[Lldb-commits] [lldb] r300138 - Use two LC_NOTE load commands for identifying the main binary that

Jason Molenda via lldb-commits lldb-commits at lists.llvm.org
Wed Apr 12 16:33:30 PDT 2017


Author: jmolenda
Date: Wed Apr 12 18:33:30 2017
New Revision: 300138

URL: http://llvm.org/viewvc/llvm-project?rev=300138&view=rev
Log:
Use two LC_NOTE load commands for identifying the main binary that
lldb should use when given a corefile.

This uses an LC_NOTE "main bin spec" or an LC_NOTE "kern ver str"
if they are present in a Mach-O core file.

Core files may have multiple different binaries -- different kernels,
or a mix of user process and kernel binaries -- and it can be
difficult for lldb to detect the correct one to use simply by looking
at the pages of memory.  These two new LC_NOTE load commands allow
for the correct binary to be recorded unambiguously.

<rdar://problem/20878266> 

Modified:
    lldb/trunk/include/lldb/Symbol/ObjectFile.h
    lldb/trunk/source/Plugins/ObjectFile/Mach-O/ObjectFileMachO.cpp
    lldb/trunk/source/Plugins/ObjectFile/Mach-O/ObjectFileMachO.h
    lldb/trunk/source/Plugins/Process/mach-core/ProcessMachCore.cpp

Modified: lldb/trunk/include/lldb/Symbol/ObjectFile.h
URL: http://llvm.org/viewvc/llvm-project/lldb/trunk/include/lldb/Symbol/ObjectFile.h?rev=300138&r1=300137&r2=300138&view=diff
==============================================================================
--- lldb/trunk/include/lldb/Symbol/ObjectFile.h (original)
+++ lldb/trunk/include/lldb/Symbol/ObjectFile.h Wed Apr 12 18:33:30 2017
@@ -18,6 +18,7 @@
 #include "lldb/Utility/DataExtractor.h"
 #include "lldb/Utility/Endian.h"
 #include "lldb/Utility/FileSpec.h"
+#include "lldb/Utility/UUID.h"
 #include "lldb/lldb-private.h"
 
 namespace lldb_private {
@@ -576,6 +577,32 @@ public:
       return std::string(); 
   }
 
+  //------------------------------------------------------------------
+  /// When the ObjectFile is a core file, lldb needs to locate the
+  /// "binary" in the core file.  lldb can iterate over the pages looking
+  /// for a valid binary, but some core files may have metadata 
+  /// describing where the main binary is exactly which removes ambiguity
+  /// when there are multiple binaries present in the captured memory pages.
+  ///
+  /// @param[out] address
+  ///   If the address of the binary is specified, this will be set.
+  ///   This is an address is the virtual address space of the core file
+  ///   memory segments; it is not an offset into the object file.
+  ///   If no address is available, will be set to LLDB_INVALID_ADDRESS.
+  ///
+  /// @param[out] uuid
+  ///   If the uuid of the binary is specified, this will be set.
+  ///   If no UUID is available, will be cleared.
+  ///
+  /// @return
+  ///   Returns true if either address or uuid has been set.
+  //------------------------------------------------------------------
+  virtual bool GetCorefileMainBinaryInfo (lldb::addr_t &address, UUID &uuid) {
+      address = LLDB_INVALID_ADDRESS;
+      uuid.Clear();
+      return false;
+  }
+
   virtual lldb::RegisterContextSP
   GetThreadContextAtIndex(uint32_t idx, lldb_private::Thread &thread) {
     return lldb::RegisterContextSP();

Modified: lldb/trunk/source/Plugins/ObjectFile/Mach-O/ObjectFileMachO.cpp
URL: http://llvm.org/viewvc/llvm-project/lldb/trunk/source/Plugins/ObjectFile/Mach-O/ObjectFileMachO.cpp?rev=300138&r1=300137&r2=300138&view=diff
==============================================================================
--- lldb/trunk/source/Plugins/ObjectFile/Mach-O/ObjectFileMachO.cpp (original)
+++ lldb/trunk/source/Plugins/ObjectFile/Mach-O/ObjectFileMachO.cpp Wed Apr 12 18:33:30 2017
@@ -60,6 +60,8 @@
 
 #ifndef __APPLE__
 #include "Utility/UuidCompatibility.h"
+#else
+#include <uuid/uuid.h>
 #endif
 
 #define THUMB_ADDRESS_BIT_MASK 0xfffffffffffffffeull
@@ -5354,23 +5356,67 @@ uint32_t ObjectFileMachO::GetNumThreadCo
   return m_thread_context_offsets.GetSize();
 }
 
