From c9851639f4830e516e48797832ec30fe82a43e28 Mon Sep 17 00:00:00 2001 From: roentgen Date: Tue, 13 Nov 2018 23:24:39 +0900 Subject: [PATCH 1/6] make them thread-safe --- Examples/cauchy_01.c | 10 ++- Examples/cauchy_02.c | 16 ++-- Examples/cauchy_03.c | 18 ++-- Examples/cauchy_04.c | 16 ++-- Examples/decoder.c | 17 ++-- Examples/encoder.c | 25 +++--- Examples/jerasure_01.c | 6 +- Examples/jerasure_02.c | 7 +- Examples/jerasure_03.c | 10 ++- Examples/jerasure_04.c | 7 +- Examples/jerasure_05.c | 13 +-- Examples/jerasure_06.c | 11 +-- Examples/jerasure_07.c | 13 +-- Examples/jerasure_08.c | 13 +-- Examples/liberation_01.c | 5 +- Examples/reed_sol_01.c | 7 +- Examples/reed_sol_02.c | 7 +- Examples/reed_sol_03.c | 7 +- Examples/reed_sol_04.c | 7 +- Examples/reed_sol_test_gf.c | 9 +- Examples/reed_sol_time_gf.c | 9 +- include/cauchy.h | 10 +-- include/galois.h | 38 +++++--- include/jerasure.h | 32 +++---- include/reed_sol.h | 16 ++-- src/cauchy.c | 36 ++++---- src/galois.c | 170 ++++++++++++++++++++---------------- src/jerasure.c | 98 +++++++++++---------- src/reed_sol.c | 55 ++++++------ 29 files changed, 376 insertions(+), 312 deletions(-) diff --git a/Examples/cauchy_01.c b/Examples/cauchy_01.c index 65419a0..1eddd0d 100644 --- a/Examples/cauchy_01.c +++ b/Examples/cauchy_01.c @@ -48,6 +48,7 @@ #include #include #include +#include "galois.h" /* gf2_t */ #include "cauchy.h" #include "jerasure.h" #include "reed_sol.h" @@ -74,6 +75,7 @@ int main(int argc, char **argv) int n; int i, no, w; int *bitmatrix; + static gf2_t g; if (argc != 3) usage(NULL); if (sscanf(argv[1], "0x%x", &n) == 0) { @@ -86,7 +88,7 @@ int main(int argc, char **argv) if (n >= (1 << w)) usage("Bad n/w combination (n not between 0 and 2^w-1)\n"); } - bitmatrix = jerasure_matrix_to_bitmatrix(1, 1, w, &n); + bitmatrix = jerasure_matrix_to_bitmatrix(&g, 1, 1, w, &n); printf("cauchy_01 %u %d\n", w, n); printf("

cauchy_01 %u %d

\n", w, n); printf("
\n");
@@ -100,13 +102,13 @@ int main(int argc, char **argv)
 
   no = 0;
   for (i = 0; i < w*w; i++) no += bitmatrix[i];
-  if (no != cauchy_n_ones(n, w)) { 
+  if (no != cauchy_n_ones(&g, n, w)) { 
     fprintf(stderr, "Jerasure error: # ones in the bitmatrix (%d) doesn't match cauchy_n_ones() (%d).\n",
-       no, cauchy_n_ones(n, w));
+            no, cauchy_n_ones(&g, n, w));
     exit(1);
   }
 
-  printf("# Ones: %d\n", cauchy_n_ones(n, w));
+  printf("# Ones: %d\n", cauchy_n_ones(&g, n, w));
 
   return 0;
 }
diff --git a/Examples/cauchy_02.c b/Examples/cauchy_02.c
index 45c78ab..47e52cc 100644
--- a/Examples/cauchy_02.c
+++ b/Examples/cauchy_02.c
@@ -50,6 +50,7 @@
 #include 
 #include 
 #include "jerasure.h"
+#include "galois.h"
 #include "cauchy.h"
 
 #define talloc(type, num) (type *) malloc(sizeof(type)*(num))
@@ -120,6 +121,7 @@ int main(int argc, char **argv)
   int *erasures, *erased;
   double mstats[3], sstats[3];
   uint32_t seed;
+  static gf2_t g;
   
   if (argc != 5) usage(NULL);
   if (sscanf(argv[1], "%d", &k) == 0 || k <= 0) usage("Bad k");
@@ -128,7 +130,7 @@ int main(int argc, char **argv)
   if (sscanf(argv[4], "%d", &seed) == 0) usage("Bad seed");
   if (w < 30 && (k+m) > (1 << w)) usage("k + m is too big");
 
-  matrix = cauchy_original_coding_matrix(k, m, w);
+  matrix = cauchy_original_coding_matrix(&g, k, m, w);
   if (matrix == NULL) {
     usage("couldn't make coding matrix");
   }
@@ -159,11 +161,11 @@ int main(int argc, char **argv)
   jerasure_print_matrix(matrix, m, k, w);
   printf("
\n"); - bitmatrix = jerasure_matrix_to_bitmatrix(k, m, w, matrix); + bitmatrix = jerasure_matrix_to_bitmatrix(&g, k, m, w, matrix); no = 0; for (i = 0; i < k*m; i++) { - no += cauchy_n_ones(matrix[i], w); + no += cauchy_n_ones(&g, matrix[i], w); } printf("The bitmatrix, which has %d one%s:

\n", no, (no == 1) ? "" : "s");
@@ -191,11 +193,11 @@ int main(int argc, char **argv)
     ccopy[i] = talloc(char, sizeof(long)*w);
   }
 
-  jerasure_bitmatrix_encode(k, m, w, bitmatrix, data, coding, w*sizeof(long), sizeof(long));
+  jerasure_bitmatrix_encode(&g, k, m, w, bitmatrix, data, coding, w*sizeof(long), sizeof(long));
   jerasure_get_stats(mstats);
 
   schedule = jerasure_smart_bitmatrix_to_schedule(k, m, w, bitmatrix);
-  jerasure_schedule_encode(k, m, w, schedule, data, ccopy, w*sizeof(long), sizeof(long));
+  jerasure_schedule_encode(&g, k, m, w, schedule, data, ccopy, w*sizeof(long), sizeof(long));
   jerasure_get_stats(sstats);
 
   printf("
Encoding with jerasure_bitmatrix_encode() - Bytes XOR'd: %.0lf.
\n", mstats[0]);
@@ -234,7 +236,7 @@ int main(int argc, char **argv)
   print_array(coding, m, sizeof(long)*w, sizeof(long), "C");
   printf("
\n");
 
-  jerasure_bitmatrix_decode(k, m, w, bitmatrix, 0, erasures, data, coding, w*sizeof(long), sizeof(long));
+  jerasure_bitmatrix_decode(&g, k, m, w, bitmatrix, 0, erasures, data, coding, w*sizeof(long), sizeof(long));
   jerasure_get_stats(mstats);
 
   printf("
Decoded with jerasure_bitmatrix_decode - Bytes XOR'd: %.0lf.
\n", mstats[0]);
@@ -250,7 +252,7 @@ int main(int argc, char **argv)
     bzero((erasures[i] < k) ? data[erasures[i]] : coding[erasures[i]-k], sizeof(long)*w);
   }
 
-  jerasure_schedule_decode_lazy(k, m, w, bitmatrix, erasures, data, coding, w*sizeof(long), sizeof(long), 1);
+  jerasure_schedule_decode_lazy(&g, k, m, w, bitmatrix, erasures, data, coding, w*sizeof(long), sizeof(long), 1);
   jerasure_get_stats(sstats);
 
   printf("jerasure_schedule_decode_lazy - Bytes XOR'd: %.0lf.
\n", sstats[0]);
diff --git a/Examples/cauchy_03.c b/Examples/cauchy_03.c
index e246e77..3868423 100644
--- a/Examples/cauchy_03.c
+++ b/Examples/cauchy_03.c
@@ -50,6 +50,7 @@
 #include 
 #include 
 #include "jerasure.h"
+#include "galois.h"
 #include "cauchy.h"
 
 #define talloc(type, num) (type *) malloc(sizeof(type)*(num))
@@ -125,6 +126,7 @@ int main(int argc, char **argv)
   double mstats[3], sstats[3];
   uint32_t seed;
   int *X, *Y;
+  static gf2_t g;
   
   if (argc != 5) usage(NULL);
   if (sscanf(argv[1], "%d", &k) == 0 || k <= 0) usage("Bad k");
@@ -138,7 +140,7 @@ int main(int argc, char **argv)
   for (i = 0; i < m; i++) X[i] = i;
   for (i = 0; i < k; i++) Y[i] = m+i;
 
-  matrix = cauchy_xy_coding_matrix(k, m, w, X, Y);
+  matrix = cauchy_xy_coding_matrix(&g, k, m, w, X, Y);
   if (matrix == NULL) {
     usage("couldn't make coding matrix");
   }
@@ -167,7 +169,7 @@ int main(int argc, char **argv)
   jerasure_print_matrix(matrix, m, k, w);
   printf("
\n"); - cauchy_improve_coding_matrix(k, m, w, matrix); + cauchy_improve_coding_matrix(&g, k, m, w, matrix); printf("
\n"); printf("Here is the matrix improved with cauchy_improve_coding_matrix().\n"); @@ -176,11 +178,11 @@ int main(int argc, char **argv) jerasure_print_matrix(matrix, m, k, w); printf("\n"); - bitmatrix = jerasure_matrix_to_bitmatrix(k, m, w, matrix); + bitmatrix = jerasure_matrix_to_bitmatrix(&g, k, m, w, matrix); no = 0; for (i = 0; i < k*m; i++) { - no += cauchy_n_ones(matrix[i], w); + no += cauchy_n_ones(&g, matrix[i], w); } printf("The bitmatrix, which has %d one%s:

\n", no, (no == 1) ? "" : "s");
@@ -209,11 +211,11 @@ int main(int argc, char **argv)
     ccopy[i] = talloc(char, sizeof(long)*w);
   }
 
-  jerasure_bitmatrix_encode(k, m, w, bitmatrix, data, coding, w*sizeof(long), sizeof(long));
+  jerasure_bitmatrix_encode(&g, k, m, w, bitmatrix, data, coding, w*sizeof(long), sizeof(long));
   jerasure_get_stats(mstats);
 
   schedule = jerasure_smart_bitmatrix_to_schedule(k, m, w, bitmatrix);
-  jerasure_schedule_encode(k, m, w, schedule, data, ccopy, w*sizeof(long), sizeof(long));
+  jerasure_schedule_encode(&g, k, m, w, schedule, data, ccopy, w*sizeof(long), sizeof(long));
   jerasure_get_stats(sstats);
 
   printf("
Encoding with jerasure_bitmatrix_encode() - Bytes XOR'd: %.0lf.
\n", mstats[0]);
@@ -252,7 +254,7 @@ int main(int argc, char **argv)
   print_array(coding, m, sizeof(long)*w, sizeof(long), "C");
   printf("
\n");
 
-  jerasure_bitmatrix_decode(k, m, w, bitmatrix, 0, erasures, data, coding, w*sizeof(long), sizeof(long));
+  jerasure_bitmatrix_decode(&g, k, m, w, bitmatrix, 0, erasures, data, coding, w*sizeof(long), sizeof(long));
   jerasure_get_stats(mstats);
 
   printf("
Decoded with jerasure_bitmatrix_decode - Bytes XOR'd: %.0lf.
\n", mstats[0]);
@@ -268,7 +270,7 @@ int main(int argc, char **argv)
     bzero((erasures[i] < k) ? data[erasures[i]] : coding[erasures[i]-k], sizeof(long)*w);
   }
 
-  jerasure_schedule_decode_lazy(k, m, w, bitmatrix, erasures, data, coding, w*sizeof(long), sizeof(long), 1);
+  jerasure_schedule_decode_lazy(&g, k, m, w, bitmatrix, erasures, data, coding, w*sizeof(long), sizeof(long), 1);
   jerasure_get_stats(sstats);
 
   printf("jerasure_schedule_decode_lazy - Bytes XOR'd: %.0lf.
\n", sstats[0]);
diff --git a/Examples/cauchy_04.c b/Examples/cauchy_04.c
index f61c614..2ff9490 100644
--- a/Examples/cauchy_04.c
+++ b/Examples/cauchy_04.c
@@ -50,6 +50,7 @@
 #include 
 #include 
 #include "jerasure.h"
+#include "galois.h"
 #include "cauchy.h"
 
 #define talloc(type, num) (type *) malloc(sizeof(type)*(num))
@@ -121,6 +122,7 @@ int main(int argc, char **argv)
   int *erasures, *erased;
   double mstats[3], sstats[3];
   uint32_t seed;
+  static gf2_t g;
   
   if (argc != 5) usage(NULL);
   if (sscanf(argv[1], "%d", &k) == 0 || k <= 0) usage("Bad k");
@@ -129,7 +131,7 @@ int main(int argc, char **argv)
   if (sscanf(argv[4], "%d", &seed) == 0) usage("Bad seed");
   if (w < 30 && (k+m) > (1 << w)) usage("k + m is too big");
 
-  matrix = cauchy_good_general_coding_matrix(k, m, w);
+  matrix = cauchy_good_general_coding_matrix(&g, k, m, w);
   if (matrix == NULL) {
     usage("couldn't make coding matrix");
   }
@@ -158,11 +160,11 @@ int main(int argc, char **argv)
   jerasure_print_matrix(matrix, m, k, w);
   printf("
\n"); - bitmatrix = jerasure_matrix_to_bitmatrix(k, m, w, matrix); + bitmatrix = jerasure_matrix_to_bitmatrix(&g, k, m, w, matrix); no = 0; for (i = 0; i < k*m; i++) { - no += cauchy_n_ones(matrix[i], w); + no += cauchy_n_ones(&g, matrix[i], w); } printf("The bitmatrix, which has %d one%s:

\n", no, (no == 1) ? "" : "s");
@@ -191,11 +193,11 @@ int main(int argc, char **argv)
     ccopy[i] = talloc(char, sizeof(long)*w);
   }
 
-  jerasure_bitmatrix_encode(k, m, w, bitmatrix, data, coding, w*sizeof(long), sizeof(long));
+  jerasure_bitmatrix_encode(&g, k, m, w, bitmatrix, data, coding, w*sizeof(long), sizeof(long));
   jerasure_get_stats(mstats);
 
   schedule = jerasure_smart_bitmatrix_to_schedule(k, m, w, bitmatrix);
-  jerasure_schedule_encode(k, m, w, schedule, data, ccopy, w*sizeof(long), sizeof(long));
+  jerasure_schedule_encode(&g, k, m, w, schedule, data, ccopy, w*sizeof(long), sizeof(long));
   jerasure_get_stats(sstats);
 
   printf("
Encoding with jerasure_bitmatrix_encode() - Bytes XOR'd: %.0lf.
\n", mstats[0]);
@@ -234,7 +236,7 @@ int main(int argc, char **argv)
   print_array(coding, m, sizeof(long)*w, sizeof(long), "C");
   printf("
\n");
 
-  jerasure_bitmatrix_decode(k, m, w, bitmatrix, 0, erasures, data, coding, w*sizeof(long), sizeof(long));
+  jerasure_bitmatrix_decode(&g, k, m, w, bitmatrix, 0, erasures, data, coding, w*sizeof(long), sizeof(long));
   jerasure_get_stats(mstats);
 
   printf("
Decoded with jerasure_bitmatrix_decode - Bytes XOR'd: %.0lf.
\n", mstats[0]);
@@ -250,7 +252,7 @@ int main(int argc, char **argv)
     bzero((erasures[i] < k) ? data[erasures[i]] : coding[erasures[i]-k], sizeof(long)*w);
   }
 
