diff --git a/repos/os/src/lib/blit/arm/blit_helper.h b/repos/os/src/lib/blit/arm/blit_helper.h
index 33e86d510c..4e10e865c0 100644
--- a/repos/os/src/lib/blit/arm/blit_helper.h
+++ b/repos/os/src/lib/blit/arm/blit_helper.h
@@ -37,9 +37,9 @@ static inline void copy_16bit_column(char const *src, int src_w,
  * \param src_w  width of source buffer in bytes
  * \param dst_w  width of destination buffer in bytes
  */
-static void copy_block_32bit(char const *src, int src_w,
-                             char *dst, int dst_w,
-                             int w, int h)
+static inline void copy_block_32bit(char const *src, int src_w,
+                                    char *dst, int dst_w,
+                                    int w, int h)
 {
 	src_w -= w*4;
 	dst_w -= w*4;
@@ -60,17 +60,28 @@ static inline void copy_block_32byte(char const *src, int src_w,
                                      char *dst, int dst_w,
                                      int w, int h)
 {
-	if (((long)src & 3) || ((long)dst & 3))
-		copy_block_32bit(src, src_w, dst, dst_w, w*8, h);
-	else {
-		src_w -= w*32;
-		dst_w -= w*32;
-		for (; h--; src += src_w, dst += dst_w)
-			for (int i = w; i--;)
+	for (; h > 0; h--) {
+		/*
+		 * Depending on 'src_w' and 'dst_w', some lines may be properly aligned,
+		 * while others may be not, so we need to check each line.
+		 */
+		if (((long)src & 3) || ((long)dst & 3)) {
+			copy_block_32bit(src, src_w, dst, dst_w, w*8, 1);
+			src += src_w;
+			dst += dst_w;
+		} else {
+			for (int i = w; i > 0; i--)
 				asm volatile ("ldmia %0!, {r3 - r10} \n\t"
 				              "stmia %1!, {r3 - r10} \n\t"
 				              : "+r" (src), "+r" (dst)
 				              :: "r3","r4","r5","r6","r7","r8","r9","r10");
+			/*
+			 * 'src' and 'dst' got auto-incremented by the copy code, so only
+			 * the remainder needs to get added
+			 */
+			src += (src_w - w*32);
+			dst += (dst_w - w*32);
+		}
 	}
 }