3 files changed, 413 insertions, 0 deletions

diff --git a/databases/caterva/files/patch-contribs_c-blosc2_blosc_bitshuffle-altivec.c b/databases/caterva/files/patch-contribs_c-blosc2_blosc_bitshuffle-altivec.c
new file mode 100644
index 000000000000..a5f4cc517c65
--- /dev/null
+++ b/databases/caterva/files/patch-contribs_c-blosc2_blosc_bitshuffle-altivec.c
@@ -0,0 +1,221 @@
+--- contribs/c-blosc2/blosc/bitshuffle-altivec.c.orig	2023-03-23 20:29:41 UTC
++++ contribs/c-blosc2/blosc/bitshuffle-altivec.c
+@@ -37,15 +37,15 @@
+ #include <stdio.h>
+ #include <string.h>
+ 
+-static void helper_print(__vector uint8_t v, char* txt){
++static void helper_print(__vector unsigned char v, char* txt){
+   printf("%s %02x %02x %02x %02x %02x %02x %02x %02x %02x %02x %02x %02x %02x %02x %02x %02x\n",txt,
+   v[0], v[1], v[2], v[3], v[4], v[5], v[6], v[7], v[8], v[9], v[10], v[11], v[12], v[13], v[14], v[15]);
+ }
+ #endif
+ 
+ 
+-static inline __vector uint8_t gen_save_mask(size_t offset){
+-  __vector uint8_t mask;
++static inline __vector unsigned char gen_save_mask(size_t offset){
++  __vector unsigned char mask;
+   size_t k;
+   for (k = 0; k < 16; k++)
+     mask[k] = (k<offset)?0:0xFF;
+@@ -54,8 +54,8 @@ static inline __vector uint8_t gen_save_mask(size_t of
+ 
+ 
+ // Build and return a bit-permutation mask
+-static __vector uint8_t make_bitperm_mask(int type_size, int bit) {
+-  __vector uint8_t result;
++static __vector unsigned char make_bitperm_mask(int type_size, int bit) {
++  __vector unsigned char result;
+   if (type_size == 1) {
+     // data_type is 8 bits long
+     for (int i = 0; i < 16; i++)
+@@ -90,29 +90,29 @@ bitunshuffle1_altivec(void* _src, void* dest, const si
+   size_t nbyte_row = size / 8;
+ 
+   // working vectors
+-  __vector uint8_t xmm0[8], xmm1[8], masks[8];
++  __vector unsigned char xmm0[8], xmm1[8], masks[8];
+   // Vector masks
+-  static const __vector uint8_t lo01 = (const __vector uint8_t) {
++  static const __vector unsigned char lo01 = (const __vector unsigned char) {
+     0x00, 0x01, 0x04, 0x05, 0x08, 0x09, 0x0c, 0x0d,
+     0x10, 0x11, 0x14, 0x15, 0x18, 0x19, 0x1c, 0x1d};
+-  static const __vector uint8_t hi01 = (const __vector uint8_t) {
++  static const __vector unsigned char hi01 = (const __vector unsigned char) {
+     0x02, 0x03, 0x06, 0x07, 0x0a, 0x0b, 0x0e, 0x0f,
+     0x12, 0x13, 0x16, 0x17, 0x1a, 0x1b, 0x1e, 0x1f};
+-  static const __vector uint8_t lo02 = (const __vector uint8_t) {
++  static const __vector unsigned char lo02 = (const __vector unsigned char) {
+     0x00, 0x01, 0x08, 0x09, 0x10, 0x11, 0x18, 0x19,
+     0x02, 0x03, 0x0a, 0x0b, 0x12, 0x13, 0x1a, 0x1b};
+-  static const __vector uint8_t hi02 = (const __vector uint8_t) {
++  static const __vector unsigned char hi02 = (const __vector unsigned char) {
+     0x04, 0x05, 0x0c, 0x0d, 0x14, 0x15, 0x1c, 0x1d,
+     0x06, 0x07, 0x0e, 0x0f, 0x16, 0x17, 0x1e, 0x1f};
+-  static const __vector uint8_t epi64_low = (const __vector uint8_t) {
