From 684edf21732316a11217966c1aa143133b4d15b9 Mon Sep 17 00:00:00 2001
From: Andrea Mazzoleni <amadvance@gmail.com>
Date: Sat, 28 Dec 2013 12:04:16 +0100
Subject: [PATCH] Updates the coding style of the RAID engine.

---
 portable.h      |   7 ++
 raid/cpu.h      |  12 +-
 raid/gf.h       |   8 +-
 raid/int.c      | 212 ++++++++++++++++----------------
 raid/internal.h | 117 +++++++++---------
 raid/intz.c     |  28 ++---
 raid/mktables.c |  79 ++++++------
 raid/module.c   |  30 ++---
 raid/raid.c     | 114 +++++++++---------
 raid/raid.h     |  23 ++--
 raid/tables.c   |  14 +--
 raid/x86.c      | 314 ++++++++++++++++++++++++++----------------------
 raid/x86z.c     |  38 +++---
 13 files changed, 515 insertions(+), 481 deletions(-)

diff --git a/portable.h b/portable.h
index b89a05f7..eb5f311c 100644
--- a/portable.h
+++ b/portable.h
@@ -180,6 +180,13 @@
 #define __always_inline inline __attribute__((always_inline))
 #endif
 
+/**
+ * Missing __aligned, like in user mode.
+ */
+#ifndef __aligned
+#define __aligned(a) __attribute__((aligned(a)))
+#endif
+
 /**
  * Another name for link() to avoid confusion with local variables called "link".
  */
diff --git a/raid/cpu.h b/raid/cpu.h
index 62504820..966055b2 100644
--- a/raid/cpu.h
+++ b/raid/cpu.h
@@ -20,7 +20,7 @@
 
 #if defined(__i386__) || defined(__x86_64__)
 
-static __always_inline void raid_cpuid(uint32_t func_eax, uint32_t sub_ecx, uint32_t* reg)
+static __always_inline void raid_cpuid(uint32_t func_eax, uint32_t sub_ecx, uint32_t *reg)
 {
 	asm volatile(
 #if defined(__i386__) && defined(__PIC__)
@@ -40,16 +40,16 @@ static __always_inline void raid_cpuid(uint32_t func_eax, uint32_t sub_ecx, uint
 
 #define CPU_VENDOR_MAX 13
 
-static __always_inline void raid_cpu_info(char* vendor, unsigned* family, unsigned* model)
+static __always_inline void raid_cpu_info(char *vendor, unsigned *family, unsigned *model)
 {
 	uint32_t reg[4];
 	unsigned f, ef, m, em;
 
 	raid_cpuid(0, 0, reg);
 
-	((uint32_t*)vendor)[0] = reg[1];
-	((uint32_t*)vendor)[1] = reg[3];
-	((uint32_t*)vendor)[2] = reg[2];
+	((uint32_t *)vendor)[0] = reg[1];
+	((uint32_t *)vendor)[1] = reg[3];
+	((uint32_t *)vendor)[2] = reg[2];
 	vendor[12] = 0;
 
 	raid_cpuid(1, 0, reg);
@@ -185,7 +185,7 @@ static __always_inline int raid_cpu_has_slowpshufb(void)
 		 * CPU GenuineIntel, family 6, model 28, flags mmx sse2 ssse3 slowmult
 		 * Memory is little-endian 64-bit
 		 * Support nanosecond timestamps with futimens()
-		 * 
+		 *
 		 * Speed test with 8 disk and 262144 buffer size, for a total of 2816 KiB...
 		 * memset0 1850 [MB/s]
 		 * CRC table 316 [MB/s]
diff --git a/raid/gf.h b/raid/gf.h
index 2eaedfe5..7d165ad1 100644
--- a/raid/gf.h
+++ b/raid/gf.h
@@ -62,7 +62,7 @@ static __always_inline uint8_t pow2(int v)
 /*
  * Gets the multiplication table for a specified value.
  */
-static __always_inline const uint8_t* table(uint8_t v)
+static __always_inline const uint8_t *table(uint8_t v)
 {
 	return gfmul[v];
 }
@@ -80,17 +80,17 @@ static __always_inline uint8_t A(int p, int d)
 /*
  * Dereference as uint8_t
  */
-#define v_8(p) (*(uint8_t*)&(p))
+#define v_8(p) (*(uint8_t *)&(p))
 
 /*
  * Dereference as uint32_t
  */
-#define v_32(p) (*(uint32_t*)&(p))
+#define v_32(p) (*(uint32_t *)&(p))
 
 /*
  * Dereference as uint64_t
  */
-#define v_64(p) (*(uint64_t*)&(p))
+#define v_64(p) (*(uint64_t *)&(p))
 
 /*
  * Multiply each byte of a uint32 by 2 in the GF(2^8).
diff --git a/raid/int.c b/raid/int.c
index 21417021..437e1acd 100644
--- a/raid/int.c
+++ b/raid/int.c
@@ -21,10 +21,10 @@
 /*
  * PAR1 (RAID5 with xor) 32bit C implementation
  */
-void raid_par1_int32(int nd, size_t size, void** vv)
+void raid_par1_int32(int nd, size_t size, void **vv)
 {
-	uint8_t** v = (uint8_t**)vv;
-	uint8_t* p;
+	uint8_t **v = (uint8_t **)vv;
+	uint8_t *p;
 	int d, l;
 	size_t i;
 
@@ -34,12 +34,12 @@ void raid_par1_int32(int nd, size_t size, void** vv)
 	l = nd - 1;
 	p = v[nd];
 
-	for(i=0;i<size;i+=8) {
+	for (i = 0; i < size; i += 8) {
 		p0 = v_32(v[l][i]);
 		p1 = v_32(v[l][i+4]);
 		/* accessing disks in backward order because the buffers */
 		/* are also in backward order */
-		for(d=l-1;d>=0;--d) {
+		for (d = l-1; d >= 0; --d) {
 			p0 ^= v_32(v[d][i]);
 			p1 ^= v_32(v[d][i+4]);
 		}
@@ -51,10 +51,10 @@ void raid_par1_int32(int nd, size_t size, void** vv)
 /*
  * PAR1 (RAID5 with xor) 64bit C implementation
  */
-void raid_par1_int64(int nd, size_t size, void** vv)
+void raid_par1_int64(int nd, size_t size, void **vv)
 {
-	uint8_t** v = (uint8_t**)vv;
-	uint8_t* p;
+	uint8_t **v = (uint8_t **)vv;
+	uint8_t *p;
 	int d, l;
 	size_t i;
 
@@ -64,12 +64,12 @@ void raid_par1_int64(int nd, size_t size, void** vv)
 	l = nd - 1;
 	p = v[nd];
 
-	for(i=0;i<size;i+=16) {
+	for (i = 0; i < size; i += 16) {
 		p0 = v_64(v[l][i]);
 		p1 = v_64(v[l][i+8]);
 		/* accessing disks in backward order because the buffers */
 		/* are also in backward order */
-		for(d=l-1;d>=0;--d) {
+		for (d = l-1; d >= 0; --d) {
 			p0 ^= v_64(v[d][i]);
 			p1 ^= v_64(v[d][i+8]);
 		}
@@ -81,11 +81,11 @@ void raid_par1_int64(int nd, size_t size, void** vv)
 /*
  * PAR2 (RAID6 with powers of 2) 32bit C implementation
  */
-void raid_par2_int32(int nd, size_t size, void** vv)
+void raid_par2_int32(int nd, size_t size, void **vv)
 {
-	uint8_t** v = (uint8_t**)vv;
-	uint8_t* p;
-	uint8_t* q;
+	uint8_t **v = (uint8_t **)vv;
+	uint8_t *p;
+	uint8_t *q;
 	int d, l;
 	size_t i;
 
@@ -96,10 +96,10 @@ void raid_par2_int32(int nd, size_t size, void** vv)
 	p = v[nd];
 	q = v[nd+1];
 
-	for(i=0;i<size;i+=8) {
+	for (i = 0; i < size; i += 8) {
 		q0 = p0 = v_32(v[l][i]);
 		q1 = p1 = v_32(v[l][i+4]);
-		for(d=l-1;d>=0;--d) {
+		for (d = l-1; d >= 0; --d) {
 			d0 = v_32(v[d][i]);
 			d1 = v_32(v[d][i+4]);
 
@@ -122,11 +122,11 @@ void raid_par2_int32(int nd, size_t size, void** vv)
 /*
  * PAR2 (RAID6 with powers of 2) 64bit C implementation
  */
-void raid_par2_int64(int nd, size_t size, void** vv)
+void raid_par2_int64(int nd, size_t size, void **vv)
 {
-	uint8_t** v = (uint8_t**)vv;
-	uint8_t* p;
-	uint8_t* q;
+	uint8_t **v = (uint8_t **)vv;
+	uint8_t *p;
+	uint8_t *q;
 	int d, l;
 	size_t i;
 
@@ -137,10 +137,10 @@ void raid_par2_int64(int nd, size_t size, void** vv)
 	p = v[nd];
 	q = v[nd+1];
 
-	for(i=0;i<size;i+=16) {
+	for (i = 0; i < size; i += 16) {
 		q0 = p0 = v_64(v[l][i]);
 		q1 = p1 = v_64(v[l][i+8]);
-		for(d=l-1;d>=0;--d) {
+		for (d = l-1; d >= 0; --d) {
 			d0 = v_64(v[d][i]);
 			d1 = v_64(v[d][i+8]);
 
@@ -162,18 +162,18 @@ void raid_par2_int64(int nd, size_t size, void** vv)
 
 /*
  * PAR3 (triple parity with Cauchy matrix) 8bit C implementation
- * 
+ *
  * Note that instead of a generic multiplicationt table, likely resulting
- * in multiple cache misses, a precomputed table could be used. 
+ * in multiple cache misses, a precomputed table could be used.
  * But this is only a kind of reference function, and we are not really
  * interested in speed.
  */
-void raid_par3_int8(int nd, size_t size, void** vv)
+void raid_par3_int8(int nd, size_t size, void **vv)
 {
-	uint8_t** v = (uint8_t**)vv;
-	uint8_t* p;
-	uint8_t* q;
-	uint8_t* r;
+	uint8_t **v = (uint8_t **)vv;
+	uint8_t *p;
+	uint8_t *q;
+	uint8_t *r;
 	int d, l;
 	size_t i;
 
@@ -184,9 +184,9 @@ void raid_par3_int8(int nd, size_t size, void** vv)
 	q = v[nd+1];
 	r = v[nd+2];
 
-	for(i=0;i<size;i+=1) {
+	for (i = 0; i < size; i += 1) {
 		p0 = q0 = r0 = 0;
-		for(d=l;d>0;--d) {
+		for (d = l; d > 0; --d) {
 			d0 = v_8(v[d][i]);
 
 			p0 ^= d0;
@@ -209,19 +209,19 @@ void raid_par3_int8(int nd, size_t size, void** vv)
 
 /*
  * PAR4 (quad parity with Cauchy matrix) 8bit C implementation
- * 
+ *
  * Note that instead of a generic multiplicationt table, likely resulting
- * in multiple cache misses, a precomputed table could be used. 
+ * in multiple cache misses, a precomputed table could be used.
  * But this is only a kind of reference function, and we are not really
  * interested in speed.
  */
-void raid_par4_int8(int nd, size_t size, void** vv)
+void raid_par4_int8(int nd, size_t size, void **vv)
 {
-	uint8_t** v = (uint8_t**)vv;
-	uint8_t* p;
-	uint8_t* q;
-	uint8_t* r;
-	uint8_t* s;
+	uint8_t **v = (uint8_t **)vv;
+	uint8_t *p;
+	uint8_t *q;
+	uint8_t *r;
+	uint8_t *s;
 	int d, l;
 	size_t i;
 
@@ -233,9 +233,9 @@ void raid_par4_int8(int nd, size_t size, void** vv)
 	r = v[nd+2];
 	s = v[nd+3];
 
-	for(i=0;i<size;i+=1) {
+	for (i = 0; i < size; i += 1) {
 		p0 = q0 = r0 = s0 = 0;
-		for(d=l;d>0;--d) {
+		for (d = l; d > 0; --d) {
 			d0 = v_8(v[d][i]);
 
 			p0 ^= d0;
@@ -261,20 +261,20 @@ void raid_par4_int8(int nd, size_t size, void** vv)
 
 /*
  * PAR5 (penta parity with Cauchy matrix) 8bit C implementation
- * 
+ *
  * Note that instead of a generic multiplicationt table, likely resulting
- * in multiple cache misses, a precomputed table could be used. 
+ * in multiple cache misses, a precomputed table could be used.
  * But this is only a kind of reference function, and we are not really
  * interested in speed.
  */
-void raid_par5_int8(int nd, size_t size, void** vv)
+void raid_par5_int8(int nd, size_t size, void **vv)
 {
-	uint8_t** v = (uint8_t**)vv;
-	uint8_t* p;
-	uint8_t* q;
-	uint8_t* r;
-	uint8_t* s;
-	uint8_t* t;
+	uint8_t **v = (uint8_t **)vv;
+	uint8_t *p;
+	uint8_t *q;
+	uint8_t *r;
+	uint8_t *s;
+	uint8_t *t;
 	int d, l;
 	size_t i;
 
@@ -287,9 +287,9 @@ void raid_par5_int8(int nd, size_t size, void** vv)
 	s = v[nd+3];
 	t = v[nd+4];
 
-	for(i=0;i<size;i+=1) {
+	for (i = 0; i < size; i += 1) {
 		p0 = q0 = r0 = s0 = t0 = 0;
-		for(d=l;d>0;--d) {
+		for (d = l; d > 0; --d) {
 			d0 = v_8(v[d][i]);
 
 			p0 ^= d0;
@@ -318,21 +318,21 @@ void raid_par5_int8(int nd, size_t size, void** vv)
 