-  jerasure_schedule_decode_lazy(k, m, w, bitmatrix, erasures, data, coding, w*sizeof(long), sizeof(long), 1);
+  jerasure_schedule_decode_lazy(&g, k, m, w, bitmatrix, erasures, data, coding, w*sizeof(long), sizeof(long), 1);
   jerasure_get_stats(sstats);
 
   printf("jerasure_schedule_decode_lazy - Bytes XOR'd: %.0lf.
\n", sstats[0]);
diff --git a/Examples/decoder.c b/Examples/decoder.c
index 229dccc..4b3da39 100644
--- a/Examples/decoder.c
+++ b/Examples/decoder.c
@@ -123,6 +123,7 @@ int main (int argc, char **argv) {
 	double tsec;
 	double totalsec;
 
+	static gf2_t g;
 	
 	signal(SIGQUIT, ctrl_bs_handler);
 
@@ -225,18 +226,18 @@ int main (int argc, char **argv) {
 		case No_Coding:
 			break;
 		case Reed_Sol_Van:
-			matrix = reed_sol_vandermonde_coding_matrix(k, m, w);
+			matrix = reed_sol_vandermonde_coding_matrix(&g, k, m, w);
 			break;
 		case Reed_Sol_R6_Op:
-			matrix = reed_sol_r6_coding_matrix(k, w);
+			matrix = reed_sol_r6_coding_matrix(&g, k, w);
 			break;
 		case Cauchy_Orig:
-			matrix = cauchy_original_coding_matrix(k, m, w);
-			bitmatrix = jerasure_matrix_to_bitmatrix(k, m, w, matrix);
+			matrix = cauchy_original_coding_matrix(&g, k, m, w);
+			bitmatrix = jerasure_matrix_to_bitmatrix(&g, k, m, w, matrix);
 			break;
 		case Cauchy_Good:
-			matrix = cauchy_good_general_coding_matrix(k, m, w);
-			bitmatrix = jerasure_matrix_to_bitmatrix(k, m, w, matrix);
+			matrix = cauchy_good_general_coding_matrix(&g, k, m, w);
+			bitmatrix = jerasure_matrix_to_bitmatrix(&g, k, m, w, matrix);
 			break;
 		case Liberation:
 			bitmatrix = liberation_coding_bitmatrix(k, w);
@@ -319,10 +320,10 @@ int main (int argc, char **argv) {
 	
 		/* Choose proper decoding method */
 		if (tech == Reed_Sol_Van || tech == Reed_Sol_R6_Op) {
-			i = jerasure_matrix_decode(k, m, w, matrix, 1, erasures, data, coding, blocksize);
+			i = jerasure_matrix_decode(&g, k, m, w, matrix, 1, erasures, data, coding, blocksize);
 		}
 		else if (tech == Cauchy_Orig || tech == Cauchy_Good || tech == Liberation || tech == Blaum_Roth || tech == Liber8tion) {
-			i = jerasure_schedule_decode_lazy(k, m, w, bitmatrix, erasures, data, coding, blocksize, packetsize, 1);
+			i = jerasure_schedule_decode_lazy(&g, k, m, w, bitmatrix, erasures, data, coding, blocksize, packetsize, 1);
 		}
 		else {
 			fprintf(stderr, "Not a valid coding technique.\n");
diff --git a/Examples/encoder.c b/Examples/encoder.c
index 367686e..f0c9d48 100644
--- a/Examples/encoder.c
+++ b/Examples/encoder.c
@@ -134,6 +134,7 @@ int main (int argc, char **argv) {
 	/* Find buffersize */
 	int up, down;
 
+	static gf2_t g;
 
 	signal(SIGQUIT, ctrl_bs_handler);
 
@@ -438,18 +439,18 @@ int main (int argc, char **argv) {
 		case No_Coding:
 			break;
 		case Reed_Sol_Van:
-			matrix = reed_sol_vandermonde_coding_matrix(k, m, w);
+			matrix = reed_sol_vandermonde_coding_matrix(&g, k, m, w);
 			break;
 		case Reed_Sol_R6_Op:
 			break;
 		case Cauchy_Orig:
-			matrix = cauchy_original_coding_matrix(k, m, w);
-			bitmatrix = jerasure_matrix_to_bitmatrix(k, m, w, matrix);
+			matrix = cauchy_original_coding_matrix(&g, k, m, w);
+			bitmatrix = jerasure_matrix_to_bitmatrix(&g, k, m, w, matrix);
 			schedule = jerasure_smart_bitmatrix_to_schedule(k, m, w, bitmatrix);
 			break;
 		case Cauchy_Good:
-			matrix = cauchy_good_general_coding_matrix(k, m, w);
-			bitmatrix = jerasure_matrix_to_bitmatrix(k, m, w, matrix);
+			matrix = cauchy_good_general_coding_matrix(&g, k, m, w);
+			bitmatrix = jerasure_matrix_to_bitmatrix(&g, k, m, w, matrix);
 			schedule = jerasure_smart_bitmatrix_to_schedule(k, m, w, bitmatrix);
 			break;	
 		case Liberation:
@@ -507,25 +508,25 @@ int main (int argc, char **argv) {
 			case No_Coding:
 				break;
 			case Reed_Sol_Van:
-				jerasure_matrix_encode(k, m, w, matrix, data, coding, blocksize);
+                jerasure_matrix_encode(&g, k, m, w, matrix, data, coding, blocksize);
 				break;
 			case Reed_Sol_R6_Op:
-				reed_sol_r6_encode(k, w, data, coding, blocksize);
+                reed_sol_r6_encode(&g, k, w, data, coding, blocksize);
 				break;
 			case Cauchy_Orig:
-				jerasure_schedule_encode(k, m, w, schedule, data, coding, blocksize, packetsize);
+				jerasure_schedule_encode(&g, k, m, w, schedule, data, coding, blocksize, packetsize);
 				break;
 			case Cauchy_Good:
-				jerasure_schedule_encode(k, m, w, schedule, data, coding, blocksize, packetsize);
+				jerasure_schedule_encode(&g, k, m, w, schedule, data, coding, blocksize, packetsize);
 				break;
 			case Liberation:
-				jerasure_schedule_encode(k, m, w, schedule, data, coding, blocksize, packetsize);
+				jerasure_schedule_encode(&g, k, m, w, schedule, data, coding, blocksize, packetsize);
 				break;
 			case Blaum_Roth:
-				jerasure_schedule_encode(k, m, w, schedule, data, coding, blocksize, packetsize);
+				jerasure_schedule_encode(&g, k, m, w, schedule, data, coding, blocksize, packetsize);
 				break;
 			case Liber8tion:
-				jerasure_schedule_encode(k, m, w, schedule, data, coding, blocksize, packetsize);
+				jerasure_schedule_encode(&g, k, m, w, schedule, data, coding, blocksize, packetsize);
 				break;
 			case RDP:
 			case EVENODD:
diff --git a/Examples/jerasure_01.c b/Examples/jerasure_01.c
index 20df225..ab448c8 100644
--- a/Examples/jerasure_01.c
+++ b/Examples/jerasure_01.c
@@ -73,11 +73,13 @@ int main(int argc, char **argv)
   matrix = talloc(int, r*c);
 
   n = 1;
+  gf2_t* g = galois_init_empty();
   for (i = 0; i < r*c; i++) {
     matrix[i] = n;
-    n = galois_single_multiply(n, 2, w);
+    n = galois_single_multiply(g, n, 2, w);
   }
-
+  galois_destroy(g);
+  
   printf("jerasure_01");
   for (i = 1; i < argc; i++) printf(" %s", argv[i]);
   printf("\n");
diff --git a/Examples/jerasure_02.c b/Examples/jerasure_02.c
index eb952ae..c8df0c7 100644
--- a/Examples/jerasure_02.c
+++ b/Examples/jerasure_02.c
@@ -71,14 +71,17 @@ int main(int argc, char **argv)
   if (sscanf(argv[3], "%d", &w) == 0 || w <= 0) usage("Bad w");
 
   matrix = talloc(int, r*c);
+  gf2_t* g = galois_init_empty();
 
   n = 1;
   for (i = 0; i < r*c; i++) {
     matrix[i] = n;
-    n = galois_single_multiply(n, 2, w);
+    n = galois_single_multiply(g, n, 2, w);
   }
 
-  bitmatrix = jerasure_matrix_to_bitmatrix(c, r, w, matrix);
+  bitmatrix = jerasure_matrix_to_bitmatrix(g, c, r, w, matrix);
+
+  galois_destroy(g);
 
   printf("jerasure_02");
   for (i = 1; i < argc; i++) printf(" %s", argv[i]);
diff --git a/Examples/jerasure_03.c b/Examples/jerasure_03.c
index 24bd15c..a2f8349 100644
--- a/Examples/jerasure_03.c
+++ b/Examples/jerasure_03.c
@@ -85,10 +85,12 @@ int main(int argc, char **argv)
   matrix_copy = talloc(int, k*k);
   inverse = talloc(int, k*k);
 
+  static gf2_t g;
+  
   for (i = 0; i < k; i++) {
     for (j = 0; j < k; j++) {
       n = i ^ ((1 << w)-1-j);
-      matrix[i*k+j] = (n == 0) ? 0 : galois_single_divide(1, n, w);
+      matrix[i*k+j] = (n == 0) ? 0 : galois_single_divide(&g, 1, n, w);
     }
   }
 
@@ -103,14 +105,14 @@ int main(int argc, char **argv)
   printf("The Cauchy Matrix:\n");
   jerasure_print_matrix(matrix, k, k, w);
   memcpy(matrix_copy, matrix, sizeof(int)*k*k);
-  i = jerasure_invertible_matrix(matrix_copy, k, w);
+  i = jerasure_invertible_matrix(&g, matrix_copy, k, w);
   printf("\nInvertible: %s\n", (i == 1) ? "Yes" : "No");
   if (i == 1) {
     printf("\nInverse:\n");
     memcpy(matrix_copy, matrix, sizeof(int)*k*k);
-    i = jerasure_invert_matrix(matrix_copy, inverse, k, w);
+    i = jerasure_invert_matrix(&g, matrix_copy, inverse, k, w);
     jerasure_print_matrix(inverse, k, k, w);
-    identity = jerasure_matrix_multiply(inverse, matrix, k, k, k, k, w);
+    identity = jerasure_matrix_multiply(&g, inverse, matrix, k, k, k, k, w);
     printf("\nInverse times matrix (should be identity):\n");
     jerasure_print_matrix(identity, k, k, w);
   }
diff --git a/Examples/jerasure_04.c b/Examples/jerasure_04.c
index 21410a5..81efc44 100644
--- a/Examples/jerasure_04.c
+++ b/Examples/jerasure_04.c
@@ -73,6 +73,7 @@ int main(int argc, char **argv)
   int *bitmatrix_copy;
   int *inverse;
   int *identity;
+  static gf2_t g;
 
   if (argc != 3) usage(NULL);
   if (sscanf(argv[1], "%d", &k) == 0 || k <= 0) usage("Bad k");
@@ -86,10 +87,10 @@ int main(int argc, char **argv)
   for (i = 0; i < k; i++) {
     for (j = 0; j < k; j++) {
       n = i ^ ((1 << w)-1-j);
-      matrix[i*k+j] = (n == 0) ? 0 : galois_single_divide(1, n, w);
+      matrix[i*k+j] = (n == 0) ? 0 : galois_single_divide(&g, 1, n, w);
     }
   }
-  bitmatrix = jerasure_matrix_to_bitmatrix(k, k, w, matrix);
+  bitmatrix = jerasure_matrix_to_bitmatrix(&g, k, k, w, matrix);
 
   printf("<HTML><TITLE>jerasure_04");
   for (i = 1; i < argc; i++) printf(" %s", argv[i]);
@@ -109,7 +110,7 @@ int main(int argc, char **argv)
     memcpy(bitmatrix_copy, bitmatrix, sizeof(int)*k*w*k*w);
     i = jerasure_invert_bitmatrix(bitmatrix_copy, inverse, k*w);
     jerasure_print_bitmatrix(inverse, k*w, k*w, w);
-    identity = jerasure_matrix_multiply(inverse, bitmatrix, k*w, k*w, k*w, k*w, 2);
+    identity = jerasure_matrix_multiply(&g, inverse, bitmatrix, k*w, k*w, k*w, k*w, 2);
     printf("\nInverse times matrix (should be identity):\n");
     jerasure_print_bitmatrix(identity, k*w, k*w, w);
   }
diff --git a/Examples/jerasure_05.c b/Examples/jerasure_05.c
index 20b059b..9e42a16 100644
--- a/Examples/jerasure_05.c
+++ b/Examples/jerasure_05.c
@@ -77,7 +77,7 @@ static void print_data_and_coding(int k, int m, int w, int size,
 {
   int i, j, x;
   int n, sp;
-
+  
   if(k > m) n = k;
   else n = m;
   sp = size * 2 + size/(w/8) + 8;
@@ -118,6 +118,7 @@ int main(int argc, char **argv)
   int *erasures, *erased;
   int *decoding_matrix, *dm_ids;
   uint32_t seed;
+  static gf2_t g;
   
   if (argc != 6) usage(NULL);
   if (sscanf(argv[1], "%d", &k) == 0 || k <= 0) usage("Bad k");
@@ -131,7 +132,7 @@ int main(int argc, char **argv)
   matrix = talloc(int, m*k);
   for (i = 0; i < m; i++) {
     for (j = 0; j < k; j++) {
-      matrix[i*k+j] = galois_single_divide(1, i ^ (m + j), w);
+      matrix[i*k+j] = galois_single_divide(&g, 1, i ^ (m + j), w);
     }
   }
 
@@ -159,7 +160,7 @@ int main(int argc, char **argv)
     coding[i] = talloc(char, size);
   }
 
-  jerasure_matrix_encode(k, m, w, matrix, data, coding, size);
+  jerasure_matrix_encode(&g, k, m, w, matrix, data, coding, size);
   
   printf("Encoding Complete:\n\n");
   print_data_and_coding(k, m, w, size, data, coding);
@@ -181,7 +182,7 @@ int main(int argc, char **argv)
   printf("Erased %d random devices:\n\n", m);
   print_data_and_coding(k, m, w, size, data, coding);
   
-  i = jerasure_matrix_decode(k, m, w, matrix, 0, erasures, data, coding, size);
+  i = jerasure_matrix_decode(&g, k, m, w, matrix, 0, erasures, data, coding, size);
 
   printf("State of the system after decoding:\n\n");
   print_data_and_coding(k, m, w, size, data, coding);
@@ -192,7 +193,7 @@ int main(int argc, char **argv)
   for (i = 0; i < m; i++) erased[i] = 1;
   for (; i < k+m; i++) erased[i] = 0;
 
