Add an internal API for converting between file and memory offsets.

Lua API.html

@@ -1763,9 +1763,17 @@ global environment, persistent between calls to the script.</p>
 <li><p class="first"><tt class="docutils literal">dfhack.internal.getVTable(name)</tt></p>
 <p>Returns the pre-extracted vtable address <tt class="docutils literal">name</tt>, or <em>nil</em>.</p>
 </li>
+<li><p class="first"><tt class="docutils literal">dfhack.internal.getImageBase()</tt></p>
+<p>Returns the mmap base of the executable.</p>
+</li>
 <li><p class="first"><tt class="docutils literal">dfhack.internal.getRebaseDelta()</tt></p>
 <p>Returns the ASLR rebase offset of the DF executable.</p>
 </li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">dfhack.internal.adjustOffset(offset[,to_file])</span></tt></p>
+<p>Returns the re-aligned offset, or <em>nil</em> if invalid.
+If <tt class="docutils literal">to_file</tt> is true, the offset is adjusted from memory to file.
+This function returns the original value everywhere except windows.</p>
+</li>
 <li><p class="first"><tt class="docutils literal">dfhack.internal.getMemRanges()</tt></p>
 <p>Returns a sequence of tables describing virtual memory ranges of the process.</p>
 </li>

Lua API.rst

@@ -1618,10 +1618,20 @@ and are only documented here for completeness:
 
   Returns the pre-extracted vtable address ``name``, or *nil*.
 
+* ``dfhack.internal.getImageBase()``
+
+  Returns the mmap base of the executable.
+
 * ``dfhack.internal.getRebaseDelta()``
 
   Returns the ASLR rebase offset of the DF executable.
 
+* ``dfhack.internal.adjustOffset(offset[,to_file])``
+
+  Returns the re-aligned offset, or *nil* if invalid.
+  If ``to_file`` is true, the offset is adjusted from memory to file.
+  This function returns the original value everywhere except windows.
+
 * ``dfhack.internal.getMemRanges()``
 
   Returns a sequence of tables describing virtual memory ranges of the process.

library/LuaApi.cpp

@@ -1718,9 +1718,11 @@ static void *checkaddr(lua_State *L, int idx, bool allow_null = false)
     return rv;
 }
 
+static uint32_t getImageBase() { return Core::getInstance().p->getBase(); }
 static int getRebaseDelta() { return Core::getInstance().vinfo->getRebaseDelta(); }
 
 static const LuaWrapper::FunctionReg dfhack_internal_module[] = {
+    WRAP(getImageBase),
     WRAP(getRebaseDelta),
     { NULL, NULL }
 };
@@ -1774,6 +1776,18 @@ static int internal_getVTable(lua_State *L)
     return 1;
 }
 
+static int internal_adjustOffset(lua_State *L)
+{
+    lua_settop(L, 2);
+    int off = luaL_checkint(L, 1);
+    int rv = Core::getInstance().p->adjustOffset(off, lua_toboolean(L, 2));
+    if (rv >= 0)
+        lua_pushinteger(L, rv);
+    else
+        lua_pushnil(L);
+    return 1;
+}
+
 static int internal_getMemRanges(lua_State *L)
 {
     std::vector<DFHack::t_memrange> ranges;
@@ -1981,6 +1995,7 @@ static const luaL_Reg dfhack_internal_funcs[] = {
     { "getAddress", internal_getAddress },
     { "setAddress", internal_setAddress },
     { "getVTable", internal_getVTable },
+    { "adjustOffset", internal_adjustOffset },
     { "getMemRanges", internal_getMemRanges },
     { "patchMemory", internal_patchMemory },
     { "patchBytes", internal_patchBytes },

library/Process-darwin.cpp

@@ -220,9 +220,14 @@ void Process::getMemRanges( vector<t_memrange> & ranges )
     }*/
 }
 
-uint32_t Process::getBase()
+uintptr_t Process::getBase()
 {
-    return 0;
+    return 0x1000000;
+}
+
+int Process::adjustOffset(int offset, bool /*to_file*/)
+{
+    return offset;
 }
 
 static int getdir (string dir, vector<string> &files)

library/Process-linux.cpp

@@ -155,9 +155,14 @@ void Process::getMemRanges( vector<t_memrange> & ranges )
     fclose(mapFile);
 }
 
-uint32_t Process::getBase()
+uintptr_t Process::getBase()
 {
-    return 0;
+    return 0x8048000;
+}
+
+int Process::adjustOffset(int offset, bool /*to_file*/)
+{
+    return offset;
 }
 
 static int getdir (string dir, vector<string> &files)

library/Process-windows.cpp

@@ -160,7 +160,7 @@ Process::Process(VersionInfoFactory * factory)
         identified = true;
         // give the process a data model and memory layout fixed for the base of first module
         my_descriptor  = new VersionInfo(*vinfo);
-        my_descriptor->rebaseTo((uint32_t)d->base);
+        my_descriptor->rebaseTo(getBase());
         for(size_t i = 0; i < threads_ids.size();i++)
         {
             HANDLE hThread = OpenThread(THREAD_ALL_ACCESS, FALSE, (DWORD) threads_ids[i]);
@@ -394,13 +394,46 @@ void Process::getMemRanges( vector<t_memrange> & ranges )
     }
 }
 
-uint32_t Process::getBase()
+uintptr_t Process::getBase()
 {
     if(d)
-        return (uint32_t) d->base;
+        return (uintptr_t) d->base;
     return 0x400000;
 }
 
+int Process::adjustOffset(int offset, bool to_file)
+{
+    if (!d)
+        return -1;
+
+    for(int i = 0; i < d->pe_header.FileHeader.NumberOfSections; i++)
+    {
+        auto &section = d->sections[i];
+
+        if (to_file)
+        {
+            unsigned delta = offset - section.VirtualAddress;
+            if (delta >= section.Misc.VirtualSize)
+                continue;
+            if (!section.PointerToRawData || delta >= section.SizeOfRawData)
+                return -1;
+            return (int)(section.PointerToRawData + delta);
+        }
+        else
+        {
+            unsigned delta = offset - section.PointerToRawData;
+            if (!section.PointerToRawData || delta >= section.SizeOfRawData)
+                continue;
+            if (delta >= section.Misc.VirtualSize)
+                return -1;
+            return (int)(section.VirtualAddress + delta);
+        }
+    }
+
+    return -1;
+}
+
+
 string Process::doReadClassName (void * vptr)
 {
     char * rtti = readPtr((char *)vptr - 0x4);

library/VersionInfoFactory.cpp

@@ -103,13 +103,13 @@ void VersionInfoFactory::ParseVersion (TiXmlElement* entry, VersionInfo* mem)
     {
         mem->setOS(OS_LINUX);
         // this is wrong... I'm not going to do base image relocation on linux though.
-        mem->setBase(0x0);
+        mem->setBase(0x8048000);
     }
     else if(os == "darwin")
     {
         mem->setOS(OS_APPLE);
         // this is wrong... I'm not going to do base image relocation on linux though.
-        mem->setBase(0x0);
+        mem->setBase(0x1000000);
     }
     else
     {

library/include/MemAccess.h

@@ -275,11 +275,13 @@ namespace DFHack
             {
                 return my_descriptor;
             };
-            uint32_t getBase();
+            uintptr_t getBase();
             /// get the DF Process ID
             int getPID();
             /// get the DF Process FilePath
             std::string getPath();
+            /// Adjust between in-memory and in-file image offset
+            int adjustOffset(int offset, bool to_file = false);
 
             /// millisecond tick count, exactly as DF uses
             uint32_t getTickCount();