 /*
  * PAR6 (hexa parity with Cauchy matrix) 8bit C implementation
- * 
+ *
  * Note that instead of a generic multiplicationt table, likely resulting
- * in multiple cache misses, a precomputed table could be used. 
+ * in multiple cache misses, a precomputed table could be used.
  * But this is only a kind of reference function, and we are not really
  * interested in speed.
  */
-void raid_par6_int8(int nd, size_t size, void** vv)
+void raid_par6_int8(int nd, size_t size, void **vv)
 {
-	uint8_t** v = (uint8_t**)vv;
-	uint8_t* p;
-	uint8_t* q;
-	uint8_t* r;
-	uint8_t* s;
-	uint8_t* t;
-	uint8_t* u;
+	uint8_t **v = (uint8_t **)vv;
+	uint8_t *p;
+	uint8_t *q;
+	uint8_t *r;
+	uint8_t *s;
+	uint8_t *t;
+	uint8_t *u;
 	int d, l;
 	size_t i;
 
@@ -346,9 +346,9 @@ void raid_par6_int8(int nd, size_t size, void** vv)
 	t = v[nd+4];
 	u = v[nd+5];
 
-	for(i=0;i<size;i+=1) {
+	for (i = 0; i < size; i += 1) {
 		p0 = q0 = r0 = s0 = t0 = u0 = 0;
-		for(d=l;d>0;--d) {
+		for (d = l; d > 0; --d) {
 			d0 = v_8(v[d][i]);
 
 			p0 ^= d0;
@@ -390,12 +390,12 @@ void raid_par6_int8(int nd, size_t size, void** vv)
  *
  * Dx = A[ip[0],id[0]]^-1 * Pd
  */
-void raid_rec1_int8(int nr, const int* id, const int* ip, int nd, size_t size, void** vv)
+void raid_rec1_int8(int nr, const int *id, const int *ip, int nd, size_t size, void **vv)
 {
-	uint8_t** v = (uint8_t**)vv;
-	uint8_t* p;
-	uint8_t* pa;
-	const uint8_t* T;
+	uint8_t **v = (uint8_t **)vv;
+	uint8_t *p;
+	uint8_t *pa;
+	const uint8_t *T;
 	uint8_t G;
 	uint8_t V;
 	size_t i;
@@ -409,7 +409,7 @@ void raid_rec1_int8(int nr, const int* id, const int* ip, int nd, size_t size, v
 	}
 
 	/* setup the coefficients matrix */
-	G = A(ip[0],id[0]);
+	G = A(ip[0], id[0]);
 
 	/* invert it to solve the system of linear equations */
 	V = inv(G);
@@ -423,7 +423,7 @@ void raid_rec1_int8(int nr, const int* id, const int* ip, int nd, size_t size, v
 	p = v[nd+ip[0]];
 	pa = v[id[0]];
 
-	for(i=0;i<size;++i) {
+	for (i = 0; i < size; ++i) {
 		/* delta */
 		uint8_t Pd = p[i] ^ pa[i];
 
@@ -443,15 +443,15 @@ void raid_rec1_int8(int nr, const int* id, const int* ip, int nd, size_t size, v
  *
  * we solve inverting the coefficients matrix.
  */
-void raid_rec2_int8(int nr, const int* id, const int* ip, int nd, size_t size, void** vv)
+void raid_rec2_int8(int nr, const int *id, const int *ip, int nd, size_t size, void **vv)
 {
-	uint8_t** v = (uint8_t**)vv;
-	uint8_t* p;
-	uint8_t* pa;
-	uint8_t* q;
-	uint8_t* qa;
+	uint8_t **v = (uint8_t **)vv;
+	uint8_t *p;
+	uint8_t *pa;
+	uint8_t *q;
+	uint8_t *qa;
 	const int N = 2;
-	const uint8_t* T[N][N];
+	const uint8_t *T[N][N];
 	uint8_t G[N*N];
 	uint8_t V[N*N];
 	size_t i;
@@ -459,28 +459,24 @@ void raid_rec2_int8(int nr, const int* id, const int* ip, int nd, size_t size, v
 
 	(void)nr; /* unused, it's always 2 */
 
-	/* if it's RAID6 uses the dedicated and faster function */
+	/* if it's RAID6 recovering with P and Q uses the dedicated and faster function */
 	if (ip[0] == 0 && ip[1] == 1) {
 		raid_rec2_par2(id, ip, nd, size, vv);
 		return;
 	}
 
 	/* setup the coefficients matrix */
-	for(j=0;j<N;++j) {
-		for(k=0;k<N;++k) {
-			G[j*N+k] = A(ip[j],id[k]);
-		}
-	}
+	for (j = 0; j < N; ++j)
+		for (k = 0; k < N; ++k)
+			G[j*N+k] = A(ip[j], id[k]);
 
 	/* invert it to solve the system of linear equations */
 	raid_invert(G, V, N);
 
 	/* get multiplication tables */
-	for(j=0;j<N;++j) {
-		for(k=0;k<N;++k) {
-			T[j][k] = table( V[j*N+k] );
-		}
-	}
+	for (j = 0; j < N; ++j)
+		for (k = 0; k < N; ++k)
+			T[j][k] = table(V[j*N+k]);
 
 	/* compute delta parity */
 	raid_delta_gen(2, id, ip, nd, size, vv);
@@ -490,7 +486,7 @@ void raid_rec2_int8(int nr, const int* id, const int* ip, int nd, size_t size, v
 	pa = v[id[0]];
 	qa = v[id[1]];
 
-	for(i=0;i<size;++i) {
+	for (i = 0; i < size; ++i) {
 		/* delta */
 		uint8_t Pd = p[i] ^ pa[i];
 		uint8_t Qd = q[i] ^ qa[i];
@@ -515,53 +511,49 @@ void raid_rec2_int8(int nr, const int* id, const int* ip, int nd, size_t size, v
  * PD[0] = Pd, PD[1] = Qd, PD[2] = Rd, ...
  * D[0] = Dx, D[1] = Dy, D[2] = Dz, ...
  */
-void raid_recX_int8(int nr, const int* id, const int* ip, int nd, size_t size, void** vv)
+void raid_recX_int8(int nr, const int *id, const int *ip, int nd, size_t size, void **vv)
 {
-	uint8_t** v = (uint8_t**)vv;
-	uint8_t* p[RAID_PARITY_MAX];
-	uint8_t* pa[RAID_PARITY_MAX];
-	const uint8_t* T[RAID_PARITY_MAX][RAID_PARITY_MAX];
+	uint8_t **v = (uint8_t **)vv;
+	uint8_t *p[RAID_PARITY_MAX];
+	uint8_t *pa[RAID_PARITY_MAX];
+	const uint8_t *T[RAID_PARITY_MAX][RAID_PARITY_MAX];
 	uint8_t G[RAID_PARITY_MAX*RAID_PARITY_MAX];
 	uint8_t V[RAID_PARITY_MAX*RAID_PARITY_MAX];
 	size_t i;
 	int j, k;
 
 	/* setup the coefficients matrix */
-	for(j=0;j<nr;++j) {
-		for(k=0;k<nr;++k) {
-			G[j*nr+k] = A(ip[j],id[k]);
-		}
-	}
+	for (j = 0; j < nr; ++j)
+		for (k = 0; k < nr; ++k)
+			G[j*nr+k] = A(ip[j], id[k]);
 
 	/* invert it to solve the system of linear equations */
 	raid_invert(G, V, nr);
 
 	/* get multiplication tables */
-	for(j=0;j<nr;++j) {
-		for(k=0;k<nr;++k) {
-			T[j][k] = table( V[j*nr+k] );
-		}
-	}
+	for (j = 0; j < nr; ++j)
+		for (k = 0; k < nr; ++k)
+			T[j][k] = table(V[j*nr+k]);
 
 	/* compute delta parity */
 	raid_delta_gen(nr, id, ip, nd, size, vv);
 
-	for(j=0;j<nr;++j) {
+	for (j = 0; j < nr; ++j) {
 		p[j] = v[nd+ip[j]];
 		pa[j] = v[id[j]];
 	}
 
-	for(i=0;i<size;++i) {
+	for (i = 0; i < size; ++i) {
 		uint8_t PD[RAID_PARITY_MAX];
 
 		/* delta */
-		for(j=0;j<nr;++j)
+		for (j = 0; j < nr; ++j)
 			PD[j] = p[j][i] ^ pa[j][i];
 
 		/* reconstruct */
-		for(j=0;j<nr;++j) {
+		for (j = 0; j < nr; ++j) {
 			uint8_t b = 0;
-			for(k=0;k<nr;++k)
+			for (k = 0; k < nr; ++k)
 				b ^= T[j][k][PD[k]];
 			pa[j][i] = b;
 		}
diff --git a/raid/internal.h b/raid/internal.h
index e1ac2663..b86e5064 100644
--- a/raid/internal.h
+++ b/raid/internal.h
@@ -27,74 +27,74 @@
  *
  * These are intented to provide access for testing.
  */
-void raid_invert(uint8_t* M, uint8_t* V, int n);
-void raid_delta_gen(int nr, const int* id, const int* ip, int nd, size_t size, void** v);
-void raid_rec1_par1(const int* id, int nd, size_t size, void** v);
-void raid_rec2_par2(const int* id, const int* ip, int nd, size_t size, void** vv);
-void raid_par1_int32(int nd, size_t size, void** vv);
-void raid_par1_int64(int nd, size_t size, void** vv);
-void raid_par1_sse2(int nd, size_t size, void** vv);
-void raid_par2_int32(int nd, size_t size, void** vv);
-void raid_par2_int64(int nd, size_t size, void** vv);
-void raid_par2_sse2(int nd, size_t size, void** vv);
-void raid_par2_sse2ext(int nd, size_t size, void** vv);
-void raid_parz_int32(int nd, size_t size, void** vv);
-void raid_parz_int64(int nd, size_t size, void** vv);
-void raid_parz_sse2(int nd, size_t size, void** vv);
-void raid_parz_sse2ext(int nd, size_t size, void** vv);
-void raid_par3_int8(int nd, size_t size, void** vv);
-void raid_par3_ssse3(int nd, size_t size, void** vv);
-void raid_par3_ssse3ext(int nd, size_t size, void** vv);
-void raid_par4_int8(int nd, size_t size, void** vv);
-void raid_par4_ssse3(int nd, size_t size, void** vv);
-void raid_par4_ssse3ext(int nd, size_t size, void** vv);
-void raid_par5_int8(int nd, size_t size, void** vv);
-void raid_par5_ssse3(int nd, size_t size, void** vv);
-void raid_par5_ssse3ext(int nd, size_t size, void** vv);
-void raid_par6_int8(int nd, size_t size, void** vv);
-void raid_par6_ssse3(int nd, size_t size, void** vv);
-void raid_par6_ssse3ext(int nd, size_t size, void** vv);
-void raid_rec1_int8(int nr, const int* id, const int* ip, int nd, size_t size, void** vv);
-void raid_rec2_int8(int nr, const int* id, const int* ip, int nd, size_t size, void** vv);
-void raid_recX_int8(int nr, const int* id, const int* ip, int nd, size_t size, void** vv);
-void raid_rec1_ssse3(int nr, const int* id, const int* ip, int nd, size_t size, void** vv);
-void raid_rec2_ssse3(int nr, const int* id, const int* ip, int nd, size_t size, void** vv);
-void raid_recX_ssse3(int nr, const int* id, const int* ip, int nd, size_t size, void** vv);
+void raid_invert(uint8_t *M, uint8_t *V, int n);
+void raid_delta_gen(int nr, const int *id, const int *ip, int nd, size_t size, void **v);
+void raid_rec1_par1(const int *id, int nd, size_t size, void **v);
+void raid_rec2_par2(const int *id, const int *ip, int nd, size_t size, void **vv);
+void raid_par1_int32(int nd, size_t size, void **vv);
+void raid_par1_int64(int nd, size_t size, void **vv);
+void raid_par1_sse2(int nd, size_t size, void **vv);
+void raid_par2_int32(int nd, size_t size, void **vv);
+void raid_par2_int64(int nd, size_t size, void **vv);
+void raid_par2_sse2(int nd, size_t size, void **vv);
+void raid_par2_sse2ext(int nd, size_t size, void **vv);
+void raid_parz_int32(int nd, size_t size, void **vv);
+void raid_parz_int64(int nd, size_t size, void **vv);
+void raid_parz_sse2(int nd, size_t size, void **vv);
+void raid_parz_sse2ext(int nd, size_t size, void **vv);
+void raid_par3_int8(int nd, size_t size, void **vv);
+void raid_par3_ssse3(int nd, size_t size, void **vv);
+void raid_par3_ssse3ext(int nd, size_t size, void **vv);
+void raid_par4_int8(int nd, size_t size, void **vv);
+void raid_par4_ssse3(int nd, size_t size, void **vv);
+void raid_par4_ssse3ext(int nd, size_t size, void **vv);
+void raid_par5_int8(int nd, size_t size, void **vv);
+void raid_par5_ssse3(int nd, size_t size, void **vv);
+void raid_par5_ssse3ext(int nd, size_t size, void **vv);
+void raid_par6_int8(int nd, size_t size, void **vv);
+void raid_par6_ssse3(int nd, size_t size, void **vv);
+void raid_par6_ssse3ext(int nd, size_t size, void **vv);
+void raid_rec1_int8(int nr, const int *id, const int *ip, int nd, size_t size, void **vv);
+void raid_rec2_int8(int nr, const int *id, const int *ip, int nd, size_t size, void **vv);
+void raid_recX_int8(int nr, const int *id, const int *ip, int nd, size_t size, void **vv);
+void raid_rec1_ssse3(int nr, const int *id, const int *ip, int nd, size_t size, void **vv);
+void raid_rec2_ssse3(int nr, const int *id, const int *ip, int nd, size_t size, void **vv);
+void raid_recX_ssse3(int nr, const int *id, const int *ip, int nd, size_t size, void **vv);
 