-  jerasure_make_decoding_matrix(k, m, w, matrix, erased, decoding_matrix, dm_ids);
+  jerasure_make_decoding_matrix(&g, k, m, w, matrix, erased, decoding_matrix, dm_ids);
 
   printf("Suppose we erase the first %d devices.  Here is the decoding matrix:\n\n", m);
   jerasure_print_matrix(decoding_matrix, k, k, w);
@@ -201,7 +202,7 @@ int main(int argc, char **argv)
   jerasure_print_matrix(dm_ids, 1, k, w);
 
   bzero(data[0], size);
-  jerasure_matrix_dotprod(k, w, decoding_matrix, dm_ids, 0, data, coding, size);
+  jerasure_matrix_dotprod(&g, k, w, decoding_matrix, dm_ids, 0, data, coding, size);
 
   printf("\nAfter calling jerasure_matrix_dotprod, we calculate the value of device #0 to be:\n\n");
   printf("D0 :");
diff --git a/Examples/jerasure_06.c b/Examples/jerasure_06.c
index 9af9ca2..755a499 100644
--- a/Examples/jerasure_06.c
+++ b/Examples/jerasure_06.c
@@ -107,6 +107,7 @@ int main(int argc, char **argv)
   int *erasures, *erased;
   int *decoding_matrix, *dm_ids;
   uint32_t seed;
+  static gf2_t g;
   
   if (argc != 6) usage(NULL);
   if (sscanf(argv[1], "%d", &k) == 0 || k <= 0) usage("Bad k");
@@ -121,10 +122,10 @@ int main(int argc, char **argv)
   matrix = talloc(int, m*k);
   for (i = 0; i < m; i++) {
     for (j = 0; j < k; j++) {
-      matrix[i*k+j] = galois_single_divide(1, i ^ (m + j), w);
+      matrix[i*k+j] = galois_single_divide(&g, 1, i ^ (m + j), w);
     }
   }
-  bitmatrix = jerasure_matrix_to_bitmatrix(k, m, w, matrix);
+  bitmatrix = jerasure_matrix_to_bitmatrix(&g, k, m, w, matrix);
 
   printf("<HTML><TITLE>jerasure_06");
   for (i = 1; i < argc; i++) printf(" %s", argv[i]);
@@ -149,7 +150,7 @@ int main(int argc, char **argv)
     coding[i] = talloc(char, psize*w);
   }
 
-  jerasure_bitmatrix_encode(k, m, w, bitmatrix, data, coding, w*psize, psize);
+  jerasure_bitmatrix_encode(&g, k, m, w, bitmatrix, data, coding, w*psize, psize);
   
   printf("Encoding Complete - Here is the state of the system\n\n");
   printf("<p>\n");
@@ -183,7 +184,7 @@ int main(int argc, char **argv)
   print_array(coding, m, psize*w, psize, "C");
   printf("<hr>\n");
 
-  i = jerasure_bitmatrix_decode(k, m, w, bitmatrix, 0, erasures, data, coding, 
+  i = jerasure_bitmatrix_decode(&g, k, m, w, bitmatrix, 0, erasures, data, coding, 
 		  w*psize, psize);
 
   printf("Here is the state of the system after decoding:\n\n");
@@ -220,7 +221,7 @@ int main(int argc, char **argv)
   printf("<hr>\n");
 
   for (i = 0; i < x; i++) {
-    jerasure_bitmatrix_dotprod(k, w, decoding_matrix+i*(k*w*w), dm_ids, i, data, coding, w*psize, psize);
+    jerasure_bitmatrix_dotprod(&g, k, w, decoding_matrix+i*(k*w*w), dm_ids, i, data, coding, w*psize, psize);
   }
 
   printf("Here is the state of the system after calling <b>jerasure_bitmatrix_dotprod()</b> %d time%s with the decoding matrix:\n\n", x, (x == 1) ? "" : "s");
diff --git a/Examples/jerasure_07.c b/Examples/jerasure_07.c
index cc57163..199eef9 100644
--- a/Examples/jerasure_07.c
+++ b/Examples/jerasure_07.c
@@ -108,6 +108,7 @@ int main(int argc, char **argv)
   int *erasures, *erased;
   double stats[3];
   uint32_t seed;
+  static gf2_t g;
   
   if (argc != 5) usage("Wrong number of arguments");
   if (sscanf(argv[1], "%d", &k) == 0 || k <= 0) usage("Bad k");
@@ -119,10 +120,10 @@ int main(int argc, char **argv)
   matrix = talloc(int, m*k);
   for (i = 0; i < m; i++) {
     for (j = 0; j < k; j++) {
-      matrix[i*k+j] = galois_single_divide(1, i ^ (m + j), w);
+      matrix[i*k+j] = galois_single_divide(&g, 1, i ^ (m + j), w);
     }
   }
-  bitmatrix = jerasure_matrix_to_bitmatrix(k, m, w, matrix);
+  bitmatrix = jerasure_matrix_to_bitmatrix(&g, k, m, w, matrix);
 
   printf("<HTML><TITLE>jerasure_07");
   for (i = 1; i < argc; i++) printf(" %s", argv[i]);
@@ -151,7 +152,7 @@ int main(int argc, char **argv)
     coding[i] = talloc(char, sizeof(long)*w);
   }
 
-  jerasure_schedule_encode(k, m, w, dumb, data, coding, w*sizeof(long), sizeof(long));
+  jerasure_schedule_encode(&g, k, m, w, dumb, data, coding, w*sizeof(long), sizeof(long));
   jerasure_get_stats(stats);
   printf("Dumb Encoding Complete: - %.0lf XOR'd bytes.  State of the system:\n\n", stats[0]);
   printf("<p>\n");
@@ -160,7 +161,7 @@ int main(int argc, char **argv)
   print_array(coding, m, sizeof(long)*w, sizeof(long), "C");
   printf("<hr>\n");
 
-  jerasure_schedule_encode(k, m, w, smart, data, coding, w*sizeof(long), sizeof(long));
+  jerasure_schedule_encode(&g, k, m, w, smart, data, coding, w*sizeof(long), sizeof(long));
   jerasure_get_stats(stats);
   printf("Smart Encoding Complete: - %.0lf XOR'd bytes\n\n", stats[0]);
   printf("<p>\n");
@@ -189,7 +190,7 @@ int main(int argc, char **argv)
   print_array(coding, m, sizeof(long)*w, sizeof(long), "C");
   printf("<hr>\n");
 
-  jerasure_schedule_decode_lazy(k, m, w, bitmatrix, erasures, data, coding, w*sizeof(long), sizeof(long), 1);
+  jerasure_schedule_decode_lazy(&g, k, m, w, bitmatrix, erasures, data, coding, w*sizeof(long), sizeof(long), 1);
   jerasure_get_stats(stats);
 
   printf("State of the system after decoding: %.0lf XOR'd bytes\n\n", stats[0]);
@@ -211,7 +212,7 @@ int main(int argc, char **argv)
   print_array(coding, m, sizeof(long)*w, sizeof(long), "C");
   printf("<hr>\n");
 
-  jerasure_do_scheduled_operations(ptrs, smart, sizeof(long));
+  jerasure_do_scheduled_operations(&g, ptrs, smart, sizeof(long));
   printf("And using <b>jerasure_do_scheduled_operations()</b>: %.0lf XOR'd bytes\n\n", stats[0]);
   printf("<p>\n");
   print_array(data, k, sizeof(long)*w, sizeof(long), "D");
diff --git a/Examples/jerasure_08.c b/Examples/jerasure_08.c
index 0eb5ace..3450661 100644
--- a/Examples/jerasure_08.c
+++ b/Examples/jerasure_08.c
@@ -111,6 +111,7 @@ int main(int argc, char **argv)
   int *erasures, *erased;
   double stats[3];
   uint32_t seed;
+  static gf2_t g;
   
   if (argc != 4) usage("Wrong number of arguments");
   if (sscanf(argv[1], "%d", &k) == 0 || k <= 0) usage("Bad k");
@@ -126,9 +127,9 @@ int main(int argc, char **argv)
   i = 1;
   for (j = 0; j < k; j++) {
     matrix[k+j] = i;
-    i = galois_single_multiply(i, 2, w);
+    i = galois_single_multiply(&g, i, 2, w);
   }
-  bitmatrix = jerasure_matrix_to_bitmatrix(k, m, w, matrix);
+  bitmatrix = jerasure_matrix_to_bitmatrix(&g, k, m, w, matrix);
 
   smart = jerasure_smart_bitmatrix_to_schedule(k, m, w, bitmatrix);
   cache = jerasure_generate_schedule_cache(k, m, w, bitmatrix, 1);
@@ -144,7 +145,7 @@ int main(int argc, char **argv)
     coding[i] = talloc(char, sizeof(long)*w);
   }
 
-  jerasure_schedule_encode(k, m, w, smart, data, coding, w*sizeof(long), sizeof(long));
+  jerasure_schedule_encode(&g, k, m, w, smart, data, coding, w*sizeof(long), sizeof(long));
   jerasure_get_stats(stats);
 
   printf("<HTML><TITLE>jerasure_08");
@@ -168,7 +169,7 @@ int main(int argc, char **argv)
   erasures[2] = -1;
   for (j = 0; j < m; j++) bzero(coding[j], sizeof(long)*w);
 
-  jerasure_schedule_decode_cache(k, m, w, cache, erasures, data, coding, w*sizeof(long), sizeof(long));
+  jerasure_schedule_decode_cache(&g, k, m, w, cache, erasures, data, coding, w*sizeof(long), sizeof(long));
   jerasure_get_stats(stats);
   printf("Encoding Using the Schedule Cache: - %.0lf XOR'd bytes\n\n", stats[0]);
   printf("<p>\n");
@@ -196,7 +197,7 @@ int main(int argc, char **argv)
   print_array(coding, m, sizeof(long)*w, sizeof(long), "C");
   printf("<hr>\n");
   
-  jerasure_schedule_decode_cache(k, m, w, cache, erasures, data, coding, w*sizeof(long), sizeof(long));
+  jerasure_schedule_decode_cache(&g, k, m, w, cache, erasures, data, coding, w*sizeof(long), sizeof(long));
   jerasure_get_stats(stats);
 
   printf("State of the system after decoding: %.0lf XOR'd bytes\n\n", stats[0]);
@@ -214,7 +215,7 @@ int main(int argc, char **argv)
   print_array(coding, m, sizeof(long)*w, sizeof(long), "C");
   printf("<hr>\n");
   
-  jerasure_do_parity(k, data, coding[0], sizeof(long)*w);
+  jerasure_do_parity(&g, k, data, coding[0], sizeof(long)*w);
   printf("State of the system after using\n");
   printf("<b>jerasure_do_parity()</b> to re-encode it:\n\n");
   printf("<p>\n");
diff --git a/Examples/liberation_01.c b/Examples/liberation_01.c
index d3b4392..2494a57 100644
--- a/Examples/liberation_01.c
+++ b/Examples/liberation_01.c
@@ -109,6 +109,7 @@ int main(int argc, char **argv)
   int *erasures, *erased;
   double stats[3];
   uint32_t seed;
+  static gf2_t g;
   
   if (argc != 4) usage("Wrong number of arguments");
   if (sscanf(argv[1], "%d", &k) == 0 || k <= 0) usage("Bad k");
@@ -149,7 +150,7 @@ int main(int argc, char **argv)
     coding[i] = talloc(char, sizeof(long)*w);
   }
 
-  jerasure_schedule_encode(k, m, w, dumb, data, coding, w*sizeof(long), sizeof(long));
+  jerasure_schedule_encode(&g, k, m, w, dumb, data, coding, w*sizeof(long), sizeof(long));
   jerasure_get_stats(stats);
   printf("Smart Encoding Complete: - %.0lf XOR'd bytes.  State of the system:\n\n", stats[0]);
   printf("<p>\n");
@@ -178,7 +179,7 @@ int main(int argc, char **argv)
   print_array(coding, m, sizeof(long)*w, sizeof(long), "C");
   printf("<hr>\n");
   
-  jerasure_schedule_decode_lazy(k, m, w, bitmatrix, erasures, data, coding, w*sizeof(long), sizeof(long), 1);
+  jerasure_schedule_decode_lazy(&g, k, m, w, bitmatrix, erasures, data, coding, w*sizeof(long), sizeof(long), 1);
   jerasure_get_stats(stats);
 
   printf("State of the system after decoding: %.0lf XOR'd bytes\n\n", stats[0]);
diff --git a/Examples/reed_sol_01.c b/Examples/reed_sol_01.c
index da46aeb..cfed7aa 100644
--- a/Examples/reed_sol_01.c
+++ b/Examples/reed_sol_01.c
@@ -116,6 +116,7 @@ int main(int argc, char **argv)
   unsigned char uc;
   int *erasures, *erased;
   uint32_t seed;
+  static gf2_t g;
   
   if (argc != 5) usage(NULL);
   if (sscanf(argv[1], "%d", &k) == 0 || k <= 0) usage("Bad k");
@@ -124,7 +125,7 @@ int main(int argc, char **argv)
   if (sscanf(argv[4], "%u", &seed) == 0) usage("Bad seed");
   if (w <= 16 && k + m > (1 << w)) usage("k + m is too big");
 
-  matrix = reed_sol_vandermonde_coding_matrix(k, m, w);
+  matrix = reed_sol_vandermonde_coding_matrix(&g, k, m, w);
 
   printf("<HTML><TITLE>reed_sol_01 %d %d %d %d\n", k, m, w, seed);
   printf("
reed_sol_01 %d %d %d %d
\n", k, m, w, seed);
@@ -153,7 +154,7 @@ int main(int argc, char **argv)
     ccopy[i] = talloc(char, sizeof(long));
   }
 
-  jerasure_matrix_encode(k, m, w, matrix, data, coding, sizeof(long));
+  jerasure_matrix_encode(&g, k, m, w, matrix, data, coding, sizeof(long));
 
   for (i = 0; i < m; i++) {
     memcpy(ccopy[i], coding[i], sizeof(long));
@@ -179,7 +180,7 @@ int main(int argc, char **argv)
   printf("Erased %d random devices:\n\n", m);
   print_data_and_coding(k, m, w, sizeof(long), data, coding);
   
-  i = jerasure_matrix_decode(k, m, w, matrix, 1, erasures, data, coding, sizeof(long));
+  i = jerasure_matrix_decode(&g, k, m, w, matrix, 1, erasures, data, coding, sizeof(long));
 
   printf("State of the system after decoding:\n\n");
   print_data_and_coding(k, m, w, sizeof(long), data, coding);
diff --git a/Examples/reed_sol_02.c b/Examples/reed_sol_02.c
index 6beb4f6..bf1bb69 100644
--- a/Examples/reed_sol_02.c
+++ b/Examples/reed_sol_02.c
@@ -73,6 +73,7 @@ int main(int argc, char **argv)
 {
   int k, w, m;
   int *matrix;
+  static gf2_t g;
   
   if (argc != 4) usage(NULL);
   if (sscanf(argv[1], "%d", &k) == 0 || k <= 0) usage("Bad k");
@@ -80,7 +81,7 @@ int main(int argc, char **argv)
   if (sscanf(argv[3], "%d", &w) == 0 || w <= 0 || w > 32) usage("Bad w");
   if (w <= 30 && k + m > (1 << w)) usage("k + m is too big");
 
-  matrix = reed_sol_extended_vandermonde_matrix(k+m, k, w);
+  matrix = reed_sol_extended_vandermonde_matrix(&g, k+m, k, w);
 
   printf("reed_sol_02 %d %d %d\n", k, m, w);
   printf("
reed_sol_02 %d %d %d
\n", k, m, w);
@@ -89,12 +90,12 @@ int main(int argc, char **argv)
   jerasure_print_matrix(matrix, k+m, k, w);
   printf("\n");
 
