[lldb/crashlog] Improve exception reporting for interactive mode

[mib]

Summary

This patch improve exception reporting when loading a crash report in a
scripted process. Now, we parse the exception dictionary from the
crash report use it the create a higher fidelity MachException stop info.

This patch also updates the test to reflect that change.

rdar://97096486

Signed-off-by: Med Ismail Bennani <[e-mail]>

[JDevlieghere]

lldb/source/Plugins/Process/Utility/StopInfoMachException.cpp

525–542

Just double checking: I assume these are always defined, even when mach/exception.h is not included?

567–568

578–610

Seems like a good candidate for a StringSwitch: https://llvm.org/doxygen/classllvm_1_1StringSwitch.html

[mib]

lldb/source/Plugins/Process/Utility/StopInfoMachException.cpp

525–542

no, I'm getting the following error if I don't include mach/exception.h

llvm.org/lldb/source/Plugins/Process/Utility/StopInfoMachException.cpp:525:8: error: use of undeclared identifier 'EXC_BAD_ACCESS'
  case EXC_BAD_ACCESS:
       ^

[mib]

Use llvm::StringSwitch to get the exception code with the name

[mib]

Remove the comment from MachException::ExceptionCode since the function returns now an llvm::Optional

[mib]

Reflect changes from D131719

[JDevlieghere]

LGTM assuming this compiles on non-apple hosts

lldb/source/Plugins/Process/Utility/StopInfoMachException.cpp

525–542

So doesn't that mean this won't compile on a non-apple platform? Is this whole block APPLE only?

lldb/examples/python/scripted_process/crashlog_scripted_process.py

def parse_crashlog(self):
             crashlog_parser = CrashLogParser(self.dbg, self.crashlog_path, False)
             crashlog = crashlog_parser.parse()
         except Exception as e:
             raise e
 
         self.pid = crashlog.process_id
         self.addr_mask = crashlog.addr_mask
         self.crashed_thread_idx = crashlog.crashed_thread_idx
+        if crashlog_parser.data and 'exception' in crashlog_parser.data:
+            self.exception = crashlog_parser.data['exception']
         self.loaded_images = []
 
         def load_images(self, images):
             #TODO: Add to self.loaded_images and load images in lldb
             if images:
                 for image in images:
                     if image not in self.loaded_images:
                         if image.uuid == uuid.UUID(int=0):
def __init__(self, target: lldb.SBTarget, args : lldb.SBStructuredData):
             if load_all_images.GetType() == lldb.eStructuredDataTypeBoolean:
                 self.load_all_images = load_all_images.GetBooleanValue()
 
         if not self.load_all_images:
             self.load_all_images = False
 
         self.pid = super().get_process_id()
         self.crashed_thread_idx = 0
+        self.exception = None
         self.parse_crashlog()
 
     def get_memory_region_containing_address(self, addr: int) -> lldb.SBMemoryRegionInfo:
         return None
 
     def get_thread_with_id(self, tid: int):
         return {}
 
class CrashLogScriptedThread(ScriptedThread):
 
     def get_state(self):
         if not self.has_crashed:
             return lldb.eStateStopped
         return lldb.eStateCrashed
 
     def get_stop_reason(self) -> Dict[str, Any]:
         if not self.has_crashed:
-            return { "type": lldb.eStopReasonNone, "data": {  }}
+            return { "type": lldb.eStopReasonNone }
         # TODO: Investigate what stop reason should be reported when crashed
-        return { "type": lldb.eStopReasonException, "data": { "desc": "EXC_BAD_ACCESS" }}
+        stop_reason = { "type": lldb.eStopReasonException, "data": {  }}
+        if self.scripted_process.exception:
+            stop_reason['data']['mach_exception'] = self.scripted_process.exception
+        return stop_reason
 
     def get_register_context(self) -> str:
         if not self.register_ctx:
             self.register_ctx = self.create_register_ctx()
 
         return struct.pack("{}Q".format(len(self.register_ctx)), *self.register_ctx.values())

lldb/source/Plugins/Process/Utility/StopInfoMachException.h

public:
   ~StopInfoMachException() override = default;
 
   lldb::StopReason GetStopReason() const override {
     return lldb::eStopReasonException;
   }
 
   const char *GetDescription() override;
 