 /*
  * Internal naming.
  *
  * These are intented to provide access for testing.
  */
-const char* raid_par1_tag(void);
-const char* raid_par2_tag(void);
-const char* raid_parz_tag(void);
-const char* raid_par3_tag(void);
-const char* raid_par4_tag(void);
-const char* raid_par5_tag(void);
-const char* raid_par6_tag(void);
-const char* raid_rec1_tag(void);
-const char* raid_rec2_tag(void);
-const char* raid_recX_tag(void);
+const char *raid_par1_tag(void);
+const char *raid_par2_tag(void);
+const char *raid_parz_tag(void);
+const char *raid_par3_tag(void);
+const char *raid_par4_tag(void);
+const char *raid_par5_tag(void);
+const char *raid_par6_tag(void);
+const char *raid_rec1_tag(void);
+const char *raid_rec2_tag(void);
+const char *raid_recX_tag(void);
 
 /*
  * Internal forwarders.
  */
-extern void (*raid_par_ptr[RAID_PARITY_MAX])(int nd, size_t size, void** vv);
-extern void (*raid_par3_ptr)(int nd, size_t size, void** vv);
-extern void (*raid_parz_ptr)(int nd, size_t size, void** vv); 
-extern void (*raid_rec_ptr[RAID_PARITY_MAX])(int nr, const int* id, const int* ip, int nd, size_t size, void** vv);
+extern void (*raid_par_ptr[RAID_PARITY_MAX])(int nd, size_t size, void **vv);
+extern void (*raid_par3_ptr)(int nd, size_t size, void **vv);
+extern void (*raid_parz_ptr)(int nd, size_t size, void **vv);
+extern void (*raid_rec_ptr[RAID_PARITY_MAX])(int nr, const int *id, const int *ip, int nd, size_t size, void **vv);
 
 /*
  * Tables.
  */
-extern const uint8_t raid_gfmul[256][256] __attribute__((aligned(256)));
-extern const uint8_t raid_gfexp[256] __attribute__((aligned(256)));
-extern const uint8_t raid_gfinv[256] __attribute__((aligned(256)));
-extern const uint8_t raid_gfvandermonde[3][256] __attribute__((aligned(256)));
-extern const uint8_t raid_gfcauchy[6][256] __attribute__((aligned(256)));
-extern const uint8_t raid_gfcauchypshufb[251][4][2][16] __attribute__((aligned(256)));
-extern const uint8_t raid_gfmulpshufb[256][2][16] __attribute__((aligned(256)));
+extern const uint8_t raid_gfmul[256][256] __aligned(256);
+extern const uint8_t raid_gfexp[256] __aligned(256);
+extern const uint8_t raid_gfinv[256] __aligned(256);
+extern const uint8_t raid_gfvandermonde[3][256] __aligned(256);
+extern const uint8_t raid_gfcauchy[6][256] __aligned(256);
+extern const uint8_t raid_gfcauchypshufb[251][4][2][16] __aligned(256);
+extern const uint8_t raid_gfmulpshufb[256][2][16] __aligned(256);
 #define gfmul raid_gfmul
 #define gfexp raid_gfexp
 #define gfinv raid_gfinv
@@ -103,5 +103,14 @@ extern const uint8_t raid_gfmulpshufb[256][2][16] __attribute__((aligned(256)));
 #define gfcauchypshufb raid_gfcauchypshufb
 #define gfmulpshufb raid_gfmulpshufb
 
+/*
+ * Assembler blocks.
+ */
+#define asm_begin() \
+	do { } while (0)
+
+#define asm_end() \
+	asm volatile("sfence" : : : "memory")
+
 #endif
 
diff --git a/raid/intz.c b/raid/intz.c
index f412f001..640037f9 100644
--- a/raid/intz.c
+++ b/raid/intz.c
@@ -21,12 +21,12 @@
 /*
  * PARz (triple parity with powers of 2^-1) 32bit C implementation
  */
-void raid_parz_int32(int nd, size_t size, void** vv)
+void raid_parz_int32(int nd, size_t size, void **vv)
 {
-	uint8_t** v = (uint8_t**)vv;
-	uint8_t* p;
-	uint8_t* q;
-	uint8_t* r;
+	uint8_t **v = (uint8_t **)vv;
+	uint8_t *p;
+	uint8_t *q;
+	uint8_t *r;
 	int d, l;
 	size_t i;
 
@@ -38,10 +38,10 @@ void raid_parz_int32(int nd, size_t size, void** vv)
 	q = v[nd+1];
 	r = v[nd+2];
 
-	for(i=0;i<size;i+=8) {
+	for (i = 0; i < size; i += 8) {
 		r0 = q0 = p0 = v_32(v[l][i]);
 		r1 = q1 = p1 = v_32(v[l][i+4]);
-		for(d=l-1;d>=0;--d) {
+		for (d = l-1; d >= 0; --d) {
 			d0 = v_32(v[d][i]);
 			d1 = v_32(v[d][i+4]);
 
@@ -72,12 +72,12 @@ void raid_parz_int32(int nd, size_t size, void** vv)
 /*
  * PARz (triple parity with powers of 2^-1) 64bit C implementation
  */
-void raid_parz_int64(int nd, size_t size, void** vv)
+void raid_parz_int64(int nd, size_t size, void **vv)
 {
-	uint8_t** v = (uint8_t**)vv;
-	uint8_t* p;
-	uint8_t* q;
-	uint8_t* r;
+	uint8_t **v = (uint8_t **)vv;
+	uint8_t *p;
+	uint8_t *q;
+	uint8_t *r;
 	int d, l;
 	size_t i;
 
@@ -89,10 +89,10 @@ void raid_parz_int64(int nd, size_t size, void** vv)
 	q = v[nd+1];
 	r = v[nd+2];
 
-	for(i=0;i<size;i+=16) {
+	for (i = 0; i < size; i += 16) {
 		r0 = q0 = p0 = v_64(v[l][i]);
 		r1 = q1 = p1 = v_64(v[l][i+8]);
-		for(d=l-1;d>=0;--d) {
+		for (d = l-1; d >= 0; --d) {
 			d0 = v_64(v[d][i]);
 			d1 = v_64(v[d][i+8]);
 
diff --git a/raid/mktables.c b/raid/mktables.c
index 932ba1fb..ae2beb47 100644
--- a/raid/mktables.c
+++ b/raid/mktables.c
@@ -63,7 +63,7 @@ uint8_t gfinv[256];
 /**
  * Setup the Cauchy matrix used to generate the parity.
  */
-static void set_cauchy(uint8_t* matrix)
+static void set_cauchy(uint8_t *matrix)
 {
 	int i, j;
 	uint8_t inv_x, y;
@@ -74,9 +74,8 @@ static void set_cauchy(uint8_t* matrix)
 	 * This is an Extended Cauchy matrix built from a Cauchy matrix
 	 * adding the first row of all 1.
 	 */
-	for(i=0;i<DISK;++i) {
+	for (i = 0; i < DISK; ++i)
 		matrix[0*DISK+i] = 1;
-	}
 
 	/*
 	 * Second row is formed by power of 2^i.
@@ -93,7 +92,7 @@ static void set_cauchy(uint8_t* matrix)
 	 * with 2^-i != 0 for any i
 	 */
 	inv_x = 1;
-	for(i=0;i<DISK;++i) {
+	for (i = 0; i < DISK; ++i) {
 		matrix[1*DISK+i] = inv_x;
 		inv_x = gfmul(2, inv_x);
 	}
@@ -108,9 +107,9 @@ static void set_cauchy(uint8_t* matrix)
 	 * with xi != yj for any i,j with i>=0,j>=1,i+j<255
 	 */
 	y = 2;
-	for(j=0;j<PARITY-2;++j) {
+	for (j = 0; j < PARITY-2; ++j) {
 		inv_x = 1;
-		for(i=0;i<DISK;++i) {
+		for (i = 0; i < DISK; ++i) {
 			uint8_t x = gfinv[inv_x];
 			matrix[(j+2)*DISK+i] = gfinv[y ^ x];
 			inv_x = gfmul(2, inv_x);
@@ -126,36 +125,34 @@ static void set_cauchy(uint8_t* matrix)
 	 * This operation doesn't invalidate the property that all the square
 	 * submatrices are not singular.
 	 */
-	for(j=0;j<PARITY-2;++j) {
+	for (j = 0; j < PARITY-2; ++j) {
 		uint8_t f = gfinv[matrix[(j+2)*DISK]];
 
-		for(i=0;i<DISK;++i) {
+		for (i = 0; i < DISK; ++i)
 			matrix[(j+2)*DISK+i] = gfmul(matrix[(j+2)*DISK+i], f);
-		}
 	}
 }
 
 /**
  * Setup the Power matrix used to generate the parity.
  */
-static void set_power(uint8_t* matrix)
+static void set_power(uint8_t *matrix)
 {
 	unsigned i;
 	uint8_t v;
 
 	v = 1;
-	for(i=0;i<DISK;++i) {
+	for (i = 0; i < DISK; ++i)
 		matrix[0*DISK+i] = v;
-	}
 
 	v = 1;
-	for(i=0;i<DISK;++i) {
+	for (i = 0; i < DISK; ++i) {
 		matrix[1*DISK+i] = v;
 		v = gfmul(2, v);
 	}
 
 	v = 1;
-	for(i=0;i<DISK;++i) {
+	for (i = 0; i < DISK; ++i) {
 		matrix[2*DISK+i] = v;
 		v = gfmul(0x8e, v);
 	}
@@ -205,11 +202,11 @@ int main(void)
 	printf("\n");
 
 	/* a*b */
-	printf("const uint8_t __attribute__((aligned(256))) raid_gfmul[256][256] =\n");
+	printf("const uint8_t __aligned(256) raid_gfmul[256][256] =\n");
 	printf("{\n");
-	for(i=0;i<256;++i) {
+	for (i = 0; i < 256; ++i) {
 		printf("\t{\n");
-		for(j=0;j<256;++j) {
+		for (j = 0; j < 256; ++j) {
 			if (j % 8 == 0)
 				printf("\t\t");
 			v = gfmul(i, j);
@@ -226,10 +223,10 @@ int main(void)
 	printf("};\n\n");
 
 	/* 2^a */
-	printf("const uint8_t __attribute__((aligned(256))) raid_gfexp[256] =\n");
+	printf("const uint8_t __aligned(256) raid_gfexp[256] =\n");
 	printf("{\n");
 	v = 1;
-	for(i=0;i<256;++i) {
+	for (i = 0; i < 256; ++i) {
 		if (i % 8 == 0)
 			printf("\t");
 		printf("0x%02x,", v);
@@ -242,10 +239,10 @@ int main(void)
 	printf("};\n\n");
 
 	/* 1/a */
-	printf("const uint8_t __attribute__((aligned(256))) raid_gfinv[256] =\n");
+	printf("const uint8_t __aligned(256) raid_gfinv[256] =\n");
 	printf("{\n");
 	printf("\t/* note that the first element is not significative */\n");
-	for(i=0;i<256;++i) {
+	for (i = 0; i < 256; ++i) {
 		if (i % 8 == 0)
 			printf("\t");
 		if (i == 0)
@@ -267,19 +264,18 @@ int main(void)
 	printf(" * Power matrix used to generate parity.\n");
 	printf(" * This matrix is valid for up to %u parity with %u data disks.\n", 3, DISK);
 	printf(" *\n");
-	for(p=0;p<3;++p) {
+	for (p = 0; p < 3; ++p) {
 		printf(" * ");
-		for(i=0;i<DISK;++i) {
+		for (i = 0; i < DISK; ++i)
 			printf("%02x ", matrix[p*DISK+i]);
-		}
 		printf("\n");
 	}
 	printf(" */\n");
-	printf("const uint8_t __attribute__((aligned(256))) raid_gfvandermonde[%u][256] =\n", 3);
+	printf("const uint8_t __aligned(256) raid_gfvandermonde[%u][256] =\n", 3);
 	printf("{\n");
-	for(p=0;p<3;++p) {
+	for (p = 0; p < 3; ++p) {
 		printf("\t{\n");
-		for(i=0;i<DISK;++i) {
+		for (i = 0; i < DISK; ++i) {
 			if (i % 8 == 0)
 				printf("\t\t");
 			printf("0x%02x,", matrix[p*DISK+i]);
@@ -299,19 +295,18 @@ int main(void)
 	printf(" * Cauchy matrix used to generate parity.\n");
 	printf(" * This matrix is valid for up to %u parity with %u data disks.\n", PARITY, DISK);
 	printf(" *\n");
-	for(p=0;p<PARITY;++p) {
+	for (p = 0; p < PARITY; ++p) {
 		printf(" * ");
-		for(i=0;i<DISK;++i) {
+		for (i = 0; i < DISK; ++i)
 			printf("%02x ", matrix[p*DISK+i]);
-		}
 		printf("\n");
 	}
 	printf(" */\n");
-	printf("const uint8_t __attribute__((aligned(256))) raid_gfcauchy[%u][256] =\n", PARITY);
+	printf("const uint8_t __aligned(256) raid_gfcauchy[%u][256] =\n", PARITY);
 	printf("{\n");
-	for(p=0;p<PARITY;++p) {
+	for (p = 0; p < PARITY; ++p) {
 		printf("\t{\n");
-		for(i=0;i<DISK;++i) {
+		for (i = 0; i < DISK; ++i) {
 			if (i % 8 == 0)
 				printf("\t\t");
 			printf("0x%02x,", matrix[p*DISK+i]);
@@ -331,15 +326,15 @@ int main(void)
 	printf(" * Indexes are [DISK][PARITY - 2][LH].\n");
 	printf(" * Where DISK is from 0 to %u, PARITY from 2 to %u, LH from 0 to 1.\n", DISK - 1, PARITY - 1);
 	printf(" */\n");
-	printf("const uint8_t __attribute__((aligned(256))) raid_gfcauchypshufb[%u][%u][2][16] =\n", DISK, np(PARITY - 2));
+	printf("const uint8_t __aligned(256) raid_gfcauchypshufb[%u][%u][2][16] =\n", DISK, np(PARITY - 2));
 	printf("{\n");
-	for(i=0;i<DISK;++i) {
+	for (i = 0; i < DISK; ++i) {
 		printf("\t{\n");
-		for(p=2;p<PARITY;++p) {
+		for (p = 2; p < PARITY; ++p) {
 			printf("\t\t{\n");
-			for(j=0;j<2;++j) {
+			for (j = 0; j < 2; ++j) {
 				printf("\t\t\t{ ");
-				for(k=0;k<16;++k) {
+				for (k = 0; k < 16; ++k) {
 					v = gfmul(matrix[p*DISK+i], k);
 					if (j == 1)
 						v = gfmul(v, 16);
@@ -363,13 +358,13 @@ int main(void)
 	printf(" * Indexes are [MULTIPLER][LH].\n");
 	printf(" * Where MULTIPLER is from 0 to 255, LH from 0 to 1.\n");
 	printf(" */\n");
-	printf("const uint8_t __attribute__((aligned(256))) raid_gfmulpshufb[256][2][16] =\n");
+	printf("const uint8_t __aligned(256) raid_gfmulpshufb[256][2][16] =\n");
 	printf("{\n");
-	for(i=0;i<256;++i) {
+	for (i = 0; i < 256; ++i) {
 		printf("\t{\n");
-		for(j=0;j<2;++j) {
+		for (j = 0; j < 2; ++j) {
 			printf("\t\t{ ");
-			for(k=0;k<16;++k) {
+			for (k = 0; k < 16; ++k) {
 				v = gfmul(i, k);
 				if (j == 1)
 					v = gfmul(v, 16);
diff --git a/raid/module.c b/raid/module.c
index db83bd16..6778eed7 100644
--- a/raid/module.c
+++ b/raid/module.c
@@ -25,7 +25,7 @@ void raid_init(void)
 	raid_par_ptr[4] = raid_par5_int8;
 	raid_par_ptr[5] = raid_par6_int8;
 
