[dpdk-dev] Clean up rte_memcpy.h file

Message ID 1429047113-11688-1-git-send-email-rkerur@gmail.com (mailing list archive)
State Superseded, archived
Headers

Commit Message

Ravi Kerur April 14, 2015, 9:31 p.m. UTC
Remove unnecessary type casting in functions.
Use loop to adjust offset during copy instead of separate invocations.

Signed-off-by: Ravi Kerur <rkerur@gmail.com>
---
 .../common/include/arch/x86/rte_memcpy.h           | 317 ++++++++++-----------
 1 file changed, 151 insertions(+), 166 deletions(-)
  

Comments

Stephen Hemminger April 15, 2015, 2:53 a.m. UTC | #1
On Tue, 14 Apr 2015 14:31:53 -0700
Ravi Kerur <rkerur@gmail.com> wrote:

> +
> +	for (i = 0; i < 2; i++)
> +		rte_mov32(dst + i * 32, src + i * 32);
>  }
Unless you force compiler to unroll the loop, it will be slower.
  
Wodkowski, PawelX April 15, 2015, 6:32 a.m. UTC | #2
On 2015-04-14 23:31, Ravi Kerur wrote:
> +
> +		for (i = 0; i < 8; i++) {
> +			ymm = _mm256_loadu_si256((const __m256i *)(src + i * 32));
> +			_mm256_storeu_si256((__m256i *)(dst + i * 32), ymm);
> +		}
> +
>   		n -= 256;
> -		ymm1 = _mm256_loadu_si256((const __m256i *)((const uint8_t *)src + 1 * 32));
> -		ymm2 = _mm256_loadu_si256((const __m256i *)((const uint8_t *)src + 2 * 32));
> -		ymm3 = _mm256_loadu_si256((const __m256i *)((const uint8_t *)src + 3 * 32));
> -		ymm4 = _mm256_loadu_si256((const __m256i *)((const uint8_t *)src + 4 * 32));
> -		ymm5 = _mm256_loadu_si256((const __m256i *)((const uint8_t *)src + 5 * 32));
> -		ymm6 = _mm256_loadu_si256((const __m256i *)((const uint8_t *)src + 6 * 32));
> -		ymm7 = _mm256_loadu_si256((const __m256i *)((const uint8_t *)src + 7 * 32));
> -		src = (const uint8_t *)src + 256;
> -		_mm256_storeu_si256((__m256i *)((uint8_t *)dst + 0 * 32), ymm0);
> -		_mm256_storeu_si256((__m256i *)((uint8_t *)dst + 1 * 32), ymm1);
> -		_mm256_storeu_si256((__m256i *)((uint8_t *)dst + 2 * 32), ymm2);
> -		_mm256_storeu_si256((__m256i *)((uint8_t *)dst + 3 * 32), ymm3);
> -		_mm256_storeu_si256((__m256i *)((uint8_t *)dst + 4 * 32), ymm4);
> -		_mm256_storeu_si256((__m256i *)((uint8_t *)dst + 5 * 32), ymm5);
> -		_mm256_storeu_si256((__m256i *)((uint8_t *)dst + 6 * 32), ymm6);
> -		_mm256_storeu_si256((__m256i *)((uint8_t *)dst + 7 * 32), ymm7);
> -		dst = (uint8_t *)dst + 256;
> +		src = src + 256;
> +		dst = dst + 256;
>   	}

Did you perform a performance test on that part?
  
Ravi Kerur April 15, 2015, 9 p.m. UTC | #3
On Tue, Apr 14, 2015 at 11:32 PM, Pawel Wodkowski <
pawelx.wodkowski@intel.com> wrote:

