ported over working martrix multiplication

Tests/isa/CPrograms/Benchmarks/Matrixmul/build.sh

@@ -0,0 +1,16 @@
+scp ../../start.S home-1:/home/akilan/cheri/output/sdk/bin/
+# scp main.c home-1:/home/akilan/cheri/output/sdk/bin/
+scp ../../malloc.c home-1:/home/akilan/cheri/output/sdk/bin/
+scp ../../malloc_test.c home-1:/home/akilan/cheri/output/sdk/bin/
+scp ../../link.ld home-1:/home/akilan/cheri/output/sdk/bin/
+
+scp matrixmul.c home-1:/home/akilan/cheri/output/sdk/bin/
+
+# ssh home-1 'cd /home/akilan/cheri/output/sdk/bin/ && ./clang   --target=riscv64-unknown-elf   -march=rv64gcxcheri -mabi=lp64d -nostdlib -nostartfiles -fno-builtin-malloc -mcmodel=medany   -Wl,-Ttext=0x80000000 -o testC start.S malloc.c memaccess.c'
+ssh home-1 'cd /home/akilan/cheri/output/sdk/bin/ && ./clang   --target=riscv64-unknown-elf   -march=rv64gcxcheri -mabi=lp64d -DDEFINE_MALLOC -DDEFINE_FREE -nostdlib -nostartfiles -fno-builtin-malloc -mcmodel=medany  -T link.ld -o testC start.S matrixmul.c malloc.c malloc_test.c'
+
+# ssh home-1 'cd /home/akilan/cheri/output/sdk/bin/ && ./llvm-objdump -d testC'
+
+scp home-1:/home/akilan/cheri/output/sdk/bin/testC ../../../
+
+