-	if (sizeof(void*) == 4) {
+	if (sizeof(void *) == 4) {
 		raid_par_ptr[0] = raid_par1_int32;
 		raid_par_ptr[1] = raid_par2_int32;
 		raid_parz_ptr = raid_parz_int32;
@@ -85,8 +85,8 @@ void raid_init(void)
 }
 
 static struct raid_func {
-	const char* name;
-	void* p;
+	const char *name;
+	void *p;
 } RAID_FUNC[] = {
 	{ "int8", raid_par3_int8 },
 	{ "int8", raid_par4_int8 },
@@ -126,9 +126,9 @@ static struct raid_func {
 	{ 0, 0 }
 };
 
-static const char* raid_tag(void* func)
+static const char *raid_tag(void *func)
 {
-	struct raid_func* i = RAID_FUNC;
+	struct raid_func *i = RAID_FUNC;
 	while (i->name != 0) {
 		if (i->p == func)
 			return i->name;
@@ -137,52 +137,52 @@ static const char* raid_tag(void* func)
 	return "unknown";
 }
 
-const char* raid_par1_tag(void)
+const char *raid_par1_tag(void)
 {
 	return raid_tag(raid_par_ptr[0]);
 }
 
-const char* raid_par2_tag(void)
+const char *raid_par2_tag(void)
 {
 	return raid_tag(raid_par_ptr[1]);
 }
 
-const char* raid_parz_tag(void)
+const char *raid_parz_tag(void)
 {
 	return raid_tag(raid_parz_ptr);
 }
 
-const char* raid_par3_tag(void)
+const char *raid_par3_tag(void)
 {
 	return raid_tag(raid_par_ptr[2]);
 }
 
-const char* raid_par4_tag(void)
+const char *raid_par4_tag(void)
 {
 	return raid_tag(raid_par_ptr[3]);
 }
 
-const char* raid_par5_tag(void)
+const char *raid_par5_tag(void)
 {
 	return raid_tag(raid_par_ptr[4]);
 }
 
-const char* raid_par6_tag(void)
+const char *raid_par6_tag(void)
 {
 	return raid_tag(raid_par_ptr[5]);
 }
 
-const char* raid_rec1_tag(void)
+const char *raid_rec1_tag(void)
 {
 	return raid_tag(raid_rec_ptr[0]);
 }
 
-const char* raid_rec2_tag(void)
+const char *raid_rec2_tag(void)
 {
 	return raid_tag(raid_rec_ptr[1]);
 }
 
-const char* raid_recX_tag(void)
+const char *raid_recX_tag(void)
 {
 	return raid_tag(raid_rec_ptr[2]);
 }
diff --git a/raid/raid.c b/raid/raid.c
index bdad2a26..714cbf8d 100644
--- a/raid/raid.c
+++ b/raid/raid.c
@@ -32,7 +32,7 @@
 
 /*
  * The RAID5 and RAID6 support is implemented using the Galois Field
- * GF(2^8) with the primitive polynomial x^8 + x^4 + x^3 + x^2 + 1 
+ * GF(2^8) with the primitive polynomial x^8 + x^4 + x^3 + x^2 + 1
  * (285 decimal).
  *
  * The parity P and Q of a set of N disk Di with 0<=i<N, is computed
@@ -42,8 +42,8 @@
  * Q = sum(2^i * Di) with 0<=i<N
  *
  * This approach is the same used by the Linux Kernel RAID, and by ZFS
- * RAIDZ2, better described in the H. Peter Anvin paper "The mathematics 
- * of RAID-6" [1]. 
+ * RAIDZ2, better described in the H. Peter Anvin paper "The mathematics
+ * of RAID-6" [1].
  *
  * To support triple parity, it was first evaluated and then dropped, an
  * extension of the same approach, with additional parity coefficients set
@@ -57,7 +57,7 @@
  * because we can implement very fast parallel multiplications and
  * divisions by 2 in GF(2^8).
  *
- * It's also similar at the approach used by ZFS RAIDZ3, with the 
+ * It's also similar at the approach used by ZFS RAIDZ3, with the
  * difference that ZFS uses powers of 4 instead of 2^-1.
  *
  * Unfortunately it doesn't work beyond triple parity, because whatever
@@ -70,13 +70,13 @@
  * [2, Chap 11, Problem 7] and this is a requirement to have
  * a MDS (Maximum Distance Separable) code [2, Chap 11, Theorem 8].
  *
- * To overcome this limitation, we use a Cauchy matrix [3][4] to compute 
+ * To overcome this limitation, we use a Cauchy matrix [3][4] to compute
  * the parity. A Cauchy matrix has the property to have all the square
  * submatrices not singular, resulting in always solvable equations,
  * for any combination of missing disks.
  *
  * The problem of this approach is that it requires the use of
- * generic multiplications, and not only by 2 or 2^-1, potentially 
+ * generic multiplications, and not only by 2 or 2^-1, potentially
  * affecting badly the performance.
  *
  * Hopefully there is a method to implement parallel multiplications
@@ -87,13 +87,13 @@
  * the first two rows with coeffients equal at the RAID5 and RAID6 approach
  * decribed, resulting in a compatible extension, and requiring SSSE3
  * instructions only if triple parity or beyond is used.
- * 
+ *
  * The matrix is also adjusted, multipling each row by a constant factor
  * to make the first column of all 1, to optimize the computation for
  * the first disk.
  *
  * This results in the matrix A[row,col] defined as:
- * 
+ *
  * 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01...
  * 01 02 04 08 10 20 40 80 1d 3a 74 e8 cd 87 13 26 4c 98 2d 5a b4 75...
  * 01 f5 d2 c4 9a 71 f1 7f fc 87 c1 c6 19 2f 40 55 3d ba 53 04 9c 61...
@@ -105,11 +105,11 @@
  * This matrix supports 6 level of parity, one for each row, for up to 251
  * data disks, one for each column, with all the 377,342,351,231 square
  * submatrices not singular, verified also with brute-force.
- * 
- * This matrix can be extended to support any number of parities, just 
+ *
+ * This matrix can be extended to support any number of parities, just
  * adding additional rows, and removing one column for each new row.
  * (see mktables.c for more details in how the matrix is generated)
- * 
+ *
  * In details, parity is computed as:
  *
  * P = sum(Di)
@@ -120,7 +120,7 @@
  * U = sum(A[5,i] * Di) with 0<=i<N
  *
  * To recover from a failure of six disks at indexes x,y,z,h,v,w,
- * with 0<=x<y<z<h<v<w<N, we compute the parity of the available N-6 
+ * with 0<=x<y<z<h<v<w<N, we compute the parity of the available N-6
  * disks as:
  *
  * Pa = sum(Di)
@@ -153,7 +153,7 @@
  *
  * Resulting speed in x64, with 8 data disks, using a stripe of 256 KiB,
  * for a Core i7-3740QM CPU @ 2.7GHz is:
- * 
+ *
  *             int8   int32   int64    sse2   sse2e   ssse3  ssse3e
  *   par1             11927   22075   36004
  *   par2              3378    5874   18235   19164
@@ -176,10 +176,10 @@
  *
  *
  * In conclusion, the use of power coefficients, and specifically powers
- * of 1,2,2^-1, is the best option to implement triple parity in CPUs 
+ * of 1,2,2^-1, is the best option to implement triple parity in CPUs
  * without SSSE3.
  * But if a modern CPU with SSSE3 (or similar) is available, the Cauchy
- * matrix is the best option because it provides a fast and general 
+ * matrix is the best option because it provides a fast and general
  * approach working for any number of parities.
  *
  * References:
@@ -212,21 +212,22 @@ void raid_mode(int mode)
 /**
  * Buffer filled with 0 used in recovering.
  */
-static void* raid_empty_zero_block;
+static void *raid_empty_zero_block;
 
-void raid_zero(void* zero)
+void raid_zero(void *zero)
 {
 	raid_empty_zero_block = zero;
 }
 
 /* internal forwarder */
-void (*raid_par_ptr[RAID_PARITY_MAX])(int nd, size_t size, void** vv);
-void (*raid_par3_ptr)(int nd, size_t size, void** vv);
-void (*raid_parz_ptr)(int nd, size_t size, void** vv);
+void (*raid_par_ptr[RAID_PARITY_MAX])(int nd, size_t size, void **vv);
+void (*raid_par3_ptr)(int nd, size_t size, void **vv);
+void (*raid_parz_ptr)(int nd, size_t size, void **vv);
 
-void raid_par(int np, int nd, size_t size, void** v)
+void raid_par(int np, int nd, size_t size, void **v)
 {
 	BUG_ON(np < 1 || np > RAID_PARITY_MAX);
+	BUG_ON(size % 64 != 0);
 
 	raid_par_ptr[np - 1](nd, size, v);
 }
@@ -235,19 +236,17 @@ void raid_par(int np, int nd, size_t size, void** v)
  * Inverts the square matrix M of size nxn into V.
  * We use Gauss elimination to invert.
  */
-void raid_invert(uint8_t* M, uint8_t* V, int n)
+void raid_invert(uint8_t *M, uint8_t *V, int n)
 {
-	int i,j,k;
+	int i, j, k;
 
 	/* set the identity matrix in V */
-	for(i=0;i<n;++i) {
-		for(j=0;j<n;++j) {
+	for (i = 0; i < n; ++i)
+		for (j = 0; j < n; ++j)
 			V[i*n+j] = i == j;
-		}
-	}
 
 	/* for each element in the diagonal */
-	for(k=0;k<n;++k) {
+	for (k = 0; k < n; ++k) {
 		uint8_t f;
 
 		/* the diagonal element cannot be 0 because */
@@ -257,17 +256,17 @@ void raid_invert(uint8_t* M, uint8_t* V, int n)
 
 		/* make the diagonal element to be 1 */
 		f = inv(M[k*n+k]);
-		for(j=0;j<n;++j) {
+		for (j = 0; j < n; ++j) {
 			M[k*n+j] = mul(f, M[k*n+j]);
 			V[k*n+j] = mul(f, V[k*n+j]);
 		}
 
 		/* make all the elements over and under the diagonal to be 0 */
-		for(i=0;i<n;++i) {
+		for (i = 0; i < n; ++i) {
 			if (i == k)
 				continue;
 			f = M[i*n+k];
-			for(j=0;j<n;++j) {
+			for (j = 0; j < n; ++j) {
 				M[i*n+j] ^= mul(f, M[k*n+j]);
 				V[i*n+j] ^= mul(f, V[k*n+j]);
 			}
@@ -285,13 +284,13 @@ void raid_invert(uint8_t* M, uint8_t* V, int n)
  * Note that all the other parities not in the ip[] vector
  * are destroyed.
  */
-void raid_delta_gen(int nr, const int* id, const int* ip, int nd, size_t size, void** v)
+void raid_delta_gen(int nr, const int *id, const int *ip, int nd, size_t size, void **v)
 {
-	void* p[RAID_PARITY_MAX];
-	void* pa[RAID_PARITY_MAX];
+	void *p[RAID_PARITY_MAX];
+	void *pa[RAID_PARITY_MAX];
 	int i;
 
-	for(i=0;i<nr;++i) {
+	for (i = 0; i < nr; ++i) {
 		/* keep a copy of the parity buffer */
 		p[i] = v[nd+ip[i]];
 
@@ -308,7 +307,7 @@ void raid_delta_gen(int nr, const int* id, const int* ip, int nd, size_t size, v
 	/* recompute the minimal parity required */
 	raid_par(ip[nr - 1] + 1, nd, size, v);
 
-	for(i=0;i<nr;++i) {
+	for (i = 0; i < nr; ++i) {
 		/* restore disk buffers as before */
 		v[id[i]] = pa[i];
 
@@ -328,10 +327,10 @@ void raid_delta_gen(int nr, const int* id, const int* ip, int nd, size_t size, v
  *
  * Dx = Pd
  */
-void raid_rec1_par1(const int* id, int nd, size_t size, void** v)
+void raid_rec1_par1(const int *id, int nd, size_t size, void **v)
 {
-	void* p;
-	void* pa;
+	void *p;
+	void *pa;
 
