diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index 50a88aa1..f0d7f41c 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -12,8 +12,8 @@ env:
   CMAKE_REQUIRED_QUIET: OFF
   DEBIAN_FRONTEND: noninteractive
   UPX_CMAKE_BUILD_FLAGS: --verbose
-  # 2023-08-26
-  ZIG_DIST_VERSION: 0.12.0-dev.170+750998eef
+  # 2023-08-31
+  ZIG_DIST_VERSION: 0.12.0-dev.244+f4c9e19bc
 
 jobs:
   job-rebuild-and-verify-stubs:
diff --git a/.github/workflows/weekly-ci-cc-zigcc.yml b/.github/workflows/weekly-ci-cc-zigcc.yml
index 0875bde5..e2e2df71 100644
--- a/.github/workflows/weekly-ci-cc-zigcc.yml
+++ b/.github/workflows/weekly-ci-cc-zigcc.yml
@@ -10,8 +10,8 @@ on:
 env:
   CMAKE_REQUIRED_QUIET: OFF
   DEBIAN_FRONTEND: noninteractive
-  # 2023-08-26
-  ZIG_DIST_VERSION: 0.12.0-dev.170+750998eef
+  # 2023-08-31
+  ZIG_DIST_VERSION: 0.12.0-dev.244+f4c9e19bc
 
 jobs:
   job-linux-zigcc: # uses cmake + make
diff --git a/misc/testsuite/upx_testsuite_1.sh b/misc/testsuite/upx_testsuite_1.sh
index 4875abec..a75d0261 100755
--- a/misc/testsuite/upx_testsuite_1.sh
+++ b/misc/testsuite/upx_testsuite_1.sh
@@ -103,6 +103,7 @@ all_errors=
 export UPX="--prefer-ucl --no-color --no-progress"
 export UPX_DEBUG_DISABLE_GITREV_WARNING=1
 export UPX_DEBUG_DOCTEST_VERBOSE=0
+export NO_COLOR=1
 
 rm -rf ./testsuite_1
 mkdir testsuite_1 || exit 1
diff --git a/src/bele.h b/src/bele.h
index e0676dfa..ad9a492e 100644
--- a/src/bele.h
+++ b/src/bele.h
@@ -69,7 +69,7 @@ struct LE32;
 struct LE64;
 
 // Note:
-//   void is explicitly allowed, but there is no automatic pointer conversion because of template
+//   void is explicitly allowed (but there is no automatic pointer conversion because of template!)
 //   char is explicitly allowed
 //   byte is explicitly allowed
 template <class T>
@@ -161,13 +161,19 @@ static forceinline upx_uint64_t bswap64(upx_uint64_t v) noexcept { return _bytes
 #else
 
 static forceinline constexpr unsigned bswap16(unsigned v) noexcept {
+#if defined(__riscv) && __riscv_xlen == 64
+    return (unsigned) __builtin_bswap64((upx_uint64_t) v << 48);
+#else
     // return __builtin_bswap16((upx_uint16_t) (v & 0xffff));
-    // return (unsigned) __builtin_bswap64((upx_uint64_t) v << 48);
     return __builtin_bswap32(v << 16);
+#endif
 }
 static forceinline constexpr unsigned bswap32(unsigned v) noexcept {
-    // return (unsigned) __builtin_bswap64((upx_uint64_t) v << 32);
+#if defined(__riscv) && __riscv_xlen == 64
+    return (unsigned) __builtin_bswap64((upx_uint64_t) v << 32);
+#else
     return __builtin_bswap32(v);
+#endif
 }
 static forceinline constexpr upx_uint64_t bswap64(upx_uint64_t v) noexcept {
     return __builtin_bswap64(v);
diff --git a/src/check/dt_check.cpp b/src/check/dt_check.cpp
index 8f2251e3..fc4486d8 100644
--- a/src/check/dt_check.cpp
+++ b/src/check/dt_check.cpp
@@ -43,6 +43,8 @@ int upx_doctest_check(int argc, char **argv) {
     return 0;
 #else
     const char *e = getenv("UPX_DEBUG_DOCTEST_DISABLE");
+    if (!e)
+        e = getenv("UPX_DEBUG_DISABLE_DOCTEST"); // allow alternate spelling
     if (e && e[0] && strcmp(e, "0") != 0)
         return 0;
     bool minimal = true;   // don't show summary
@@ -319,7 +321,7 @@ struct TestIntegerWrap {
     static inline bool neg_eq(const T x) noexcept { return T(0) - x == x; }
 };
 