+  struct MachException {
+    static const char *Name(uint32_t exc_type);
+    static uint32_t ExceptionCode(const char *name);
+  };
+
   // Since some mach exceptions will be reported as breakpoints, signals,
   // or trace, we use this static accessor which will translate the mach
   // exception into the correct StopInfo.
   static lldb::StopInfoSP CreateStopReasonWithMachException(
       Thread &thread, uint32_t exc_type, uint32_t exc_data_count,
       uint64_t exc_code, uint64_t exc_sub_code, uint64_t exc_sub_sub_code,
       bool pc_already_adjusted = true, bool adjust_pc_if_needed = false);
 

lldb/source/Plugins/Process/Utility/StopInfoMachException.cpp

 //===-- StopInfoMachException.cpp -----------------------------------------===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #include "StopInfoMachException.h"
 
 #include "lldb/lldb-forward.h"
 
 #if defined(__APPLE__)
 // Needed for the EXC_RESOURCE interpretation macros
 #include <kern/exc_resource.h>
+#include <mach/exception.h>
 #endif
 
 #include "lldb/Breakpoint/Watchpoint.h"
 #include "lldb/Symbol/Symbol.h"
 #include "lldb/Target/ABI.h"
 #include "lldb/Target/DynamicLoader.h"
 #include "lldb/Target/ExecutionContext.h"
 #include "lldb/Target/Process.h"
if (bp_sp && bp_sp->IsEnabled()) {
       return StopInfo::CreateStopReasonWithBreakpointSiteID(thread,
                                                             bp_sp->GetID());
     }
   }
 
   return nullptr;
 }
 
+const char *StopInfoMachException::MachException::Name(uint32_t exc_type) {
+  switch (exc_type) {
+  case EXC_BAD_ACCESS:
+    return "EXC_BAD_ACCESS";
+  case EXC_BAD_INSTRUCTION:
+    return "EXC_BAD_INSTRUCTION";
+  case EXC_ARITHMETIC:
+    return "EXC_ARITHMETIC";
+  case EXC_EMULATION:
+    return "EXC_EMULATION";
+  case EXC_SOFTWARE:
+    return "EXC_SOFTWARE";
+  case EXC_BREAKPOINT:
+    return "EXC_BREAKPOINT";
+  case EXC_SYSCALL:
+    return "EXC_SYSCALL";
+  case EXC_MACH_SYSCALL:
+    return "EXC_MACH_SYSCALL";
+  case EXC_RPC_ALERT:
+    return "EXC_RPC_ALERT";
+#ifdef EXC_CRASH
+  case EXC_CRASH:
+    return "EXC_CRASH";
+#endif
+  case EXC_RESOURCE:
+    return "EXC_RESOURCE";
+#ifdef EXC_GUARD
+  case EXC_GUARD:
+    return "EXC_GUARD";
+#endif
+#ifdef EXC_CORPSE_NOTIFY
+  case EXC_CORPSE_NOTIFY:
+    return "EXC_CORPSE_NOTIFY";
+#endif
+#ifdef EXC_CORPSE_VARIANT_BIT
+  case EXC_CORPSE_VARIANT_BIT:
+    return "EXC_CORPSE_VARIANT_BIT";
+#endif
+  default:
+    break;
+  }
+  return NULL;
+}
+
+// Returns the exception code for a given exception name.
+// 0 is not a legit code, so we return that in the case of an error.
+uint32_t StopInfoMachException::MachException::ExceptionCode(const char *name) {
+  static const char *exception_prefix = "EXC_";
+  static const int prefix_len = strlen(exception_prefix);
+
+  // All mach exceptions start with this prefix:
+  if (strstr(name, exception_prefix) != name)
+    return 0;
+
+  name += prefix_len;
+  std::string name_str = name;
+  if (name_str == "BAD_ACCESS")
+    return EXC_BAD_ACCESS;
+  if (name_str == "BAD_INSTRUCTION")
+    return EXC_BAD_INSTRUCTION;
+  if (name_str == "ARITHMETIC")
+    return EXC_ARITHMETIC;
+  if (name_str == "EMULATION")
+    return EXC_EMULATION;
+  if (name_str == "SOFTWARE")
+    return EXC_SOFTWARE;
+  if (name_str == "BREAKPOINT")
+    return EXC_BREAKPOINT;
+  if (name_str == "SYSCALL")
+    return EXC_SYSCALL;
+  if (name_str == "MACH_SYSCALL")
+    return EXC_MACH_SYSCALL;
+  if (name_str == "RPC_ALERT")
+    return EXC_RPC_ALERT;
+#ifdef EXC_CRASH
+  if (name_str == "CRASH")
+    return EXC_CRASH;
+#endif
+  if (name_str == "RESOURCE")
+    return EXC_RESOURCE;
+#ifdef EXC_GUARD
+  if (name_str == "GUARD")
+    return EXC_GUARD;
+#endif
+#ifdef EXC_CORPSE_NOTIFY
+  if (name_str == "CORPSE_NOTIFY")
+    return EXC_CORPSE_NOTIFY;
+#endif
+  return 0;
+}
+
 StopInfoSP StopInfoMachException::CreateStopReasonWithMachException(
     Thread &thread, uint32_t exc_type, uint32_t exc_data_count,
     uint64_t exc_code, uint64_t exc_sub_code, uint64_t exc_sub_sub_code,
     bool pc_already_adjusted, bool adjust_pc_if_needed) {
   if (exc_type == 0)
     return StopInfoSP();
 