-  matrix = reed_sol_big_vandermonde_distribution_matrix(k+m, k, w);
+  matrix = reed_sol_big_vandermonde_distribution_matrix(&g, k+m, k, w);
   printf("Vandermonde Generator Matrix (G^T):\n\n");
   jerasure_print_matrix(matrix, k+m, k, w);
   printf("\n");
 
-  matrix = reed_sol_vandermonde_coding_matrix(k, m, w);
+  matrix = reed_sol_vandermonde_coding_matrix(&g, k, m, w);
   printf("Vandermonde Coding Matrix:\n\n");
   jerasure_print_matrix(matrix, m, k, w);
   printf("\n");
diff --git a/Examples/reed_sol_03.c b/Examples/reed_sol_03.c
index 187cebd..b708b50 100644
--- a/Examples/reed_sol_03.c
+++ b/Examples/reed_sol_03.c
@@ -116,6 +116,7 @@ int main(int argc, char **argv)
   char **data, **coding, **dcopy, **ccopy;
   int *erasures, *erased;
   uint32_t seed;
+  static gf2_t g;
   
   if (argc != 4) usage(NULL);
   if (sscanf(argv[1], "%d", &k) == 0 || k <= 0) usage("Bad k");
@@ -125,7 +126,7 @@ int main(int argc, char **argv)
   if (w <= 16 && k + m > (1 << w)) usage("k + m is too big");
 
   MOA_Seed(seed);
-  matrix = reed_sol_r6_coding_matrix(k, w);
+  matrix = reed_sol_r6_coding_matrix(&g, k, w);
 
   printf("reed_sol_03 %d %d %d\n", k, w, seed);
   printf("
reed_sol_03 %d %d %d
\n", k, w, seed);
@@ -154,7 +155,7 @@ int main(int argc, char **argv)
     ccopy[i] = talloc(char, sizeof(long));
   }
 
-  reed_sol_r6_encode(k, w, data, coding, sizeof(long));
+  reed_sol_r6_encode(&g, k, w, data, coding, sizeof(long));
   for (i = 0; i < m; i++) {
     memcpy(ccopy[i], coding[i], sizeof(long));
   }
@@ -179,7 +180,7 @@ int main(int argc, char **argv)
   printf("Erased %d random devices:\n\n", m);
   print_data_and_coding(k, m, w, sizeof(long), data, coding);
   
-  i = jerasure_matrix_decode(k, m, w, matrix, 1, erasures, data, coding, sizeof(long));
+  i = jerasure_matrix_decode(&g, k, m, w, matrix, 1, erasures, data, coding, sizeof(long));
 
   printf("State of the system after decoding:\n\n");
   print_data_and_coding(k, m, w, sizeof(long), data, coding);
diff --git a/Examples/reed_sol_04.c b/Examples/reed_sol_04.c
index f763fb0..fe6141d 100644
--- a/Examples/reed_sol_04.c
+++ b/Examples/reed_sol_04.c
@@ -77,6 +77,7 @@ int main(int argc, char **argv)
   int *a32, *copy;
   int i;
   int w;
+  static gf2_t g;
   
   if (argc != 3) usage(NULL);
   if (sscanf(argv[1], "%d", &w) == 0 || (w != 8 && w != 16 && w != 32)) usage("Bad w");
@@ -96,21 +97,21 @@ int main(int argc, char **argv)
   if (w == 8) {
     x = (unsigned char *) copy;
     y = (unsigned char *) a32;
-    reed_sol_galois_w08_region_multby_2((char *) a32, sizeof(int)*4);
+    reed_sol_galois_w08_region_multby_2(&g, (char *) a32, sizeof(int)*4);
     for (i = 0; i < 4*sizeof(int)/sizeof(char); i++) {
        printf("Char %2d: %3u *2 = %3u\n", i, x[i], y[i]);
     }
   } else if (w == 16) {
     xs = (unsigned short *) copy;
     ys = (unsigned short *) a32;
-    reed_sol_galois_w16_region_multby_2((char *) a32, sizeof(int)*4);
+    reed_sol_galois_w16_region_multby_2(&g, (char *) a32, sizeof(int)*4);
     for (i = 0; i < 4*sizeof(int)/sizeof(short); i++) {
        printf("Short %2d: %5u *2 = %5u\n", i, xs[i], ys[i]);
     }
   } else if (w == 32) {
     xi = (unsigned int *) copy;
     yi = (unsigned int *) a32;
-    reed_sol_galois_w16_region_multby_2((char *) a32, sizeof(int)*4);
+    reed_sol_galois_w16_region_multby_2(&g, (char *) a32, sizeof(int)*4);
     for (i = 0; i < 4*sizeof(int)/sizeof(int); i++) {
        printf("Int %2d: %10u *2 = %10u\n", i, xi[i], yi[i]);
     }
diff --git a/Examples/reed_sol_test_gf.c b/Examples/reed_sol_test_gf.c
index 9e8856c..a1cb9e1 100644
--- a/Examples/reed_sol_test_gf.c
+++ b/Examples/reed_sol_test_gf.c
@@ -109,6 +109,7 @@ int main(int argc, char **argv)
   int *erasures, *erased;
   gf_t *gf = NULL;
   uint32_t seed;
+  static gf2_t g;
   
   if (argc < 6) usage("Not enough command line arguments");  
   if (sscanf(argv[1], "%d", &k) == 0 || k <= 0) usage("Bad k");
@@ -125,9 +126,9 @@ int main(int argc, char **argv)
     usage("Invalid arguments given for GF!\n");
   }
 
-  galois_change_technique(gf, w); 
+  galois_change_technique(&g, gf, w); 
 
-  matrix = reed_sol_vandermonde_coding_matrix(k, m, w);
+  matrix = reed_sol_vandermonde_coding_matrix(&g, k, m, w);
 
   printf("reed_sol_test_gf");
   for (i = 1; i < argc; i++) printf(" %s", argv[i]);
@@ -154,7 +155,7 @@ int main(int argc, char **argv)
     old_values[i] = talloc(char, BUFSIZE);
   }
 
-  jerasure_matrix_encode(k, m, w, matrix, data, coding, BUFSIZE);
+  jerasure_matrix_encode(&g, k, m, w, matrix, data, coding, BUFSIZE);
   
   erasures = talloc(int, (m+1));
   erased = talloc(int, (k+m));
@@ -170,7 +171,7 @@ int main(int argc, char **argv)
   }
   erasures[i] = -1;
 
-  i = jerasure_matrix_decode(k, m, w, matrix, 1, erasures, data, coding, BUFSIZE);
+  i = jerasure_matrix_decode(&g, k, m, w, matrix, 1, erasures, data, coding, BUFSIZE);
 
   for (i = 0; i < m; i++) {
     if (erasures[i] < k) {
diff --git a/Examples/reed_sol_time_gf.c b/Examples/reed_sol_time_gf.c
index 175de8b..8f3f41a 100644
--- a/Examples/reed_sol_time_gf.c
+++ b/Examples/reed_sol_time_gf.c
@@ -109,6 +109,7 @@ int main(int argc, char **argv)
   uint32_t seed;
   double t = 0, total_time = 0;
   gf_t *gf = NULL;
+  static gf2_t g;
   
   if (argc < 8) usage(NULL);  
   if (sscanf(argv[1], "%d", &k) == 0 || k <= 0) usage("Bad k");
@@ -127,9 +128,9 @@ int main(int argc, char **argv)
     usage("Invalid arguments given for GF!\n");
   }
 
-  galois_change_technique(gf, w); 
+  galois_change_technique(&g, gf, w); 
 
-  matrix = reed_sol_vandermonde_coding_matrix(k, m, w);
+  matrix = reed_sol_vandermonde_coding_matrix(&g, k, m, w);
 
   printf("<HTML><TITLE>reed_sol_time_gf");
   for (i = 1; i < argc; i++) printf(" %s", argv[i]);
@@ -158,7 +159,7 @@ int main(int argc, char **argv)
 
   for (i = 0; i < iterations; i++) {
     t = timing_now();
-    jerasure_matrix_encode(k, m, w, matrix, data, coding, bufsize);
+    jerasure_matrix_encode(&g, k, m, w, matrix, data, coding, bufsize);
     total_time += timing_now() - t;
   }
 
@@ -180,7 +181,7 @@ int main(int argc, char **argv)
 
   for (i = 0; i < iterations; i++) {
     t = timing_now();
-    jerasure_matrix_decode(k, m, w, matrix, 1, erasures, data, coding, bufsize);
+    jerasure_matrix_decode(&g, k, m, w, matrix, 1, erasures, data, coding, bufsize);
     total_time += timing_now() - t;
   }
   
diff --git a/include/cauchy.h b/include/cauchy.h
index f8d1ae3..c837b80 100644
--- a/include/cauchy.h
+++ b/include/cauchy.h
@@ -43,11 +43,11 @@
 extern "C" {
 #endif
 
-extern int *cauchy_original_coding_matrix(int k, int m, int w);
-extern int *cauchy_xy_coding_matrix(int k, int m, int w, int *x, int *y);
-extern void cauchy_improve_coding_matrix(int k, int m, int w, int *matrix);
-extern int *cauchy_good_general_coding_matrix(int k, int m, int w);
-extern int cauchy_n_ones(int n, int w);
+extern int *cauchy_original_coding_matrix(gf2_t* g, int k, int m, int w);
+extern int *cauchy_xy_coding_matrix(gf2_t* g, int k, int m, int w, int *x, int *y);
+extern void cauchy_improve_coding_matrix(gf2_t* g, int k, int m, int w, int *matrix);
+extern int *cauchy_good_general_coding_matrix(gf2_t* g, int k, int m, int w);
+extern int cauchy_n_ones(gf2_t* g, int n, int w);
 
 #ifdef __cplusplus
 }
diff --git a/include/galois.h b/include/galois.h
index 78e6567..cc42f25 100644
--- a/include/galois.h
+++ b/include/galois.h
@@ -46,43 +46,55 @@
 extern "C" {
 #endif
 
-extern int galois_init_default_field(int w);
-extern int galois_uninit_field(int w);
-extern void galois_change_technique(gf_t *gf, int w);
+typedef struct gf2_t_ {
+	gf_t* gf;
+	gf_t* gfp_array[64];
+	int gfp_is_composite[64];
+} gf2_t;
 
-extern int galois_single_multiply(int a, int b, int w);
-extern int galois_single_divide(int a, int b, int w);
-extern int galois_inverse(int x, int w);
+extern gf2_t* galois_init_empty();
+extern void galois_destroy(gf2_t*);
+extern int galois_init_default_field(gf2_t*, int w);
+extern int galois_uninit_field(gf2_t*, int w);
+extern void galois_change_technique(gf2_t *g, gf_t* gf, int w);
 
-void galois_region_xor(           char *src,         /* Source Region */
+extern int galois_single_multiply(gf2_t*, int a, int b, int w);
+extern int galois_single_divide(gf2_t*, int a, int b, int w);
+extern int galois_inverse(gf2_t*, int x, int w);
+
+void galois_region_xor(           gf2_t*,
+	                              char *src,         /* Source Region */
                                   char *dest,        /* Dest Region (holds result) */
                                   int nbytes);      /* Number of bytes in region */
 
 /* These multiply regions in w=8, w=16 and w=32.  They are much faster
    than calling galois_single_multiply.  The regions must be long word aligned. */
 
-void galois_w08_region_multiply(char *region,       /* Region to multiply */
+void galois_w08_region_multiply(gf2_t*,
+                                  char *region,       /* Region to multiply */
                                   int multby,       /* Number to multiply by */
                                   int nbytes,       /* Number of bytes in region */
                                   char *r2,         /* If r2 != NULL, products go here.  
                                                        Otherwise region is overwritten */
                                   int add);         /* If (r2 != NULL && add) the produce is XOR'd with r2 */
 
-void galois_w16_region_multiply(char *region,       /* Region to multiply */
+void galois_w16_region_multiply(gf2_t*,
+	char *region,       /* Region to multiply */
                                   int multby,       /* Number to multiply by */
                                   int nbytes,       /* Number of bytes in region */
                                   char *r2,         /* If r2 != NULL, products go here.  
                                                        Otherwise region is overwritten */
                                   int add);         /* If (r2 != NULL && add) the produce is XOR'd with r2 */
 
-void galois_w32_region_multiply(char *region,       /* Region to multiply */
+void galois_w32_region_multiply(gf2_t*,
+								char *region,       /* Region to multiply */
                                   int multby,       /* Number to multiply by */
                                   int nbytes,       /* Number of bytes in region */
                                   char *r2,         /* If r2 != NULL, products go here.  
                                                        Otherwise region is overwritten */
                                   int add);         /* If (r2 != NULL && add) the produce is XOR'd with r2 */
 
-gf_t* galois_init_field(int w,
+gf2_t* galois_init_field(int w,
                              int mult_type,
                              int region_type,
                              int divide_type,
@@ -90,13 +102,13 @@ gf_t* galois_init_field(int w,
                              int arg1,
                              int arg2);
 
-gf_t* galois_init_composite_field(int w,
+gf2_t* galois_init_composite_field(int w,
                                 int region_type,
                                 int divide_type,
                                 int degree,
                                 gf_t* base_gf);
 
-gf_t * galois_get_field_ptr(int w);
+gf_t * galois_get_field_ptr(gf2_t*, int w);
 
 #ifdef __cplusplus
 }
diff --git a/include/jerasure.h b/include/jerasure.h
index de874e1..f311458 100644
--- a/include/jerasure.h
+++ b/include/jerasure.h
@@ -123,7 +123,7 @@ extern "C" {
                               jerasure_generate_schedule_cache.
  */
 
-int *jerasure_matrix_to_bitmatrix(int k, int m, int w, int *matrix);
+int *jerasure_matrix_to_bitmatrix(gf2_t* g, int k, int m, int w, int *matrix);
 int **jerasure_dumb_bitmatrix_to_schedule(int k, int m, int w, int *bitmatrix);
 int **jerasure_smart_bitmatrix_to_schedule(int k, int m, int w, int *bitmatrix);
 int ***jerasure_generate_schedule_cache(int k, int m, int w, int *bitmatrix, int smart);
@@ -136,15 +136,15 @@ void jerasure_free_schedule_cache(int k, int m, int ***cache);
 /* Encoding - these are all straightforward.  jerasure_matrix_encode only 
    works with w = 8|16|32.  */
 
-void jerasure_do_parity(int k, char **data_ptrs, char *parity_ptr, int size);
+void jerasure_do_parity(gf2_t* g, int k, char **data_ptrs, char *parity_ptr, int size);
 
-void jerasure_matrix_encode(int k, int m, int w, int *matrix,
+void jerasure_matrix_encode(gf2_t* g, int k, int m, int w, int *matrix,
                           char **data_ptrs, char **coding_ptrs, int size);
 
-void jerasure_bitmatrix_encode(int k, int m, int w, int *bitmatrix,
+void jerasure_bitmatrix_encode(gf2_t* g, int k, int m, int w, int *bitmatrix,
                             char **data_ptrs, char **coding_ptrs, int size, int packetsize);
 
-void jerasure_schedule_encode(int k, int m, int w, int **schedule,
+void jerasure_schedule_encode(gf2_t* g, int k, int m, int w, int **schedule,
                                   char **data_ptrs, char **coding_ptrs, int size, int packetsize);
 
 /* ------------------------------------------------------------ */
@@ -183,22 +183,22 @@ void jerasure_schedule_encode(int k, int m, int w, int **schedule,
     