 	/* for PAR1 we can directly compute the missing block */
 	/* and we don't need to use the zero buffer */
@@ -375,19 +374,19 @@ void raid_rec1_par1(const int* id, int nd, size_t size, void** v)
  *
  * That are always satisfied for any 0<=id[0]<id[1]<255.
  */
-void raid_rec2_par2(const int* id, const int* ip, int nd, size_t size, void** vv)
+void raid_rec2_par2(const int *id, const int *ip, int nd, size_t size, void **vv)
 {
-	uint8_t** v = (uint8_t**)vv;
+	uint8_t **v = (uint8_t **)vv;
 	size_t i;
-	uint8_t* p;
-	uint8_t* pa;
-	uint8_t* q;
-	uint8_t* qa;
-	const uint8_t* T[2];
+	uint8_t *p;
+	uint8_t *pa;
+	uint8_t *q;
+	uint8_t *qa;
+	const uint8_t *T[2];
 
 	/* get multiplication tables */
-	T[0] = table( inv(pow2(id[1]-id[0]) ^ 1) );
-	T[1] = table( inv(pow2(id[0]) ^ pow2(id[1])) );
+	T[0] = table(inv(pow2(id[1]-id[0]) ^ 1));
+	T[1] = table(inv(pow2(id[0]) ^ pow2(id[1])));
 
 	/* compute delta parity */
 	raid_delta_gen(2, id, ip, nd, size, vv);
@@ -397,7 +396,7 @@ void raid_rec2_par2(const int* id, const int* ip, int nd, size_t size, void** vv
 	pa = v[id[0]];
 	qa = v[id[1]];
 
-	for(i=0;i<size;++i) {
+	for (i = 0; i < size; ++i) {
 		/* delta */
 		uint8_t Pd = p[i] ^ pa[i];
 		uint8_t Qd = q[i] ^ qa[i];
@@ -413,22 +412,23 @@ void raid_rec2_par2(const int* id, const int* ip, int nd, size_t size, void** vv
 }
 
 /* internal forwarder */
-void (*raid_rec_ptr[RAID_PARITY_MAX])(int nr, const int* id, const int* ip, int nd, size_t size, void** vv);
+void (*raid_rec_ptr[RAID_PARITY_MAX])(int nr, const int *id, const int *ip, int nd, size_t size, void **vv);
 
-void raid_rec(int nr, const int* id, const int* ip, int nd, size_t size, void** v)
+void raid_rec(int nr, const int *id, const int *ip, int nd, size_t size, void **v)
 {
 	BUG_ON(nr > nd);
+	BUG_ON(size % 64 != 0);
 
 	/* if failed data is present */
-	if (nr != 0) {
+	if (nr != 0)
 		raid_rec_ptr[nr - 1](nr, id, ip, nd, size, v);
-	}
 }
 
-void raid_recpar(int nrd, const int* id, int nrp, int* ip, int np, int nd, size_t size, void** v)
+void raid_recpar(int nrd, const int *id, int nrp, int *ip, int np, int nd, size_t size, void **v)
 {
 	BUG_ON(nrd > nd);
 	BUG_ON(nrd + nrp > np);
+	BUG_ON(size % 64 != 0);
 
 	/* if failed data is present */
 	if (nrd != 0) {
@@ -436,7 +436,7 @@ void raid_recpar(int nrd, const int* id, int nrp, int* ip, int np, int nd, size_
 		int i, j, k;
 
 		/* setup the vector of parities to use */
-		for(i=0,j=0,k=0;i<np;++i) {
+		for (i = 0, j = 0, k = 0; i < np; ++i) {
 			if (ip[j] == i) {
 				++j;
 			} else {
@@ -450,9 +450,7 @@ void raid_recpar(int nrd, const int* id, int nrp, int* ip, int np, int nd, size_
 	}
 
 	/* recompute all the parities up to the last bad one */
-	if (nrp != 0) {
+	if (nrp != 0)
 		raid_par(ip[nrp - 1] + 1, nd, size, v);
-	}
 }
 
-
diff --git a/raid/raid.h b/raid/raid.h
index bbab02b4..0782db38 100644
--- a/raid/raid.h
+++ b/raid/raid.h
@@ -21,7 +21,7 @@
 /**
  * RAID mode supporting up to 6 parities.
  *
- * It requires SSSE3 to get good performance with triple or more 
+ * It requires SSSE3 to get good performance with triple or more
  * parities.
  *
  * This is the default mode used by SnapRAID.
@@ -70,7 +70,7 @@ void raid_mode(int mode);
  * Before calling raid_rec() and raid_recpar() you must provide a memory
  * buffer filled with zero with the same size of the blocks to recover.
  */
-void raid_zero(void* zero);
+void raid_zero(void *zero);
 
 /**
  * Computes the parity.
@@ -79,10 +79,11 @@ void raid_zero(void* zero);
  * \param nd Number of data disks.
  * \param size Size of the blocks pointed by vv. It must be a multipler of 64.
  * \param vv Vector of pointers to the blocks for disks and parities.
- *   It has (::nd + ::np) elements.
- *   Each element points to a buffer of ::size bytes.
+ *   It has (::nd + ::np) elements. The first elements are the blocks for
+ *   data, following with the parity blocks.
+ *   Each blocks has ::size bytes.
  */
-void raid_par(int np, int nd, size_t size, void** vv);
+void raid_par(int np, int nd, size_t size, void **vv);
 
 /**
  * Recovers failures of data disks using the specified parities.
@@ -98,10 +99,11 @@ void raid_par(int np, int nd, size_t size, void** vv);
  * \param nd Number of data disks.
  * \param size Size of the blocks pointed by vv. It must be a multipler of 64.
  * \param vv Vector of pointers to the blocks for disks and parities.
- *   It has (::nd + ::ip[::nr - 1] + 1) elements. The first elements are the blocks
- *   for data, following with the parity blocks. Each blocks has ::size bytes.
+ *   It has (::nd + ::ip[::nr - 1] + 1) elements. The first elements are the
+ *   blocks for data, following with the parity blocks.
+ *   Each blocks has ::size bytes.
  */
-void raid_rec(int nr, const int* id, const int* ip, int nd, size_t size, void** vv);
+void raid_rec(int nr, const int *id, const int *ip, int nd, size_t size, void **vv);
 
 /**
  * Recovers failures of data and parity disks.
@@ -127,8 +129,9 @@ void raid_rec(int nr, const int* id, const int* ip, int nd, size_t size, void**
  * \param size Size of the blocks pointed by vv. It must be a multipler of 64.
  * \param vv Vector of pointers to the blocks for disks and parities.
  *   It has (::nd + ::np) elements. The first elements are the blocks
- *   for data, following with the parity blocks. Each blocks has ::size bytes. 
+ *   for data, following with the parity blocks.
+ *   Each blocks has ::size bytes.
  */
-void raid_recpar(int nrd, const int* id, int nrp, int* ip, int np, int nd, size_t size, void** vv);
+void raid_recpar(int nrd, const int *id, int nrp, int *ip, int np, int nd, size_t size, void **vv);
 
 #endif
diff --git a/raid/tables.c b/raid/tables.c
index cb5bda37..cb41a9c9 100644
--- a/raid/tables.c
+++ b/raid/tables.c
@@ -17,7 +17,7 @@
 
 #include "internal.h"
 
-const uint8_t __attribute__((aligned(256))) raid_gfmul[256][256] =
+const uint8_t __aligned(256) raid_gfmul[256][256] =
 {
 	{
 		0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
@@ -8725,7 +8725,7 @@ const uint8_t __attribute__((aligned(256))) raid_gfmul[256][256] =
 	},
 };
 
-const uint8_t __attribute__((aligned(256))) raid_gfexp[256] =
+const uint8_t __aligned(256) raid_gfexp[256] =
 {
 	0x01, 0x02, 0x04, 0x08, 0x10, 0x20, 0x40, 0x80,
 	0x1d, 0x3a, 0x74, 0xe8, 0xcd, 0x87, 0x13, 0x26,
@@ -8761,7 +8761,7 @@ const uint8_t __attribute__((aligned(256))) raid_gfexp[256] =
 	0x1b, 0x36, 0x6c, 0xd8, 0xad, 0x47, 0x8e, 0x01,
 };
 
-const uint8_t __attribute__((aligned(256))) raid_gfinv[256] =
+const uint8_t __aligned(256) raid_gfinv[256] =
 {
 	/* note that the first element is not significative */
 	0x00, 0x01, 0x8e, 0xf4, 0x47, 0xa7, 0x7a, 0xba,
@@ -8806,7 +8806,7 @@ const uint8_t __attribute__((aligned(256))) raid_gfinv[256] =
  * 01 02 04 08 10 20 40 80 1d 3a 74 e8 cd 87 13 26 4c 98 2d 5a b4 75 ea c9 8f 03 06 0c 18 30 60 c0 9d 27 4e 9c 25 4a 94 35 6a d4 b5 77 ee c1 9f 23 46 8c 05 0a 14 28 50 a0 5d ba 69 d2 b9 6f de a1 5f be 61 c2 99 2f 5e bc 65 ca 89 0f 1e 3c 78 f0 fd e7 d3 bb 6b d6 b1 7f fe e1 df a3 5b b6 71 e2 d9 af 43 86 11 22 44 88 0d 1a 34 68 d0 bd 67 ce 81 1f 3e 7c f8 ed c7 93 3b 76 ec c5 97 33 66 cc 85 17 2e 5c b8 6d da a9 4f 9e 21 42 84 15 2a 54 a8 4d 9a 29 52 a4 55 aa 49 92 39 72 e4 d5 b7 73 e6 d1 bf 63 c6 91 3f 7e fc e5 d7 b3 7b f6 f1 ff e3 db ab 4b 96 31 62 c4 95 37 6e dc a5 57 ae 41 82 19 32 64 c8 8d 07 0e 1c 38 70 e0 dd a7 53 a6 51 a2 59 b2 79 f2 f9 ef c3 9b 2b 56 ac 45 8a 09 12 24 48 90 3d 7a f4 f5 f7 f3 fb eb cb 8b 0b 16 2c 58 b0 7d fa e9 cf 83 1b 36 6c 
  * 01 8e 47 ad d8 6c 36 1b 83 cf e9 fa 7d b0 58 2c 16 0b 8b cb eb fb f3 f7 f5 f4 7a 3d 90 48 24 12 09 8a 45 ac 56 2b 9b c3 ef f9 f2 79 b2 59 a2 51 a6 53 a7 dd e0 70 38 1c 0e 07 8d c8 64 32 19 82 41 ae 57 a5 dc 6e 37 95 c4 62 31 96 4b ab db e3 ff f1 f6 7b b3 d7 e5 fc 7e 3f 91 c6 63 bf d1 e6 73 b7 d5 e4 72 39 92 49 aa 55 a4 52 29 9a 4d a8 54 2a 15 84 42 21 9e 4f a9 da 6d b8 5c 2e 17 85 cc 66 33 97 c5 ec 76 3b 93 c7 ed f8 7c 3e 1f 81 ce 67 bd d0 68 34 1a 0d 88 44 22 11 86 43 af d9 e2 71 b6 5b a3 df e1 fe 7f b1 d6 6b bb d3 e7 fd f0 78 3c 1e 0f 89 ca 65 bc 5e 2f 99 c2 61 be 5f a1 de 6f b9 d2 69 ba 5d a0 50 28 14 0a 05 8c 46 23 9f c1 ee 77 b5 d4 6a 35 94 4a 25 9c 4e 27 9d c0 60 30 18 0c 06 03 8f c9 ea 75 b4 5a 2d 98 4c 26 13 87 cd e8 74 3a 1d 80 40 20 
  */
-const uint8_t __attribute__((aligned(256))) raid_gfvandermonde[3][256] =
+const uint8_t __aligned(256) raid_gfvandermonde[3][256] =
 {
 	{
 		0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,
@@ -8923,7 +8923,7 @@ const uint8_t __attribute__((aligned(256))) raid_gfvandermonde[3][256] =
  * 01 97 7f 9c 7c 18 bd a2 58 1a da 74 70 a3 e5 47 29 07 f5 80 23 e9 fa 46 54 a0 99 95 53 9b 0b c7 09 c0 78 89 92 e3 0d b0 2a 8c fb 17 3f 26 65 87 27 5c 66 61 79 4d 32 b3 8d 52 e2 82 3d f9 c5 02 bc 4c 73 48 62 af ba 41 d9 c4 2f b1 33 b8 15 7d cf 3a a9 5f 84 6d 34 1b 44 94 72 81 42 be cc 4b 0a 6f 5a 22 36 b5 3c 9d 13 7e 08 dd d6 5e 04 fc 5b ec ef f1 6e 1e 77 24 e6 c6 aa cb fd 51 67 06 6a 4a 88 db b2 c2 5d 43 40 f7 50 a8 f2 7a 71 a4 d2 bf 31 90 19 9a 8e f6 c3 a6 e7 60 12 ee 2d de 38 e8 b7 98 c1 28 f3 05 96 63 d1 b9 14 9f 1d 83 68 75 ed 16 03 ce e4 df e0 10 ae 69 55 91 2e 4e fe 21 1f 9e e1 d5 cd ca f0 8b 2b c9 8a 93 bb 57 20 86 1c a1 4f 3e 25 d4 6c a5 6b a7 37 ff 39 35 0c f8 ea 56 45 8f 2c 59 ab 85 eb 49 0f dc d8 76 b6 f4 0e 11 b4 d0 30 d3 3b ad d7 
  * 01 2b 3f cf 73 2c d6 ed cb 74 15 78 8a c1 17 c9 89 68 21 ab 76 3b 4b 5a 6e 0e b9 d3 b6 3e 36 86 bf a2 a7 30 14 eb c7 2d 96 67 20 b5 9a e0 a8 c6 80 04 8d fe 75 5e 23 ca 8f 48 99 0d df 8e b8 70 29 9c 44 69 3d a5 c2 90 d2 1c 9b 02 1d 98 93 ec 84 e8 64 4c 3a 8b 97 f3 e5 c0 7d 26 c8 08 a0 62 82 55 f7 33 f6 51 63 4d 77 da fd c3 38 6d ee 09 47 a3 05 de a6 f1 22 25 6a 0c 81 b2 6b 58 d5 b3 fc fb 28 7f 07 dc 7a 9e d0 37 b4 e1 1a 24 03 ae 94 ba 88 2f ea 2e 8c 5b bb 79 d1 11 ff a4 19 3c 2a 4e 52 e3 95 bd 31 5d 35 4a 41 c4 db 42 c5 0b 49 1b 7c e4 b0 9d 45 f0 a9 61 57 06 d4 40 91 56 13 fa 87 ac 27 54 dd 59 1f 71 39 43 6c f9 be 4f f4 1e 32 cd e9 7e 7b 66 5f ef e7 6f 0a 60 d7 b7 83 92 e2 af 72 f8 b1 50 10 ce 18 53 a1 cc ad 12 34 0f f5 aa 16 e6 f2 d8 85 9f bc 
  */
-const uint8_t __attribute__((aligned(256))) raid_gfcauchy[6][256] =
+const uint8_t __aligned(256) raid_gfcauchy[6][256] =
 {
 	{
 		0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,
@@ -9138,7 +9138,7 @@ const uint8_t __attribute__((aligned(256))) raid_gfcauchy[6][256] =
  * Indexes are [DISK][PARITY - 2][LH].
  * Where DISK is from 0 to 250, PARITY from 2 to 5, LH from 0 to 1.
  */
-const uint8_t __attribute__((aligned(256))) raid_gfcauchypshufb[251][4][2][16] =
+const uint8_t __aligned(256) raid_gfcauchypshufb[251][4][2][16] =
 {
 	{
 		{
@@ -13668,7 +13668,7 @@ const uint8_t __attribute__((aligned(256))) raid_gfcauchypshufb[251][4][2][16] =
  * Indexes are [MULTIPLER][LH].
  * Where MULTIPLER is from 0 to 255, LH from 0 to 1.
  */
-const uint8_t __attribute__((aligned(256))) raid_gfmulpshufb[256][2][16] =
+const uint8_t __aligned(256) raid_gfmulpshufb[256][2][16] =
 {
 	{
 		{ 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00 },
diff --git a/raid/x86.c b/raid/x86.c
index 29f1922d..c0f6ad4e 100644
--- a/raid/x86.c
+++ b/raid/x86.c
@@ -27,24 +27,26 @@
  * the typical cache block, and processing 128 bytes doesn't increase
  * performance.
  */
-void raid_par1_sse2(int nd, size_t size, void** vv)
+void raid_par1_sse2(int nd, size_t size, void **vv)
 {
-	uint8_t** v = (uint8_t**)vv;
-	uint8_t* p;
+	uint8_t **v = (uint8_t **)vv;
+	uint8_t *p;
 	int d, l;
 	size_t i;
 