++  static const __vector unsigned char epi64_low = (const __vector unsigned char) {
+     0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07,
+     0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17};
+-  static const __vector uint8_t epi64_hi = (const __vector uint8_t) {
++  static const __vector unsigned char epi64_hi = (const __vector unsigned char) {
+     0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f,
+     0x18, 0x19, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f};
+ 
+   for (kk = 0; kk < 8; kk++){
+-    __vector uint8_t msk;
++    __vector unsigned char msk;
+     for (ii = 0; ii < 8; ii++){
+       msk[ii] = 127-(16*ii+2*kk);
+       msk[ii+8] = 127-(16*ii+2*kk+1);
+@@ -161,13 +161,13 @@ bitunshuffle1_altivec(void* _src, void* dest, const si
+ 
+       // At this stage each vector xmm1 contains the data from 16 adjacent bytes
+       for (int ll = 0; ll < 8; ll++){
+-        __vector uint8_t xmm = xmm1[ll];
++        __vector unsigned char xmm = xmm1[ll];
+         //helper_print(xmm, "vector transposed");
+         for (kk = 0; kk < 8; kk++) {
+-           __vector uint16_t tmp;
+-           tmp = (__vector uint16_t) vec_bperm(xmm, masks[kk]);
++           __vector unsigned short tmp;
++           tmp = (__vector unsigned short) vec_bperm(xmm, masks[kk]);
+            //printf("%d %d\n", vp, tmp[4]);
+-           //helper_print((__vector uint8_t)tmp, "tmp");
++           //helper_print((__vector unsigned char)tmp, "tmp");
+            out_s[vp++] = tmp[4];
+         }
+       }
+@@ -179,7 +179,7 @@ int64_t bshuf_trans_byte_elem_16(void* in, void* out, 
+ /* Transpose bytes within elements for 16 bit elements. */
+ int64_t bshuf_trans_byte_elem_16(void* in, void* out, const size_t size) {
+   static const uint8_t bytesoftype = 2;
+-  __vector uint8_t xmm0[2];
++  __vector unsigned char xmm0[2];
+ 
+   for (size_t i = 0; i + 15 < size; i += 16) {
+     for (int j = 0; j < bytesoftype; j++)
+@@ -199,7 +199,7 @@ int64_t bshuf_trans_byte_elem_32(void* in, void* out, 
+ /* Transpose bytes within elements for 32 bit elements. */
+ int64_t bshuf_trans_byte_elem_32(void* in, void* out, const size_t size) {
+   static const uint8_t bytesoftype = 4;
+-  __vector uint8_t xmm0[4];
++  __vector unsigned char xmm0[4];
+ 
+   for (size_t i = 0; i + 15 < size; i += 16) {
+     for (int j = 0; j < bytesoftype; j++)
+@@ -219,7 +219,7 @@ int64_t bshuf_trans_byte_elem_64(void* in, void* out, 
+ /* Transpose bytes within elements for 64 bit elements. */
+ int64_t bshuf_trans_byte_elem_64(void* in, void* out, const size_t size) {
+   static const uint8_t bytesoftype = 8;
+-  __vector uint8_t xmm0[8];
++  __vector unsigned char xmm0[8];
+ 
+   for (size_t i = 0; i + 15 < size; i += 16) {
+     for (int j = 0; j < bytesoftype; j++)
+@@ -239,7 +239,7 @@ int64_t bshuf_trans_byte_elem_128(void* in, void* out,
+ /* Transpose bytes within elements for 128 bit elements. */
+ int64_t bshuf_trans_byte_elem_128(void* in, void* out, const size_t size) {