Tests/isa/CPrograms/Benchmarks/Matrixmul/matrixmul.c

@@ -0,0 +1,215 @@
+// Source: https://github.com/ivanbgd/Matrix-Multiplication-MatMul-C/blob/master/matmul_1d_seq.c
+#define MATMUL_1D
+#ifdef MATMUL_1D
+
+/* Matrices are represented as 1-D arrays in memory.
+ * That means they are contiguous in memory.
+ * Minimum dimension is 1, not 0, and internal dimensions must match. */
+
+// #include <math.h>
+// #include <omp.h>
+// #include <stdio.h>
+// #include <stdlib.h>
+// #include <string.h>
+// #include <time.h>
+#include <cheriintrin.h>
+#include <stddef.h>
+#include <stdint.h>
+
+void free(void * __capability ptr);
+void * __capability malloc(size_t size);
+
+/* Initializes vector or matrix, sequentially, with indices. */
+void init_seq(double * __capability a, const unsigned n_rows_a, const unsigned n_cols_a) {
+    for (size_t i = 0; i < n_rows_a; i++) {
+        for (size_t j = 0; j < n_cols_a; j++) {
+            a[i*n_cols_a + j] = 2.0;
+        }
+    }
+}
+
+// static unsigned int seed = 123456789;  // must be initialized somehow
+
+// unsigned int rand_test(void)
+// {
+//     seed = (1103515245 * seed + 12345);
+//     return (seed >> 16) & 0x7FFF; // 15-bit result
+// }
+
+// /* Initializes vector or matrix, randomly. */
+// void init_rand(double *a, const unsigned n_rows_a, const unsigned n_cols_a) {
+//     for (size_t i = 0; i < n_rows_a; i++) {
+//         for (size_t j = 0; j < n_cols_a; j++) {
+//             a[i*n_cols_a + j] = rand_test() / 32767.0;
+//             // rand_test() / (double)RAND_MAX;
+//         }
+//     }
+// }
+
+/*  Takes and returns a new matrix, t, which is a transpose of the original one, m.
+    It's also flat in memory, i.e., 1-D, but it should be looked at as a transpose
+    of m, meaning, n_rows_t == n_cols_m, and n_cols_t == n_rows_m.
+    The original matrix m stays intact. */
+double __capability * transpose(const double __capability *m, const unsigned n_rows_m, const unsigned n_cols_m, double __capability *t) {
+    for (size_t i = 0; i < n_rows_m; i++) {
+        for (size_t j = 0; j < n_cols_m; j++) {
+            t[j*n_rows_m + i] = m[i*n_cols_m + j];
+        }
+    }
+
+    return t;
+}
+
+/* Dot product of two arrays, or matrix product
+ * Allocates and returns an array.
+ * This variant doesn't transpose matrix b, and it's a lot slower. */
+double * __capability dot_simple(const double * __capability a, const unsigned n_rows_a, const unsigned n_cols_a,\
+                   const double * __capability b, const unsigned n_rows_b, const unsigned n_cols_b) {
+
+    if (n_cols_a != n_rows_b) {
+        // printf("#columns A must be equal to #rows B!\n");
+        // system("pause");
+        // exit(-2);
+    }
+
+    double *__capability c = malloc(n_rows_a * n_cols_b * sizeof(*c));
+    if (c == NULL) {
+        // printf("Couldn't allocate memory!\n");
+        // system("pause");
+        // exit(-1);
+    }
+
+    for (size_t i = 0; i < n_rows_a; i++) {
+        for (size_t k = 0; k < n_cols_b; k++) {
+            double sum = 0.0;
+            for (size_t j = 0; j < n_cols_a; j++) {
+                sum += a[i*n_cols_a + j] * b[j*n_cols_b + k];
+            }
+            c[i*n_cols_b + k] = sum;
+        }
+    }
+
+    return c;
+}
+
+/* Dot product of two arrays, or matrix product
+ * Allocates and returns an array.
+ * This variant transposes matrix b, and it's a lot faster. */
+double * __capability dot(const double  __capability *a, const unsigned n_rows_a, const unsigned n_cols_a, \
+            const double * __capability b, const unsigned n_rows_b, const unsigned n_cols_b) {
+
+    if (n_cols_a != n_rows_b) {
+        // printf("#columns A must be equal to #rows B!\n");
+        // system("pause");
+        // exit(-2);
+    }
+
+    double * __capability bt = malloc(n_rows_b * n_cols_b * sizeof(*b));
+
+    double * __capability c = malloc(n_rows_a * n_cols_b * sizeof(*c));    
+
+    if ((c == NULL) || (bt == NULL)) {
+        // printf("Couldn't allocate memory!\n");
+        // system("pause");
+        // exit(-1);
+    }
+
+    bt = transpose(b, n_rows_b, n_cols_b, bt);
+
+    for (size_t i = 0; i < n_rows_a; i++) {
+        for (size_t k = 0; k < n_cols_b; k++) {
+            double sum = 0.0;
+            for (size_t j = 0; j < n_cols_a; j++) {
+                sum += a[i*n_cols_a + j] * bt[k*n_rows_b + j];
+            }
+            c[i*n_cols_b + k] = sum;
+        }
+    }
+
+    free(bt);
+
+    return c;
+}
+
+/* Prints vector, or matrix. */
+// void print(const double *a, const unsigned n_rows_a, const unsigned n_cols_a) {
+//     for (size_t i = 0; i < n_rows_a; i++) {
+//         for (size_t j = 0; j < n_cols_a; j++) {
+//             printf("%8.3f ", a[i*n_cols_a + j]);
+//         }
+//         printf("\n");
+//     }
+//     printf("\n");
+// }
+
+int main(void) {
+    /* Intializes random number generator */
+    // time_t t;
+    // srand((unsigned)time(&t));
+    // srand(0);
+
+    /* For measuring time */
+    double t0, t1;
+
+    const unsigned scale = 14;
+    const unsigned n_rows_a = 4 * scale;
+    const unsigned n_cols_a = 3 * scale;
+    const unsigned n_rows_b = 3 * scale;
+    const unsigned n_cols_b = 2 * scale;
+
+    double __capability *a = malloc(n_rows_a * n_cols_a * sizeof(*a));
+    double __capability *b = malloc(n_rows_b * n_cols_b * sizeof(*b));
+    double __capability *c = NULL;
+    double __capability *d = NULL;
+
+    if (!a || !b) {
+        // printf("Couldn't allocate memory!\n");
+        while(1);
+        // system("pause");
+        // exit(-1);
+    }
+
+    // init_rand(a, n_rows_a, n_cols_a);
+    // init_rand(b, n_rows_b, n_cols_b);
+
+    init_seq(a, n_rows_a, n_cols_a);
+    init_seq(b, n_rows_b, n_cols_b);
+
+    // for (size_t i = 0; i < n_rows_a; i++) {
+    //     for (size_t j = 0; j < n_cols_a; j++) {
+    //         a[i*n_cols_a + j] = i*n_cols_a + j;
+    //     }
+    // }
+
+    // t0 = omp_get_wtime();
+    c = dot_simple(a, n_rows_a, n_cols_a, b, n_rows_b, n_cols_b);
+    // t1 = omp_get_wtime();
+    // printf("Dot Simple: Elapsed time %.3f s\n", t1 - t0);
+
+    // // t0 = omp_get_wtime();
+    d = dot(a, n_rows_a, n_cols_a, b, n_rows_b, n_cols_b);
+    // t1 = omp_get_wtime();
+    // printf("Dot: Elapsed time %.3f s\n", t1 - t0);
+
+    // if (scale == 1) {
+    //     // printf("Matrix A:\n");
+    //     // print(a, n_rows_a, n_cols_a);
+    //     // // printf("Matrix B:\n");
+    //     // print(b, n_rows_b, n_cols_b);
+    //     // // printf("Matrix C:\n");
+    //     // print(c, n_rows_a, n_cols_b);
+    //     // // printf("Matrix D:\n");
+    //     // print(d, n_rows_a, n_cols_b);
+    // }
+
+    free(a);
+    free(b);
+    free(c);
+    free(d);
+
+    // system("pause");
+
+    return(0);
+}
+
+#endif // MATMUL_1D