 	l = nd - 1;
 	p = v[nd];
 
-	for(i=0;i<size;i+=64) {
+	asm_begin();
+
+	for (i = 0; i < size; i += 64) {
 		asm volatile("movdqa %0,%%xmm0" : : "m" (v[l][i]));
 		asm volatile("movdqa %0,%%xmm1" : : "m" (v[l][i+16]));
 		asm volatile("movdqa %0,%%xmm2" : : "m" (v[l][i+32]));
 		asm volatile("movdqa %0,%%xmm3" : : "m" (v[l][i+48]));
 		/* accessing disks in backward order because the buffers */
 		/* are also in backward order */
-		for(d=l-1;d>=0;--d) {
+		for (d = l-1; d >= 0; --d) {
 			asm volatile("movdqa %0,%%xmm4" : : "m" (v[d][i]));
 			asm volatile("movdqa %0,%%xmm5" : : "m" (v[d][i+16]));
 			asm volatile("movdqa %0,%%xmm6" : : "m" (v[d][i+32]));
@@ -60,7 +62,7 @@ void raid_par1_sse2(int nd, size_t size, void** vv)
 		asm volatile("movntdq %%xmm3,%0" : "=m" (p[i+48]));
 	}
 
-	asm volatile("sfence" : : : "memory");
+	asm_end();
 }
 #endif
 
@@ -68,19 +70,21 @@ void raid_par1_sse2(int nd, size_t size, void** vv)
 static const struct gfconst16 {
 	uint8_t poly[16];
 	uint8_t low4[16];
-} gfconst16  __attribute__((aligned(64))) = {
+} gfconst16  __aligned(32) = {
 	{ 0x1d, 0x1d, 0x1d, 0x1d, 0x1d, 0x1d, 0x1d, 0x1d, 0x1d, 0x1d, 0x1d, 0x1d, 0x1d, 0x1d, 0x1d, 0x1d },
 	{ 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f },
 };
+#endif
 
+#if defined(__i386__) || defined(__x86_64__)
 /*
  * PAR2 (RAID6 with powers of 2) SSE2 implementation
  */
-void raid_par2_sse2(int nd, size_t size, void** vv)
+void raid_par2_sse2(int nd, size_t size, void **vv)
 {
-	uint8_t** v = (uint8_t**)vv;
-	uint8_t* p;
-	uint8_t* q;
+	uint8_t **v = (uint8_t **)vv;
+	uint8_t *p;
+	uint8_t *q;
 	int d, l;
 	size_t i;
 
@@ -88,14 +92,16 @@ void raid_par2_sse2(int nd, size_t size, void** vv)
 	p = v[nd];
 	q = v[nd+1];
 
+	asm_begin();
+
 	asm volatile("movdqa %0,%%xmm7" : : "m" (gfconst16.poly[0]));
 
-	for(i=0;i<size;i+=32) {
+	for (i = 0; i < size; i += 32) {
 		asm volatile("movdqa %0,%%xmm0" : : "m" (v[l][i]));
 		asm volatile("movdqa %0,%%xmm1" : : "m" (v[l][i+16]));
 		asm volatile("movdqa %xmm0,%xmm2");
 		asm volatile("movdqa %xmm1,%xmm3");
-		for(d=l-1;d>=0;--d) {
+		for (d = l-1; d >= 0; --d) {
 			asm volatile("pxor %xmm4,%xmm4");
 			asm volatile("pxor %xmm5,%xmm5");
 			asm volatile("pcmpgtb %xmm2,%xmm4");
@@ -120,7 +126,7 @@ void raid_par2_sse2(int nd, size_t size, void** vv)
 		asm volatile("movntdq %%xmm3,%0" : "=m" (q[i+16]));
 	}
 
-	asm volatile("sfence" : : : "memory");
+	asm_end();
 }
 #endif
 
@@ -130,11 +136,11 @@ void raid_par2_sse2(int nd, size_t size, void** vv)
  *
  * Note that it uses 16 registers, meaning that x64 is required.
  */
-void raid_par2_sse2ext(int nd, size_t size, void** vv)
+void raid_par2_sse2ext(int nd, size_t size, void **vv)
 {
-	uint8_t** v = (uint8_t**)vv;
-	uint8_t* p;
-	uint8_t* q;
+	uint8_t **v = (uint8_t **)vv;
+	uint8_t *p;
+	uint8_t *q;
 	int d, l;
 	size_t i;
 
@@ -142,9 +148,11 @@ void raid_par2_sse2ext(int nd, size_t size, void** vv)
 	p = v[nd];
 	q = v[nd+1];
 
+	asm_begin();
+
 	asm volatile("movdqa %0,%%xmm15" : : "m" (gfconst16.poly[0]));
 
-	for(i=0;i<size;i+=64) {
+	for (i = 0; i < size; i += 64) {
 		asm volatile("movdqa %0,%%xmm0" : : "m" (v[l][i]));
 		asm volatile("movdqa %0,%%xmm1" : : "m" (v[l][i+16]));
 		asm volatile("movdqa %0,%%xmm2" : : "m" (v[l][i+32]));
@@ -153,7 +161,7 @@ void raid_par2_sse2ext(int nd, size_t size, void** vv)
 		asm volatile("movdqa %xmm1,%xmm5");
 		asm volatile("movdqa %xmm2,%xmm6");
 		asm volatile("movdqa %xmm3,%xmm7");
-		for(d=l-1;d>=0;--d) {
+		for (d = l-1; d >= 0; --d) {
 			asm volatile("pxor %xmm8,%xmm8");
 			asm volatile("pxor %xmm9,%xmm9");
 			asm volatile("pxor %xmm10,%xmm10");
@@ -198,7 +206,7 @@ void raid_par2_sse2ext(int nd, size_t size, void** vv)
 		asm volatile("movntdq %%xmm7,%0" : "=m" (q[i+48]));
 	}
 
-	asm volatile("sfence" : : : "memory");
+	asm_end();
 }
 #endif
 
@@ -206,12 +214,12 @@ void raid_par2_sse2ext(int nd, size_t size, void** vv)
 /*
  * PAR3 (triple parity with Cauchy matrix) SSSE3 implementation
  */
-void raid_par3_ssse3(int nd, size_t size, void** vv)
+void raid_par3_ssse3(int nd, size_t size, void **vv)
 {
-	uint8_t** v = (uint8_t**)vv;
-	uint8_t* p;
-	uint8_t* q;
-	uint8_t* r;
+	uint8_t **v = (uint8_t **)vv;
+	uint8_t *p;
+	uint8_t *q;
+	uint8_t *r;
 	int d, l;
 	size_t i;
 
@@ -222,16 +230,18 @@ void raid_par3_ssse3(int nd, size_t size, void** vv)
 
 	/* special case with only one data disk */
 	if (l == 0) {
-		for(i=0;i<3;++i)
+		for (i = 0; i < 3; ++i)
 			memcpy(v[1+i], v[0], size);
 		return;
 	}
 
+	asm_begin();
+
 	/* generic case with at least two data disks */
 	asm volatile("movdqa %0,%%xmm3" : : "m" (gfconst16.poly[0]));
 	asm volatile("movdqa %0,%%xmm7" : : "m" (gfconst16.low4[0]));
 
-	for(i=0;i<size;i+=16) {
+	for (i = 0; i < size; i += 16) {
 		/* last disk without the by two multiplication */
 		asm volatile("movdqa %0,%%xmm4" : : "m" (v[l][i]));
 
@@ -250,7 +260,7 @@ void raid_par3_ssse3(int nd, size_t size, void** vv)
 		asm volatile("pxor   %xmm6,%xmm2");
 
 		/* intermediate disks */
-		for(d=l-1;d>0;--d) {
+		for (d = l-1; d > 0; --d) {
 			asm volatile("movdqa %0,%%xmm4" : : "m" (v[d][i]));
 
 			asm volatile("pxor %xmm5,%xmm5");
@@ -287,13 +297,13 @@ void raid_par3_ssse3(int nd, size_t size, void** vv)
 		asm volatile("pxor %xmm4,%xmm0");
 		asm volatile("pxor %xmm4,%xmm1");
 		asm volatile("pxor %xmm4,%xmm2");
-		
+
 		asm volatile("movntdq %%xmm0,%0" : "=m" (p[i]));
 		asm volatile("movntdq %%xmm1,%0" : "=m" (q[i]));
 		asm volatile("movntdq %%xmm2,%0" : "=m" (r[i]));
 	}
 
-	asm volatile("sfence" : : : "memory");
+	asm_end();
 }
 #endif
 
@@ -303,12 +313,12 @@ void raid_par3_ssse3(int nd, size_t size, void** vv)
  *
  * Note that it uses 16 registers, meaning that x64 is required.
  */
-void raid_par3_ssse3ext(int nd, size_t size, void** vv)
+void raid_par3_ssse3ext(int nd, size_t size, void **vv)
 {
-	uint8_t** v = (uint8_t**)vv;
-	uint8_t* p;
-	uint8_t* q;
-	uint8_t* r;
+	uint8_t **v = (uint8_t **)vv;
+	uint8_t *p;
+	uint8_t *q;
+	uint8_t *r;
 	int d, l;
 	size_t i;
 
@@ -319,16 +329,18 @@ void raid_par3_ssse3ext(int nd, size_t size, void** vv)
 
 	/* special case with only one data disk */
 	if (l == 0) {
-		for(i=0;i<3;++i)
+		for (i = 0; i < 3; ++i)
 			memcpy(v[1+i], v[0], size);
 		return;
 	}
 
+	asm_begin();
+
 	/* generic case with at least two data disks */
 	asm volatile("movdqa %0,%%xmm3" : : "m" (gfconst16.poly[0]));
 	asm volatile("movdqa %0,%%xmm11" : : "m" (gfconst16.low4[0]));
 
-	for(i=0;i<size;i+=32) {
+	for (i = 0; i < size; i += 32) {
 		/* last disk without the by two multiplication */
 		asm volatile("movdqa %0,%%xmm4" : : "m" (v[l][i]));
 		asm volatile("movdqa %0,%%xmm12" : : "m" (v[l][i+16]));
@@ -359,7 +371,7 @@ void raid_par3_ssse3ext(int nd, size_t size, void** vv)
 		asm volatile("pxor   %xmm15,%xmm10");
 
 		/* intermediate disks */
-		for(d=l-1;d>0;--d) {
+		for (d = l-1; d > 0; --d) {
 			asm volatile("movdqa %0,%%xmm4" : : "m" (v[d][i]));
 			asm volatile("movdqa %0,%%xmm12" : : "m" (v[d][i+16]));
 
@@ -432,7 +444,7 @@ void raid_par3_ssse3ext(int nd, size_t size, void** vv)
 		asm volatile("movntdq %%xmm10,%0" : "=m" (r[i+16]));
 	}
 
-	asm volatile("sfence" : : : "memory");
+	asm_end();
 }
 #endif
 
@@ -440,13 +452,13 @@ void raid_par3_ssse3ext(int nd, size_t size, void** vv)
 /*
  * PAR4 (quad parity with Cauchy matrix) SSSE3 implementation
  */
-void raid_par4_ssse3(int nd, size_t size, void** vv)
+void raid_par4_ssse3(int nd, size_t size, void **vv)
 {
-	uint8_t** v = (uint8_t**)vv;
-	uint8_t* p;
-	uint8_t* q;
-	uint8_t* r;
-	uint8_t* s;
+	uint8_t **v = (uint8_t **)vv;
+	uint8_t *p;
+	uint8_t *q;
+	uint8_t *r;
+	uint8_t *s;
 	int d, l;
 	size_t i;
 