+   static const uint8_t bytesoftype = 16;
+-  __vector uint8_t xmm0[16];
++  __vector unsigned char xmm0[16];
+ 
+   for (size_t i = 0; i + 15 < size; i += 16) {
+     for (int j = 0; j < bytesoftype; j++)
+@@ -306,7 +306,7 @@ int64_t bshuf_trans_byte_elem_altivec(void* in, void* 
+ 
+     if ((elem_size % 16) == 0) {
+       nchunk_elem = elem_size / 16;
+-      TRANS_ELEM_TYPE(in, out, size, nchunk_elem, __vector uint8_t);
++      TRANS_ELEM_TYPE(in, out, size, nchunk_elem, __vector unsigned char);
+       count = bshuf_trans_byte_elem_128(out, tmp_buf,
+                                         size * nchunk_elem);
+       bshuf_trans_elem(tmp_buf, out, 16, nchunk_elem, size);
+@@ -343,7 +343,7 @@ int64_t bshuf_trans_bit_byte_altivec(void* in, void* o
+   uint8_t* out_b = (uint8_t*)out;
+   int64_t count;
+   size_t nbyte = elem_size * size;
+-  __vector uint8_t data, masks[8];
++  __vector unsigned char data, masks[8];
+   size_t ii, kk;
+ 
+   CHECK_MULT_EIGHT(nbyte);
+@@ -356,9 +356,9 @@ int64_t bshuf_trans_bit_byte_altivec(void* in, void* o
+   for (ii = 0; ii + 15 < nbyte; ii += 16) {
+     data = vec_xl(ii, in_b);
+     for (kk = 0; kk < 8; kk++) {
+-      __vector uint16_t tmp;
++      __vector unsigned short tmp;
+       uint16_t* oui16;
+-      tmp = (__vector uint16_t) vec_bperm(data, masks[kk]);
++      tmp = (__vector unsigned short) vec_bperm(data, masks[kk]);
+       oui16 = (uint16_t*)&out_b[(ii + kk*nbyte) >> 3];
+       *oui16 = tmp[4];
+     }
+@@ -390,28 +390,28 @@ int64_t bshuf_trans_byte_bitrow_altivec(void* in, void
+  * the bytes. */
+ int64_t bshuf_trans_byte_bitrow_altivec(void* in, void* out, const size_t size,
+                                         const size_t elem_size) {
+-  static const __vector uint8_t epi8_low = (const __vector uint8_t) {
++  static const __vector unsigned char epi8_low = (const __vector unsigned char) {
+     0x00, 0x10, 0x01, 0x11, 0x02, 0x12, 0x03, 0x13,
+     0x04, 0x14, 0x05, 0x15, 0x06, 0x16, 0x07, 0x17};
+-  static const __vector uint8_t epi8_hi = (const __vector uint8_t) {
++  static const __vector unsigned char epi8_hi = (const __vector unsigned char) {
+     0x08, 0x18, 0x09, 0x19, 0x0a, 0x1a, 0x0b, 0x1b,
+     0x0c, 0x1c, 0x0d, 0x1d, 0x0e, 0x1e, 0x0f, 0x1f};
+-  static const __vector uint8_t epi16_low = (const __vector uint8_t) {
++  static const __vector unsigned char epi16_low = (const __vector unsigned char) {
+     0x00, 0x01, 0x10, 0x11, 0x02, 0x03, 0x12, 0x13,
+     0x04, 0x05, 0x14, 0x15, 0x06, 0x07, 0x16, 0x17};
+-  static const __vector uint8_t epi16_hi = (const __vector uint8_t) {
++  static const __vector unsigned char epi16_hi = (const __vector unsigned char) {
+     0x08, 0x09, 0x18, 0x19, 0x0a, 0x0b, 0x1a, 0x1b,
+     0x0c, 0x0d, 0x1c, 0x1d, 0x0e, 0x0f, 0x1e, 0x1f};
+-  static const __vector uint8_t epi32_low = (const __vector uint8_t) {
++  static const __vector unsigned char epi32_low = (const __vector unsigned char) {
+     0x00, 0x01, 0x02, 0x03, 0x10, 0x11, 0x12, 0x13,
