[0.2.x] ~load symbols by RVA-read executable

git-svn-id: https://pykd.svn.codeplex.com/svn@80112 9b283d60-5439-405e-af05-b73fd8c4d996

pykd/dbgmem.h

@@ -5,6 +5,7 @@ namespace pykd {
 ///////////////////////////////////////////////////////////////////////////////////
 
 ULONG64 addr64( ULONG64 offset );
+HRESULT readMemoryImpl(ULONG64 offset, PVOID buffer, ULONG length, ULONG *readed = NULL, bool phyAddr = false);
 void readMemory( ULONG64 offset, PVOID buffer, ULONG length, bool phyAddr = FALSE );
 bool isVaValid( ULONG64 addr );
 bool compareMemory( ULONG64 addr1, ULONG64 addr2, ULONG length, bool phyAddr = FALSE );

pykd/dia/diacallback.h

@@ -0,0 +1,100 @@
+// 
+// DIA load callback
+// 
+
+#pragma once
+
+namespace pykd {
+
+class DiaLoadCallback2 : public IDiaLoadCallback2 {
+    int m_nRefCount;
+    std::string *m_openedSymbolFile;
+public:
+    DiaLoadCallback2(std::string *openedSymbolFile = NULL) 
+        : m_nRefCount(1), m_openedSymbolFile(openedSymbolFile)
+    {
+    }
+
+    ULONG STDMETHODCALLTYPE AddRef() {
+        m_nRefCount++;
+        return m_nRefCount;
+    }
+    ULONG STDMETHODCALLTYPE Release() {
+        const int nRefCount = (--m_nRefCount);
+        if (!nRefCount)
+            delete this;
+        return nRefCount;
+    }
+    HRESULT STDMETHODCALLTYPE QueryInterface( REFIID rid, void **ppUnk ) {
+        if ( ppUnk == NULL )
+            return E_INVALIDARG;
+
+        if (rid == __uuidof( IDiaLoadCallback2 ) )
+            *ppUnk = (IDiaLoadCallback2 *)this;
+        else if (rid == __uuidof( IDiaLoadCallback ) )
+            *ppUnk = (IDiaLoadCallback *)this;
+        else if (rid == __uuidof( IUnknown ) )
+            *ppUnk = (IUnknown *)this;
+        else
+            *ppUnk = NULL;
+
+        if ( *ppUnk != NULL ) {
+            AddRef();
+            return S_OK;
+        }
+        return E_NOINTERFACE;
+    }
+
+    HRESULT STDMETHODCALLTYPE NotifyDebugDir(
+        BOOL fExecutable, 
+        DWORD cbData,
+        BYTE data[]
+    ) 
+    {
+        DBG_UNREFERENCED_PARAMETER(fExecutable);
+        DBG_UNREFERENCED_PARAMETER(cbData);
+        DBG_UNREFERENCED_PARAMETER(data);
+        return S_OK;
+    }
+    HRESULT STDMETHODCALLTYPE NotifyOpenDBG(
+        LPCOLESTR dbgPath, 
+        HRESULT resultCode
+    )
+    {
+        if (S_OK == resultCode && m_openedSymbolFile)
+            *m_openedSymbolFile = autoBstr::asStr((BSTR)dbgPath);
+
+        return S_OK;
+    }
+
+    HRESULT STDMETHODCALLTYPE NotifyOpenPDB(
+        LPCOLESTR pdbPath, 
+        HRESULT resultCode
+    )
+    {
+        if (S_OK == resultCode && m_openedSymbolFile)
+            *m_openedSymbolFile = autoBstr::asStr((BSTR)pdbPath);
+
+        return S_OK;
+    }
+    HRESULT STDMETHODCALLTYPE RestrictRegistryAccess() {
+        return S_OK;
+    }
+    HRESULT STDMETHODCALLTYPE RestrictSymbolServerAccess() {
+      return S_OK;
+    }
+    HRESULT STDMETHODCALLTYPE RestrictOriginalPathAccess() {
+        return S_OK;
+    }
+    HRESULT STDMETHODCALLTYPE RestrictReferencePathAccess() {
+        return S_OK;
+    }
+    HRESULT STDMETHODCALLTYPE RestrictDBGAccess() {
+        return S_OK;
+    }
+    HRESULT STDMETHODCALLTYPE RestrictSystemRootAccess() {
+        return S_OK;
+    }
+};
+
+}

pykd/dia/diawrapper.cpp

@@ -3,6 +3,8 @@
 #include "dbghelp.h"
 #include "dia/diawrapper.h"
 #include "win/utils.h"
+#include "dbgengine.h"
+#include "diacallback.h"
 
 namespace pykd {
 
@@ -120,7 +122,18 @@ std::string DiaException::makeFullDesc(const std::string &desc, HRESULT hres, ID
 