  */
 
-int jerasure_matrix_decode(int k, int m, int w, 
+int jerasure_matrix_decode(gf2_t* g, int k, int m, int w, 
                           int *matrix, int row_k_ones, int *erasures,
                           char **data_ptrs, char **coding_ptrs, int size);
                           
-int jerasure_bitmatrix_decode(int k, int m, int w, 
+int jerasure_bitmatrix_decode(gf2_t* g, int k, int m, int w, 
                             int *bitmatrix, int row_k_ones, int *erasures,
                             char **data_ptrs, char **coding_ptrs, int size, int packetsize);
 
-int jerasure_schedule_decode_lazy(int k, int m, int w, int *bitmatrix, int *erasures,
+int jerasure_schedule_decode_lazy(gf2_t* g, int k, int m, int w, int *bitmatrix, int *erasures,
                             char **data_ptrs, char **coding_ptrs, int size, int packetsize,
                             int smart);
 
-int jerasure_schedule_decode_cache(int k, int m, int w, int ***scache, int *erasures,
+int jerasure_schedule_decode_cache(gf2_t* g, int k, int m, int w, int ***scache, int *erasures,
                             char **data_ptrs, char **coding_ptrs, int size, int packetsize);
 
-int jerasure_make_decoding_matrix(int k, int m, int w, int *matrix, int *erased, 
+int jerasure_make_decoding_matrix(gf2_t* g, int k, int m, int w, int *matrix, int *erased, 
                                   int *decoding_matrix, int *dm_ids);
 
 int jerasure_make_decoding_bitmatrix(int k, int m, int w, int *matrix, int *erased, 
@@ -221,15 +221,15 @@ int *jerasure_erasures_to_erased(int k, int m, int *erasures);
 
  */
  
-void jerasure_matrix_dotprod(int k, int w, int *matrix_row,
+void jerasure_matrix_dotprod(gf2_t* g, int k, int w, int *matrix_row,
                           int *src_ids, int dest_id,
                           char **data_ptrs, char **coding_ptrs, int size);
 
-void jerasure_bitmatrix_dotprod(int k, int w, int *bitmatrix_row,
+void jerasure_bitmatrix_dotprod(gf2_t* g, int k, int w, int *bitmatrix_row,
                              int *src_ids, int dest_id,
                              char **data_ptrs, char **coding_ptrs, int size, int packetsize);
 
-void jerasure_do_scheduled_operations(char **ptrs, int **schedule, int packetsize);
+void jerasure_do_scheduled_operations(gf2_t* g, char **ptrs, int **schedule, int packetsize);
 
 /* ------------------------------------------------------------ */
 /* Matrix Inversion ------------------------------------------- */
@@ -255,9 +255,9 @@ void jerasure_do_scheduled_operations(char **ptrs, int **schedule, int packetsiz
    invertible.  (0 or 1). Mat will be destroyed.
  */
 
-int jerasure_invert_matrix(int *mat, int *inv, int rows, int w);
+int jerasure_invert_matrix(gf2_t* g, int *mat, int *inv, int rows, int w);
 int jerasure_invert_bitmatrix(int *mat, int *inv, int rows);
-int jerasure_invertible_matrix(int *mat, int rows, int w);
+int jerasure_invertible_matrix(gf2_t* g, int *mat, int rows, int w);
 int jerasure_invertible_bitmatrix(int *mat, int rows);
 
 /* ------------------------------------------------------------ */
@@ -277,7 +277,7 @@ void jerasure_print_matrix(int *matrix, int rows, int cols, int w);
 void jerasure_print_bitmatrix(int *matrix, int rows, int cols, int w);
 
 
-int *jerasure_matrix_multiply(int *m1, int *m2, int r1, int c1, int r2, int c2, int w);
+int *jerasure_matrix_multiply(gf2_t* g, int *m1, int *m2, int r1, int c1, int r2, int c2, int w);
 
 /* ------------------------------------------------------------ */
 /* Stats ------------------------------------------------------ */
diff --git a/include/reed_sol.h b/include/reed_sol.h
index 48fc4e6..c47b26f 100644
--- a/include/reed_sol.h
+++ b/include/reed_sol.h
@@ -43,16 +43,16 @@
 extern "C" {
 #endif
 
-extern int *reed_sol_vandermonde_coding_matrix(int k, int m, int w);
-extern int *reed_sol_extended_vandermonde_matrix(int rows, int cols, int w);
-extern int *reed_sol_big_vandermonde_distribution_matrix(int rows, int cols, int w);
+extern int *reed_sol_vandermonde_coding_matrix(gf2_t* g, int k, int m, int w);
+extern int *reed_sol_extended_vandermonde_matrix(gf2_t* g, int rows, int cols, int w);
+extern int *reed_sol_big_vandermonde_distribution_matrix(gf2_t* g, int rows, int cols, int w);
 
-extern int reed_sol_r6_encode(int k, int w, char **data_ptrs, char **coding_ptrs, int size);
-extern int *reed_sol_r6_coding_matrix(int k, int w);
+extern int reed_sol_r6_encode(gf2_t* g, int k, int w, char **data_ptrs, char **coding_ptrs, int size);
+extern int *reed_sol_r6_coding_matrix(gf2_t* g, int k, int w);
 
-extern void reed_sol_galois_w08_region_multby_2(char *region, int nbytes);
-extern void reed_sol_galois_w16_region_multby_2(char *region, int nbytes);
-extern void reed_sol_galois_w32_region_multby_2(char *region, int nbytes);
+extern void reed_sol_galois_w08_region_multby_2(gf2_t* g, char *region, int nbytes);
+extern void reed_sol_galois_w16_region_multby_2(gf2_t* g, char *region, int nbytes);
+extern void reed_sol_galois_w32_region_multby_2(gf2_t* g, char *region, int nbytes);
 
 #ifdef __cplusplus
 }
diff --git a/src/cauchy.c b/src/cauchy.c
index f63dfb7..a5c859f 100644
--- a/src/cauchy.c
+++ b/src/cauchy.c
@@ -86,7 +86,7 @@ static int *cbest_all[33];
 
 #define talloc(type, num) (type *) malloc(sizeof(type)*(num))
 
-int cauchy_n_ones(int n, int w)
+int cauchy_n_ones(gf2_t* g, int n, int w)
 {
   int no;
   int cno;
@@ -98,7 +98,7 @@ int cauchy_n_ones(int n, int w)
 
   if (PPs[w] == -1) {
     nones = 0;
-    PPs[w] = galois_single_multiply(highbit, 2, w);
+    PPs[w] = galois_single_multiply(g, highbit, 2, w);
     for (i = 0; i < w; i++) {
       if (PPs[w] & (1 << i)) {
         ONEs[w][nones] = (1 << i);
@@ -128,7 +128,7 @@ int cauchy_n_ones(int n, int w)
   return no;
 }
   
-int *cauchy_original_coding_matrix(int k, int m, int w)
+int *cauchy_original_coding_matrix(gf2_t* g, int k, int m, int w)
 {
   int *matrix;
   int i, j, index;
@@ -139,14 +139,14 @@ int *cauchy_original_coding_matrix(int k, int m, int w)
   index = 0;
   for (i = 0; i < m; i++) {
     for (j = 0; j < k; j++) {
-      matrix[index] = galois_single_divide(1, (i ^ (m+j)), w);
+      matrix[index] = galois_single_divide(g, 1, (i ^ (m+j)), w);
       index++;
     }
   }
   return matrix;
 }
 
-int *cauchy_xy_coding_matrix(int k, int m, int w, int *X, int *Y)
+int *cauchy_xy_coding_matrix(gf2_t* g, int k, int m, int w, int *X, int *Y)
 {
   int index, i, j;
   int *matrix;
@@ -156,14 +156,14 @@ int *cauchy_xy_coding_matrix(int k, int m, int w, int *X, int *Y)
   index = 0;
   for (i = 0; i < m; i++) {
     for (j = 0; j < k; j++) {
-      matrix[index] = galois_single_divide(1, (X[i] ^ Y[j]), w);
+      matrix[index] = galois_single_divide(g, 1, (X[i] ^ Y[j]), w);
       index++;
     }
   }
   return matrix;
 }
 
-void cauchy_improve_coding_matrix(int k, int m, int w, int *matrix)
+void cauchy_improve_coding_matrix(gf2_t* g, int k, int m, int w, int *matrix)
 {
   int index, i, j, x;
   int tmp;
@@ -171,10 +171,10 @@ void cauchy_improve_coding_matrix(int k, int m, int w, int *matrix)
 
   for (j = 0; j < k; j++) {
     if (matrix[j] != 1) {
-      tmp = galois_single_divide(1, matrix[j], w);
+      tmp = galois_single_divide(g, 1, matrix[j], w);
       index = j;
       for (i = 0; i < m; i++) {
-        matrix[index] = galois_single_multiply(matrix[index], tmp, w);
+        matrix[index] = galois_single_multiply(g, matrix[index], tmp, w);
         index += k;
       }
     }
@@ -182,14 +182,14 @@ void cauchy_improve_coding_matrix(int k, int m, int w, int *matrix)
   for (i = 1; i < m; i++) {
     bno = 0;
     index = i*k;
-    for (j = 0; j < k; j++) bno += cauchy_n_ones(matrix[index+j], w);
+    for (j = 0; j < k; j++) bno += cauchy_n_ones(g, matrix[index+j], w);
     bno_index = -1;
     for (j = 0; j < k; j++) {
       if (matrix[index+j] != 1) {
-        tmp = galois_single_divide(1, matrix[index+j], w);
+        tmp = galois_single_divide(g, 1, matrix[index+j], w);
         tno = 0;
         for (x = 0; x < k; x++) {
-          tno += cauchy_n_ones(galois_single_multiply(matrix[index+x], tmp, w), w);
+          tno += cauchy_n_ones(g, galois_single_multiply(g, matrix[index+x], tmp, w), w);
         }
         if (tno < bno) {
           bno = tno;
@@ -198,22 +198,22 @@ void cauchy_improve_coding_matrix(int k, int m, int w, int *matrix)
       }
     }
     if (bno_index != -1) {
-      tmp = galois_single_divide(1, matrix[index+bno_index], w);
+      tmp = galois_single_divide(g, 1, matrix[index+bno_index], w);
       for (j = 0; j < k; j++) {
-        matrix[index+j] = galois_single_multiply(matrix[index+j], tmp, w);
+        matrix[index+j] = galois_single_multiply(g, matrix[index+j], tmp, w);
       }
     }
   }
 }
 
-int *cauchy_good_general_coding_matrix(int k, int m, int w)
+int *cauchy_good_general_coding_matrix(gf2_t* g, int k, int m, int w)
 {
   int *matrix, i;
 
   if (m == 2 && k <= cbest_max_k[w]) {
     matrix = talloc(int, k*m);
     if (matrix == NULL) return NULL;
-    if (!cbest_init) {
+    if (!cbest_init) { /* not safe for multithreads. need CAS */
       cbest_init = 1;
       cbest_all[0] = cbest_0; cbest_all[1] = cbest_1; cbest_all[2] = cbest_2; cbest_all[3] = cbest_3; cbest_all[4] =
       cbest_4; cbest_all[5] = cbest_5; cbest_all[6] = cbest_6; cbest_all[7] = cbest_7; cbest_all[8] = cbest_8;
@@ -230,9 +230,9 @@ int *cauchy_good_general_coding_matrix(int k, int m, int w)
     }
     return matrix;
   } else {
-    matrix = cauchy_original_coding_matrix(k, m, w);
+    matrix = cauchy_original_coding_matrix(g, k, m, w);
     if (matrix == NULL) return NULL;
-    cauchy_improve_coding_matrix(k, m, w, matrix);
+    cauchy_improve_coding_matrix(g, k, m, w, matrix);
     return matrix;
   }
 }
diff --git a/src/galois.c b/src/galois.c
index cd9faa8..b900b32 100644
--- a/src/galois.c
+++ b/src/galois.c
@@ -53,19 +53,34 @@
 #include "galois.h"
 
 #define MAX_GF_INSTANCES 64
-gf_t *gfp_array[MAX_GF_INSTANCES] = { 0 };
-int  gfp_is_composite[MAX_GF_INSTANCES] = { 0 };
 
-gf_t *galois_get_field_ptr(int w)
+gf_t *galois_get_field_ptr(gf2_t* g, int w)
 {
-  if (gfp_array[w] != NULL) {
-    return gfp_array[w];
+  if (g->gfp_array[w] != NULL) {
+    return g->gfp_array[w];
   }
 
   return NULL;
 }
 
-gf_t* galois_init_field(int w,
+gf2_t* galois_init_empty()
+{
+  gf2_t* ptr =  (gf2_t *) malloc(sizeof(gf2_t));
+  memset(ptr, sizeof(gf2_t), 0);
+  return ptr;
+}
+
+void galois_destroy(gf2_t* g)
+{
+	if (g) {
+		free(g->gf);
+		for (int i = 0; i < MAX_GF_INSTANCES; i ++)
+			free(galois_get_field_ptr(g, i));
+		free(g);
+	}
+}
+
+gf2_t* galois_init_field(int w,
                         int mult_type,
                         int region_type,
                         int divide_type,
@@ -75,15 +90,16 @@ gf_t* galois_init_field(int w,
 {
   int scratch_size;
   void *scratch_memory;
-  gf_t *gfp;
+  gf2_t *gf2p;
 
   if (w <= 0 || w > 32) {
     fprintf(stderr, "ERROR -- cannot init default Galois field for w=%d\n", w);
     assert(0);
   }
 
-  gfp = (gf_t *) malloc(sizeof(gf_t));
-  if (!gfp) {
+  gf2p = (gf2_t *) malloc(sizeof(gf2_t));
+  gf2p->gf = (gf_t *) malloc(sizeof(gf_t));
+  if (!gf2p || gf2p->gf) {
     fprintf(stderr, "ERROR -- cannot allocate memory for Galois field w=%d\n", w);
     assert(0);
   }
@@ -100,7 +116,7 @@ gf_t* galois_init_field(int w,
     assert(0);
   }
 
-  if(!gf_init_hard(gfp,
+  if(!gf_init_hard(gf2p->gf,
                    w, 
                    mult_type, 
                    region_type, 
@@ -115,27 +131,28 @@ gf_t* galois_init_field(int w,
     assert(0);
   }
 
-  gfp_is_composite[w] = 0;
-  return gfp;
+  gf2p->gfp_is_composite[w] = 0;
+  return gf2p;
 }
 
-gf_t* galois_init_composite_field(int w,
-                                int region_type,
-                                int divide_type,
-                                int degree,
-                                gf_t* base_gf)
+gf2_t* galois_init_composite_field(int w,
+								   int region_type,
+								   int divide_type,
+								   int degree,
+								   gf_t* base_gf)
 {
   int scratch_size;
   void *scratch_memory;
-  gf_t *gfp;
+  gf2_t *gf2p;
   
   if (w <= 0 || w > 32) {
     fprintf(stderr, "ERROR -- cannot init composite field for w=%d\n", w);
     assert(0);
   }
   
-  gfp = (gf_t *) malloc(sizeof(gf_t));
-  if (!gfp) {
+  gf2p = (gf2_t *) malloc(sizeof(gf2_t));
+  gf2p->gf = (gf_t *) malloc(sizeof(gf_t));
+  if (!gf2p || gf2p->gf) {
     fprintf(stderr, "ERROR -- cannot allocate memory for Galois field w=%d\n", w);
     assert(0);
   }
@@ -152,7 +169,7 @@ gf_t* galois_init_composite_field(int w,
     assert(0);
   }
 