Tests/isa/CPrograms/Benchmarks/Memaccess/memaccess.c

@@ -27,10 +27,10 @@ All rights reserved.
 
 #define NPAD 7
 #define MIN_WSS sizeof(struct l)
-#define MAX_WSS 32 //2^32
+#define MAX_WSS 64 //2^32
 
-void free(void * __capability ptr);
-void * __capability malloc(size_t size);
+void free_test(void * __capability ptr);
+void * __capability malloc_test(size_t size);
 
 
 // struct l{
@@ -278,7 +278,7 @@ void * walk(void * __capability param)
 
 		// printf("%d,%ld,%ld,%ld\n",data->thread_index,data->working_set_size, end-start, data->working_set_size/sizeof(struct l));
 
-		free(data);
+		free_test(data);
 		return NULL;
 }
 
@@ -298,7 +298,7 @@ int main(void)
 
 	// pthread_t threads[NTHREADS];
 
-	root=malloc(MAX_WSS/sizeof(struct l) * sizeof(struct l));	
+	root=malloc_test(MAX_WSS/sizeof(struct l) * sizeof(struct l));	
 		
 	for(working_set_size=MIN_WSS;working_set_size<MAX_WSS;working_set_size=working_set_size<<1)
 	{
@@ -331,7 +331,7 @@ int main(void)
 // 		for(i=0;i<NTHREADS;i++)
 // 			pthread_join(threads[i],NULL);					
 // #else
-		tdata = malloc(sizeof(struct thread_start_data));
+		tdata = malloc_test(sizeof(struct thread_start_data));
 		tdata->thread_index=0;
 		tdata->root=root;
 		tdata->working_set_size=working_set_size;
@@ -348,7 +348,7 @@ int main(void)
 	// printf("-----------------------------------------------");
 