+     0x04, 0x05, 0x06, 0x07, 0x14, 0x15, 0x16, 0x17};
+-  static const __vector uint8_t epi32_hi = (const __vector uint8_t) {
++  static const __vector unsigned char epi32_hi = (const __vector unsigned char) {
+     0x08, 0x09, 0x0a, 0x0b, 0x18, 0x19, 0x1a, 0x1b,
+     0x0c, 0x0d, 0x0e, 0x0f, 0x1c, 0x1d, 0x1e, 0x1f};
+-  static const __vector uint8_t epi64_low = (const __vector uint8_t) {
++  static const __vector unsigned char epi64_low = (const __vector unsigned char) {
+     0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07,
+     0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17};
+-  static const __vector uint8_t epi64_hi = (const __vector uint8_t) {
++  static const __vector unsigned char epi64_hi = (const __vector unsigned char) {
+     0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f,
+     0x18, 0x19, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f};
+ 
+@@ -419,7 +419,7 @@ int64_t bshuf_trans_byte_bitrow_altivec(void* in, void
+   uint8_t* out_b = (uint8_t*)out;
+   size_t nrows = 8 * elem_size;
+   size_t nbyte_row = size / 8;
+-  __vector uint8_t xmm0[16], xmm1[16];
++  __vector unsigned char xmm0[16], xmm1[16];
+ 
+   CHECK_MULT_EIGHT(size);
+ 
+@@ -546,7 +546,7 @@ int64_t bshuf_shuffle_bit_eightelem_altivec(void* in, 
+   const uint8_t* in_b = (const uint8_t*)in;
+   uint8_t* out_b = (uint8_t*)out;
+   size_t nbyte = elem_size * size;
+-  __vector uint8_t masks[8], data;
++  __vector unsigned char masks[8], data;
+ 
+   CHECK_MULT_EIGHT(size);
+ 
+@@ -563,9 +563,9 @@ int64_t bshuf_shuffle_bit_eightelem_altivec(void* in, 
+       for (size_t jj = 0; jj + 15 < 8 * elem_size; jj += 16) {
+         data = vec_xl(ii + jj, in_b);
+         for (size_t kk = 0; kk < 8; kk++) {
+-          __vector uint16_t tmp;
++          __vector unsigned short tmp;
+           uint16_t* oui16;
+-          tmp = (__vector uint16_t) vec_bperm(data, masks[kk]);
++          tmp = (__vector unsigned short) vec_bperm(data, masks[kk]);
+           oui16 = (uint16_t*)&out_b[ii + (jj>>3) + kk * elem_size];
+           *oui16 = tmp[4];
+         }
diff --git a/databases/caterva/files/patch-contribs_c-blosc2_blosc_shuffle-altivec.c b/databases/caterva/files/patch-contribs_c-blosc2_blosc_shuffle-altivec.c
new file mode 100644
index 000000000000..69a9ed1dbe69
--- /dev/null
+++ b/databases/caterva/files/patch-contribs_c-blosc2_blosc_shuffle-altivec.c
@@ -0,0 +1,134 @@
+--- contribs/c-blosc2/blosc/shuffle-altivec.c.orig	2023-03-23 20:25:30 UTC
++++ contribs/c-blosc2/blosc/shuffle-altivec.c
+@@ -25,7 +25,7 @@ shuffle2_altivec(uint8_t* const dest, const uint8_t* c
+                  const int32_t vectorizable_elements, const int32_t total_elements){
+   static const int32_t bytesoftype = 2;
+   uint32_t i, j;
+-  __vector uint8_t xmm0[2];
++  __vector unsigned char xmm0[2];
+ 
+   for (j = 0; j < vectorizable_elements; j += 16){
+     /* Fetch 16 elements (32 bytes) */
+@@ -47,7 +47,7 @@ shuffle4_altivec(uint8_t* const dest, const uint8_t* c