 ////////////////////////////////////////////////////////////////////////////////
 
-SymbolSessionPtr  loadSymbolFile(const std::string &filePath, ULONGLONG loadBase )
+interface IDiaLoadDataImpl
+{
+    virtual ~IDiaLoadDataImpl() {}
+    virtual HRESULT load(__inout IDiaDataSource &DiaDataSource) = 0;
+};
+
+////////////////////////////////////////////////////////////////////////////////
+
+static SymbolSessionPtr loadDebugSymbols(
+    IDiaLoadDataImpl &LoadDataImpl,
+    ULONGLONG loadBase
+)
 {
     HRESULT hres;
     DiaDataSourcePtr dataSource;
@@ -130,9 +143,9 @@ SymbolSessionPtr  loadSymbolFile(const std::string &filePath, ULONGLONG loadBase
     if ( S_OK != hres )
         throw DiaException("Call ::CoCreateInstance", hres);
 
-    hres = dataSource->loadDataFromPdb( toWStr(filePath) );
+    hres = LoadDataImpl.load(*dataSource);
     if ( S_OK != hres )
-        throw DiaException("Call IDiaDataSource::loadDataFromPdb", hres);
+        throw DiaException("Call IDiaDataSource::loadDataXxx", hres);
 
     DiaSessionPtr _session;
     hres = dataSource->openSession(&_session);
@@ -151,6 +164,119 @@ SymbolSessionPtr  loadSymbolFile(const std::string &filePath, ULONGLONG loadBase
     return SymbolSessionPtr( new DiaSession( _session, _globalScope ) );
 }
 
+////////////////////////////////////////////////////////////////////////////////
+
+SymbolSessionPtr  loadSymbolFile(const std::string &filePath, ULONGLONG loadBase )
+{
+    struct LoadDataFromPdb : IDiaLoadDataImpl, boost::noncopyable
+    {
+        const std::string &m_filePath;
+        LoadDataFromPdb(const std::string &filePath) : m_filePath(filePath) {}
+
+        virtual HRESULT load(__inout IDiaDataSource &dataSource) override
+        {
+            return dataSource.loadDataFromPdb( toWStr(m_filePath) );
+        }
+    };
+
+    LoadDataFromPdb loadDataFromPdb(filePath);
+
+    return loadDebugSymbols(loadDataFromPdb, loadBase);
+}
+
+//////////////////////////////////////////////////////////////////////////////////
+
+SymbolSessionPtr loadSymbolFile(
+    __in ULONGLONG loadBase,
+    __in const std::string &executable,
+    __out std::string &loadedSymbolFile,
+    __in_opt std::string symbolSearchPath /*= std::string()*/
+)
+{
+    if (symbolSearchPath.empty())
+        symbolSearchPath = getSymbolPath();
+
+    class ReadExeAtRVACallback : public IDiaReadExeAtRVACallback
+    {
+        ULONGLONG m_loadBase;
+        int m_nRefCount;
+        CComPtr< IDiaLoadCallback2 > m_diaLoadCallback2;
+public:
+        ReadExeAtRVACallback(ULONGLONG loadBase, std::string &openedSymbolFile) 
+            : m_loadBase(loadBase), m_nRefCount(1), m_diaLoadCallback2(new DiaLoadCallback2(&openedSymbolFile))
+        {
+        }
+
+        ULONG STDMETHODCALLTYPE AddRef() {
+            m_nRefCount++;
+            return m_nRefCount;
+        }
+        ULONG STDMETHODCALLTYPE Release() {
+            const int nRefCount = (--m_nRefCount);
+            if (!nRefCount)
+                delete this;
+            return nRefCount;
+        }
+
+        HRESULT STDMETHODCALLTYPE QueryInterface( REFIID rid, void **ppUnk ) {
+            if ( ppUnk == NULL )
+                return E_INVALIDARG;
+
+            if (rid == __uuidof(IDiaReadExeAtRVACallback))
+            {
+                *ppUnk = this;
+                return S_OK;
+            }
+
+            return m_diaLoadCallback2->QueryInterface(rid, ppUnk);
+        }
+
+        virtual HRESULT STDMETHODCALLTYPE ReadExecutableAtRVA( 
+            /* [in] */ DWORD relativeVirtualAddress,
+            /* [in] */ DWORD cbData,
+            /* [out] */ DWORD *pcbData,
+            /* [size_is][out] */ BYTE *pbData
+        ) override
+        {
+            return 
+                readMemoryImpl(
+                    m_loadBase + relativeVirtualAddress,
+                    pbData,
+                    cbData,
+                    pcbData);
+        }
+    };
+
+    struct LoadDataForExeByRva : IDiaLoadDataImpl, boost::noncopyable
+    {
+        ULONGLONG m_loadBase;
+        const std::string &m_executable;
+        const std::string &m_symbolSearchPath;
+        std::string m_openedSymbolFile;
+
+        LoadDataForExeByRva(ULONGLONG loadBase, const std::string &executable, std::string symbolSearchPath)
+            : m_loadBase(loadBase), m_executable(executable), m_symbolSearchPath(symbolSearchPath)
+        {
+        }
+
+        HRESULT STDMETHODCALLTYPE load(__inout IDiaDataSource &dataSource) override
+        {
+            CComPtr< IUnknown > readExeAtRVACallback(new ReadExeAtRVACallback(m_loadBase, m_openedSymbolFile) );
+            return 
+                dataSource.loadDataForExe(
+                    toWStr(m_executable),
+                    toWStr(m_symbolSearchPath),
+                    readExeAtRVACallback);
+        }
+    };
+
+    LoadDataForExeByRva loadDataForExeByRva(loadBase, executable, symbolSearchPath);
+
+    SymbolSessionPtr symSession = loadDebugSymbols(loadDataForExeByRva, loadBase);
+    loadedSymbolFile = loadDataForExeByRva.m_openedSymbolFile;
+    return symSession;
+}
+
 //////////////////////////////////////////////////////////////////////////////////
 