-  if(!gf_init_hard(gfp,
+  if(!gf_init_hard(gf2p->gf,
                    w,
                    GF_MULT_COMPOSITE,
                    region_type,
@@ -166,42 +183,42 @@ gf_t* galois_init_composite_field(int w,
     fprintf(stderr, "ERROR -- cannot init default composite field for w=%d\n", w);
     assert(0);
   }
-  gfp_is_composite[w] = 1;
-  return gfp;
+  gf2p->gfp_is_composite[w] = 1;
+  return gf2p;
 }
 
-int galois_init_default_field(int w)
+int galois_init_default_field(gf2_t* g, int w)
 {
-  if (gfp_array[w] == NULL) {
-    gfp_array[w] = (gf_t*)malloc(sizeof(gf_t));
-    if(gfp_array[w] == NULL)
+  if (g->gfp_array[w] == NULL) {
+    g->gfp_array[w] = (gf_t*)malloc(sizeof(gf_t));
+    if(g->gfp_array[w] == NULL)
       return ENOMEM;
-    if (!gf_init_easy(gfp_array[w], w))
+    if (!gf_init_easy(g->gfp_array[w], w))
       return EINVAL;
   }
   return 0;
 }
 
-int galois_uninit_field(int w)
+int galois_uninit_field(gf2_t* g, int w)
 {
   int ret = 0;
-  if (gfp_array[w] != NULL) {
+  if (g->gfp_array[w] != NULL) {
     int recursive = 1;
-    ret = gf_free(gfp_array[w], recursive);
-    free(gfp_array[w]);
-    gfp_array[w] = NULL;
+    ret = gf_free(g->gfp_array[w], recursive);
+    free(g->gfp_array[w]);
+    g->gfp_array[w] = NULL;
   }
   return ret;
 }
 
-static void galois_init(int w)
+static void galois_init(gf2_t* g, int w)
 {
   if (w <= 0 || w > 32) {
     fprintf(stderr, "ERROR -- cannot init default Galois field for w=%d\n", w);
     assert(0);
   }
 
-  switch (galois_init_default_field(w)) {
+  switch (galois_init_default_field(g, w)) {
   case ENOMEM:
     fprintf(stderr, "ERROR -- cannot allocate memory for Galois field w=%d\n", w);
     assert(0);
@@ -243,7 +260,7 @@ static int is_valid_gf(gf_t *gf, int w)
   return 1;
 }
 
-void galois_change_technique(gf_t *gf, int w)
+void galois_change_technique(gf2_t *g, gf_t* gf, int w)
 {
   if (w <= 0 || w > 32) {
     fprintf(stderr, "ERROR -- cannot support Galois field for w=%d\n", w);
@@ -255,111 +272,114 @@ void galois_change_technique(gf_t *gf, int w)
     assert(0);
   }
 
-  if (gfp_array[w] != NULL) {
-    gf_free(gfp_array[w], gfp_is_composite[w]);
+  if (g->gfp_array[w] != NULL) {
+    gf_free(g->gfp_array[w], g->gfp_is_composite[w]);
   }
 
-  gfp_array[w] = gf;
+  g->gfp_array[w] = g;
 }
 
-int galois_single_multiply(int x, int y, int w)
+int galois_single_multiply(gf2_t* g, int x, int y, int w)
 {
   if (x == 0 || y == 0) return 0;
   
-  if (gfp_array[w] == NULL) {
-    galois_init(w);
+  if (g->gfp_array[w] == NULL) {
+	  galois_init(g, w);
   }
 
   if (w <= 32) {
-    return gfp_array[w]->multiply.w32(gfp_array[w], x, y);
+    return g->gfp_array[w]->multiply.w32(g->gfp_array[w], x, y);
   } else {
     fprintf(stderr, "ERROR -- Galois field not implemented for w=%d\n", w);
     return 0;
   }
 }
 
-int galois_single_divide(int x, int y, int w)
+int galois_single_divide(gf2_t* g, int x, int y, int w)
 {
   if (x == 0) return 0;
   if (y == 0) return -1;
 
-  if (gfp_array[w] == NULL) {
-    galois_init(w);
+  if (g->gfp_array[w] == NULL) {
+	  galois_init(g, w);
   }
 
   if (w <= 32) {
-    return gfp_array[w]->divide.w32(gfp_array[w], x, y);
+    return g->gfp_array[w]->divide.w32(g->gfp_array[w], x, y);
   } else {
     fprintf(stderr, "ERROR -- Galois field not implemented for w=%d\n", w);
     return 0;
   }
 }
 
-void galois_w08_region_multiply(char *region,      /* Region to multiply */
+void galois_w08_region_multiply(gf2_t* g,
+								  char *region,      /* Region to multiply */
                                   int multby,       /* Number to multiply by */
                                   int nbytes,        /* Number of bytes in region */
                                   char *r2,          /* If r2 != NULL, products go here */
                                   int add)
 {
-  if (gfp_array[8] == NULL) {
-    galois_init(8);
+  if (g->gfp_array[8] == NULL) {
+	  galois_init(g, 8);
   }
-  gfp_array[8]->multiply_region.w32(gfp_array[8], region, r2, multby, nbytes, add);
+  g->gfp_array[8]->multiply_region.w32(g->gfp_array[8], region, r2, multby, nbytes, add);
 }
 
-void galois_w16_region_multiply(char *region,      /* Region to multiply */
+void galois_w16_region_multiply(gf2_t* g,
+								char *region,      /* Region to multiply */
                                   int multby,       /* Number to multiply by */
                                   int nbytes,        /* Number of bytes in region */
                                   char *r2,          /* If r2 != NULL, products go here */
                                   int add)
 {
-  if (gfp_array[16] == NULL) {
-    galois_init(16);
+  if (g->gfp_array[16] == NULL) {
+	  galois_init(g, 16);
   }
-  gfp_array[16]->multiply_region.w32(gfp_array[16], region, r2, multby, nbytes, add);
+  g->gfp_array[16]->multiply_region.w32(g->gfp_array[16], region, r2, multby, nbytes, add);
 }
 
 
-void galois_w32_region_multiply(char *region,      /* Region to multiply */
+void galois_w32_region_multiply(gf2_t* g,
+								char *region,      /* Region to multiply */
                                   int multby,       /* Number to multiply by */
                                   int nbytes,        /* Number of bytes in region */
                                   char *r2,          /* If r2 != NULL, products go here */
                                   int add)
 {
-  if (gfp_array[32] == NULL) {
-    galois_init(32);
+  if (g->gfp_array[32] == NULL) {
+	  galois_init(g, 32);
   }
-  gfp_array[32]->multiply_region.w32(gfp_array[32], region, r2, multby, nbytes, add);
+  g->gfp_array[32]->multiply_region.w32(g->gfp_array[32], region, r2, multby, nbytes, add);
 }
 
-void galois_w8_region_xor(void *src, void *dest, int nbytes)
+void galois_w8_region_xor(gf2_t* g, void *src, void *dest, int nbytes)
 {
-  if (gfp_array[8] == NULL) {
-    galois_init(8);
+  if (g->gfp_array[8] == NULL) {
+	  galois_init(g, 8);
   }
-  gfp_array[8]->multiply_region.w32(gfp_array[32], src, dest, 1, nbytes, 1);
+  g->gfp_array[8]->multiply_region.w32(g->gfp_array[32], src, dest, 1, nbytes, 1);
 }
 
-void galois_w16_region_xor(void *src, void *dest, int nbytes)
+void galois_w16_region_xor(gf2_t* g, void *src, void *dest, int nbytes)
 {
-  if (gfp_array[16] == NULL) {
-    galois_init(16);
+  if (g->gfp_array[16] == NULL) {
+	  galois_init(g, 16);
   }
-  gfp_array[16]->multiply_region.w32(gfp_array[16], src, dest, 1, nbytes, 1);
+  g->gfp_array[16]->multiply_region.w32(g->gfp_array[16], src, dest, 1, nbytes, 1);
 }
 
-void galois_w32_region_xor(void *src, void *dest, int nbytes)
+void galois_w32_region_xor(gf2_t* g, void *src, void *dest, int nbytes)
 {
-  if (gfp_array[32] == NULL) {
-    galois_init(32);
+  if (g->gfp_array[32] == NULL) {
+	  galois_init(g, 32);
   }
-  gfp_array[32]->multiply_region.w32(gfp_array[32], src, dest, 1, nbytes, 1);
+  g->gfp_array[32]->multiply_region.w32(g->gfp_array[32], src, dest, 1, nbytes, 1);
 }
 
-void galois_region_xor(char *src, char *dest, int nbytes)
+void galois_region_xor(gf2_t* g, char *src, char *dest, int nbytes)
 {
   if (nbytes >= 16) {
-    galois_w32_region_xor(src, dest, nbytes);
+	  galois_w32_region_xor(g, src, dest, nbytes);
   } else {
     int i = 0;
     for (i = 0; i < nbytes; i++) {
@@ -370,8 +390,8 @@ void galois_region_xor(char *src, char *dest, int nbytes)
   }
 }
 
-int galois_inverse(int y, int w)
+int galois_inverse(gf2_t* g, int y, int w)
 {
   if (y == 0) return -1;
-  return galois_single_divide(1, y, w);
+  return galois_single_divide(g, 1, y, w);
 }
diff --git a/src/jerasure.c b/src/jerasure.c
index 6874a8a..8ef5b25 100644
--- a/src/jerasure.c
+++ b/src/jerasure.c
@@ -96,7 +96,7 @@ void jerasure_print_bitmatrix(int *m, int rows, int cols, int w)
   }
 }
 
-int jerasure_make_decoding_matrix(int k, int m, int w, int *matrix, int *erased, int *decoding_matrix, int *dm_ids)
+int jerasure_make_decoding_matrix(gf2_t* g, int k, int m, int w, int *matrix, int *erased, int *decoding_matrix, int *dm_ids)
 {
   int i, j, *tmpmat;
 
@@ -121,7 +121,7 @@ int jerasure_make_decoding_matrix(int k, int m, int w, int *matrix, int *erased,
     }
   }
 
-  i = jerasure_invert_matrix(tmpmat, decoding_matrix, k, w);
+  i = jerasure_invert_matrix(g, tmpmat, decoding_matrix, k, w);
   free(tmpmat);
   return i;
 }
@@ -165,7 +165,7 @@ int jerasure_make_decoding_bitmatrix(int k, int m, int w, int *matrix, int *eras
   return i;
 }
 
-int jerasure_matrix_decode(int k, int m, int w, int *matrix, int row_k_ones, int *erasures,
+int jerasure_matrix_decode(gf2_t* g, int k, int m, int w, int *matrix, int row_k_ones, int *erasures,
                           char **data_ptrs, char **coding_ptrs, int size)
 {
   int i, edd, lastdrive;
@@ -221,7 +221,7 @@ int jerasure_matrix_decode(int k, int m, int w, int *matrix, int row_k_ones, int
       return -1;
     }
 
-    if (jerasure_make_decoding_matrix(k, m, w, matrix, erased, decoding_matrix, dm_ids) < 0) {
+    if (jerasure_make_decoding_matrix(g, k, m, w, matrix, erased, decoding_matrix, dm_ids) < 0) {
       free(erased);
       free(dm_ids);
       free(decoding_matrix);
@@ -237,7 +237,7 @@ int jerasure_matrix_decode(int k, int m, int w, int *matrix, int row_k_ones, int
 
   for (i = 0; edd > 0 && i < lastdrive; i++) {
     if (erased[i]) {
-      jerasure_matrix_dotprod(k, w, decoding_matrix+(i*k), dm_ids, i, data_ptrs, coding_ptrs, size);
+      jerasure_matrix_dotprod(g, k, w, decoding_matrix+(i*k), dm_ids, i, data_ptrs, coding_ptrs, size);
       edd--;
     }
   }
@@ -255,7 +255,7 @@ int jerasure_matrix_decode(int k, int m, int w, int *matrix, int row_k_ones, int
     for (i = 0; i < k; i++) {
       tmpids[i] = (i < lastdrive) ? i : i+1;
     }
-    jerasure_matrix_dotprod(k, w, matrix, tmpids, lastdrive, data_ptrs, coding_ptrs, size);
+    jerasure_matrix_dotprod(g, k, w, matrix, tmpids, lastdrive, data_ptrs, coding_ptrs, size);
     free(tmpids);
   }
   
@@ -263,7 +263,7 @@ int jerasure_matrix_decode(int k, int m, int w, int *matrix, int row_k_ones, int
 
   for (i = 0; i < m; i++) {
     if (erased[k+i]) {
-      jerasure_matrix_dotprod(k, w, matrix+(i*k), NULL, i+k, data_ptrs, coding_ptrs, size);
+      jerasure_matrix_dotprod(g, k, w, matrix+(i*k), NULL, i+k, data_ptrs, coding_ptrs, size);
     }
   }
 
@@ -275,7 +275,7 @@ int jerasure_matrix_decode(int k, int m, int w, int *matrix, int row_k_ones, int
 }
 
 
-int *jerasure_matrix_to_bitmatrix(int k, int m, int w, int *matrix) 
+int *jerasure_matrix_to_bitmatrix(gf2_t* g, int k, int m, int w, int *matrix) 
 {
   int *bitmatrix;
   int rowelts, rowindex, colindex, elt, i, j, l, x;
@@ -296,7 +296,7 @@ int *jerasure_matrix_to_bitmatrix(int k, int m, int w, int *matrix)
         for (l = 0; l < w; l++) {
           bitmatrix[colindex+x+l*rowelts] = ((elt & (1 << l)) ? 1 : 0);
         }
-        elt = galois_single_multiply(elt, 2, w);
+        elt = galois_single_multiply(g, elt, 2, w);
       }
       colindex += w;
     }
@@ -305,7 +305,7 @@ int *jerasure_matrix_to_bitmatrix(int k, int m, int w, int *matrix)
   return bitmatrix;
 }
 
-void jerasure_matrix_encode(int k, int m, int w, int *matrix,
+void jerasure_matrix_encode(gf2_t* g, int k, int m, int w, int *matrix,
                           char **data_ptrs, char **coding_ptrs, int size)
 {
   int i;
@@ -316,11 +316,11 @@ void jerasure_matrix_encode(int k, int m, int w, int *matrix,
   }
 
   for (i = 0; i < m; i++) {
-    jerasure_matrix_dotprod(k, w, matrix+(i*k), NULL, k+i, data_ptrs, coding_ptrs, size);
+    jerasure_matrix_dotprod(g, k, w, matrix+(i*k), NULL, k+i, data_ptrs, coding_ptrs, size);
   }
 }
 
-void jerasure_bitmatrix_dotprod(int k, int w, int *bitmatrix_row,
+void jerasure_bitmatrix_dotprod(gf2_t* g, int k, int w, int *bitmatrix_row,
                              int *src_ids, int dest_id,
                              char **data_ptrs, char **coding_ptrs, int size, int packetsize)
 {
@@ -355,7 +355,7 @@ void jerasure_bitmatrix_dotprod(int k, int w, int *bitmatrix_row,
               jerasure_total_memcpy_bytes += packetsize;
               pstarted = 1;
             } else {
-              galois_region_xor(dptr, pptr, packetsize);
+              galois_region_xor(g, dptr, pptr, packetsize);
               jerasure_total_xor_bytes += packetsize;
             }
           }
@@ -366,7 +366,7 @@ void jerasure_bitmatrix_dotprod(int k, int w, int *bitmatrix_row,
   }
 }
 