-// The LC_IDENT load command has been obsoleted for a very
-// long time and it should not occur in Mach-O files.  But
-// if it is there, it may contain a hint about where to find
-// the main binary in a core file, so we'll use it.
 std::string ObjectFileMachO::GetIdentifierString() {
   std::string result;
   ModuleSP module_sp(GetModule());
   if (module_sp) {
     std::lock_guard<std::recursive_mutex> guard(module_sp->GetMutex());
+
+    // First, look over the load commands for an LC_NOTE load command
+    // with data_owner string "kern ver str" & use that if found.
     lldb::offset_t offset = MachHeaderSizeFromMagic(m_header.magic);
     for (uint32_t i = 0; i < m_header.ncmds; ++i) {
       const uint32_t cmd_offset = offset;
+      load_command lc;
+      if (m_data.GetU32(&offset, &lc.cmd, 2) == NULL)
+          break;
+      if (lc.cmd == LC_NOTE)
+      {
+          char data_owner[17];
+          m_data.CopyData (offset, 16, data_owner);
+          data_owner[16] = '\0';
+          offset += 16;
+          uint64_t fileoff = m_data.GetU64_unchecked (&offset);
+          uint64_t size = m_data.GetU64_unchecked (&offset);
+
+          // "kern ver str" has a uint32_t version and then a
+          // nul terminated c-string.
+          if (strcmp ("kern ver str", data_owner) == 0)
+          {
+              offset = fileoff;
+              uint32_t version;
+              if (m_data.GetU32 (&offset, &version, 1) != nullptr)
+              {
+                  if (version == 1)
+                  {
+                      uint32_t strsize = size - sizeof (uint32_t);
+                      char *buf = (char*) malloc (strsize);
+                      if (buf)
+                      {
+                          m_data.CopyData (offset, strsize, buf);
+                          buf[strsize - 1] = '\0';
+                          result = buf;
+                          if (buf)
+                              free (buf);
+                          return result;
+                      }
+                  }
+              }
+          }
+      }
+      offset = cmd_offset + lc.cmdsize;
+    }
+
+    // Second, make a pass over the load commands looking for an
+    // obsolete LC_IDENT load command.
+    offset = MachHeaderSizeFromMagic(m_header.magic);
+    for (uint32_t i = 0; i < m_header.ncmds; ++i) {
+      const uint32_t cmd_offset = offset;
       struct ident_command ident_command;
       if (m_data.GetU32(&offset, &ident_command, 2) == NULL)
         break;
       if (ident_command.cmd == LC_IDENT && ident_command.cmdsize != 0) {
-        char *buf = (char *)malloc (ident_command.cmdsize);
+        char *buf = (char *) malloc (ident_command.cmdsize);
         if (buf != nullptr 
             && m_data.CopyData (offset, ident_command.cmdsize, buf) == ident_command.cmdsize) {
           buf[ident_command.cmdsize - 1] = '\0';
@@ -5381,10 +5427,65 @@ std::string ObjectFileMachO::GetIdentifi
       }
       offset = cmd_offset + ident_command.cmdsize;
     }
+
   }
   return result;
 }
 
+bool ObjectFileMachO::GetCorefileMainBinaryInfo (addr_t &address, UUID &uuid) {
+  address = LLDB_INVALID_ADDRESS;
+  uuid.Clear();
+  ModuleSP module_sp(GetModule());
+  if (module_sp) {
+    std::lock_guard<std::recursive_mutex> guard(module_sp->GetMutex());
+    lldb::offset_t offset = MachHeaderSizeFromMagic(m_header.magic);
+    for (uint32_t i = 0; i < m_header.ncmds; ++i) {
+      const uint32_t cmd_offset = offset;
+      load_command lc;
+      if (m_data.GetU32(&offset, &lc.cmd, 2) == NULL)
+          break;
+      if (lc.cmd == LC_NOTE)
+      {
+          char data_owner[17];
+          memset (data_owner, 0, sizeof (data_owner));
+          m_data.CopyData (offset, 16, data_owner);
+          offset += 16;
+          uint64_t fileoff = m_data.GetU64_unchecked (&offset);
+          uint64_t size = m_data.GetU64_unchecked (&offset);
+
+          // "main bin spec" (main binary specification) data payload is formatted:
+          //    uint32_t version       [currently 1]
+          //    uint32_t type          [0 == unspecified, 1 == kernel, 2 == user process]
+          //    uint64_t address       [ UINT64_MAX if address not specified ]
+          //    uuid_t   uuid          [ all zero's if uuid not specified ]
+          //    uint32_t log2_pagesize [ process page size in log base 2, e.g. 4k pages are 12.  0 for unspecified ]
+
+          if (strcmp ("main bin spec", data_owner) == 0 && size >= 32)
+          {
+              offset = fileoff;
+              uint32_t version;
+              if (m_data.GetU32 (&offset, &version, 1) != nullptr && version == 1)
+              {
+                  uint32_t type = 0;
+                  uuid_t raw_uuid;
+                  uuid_clear (raw_uuid);
+
+                  if (m_data.GetU32 (&offset, &type, 1)
+                      && m_data.GetU64 (&offset, &address, 1)
+                      && m_data.CopyData (offset, sizeof (uuid_t), raw_uuid) != 0
+                      && uuid.SetBytes (raw_uuid, sizeof (uuid_t)))
+                  {
+                      return true;
+                  }
+              }
+          }
+      }
+      offset = cmd_offset + lc.cmdsize;
+    }
+  }
+  return false;
+}
+
 lldb::RegisterContextSP
 ObjectFileMachO::GetThreadContextAtIndex(uint32_t idx,
                                          lldb_private::Thread &thread) {

