py/objint_longlong: Add arithmetic overflow checks.

Long long big integer support now raises an exception on overflow rather
than returning an undefined result.

Also adds an error when shifting by a negative value.

The new arithmetic checks are added in the misc.h header.

This work was funded through GitHub Sponsors.

Signed-off-by: Angus Gratton <angus@redyak.com.au>

py/misc.h

@@ -33,10 +33,15 @@
 #include <stdbool.h>
 #include <stdint.h>
 #include <stddef.h>
+#include <limits.h>
 
 typedef unsigned char byte;
 typedef unsigned int uint;
 
+#ifndef __has_builtin
+#define __has_builtin(x) (0)
+#endif
+
 /** generic ops *************************************************/
 
 #ifndef MIN
@@ -374,26 +379,23 @@ static inline bool mp_check(bool value) {
 static inline uint32_t mp_popcount(uint32_t x) {
     return __popcnt(x);
 }
-#else
+#else // _MSC_VER
 #define mp_clz(x) __builtin_clz(x)
 #define mp_clzl(x) __builtin_clzl(x)
 #define mp_clzll(x) __builtin_clzll(x)
 #define mp_ctz(x) __builtin_ctz(x)
 #define mp_check(x) (x)
-#if defined __has_builtin
 #if __has_builtin(__builtin_popcount)
 #define mp_popcount(x) __builtin_popcount(x)
-#endif
+#else
-#endif
-#if !defined(mp_popcount)
 static inline uint32_t mp_popcount(uint32_t x) {
     x = x - ((x >> 1) & 0x55555555);
     x = (x & 0x33333333) + ((x >> 2) & 0x33333333);
     x = (x + (x >> 4)) & 0x0F0F0F0F;
     return (x * 0x01010101) >> 24;
 }
-#endif
+#endif // __has_builtin(__builtin_popcount)
-#endif
+#endif // _MSC_VER
 
 #define MP_FIT_UNSIGNED(bits, value) (((value) & (~0U << (bits))) == 0)
 #define MP_FIT_SIGNED(bits, value) \
@@ -426,4 +428,93 @@ static inline uint32_t mp_clz_mpi(mp_int_t x) {
     #endif
 }
 
+// Overflow-checked operations for long long
+
+// Integer overflow builtins were added to GCC 5, but __has_builtin only in GCC 10
+//
+// Note that the builtins has a defined result when overflow occurs, whereas the custom
+// functions below don't update the result if an overflow would occur (to avoid UB).
+#define MP_GCC_HAS_BUILTIN_OVERFLOW (__GNUC__ >= 5)
+
+#if __has_builtin(__builtin_umulll_overflow) || MP_GCC_HAS_BUILTIN_OVERFLOW
+#define mp_mul_ull_overflow __builtin_umulll_overflow
+#else
+inline static bool mp_mul_ull_overflow(unsigned long long int x, unsigned long long int y, unsigned long long int *res) {
+    if (y > 0 && x > (ULLONG_MAX / y)) {
+        return true; // overflow
+    }
+    *res = x * y;
+    return false;
+}
+#endif
+
+#if __has_builtin(__builtin_smulll_overflow) || MP_GCC_HAS_BUILTIN_OVERFLOW
+#define mp_mul_ll_overflow __builtin_smulll_overflow
+#else
+inline static bool mp_mul_ll_overflow(long long int x, long long int y, long long int *res) {
+    bool overflow;
+
+    // Check for multiply overflow; see CERT INT32-C
+    if (x > 0) { // x is positive
+        if (y > 0) { // x and y are positive
+            overflow = (x > (LLONG_MAX / y));
+        } else { // x positive, y nonpositive
+            overflow = (y < (LLONG_MIN / x));
+        } // x positive, y nonpositive
+    } else { // x is nonpositive
+        if (y > 0) { // x is nonpositive, y is positive
+            overflow = (x < (LLONG_MIN / y));
+        } else { // x and y are nonpositive
+            overflow = (x != 0 && y < (LLONG_MAX / x));
+        } // End if x and y are nonpositive
+    } // End if x is nonpositive
+
+    if (!overflow) {
+        *res = x * y;
+    }
+
+    return overflow;
+}
+#endif
+
+#if __has_builtin(__builtin_saddll_overflow) || MP_GCC_HAS_BUILTIN_OVERFLOW
+#define mp_add_ll_overflow __builtin_saddll_overflow
+#else
+inline static bool mp_add_ll_overflow(long long int lhs, long long int rhs, long long int *res) {
+    bool overflow;
+
+    if (rhs > 0) {
+        overflow = (lhs > LLONG_MAX - rhs);
+    } else {
+        overflow = (lhs < LLONG_MIN - rhs);
+    }
+
+    if (!overflow) {
+        *res = lhs + rhs;
+    }
+
+    return overflow;
+}
+#endif
+
+#if __has_builtin(__builtin_ssubll_overflow) || MP_GCC_HAS_BUILTIN_OVERFLOW
+#define mp_sub_ll_overflow __builtin_ssubll_overflow
+#else
+inline static bool mp_sub_ll_overflow(long long int lhs, long long int rhs, long long int *res) {
+    bool overflow;
+
+    if (rhs > 0) {
+        overflow = (lhs < LLONG_MIN + rhs);
+    } else {
+        overflow = (lhs > LLONG_MAX + rhs);
+    }
+
+    if (!overflow) {
+        *res = lhs - rhs;
+    }
+
+    return overflow;
+}
+#endif
+
 #endif // MICROPY_INCLUDED_PY_MISC_H