-void jerasure_do_parity(int k, char **data_ptrs, char *parity_ptr, int size) 
+void jerasure_do_parity(gf2_t* g, int k, char **data_ptrs, char *parity_ptr, int size) 
 {
   int i;
 
@@ -374,12 +374,12 @@ void jerasure_do_parity(int k, char **data_ptrs, char *parity_ptr, int size)
   jerasure_total_memcpy_bytes += size;
   
   for (i = 1; i < k; i++) {
-    galois_region_xor(data_ptrs[i], parity_ptr, size);
+    galois_region_xor(g, data_ptrs[i], parity_ptr, size);
     jerasure_total_xor_bytes += size;
   }
 }
 
-int jerasure_invert_matrix(int *mat, int *inv, int rows, int w)
+int jerasure_invert_matrix(gf2_t* g, int *mat, int *inv, int rows, int w)
 {
   int cols, i, j, k, x, rs2;
   int row_start, tmp, inverse;
@@ -418,10 +418,10 @@ int jerasure_invert_matrix(int *mat, int *inv, int rows, int w)
     /* Multiply the row by 1/element i,i  */
     tmp = mat[row_start+i];
     if (tmp != 1) {
-      inverse = galois_single_divide(1, tmp, w);
+      inverse = galois_single_divide(g, 1, tmp, w);
       for (j = 0; j < cols; j++) { 
-        mat[row_start+j] = galois_single_multiply(mat[row_start+j], inverse, w);
-        inv[row_start+j] = galois_single_multiply(inv[row_start+j], inverse, w);
+        mat[row_start+j] = galois_single_multiply(g, mat[row_start+j], inverse, w);
+        inv[row_start+j] = galois_single_multiply(g, inv[row_start+j], inverse, w);
       }
     }
 
@@ -440,8 +440,8 @@ int jerasure_invert_matrix(int *mat, int *inv, int rows, int w)
           tmp = mat[k];
           rs2 = cols*j;
           for (x = 0; x < cols; x++) {
-            mat[rs2+x] ^= galois_single_multiply(tmp, mat[row_start+x], w);
-            inv[rs2+x] ^= galois_single_multiply(tmp, inv[row_start+x], w);
+            mat[rs2+x] ^= galois_single_multiply(g, tmp, mat[row_start+x], w);
+            inv[rs2+x] ^= galois_single_multiply(g, tmp, inv[row_start+x], w);
           }
         }
       }
@@ -458,7 +458,7 @@ int jerasure_invert_matrix(int *mat, int *inv, int rows, int w)
         tmp = mat[rs2+i];
         mat[rs2+i] = 0; 
         for (k = 0; k < cols; k++) {
-          inv[rs2+k] ^= galois_single_multiply(tmp, inv[row_start+k], w);
+          inv[rs2+k] ^= galois_single_multiply(g, tmp, inv[row_start+k], w);
         }
       }
     }
@@ -466,7 +466,7 @@ int jerasure_invert_matrix(int *mat, int *inv, int rows, int w)
   return 0;
 }
 
-int jerasure_invertible_matrix(int *mat, int rows, int w)
+int jerasure_invertible_matrix(gf2_t* g, int *mat, int rows, int w)
 {
   int cols, i, j, k, x, rs2;
   int row_start, tmp, inverse;
@@ -494,9 +494,9 @@ int jerasure_invertible_matrix(int *mat, int rows, int w)
     /* Multiply the row by 1/element i,i  */
     tmp = mat[row_start+i];
     if (tmp != 1) {
-      inverse = galois_single_divide(1, tmp, w);
+      inverse = galois_single_divide(g, 1, tmp, w);
       for (j = 0; j < cols; j++) { 
-        mat[row_start+j] = galois_single_multiply(mat[row_start+j], inverse, w);
+        mat[row_start+j] = galois_single_multiply(g, mat[row_start+j], inverse, w);
       }
     }
 
@@ -514,7 +514,7 @@ int jerasure_invertible_matrix(int *mat, int rows, int w)
           tmp = mat[k];
           rs2 = cols*j;
           for (x = 0; x < cols; x++) {
-            mat[rs2+x] ^= galois_single_multiply(tmp, mat[row_start+x], w);
+            mat[rs2+x] ^= galois_single_multiply(g, tmp, mat[row_start+x], w);
           }
         }
       }
@@ -580,7 +580,7 @@ void jerasure_free_schedule_cache(int k, int m, int ***cache)
   free(cache);
 }
 
-void jerasure_matrix_dotprod(int k, int w, int *matrix_row,
+void jerasure_matrix_dotprod(gf2_t* g, int k, int w, int *matrix_row,
                           int *src_ids, int dest_id,
                           char **data_ptrs, char **coding_ptrs, int size)
 {
@@ -613,7 +613,7 @@ void jerasure_matrix_dotprod(int k, int w, int *matrix_row,
         jerasure_total_memcpy_bytes += size;
         init = 1;
       } else {
-        galois_region_xor(sptr, dptr, size);
+        galois_region_xor(g, sptr, dptr, size);
         jerasure_total_xor_bytes += size;
       }
     }
@@ -631,9 +631,9 @@ void jerasure_matrix_dotprod(int k, int w, int *matrix_row,
         sptr = coding_ptrs[src_ids[i]-k];
       }
       switch (w) {
-        case 8:  galois_w08_region_multiply(sptr, matrix_row[i], size, dptr, init); break;
-        case 16: galois_w16_region_multiply(sptr, matrix_row[i], size, dptr, init); break;
-        case 32: galois_w32_region_multiply(sptr, matrix_row[i], size, dptr, init); break;
+	  case 8:  galois_w08_region_multiply(g, sptr, matrix_row[i], size, dptr, init); break;
+	  case 16: galois_w16_region_multiply(g, sptr, matrix_row[i], size, dptr, init); break;
+	  case 32: galois_w32_region_multiply(g, sptr, matrix_row[i], size, dptr, init); break;
       }
       jerasure_total_gf_bytes += size;
       init = 1;
@@ -642,7 +642,7 @@ void jerasure_matrix_dotprod(int k, int w, int *matrix_row,
 }
 
 
-int jerasure_bitmatrix_decode(int k, int m, int w, int *bitmatrix, int row_k_ones, int *erasures,
+int jerasure_bitmatrix_decode(gf2_t* g, int k, int m, int w, int *bitmatrix, int row_k_ones, int *erasures,
                             char **data_ptrs, char **coding_ptrs, int size, int packetsize)
 {
   int i;
@@ -697,7 +697,7 @@ int jerasure_bitmatrix_decode(int k, int m, int w, int *bitmatrix, int row_k_one
 
   for (i = 0; edd > 0 && i < lastdrive; i++) {
     if (erased[i]) {
-      jerasure_bitmatrix_dotprod(k, w, decoding_matrix+i*k*w*w, dm_ids, i, data_ptrs, coding_ptrs, size, packetsize);
+      jerasure_bitmatrix_dotprod(g, k, w, decoding_matrix+i*k*w*w, dm_ids, i, data_ptrs, coding_ptrs, size, packetsize);
       edd--;
     }
   }
@@ -713,13 +713,13 @@ int jerasure_bitmatrix_decode(int k, int m, int w, int *bitmatrix, int row_k_one
     for (i = 0; i < k; i++) {
       tmpids[i] = (i < lastdrive) ? i : i+1;
     }
-    jerasure_bitmatrix_dotprod(k, w, bitmatrix, tmpids, lastdrive, data_ptrs, coding_ptrs, size, packetsize);
+    jerasure_bitmatrix_dotprod(g, k, w, bitmatrix, tmpids, lastdrive, data_ptrs, coding_ptrs, size, packetsize);
     free(tmpids);
   }
 
   for (i = 0; i < m; i++) {
     if (erased[k+i]) {
-      jerasure_bitmatrix_dotprod(k, w, bitmatrix+i*k*w*w, NULL, k+i, data_ptrs, coding_ptrs, size, packetsize);
+      jerasure_bitmatrix_dotprod(g, k, w, bitmatrix+i*k*w*w, NULL, k+i, data_ptrs, coding_ptrs, size, packetsize);
     }
   }
 
@@ -988,7 +988,7 @@ static int **jerasure_generate_decoding_schedule(int k, int m, int w, int *bitma
   return schedule;
 }
 
-int jerasure_schedule_decode_lazy(int k, int m, int w, int *bitmatrix, int *erasures,
+int jerasure_schedule_decode_lazy(gf2_t* g, int k, int m, int w, int *bitmatrix, int *erasures,
                             char **data_ptrs, char **coding_ptrs, int size, int packetsize, 
                             int smart)
 {
@@ -1006,7 +1006,7 @@ int jerasure_schedule_decode_lazy(int k, int m, int w, int *bitmatrix, int *eras
   }
 
   for (tdone = 0; tdone < size; tdone += packetsize*w) {
-  jerasure_do_scheduled_operations(ptrs, schedule, packetsize);
+    jerasure_do_scheduled_operations(g, ptrs, schedule, packetsize);
     for (i = 0; i < k+m; i++) ptrs[i] += (packetsize*w);
   }
 
@@ -1016,7 +1016,7 @@ int jerasure_schedule_decode_lazy(int k, int m, int w, int *bitmatrix, int *eras
   return 0;
 }
 
-int jerasure_schedule_decode_cache(int k, int m, int w, int ***scache, int *erasures,
+int jerasure_schedule_decode_cache(gf2_t* g, int k, int m, int w, int ***scache, int *erasures,
                             char **data_ptrs, char **coding_ptrs, int size, int packetsize)
 {
   int i, tdone;
@@ -1039,7 +1039,7 @@ int jerasure_schedule_decode_cache(int k, int m, int w, int ***scache, int *eras
 
 
   for (tdone = 0; tdone < size; tdone += packetsize*w) {
-  jerasure_do_scheduled_operations(ptrs, schedule, packetsize);
+    jerasure_do_scheduled_operations(g, ptrs, schedule, packetsize);
     for (i = 0; i < k+m; i++) ptrs[i] += (packetsize*w);
   }
 
@@ -1176,7 +1176,7 @@ int jerasure_invertible_bitmatrix(int *mat, int rows)
 }
 
   
-int *jerasure_matrix_multiply(int *m1, int *m2, int r1, int c1, int r2, int c2, int w)
+int *jerasure_matrix_multiply(gf2_t* g, int *m1, int *m2, int r1, int c1, int r2, int c2, int w)
 {
   int *product, i, j, k;
 
@@ -1186,7 +1186,7 @@ int *jerasure_matrix_multiply(int *m1, int *m2, int r1, int c1, int r2, int c2,
   for (i = 0; i < r1; i++) {
     for (j = 0; j < c2; j++) {
       for (k = 0; k < r2; k++) {
-        product[i*c2+j] ^= galois_single_multiply(m1[i*c1+k], m2[k*c2+j], w);
+        product[i*c2+j] ^= galois_single_multiply(g, m1[i*c1+k], m2[k*c2+j], w);
       }
     }
   }
@@ -1203,7 +1203,7 @@ void jerasure_get_stats(double *fill_in)
   jerasure_total_memcpy_bytes = 0;
 }
 
-void jerasure_do_scheduled_operations(char **ptrs, int **operations, int packetsize)
+void jerasure_do_scheduled_operations(gf2_t* g, char **ptrs, int **operations, int packetsize)
 {
   char *sptr;
   char *dptr;
@@ -1218,7 +1218,7 @@ void jerasure_do_scheduled_operations(char **ptrs, int **operations, int packets
       operations[op][2], 
       operations[op][3]); 
       printf("xor(0x%x, 0x%x -> 0x%x, %d)\n", sptr, dptr, dptr, packetsize); */
-      galois_region_xor(sptr, dptr, packetsize);
+      galois_region_xor(g, sptr, dptr, packetsize);
       jerasure_total_xor_bytes += packetsize;
     } else {
 /*      printf("memcpy(0x%x <- 0x%x)\n", dptr, sptr); */
@@ -1228,7 +1228,7 @@ void jerasure_do_scheduled_operations(char **ptrs, int **operations, int packets
   }  
 }
 
-void jerasure_schedule_encode(int k, int m, int w, int **schedule,
+void jerasure_schedule_encode(gf2_t* g, int k, int m, int w, int **schedule,
                                    char **data_ptrs, char **coding_ptrs, int size, int packetsize)
 {
   char **ptr_copy;
@@ -1238,7 +1238,7 @@ void jerasure_schedule_encode(int k, int m, int w, int **schedule,
   for (i = 0; i < k; i++) ptr_copy[i] = data_ptrs[i];
   for (i = 0; i < m; i++) ptr_copy[i+k] = coding_ptrs[i];
   for (tdone = 0; tdone < size; tdone += packetsize*w) {
-    jerasure_do_scheduled_operations(ptr_copy, schedule, packetsize);
+    jerasure_do_scheduled_operations(g, ptr_copy, schedule, packetsize);
     for (i = 0; i < k+m; i++) ptr_copy[i] += (packetsize*w);
   }
   free(ptr_copy);
@@ -1450,7 +1450,7 @@ int **jerasure_smart_bitmatrix_to_schedule(int k, int m, int w, int *bitmatrix)
   return NULL;
 }
 
-void jerasure_bitmatrix_encode(int k, int m, int w, int *bitmatrix,
+void jerasure_bitmatrix_encode(gf2_t* g, int k, int m, int w, int *bitmatrix,
                             char **data_ptrs, char **coding_ptrs, int size, int packetsize)
 {
   int i;
@@ -1466,7 +1466,7 @@ void jerasure_bitmatrix_encode(int k, int m, int w, int *bitmatrix,
   }
 
   for (i = 0; i < m; i++) {
-    jerasure_bitmatrix_dotprod(k, w, bitmatrix+i*k*w*w, NULL, k+i, data_ptrs, coding_ptrs, size, packetsize);
+    jerasure_bitmatrix_dotprod(g, k, w, bitmatrix+i*k*w*w, NULL, k+i, data_ptrs, coding_ptrs, size, packetsize);
   }
 }
 
@@ -1476,7 +1476,9 @@ void jerasure_bitmatrix_encode(int k, int m, int w, int *bitmatrix,
  */
 int jerasure_autoconf_test()
 {
-  int x = galois_single_multiply(1, 2, 8);
+  gf2_t* g = galois_init_empty();
+  int x = galois_single_multiply(g, 1, 2, 8);
+  galois_destroy(g);
   if (x != 2) {
     return -1;
   }
diff --git a/src/reed_sol.c b/src/reed_sol.c
index 72cbff8..8fe4ca6 100644
--- a/src/reed_sol.c
+++ b/src/reed_sol.c
@@ -56,7 +56,7 @@
 
 #define talloc(type, num) (type *) malloc(sizeof(type)*(num))
 
-int *reed_sol_r6_coding_matrix(int k, int w)
+int *reed_sol_r6_coding_matrix(gf2_t* g, int k, int w)
 {
   int *matrix;
   int i, tmp;
@@ -70,18 +70,18 @@ int *reed_sol_r6_coding_matrix(int k, int w)
   matrix[k] = 1;
   tmp = 1;
   for (i = 1; i < k; i++) {
-    tmp = galois_single_multiply(tmp, 2, w);
+    tmp = galois_single_multiply(g, tmp, 2, w);
     matrix[k+i] = tmp;
   }
   return matrix;
 }
 