> On 2015-04-14 23:31, Ravi Kerur wrote:
>
>> +
>> +               for (i = 0; i < 8; i++) {
>> +                       ymm = _mm256_loadu_si256((const __m256i *)(src +
>> i * 32));
>> +                       _mm256_storeu_si256((__m256i *)(dst + i * 32),
>> ymm);
>> +               }
>> +
>>                 n -= 256;
>> -               ymm1 = _mm256_loadu_si256((const __m256i *)((const
>> uint8_t *)src + 1 * 32));
>> -               ymm2 = _mm256_loadu_si256((const __m256i *)((const
>> uint8_t *)src + 2 * 32));
>> -               ymm3 = _mm256_loadu_si256((const __m256i *)((const
>> uint8_t *)src + 3 * 32));
>> -               ymm4 = _mm256_loadu_si256((const __m256i *)((const
>> uint8_t *)src + 4 * 32));
>> -               ymm5 = _mm256_loadu_si256((const __m256i *)((const
>> uint8_t *)src + 5 * 32));
>> -               ymm6 = _mm256_loadu_si256((const __m256i *)((const
>> uint8_t *)src + 6 * 32));
>> -               ymm7 = _mm256_loadu_si256((const __m256i *)((const
>> uint8_t *)src + 7 * 32));
>> -               src = (const uint8_t *)src + 256;
>> -               _mm256_storeu_si256((__m256i *)((uint8_t *)dst + 0 * 32),
>> ymm0);
>> -               _mm256_storeu_si256((__m256i *)((uint8_t *)dst + 1 * 32),
>> ymm1);
>> -               _mm256_storeu_si256((__m256i *)((uint8_t *)dst + 2 * 32),
>> ymm2);
>> -               _mm256_storeu_si256((__m256i *)((uint8_t *)dst + 3 * 32),
>> ymm3);
>> -               _mm256_storeu_si256((__m256i *)((uint8_t *)dst + 4 * 32),
>> ymm4);
>> -               _mm256_storeu_si256((__m256i *)((uint8_t *)dst + 5 * 32),
>> ymm5);
>> -               _mm256_storeu_si256((__m256i *)((uint8_t *)dst + 6 * 32),
>> ymm6);
>> -               _mm256_storeu_si256((__m256i *)((uint8_t *)dst + 7 * 32),
>> ymm7);
>> -               dst = (uint8_t *)dst + 256;
>> +               src = src + 256;
>> +               dst = dst + 256;
>>         }
>>
>
> Did you perform a performance test on that part?
>
>
I ran "make test" which runs "memcpy perf"  results  were given in
"cover-letter". I am pasting it here again.

/**********************With changes*************************************/
Start memcpy_perf:             Success                       [00m 00s]
Memcpy performance autotest:   Success               [09m 36s] [17m 45s]
/**********************Without changes**********************************/
Start memcpy_perf:             Success                       [00m 00s]
Memcpy performance autotest:   Success               [09m 35s] [13m 57s]
  
Ravi Kerur April 15, 2015, 9:04 p.m. UTC | #4
On Tue, Apr 14, 2015 at 7:53 PM, Stephen Hemminger <
stephen@networkplumber.org> wrote:

> On Tue, 14 Apr 2015 14:31:53 -0700
> Ravi Kerur <rkerur@gmail.com> wrote:
>
> > +
> > +     for (i = 0; i < 2; i++)
> > +             rte_mov32(dst + i * 32, src + i * 32);
> >  }
> Unless you force compiler to unroll the loop, it will be slower.
>

I had done following things

1. Use sample code from Intel to make sure CPU supports those instructions.
2. Check generated code with and without loop using (gcc -O3 -m64 -S), gcc
version is 4.8.2
No difference in code generated between "loop" and "no-loop". At least I
was expecting difference in the code.

3. Run "make test" and compare "memcpy perf" results.
  

Patch

diff --git a/lib/librte_eal/common/include/arch/x86/rte_memcpy.h b/lib/librte_eal/common/include/arch/x86/rte_memcpy.h
index 6a57426..921e990 100644
--- a/lib/librte_eal/common/include/arch/x86/rte_memcpy.h
+++ b/lib/librte_eal/common/include/arch/x86/rte_memcpy.h
@@ -106,8 +106,10 @@  rte_mov32(uint8_t *dst, const uint8_t *src)
 static inline void
 rte_mov64(uint8_t *dst, const uint8_t *src)
 {
-	rte_mov32((uint8_t *)dst + 0 * 32, (const uint8_t *)src + 0 * 32);
-	rte_mov32((uint8_t *)dst + 1 * 32, (const uint8_t *)src + 1 * 32);
+	uint8_t i;
+
+	for (i = 0; i < 2; i++)
+		rte_mov32(dst + i * 32, src + i * 32);
 }
 
 /**
@@ -117,10 +119,10 @@  rte_mov64(uint8_t *dst, const uint8_t *src)
 static inline void
 rte_mov128(uint8_t *dst, const uint8_t *src)
 {
-	rte_mov32((uint8_t *)dst + 0 * 32, (const uint8_t *)src + 0 * 32);
-	rte_mov32((uint8_t *)dst + 1 * 32, (const uint8_t *)src + 1 * 32);
-	rte_mov32((uint8_t *)dst + 2 * 32, (const uint8_t *)src + 2 * 32);
-	rte_mov32((uint8_t *)dst + 3 * 32, (const uint8_t *)src + 3 * 32);
+	uint8_t i;
+
+	for (i = 0; i < 4; i++)
+		rte_mov32(dst + i * 32, src + i * 32);
 }
 