+                  const int32_t vectorizable_elements, const int32_t total_elements){
+   static const int32_t bytesoftype = 4;
+   int32_t i, j;
+-  __vector uint8_t xmm0[4];
++  __vector unsigned char xmm0[4];
+ 
+   for (j = 0; j < vectorizable_elements; j += 16)
+   {
+@@ -73,7 +73,7 @@ shuffle8_altivec(uint8_t* const dest, const uint8_t* c
+                  const int32_t vectorizable_elements, const int32_t total_elements) {
+   static const uint8_t bytesoftype = 8;
+   int32_t i, j;
+-  __vector uint8_t xmm0[8];
++  __vector unsigned char xmm0[8];
+ 
+   for (j = 0; j < vectorizable_elements; j += 16)
+   {
+@@ -96,7 +96,7 @@ shuffle16_altivec(uint8_t* const dest, const uint8_t* 
+                   const int32_t vectorizable_elements, const int32_t total_elements) {
+   static const int32_t bytesoftype = 16;
+   int32_t i, j;
+-  __vector uint8_t xmm0[16];
++  __vector unsigned char xmm0[16];
+ 
+   for (j = 0; j < vectorizable_elements; j += 16)
+   {
+@@ -121,7 +121,7 @@ shuffle16_tiled_altivec(uint8_t* const dest, const uin
+                         const int32_t bytesoftype) {
+   int32_t j, k;
+   const int32_t vecs_per_el_rem = bytesoftype & 0xF;
+-  __vector uint8_t xmm[16];
++  __vector unsigned char xmm[16];
+ 
+   for (j = 0; j < vectorizable_elements; j += 16) {
+     /* Advance the offset into the type by the vector size (in bytes), unless this is
+@@ -152,7 +152,7 @@ unshuffle2_altivec(uint8_t* const dest, const uint8_t*
+                    const int32_t vectorizable_elements, const int32_t total_elements) {
+   static const int32_t bytesoftype = 2;
+   uint32_t i, j;
+-  __vector uint8_t xmm0[2], xmm1[2];
++  __vector unsigned char xmm0[2], xmm1[2];
+ 
+   for (j = 0; j < vectorizable_elements; j += 16) {
+     /* Load 16 elements (32 bytes) into 2 vectors registers. */
+@@ -176,7 +176,7 @@ unshuffle4_altivec(uint8_t* const dest, const uint8_t*
+                    const int32_t vectorizable_elements, const int32_t total_elements) {
+   static const int32_t bytesoftype = 4;
+   uint32_t i, j;
+-  __vector uint8_t xmm0[4], xmm1[4];
++  __vector unsigned char xmm0[4], xmm1[4];
+ 
+   for (j = 0; j < vectorizable_elements; j += 16) {
+     /* Load 16 elements (64 bytes) into 4 vectors registers. */
+@@ -191,11 +191,11 @@ unshuffle4_altivec(uint8_t* const dest, const uint8_t*
+     /* Shuffle 2-byte words */
+     for (i = 0; i < 2; i++) {
+       /* Compute the low 32 bytes */
+-      xmm0[i] = (__vector uint8_t) vec_vmrghh((__vector uint16_t)xmm1[i * 2],
+-                                              (__vector uint16_t) xmm1[i * 2 + 1]);
++      xmm0[i] = (__vector unsigned char) vec_vmrghh((__vector unsigned short)xmm1[i * 2],
++                                              (__vector unsigned short) xmm1[i * 2 + 1]);
+       /* Compute the hi 32 bytes */
+-      xmm0[i+2] = (__vector uint8_t) vec_vmrglh((__vector uint16_t)xmm1[i * 2],
+-                                                (__vector uint16_t)xmm1[i * 2 + 1]);
