--- a/src/gui/painting/qdrawhelper_arm_gnu.cpp
+++ b/src/gui/painting/qdrawhelper_arm_gnu.cpp
@@ -122,6 +122,27 @@
 // -- I could not verify this on a Cortex-A8 with the above UNROLL_LOOP_WITH_PLD
 #    define PRELOAD_COND2(x,y) if (((uint)&x[i])%L2CacheLineLength == 0) preload(&x[i] + L2CacheLineLengthInInts); \
          if (((uint)&y[i])%L2CacheLineLength == 16) preload(&y[i] + L2CacheLineLengthInInts);
+
+Q_STATIC_INLINE_FUNCTION uint BYTE_MUL_ARM(uint x, uint a) {
+    switch (a) {
+    case 255:
+        return x;
+    case 0:
+        return 0;
+    default: {
+        uint t = (x & 0xff00ff) * a;
+        t = (t + ((t >> 8) & 0xff00ff) + 0x800080) >> 8;
+        t &= 0xff00ff;
+
+        x = ((x >> 8) & 0xff00ff) * a;
+        x = (x + ((x >> 8) & 0xff00ff) + 0x800080);
+        x &= 0xff00ff00;
+        x |= t;
+        return x;
+    }
+    }
+}
+
 void comp_func_SourceOver_arm(uint *dest, const uint *src, int length, uint const_alpha)
 {
     PRELOAD_INIT2(dest, src)
@@ -134,7 +155,7 @@
                     dest[i] = src[i];
                 } else {
                     uint s = src[i];
-                    dest[i] = s + BYTE_MUL(dest[i], qAlpha(~s));
+                    dest[i] = s + BYTE_MUL_ARM(dest[i], qAlpha(~s));
                 }
 
                 ++i;
@@ -143,7 +164,7 @@
         int i = 0;
         while (i < length) {
             uint s = BYTE_MUL(src[i], const_alpha);
-            dest[i] = s + BYTE_MUL(dest[i], qAlpha(~s));
+            dest[i] = s + BYTE_MUL_ARM(dest[i], qAlpha(~s));
             PRELOAD_COND2(dest, src)
             ++i;
         }