 /**
@@ -130,14 +132,10 @@  rte_mov128(uint8_t *dst, const uint8_t *src)
 static inline void
 rte_mov256(uint8_t *dst, const uint8_t *src)
 {
-	rte_mov32((uint8_t *)dst + 0 * 32, (const uint8_t *)src + 0 * 32);
-	rte_mov32((uint8_t *)dst + 1 * 32, (const uint8_t *)src + 1 * 32);
-	rte_mov32((uint8_t *)dst + 2 * 32, (const uint8_t *)src + 2 * 32);
-	rte_mov32((uint8_t *)dst + 3 * 32, (const uint8_t *)src + 3 * 32);
-	rte_mov32((uint8_t *)dst + 4 * 32, (const uint8_t *)src + 4 * 32);
-	rte_mov32((uint8_t *)dst + 5 * 32, (const uint8_t *)src + 5 * 32);
-	rte_mov32((uint8_t *)dst + 6 * 32, (const uint8_t *)src + 6 * 32);
-	rte_mov32((uint8_t *)dst + 7 * 32, (const uint8_t *)src + 7 * 32);
+	uint8_t i;
+
+	for (i = 0; i < 8; i++)
+		rte_mov32(dst + i * 32, src + i * 32);
 }
 
 /**
@@ -147,16 +145,19 @@  rte_mov256(uint8_t *dst, const uint8_t *src)
 static inline void
 rte_mov64blocks(uint8_t *dst, const uint8_t *src, size_t n)
 {
-	__m256i ymm0, ymm1;
+	__m256i ymm;
+	uint8_t i;
 
 	while (n >= 64) {
-		ymm0 = _mm256_loadu_si256((const __m256i *)((const uint8_t *)src + 0 * 32));
+
+		for (i = 0; i < 2; i++) {
+			ymm = _mm256_loadu_si256((const __m256i *)(src + i * 32));
+			_mm256_storeu_si256((__m256i *)(dst + i * 32), ymm);
+		}
+
 		n -= 64;
-		ymm1 = _mm256_loadu_si256((const __m256i *)((const uint8_t *)src + 1 * 32));
-		src = (const uint8_t *)src + 64;
-		_mm256_storeu_si256((__m256i *)((uint8_t *)dst + 0 * 32), ymm0);
-		_mm256_storeu_si256((__m256i *)((uint8_t *)dst + 1 * 32), ymm1);
-		dst = (uint8_t *)dst + 64;
+		src = src + 64;
+		dst = dst + 64;
 	}
 }
 
@@ -167,37 +168,30 @@  rte_mov64blocks(uint8_t *dst, const uint8_t *src, size_t n)
 static inline void
 rte_mov256blocks(uint8_t *dst, const uint8_t *src, size_t n)
 {
-	__m256i ymm0, ymm1, ymm2, ymm3, ymm4, ymm5, ymm6, ymm7;
+	__m256i ymm;
+	uint8_t i;
 
 	while (n >= 256) {
-		ymm0 = _mm256_loadu_si256((const __m256i *)((const uint8_t *)src + 0 * 32));
+
+		for (i = 0; i < 8; i++) {
+			ymm = _mm256_loadu_si256((const __m256i *)(src + i * 32));
+			_mm256_storeu_si256((__m256i *)(dst + i * 32), ymm);
+		}
+
 		n -= 256;
-		ymm1 = _mm256_loadu_si256((const __m256i *)((const uint8_t *)src + 1 * 32));
-		ymm2 = _mm256_loadu_si256((const __m256i *)((const uint8_t *)src + 2 * 32));
-		ymm3 = _mm256_loadu_si256((const __m256i *)((const uint8_t *)src + 3 * 32));
-		ymm4 = _mm256_loadu_si256((const __m256i *)((const uint8_t *)src + 4 * 32));
-		ymm5 = _mm256_loadu_si256((const __m256i *)((const uint8_t *)src + 5 * 32));
-		ymm6 = _mm256_loadu_si256((const __m256i *)((const uint8_t *)src + 6 * 32));
-		ymm7 = _mm256_loadu_si256((const __m256i *)((const uint8_t *)src + 7 * 32));
-		src = (const uint8_t *)src + 256;
-		_mm256_storeu_si256((__m256i *)((uint8_t *)dst + 0 * 32), ymm0);
-		_mm256_storeu_si256((__m256i *)((uint8_t *)dst + 1 * 32), ymm1);
-		_mm256_storeu_si256((__m256i *)((uint8_t *)dst + 2 * 32), ymm2);
-		_mm256_storeu_si256((__m256i *)((uint8_t *)dst + 3 * 32), ymm3);
-		_mm256_storeu_si256((__m256i *)((uint8_t *)dst + 4 * 32), ymm4);
-		_mm256_storeu_si256((__m256i *)((uint8_t *)dst + 5 * 32), ymm5);
-		_mm256_storeu_si256((__m256i *)((uint8_t *)dst + 6 * 32), ymm6);
-		_mm256_storeu_si256((__m256i *)((uint8_t *)dst + 7 * 32), ymm7);
-		dst = (uint8_t *)dst + 256;
+		src = src + 256;
+		dst = dst + 256;
 	}
 }
 