Modified: lldb/trunk/source/Plugins/ObjectFile/Mach-O/ObjectFileMachO.h
URL: http://llvm.org/viewvc/llvm-project/lldb/trunk/source/Plugins/ObjectFile/Mach-O/ObjectFileMachO.h?rev=300138&r1=300137&r2=300138&view=diff
==============================================================================
--- lldb/trunk/source/Plugins/ObjectFile/Mach-O/ObjectFileMachO.h (original)
+++ lldb/trunk/source/Plugins/ObjectFile/Mach-O/ObjectFileMachO.h Wed Apr 12 18:33:30 2017
@@ -20,6 +20,7 @@
 #include "lldb/Symbol/ObjectFile.h"
 #include "lldb/Utility/FileSpec.h"
 #include "lldb/Utility/SafeMachO.h"
+#include "lldb/Utility/UUID.h"
 
 //----------------------------------------------------------------------
 // This class needs to be hidden as eventually belongs in a plugin that
@@ -113,6 +114,8 @@ public:
 
   std::string GetIdentifierString() override;
 
+  bool GetCorefileMainBinaryInfo (lldb::addr_t &address, lldb_private::UUID &uuid) override;
+
   lldb::RegisterContextSP
   GetThreadContextAtIndex(uint32_t idx, lldb_private::Thread &thread) override;
 

Modified: lldb/trunk/source/Plugins/Process/mach-core/ProcessMachCore.cpp
URL: http://llvm.org/viewvc/llvm-project/lldb/trunk/source/Plugins/Process/mach-core/ProcessMachCore.cpp?rev=300138&r1=300137&r2=300138&view=diff
==============================================================================
--- lldb/trunk/source/Plugins/Process/mach-core/ProcessMachCore.cpp (original)
+++ lldb/trunk/source/Plugins/Process/mach-core/ProcessMachCore.cpp Wed Apr 12 18:33:30 2017
@@ -294,11 +294,26 @@ Error ProcessMachCore::DoLoadCore() {
 
   bool found_main_binary_definitively = false;
 
+  addr_t objfile_binary_addr;
+  UUID objfile_binary_uuid;
+  if (core_objfile->GetCorefileMainBinaryInfo (objfile_binary_addr, objfile_binary_uuid))
+  {
+    if (objfile_binary_addr != LLDB_INVALID_ADDRESS)
+    {
+        m_mach_kernel_addr = objfile_binary_addr;
+        found_main_binary_definitively = true;
+        if (log)
+            log->Printf ("ProcessMachCore::DoLoadCore: using kernel address 0x%" PRIx64
+                         " from LC_NOTE 'main bin spec' load command.", m_mach_kernel_addr);
+    }
+  }
+  
   // This checks for the presence of an LC_IDENT string in a core file;
   // LC_IDENT is very obsolete and should not be used in new code, but
   // if the load command is present, let's use the contents.
   std::string corefile_identifier = core_objfile->GetIdentifierString();
-  if (corefile_identifier.find("Darwin Kernel") != std::string::npos) {
+  if (found_main_binary_definitively == false 
+      && corefile_identifier.find("Darwin Kernel") != std::string::npos) {
       UUID uuid;
       addr_t addr = LLDB_INVALID_ADDRESS;
       if (corefile_identifier.find("UUID=") != std::string::npos) {
@@ -320,13 +335,13 @@ Error ProcessMachCore::DoLoadCore() {
           found_main_binary_definitively = true;
           if (log)
             log->Printf("ProcessMachCore::DoLoadCore: Using the kernel address 0x%" PRIx64
-                        "from LC_IDENT string '%s'", addr, corefile_identifier.c_str());
+                        " from LC_IDENT/LC_NOTE 'kern ver str' string: '%s'", addr, corefile_identifier.c_str());
       }
   }
 
-  if (found_main_binary_definitively == false &&
-      (m_dyld_addr == LLDB_INVALID_ADDRESS ||
-      m_mach_kernel_addr == LLDB_INVALID_ADDRESS)) {
+  if (found_main_binary_definitively == false
+      && (m_dyld_addr == LLDB_INVALID_ADDRESS
+          || m_mach_kernel_addr == LLDB_INVALID_ADDRESS)) {
     // We need to locate the main executable in the memory ranges
     // we have in the core file.  We need to search for both a user-process dyld
     // binary




More information about the lldb-commits mailing list