++      xmm0[i+2] = (__vector unsigned char) vec_vmrglh((__vector unsigned short)xmm1[i * 2],
++                                                (__vector unsigned short)xmm1[i * 2 + 1]);
+     }
+     /* Store the result vectors in proper order */
+     vec_xst(xmm0[0], bytesoftype * j, dest);
+@@ -211,7 +211,7 @@ unshuffle8_altivec(uint8_t* const dest, const uint8_t*
+                    const int32_t vectorizable_elements, const int32_t total_elements) {
+   static const uint8_t bytesoftype = 8;
+   uint32_t i, j;
+-  __vector uint8_t xmm0[8], xmm1[8];
++  __vector unsigned char xmm0[8], xmm1[8];
+ 
+   // Initialize permutations for writing
+   for (j = 0; j < vectorizable_elements; j += 16) {
+@@ -225,17 +225,17 @@ unshuffle8_altivec(uint8_t* const dest, const uint8_t*
+     }
+     /* Shuffle 2-byte words */
+     for (i = 0; i < 4; i++) {
+-      xmm0[i] = (__vector uint8_t)vec_vmrghh((__vector uint16_t)xmm1[i * 2],
+-                                             (__vector uint16_t)xmm1[i * 2 + 1]);
+-      xmm0[4 + i] = (__vector uint8_t)vec_vmrglh((__vector uint16_t)xmm1[i * 2],
+-                                                 (__vector uint16_t)xmm1[i * 2 + 1]);
++      xmm0[i] = (__vector unsigned char)vec_vmrghh((__vector unsigned short)xmm1[i * 2],
++                                             (__vector unsigned short)xmm1[i * 2 + 1]);
++      xmm0[4 + i] = (__vector unsigned char)vec_vmrglh((__vector unsigned short)xmm1[i * 2],
++                                                 (__vector unsigned short)xmm1[i * 2 + 1]);
+     }
+     /* Shuffle 4-byte dwords */
+     for (i = 0; i < 4; i++) {
+-      xmm1[i] = (__vector uint8_t)vec_vmrghw((__vector uint32_t)xmm0[i * 2],
+-                                             (__vector uint32_t)xmm0[i * 2 + 1]);
+-      xmm1[4 + i] = (__vector uint8_t)vec_vmrglw((__vector uint32_t)xmm0[i * 2],
+-                                                 (__vector uint32_t)xmm0[i * 2 + 1]);
++      xmm1[i] = (__vector unsigned char)vec_vmrghw((__vector unsigned int)xmm0[i * 2],
++                                             (__vector unsigned int)xmm0[i * 2 + 1]);
++      xmm1[4 + i] = (__vector unsigned char)vec_vmrglw((__vector unsigned int)xmm0[i * 2],
++                                                 (__vector unsigned int)xmm0[i * 2 + 1]);
+     }
+     /* Store the result vectors in proper order */
+     vec_xst(xmm1[0], bytesoftype * j, dest);
+@@ -256,7 +256,7 @@ unshuffle16_altivec(uint8_t* const dest, const uint8_t
+                     const int32_t vectorizable_elements, const int32_t total_elements) {
+   static const int32_t bytesoftype = 16;
+   uint32_t i, j;
+-  __vector uint8_t xmm0[16];
++  __vector unsigned char xmm0[16];
+ 
+   for (j = 0; j < vectorizable_elements; j += 16) {
+     /* Load 16 elements (64 bytes) into 4 vectors registers. */
+@@ -280,7 +280,7 @@ unshuffle16_tiled_altivec(uint8_t* const dest, const u
