Add offsets for bytes objects

This continues laying the groundwork for supporting a free-threading interpreter. Signed-off-by: Matt Wozniski <[email protected]>
bloomberg · Aug 13, 2024 · d6b1102 · d6b1102
1 parent 4b4e0bd
commit d6b1102
Show file tree

Hide file tree

Showing 4 changed files with 44 additions and 5 deletions.
diff --git a/src/pystack/_pystack/cpython/string.h b/src/pystack/_pystack/cpython/string.h
@@ -12,11 +12,13 @@ typedef uint8_t Py_UCS1;
 typedef Py_UCS4 Py_UNICODE;
 typedef Py_ssize_t Py_hash_t;
 
+namespace Python3 {
 typedef struct
 {
     PyObject_VAR_HEAD Py_hash_t ob_shash;
     char ob_sval[1];
 } PyBytesObject;
+}  // namespace Python3
 
 namespace Python2 {
 typedef struct
@@ -107,6 +109,10 @@ typedef struct
 
 }  // namespace Python3_12
 
+typedef union {
+    Python3::PyBytesObject v3;
+} PyBytesObject;
+
 typedef union {
     Python2::PyUnicodeObject v2;
     Python3::PyUnicodeObject v3;

diff --git a/src/pystack/_pystack/process.cpp b/src/pystack/_pystack/process.cpp
@@ -471,13 +471,14 @@ AbstractProcessManager::getBytesFromAddress(remote_addr_t addr) const
                    << addr;
         PyBytesObject bytes;
 
-        copyObjectFromProcess(addr, &bytes);
-
-        if ((len = bytes.ob_base.ob_size + 1) < 1) {
-            throw std::runtime_error("Incorrect size of the fetches bytes object");
+        copyMemoryFromProcess(addr, offsets().py_bytes.size, &bytes);
+        len = getField(bytes, &py_bytes_v::o_ob_size) + 1;
+        if (len < 1) {
+            throw std::runtime_error("Incorrect size of the fetched bytes object");
         }
         buffer.resize(len);
-        data_addr = (remote_addr_t)((char*)addr + offsetof(PyBytesObject, ob_sval));
+        data_addr = addr + getFieldOffset(&py_bytes_v::o_ob_sval);
+
         LOG(DEBUG) << std::hex << std::showbase << "Copying data for bytes object from address "
                    << data_addr;
         copyMemoryFromProcess(data_addr, len, buffer.data());

diff --git a/src/pystack/_pystack/version.cpp b/src/pystack/_pystack/version.cpp
@@ -297,6 +297,17 @@ py_object()
     };
 }
 