 static inline void *
-rte_memcpy(void *dst, const void *src, size_t n)
+rte_memcpy(void *_dst, const void *_src, size_t n)
 {
-	uintptr_t dstu = (uintptr_t)dst;
-	uintptr_t srcu = (uintptr_t)src;
-	void *ret = dst;
+	const uint8_t *src = (const uint8_t *)_src;
+	uint8_t *dst = (uint8_t *)_dst;
+	uintptr_t dstu = (uintptr_t)_dst;
+	uintptr_t srcu = (uintptr_t)_src;
+	void *ret = _dst;
 	size_t dstofss;
 	size_t bits;
 
@@ -230,43 +224,44 @@  rte_memcpy(void *dst, const void *src, size_t n)
 	 * Fast way when copy size doesn't exceed 512 bytes
 	 */
 	if (n <= 32) {
-		rte_mov16((uint8_t *)dst, (const uint8_t *)src);
-		rte_mov16((uint8_t *)dst - 16 + n, (const uint8_t *)src - 16 + n);
+		rte_mov16(dst, src);
+		rte_mov16(dst - 16 + n, src - 16 + n);
 		return ret;
 	}
 	if (n <= 64) {
-		rte_mov32((uint8_t *)dst, (const uint8_t *)src);
-		rte_mov32((uint8_t *)dst - 32 + n, (const uint8_t *)src - 32 + n);
+		rte_mov32(dst, src);
+		rte_mov32(dst - 32 + n, src - 32 + n);
 		return ret;
 	}
 	if (n <= 512) {
 		if (n >= 256) {
 			n -= 256;
-			rte_mov256((uint8_t *)dst, (const uint8_t *)src);
-			src = (const uint8_t *)src + 256;
-			dst = (uint8_t *)dst + 256;
+			rte_mov256(dst, src);
+			src = src + 256;
+			dst = dst + 256;
 		}
 		if (n >= 128) {
 			n -= 128;
-			rte_mov128((uint8_t *)dst, (const uint8_t *)src);
-			src = (const uint8_t *)src + 128;
-			dst = (uint8_t *)dst + 128;
+			rte_mov128(dst, src);
+			src = src + 128;
+			dst = dst + 128;
 		}
 		if (n >= 64) {
 			n -= 64;
-			rte_mov64((uint8_t *)dst, (const uint8_t *)src);
-			src = (const uint8_t *)src + 64;
-			dst = (uint8_t *)dst + 64;
+			rte_mov64(dst, src);
+			src = src + 64;
+			dst = dst + 64;
 		}
+
 COPY_BLOCK_64_BACK31:
 		if (n > 32) {
-			rte_mov32((uint8_t *)dst, (const uint8_t *)src);
-			rte_mov32((uint8_t *)dst - 32 + n, (const uint8_t *)src - 32 + n);
+			rte_mov32(dst, src);
+			rte_mov32(dst - 32 + n, src - 32 + n);
 			return ret;
 		}
-		if (n > 0) {
-			rte_mov32((uint8_t *)dst - 32 + n, (const uint8_t *)src - 32 + n);
-		}
+		if (n > 0)
+			rte_mov32(dst - 32 + n, src - 32 + n);
+
 		return ret;
 	}
 
@@ -275,21 +270,21 @@  COPY_BLOCK_64_BACK31:
 	 */
 	dstofss = 32 - ((uintptr_t)dst & 0x1F);
 	n -= dstofss;
-	rte_mov32((uint8_t *)dst, (const uint8_t *)src);
-	src = (const uint8_t *)src + dstofss;
-	dst = (uint8_t *)dst + dstofss;
+	rte_mov32(dst, src);
+	src = src + dstofss;
+	dst = dst + dstofss;
 
 	/**
 	 * Copy 256-byte blocks.
 	 * Use copy block function for better instruction order control,
 	 * which is important when load is unaligned.
 	 */
-	rte_mov256blocks((uint8_t *)dst, (const uint8_t *)src, n);
+	rte_mov256blocks(dst, src, n);
 	bits = n;
 	n = n & 255;
 	bits -= n;
-	src = (const uint8_t *)src + bits;
-	dst = (uint8_t *)dst + bits;
+	src = src + bits;
+	dst = dst + bits;
 