-static noinline void throwSomeValue(int x) {
+static noinline void throwSomeValue(int x) may_throw {
     if (x < 0)
         throw int(x);
     else
@@ -346,6 +348,8 @@ static noinline void check_basic_cxx_exception_handling(void (*func)(int)) noexc
 
 void upx_compiler_sanity_check(void) noexcept {
     const char *e = getenv("UPX_DEBUG_DOCTEST_DISABLE");
+    if (!e)
+        e = getenv("UPX_DEBUG_DISABLE_DOCTEST"); // allow alternate spelling
     if (e && e[0] && strcmp(e, "0") != 0) {
         // If UPX_DEBUG_DOCTEST_DISABLE is set then we don't want to throw any
         // exceptions in order to improve debugging experience.
@@ -622,4 +626,36 @@ TEST_CASE("libc snprintf") {
     CHECK_EQ(strcmp(buf, "-7.0.0.0.0.0.0.0.7.0xffffffffffffffff"), 0);
 }
 
+TEST_CASE("libc qsort") {
+    // runtime check that libc qsort() never compares identical objects
+    struct Elem {
+        upx_uint16_t id;
+        upx_uint16_t value;
+        static int compare(const void *aa, const void *bb) noexcept {
+            const Elem *a = (const Elem *) aa;
+            const Elem *b = (const Elem *) bb;
+            assert_noexcept(a->id != b->id); // check not IDENTICAL
+            return a->value < b->value ? -1 : (a->value == b->value ? 0 : 1);
+        }
+        static bool qsort_check(Elem *e, size_t n) {
+            upx_uint32_t x = n + 5381;
+            for (size_t i = 0; i < n; i++) {
+                e[i].id = (upx_uint16_t) i;
+                x = x * 33 + (i & 255);
+                e[i].value = (upx_uint16_t) x;
+            }
+            qsort(e, n, sizeof(*e), Elem::compare);
+            bool sorted_ok = true;
+            for (size_t i = 1; i < n; i++)
+                sorted_ok &= e[i - 1].value <= e[i].value;
+            return sorted_ok;
+        }
+    };
+
+    constexpr size_t N = 4096;
+    Elem e[N];
+    for (size_t n = 1; n <= N; n <<= 1)
+        CHECK(Elem::qsort_check(e, n));
+}
+
 /* vim:set ts=4 sw=4 et: */
diff --git a/src/except.h b/src/except.h
index f8e59f1b..ac7fcd42 100644
--- a/src/except.h
+++ b/src/except.h
@@ -186,32 +186,32 @@ public:
 #define NORET noinline
 #endif
 
-NORET void throwCantPack(const char *msg);
-NORET void throwCantPackExact();
-NORET void throwUnknownExecutableFormat(const char *msg = nullptr, bool warn = false);
-NORET void throwNotCompressible(const char *msg = nullptr);
-NORET void throwAlreadyPacked(const char *msg = nullptr);
-NORET void throwAlreadyPackedByUPX(const char *msg = nullptr);
-NORET void throwCantUnpack(const char *msg);
-NORET void throwNotPacked(const char *msg = nullptr);
-NORET void throwFilterException();
-NORET void throwBadLoader();
-NORET void throwChecksumError();
-NORET void throwCompressedDataViolation();
-NORET void throwInternalError(const char *msg);
-NORET void throwOutOfMemoryException(const char *msg = nullptr);
-NORET void throwIOException(const char *msg = nullptr, int e = 0);
-NORET void throwEOFException(const char *msg = nullptr, int e = 0);
+NORET void throwCantPack(const char *msg) may_throw;
+NORET void throwCantPackExact() may_throw;
+NORET void throwUnknownExecutableFormat(const char *msg = nullptr, bool warn = false) may_throw;
+NORET void throwNotCompressible(const char *msg = nullptr) may_throw;
+NORET void throwAlreadyPacked(const char *msg = nullptr) may_throw;
+NORET void throwAlreadyPackedByUPX(const char *msg = nullptr) may_throw;
+NORET void throwCantUnpack(const char *msg) may_throw;
+NORET void throwNotPacked(const char *msg = nullptr) may_throw;
+NORET void throwFilterException() may_throw;
+NORET void throwBadLoader() may_throw;
+NORET void throwChecksumError() may_throw;
+NORET void throwCompressedDataViolation() may_throw;
+NORET void throwInternalError(const char *msg) may_throw;
+NORET void throwOutOfMemoryException(const char *msg = nullptr) may_throw;
+NORET void throwIOException(const char *msg = nullptr, int e = 0) may_throw;
+NORET void throwEOFException(const char *msg = nullptr, int e = 0) may_throw;
 
 // some C++ template wizardry is needed to overload throwCantPack() for varargs
 template <class T>
 void throwCantPack(const T *, ...) DELETED_FUNCTION;
 template <>
-NORET void throwCantPack(const char *format, ...) attribute_format(1, 2);
+NORET void throwCantPack(const char *format, ...) may_throw attribute_format(1, 2);
 template <class T>
 void throwCantUnpack(const T *, ...) DELETED_FUNCTION;
 template <>
-NORET void throwCantUnpack(const char *format, ...) attribute_format(1, 2);
+NORET void throwCantUnpack(const char *format, ...) may_throw attribute_format(1, 2);
 
 #undef NORET
 
diff --git a/src/p_vmlinx.cpp b/src/p_vmlinx.cpp
index be5701a2..57170f50 100644
--- a/src/p_vmlinx.cpp
+++ b/src/p_vmlinx.cpp
@@ -1238,8 +1238,9 @@ Linker* PackVmlinuxAMD64::newLinker() const
 