py/objint_longlong.c

@@ -31,6 +31,7 @@
 #include "py/smallint.h"
 #include "py/objint.h"
 #include "py/runtime.h"
+#include "py/misc.h"
 
 #if MICROPY_PY_BUILTINS_FLOAT
 #include <math.h>
@@ -43,6 +44,10 @@
 const mp_obj_int_t mp_sys_maxsize_obj = {{&mp_type_int}, MP_SSIZE_MAX};
 #endif
 
+static void raise_long_long_overflow(void) {
+    mp_raise_msg(&mp_type_OverflowError, MP_ERROR_TEXT("result overflows long long storage"));
+}
+
 mp_obj_t mp_obj_int_from_bytes_impl(bool big_endian, size_t len, const byte *buf) {
     int delta = 1;
     if (!big_endian) {
@@ -120,7 +125,6 @@ mp_obj_t mp_obj_int_unary_op(mp_unary_op_t op, mp_obj_t o_in) {
         // small int if the value fits without truncation
         case MP_UNARY_OP_HASH:
             return MP_OBJ_NEW_SMALL_INT((mp_int_t)o->val);
-
         case MP_UNARY_OP_POSITIVE:
             return o_in;
         case MP_UNARY_OP_NEGATIVE:
@@ -147,6 +151,8 @@ mp_obj_t mp_obj_int_unary_op(mp_unary_op_t op, mp_obj_t o_in) {
 mp_obj_t mp_obj_int_binary_op(mp_binary_op_t op, mp_obj_t lhs_in, mp_obj_t rhs_in) {
     long long lhs_val;
     long long rhs_val;
+    bool overflow = false;
+    long long result;
 
     if (mp_obj_is_small_int(lhs_in)) {
         lhs_val = MP_OBJ_SMALL_INT_VALUE(lhs_in);
@@ -167,13 +173,16 @@ mp_obj_t mp_obj_int_binary_op(mp_binary_op_t op, mp_obj_t lhs_in, mp_obj_t rhs_i
     switch (op) {
         case MP_BINARY_OP_ADD:
         case MP_BINARY_OP_INPLACE_ADD:
-            return mp_obj_new_int_from_ll(lhs_val + rhs_val);
+            overflow = mp_add_ll_overflow(lhs_val, rhs_val, &result);
+            break;
         case MP_BINARY_OP_SUBTRACT:
         case MP_BINARY_OP_INPLACE_SUBTRACT:
-            return mp_obj_new_int_from_ll(lhs_val - rhs_val);
+            overflow = mp_sub_ll_overflow(lhs_val, rhs_val, &result);
+            break;
         case MP_BINARY_OP_MULTIPLY:
         case MP_BINARY_OP_INPLACE_MULTIPLY:
-            return mp_obj_new_int_from_ll(lhs_val * rhs_val);
+            overflow = mp_mul_ll_overflow(lhs_val, rhs_val, &result);
+            break;
         case MP_BINARY_OP_FLOOR_DIVIDE:
         case MP_BINARY_OP_INPLACE_FLOOR_DIVIDE:
             if (rhs_val == 0) {
@@ -199,9 +208,21 @@ mp_obj_t mp_obj_int_binary_op(mp_binary_op_t op, mp_obj_t lhs_in, mp_obj_t rhs_i
 