-int *reed_sol_vandermonde_coding_matrix(int k, int m, int w)
+int *reed_sol_vandermonde_coding_matrix(gf2_t* g, int k, int m, int w)
 {
   int i, j;
   int *vdm, *dist;
 
-  vdm = reed_sol_big_vandermonde_distribution_matrix(k+m, k, w);
+  vdm = reed_sol_big_vandermonde_distribution_matrix(g, k+m, k, w);
   if (vdm == NULL) return NULL;
   dist = talloc(int, m*k);
   if (dist == NULL) {
@@ -101,10 +101,10 @@ int *reed_sol_vandermonde_coding_matrix(int k, int m, int w)
 static int prim08 = -1;
 static gf_t GF08;
 
-void reed_sol_galois_w08_region_multby_2(char *region, int nbytes)
+void reed_sol_galois_w08_region_multby_2(gf2_t* g, char *region, int nbytes)
 {
   if (prim08 == -1) {
-    prim08 = galois_single_multiply((1 << 7), 2, 8);
+    prim08 = galois_single_multiply(g, (1 << 7), 2, 8);
     if (!gf_init_hard(&GF08, 8, GF_MULT_BYTWO_b, GF_REGION_DEFAULT, GF_DIVIDE_DEFAULT,
                       prim08, 0, 0, NULL, NULL)) {
       fprintf(stderr, "Error: Can't initialize the GF for reed_sol_galois_w08_region_multby_2\n");
@@ -117,10 +117,10 @@ void reed_sol_galois_w08_region_multby_2(char *region, int nbytes)
 static int prim16 = -1;
 static gf_t GF16;
 
-void reed_sol_galois_w16_region_multby_2(char *region, int nbytes)
+void reed_sol_galois_w16_region_multby_2(gf2_t* g, char *region, int nbytes)
 {
   if (prim16 == -1) {
-    prim16 = galois_single_multiply((1 << 15), 2, 16);
+    prim16 = galois_single_multiply(g, (1 << 15), 2, 16);
     if (!gf_init_hard(&GF16, 16, GF_MULT_BYTWO_b, GF_REGION_DEFAULT, GF_DIVIDE_DEFAULT,
                       prim16, 0, 0, NULL, NULL)) {
       fprintf(stderr, "Error: Can't initialize the GF for reed_sol_galois_w16_region_multby_2\n");
@@ -133,10 +133,10 @@ void reed_sol_galois_w16_region_multby_2(char *region, int nbytes)
 static int prim32 = -1;
 static gf_t GF32;
 
-void reed_sol_galois_w32_region_multby_2(char *region, int nbytes)
+void reed_sol_galois_w32_region_multby_2(gf2_t* g, char *region, int nbytes)
 {
   if (prim32 == -1) {
-    prim32 = galois_single_multiply(((gf_val_32_t)1 << 31), 2, 32);
+    prim32 = galois_single_multiply(g, ((gf_val_32_t)1 << 31), 2, 32);
     if (!gf_init_hard(&GF32, 32, GF_MULT_BYTWO_b, GF_REGION_DEFAULT, GF_DIVIDE_DEFAULT,
                       prim32, 0, 0, NULL, NULL)) {
       fprintf(stderr, "Error: Can't initialize the GF for reed_sol_galois_w32_region_multby_2\n");
@@ -146,7 +146,7 @@ void reed_sol_galois_w32_region_multby_2(char *region, int nbytes)
   GF32.multiply_region.w32(&GF32, region, region, 2, nbytes, 0);
 }
 
-int reed_sol_r6_encode(int k, int w, char **data_ptrs, char **coding_ptrs, int size)
+int reed_sol_r6_encode(gf2_t* g, int k, int w, char **data_ptrs, char **coding_ptrs, int size)
 {
   int i;
 
@@ -154,26 +154,27 @@ int reed_sol_r6_encode(int k, int w, char **data_ptrs, char **coding_ptrs, int s
 
   memcpy(coding_ptrs[0], data_ptrs[0], size);
 
-  for (i = 1; i < k; i++) galois_region_xor(data_ptrs[i], coding_ptrs[0], size);
+  for (i = 1; i < k; i++) galois_region_xor(g, data_ptrs[i], coding_ptrs[0], size);
 
   /* Next, put the sum of (2^j)*Dj into coding region 1 */
 
   memcpy(coding_ptrs[1], data_ptrs[k-1], size);
 
+  /* reed_sol_galois_w**_region_multby_2()s are not safe for multithread even if it has the context(gf2_t) */
   for (i = k-2; i >= 0; i--) {
     switch (w) {
-      case 8:  reed_sol_galois_w08_region_multby_2(coding_ptrs[1], size); break;
-      case 16: reed_sol_galois_w16_region_multby_2(coding_ptrs[1], size); break;
-      case 32: reed_sol_galois_w32_region_multby_2(coding_ptrs[1], size); break;
+      case 8:  reed_sol_galois_w08_region_multby_2(g, coding_ptrs[1], size); break;
+      case 16: reed_sol_galois_w16_region_multby_2(g, coding_ptrs[1], size); break;
+      case 32: reed_sol_galois_w32_region_multby_2(g, coding_ptrs[1], size); break;
       default: return 0;
     }
 
-    galois_region_xor(data_ptrs[i], coding_ptrs[1], size);
+    galois_region_xor(g, data_ptrs[i], coding_ptrs[1], size);
   }
   return 1;
 }
 
-int *reed_sol_extended_vandermonde_matrix(int rows, int cols, int w)
+int *reed_sol_extended_vandermonde_matrix(gf2_t* g, int rows, int cols, int w)
 {
   int *vdm;
   int i, j, k;
@@ -197,13 +198,13 @@ int *reed_sol_extended_vandermonde_matrix(int rows, int cols, int w)
     k = 1;
     for (j = 0; j < cols; j++) {
       vdm[i*cols+j] = k;
-      k = galois_single_multiply(k, i, w);
+      k = galois_single_multiply(g, k, i, w);
     }
   }
   return vdm;
 }
 
-int *reed_sol_big_vandermonde_distribution_matrix(int rows, int cols, int w)
+int *reed_sol_big_vandermonde_distribution_matrix(gf2_t* g, int rows, int cols, int w)
 {
   int *dist;
   int i, j, k;
@@ -211,7 +212,7 @@ int *reed_sol_big_vandermonde_distribution_matrix(int rows, int cols, int w)
 
   if (cols >= rows) return NULL;
   
-  dist = reed_sol_extended_vandermonde_matrix(rows, cols, w);
+  dist = reed_sol_extended_vandermonde_matrix(g, rows, cols, w);
   if (dist == NULL) return NULL;
 
   sindex = 0;
@@ -240,10 +241,10 @@ int *reed_sol_big_vandermonde_distribution_matrix(int rows, int cols, int w)
     /* If Element i,i is not equal to 1, multiply the column by 1/i */
 
     if (dist[sindex+i] != 1) {
-      tmp = galois_single_divide(1, dist[sindex+i], w);
+      tmp = galois_single_divide(g, 1, dist[sindex+i], w);
       srindex = i;
       for (j = 0; j < rows; j++) {
-        dist[srindex] = galois_single_multiply(tmp, dist[srindex], w);
+        dist[srindex] = galois_single_multiply(g, tmp, dist[srindex], w);
         srindex += cols;
       }
     }
@@ -260,7 +261,7 @@ int *reed_sol_big_vandermonde_distribution_matrix(int rows, int cols, int w)
         srindex = j;
         siindex = i;
         for (k = 0; k < rows; k++) {
-          dist[srindex] = dist[srindex] ^ galois_single_multiply(tmp, dist[siindex], w);
+          dist[srindex] = dist[srindex] ^ galois_single_multiply(g, tmp, dist[siindex], w);
           srindex += cols;
           siindex += cols;
         }
@@ -274,10 +275,10 @@ int *reed_sol_big_vandermonde_distribution_matrix(int rows, int cols, int w)
   for (j = 0; j < cols; j++) {
     tmp = dist[sindex];
     if (tmp != 1) { 
-      tmp = galois_single_divide(1, tmp, w);
+      tmp = galois_single_divide(g, 1, tmp, w);
       srindex = sindex;
       for (i = cols; i < rows; i++) {
-        dist[srindex] = galois_single_multiply(tmp, dist[srindex], w);
+        dist[srindex] = galois_single_multiply(g, tmp, dist[srindex], w);
         srindex += cols;
       }
     }
@@ -291,8 +292,8 @@ int *reed_sol_big_vandermonde_distribution_matrix(int rows, int cols, int w)
   for (i = cols+1; i < rows; i++) {
     tmp = dist[sindex];
     if (tmp != 1) { 
-      tmp = galois_single_divide(1, tmp, w);
-      for (j = 0; j < cols; j++) dist[sindex+j] = galois_single_multiply(dist[sindex+j], tmp, w);
+      tmp = galois_single_divide(g, 1, tmp, w);
+      for (j = 0; j < cols; j++) dist[sindex+j] = galois_single_multiply(g, dist[sindex+j], tmp, w);
     }
     sindex += cols;
   }

From 084cbd45c7926c0a45b07bbdfd3391045d9ffde5 Mon Sep 17 00:00:00 2001
From: roentgen <fukasawa666@gmail.com>
Date: Fri, 16 Nov 2018 20:19:57 +0900
Subject: [PATCH 2/6] fixed SEGV

---
 src/galois.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/galois.c b/src/galois.c
index b900b32..2b71523 100644
--- a/src/galois.c
+++ b/src/galois.c
@@ -66,7 +66,7 @@ gf_t *galois_get_field_ptr(gf2_t* g, int w)
 gf2_t* galois_init_empty()
 {
   gf2_t* ptr =  (gf2_t *) malloc(sizeof(gf2_t));
-  memset(ptr, sizeof(gf2_t), 0);
+  memset(ptr, 0, sizeof(gf2_t));
   return ptr;
 }
 

From 5cca7df46ba83459fe9c7462d57b06af670838a2 Mon Sep 17 00:00:00 2001
From: yuuya uezato <yuuya_uezato@dwango.co.jp>
Date: Mon, 19 Nov 2018 19:25:03 +0900
Subject: [PATCH 3/6] fix a misstake in the galois_change_technique function

---
 src/galois.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/galois.c b/src/galois.c
index 2b71523..1d530ab 100644
--- a/src/galois.c
+++ b/src/galois.c
@@ -276,7 +276,7 @@ void galois_change_technique(gf2_t *g, gf_t* gf, int w)
     gf_free(g->gfp_array[w], g->gfp_is_composite[w]);
   }
 
-  g->gfp_array[w] = g;
+  g->gfp_array[w] = gf;
 }
 
 int galois_single_multiply(gf2_t* g, int x, int y, int w)

From a53f2e2ba3b9a3861ef2aec754c670a5f6e0820c Mon Sep 17 00:00:00 2001
From: yuuya uezato <yuuya_uezato@dwango.co.jp>
Date: Mon, 19 Nov 2018 19:45:45 +0900
Subject: [PATCH 4/6] bypass the C99 feature, loop initial declarations

---
 src/galois.c | 13 +++++++------
 1 file changed, 7 insertions(+), 6 deletions(-)

diff --git a/src/galois.c b/src/galois.c
index 1d530ab..d35751f 100644
--- a/src/galois.c
+++ b/src/galois.c
@@ -72,12 +72,13 @@ gf2_t* galois_init_empty()
 
 void galois_destroy(gf2_t* g)
 {
-	if (g) {
-		free(g->gf);
-		for (int i = 0; i < MAX_GF_INSTANCES; i ++)
-			free(galois_get_field_ptr(g, i));
-		free(g);
-	}
+  if (g) {
+    free(g->gf);
+    int i;
+    for (i = 0; i < MAX_GF_INSTANCES; i++)
+      free(galois_get_field_ptr(g, i));
+    free(g);
+  }
 }
 
 gf2_t* galois_init_field(int w,

From 77c90fd5cc051182c2bc04828a3ed4ac84e45e7c Mon Sep 17 00:00:00 2001
From: yuuya uezato <yuuya_uezato@dwango.co.jp>
Date: Fri, 30 Nov 2018 20:02:24 +0900
Subject: [PATCH 5/6] rename libJerasure.* -> libFrugalosJerasure.*

---
 Examples/Makefile.am | 2 +-
 src/Makefile.am      | 8 ++++----
 2 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/Examples/Makefile.am b/Examples/Makefile.am
index defd139..61ba7d5 100644
--- a/Examples/Makefile.am
+++ b/Examples/Makefile.am
@@ -61,7 +61,7 @@ liberation_01_SOURCES = liberation_01.c
 decoder_SOURCES = decoder.c
 encoder_SOURCES = encoder.c
 
-LDADD = ../src/libJerasure.la
+LDADD = ../src/libFrugalosJerasure.la
 decoder_LDADD = $(LDADD) ../src/libtiming.a
 encoder_LDADD = $(LDADD) ../src/libtiming.a
 reed_sol_time_gf_LDADD = $(LDADD) ../src/libtiming.a
diff --git a/src/Makefile.am b/src/Makefile.am
index 070f545..8a7ce22 100644
--- a/src/Makefile.am
+++ b/src/Makefile.am
@@ -3,10 +3,10 @@
 AM_CPPFLAGS = -I$(top_srcdir)/include
 AM_CFLAGS = $(SIMD_FLAGS)
 
-lib_LTLIBRARIES = libJerasure.la
-libJerasure_la_SOURCES = galois.c jerasure.c reed_sol.c cauchy.c liberation.c
-libJerasure_la_LDFLAGS = -version-info 2:0:0
-libJerasure_la_LIBADD = -lgf_complete
+lib_LTLIBRARIES = libFrugalosJerasure.la
+libFrugalosJerasure_la_SOURCES = galois.c jerasure.c reed_sol.c cauchy.c liberation.c
+libFrugalosJerasure_la_LDFLAGS = -version-info 2:0:0
+libFrugalosJerasure_la_LIBADD = -lgf_complete
 include_HEADERS = ../include/jerasure.h
 
 # Install additional Jerasure header files in their own directory.

From 7e37185929355cbd1b4f06e18c851958577f5b23 Mon Sep 17 00:00:00 2001
From: roentgen <fukasawa666@gmail.com>
Date: Tue, 4 Dec 2018 18:43:07 +0900
Subject: [PATCH 6/6] fixed wrong checking of badalloc

---
 src/galois.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/src/galois.c b/src/galois.c
index d35751f..dd45d4c 100644
--- a/src/galois.c
+++ b/src/galois.c
@@ -100,7 +100,7 @@ gf2_t* galois_init_field(int w,
 
   gf2p = (gf2_t *) malloc(sizeof(gf2_t));
   gf2p->gf = (gf_t *) malloc(sizeof(gf_t));
-  if (!gf2p || gf2p->gf) {
+  if (!gf2p || !gf2p->gf) {
     fprintf(stderr, "ERROR -- cannot allocate memory for Galois field w=%d\n", w);
     assert(0);
   }
@@ -153,7 +153,7 @@ gf2_t* galois_init_composite_field(int w,
   
   gf2p = (gf2_t *) malloc(sizeof(gf2_t));
   gf2p->gf = (gf_t *) malloc(sizeof(gf_t));
-  if (!gf2p || gf2p->gf) {
+  if (!gf2p || !gf2p->gf) {
     fprintf(stderr, "ERROR -- cannot allocate memory for Galois field w=%d\n", w);
     assert(0);
   }