 
 // instantiate instances
-template class PackVmlinuxBase<ElfClass_LE32>;
 template class PackVmlinuxBase<ElfClass_BE32>;
+// template class PackVmlinuxBase<ElfClass_BE64>; // not used
+template class PackVmlinuxBase<ElfClass_LE32>;
 template class PackVmlinuxBase<ElfClass_LE64>;
 
 /* vim:set ts=4 sw=4 et: */
diff --git a/src/packmast.cpp b/src/packmast.cpp
index aee03e60..94b4f8a9 100644
--- a/src/packmast.cpp
+++ b/src/packmast.cpp
@@ -91,7 +91,7 @@ PackMaster::~PackMaster() noexcept {
 //
 **************************************************************************/
 
-static tribool try_can_pack(PackerBase *pb, void *user) may_throw {
+static noinline tribool try_can_pack(PackerBase *pb, void *user) may_throw {
     InputFile *f = (InputFile *) user;
     try {
         pb->initPackHeader();
@@ -111,7 +111,7 @@ static tribool try_can_pack(PackerBase *pb, void *user) may_throw {
     return false;
 }
 
-static tribool try_can_unpack(PackerBase *pb, void *user) may_throw {
+static noinline tribool try_can_unpack(PackerBase *pb, void *user) may_throw {
     InputFile *f = (InputFile *) user;
     try {
         pb->initPackHeader();
diff --git a/src/packmast.h b/src/packmast.h
index 1508b39c..36958227 100644
--- a/src/packmast.h
+++ b/src/packmast.h
@@ -40,11 +40,11 @@ public:
     explicit PackMaster(InputFile *f, Options *o = nullptr) noexcept;
     ~PackMaster() noexcept;
 
-    void pack(OutputFile *fo);
-    void unpack(OutputFile *fo);
-    void test();
-    void list();
-    void fileInfo();
+    void pack(OutputFile *fo) may_throw;
+    void unpack(OutputFile *fo) may_throw;
+    void test() may_throw;
+    void list() may_throw;
+    void fileInfo() may_throw;
 
     typedef tribool (*visit_func_t)(PackerBase *pb, void *user);
     static PackerBase *visitAllPackers(visit_func_t, InputFile *f, const Options *, void *user)
@@ -54,8 +54,8 @@ private:
     OwningPointer(PackerBase) packer = nullptr; // owner
     InputFile *const fi;                        // reference, required
 
-    static PackerBase *getPacker(InputFile *f);
-    static PackerBase *getUnpacker(InputFile *f);
+    static PackerBase *getPacker(InputFile *f) may_throw;
+    static PackerBase *getUnpacker(InputFile *f) may_throw;
 
     // setup local options for each file
     Options local_options;
diff --git a/src/pefile.cpp b/src/pefile.cpp
index f5e3ed71..0dd745be 100644
--- a/src/pefile.cpp
+++ b/src/pefile.cpp
@@ -704,8 +704,9 @@ class PeFile::ImportLinker final : public ElfLinkerAMD64 {
         if (rc != 0)
             return rc;
         // What could remain?
+        // make sort order deterministic
         assert_noexcept(a->sort_id != b->sort_id);
-        return a->sort_id < b->sort_id ? -1 : 1; // make sort order deterministic
+        return a->sort_id < b->sort_id ? -1 : 1;
     }
 
     virtual void alignCode(unsigned len) override { alignWithByte(len, 0); }
diff --git a/src/util/xspan_impl.h b/src/util/xspan_impl.h
index 2ddec6f6..ea3d3bbd 100644
--- a/src/util/xspan_impl.h
+++ b/src/util/xspan_impl.h
@@ -42,13 +42,13 @@
 XSPAN_NAMESPACE_BEGIN
 
 // HINT: set env-var "UPX_DEBUG_DOCTEST_DISABLE=1" for improved debugging experience
-noinline void xspan_fail_nullptr(void);
-noinline void xspan_fail_nullbase(void);
-noinline void xspan_fail_not_same_base(void);
-noinline void xspan_fail_range_nullptr(void);
-noinline void xspan_fail_range_nullbase(void);
-noinline void xspan_fail_range_range(void);
-void xspan_check_range(const void *ptr, const void *base, ptrdiff_t size_in_bytes);
+noinline void xspan_fail_nullptr(void) may_throw;
+noinline void xspan_fail_nullbase(void) may_throw;
+noinline void xspan_fail_not_same_base(void) may_throw;
+noinline void xspan_fail_range_nullptr(void) may_throw;
+noinline void xspan_fail_range_nullbase(void) may_throw;
+noinline void xspan_fail_range_range(void) may_throw;
+void xspan_check_range(const void *ptr, const void *base, ptrdiff_t size_in_bytes) may_throw;
 
 // help constructor to distinguish between number of elements and bytes
 struct XSpanCount {