 
-	free(root);	
+	free_test(root);	
     // root = NULL;
 
 	// while(1);

Tests/isa/CPrograms/Benchmarks/Richards/build.sh

@@ -7,7 +7,7 @@ scp ../../link.ld home-1:/home/akilan/cheri/output/sdk/bin/
 scp richards.c home-1:/home/akilan/cheri/output/sdk/bin/
 
 # ssh home-1 'cd /home/akilan/cheri/output/sdk/bin/ && ./clang   --target=riscv64-unknown-elf   -march=rv64gcxcheri -mabi=lp64d -nostdlib -nostartfiles -fno-builtin-malloc -mcmodel=medany   -Wl,-Ttext=0x80000000 -o testC start.S malloc.c memaccess.c'
-ssh home-1 'cd /home/akilan/cheri/output/sdk/bin/ && ./clang   --target=riscv64-unknown-elf   -march=rv64gcxcheri -mabi=lp64d -DDEFINE_MALLOC -DDEFINE_FREE -nostdlib -nostartfiles -fno-builtin-malloc -mcmodel=medany  -T link.ld -o testC start.S richards.c malloc.c malloc_test.c'
+ssh home-1 'cd /home/akilan/cheri/output/sdk/bin/ && ./clang   --target=riscv64-unknown-elf   -march=rv64gcxcheri -mabi=lp64d -DDEFINE_MALLOC -DDEFINE_FREE -nostartfiles -fno-builtin-malloc -mcmodel=medany  -T link.ld -o testC start.S richards.c malloc.c malloc_test.c'
 