@@ -458,13 +470,15 @@ void raid_par4_ssse3(int nd, size_t size, void** vv)
 
 	/* special case with only one data disk */
 	if (l == 0) {
-		for(i=0;i<4;++i)
+		for (i = 0; i < 4; ++i)
 			memcpy(v[1+i], v[0], size);
 		return;
 	}
 
+	asm_begin();
+
 	/* generic case with at least two data disks */
-	for(i=0;i<size;i+=16) {
+	for (i = 0; i < size; i += 16) {
 		/* last disk without the by two multiplication */
 		asm volatile("movdqa %0,%%xmm7" : : "m" (gfconst16.low4[0]));
 		asm volatile("movdqa %0,%%xmm4" : : "m" (v[l][i]));
@@ -490,7 +504,7 @@ void raid_par4_ssse3(int nd, size_t size, void** vv)
 		asm volatile("pxor   %xmm7,%xmm3");
 
 		/* intermediate disks */
-		for(d=l-1;d>0;--d) {
+		for (d = l-1; d > 0; --d) {
 			asm volatile("movdqa %0,%%xmm7" : : "m" (gfconst16.poly[0]));
 			asm volatile("movdqa %0,%%xmm4" : : "m" (v[d][i]));
 
@@ -546,7 +560,7 @@ void raid_par4_ssse3(int nd, size_t size, void** vv)
 		asm volatile("movntdq %%xmm3,%0" : "=m" (s[i]));
 	}
 
-	asm volatile("sfence" : : : "memory");
+	asm_end();
 }
 #endif
 
@@ -556,13 +570,13 @@ void raid_par4_ssse3(int nd, size_t size, void** vv)
  *
  * Note that it uses 16 registers, meaning that x64 is required.
  */
-void raid_par4_ssse3ext(int nd, size_t size, void** vv)
+void raid_par4_ssse3ext(int nd, size_t size, void **vv)
 {
-	uint8_t** v = (uint8_t**)vv;
-	uint8_t* p;
-	uint8_t* q;
-	uint8_t* r;
-	uint8_t* s;
+	uint8_t **v = (uint8_t **)vv;
+	uint8_t *p;
+	uint8_t *q;
+	uint8_t *r;
+	uint8_t *s;
 	int d, l;
 	size_t i;
 
@@ -574,13 +588,15 @@ void raid_par4_ssse3ext(int nd, size_t size, void** vv)
 
 	/* special case with only one data disk */
 	if (l == 0) {
-		for(i=0;i<4;++i)
+		for (i = 0; i < 4; ++i)
 			memcpy(v[1+i], v[0], size);
 		return;
 	}
 
+	asm_begin();
+
 	/* generic case with at least two data disks */
-	for(i=0;i<size;i+=32) {
+	for (i = 0; i < size; i += 32) {
 		/* last disk without the by two multiplication */
 		asm volatile("movdqa %0,%%xmm15" : : "m" (gfconst16.low4[0]));
 		asm volatile("movdqa %0,%%xmm4" : : "m" (v[l][i]));
@@ -623,7 +639,7 @@ void raid_par4_ssse3ext(int nd, size_t size, void** vv)
 		asm volatile("pxor   %xmm15,%xmm11");
 
 		/* intermediate disks */
-		for(d=l-1;d>0;--d) {
+		for (d = l-1; d > 0; --d) {
 			asm volatile("movdqa %0,%%xmm7" : : "m" (gfconst16.poly[0]));
 			asm volatile("movdqa %0,%%xmm15" : : "m" (gfconst16.low4[0]));
 			asm volatile("movdqa %0,%%xmm4" : : "m" (v[d][i]));
@@ -717,7 +733,7 @@ void raid_par4_ssse3ext(int nd, size_t size, void** vv)
 		asm volatile("movntdq %%xmm11,%0" : "=m" (s[i+16]));
 	}
 
-	asm volatile("sfence" : : : "memory");
+	asm_end();
 }
 #endif
 
@@ -729,18 +745,18 @@ void raid_par4_ssse3ext(int nd, size_t size, void** vv)
 /* ensures that stack is aligned at 16 bytes because we allocate SSE registers in it */
 __attribute__((force_align_arg_pointer))
 #endif
-void raid_par5_ssse3(int nd, size_t size, void** vv)
+void raid_par5_ssse3(int nd, size_t size, void **vv)
 /* ensures that stack is aligned at 16 bytes because we allocate SSE registers in it */
 {
-	uint8_t** v = (uint8_t**)vv;
-	uint8_t* p;
-	uint8_t* q;
-	uint8_t* r;
-	uint8_t* s;
-	uint8_t* t;
+	uint8_t **v = (uint8_t **)vv;
+	uint8_t *p;
+	uint8_t *q;
+	uint8_t *r;
+	uint8_t *s;
+	uint8_t *t;
 	int d, l;
 	size_t i;
-	uint8_t p0[16] __attribute__((aligned(16)));
+	uint8_t p0[16] __aligned(16);
 
 	l = nd - 1;
 	p = v[nd];
@@ -751,13 +767,15 @@ void raid_par5_ssse3(int nd, size_t size, void** vv)
 
 	/* special case with only one data disk */
 	if (l == 0) {
-		for(i=0;i<5;++i)
+		for (i = 0; i < 5; ++i)
 			memcpy(v[1+i], v[0], size);
 		return;
 	}
 
+	asm_begin();
+
 	/* generic case with at least two data disks */
-	for(i=0;i<size;i+=16) {
+	for (i = 0; i < size; i += 16) {
 		/* last disk without the by two multiplication */
 		asm volatile("movdqa %0,%%xmm4" : : "m" (v[l][i]));
 
@@ -789,7 +807,7 @@ void raid_par5_ssse3(int nd, size_t size, void** vv)
 		asm volatile("pxor   %xmm7,%xmm3");
 
 		/* intermediate disks */
-		for(d=l-1;d>0;--d) {
+		for (d = l-1; d > 0; --d) {
 			asm volatile("movdqa %0,%%xmm4" : : "m" (v[d][i]));
 			asm volatile("movdqa %0,%%xmm6" : : "m" (p0[0]));
 			asm volatile("movdqa %0,%%xmm7" : : "m" (gfconst16.poly[0]));
@@ -856,7 +874,7 @@ void raid_par5_ssse3(int nd, size_t size, void** vv)
 		asm volatile("movntdq %%xmm3,%0" : "=m" (t[i]));
 	}
 
-	asm volatile("sfence" : : : "memory");
+	asm_end();
 }
 #endif
 
@@ -866,14 +884,14 @@ void raid_par5_ssse3(int nd, size_t size, void** vv)
  *
  * Note that it uses 16 registers, meaning that x64 is required.
  */
-void raid_par5_ssse3ext(int nd, size_t size, void** vv)
+void raid_par5_ssse3ext(int nd, size_t size, void **vv)
 {
-	uint8_t** v = (uint8_t**)vv;
-	uint8_t* p;
-	uint8_t* q;
-	uint8_t* r;
-	uint8_t* s;
-	uint8_t* t;
+	uint8_t **v = (uint8_t **)vv;
+	uint8_t *p;
+	uint8_t *q;
+	uint8_t *r;
+	uint8_t *s;
+	uint8_t *t;
 	int d, l;
 	size_t i;
 
@@ -886,16 +904,18 @@ void raid_par5_ssse3ext(int nd, size_t size, void** vv)
 
 	/* special case with only one data disk */
 	if (l == 0) {
-		for(i=0;i<5;++i)
+		for (i = 0; i < 5; ++i)
 			memcpy(v[1+i], v[0], size);
 		return;
 	}
 
+	asm_begin();
+
 	/* generic case with at least two data disks */
 	asm volatile("movdqa %0,%%xmm14" : : "m" (gfconst16.poly[0]));
 	asm volatile("movdqa %0,%%xmm15" : : "m" (gfconst16.low4[0]));
 
-	for(i=0;i<size;i+=16) {
+	for (i = 0; i < size; i += 16) {
 		/* last disk without the by two multiplication */
 		asm volatile("movdqa %0,%%xmm10" : : "m" (v[l][i]));
 
@@ -926,7 +946,7 @@ void raid_par5_ssse3ext(int nd, size_t size, void** vv)
 		asm volatile("pxor   %xmm13,%xmm4");
 
 		/* intermediate disks */
-		for(d=l-1;d>0;--d) {
+		for (d = l-1; d > 0; --d) {
 			asm volatile("movdqa %0,%%xmm10" : : "m" (v[d][i]));
 
 			asm volatile("pxor %xmm11,%xmm11");
@@ -987,7 +1007,7 @@ void raid_par5_ssse3ext(int nd, size_t size, void** vv)
 		asm volatile("movntdq %%xmm4,%0" : "=m" (t[i]));
 	}
 
-	asm volatile("sfence" : : : "memory");
+	asm_end();
 }
 #endif
 
@@ -999,19 +1019,19 @@ void raid_par5_ssse3ext(int nd, size_t size, void** vv)
 /* ensures that stack is aligned at 16 bytes because we allocate SSE registers in it */
 __attribute__((force_align_arg_pointer))
 #endif
-void raid_par6_ssse3(int nd, size_t size, void** vv)
+void raid_par6_ssse3(int nd, size_t size, void **vv)
 {
-	uint8_t** v = (uint8_t**)vv;
-	uint8_t* p;
-	uint8_t* q;
-	uint8_t* r;
-	uint8_t* s;
-	uint8_t* t;
-	uint8_t* u;
+	uint8_t **v = (uint8_t **)vv;
+	uint8_t *p;
+	uint8_t *q;
+	uint8_t *r;
+	uint8_t *s;
+	uint8_t *t;
+	uint8_t *u;
 	int d, l;
 	size_t i;
-	uint8_t p0[16] __attribute__((aligned(16)));
-	uint8_t q0[16] __attribute__((aligned(16))); 
+	uint8_t p0[16] __aligned(16);
+	uint8_t q0[16] __aligned(16);
 
 	l = nd - 1;
 	p = v[nd];
@@ -1023,13 +1043,15 @@ void raid_par6_ssse3(int nd, size_t size, void** vv)
 
 	/* special case with only one data disk */
 	if (l == 0) {
-		for(i=0;i<6;++i)
+		for (i = 0; i < 6; ++i)
 			memcpy(v[1+i], v[0], size);
 		return;
 	}
 
+	asm_begin();
+
 	/* generic case with at least two data disks */
-	for(i=0;i<size;i+=16) {
+	for (i = 0; i < size; i += 16) {
 		/* last disk without the by two multiplication */
 		asm volatile("movdqa %0,%%xmm4" : : "m" (v[l][i]));
 
@@ -1067,7 +1089,7 @@ void raid_par6_ssse3(int nd, size_t size, void** vv)
 		asm volatile("pxor   %xmm7,%xmm3");
 
 		/* intermediate disks */
-		for(d=l-1;d>0;--d) {
+		for (d = l-1; d > 0; --d) {
 			asm volatile("movdqa %0,%%xmm5" : : "m" (p0[0]));
 			asm volatile("movdqa %0,%%xmm6" : : "m" (q0[0]));
 			asm volatile("movdqa %0,%%xmm7" : : "m" (gfconst16.poly[0]));
@@ -1147,7 +1169,7 @@ void raid_par6_ssse3(int nd, size_t size, void** vv)
 		asm volatile("movntdq %%xmm3,%0" : "=m" (u[i]));
 	}
 
-	asm volatile("sfence" : : : "memory");
+	asm_end();
 }
 #endif
 
@@ -1157,15 +1179,15 @@ void raid_par6_ssse3(int nd, size_t size, void** vv)
  *
  * Note that it uses 16 registers, meaning that x64 is required.
  */
-void raid_par6_ssse3ext(int nd, size_t size, void** vv)
+void raid_par6_ssse3ext(int nd, size_t size, void **vv)
 {
-	uint8_t** v = (uint8_t**)vv;
-	uint8_t* p;
-	uint8_t* q;
-	uint8_t* r;
-	uint8_t* s;
-	uint8_t* t;
-	uint8_t* u;
+	uint8_t **v = (uint8_t **)vv;
+	uint8_t *p;
+	uint8_t *q;
+	uint8_t *r;
+	uint8_t *s;
+	uint8_t *t;
+	uint8_t *u;
 	int d, l;
 	size_t i;
 
@@ -1179,16 +1201,18 @@ void raid_par6_ssse3ext(int nd, size_t size, void** vv)
 
 	/* special case with only one data disk */
 	if (l == 0) {
-		for(i=0;i<6;++i)
+		for (i = 0; i < 6; ++i)
 			memcpy(v[1+i], v[0], size);
 		return;
 	}
 
+	asm_begin();
+
 	/* generic case with at least two data disks */
 	asm volatile("movdqa %0,%%xmm14" : : "m" (gfconst16.poly[0]));
 	asm volatile("movdqa %0,%%xmm15" : : "m" (gfconst16.low4[0]));
 
-	for(i=0;i<size;i+=16) {
+	for (i = 0; i < size; i += 16) {
 		/* last disk without the by two multiplication */
 		asm volatile("movdqa %0,%%xmm10" : : "m" (v[l][i]));
 
@@ -1225,7 +1249,7 @@ void raid_par6_ssse3ext(int nd, size_t size, void** vv)
 		asm volatile("pxor   %xmm13,%xmm5");
 
 		/* intermediate disks */
-		for(d=l-1;d>0;--d) {
+		for (d = l-1; d > 0; --d) {
 			asm volatile("movdqa %0,%%xmm10" : : "m" (v[d][i]));
 
 			asm volatile("pxor %xmm11,%xmm11");
@@ -1295,7 +1319,7 @@ void raid_par6_ssse3ext(int nd, size_t size, void** vv)
 		asm volatile("movntdq %%xmm5,%0" : "=m" (u[i]));
 	}
 