+                           const int32_t bytesoftype) {
+   int32_t i, j, offset_into_type;
+   const int32_t vecs_per_el_rem = bytesoftype &  0xF;
+-  __vector uint8_t xmm[16];
++  __vector unsigned char xmm[16];
+ 
+ 
+   /* Advance the offset into the type by the vector size (in bytes), unless this is
diff --git a/databases/caterva/files/patch-contribs_c-blosc2_blosc_transpose-altivec.h b/databases/caterva/files/patch-contribs_c-blosc2_blosc_transpose-altivec.h
new file mode 100644
index 000000000000..060dceb6dde1
--- /dev/null
+++ b/databases/caterva/files/patch-contribs_c-blosc2_blosc_transpose-altivec.h
@@ -0,0 +1,58 @@
+--- contribs/c-blosc2/blosc/transpose-altivec.h.orig	2023-03-23 20:13:07 UTC
++++ contribs/c-blosc2/blosc/transpose-altivec.h
+@@ -15,18 +15,18 @@ extern "C" {
+ extern "C" {
+ #endif
+ 
+-static const __vector uint8_t even = (const __vector uint8_t) {
++static const __vector unsigned char even = (const __vector unsigned char) {
+   0x00, 0x02, 0x04, 0x06, 0x08, 0x0a, 0x0c, 0x0e,
+   0x10, 0x12, 0x14, 0x16, 0x18, 0x1a, 0x1c, 0x1e};
+ 
+-static const __vector uint8_t odd = (const __vector uint8_t) {
++static const __vector unsigned char odd = (const __vector unsigned char) {
+   0x01, 0x03, 0x05, 0x07, 0x09, 0x0b, 0x0d, 0x0f,
+   0x11, 0x13, 0x15, 0x17, 0x19, 0x1b, 0x1d, 0x1f};
+ 
+ 
+ /* Transpose inplace 2 vectors of 16 bytes in src into dst. */
+-static void transpose2x16(__vector uint8_t *xmm0) {
+-  __vector uint8_t xmm1[2];
++static void transpose2x16(__vector unsigned char *xmm0) {
++  __vector unsigned char xmm1[2];
+   xmm1[0] = vec_perm(xmm0[0], xmm0[1], even);
+   xmm1[1] = vec_perm(xmm0[0], xmm0[1], odd);
+ 
+@@ -38,8 +38,8 @@ static void transpose2x16(__vector uint8_t *xmm0) {
+ 
+ /* Transpose inplace 4 vectors of 16 bytes in src into dst.
+  * Total cost: 8 calls to vec_perm. */
+-static void transpose4x16(__vector uint8_t *xmm0) {
+-  __vector uint8_t xmm1[4];
++static void transpose4x16(__vector unsigned char *xmm0) {
++  __vector unsigned char xmm1[4];
+ 
+   /* Transpose vectors 0-1*/
+   xmm1[0] = vec_perm(xmm0[0], xmm0[1], even);
+@@ -56,8 +56,8 @@ static void transpose4x16(__vector uint8_t *xmm0) {
+ 
+ /* Transpose inplace 8 vectors of 16 bytes in src into dst.
+  * Total cost: 24 calls to vec_perm. */
+-static void transpose8x16(__vector uint8_t *xmm0) {
+-  __vector uint8_t xmm1[8];
++static void transpose8x16(__vector unsigned char *xmm0) {
++  __vector unsigned char xmm1[8];
+ 
+   /* Transpose vectors 0-1*/
+   for (int i = 0; i < 8; i += 2){
+@@ -85,8 +85,8 @@ static void transpose8x16(__vector uint8_t *xmm0) {
+ 
+ /* Transpose inplace 16 vectors of 16 bytes in src into dst.
+  * Total cost: 64 calls to vec_perm. */
+-static void transpose16x16(__vector uint8_t * xmm0){
+-  __vector uint8_t xmm1[16];
++static void transpose16x16(__vector unsigned char * xmm0){
++  __vector unsigned char xmm1[16];
+   /* Transpose vectors 0-1*/
+   for (int i = 0; i < 16; i += 2){
+     xmm1[i] = vec_perm(xmm0[i], xmm0[i+1], even);