 	/**
 	 * Copy 64-byte blocks.
@@ -297,12 +292,12 @@  COPY_BLOCK_64_BACK31:
 	 * which is important when load is unaligned.
 	 */
 	if (n >= 64) {
-		rte_mov64blocks((uint8_t *)dst, (const uint8_t *)src, n);
+		rte_mov64blocks(dst, src, n);
 		bits = n;
 		n = n & 63;
 		bits -= n;
-		src = (const uint8_t *)src + bits;
-		dst = (uint8_t *)dst + bits;
+		src = src + bits;
+		dst = dst + bits;
 	}
 
 	/**
@@ -337,8 +332,10 @@  rte_mov16(uint8_t *dst, const uint8_t *src)
 static inline void
 rte_mov32(uint8_t *dst, const uint8_t *src)
 {
-	rte_mov16((uint8_t *)dst + 0 * 16, (const uint8_t *)src + 0 * 16);
-	rte_mov16((uint8_t *)dst + 1 * 16, (const uint8_t *)src + 1 * 16);
+	uint8_t i;
+
+	for (i = 0; i < 2; i++)
+		rte_mov16(dst + i * 16, src + i * 16);
 }
 
 /**
@@ -348,10 +345,10 @@  rte_mov32(uint8_t *dst, const uint8_t *src)
 static inline void
 rte_mov64(uint8_t *dst, const uint8_t *src)
 {
-	rte_mov16((uint8_t *)dst + 0 * 16, (const uint8_t *)src + 0 * 16);
-	rte_mov16((uint8_t *)dst + 1 * 16, (const uint8_t *)src + 1 * 16);
-	rte_mov16((uint8_t *)dst + 2 * 16, (const uint8_t *)src + 2 * 16);
-	rte_mov16((uint8_t *)dst + 3 * 16, (const uint8_t *)src + 3 * 16);
+	uint8_t i;
+
+	for (i = 0; i < 4; i++)
+		rte_mov16(dst + i * 16, src + i * 16);
 }
 
 /**
@@ -361,14 +358,10 @@  rte_mov64(uint8_t *dst, const uint8_t *src)
 static inline void
 rte_mov128(uint8_t *dst, const uint8_t *src)
 {
-	rte_mov16((uint8_t *)dst + 0 * 16, (const uint8_t *)src + 0 * 16);
-	rte_mov16((uint8_t *)dst + 1 * 16, (const uint8_t *)src + 1 * 16);
-	rte_mov16((uint8_t *)dst + 2 * 16, (const uint8_t *)src + 2 * 16);
-	rte_mov16((uint8_t *)dst + 3 * 16, (const uint8_t *)src + 3 * 16);
-	rte_mov16((uint8_t *)dst + 4 * 16, (const uint8_t *)src + 4 * 16);
-	rte_mov16((uint8_t *)dst + 5 * 16, (const uint8_t *)src + 5 * 16);
-	rte_mov16((uint8_t *)dst + 6 * 16, (const uint8_t *)src + 6 * 16);
-	rte_mov16((uint8_t *)dst + 7 * 16, (const uint8_t *)src + 7 * 16);
+	uint8_t i;
+
+	for (i = 0; i < 8; i++)
+		rte_mov16(dst + i * 16, src + i * 16);
 }
 
 /**
@@ -378,22 +371,10 @@  rte_mov128(uint8_t *dst, const uint8_t *src)
 static inline void
 rte_mov256(uint8_t *dst, const uint8_t *src)
 {
-	rte_mov16((uint8_t *)dst + 0 * 16, (const uint8_t *)src + 0 * 16);
-	rte_mov16((uint8_t *)dst + 1 * 16, (const uint8_t *)src + 1 * 16);
-	rte_mov16((uint8_t *)dst + 2 * 16, (const uint8_t *)src + 2 * 16);
-	rte_mov16((uint8_t *)dst + 3 * 16, (const uint8_t *)src + 3 * 16);
-	rte_mov16((uint8_t *)dst + 4 * 16, (const uint8_t *)src + 4 * 16);
-	rte_mov16((uint8_t *)dst + 5 * 16, (const uint8_t *)src + 5 * 16);
-	rte_mov16((uint8_t *)dst + 6 * 16, (const uint8_t *)src + 6 * 16);
-	rte_mov16((uint8_t *)dst + 7 * 16, (const uint8_t *)src + 7 * 16);
-	rte_mov16((uint8_t *)dst + 8 * 16, (const uint8_t *)src + 8 * 16);
-	rte_mov16((uint8_t *)dst + 9 * 16, (const uint8_t *)src + 9 * 16);
-	rte_mov16((uint8_t *)dst + 10 * 16, (const uint8_t *)src + 10 * 16);
-	rte_mov16((uint8_t *)dst + 11 * 16, (const uint8_t *)src + 11 * 16);
-	rte_mov16((uint8_t *)dst + 12 * 16, (const uint8_t *)src + 12 * 16);
-	rte_mov16((uint8_t *)dst + 13 * 16, (const uint8_t *)src + 13 * 16);
-	rte_mov16((uint8_t *)dst + 14 * 16, (const uint8_t *)src + 14 * 16);
-	rte_mov16((uint8_t *)dst + 15 * 16, (const uint8_t *)src + 15 * 16);
+	uint8_t i;
+
+	for (i = 0; i < 16; i++)
+		rte_mov16(dst + i * 16,   src + i * 16);
 }
 