-	asm volatile("sfence" : : : "memory");
+	asm_end();
 }
 #endif
 
@@ -1303,11 +1327,11 @@ void raid_par6_ssse3ext(int nd, size_t size, void** vv)
 /*
  * RAID recovering for one disk SSSE3 implementation
  */
-void raid_rec1_ssse3(int nr, const int* id, const int* ip, int nd, size_t size, void** vv)
+void raid_rec1_ssse3(int nr, const int *id, const int *ip, int nd, size_t size, void **vv)
 {
-	uint8_t** v = (uint8_t**)vv;
-	uint8_t* p;
-	uint8_t* pa;
+	uint8_t **v = (uint8_t **)vv;
+	uint8_t *p;
+	uint8_t *pa;
 	uint8_t G;
 	uint8_t V;
 	size_t i;
@@ -1321,7 +1345,7 @@ void raid_rec1_ssse3(int nr, const int* id, const int* ip, int nd, size_t size,
 	}
 
 	/* setup the coefficients matrix */
-	G = A(ip[0],id[0]);
+	G = A(ip[0], id[0]);
 
 	/* invert it to solve the system of linear equations */
 	V = inv(G);
@@ -1332,11 +1356,13 @@ void raid_rec1_ssse3(int nr, const int* id, const int* ip, int nd, size_t size,
 	p = v[nd+ip[0]];
 	pa = v[id[0]];
 
+	asm_begin();
+
 	asm volatile("movdqa %0,%%xmm7" : : "m" (gfconst16.low4[0]));
 	asm volatile("movdqa %0,%%xmm4" : : "m" (gfmulpshufb[V][0][0]));
 	asm volatile("movdqa %0,%%xmm5" : : "m" (gfmulpshufb[V][1][0]));
 
-	for(i=0;i<size;i+=16) {
+	for (i = 0; i < size; i += 16) {
 		asm volatile("movdqa %0,%%xmm0" : : "m" (p[i]));
 		asm volatile("movdqa %0,%%xmm1" : : "m" (pa[i]));
 		asm volatile("movdqa %xmm4,%xmm2");
@@ -1352,7 +1378,7 @@ void raid_rec1_ssse3(int nr, const int* id, const int* ip, int nd, size_t size,
 		asm volatile("movdqa %%xmm2,%0" : "=m" (pa[i]));
 	}
 
-	asm volatile("sfence" : : : "memory");
+	asm_end();
 }
 #endif
 
@@ -1360,12 +1386,12 @@ void raid_rec1_ssse3(int nr, const int* id, const int* ip, int nd, size_t size,
 /*
  * RAID recovering for two disks SSSE3 implementation
  */
-void raid_rec2_ssse3(int nr, const int* id, const int* ip, int nd, size_t size, void** vv)
+void raid_rec2_ssse3(int nr, const int *id, const int *ip, int nd, size_t size, void **vv)
 {
-	uint8_t** v = (uint8_t**)vv;
+	uint8_t **v = (uint8_t **)vv;
 	const int N = 2;
-	uint8_t* p[N];
-	uint8_t* pa[N];
+	uint8_t *p[N];
+	uint8_t *pa[N];
 	uint8_t G[N*N];
 	uint8_t V[N*N];
 	size_t i;
@@ -1374,11 +1400,9 @@ void raid_rec2_ssse3(int nr, const int* id, const int* ip, int nd, size_t size,
 	(void)nr; /* unused, it's always 2 */
 
 	/* setup the coefficients matrix */
-	for(j=0;j<N;++j) {
-		for(k=0;k<N;++k) {
-			G[j*N+k] = A(ip[j],id[k]);
-		}
-	}
+	for (j = 0; j < N; ++j)
+		for (k = 0; k < N; ++k)
+			G[j*N+k] = A(ip[j], id[k]);
 
 	/* invert it to solve the system of linear equations */
 	raid_invert(G, V, N);
@@ -1386,14 +1410,16 @@ void raid_rec2_ssse3(int nr, const int* id, const int* ip, int nd, size_t size,
 	/* compute delta parity */
 	raid_delta_gen(N, id, ip, nd, size, vv);
 
-	for(j=0;j<N;++j) {
+	for (j = 0; j < N; ++j) {
 		p[j] = v[nd+ip[j]];
 		pa[j] = v[id[j]];
 	}
 
+	asm_begin();
+
 	asm volatile("movdqa %0,%%xmm7" : : "m" (gfconst16.low4[0]));
 
-	for(i=0;i<size;i+=16) {
+	for (i = 0; i < size; i += 16) {
 		asm volatile("movdqa %0,%%xmm0" : : "m" (p[0][i]));
 		asm volatile("movdqa %0,%%xmm2" : : "m" (pa[0][i]));
 		asm volatile("movdqa %0,%%xmm1" : : "m" (p[1][i]));
@@ -1458,7 +1484,7 @@ void raid_rec2_ssse3(int nr, const int* id, const int* ip, int nd, size_t size,
 		asm volatile("movdqa %%xmm6,%0" : "=m" (pa[1][i]));
 	}
 
-	asm volatile("sfence" : : : "memory");
+	asm_end();
 }
 #endif
 
@@ -1466,23 +1492,21 @@ void raid_rec2_ssse3(int nr, const int* id, const int* ip, int nd, size_t size,
 /*
  * RAID recovering SSSE3 implementation
  */
-void raid_recX_ssse3(int nr, const int* id, const int* ip, int nd, size_t size, void** vv)
+void raid_recX_ssse3(int nr, const int *id, const int *ip, int nd, size_t size, void **vv)
 {
-	uint8_t** v = (uint8_t**)vv;
+	uint8_t **v = (uint8_t **)vv;
 	int N = nr;
-	uint8_t* p[RAID_PARITY_MAX];
-	uint8_t* pa[RAID_PARITY_MAX];
+	uint8_t *p[RAID_PARITY_MAX];
+	uint8_t *pa[RAID_PARITY_MAX];
 	uint8_t G[RAID_PARITY_MAX*RAID_PARITY_MAX];
 	uint8_t V[RAID_PARITY_MAX*RAID_PARITY_MAX];
 	size_t i;
 	int j, k;
 
 	/* setup the coefficients matrix */
-	for(j=0;j<N;++j) {
-		for(k=0;k<N;++k) {
-			G[j*N+k] = A(ip[j],id[k]);
-		}
-	}
+	for (j = 0; j < N; ++j)
+		for (k = 0; k < N; ++k)
+			G[j*N+k] = A(ip[j], id[k]);
 
 	/* invert it to solve the system of linear equations */
 	raid_invert(G, V, N);
@@ -1490,18 +1514,20 @@ void raid_recX_ssse3(int nr, const int* id, const int* ip, int nd, size_t size,
 	/* compute delta parity */
 	raid_delta_gen(N, id, ip, nd, size, vv);
 
-	for(j=0;j<N;++j) {
+	for (j = 0; j < N; ++j) {
 		p[j] = v[nd+ip[j]];
 		pa[j] = v[id[j]];
 	}
 
+	asm_begin();
+
 	asm volatile("movdqa %0,%%xmm7" : : "m" (gfconst16.low4[0]));
 
-	for(i=0;i<size;i+=16) {
-		uint8_t PD[RAID_PARITY_MAX][16] __attribute__((aligned(16)));
+	for (i = 0; i < size; i += 16) {
+		uint8_t PD[RAID_PARITY_MAX][16] __aligned(16);
 
 		/* delta */
-		for(j=0;j<N;++j) {
+		for (j = 0; j < N; ++j) {
 			asm volatile("movdqa %0,%%xmm0" : : "m" (p[j][i]));
 			asm volatile("movdqa %0,%%xmm1" : : "m" (pa[j][i]));
 			asm volatile("pxor   %xmm1,%xmm0");
@@ -1509,11 +1535,11 @@ void raid_recX_ssse3(int nr, const int* id, const int* ip, int nd, size_t size,
 		}
 
 		/* reconstruct */
-		for(j=0;j<N;++j) {
+		for (j = 0; j < N; ++j) {
 			asm volatile("pxor %xmm0,%xmm0");
 			asm volatile("pxor %xmm1,%xmm1");
 
-			for(k=0;k<N;++k) {
+			for (k = 0; k < N; ++k) {
 				uint8_t m = V[j*N+k];
 
 				asm volatile("movdqa %0,%%xmm2" : : "m" (gfmulpshufb[m][0][0]));
@@ -1534,7 +1560,7 @@ void raid_recX_ssse3(int nr, const int* id, const int* ip, int nd, size_t size,
 		}
 	}
 
-	asm volatile("sfence" : : : "memory");
+	asm_end();
 }
 #endif
 
diff --git a/raid/x86z.c b/raid/x86z.c
index bb96eb60..f883846c 100644
--- a/raid/x86z.c
+++ b/raid/x86z.c
@@ -22,7 +22,7 @@ static const struct gfzconst16 {
 	uint8_t poly[16];
 	uint8_t half[16];
 	uint8_t low7[16];
-} gfzconst16  __attribute__((aligned(64))) = {
+} gfzconst16  __aligned(64) = {
 	{ 0x1d, 0x1d, 0x1d, 0x1d, 0x1d, 0x1d, 0x1d, 0x1d, 0x1d, 0x1d, 0x1d, 0x1d, 0x1d, 0x1d, 0x1d, 0x1d },
 	{ 0x8e, 0x8e, 0x8e, 0x8e, 0x8e, 0x8e, 0x8e, 0x8e, 0x8e, 0x8e, 0x8e, 0x8e, 0x8e, 0x8e, 0x8e, 0x8e },
 	{ 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f }
@@ -31,12 +31,12 @@ static const struct gfzconst16 {
 /*
  * PARz (triple parity with powers of 2^-1) SSE2 implementation
  */
-void raid_parz_sse2(int nd, size_t size, void** vv)
+void raid_parz_sse2(int nd, size_t size, void **vv)
 {
-	uint8_t** v = (uint8_t**)vv;
-	uint8_t* p;
-	uint8_t* q;
-	uint8_t* r;
+	uint8_t **v = (uint8_t **)vv;
+	uint8_t *p;
+	uint8_t *q;
+	uint8_t *r;
 	int d, l;
 	size_t i;
 
@@ -45,15 +45,17 @@ void raid_parz_sse2(int nd, size_t size, void** vv)
 	q = v[nd+1];
 	r = v[nd+2];
 
+	asm_begin();
+
 	asm volatile("movdqa %0,%%xmm7" : : "m" (gfzconst16.poly[0]));
 	asm volatile("movdqa %0,%%xmm3" : : "m" (gfzconst16.half[0]));
 	asm volatile("movdqa %0,%%xmm6" : : "m" (gfzconst16.low7[0]));
 
-	for(i=0;i<size;i+=16) {
+	for (i = 0; i < size; i += 16) {
 		asm volatile("movdqa %0,%%xmm0" : : "m" (v[l][i]));
 		asm volatile("movdqa %xmm0,%xmm1");
 		asm volatile("movdqa %xmm0,%xmm2");
-		for(d=l-1;d>=0;--d) {
+		for (d = l-1; d >= 0; --d) {
 			asm volatile("pxor %xmm4,%xmm4");
 			asm volatile("pcmpgtb %xmm1,%xmm4");
 			asm volatile("paddb %xmm1,%xmm1");
@@ -79,7 +81,7 @@ void raid_parz_sse2(int nd, size_t size, void** vv)
 		asm volatile("movntdq %%xmm2,%0" : "=m" (r[i]));
 	}
 
-	asm volatile("sfence" : : : "memory");
+	asm_end();
 }
 #endif
 
@@ -89,12 +91,12 @@ void raid_parz_sse2(int nd, size_t size, void** vv)
  *
  * Note that it uses 16 registers, meaning that x64 is required.
  */
-void raid_parz_sse2ext(int nd, size_t size, void** vv)
+void raid_parz_sse2ext(int nd, size_t size, void **vv)
 {
-	uint8_t** v = (uint8_t**)vv;
-	uint8_t* p;
-	uint8_t* q;
-	uint8_t* r;
+	uint8_t **v = (uint8_t **)vv;
+	uint8_t *p;
+	uint8_t *q;
+	uint8_t *r;
 	int d, l;
 	size_t i;
 
@@ -103,18 +105,20 @@ void raid_parz_sse2ext(int nd, size_t size, void** vv)
 	q = v[nd+1];
 	r = v[nd+2];
 
+	asm_begin();
+
 	asm volatile("movdqa %0,%%xmm7" : : "m" (gfzconst16.poly[0]));
 	asm volatile("movdqa %0,%%xmm3" : : "m" (gfzconst16.half[0]));
 	asm volatile("movdqa %0,%%xmm11" : : "m" (gfzconst16.low7[0]));
 
-	for(i=0;i<size;i+=32) {
+	for (i = 0; i < size; i += 32) {
 		asm volatile("movdqa %0,%%xmm0" : : "m" (v[l][i]));
 		asm volatile("movdqa %0,%%xmm8" : : "m" (v[l][i+16]));
 		asm volatile("movdqa %xmm0,%xmm1");
 		asm volatile("movdqa %xmm8,%xmm9");
 		asm volatile("movdqa %xmm0,%xmm2");
 		asm volatile("movdqa %xmm8,%xmm10");
-		for(d=l-1;d>=0;--d) {
+		for (d = l-1; d >= 0; --d) {
 			asm volatile("movdqa %xmm2,%xmm6");
 			asm volatile("movdqa %xmm10,%xmm14");
 			asm volatile("pxor %xmm4,%xmm4");
@@ -159,7 +163,7 @@ void raid_parz_sse2ext(int nd, size_t size, void** vv)
 		asm volatile("movntdq %%xmm10,%0" : "=m" (r[i+16]));
 	}
 
-	asm volatile("sfence" : : : "memory");
+	asm_end();
 }
 #endif