 std::string getBasicTypeName( ULONG basicType )

pykd/module.cpp

@@ -22,7 +22,6 @@ Module::Module(const std::string &moduleName )
 {
     m_base = findModuleBase( moduleName );
     m_name = moduleName;
-    m_symfile = getModuleSymbolFileName( m_base );
     m_imageName = getModuleImageName( m_base );
     m_timeDataStamp = getModuleTimeStamp( m_base );
     m_checkSum = getModuleCheckSum( m_base );
@@ -35,7 +34,6 @@ Module::Module(ULONG64 offset )
 {
     m_base = findModuleBase( addr64(offset) );
     m_name = getModuleName( m_base );
-    m_symfile = getModuleSymbolFileName( m_base );
     m_imageName = getModuleImageName( m_base );
     m_timeDataStamp = getModuleTimeStamp( m_base );
     m_checkSum = getModuleCheckSum( m_base );
@@ -46,20 +44,15 @@ Module::Module(ULONG64 offset )
 
 SymbolSessionPtr& Module::getSymSession()
 {
-    do {
+    m_symSession = loadSymbolFile(m_base, m_imageName, m_symfile);
-
-        if ( m_symSession )
-            return m_symSession;
-
-        if ( m_symfile.empty() )
-            break;
-
-        m_symSession = loadSymbolFile( m_symfile, m_base );
-
-    } while( false );
 
     if ( !m_symSession )
-        throw SymbolException( "failed to find symbol file" );
+    {
+        m_symfile = getModuleSymbolFileName(m_base);
+        m_symSession = loadSymbolFile(m_symfile, m_base);
+        if (m_symSession)
+            throw SymbolException( "failed to load symbol file" );
+    }
 
     return m_symSession;
 }
@@ -75,8 +68,8 @@ SymbolPtr& Module::getSymScope()
 
 void Module::reloadSymbols()
 {
-    m_symfile = getModuleSymbolFileName( m_base );
     m_symSession.reset();
+    getSymSession();
 }
 
 /////////////////////////////////////////////////////////////////////////////////////
@@ -114,6 +107,8 @@ std::string Module::print()
 {
     std::stringstream   sstr;
 