 /**
@@ -411,48 +392,50 @@  rte_mov256(uint8_t *dst, const uint8_t *src)
 ({                                                                                                          \
     int tmp;                                                                                                \
     while (len >= 128 + 16 - offset) {                                                                      \
-        xmm0 = _mm_loadu_si128((const __m128i *)((const uint8_t *)src - offset + 0 * 16));                  \
+        xmm0 = _mm_loadu_si128((const __m128i *)(src - offset + 0 * 16));                  \
         len -= 128;                                                                                         \
-        xmm1 = _mm_loadu_si128((const __m128i *)((const uint8_t *)src - offset + 1 * 16));                  \
-        xmm2 = _mm_loadu_si128((const __m128i *)((const uint8_t *)src - offset + 2 * 16));                  \
-        xmm3 = _mm_loadu_si128((const __m128i *)((const uint8_t *)src - offset + 3 * 16));                  \
-        xmm4 = _mm_loadu_si128((const __m128i *)((const uint8_t *)src - offset + 4 * 16));                  \
-        xmm5 = _mm_loadu_si128((const __m128i *)((const uint8_t *)src - offset + 5 * 16));                  \
-        xmm6 = _mm_loadu_si128((const __m128i *)((const uint8_t *)src - offset + 6 * 16));                  \
-        xmm7 = _mm_loadu_si128((const __m128i *)((const uint8_t *)src - offset + 7 * 16));                  \
-        xmm8 = _mm_loadu_si128((const __m128i *)((const uint8_t *)src - offset + 8 * 16));                  \
-        src = (const uint8_t *)src + 128;                                                                   \
-        _mm_storeu_si128((__m128i *)((uint8_t *)dst + 0 * 16), _mm_alignr_epi8(xmm1, xmm0, offset));        \
-        _mm_storeu_si128((__m128i *)((uint8_t *)dst + 1 * 16), _mm_alignr_epi8(xmm2, xmm1, offset));        \
-        _mm_storeu_si128((__m128i *)((uint8_t *)dst + 2 * 16), _mm_alignr_epi8(xmm3, xmm2, offset));        \
-        _mm_storeu_si128((__m128i *)((uint8_t *)dst + 3 * 16), _mm_alignr_epi8(xmm4, xmm3, offset));        \
-        _mm_storeu_si128((__m128i *)((uint8_t *)dst + 4 * 16), _mm_alignr_epi8(xmm5, xmm4, offset));        \
-        _mm_storeu_si128((__m128i *)((uint8_t *)dst + 5 * 16), _mm_alignr_epi8(xmm6, xmm5, offset));        \
-        _mm_storeu_si128((__m128i *)((uint8_t *)dst + 6 * 16), _mm_alignr_epi8(xmm7, xmm6, offset));        \
-        _mm_storeu_si128((__m128i *)((uint8_t *)dst + 7 * 16), _mm_alignr_epi8(xmm8, xmm7, offset));        \
-        dst = (uint8_t *)dst + 128;                                                                         \
+        xmm1 = _mm_loadu_si128((const __m128i *)(src - offset + 1 * 16));                  \
+        xmm2 = _mm_loadu_si128((const __m128i *)(src - offset + 2 * 16));                  \
+        xmm3 = _mm_loadu_si128((const __m128i *)(src - offset + 3 * 16));                  \
+        xmm4 = _mm_loadu_si128((const __m128i *)(src - offset + 4 * 16));                  \
+        xmm5 = _mm_loadu_si128((const __m128i *)(src - offset + 5 * 16));                  \
+        xmm6 = _mm_loadu_si128((const __m128i *)(src - offset + 6 * 16));                  \
+        xmm7 = _mm_loadu_si128((const __m128i *)(src - offset + 7 * 16));                  \
+        xmm8 = _mm_loadu_si128((const __m128i *)(src - offset + 8 * 16));                  \
+        src = src + 128;	\
+        _mm_storeu_si128((__m128i *)(dst + 0 * 16), _mm_alignr_epi8(xmm1, xmm0, offset));        \
+        _mm_storeu_si128((__m128i *)(dst + 1 * 16), _mm_alignr_epi8(xmm2, xmm1, offset));        \
+        _mm_storeu_si128((__m128i *)(dst + 2 * 16), _mm_alignr_epi8(xmm3, xmm2, offset));        \
+        _mm_storeu_si128((__m128i *)(dst + 3 * 16), _mm_alignr_epi8(xmm4, xmm3, offset));        \