         case MP_BINARY_OP_LSHIFT:
         case MP_BINARY_OP_INPLACE_LSHIFT:
-            return mp_obj_new_int_from_ll(lhs_val << (int)rhs_val);
+            if ((int)rhs_val < 0) {
+                // negative shift not allowed
+                mp_raise_ValueError(MP_ERROR_TEXT("negative shift count"));
+            }
+            result = lhs_val << (int)rhs_val;
+            // Left-shifting of negative values is implementation defined in C, but assume compiler
+            // will give us typical 2s complement behaviour unless the value overflows
+            overflow = rhs_val > 0 && ((lhs_val >= 0 && result < lhs_val) || (lhs_val < 0 && result > lhs_val));
+            break;
         case MP_BINARY_OP_RSHIFT:
         case MP_BINARY_OP_INPLACE_RSHIFT:
+            if ((int)rhs_val < 0) {
+                // negative shift not allowed
+                mp_raise_ValueError(MP_ERROR_TEXT("negative shift count"));
+            }
             return mp_obj_new_int_from_ll(lhs_val >> (int)rhs_val);
 
         case MP_BINARY_OP_POWER:
@@ -213,18 +234,18 @@ mp_obj_t mp_obj_int_binary_op(mp_binary_op_t op, mp_obj_t lhs_in, mp_obj_t rhs_i
                 mp_raise_ValueError(MP_ERROR_TEXT("negative power with no float support"));
                 #endif
             }
-            long long ans = 1;
+            result = 1;
-            while (rhs_val > 0) {
+            while (rhs_val > 0 && !overflow) {
                 if (rhs_val & 1) {
-                    ans *= lhs_val;
+                    overflow = mp_mul_ll_overflow(result, lhs_val, &result);
                 }
-                if (rhs_val == 1) {
+                if (rhs_val == 1 || overflow) {
                     break;
                 }
                 rhs_val /= 2;
-                lhs_val *= lhs_val;
+                overflow = mp_mul_ll_overflow(lhs_val, lhs_val, &lhs_val);
             }
-            return mp_obj_new_int_from_ll(ans);
+            break;
         }
 
         case MP_BINARY_OP_LESS:
@@ -242,6 +263,12 @@ mp_obj_t mp_obj_int_binary_op(mp_binary_op_t op, mp_obj_t lhs_in, mp_obj_t rhs_i
             return MP_OBJ_NULL; // op not supported
     }
 
+    if (overflow) {
+        raise_long_long_overflow();
+    }
+
+    return mp_obj_new_int_from_ll(result);
+
 zero_division:
     mp_raise_msg(&mp_type_ZeroDivisionError, MP_ERROR_TEXT("divide by zero"));
 }
@@ -267,7 +294,7 @@ mp_obj_t mp_obj_new_int_from_ll(long long val) {
 mp_obj_t mp_obj_new_int_from_ull(unsigned long long val) {
     // TODO raise an exception if the unsigned long long won't fit
     if (val >> (sizeof(unsigned long long) * 8 - 1) != 0) {
-        mp_raise_msg(&mp_type_OverflowError, MP_ERROR_TEXT("ulonglong too large"));
+        raise_long_long_overflow();
     }
     return mp_obj_new_int_from_ll(val);
 }

tests/basics/int_64_basics.py

@@ -117,10 +117,21 @@ x = -4611686018427387904 # big
 # sys.maxsize is a constant bigint, so test it's compatible with dynamic ones
 import sys
 if hasattr(sys, "maxsize"):
-    print(sys.maxsize + 1 - 1 == sys.maxsize)
+    print(sys.maxsize - 1 + 1 == sys.maxsize)
 else:
     print(True)  # No maxsize property in this config
 
 # test extraction of big int value via mp_obj_get_int_maybe
 x = 1 << 62
 print('a' * (x + 4 - x))
+
+# negative shifts are invalid
+try:
+    print((1 << 48) >> -4)
+except ValueError as e:
+    print(e)
+
+try:
+    print((1 << 48) << -6)
+except ValueError as e:
+    print(e)

tests/extmod/uctypes_addressof.py

@@ -12,5 +12,8 @@ for i in range(8):
     print(uctypes.addressof(uctypes.bytearray_at(1 << i, 8)))
 
 # Test address that is bigger than the greatest small-int but still within the address range.
+try:
     large_addr = maxsize + 1
     print(uctypes.addressof(uctypes.bytearray_at(large_addr, 8)) == large_addr)
+except OverflowError:
+    print(True)  # systems with 64-bit bigints will overflow on the above operation