diff --git a/src/libFLAC/lpc.c b/src/libFLAC/lpc.c
index 8225cffb..a996c39a 100644
--- a/src/libFLAC/lpc.c
+++ b/src/libFLAC/lpc.c
@@ -220,9 +220,8 @@ void FLAC__lpc_compute_residual_from_qlp_coefficients(const FLAC__int32 data[],
 			sum += qlp_coeff[j] * (*(--history));
 #ifdef FLAC__OVERFLOW_DETECT
 			sumo += (FLAC__int64)qlp_coeff[j] * (FLAC__int64)(*history);
-#if defined _MSC_VER || defined __MINGW32__ /* don't know how to do 64-bit literals in VC++ */
-			if(sumo < 0) sumo = -sumo;
-			if(sumo > 2147483647)
+#if defined _MSC_VER /* don't know how to do 64-bit literals in VC++ */
+			if(sumo > 2147483647I64 || sumo < -2147483648I64)
 #else
 			if(sumo > 2147483647ll || sumo < -2147483648ll)
 #endif
@@ -304,9 +303,8 @@ void FLAC__lpc_restore_signal(const FLAC__int32 residual[], unsigned data_len, c
 			sum += qlp_coeff[j] * (*(--history));
 #ifdef FLAC__OVERFLOW_DETECT
 			sumo += (FLAC__int64)qlp_coeff[j] * (FLAC__int64)(*history);
-#if defined _MSC_VER || defined __MINGW32__ /* don't know how to do 64-bit literals in VC++ */
-			if(sumo < 0) sumo = -sumo;
-			if(sumo > 2147483647)
+#if defined _MSC_VER /* don't know how to do 64-bit literals in VC++ */
+			if(sumo > 2147483647I64 || sumo < -2147483648I64)
 #else
 			if(sumo > 2147483647ll || sumo < -2147483648ll)
 #endif