+    prepareSymbolFile();
+
     sstr << "Module: " << m_name <<  std::endl;
     sstr << "Start: " << std::hex << m_base << " End: " << getEnd() << " Size: " << m_size << std::endl;
     sstr << "Image: " << m_imageName << std::endl;
@@ -248,6 +243,19 @@ std::string Module::getSourceFile( ULONG64 offset )
     return fileName;
 }
 
+void Module::prepareSymbolFile()
+{
+    try
+    {
+        // load symbols for modules, if need
+        getSymSession();
+    }
+    catch (const SymbolException &e)
+    {
+        DBG_UNREFERENCED_LOCAL_VARIABLE(e);
+    }
+}
+
 ///////////////////////////////////////////////////////////////////////////////////
 
 }; // end of namespace pykd

pykd/module.h

@@ -46,7 +46,8 @@ public:
         return m_size;
     }
 
-    std::string getSymFile() const {
+    std::string getSymFile() {
+        prepareSymbolFile();
         return m_symfile;
     }
 
@@ -116,6 +117,8 @@ private:
 
     ULONG getRvaByName(const std::string &symName);
 
+    void prepareSymbolFile();
+
     std::string             m_name;
     std::string             m_imageName;
     std::string             m_symfile;

pykd/pykd_2008.vcproj

@@ -1,7 +1,7 @@
 <?xml version="1.0" encoding="windows-1251"?>
 <VisualStudioProject
 	ProjectType="Visual C++"
-	Version="9.00"
+	Version="9,00"
 	Name="pykd"
 	ProjectGUID="{FE961905-666F-4908-A212-961465F46F13}"
 	RootNamespace="pykd"
@@ -609,6 +609,10 @@
 		<Filter
 			Name="dia"
 			>
+			<File
+				RelativePath=".\dia\diacallback.h"
+				>
+			</File>
 			<File
 				RelativePath=".\dia\diadata.cpp"
 				>

pykd/symengine.h

@@ -172,7 +172,14 @@ public:
 
 std::string getBasicTypeName( ULONG basicType );
 
-SymbolSessionPtr  loadSymbolFile(const std::string &filePath, ULONGLONG loadBase = 0);
+SymbolSessionPtr loadSymbolFile(const std::string &filePath, ULONGLONG loadBase = 0);
+
+SymbolSessionPtr loadSymbolFile(
+    __in ULONGLONG loadBase,
+    __in const std::string &executable,
+    __out std::string &loadedSymbolFile,
+    __in_opt std::string symbolSearchPath = std::string()
+);
 
 ////////////////////////////////////////////////////////////////////////////////
 

pykd/win/memory.cpp

@@ -65,12 +65,9 @@ bool isVaValid( ULONG64 addr )
 
 ///////////////////////////////////////////////////////////////////////////////////
 
-void readMemory( ULONG64 offset, PVOID buffer, ULONG length, bool phyAddr )
+HRESULT readMemoryImpl(ULONG64 offset, PVOID buffer, ULONG length, ULONG *readed, bool phyAddr)
 {
-    PyThread_StateRestore pyThreadRestore( g_dbgEng->pystate );
+    HRESULT hres;
-
-    HRESULT     hres;
-
     if ( phyAddr == false )
     {
         offset = addr64NoSafe( offset );
@@ -81,13 +78,24 @@ void readMemory( ULONG64 offset, PVOID buffer, ULONG length, bool phyAddr )
 
         DBG_UNREFERENCED_LOCAL_VARIABLE(nextAddress);
 
-        hres = g_dbgEng->dataspace->ReadVirtual( offset, buffer, length, NULL );
+        hres = g_dbgEng->dataspace->ReadVirtual( offset, buffer, length, readed );
     }
     else
     {
-        hres = g_dbgEng->dataspace->ReadPhysical( offset, buffer, length, NULL );
+        hres = g_dbgEng->dataspace->ReadPhysical( offset, buffer, length, readed );
     }
 
+    return hres;
+}
+
+///////////////////////////////////////////////////////////////////////////////////
+
+void readMemory( ULONG64 offset, PVOID buffer, ULONG length, bool phyAddr )
+{
+    PyThread_StateRestore pyThreadRestore( g_dbgEng->pystate );
+
+    HRESULT hres = readMemoryImpl(offset, buffer, length, NULL, phyAddr);
+
     if ( FAILED( hres ) )
         throw MemoryException( offset, phyAddr );
 }