diff --git a/arch/arm/include/asm/div64.h b/arch/arm/include/asm/div64.h
index 562d5376ae7a..d3ef8e416b27 100644
--- a/arch/arm/include/asm/div64.h
+++ b/arch/arm/include/asm/div64.h
@@ -52,7 +52,12 @@ static inline uint32_t __div64_32(uint64_t *n, uint32_t base)
 
 #else
 
-static inline uint64_t __arch_xprod_64(uint64_t m, uint64_t n, bool bias)
+#ifdef CONFIG_CC_OPTIMIZE_FOR_PERFORMANCE
+static __always_inline
+#else
+static inline
+#endif
+uint64_t __arch_xprod_64(uint64_t m, uint64_t n, bool bias)
 {
 	unsigned long long res;
 	register unsigned int tmp asm("ip") = 0;
diff --git a/include/asm-generic/div64.h b/include/asm-generic/div64.h
index 5d59cf7e73d9..25e7b4b58dcf 100644
--- a/include/asm-generic/div64.h
+++ b/include/asm-generic/div64.h
@@ -134,7 +134,12 @@
  * Hoping for compile-time optimization of  conditional code.
  * Architectures may provide their own optimized assembly implementation.
  */
-static inline uint64_t __arch_xprod_64(const uint64_t m, uint64_t n, bool bias)
+#ifdef CONFIG_CC_OPTIMIZE_FOR_PERFORMANCE
+static __always_inline
+#else
+static inline
+#endif
+uint64_t __arch_xprod_64(const uint64_t m, uint64_t n, bool bias)
 {
 	uint32_t m_lo = m;
 	uint32_t m_hi = m >> 32;