Regex: Use memcpy to write/read offsets from bytecode

reinterpret_cast was undefined behaviour as we do not guarantee
that offsets are going to be stored properly aligned.

2 files changed, 18 insertions, 9 deletions

diff --git a/src/regex_impl.cc b/src/regex_impl.cc
index eacbc1a6..66426aae 100644
--- a/src/regex_impl.cc
+++ b/src/regex_impl.cc
@@ -566,7 +566,7 @@ private:
                 goto_inner_end_offsets.push_back(alloc_offset());
 
                 auto right_pos = compile_node(children[1]);
-                get_offset(offset) = right_pos;
+                set_offset(offset, right_pos);
 
                 break;
             }
@@ -611,7 +611,7 @@ private:
         }
 
         for (auto& offset : goto_inner_end_offsets)
-            get_offset(offset) =  m_program.bytecode.size();
+            set_offset(offset, m_program.bytecode.size());
 
         if (capture != -1)
         {
@@ -645,7 +645,7 @@ private:
         {
             push_op(quantifier.greedy ? CompiledRegex::Split_PrioritizeChild
                                       : CompiledRegex::Split_PrioritizeParent);
-            get_offset(alloc_offset()) = inner_pos;
+            set_offset(alloc_offset(), inner_pos);
         }
         // Write the node as an optional match for the min -> max counts
         else for (int i = std::max(1, quantifier.min); // STILL UGLY !
@@ -658,7 +658,7 @@ private:
         }
 
         for (auto offset : goto_end_offsets)
-            get_offset(offset) = m_program.bytecode.size();
+            set_offset(offset, m_program.bytecode.size());
 
         return pos;
     }
@@ -670,9 +670,9 @@ private:
         return pos;
     }
 
-    Offset& get_offset(Offset pos)
+    void set_offset(Offset pos, Offset value)
     {
-        return *reinterpret_cast<Offset*>(&m_program.bytecode[pos]);
+        memcpy(&m_program.bytecode[pos], &value, sizeof(Offset));
     }
 
     void push_op(CompiledRegex::Op op)
diff --git a/src/regex_impl.hh b/src/regex_impl.hh
index 84b90fa7..0cc1d23f 100644
--- a/src/regex_impl.hh
+++ b/src/regex_impl.hh
@@ -8,6 +8,8 @@
 #include "flags.hh"
 #include "ref_ptr.hh"
 
+#include <string.h>
+
 namespace Kakoune
 {
 
@@ -199,12 +201,12 @@ private:
                 case CompiledRegex::AnyChar:
                     return StepResult::Consumed;
                 case CompiledRegex::Jump:
-                    thread.inst = prog_start + *reinterpret_cast<const CompiledRegex::Offset*>(thread.inst);
+                    thread.inst = prog_start + get_offset(thread.inst);
                     break;
                 case CompiledRegex::Split_PrioritizeParent:
                 {
                     auto parent = thread.inst + sizeof(CompiledRegex::Offset);
-                    auto child = prog_start + *reinterpret_cast<const CompiledRegex::Offset*>(thread.inst);
+                    auto child = prog_start + get_offset(thread.inst);
                     thread.inst = parent;
                     if (thread.saves)
                         ++thread.saves->refcount;
@@ -214,7 +216,7 @@ private:
                 case CompiledRegex::Split_PrioritizeChild:
                 {
                     auto parent = thread.inst + sizeof(CompiledRegex::Offset);
-                    auto child = prog_start + *reinterpret_cast<const CompiledRegex::Offset*>(thread.inst);
+                    auto child = prog_start + get_offset(thread.inst);
                     thread.inst = child;
                     if (thread.saves)
                         ++thread.saves->refcount;
@@ -373,6 +375,13 @@ private:
             ++start;
     }
 
+    static CompiledRegex::Offset get_offset(const char* ptr)
+    {
+        CompiledRegex::Offset res;
+        memcpy(&res, ptr, sizeof(CompiledRegex::Offset));
+        return res;
+    }
+
     bool is_line_start(const Utf8It& pos) const
     {
         return (pos == m_begin and not (m_flags & RegexExecFlags::NotBeginOfLine)) or