From 3db2f8e9b17027592a24cf677656078b9324d816 Mon Sep 17 00:00:00 2001
From: Matt Wozniski <mwozniski@bloomberg.net>
Date: Wed, 11 Sep 2024 13:43:21 -0400
Subject: [PATCH] Fix decoding Python 3.13 stacks without symbols

The eager `isValid` check that we implemented originally fails when
symbols aren't available, as we're unable to look up `PyCode_Type` to
compare the type of `f_executable` against it.

Instead, let's treat every `f_executable` as though it's a `PyCode_Type`
without checking in advance. If it's not able to be interpreted as
a `PyCode_Type`, we'll expect a `RemoteMemCopyError` from chasing an
invalid pointer at some point, and can handle that the way that we have
been handling `!isValid`.

Signed-off-by: Matt Wozniski <mwozniski@bloomberg.net>
---
 src/pystack/_pystack/pycode.cpp  | 39 ++++++++------------------------
 src/pystack/_pystack/pycode.h    |  3 +--
 src/pystack/_pystack/pyframe.cpp |  9 +++++++-
 3 files changed, 18 insertions(+), 33 deletions(-)

diff --git a/src/pystack/_pystack/pycode.cpp b/src/pystack/_pystack/pycode.cpp
index 2420937f..8d8c9505 100644
--- a/src/pystack/_pystack/pycode.cpp
+++ b/src/pystack/_pystack/pycode.cpp
@@ -162,41 +162,11 @@ getLocationInfo(
     return location_info;
 }
 
-static bool
-isValid(const std::shared_ptr<const AbstractProcessManager>& manager, remote_addr_t addr)
-{
-    if (manager->versionIsAtLeast(3, 13)) {
-        // In Python 3.13, the frame f_executable field can be a code object or a bunch
-        // of other possible types (including None). We consider valid only the cases
-        // where it is a code object.
-        remote_addr_t pycodeobject_addr = manager->getAddressFromCache("PyCode_Type");
-        if (pycodeobject_addr == 0) {
-            Object code_obj(manager, addr);
-            if (code_obj.objectType() == Object::ObjectType::CODE) {
-                manager->registerAddressInCache("PyCode_Type", code_obj.typeAddr());
-                return true;
-            }
-            return false;
-        } else {
-            Structure<py_object_v> obj(manager, addr);
-            return obj.getField(&py_object_v::o_ob_type) == pycodeobject_addr;
-        }
-    }
-    return true;
-}
-
 CodeObject::CodeObject(
         const std::shared_ptr<const AbstractProcessManager>& manager,
         remote_addr_t addr,
         uintptr_t lasti)
 {
-    if (!isValid(manager, addr)) {
-        d_filename = "???";
-        d_scope = "???";
-        d_location_info = LocationInfo{0, 0, 0, 0};
-        d_narguments = 0;
-        return;
-    }
     LOG(DEBUG) << std::hex << std::showbase << "Copying code struct from address " << addr;
     Structure<py_code_v> code(manager, addr);
 
@@ -235,6 +205,15 @@ CodeObject::CodeObject(
             });
 }
 
+CodeObject::CodeObject(std::string filename, std::string scope, LocationInfo location_info)
+: d_filename(filename)
+, d_scope(scope)
+, d_location_info(location_info)
+, d_narguments()
+, d_varnames()
+{
+}
+
 std::string
 CodeObject::Filename() const
 {
diff --git a/src/pystack/_pystack/pycode.h b/src/pystack/_pystack/pycode.h
index 32c5531e..c4a2706b 100644
--- a/src/pystack/_pystack/pycode.h
+++ b/src/pystack/_pystack/pycode.h
@@ -27,6 +27,7 @@ class CodeObject
             const std::shared_ptr<const AbstractProcessManager>& manager,
             remote_addr_t addr,
             uintptr_t lastli);
+    CodeObject(std::string filename, std::string scope, LocationInfo location_info);
 
     // Getters
     std::string Filename() const;
@@ -41,8 +42,6 @@ class CodeObject
     std::string d_scope;
     LocationInfo d_location_info;
     int d_narguments;
-
-  private:
     std::vector<std::string> d_varnames;
 };
 }  // namespace pystack
diff --git a/src/pystack/_pystack/pyframe.cpp b/src/pystack/_pystack/pyframe.cpp
index c752ddd9..e51b43f9 100644
--- a/src/pystack/_pystack/pyframe.cpp
+++ b/src/pystack/_pystack/pyframe.cpp
@@ -71,7 +71,14 @@ FrameObject::getCode(
     } else {
         last_instruction = frame.getField(&py_frame_v::o_lasti);
     }
-    return std::make_unique<CodeObject>(manager, py_code_addr, last_instruction);
+    try {
+        return std::make_unique<CodeObject>(manager, py_code_addr, last_instruction);
+    } catch (const RemoteMemCopyError& ex) {
+        // This may not have been a code object at all, or it may have been
+        // trashed by memory corruption. Either way, indicate that we failed
+        // to understand what code this frame is running.
+        return std::make_unique<CodeObject>("???", "???", LocationInfo{0, 0, 0, 0});
+    }
 }
 
 bool