+        _mm_storeu_si128((__m128i *)(dst + 4 * 16), _mm_alignr_epi8(xmm5, xmm4, offset));        \
+        _mm_storeu_si128((__m128i *)(dst + 5 * 16), _mm_alignr_epi8(xmm6, xmm5, offset));        \
+        _mm_storeu_si128((__m128i *)(dst + 6 * 16), _mm_alignr_epi8(xmm7, xmm6, offset));        \
+        _mm_storeu_si128((__m128i *)(dst + 7 * 16), _mm_alignr_epi8(xmm8, xmm7, offset));        \
+        dst = dst + 128;                                                                         \
     }                                                                                                       \
     tmp = len;                                                                                              \
     len = ((len - 16 + offset) & 127) + 16 - offset;                                                        \
     tmp -= len;                                                                                             \
-    src = (const uint8_t *)src + tmp;                                                                       \
-    dst = (uint8_t *)dst + tmp;                                                                             \
+    src = src + tmp;      \
+                          \
+    dst = dst + tmp;      \
+                          \
     if (len >= 32 + 16 - offset) {                                                                          \
         while (len >= 32 + 16 - offset) {                                                                   \
-            xmm0 = _mm_loadu_si128((const __m128i *)((const uint8_t *)src - offset + 0 * 16));              \
+            xmm0 = _mm_loadu_si128((const __m128i *)(src - offset + 0 * 16));              \
             len -= 32;                                                                                      \
-            xmm1 = _mm_loadu_si128((const __m128i *)((const uint8_t *)src - offset + 1 * 16));              \
-            xmm2 = _mm_loadu_si128((const __m128i *)((const uint8_t *)src - offset + 2 * 16));              \
-            src = (const uint8_t *)src + 32;                                                                \
-            _mm_storeu_si128((__m128i *)((uint8_t *)dst + 0 * 16), _mm_alignr_epi8(xmm1, xmm0, offset));    \
-            _mm_storeu_si128((__m128i *)((uint8_t *)dst + 1 * 16), _mm_alignr_epi8(xmm2, xmm1, offset));    \
-            dst = (uint8_t *)dst + 32;                                                                      \
+            xmm1 = _mm_loadu_si128((const __m128i *)(src - offset + 1 * 16));              \
+            xmm2 = _mm_loadu_si128((const __m128i *)(src - offset + 2 * 16));              \
+            src = src + 32;                                                                \
+            _mm_storeu_si128((__m128i *)(dst + 0 * 16), _mm_alignr_epi8(xmm1, xmm0, offset));    \
+            _mm_storeu_si128((__m128i *)(dst + 1 * 16), _mm_alignr_epi8(xmm2, xmm1, offset));    \
+            dst = dst + 32;                                                                      \
         }                                                                                                   \
         tmp = len;                                                                                          \
         len = ((len - 16 + offset) & 31) + 16 - offset;                                                     \
         tmp -= len;                                                                                         \
-        src = (const uint8_t *)src + tmp;                                                                   \
-        dst = (uint8_t *)dst + tmp;                                                                         \
+        src = src + tmp;  \
+        dst = dst + tmp;  \
     }                                                                                                       \
 })
 