   uint32_t pc_decrement = 0;

lldb/source/Plugins/Process/scripted/ScriptedThread.cpp

 //===-- ScriptedThread.cpp ------------------------------------------------===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 
 #include "ScriptedThread.h"
 
 #include "Plugins/Process/Utility/RegisterContextThreadMemory.h"
+#include "Plugins/Process/Utility/StopInfoMachException.h"
 #include "lldb/Target/OperatingSystem.h"
 #include "lldb/Target/Process.h"
 #include "lldb/Target/RegisterContext.h"
 #include "lldb/Target/StopInfo.h"
 #include "lldb/Target/Unwind.h"
 #include "lldb/Utility/DataBufferHeap.h"
 #include "lldb/Utility/LLDBLog.h"
 #include <memory>
case lldb::eStopReasonSignal: {
     llvm::StringRef description;
     data_dict->GetValueForKeyAsInteger("signal", signal,
                                        LLDB_INVALID_SIGNAL_NUMBER);
     data_dict->GetValueForKeyAsString("desc", description);
     stop_info_sp =
         StopInfo::CreateStopReasonWithSignal(*this, signal, description.data());
   } break;
   case lldb::eStopReasonException: {
-    llvm::StringRef description;
-    data_dict->GetValueForKeyAsString("desc", description);
-
+    StructuredData::Dictionary *mach_exception;
+    if (!data_dict->GetValueForKeyAsDictionary("mach_exception",
+                                               mach_exception)) {
       stop_info_sp =
-        StopInfo::CreateStopReasonWithException(*this, description.data());
+          StopInfo::CreateStopReasonWithException(*this, "EXC_BAD_ACCESS");
+    } else {
+      llvm::StringRef value;
+      mach_exception->GetValueForKeyAsString("type", value);
+      uint32_t exc_type =
+          StopInfoMachException::MachException::ExceptionCode(value.data());
+
+      StructuredData::Array *exc_rawcodes;
+      mach_exception->GetValueForKeyAsArray("rawCodes", exc_rawcodes);
+      llvm::SmallVector<uint64_t, 3> raw_codes;
+
+      auto fetch_data = [&raw_codes](StructuredData::Object *obj) {
+        if (!obj)
+          return false;
+        raw_codes.push_back(obj->GetIntegerValue());
+        return true;
+      };
+
+      exc_rawcodes->ForEach(fetch_data);
+
+      uint32_t exc_data_size = raw_codes.size();
+
+      stop_info_sp = StopInfoMachException::CreateStopReasonWithMachException(
+          *this, exc_type, exc_data_size, exc_data_size >= 1 ? raw_codes[0] : 0,
+          exc_data_size >= 2 ? raw_codes[1] : 0,
+          exc_data_size >= 3 ? raw_codes[2] : 0);
+    }
+
   } break;
   default:
     return ScriptedInterface::ErrorWithMessage<bool>(
         LLVM_PRETTY_FUNCTION,
         llvm::Twine("Unsupported stop reason type (" +
                     llvm::Twine(stop_reason_type) + llvm::Twine(")."))
             .str(),
         error, LLDBLog::Thread);

lldb/test/Shell/ScriptInterpreter/Python/Crashlog/scripted_crashlog_json.test

 # REQUIRES: python, native && target-aarch64 && system-darwin
 
 # RUN: mkdir -p %t.dir
 # RUN: yaml2obj %S/Inputs/interactive_crashlog/multithread-test.yaml > %t.dir/multithread-test
 # RUN: %lldb -o 'command script import lldb.macosx.crashlog' \
 # RUN: -o 'crashlog -a -i -t %t.dir/multithread-test %S/Inputs/interactive_crashlog/multithread-test.ips' \
 # RUN: -o "thread list" -o "bt all" 2>&1 | FileCheck %s
 
