| }; |
| |
| class Dump: public test_assembler::Section { |
| public: |
| |
| // Create a test_assembler::Section containing a minidump file whose |
| // header uses the given values. ENDIANNESS determines the |
| // endianness of the signature; we set this section's default |
| // endianness by this. |
| Dump(uint64_t flags, |
| Endianness endianness = kLittleEndian, |
| uint32_t version = MD_HEADER_VERSION, |
| uint32_t date_time_stamp = 1262805309); |
| |
| // The following functions call OBJECT->Finish(), and append the |
| // contents of OBJECT to this minidump. They also record OBJECT in |
| // whatever directory or list is appropriate for its type. The |
| // stream directory, memory list, thread list, and module list are |
| // accumulated this way. |
| Dump &Add(SynthMinidump::Section *object); // simply append data |
| Dump &Add(Stream *object); // append, record in stream directory |
| Dump &Add(Memory *object); // append, record in memory list |
| Dump &Add(Thread *object); // append, record in thread list |
| Dump &Add(Module *object); // append, record in module list |
| Dump &Add(UnloadedModule *object); // append, record in unloaded module list |
| |
| // Complete the construction of the minidump, given the Add calls |
| // we've seen up to this point. After this call, this Dump's |
| // contents are complete, all labels should be defined if everything |
| // Cited has been Added, and you may call GetContents on it. |
| void Finish(); |
| |
| private: |
| // A label representing the start of the minidump file. |
| Label file_start_; |
| |
| // The stream directory. We construct this incrementally from |
| // Add(Stream *) calls. |
| SynthMinidump::Section stream_directory_; // The directory's contents. |
| size_t stream_count_; // The number of streams so far. |
| Label stream_count_label_; // Cited in file header. |
| Label stream_directory_rva_; // The directory's file offset. |
| |
| // This minidump's thread list. We construct this incrementally from |
| // Add(Thread *) calls. |
| List<Thread> thread_list_; |
| |
| // This minidump's module list. We construct this incrementally from |
| // Add(Module *) calls. |
| List<Module> module_list_; |
| |
| // This minidump's unloaded module list. We construct this incrementally from |
| // Add(UnloadedModule *) calls. |
| UnloadedModuleList unloaded_module_list_; |
| |
| // This minidump's memory list. We construct this incrementally from |
| // Add(Memory *) calls. This is actually a list of MDMemoryDescriptors, |
| // not memory ranges --- thus the odd type. |
| List<SynthMinidump::Section> memory_list_; |
| }; |
| |
| } // namespace SynthMinidump |
| |
| } // namespace google_breakpad |
| |
| #endif // PROCESSOR_SYNTH_MINIDUMP_H_ |