From 08d222fe084c5e3f5e65013e0e976f29fc5d2c4f Mon Sep 17 00:00:00 2001
From: Niklas Uhl <uhl@kit.edu>
Date: Thu, 4 Dec 2025 16:21:02 +0100
Subject: [PATCH 1/3] Avoid UB in MSB computation in feistel permutation.

While this has no impact on the actual implementation, the missing
check resulted in a failed assertion

https://github.com/KarlsruheGraphGeneration/KaGen/blob/1118666271f36a8c71018768e2de459f0e5ff1c3/kagen/tools/random_permutation.h#L297

When the left half is all zeros, `most_significant_bit_set` is
undefined and may return arbitrary large values, making this assertion
fail in debug builds.
---
 kagen/tools/random_permutation.h | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/kagen/tools/random_permutation.h b/kagen/tools/random_permutation.h
index dead687..1c3faa2 100644
--- a/kagen/tools/random_permutation.h
+++ b/kagen/tools/random_permutation.h
@@ -84,6 +84,11 @@ std::uint8_t most_significant_bit_set(const Data arg) {
     constexpr std::size_t arg_width = std::numeric_limits<Data>::digits;
     auto                  log2      = static_cast<Data>(arg_width);
 
+    // __builtin_clz(l) is undefined for 0
+    if (arg == 0) {
+      return 0;
+    }
+
     if constexpr (arg_width == std::numeric_limits<unsigned int>::digits) {
         log2 -= __builtin_clz(arg);
     } else {

From ec7c0af481afb65eef614857a9244fa64d09ae9b Mon Sep 17 00:00:00 2001
From: Niklas Uhl <uhl@kit.edu>
Date: Thu, 4 Dec 2025 17:14:12 +0100
Subject: [PATCH 2/3] Make msb function less easy to misuse.

---
 kagen/tools/random_permutation.h | 27 +++++++++++++++++----------
 1 file changed, 17 insertions(+), 10 deletions(-)

diff --git a/kagen/tools/random_permutation.h b/kagen/tools/random_permutation.h
index 1c3faa2..d4e44dc 100644
--- a/kagen/tools/random_permutation.h
+++ b/kagen/tools/random_permutation.h
@@ -79,24 +79,31 @@ inline uint8_t most_significant_bit_set(Data bytes) {
     asm("bsr %1,%0" : "=r"(msb) : "r"(bytes));
     return asserting_cast<uint8_t>(msb);
 }*/
-template <typename Data>
+template <
+    typename Data, typename Enable = std::enable_if_t<
+                       std::is_same_v<Data, unsigned int> || std::is_same_v<Data, unsigned long>
+                       || std::is_same_v<Data, unsigned long long>>>
 std::uint8_t most_significant_bit_set(const Data arg) {
-    constexpr std::size_t arg_width = std::numeric_limits<Data>::digits;
-    auto                  log2      = static_cast<Data>(arg_width);
+    constexpr std::size_t width = std::numeric_limits<Data>::digits;
+
 
-    // __builtin_clz(l) is undefined for 0
+    // __builtin_clz* is undefined for 0
     if (arg == 0) {
       return 0;
     }
 
-    if constexpr (arg_width == std::numeric_limits<unsigned int>::digits) {
-        log2 -= __builtin_clz(arg);
-    } else {
-        static_assert(arg_width == std::numeric_limits<unsigned long>::digits, "unsupported data type width");
+    std::size_t leading_zeros = 0;
+
+    if constexpr (std::is_same_v<Data, unsigned int>) {
+        leading_zeros = __builtin_clz(arg);
+    } else if constexpr (std::is_same_v<Data, unsigned long>) {
         log2 -= __builtin_clzl(arg);
+        leading_zeros = __builtin_clzl(arg);
+    } else {
+        static_assert(std::is_same_v<Data, unsigned long long>);
+        leading_zeros = __builtin_clzll(arg);
     }
-
-    return log2 - 1;
+    return asserting_cast<std::uint8_t>(width - leading_zeros - 1);
 }
 
 #define UNUSED(expr) (void)(expr)

From f0a08821c082ae7601fce43595b16cfc4ea6e2e0 Mon Sep 17 00:00:00 2001
From: Niklas Uhl <uhl@kit.edu>
Date: Thu, 4 Dec 2025 17:17:08 +0100
Subject: [PATCH 3/3] fixup! Make msb function less easy to misuse.

---
 kagen/tools/random_permutation.h | 1 -
 1 file changed, 1 deletion(-)

diff --git a/kagen/tools/random_permutation.h b/kagen/tools/random_permutation.h
index d4e44dc..c4ea223 100644
--- a/kagen/tools/random_permutation.h
+++ b/kagen/tools/random_permutation.h
@@ -97,7 +97,6 @@ std::uint8_t most_significant_bit_set(const Data arg) {
     if constexpr (std::is_same_v<Data, unsigned int>) {
         leading_zeros = __builtin_clz(arg);
     } else if constexpr (std::is_same_v<Data, unsigned long>) {
-        log2 -= __builtin_clzl(arg);
         leading_zeros = __builtin_clzl(arg);
     } else {
         static_assert(std::is_same_v<Data, unsigned long long>);