+template<class T>
+constexpr py_bytes_v
+py_bytes()
+{
+    return {
+            sizeof(T),
+            offsetof(T, ob_base.ob_size),
+            offsetof(T, ob_sval),
+    };
+}
+
 template<class T>
 constexpr py_unicode_v
 py_unicode()
@@ -410,6 +421,7 @@ python_v python_v2 = {
         py_float<PyFloatObject>(),
         py_long<_PyLongObject>(),
         {},
+        {},
         py_object<PyObject>(),
         py_type<Python2::PyTypeObject>(),
         py_code<Python2::PyCodeObject>(),
@@ -428,6 +440,7 @@ python_v python_v3_3 = {
         py_dictvalues<Python3::PyDictValuesObject>(),
         py_float<PyFloatObject>(),
         py_long<_PyLongObject>(),
+        py_bytes<Python3::PyBytesObject>(),
         py_unicode<Python3::PyUnicodeObject>(),
         py_object<PyObject>(),
         py_type<Python3_3::PyTypeObject>(),
@@ -447,6 +460,7 @@ python_v python_v3_4 = {
         py_dictvalues<Python3::PyDictValuesObject>(),
         py_float<PyFloatObject>(),
         py_long<_PyLongObject>(),
+        py_bytes<Python3::PyBytesObject>(),
         py_unicode<Python3::PyUnicodeObject>(),
         py_object<PyObject>(),
         py_type<Python3_3::PyTypeObject>(),
@@ -466,6 +480,7 @@ python_v python_v3_6 = {
         py_dictvalues<Python3::PyDictValuesObject>(),
         py_float<PyFloatObject>(),
         py_long<_PyLongObject>(),
+        py_bytes<Python3::PyBytesObject>(),
         py_unicode<Python3::PyUnicodeObject>(),
         py_object<PyObject>(),
         py_type<Python3_3::PyTypeObject>(),
@@ -485,6 +500,7 @@ python_v python_v3_7 = {
         py_dictvalues<Python3::PyDictValuesObject>(),
         py_float<PyFloatObject>(),
         py_long<_PyLongObject>(),
+        py_bytes<Python3::PyBytesObject>(),
         py_unicode<Python3::PyUnicodeObject>(),
         py_object<PyObject>(),
         py_type<Python3_3::PyTypeObject>(),
@@ -506,6 +522,7 @@ python_v python_v3_8 = {
         py_dictvalues<Python3::PyDictValuesObject>(),
         py_float<PyFloatObject>(),
         py_long<_PyLongObject>(),
+        py_bytes<Python3::PyBytesObject>(),
         py_unicode<Python3::PyUnicodeObject>(),
         py_object<PyObject>(),
         py_type<Python3_8::PyTypeObject>(),
@@ -527,6 +544,7 @@ python_v python_v3_9 = {
         py_dictvalues<Python3::PyDictValuesObject>(),
         py_float<PyFloatObject>(),
         py_long<_PyLongObject>(),
+        py_bytes<Python3::PyBytesObject>(),
         py_unicode<Python3::PyUnicodeObject>(),
         py_object<PyObject>(),
         py_type<Python3_8::PyTypeObject>(),
@@ -548,6 +566,7 @@ python_v python_v3_10 = {
         py_dictvalues<Python3::PyDictValuesObject>(),
         py_float<PyFloatObject>(),
         py_long<_PyLongObject>(),
+        py_bytes<Python3::PyBytesObject>(),
         py_unicode<Python3::PyUnicodeObject>(),
         py_object<PyObject>(),
         py_type<Python3_8::PyTypeObject>(),
@@ -569,6 +588,7 @@ python_v python_v3_11 = {
         py_dictvalues<Python3::PyDictValuesObject>(),
         py_float<PyFloatObject>(),
         py_long<_PyLongObject>(),
+        py_bytes<Python3::PyBytesObject>(),
         py_unicode<Python3::PyUnicodeObject>(),
         py_object<PyObject>(),
         py_type<Python3_8::PyTypeObject>(),
@@ -591,6 +611,7 @@ python_v python_v3_12 = {
         py_dictvalues<Python3::PyDictValuesObject>(),
         py_float<PyFloatObject>(),
         py_long<_PyLongObject>(),
+        py_bytes<Python3::PyBytesObject>(),
         py_unicode<Python3_12::PyUnicodeObject>(),
         py_object<PyObject>(),
         py_type<Python3_8::PyTypeObject>(),
@@ -613,6 +634,7 @@ python_v python_v3_13 = {
         py_dictvalues<Python3_13::PyDictValuesObject>(),
         py_float<PyFloatObject>(),
         py_long<_PyLongObject>(),
+        py_bytes<Python3::PyBytesObject>(),
         py_unicode<Python3_12::PyUnicodeObject>(),
         py_object<PyObject>(),
         py_type<Python3_8::PyTypeObject>(),

diff --git a/src/pystack/_pystack/version.h b/src/pystack/_pystack/version.h
@@ -74,6 +74,14 @@ struct py_long_v
     FieldOffset<digit[1]> o_ob_digit;
 };
 
+struct py_bytes_v
+{
+    typedef PyBytesObject Structure;
+    ssize_t size;
+    FieldOffset<Py_ssize_t> o_ob_size;
+    FieldOffset<char[1]> o_ob_sval;
+};
+
 struct py_unicode_v
 {
     typedef PyUnicodeObject Structure;
@@ -244,6 +252,7 @@ struct python_v
     py_dictvalues_v py_dictvalues;
     py_float_v py_float;
     py_long_v py_long;
+    py_bytes_v py_bytes;
     py_unicode_v py_unicode;
     py_object_v py_object;
     py_type_v py_type;
@@ -273,6 +282,7 @@ define_python_v_get_specialization(py_dictkeys);
 define_python_v_get_specialization(py_dictvalues);
 define_python_v_get_specialization(py_float);
 define_python_v_get_specialization(py_long);
+define_python_v_get_specialization(py_bytes);
 define_python_v_get_specialization(py_unicode);
 define_python_v_get_specialization(py_object);
 define_python_v_get_specialization(py_type);