 # CHECK: "crashlog" {{.*}} commands have been installed, use the "--help" options on these commands
 
 # CHECK: (lldb) process status
 # CHECK-NEXT: Process 22511 stopped
-# CHECK-NEXT: * thread #3, stop reason = EXC_BAD_ACCESS
+# CHECK-NEXT: * thread #3, stop reason = EXC_BAD_ACCESS (code=1, address=0x0)
 # CHECK-NEXT:     frame #0: 0x0000000100ec58f4 multithread-test`bar
 
 # CHECK: (lldb) thread backtrace
-# CHECK-NEXT: * thread #3, stop reason = EXC_BAD_ACCESS
+# CHECK-NEXT: * thread #3, stop reason = EXC_BAD_ACCESS (code=1, address=0x0)
 # CHECK-NEXT:   * frame #0: 0x0000000100ec58f4 multithread-test`bar{{.*}} [artificial]
 # CHECK-NEXT:     frame #1: 0x0000000100ec591b multithread-test`foo{{.*}} [artificial]
 # CHECK-NEXT:     frame #2: 0x0000000100ec5a87 multithread-test`compute_pow{{.*}} [artificial]
 
 # CHECK: (lldb) thread list
 # CHECK-NEXT: Process 22511 stopped
 # CHECK-NEXT:   thread #1: tid = 0x23c7fe, 0x000000019cc40b84 libsystem_kernel.dylib`__ulock_wait{{.*}}, queue = 'com.apple.main-thread'
 # CHECK-NEXT:   thread #2: tid = 0x23c800, 0x000000019cc42c9c libsystem_kernel.dylib`{{.*}}
-# CHECK-NEXT: * thread #3: tid = 0x23c801, 0x0000000100ec58f4 multithread-test`bar{{.*}}, stop reason = EXC_BAD_ACCESS
+# CHECK-NEXT: * thread #3: tid = 0x23c801, 0x0000000100ec58f4 multithread-test`bar{{.*}}, stop reason = EXC_BAD_ACCESS (code=1, address=0x0)
 
 # CHECK: (lldb) bt all
 # CHECK:  thread #1, queue = 'com.apple.main-thread'
 # CHECK:    frame #{{[0-9]+}}: 0x000000019cc40b84 libsystem_kernel.dylib`__ulock_wait{{.*}} [artificial]
 # CHECK:    frame #{{[0-9]+}}: 0x0000000100ec5b3b multithread-test`main{{.*}} [artificial]
 # CHECK:    frame #{{[0-9]+}}: 0x00000002230f8da7 dyld`start{{.*}} [artificial]
 # CHECK-NEXT:  thread #2
 # CHECK-NEXT:    frame #0: 0x000000019cc42c9c libsystem_kernel.dylib`__write_nocancel{{.*}} [artificial]
 # CHECK:    frame #{{[0-9]+}}: 0x0000000100ec5957 multithread-test`call_and_wait{{.*}} [artificial]
 # CHECK:    frame #{{[0-9]+}}: 0x000000019cc7e06b libsystem_pthread.dylib`_pthread_start{{.*}} [artificial]
 # CHECK:    frame #{{[0-9]+}}: 0x000000019cc78e2b libsystem_pthread.dylib`thread_start{{.*}} [artificial]
-# CHECK-NEXT:* thread #3, stop reason = EXC_BAD_ACCESS
+# CHECK-NEXT:* thread #3, stop reason = EXC_BAD_ACCESS (code=1, address=0x0)
 # CHECK-NEXT:  * frame #0: 0x0000000100ec58f4 multithread-test`bar{{.*}} [artificial]
 # CHECK-NEXT:    frame #1: 0x0000000100ec591b multithread-test`foo{{.*}} [artificial]
 # CHECK-NEXT:    frame #2: 0x0000000100ec5a87 multithread-test`compute_pow{{.*}} [artificial]
 # CHECK:    frame #{{[0-9]+}}: 0x000000019cc7e06b libsystem_pthread.dylib`_pthread_start{{.*}} [artificial]
 # CHECK:    frame #{{[0-9]+}}: 0x000000019cc78e2b libsystem_pthread.dylib`thread_start{{.*}} [artificial]

lldb/test/Shell/ScriptInterpreter/Python/Crashlog/skipped_status_interactive_crashlog.test