@@ -491,12 +474,14 @@  rte_mov256(uint8_t *dst, const uint8_t *src)
 })
 
 static inline void *
-rte_memcpy(void *dst, const void *src, size_t n)
+rte_memcpy(void *_dst, const void *_src, size_t n)
 {
 	__m128i xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, xmm8;
-	uintptr_t dstu = (uintptr_t)dst;
-	uintptr_t srcu = (uintptr_t)src;
-	void *ret = dst;
+	const uint8_t *src  = (const uint8_t *)_src;
+	uint8_t *dst = (uint8_t *)_dst;
+	uintptr_t dstu = (uintptr_t)_dst;
+	uintptr_t srcu = (uintptr_t)_src;
+	void *ret = _dst;
 	size_t dstofss;
 	size_t srcofs;
 
@@ -529,19 +514,19 @@  rte_memcpy(void *dst, const void *src, size_t n)
 	 * Fast way when copy size doesn't exceed 512 bytes
 	 */
 	if (n <= 32) {
-		rte_mov16((uint8_t *)dst, (const uint8_t *)src);
-		rte_mov16((uint8_t *)dst - 16 + n, (const uint8_t *)src - 16 + n);
+		rte_mov16(dst, src);
+		rte_mov16(dst - 16 + n, src - 16 + n);
 		return ret;
 	}
 	if (n <= 48) {
-		rte_mov32((uint8_t *)dst, (const uint8_t *)src);
-		rte_mov16((uint8_t *)dst - 16 + n, (const uint8_t *)src - 16 + n);
+		rte_mov32(dst, src);
+		rte_mov16(dst - 16 + n, src - 16 + n);
 		return ret;
 	}
 	if (n <= 64) {
-		rte_mov32((uint8_t *)dst, (const uint8_t *)src);
-		rte_mov16((uint8_t *)dst + 32, (const uint8_t *)src + 32);
-		rte_mov16((uint8_t *)dst - 16 + n, (const uint8_t *)src - 16 + n);
+		rte_mov32(dst, src);
+		rte_mov16(dst + 32, src + 32);
+		rte_mov16(dst - 16 + n, src - 16 + n);
 		return ret;
 	}
 	if (n <= 128) {
@@ -550,40 +535,40 @@  rte_memcpy(void *dst, const void *src, size_t n)
 	if (n <= 512) {
 		if (n >= 256) {
 			n -= 256;
-			rte_mov128((uint8_t *)dst, (const uint8_t *)src);
-			rte_mov128((uint8_t *)dst + 128, (const uint8_t *)src + 128);
-			src = (const uint8_t *)src + 256;
-			dst = (uint8_t *)dst + 256;
+			rte_mov128(dst, src);
+			rte_mov128(dst + 128, src + 128);
+			src = src + 256;
+			dst = dst + 256;
 		}
 COPY_BLOCK_255_BACK15:
 		if (n >= 128) {
 			n -= 128;
-			rte_mov128((uint8_t *)dst, (const uint8_t *)src);
-			src = (const uint8_t *)src + 128;
-			dst = (uint8_t *)dst + 128;
+			rte_mov128(dst, src);
+			src = src + 128;
+			dst = dst + 128;
 		}
 COPY_BLOCK_128_BACK15:
 		if (n >= 64) {
 			n -= 64;
-			rte_mov64((uint8_t *)dst, (const uint8_t *)src);
-			src = (const uint8_t *)src + 64;
-			dst = (uint8_t *)dst + 64;
+			rte_mov64(dst, src);
+			src = src + 64;
+			dst = dst + 64;
 		}
 COPY_BLOCK_64_BACK15:
 		if (n >= 32) {
 			n -= 32;
-			rte_mov32((uint8_t *)dst, (const uint8_t *)src);
-			src = (const uint8_t *)src + 32;
-			dst = (uint8_t *)dst + 32;
+			rte_mov32(dst, src);
+			src = src + 32;
+			dst = dst + 32;
 		}
 		if (n > 16) {
-			rte_mov16((uint8_t *)dst, (const uint8_t *)src);
-			rte_mov16((uint8_t *)dst - 16 + n, (const uint8_t *)src - 16 + n);
+			rte_mov16(dst, src);
+			rte_mov16(dst - 16 + n, src - 16 + n);
 			return ret;
 		}
-		if (n > 0) {
-			rte_mov16((uint8_t *)dst - 16 + n, (const uint8_t *)src - 16 + n);
-		}
+		if (n > 0)
+			rte_mov16(dst - 16 + n, src - 16 + n);
+
 		return ret;
 	}
 
@@ -595,9 +580,9 @@  COPY_BLOCK_64_BACK15:
 	 */
 	dstofss = 16 - ((uintptr_t)dst & 0x0F) + 16;
 	n -= dstofss;
-	rte_mov32((uint8_t *)dst, (const uint8_t *)src);
-	src = (const uint8_t *)src + dstofss;
-	dst = (uint8_t *)dst + dstofss;
+	rte_mov32(dst, src);
+	src = src + dstofss;
+	dst = dst + dstofss;
 	srcofs = ((uintptr_t)src & 0x0F);
 
 	/**
@@ -608,9 +593,9 @@  COPY_BLOCK_64_BACK15:
 		 * Copy 256-byte blocks
 		 */
 		for (; n >= 256; n -= 256) {
-			rte_mov256((uint8_t *)dst, (const uint8_t *)src);
-			dst = (uint8_t *)dst + 256;
-			src = (const uint8_t *)src + 256;
+			rte_mov256(dst, src);
+			dst = dst + 256;
+			src = src + 256;
 		}
 
 		/**