 # REQUIRES: python, native && target-aarch64 && system-darwin
 
 # RUN: mkdir -p %t.dir
 # RUN: yaml2obj %S/Inputs/interactive_crashlog/multithread-test.yaml > %t.dir/multithread-test
 # RUN: %lldb -b -o 'command script import lldb.macosx.crashlog' \
 # RUN: -o 'crashlog -a -i -s -t %t.dir/multithread-test %S/Inputs/interactive_crashlog/multithread-test.ips' \
 # RUN: -o 'command source -s 0 %s' 2>&1 | FileCheck %s
 
 # CHECK: "crashlog" {{.*}} commands have been installed, use the "--help" options on these commands
 # CHECK: Note: Skipping null image
 
 process status
 # CHECK: Process 22511 stopped
-# CHECK-NEXT: * thread #3, stop reason = EXC_BAD_ACCESS
+# CHECK-NEXT: * thread #3, stop reason = EXC_BAD_ACCESS (code=1, address=0x0)
 # CHECK-NEXT:     frame #0: 0x0000000100ec58f4 multithread-test`bar
 
 thread backtrace
-# CHECK: * thread #3, stop reason = EXC_BAD_ACCESS
+# CHECK: * thread #3, stop reason = EXC_BAD_ACCESS (code=1, address=0x0)
 # CHECK-NEXT:   * frame #0: 0x0000000100ec58f4 multithread-test`bar{{.*}} [artificial]
 # CHECK-NEXT:     frame #1: 0x0000000100ec591b multithread-test`foo{{.*}} [artificial]
 # CHECK-NEXT:     frame #2: 0x0000000100ec5a87 multithread-test`compute_pow{{.*}} [artificial]
 
 thread list
 # CHECK: Process 22511 stopped
 # CHECK-NEXT:   thread #1: tid = 0x23c7fe, 0x000000019cc40b84 libsystem_kernel.dylib`__ulock_wait{{.*}}, queue = 'com.apple.main-thread'
 # CHECK-NEXT:   thread #2: tid = 0x23c800, 0x000000019cc42c9c libsystem_kernel.dylib`{{.*}}
-# CHECK-NEXT: * thread #3: tid = 0x23c801, 0x0000000100ec58f4 multithread-test`bar{{.*}}, stop reason = EXC_BAD_ACCESS
+# CHECK-NEXT: * thread #3: tid = 0x23c801, 0x0000000100ec58f4 multithread-test`bar{{.*}}, stop reason = EXC_BAD_ACCESS (code=1, address=0x0)
 
 bt all
 # CHECK:  thread #1, queue = 'com.apple.main-thread'
 # CHECK:    frame #{{[0-9]+}}: 0x000000019cc40b84 libsystem_kernel.dylib`__ulock_wait{{.*}} [artificial]
 # CHECK:    frame #{{[0-9]+}}: 0x0000000100ec5b3b multithread-test`main{{.*}} [artificial]
 # CHECK:    frame #{{[0-9]+}}: 0x00000002230f8da7 dyld`start{{.*}} [artificial]
 # CHECK-NEXT:  thread #2
 # CHECK-NEXT:    frame #0: 0x000000019cc42c9c libsystem_kernel.dylib`__write_nocancel{{.*}} [artificial]
 # CHECK:    frame #{{[0-9]+}}: 0x0000000100ec5957 multithread-test`call_and_wait{{.*}} [artificial]
 # CHECK:    frame #{{[0-9]+}}: 0x000000019cc7e06b libsystem_pthread.dylib`_pthread_start{{.*}} [artificial]
 # CHECK:    frame #{{[0-9]+}}: 0x000000019cc78e2b libsystem_pthread.dylib`thread_start{{.*}} [artificial]
-# CHECK-NEXT:* thread #3, stop reason = EXC_BAD_ACCESS
+# CHECK-NEXT:* thread #3, stop reason = EXC_BAD_ACCESS (code=1, address=0x0)
 # CHECK-NEXT:  * frame #0: 0x0000000100ec58f4 multithread-test`bar{{.*}} [artificial]
 # CHECK-NEXT:    frame #1: 0x0000000100ec591b multithread-test`foo{{.*}} [artificial]
 # CHECK-NEXT:    frame #2: 0x0000000100ec5a87 multithread-test`compute_pow{{.*}} [artificial]
 # CHECK:    frame #{{[0-9]+}}: 0x000000019cc7e06b libsystem_pthread.dylib`_pthread_start{{.*}} [artificial]
 # CHECK:    frame #{{[0-9]+}}: 0x000000019cc78e2b libsystem_pthread.dylib`thread_start{{.*}} [artificial]