 # ssh home-1 'cd /home/akilan/cheri/output/sdk/bin/ && ./llvm-objdump -d testC'
 

Tests/isa/CPrograms/Benchmarks/barnes/barnes.c

@@ -0,0 +1,495 @@
+/*
+ * File: barnes_hut.c
+ * --------------
+ * Implements the Barnes Hut algorithm for n-body
+ * simulation with galaxy-like initial conditions.
+ * Uses OpenGl and GLUT for graphics. Written by:
+ * Joel Backsell, Kim Torberntsson & Alexander Bilock.
+ * Source: https://github.com/KimTorberntsson/Barnes-Hut/blob/master/barnes_hut.c
+ */
+
+// #include <stdlib.h>
+// #include <math.h>
+// #include <stdio.h>
+// #include <math.h>
+// #include <time.h>
+// #include <GLUT/glut.h>
+
+#include <cheriintrin.h>
+#include <stddef.h>
+#include <stdint.h>
+
+#include "barnes_hut.h"
+// #include "graphics.h"
+
+// void free(void *ptr);
+// void *malloc(size_t size);
+
+void free(void * __capability ptr);
+void * __capability malloc(size_t size);
+
+//Some constants and global variables
+int N = 10;
+const double L = 1, W = 1, dt = 1e-3, alpha = 0.25, V = 50, epsilon = 1e-1, grav = 0.04; //grav should be 100/N
+double __capability *x;
+double __capability *y;
+double __capability *u;
+double __capability *v;
+double __capability *force_x;
+double __capability *force_y;
+double __capability *mass;
+struct node_t __capability *root;
+
+// -------- Modified by Akilan ----------
+
+double my_sqrt(double x) {
+    if (x <= 0) return 0;
+
+    double guess = x;
+    for (int i = 0; i < 10; i++) {
+        guess = 0.5 * (guess + x / guess);
+    }
+    return guess;
+}
+
+#define PI 3.141592653589793
+
+double wrap_angle(double x) {
+    while (x > PI)  x -= 2*PI;
+    while (x < -PI) x += 2*PI;
+    return x;
+}
+
+double my_sin(double x) {
+    x = wrap_angle(x);
+
+    double x2 = x * x;
+    return x
+         - (x * x2) / 6
+         + (x * x2 * x2) / 120
+         - (x * x2 * x2 * x2) / 5040;
+}
+
+double my_cos(double x) {
+    x = wrap_angle(x);
+
+    double x2 = x * x;
+    return 1
+         - x2 / 2
+         + (x2 * x2) / 24
+         - (x2 * x2 * x2) / 720;
+}
+
+// --------------------------------------
+
+// -------- Modified by Akilan ----------
+
+static unsigned int seed = 123456789;  // must be initialized somehow
+
+unsigned int rand_test(void)
+{
+    seed = (1103515245 * seed + 12345);
+    return (seed >> 16) & 0x7FFF; // 15-bit result
+}
+
+// ---------------------------------------
+
+/*
+ * Function for producing a random number between two double values
+ */
+double frand(double xmin, double xmax) {
+    return xmin + (xmax-xmin)*rand_test()/32767;
+}
+
+/*
+ * Prints the time between two clocks in seconds
+ */
+// void print_time(clock_t s, clock_t e)
+// {
+//     printf("Time: %f seconds\n", (double)(e-s)/CLOCKS_PER_SEC);
+// }
+
+/*
+ * This function is called every time GLUT refreshes the display.
+ */
+// void display(void) {
+// 	//Do a time step
+// 	time_step();
+
+// 	//Draw points
+// 	drawPoints(x, y, N);
+// }
+
+/*
+ * Updates the positions of the particles of a time step.
+ */
+void time_step(void) {
+	//Allocate memory for root
+	root = malloc(sizeof(struct node_t));
+	set_node(root);
+	root->min_x = 0; root->max_x = 1; root->min_y = 0; root->max_y = 1;
+
+	//Put particles in tree
+	for(int i = 0; i < N; i++) {
+		put_particle_in_tree(i, root);
+	}
+
+	//Calculate mass and center of mass
+	calculate_mass(root);
+	calculate_center_of_mass_x(root);
+	calculate_center_of_mass_y(root);
+
+	//Calculate forces
+	update_forces();
+
+	//Update velocities and positions
+	for(int i = 0; i < N; i++) {
+		double ax = force_x[i]/mass[i];
+		double ay = force_y[i]/mass[i];
+		u[i] += ax*dt;
+		v[i] += ay*dt;
+		x[i] += u[i]*dt;
+		y[i] += v[i]*dt;
+
+		/* This of course doesn't make any sense physically,
+		 * but makes sure that the particles stay within the
+		 * bounds. Normally the particles won't leave the
+		 * area anyway.
+		*/
+		bounce(&x[i], &y[i], &u[i], &v[i]);
+	}
+
+	//Free memory
+	free_node(root);
+	free(root);
+}
+
+/*
+ * If a particle moves beyond any of the boundaries then bounce it back
+ */
+void bounce(double __capability *x, double __capability *y, double __capability *u, double __capability *v) {
+	double W = 1.0f, H = 1.0f;
+	if(*x > W) {
+		*x = 2*W - *x;
+		*u = -*u;
+	}
+
+	if(*x < 0) {
+		*x = -*x;
+		*u = -*u;
+	}
+
+	if(*y > H) {
+		*y = 2*H - *y;
+		*v = -*v;
+	}
+
+	if(*y < 0) {
+		*y = -*y;
+		*v = -*v;
+	}
+}
+
+/*
+ * Puts a particle recursively in the Barnes Hut quad-tree.
+ */
+void put_particle_in_tree(int new_particle, struct node_t __capability *node) {
+	//If no particle is assigned to the node
+	if(!node->has_particle) {
+		node->particle = new_particle;
+		node->has_particle = 1;
+	}
+	//If the node has no children
+	else if(!node->has_children) {
+		//Allocate and initiate children
+		node->children = malloc(4*sizeof(struct node_t));
+		for(int i = 0; i < 4; i++) {
+			set_node(&node->children[i]);
+		}
+
+		//Set boundaries for the children
+		node->children[0].min_x = node->min_x; node->children[0].max_x = (node->min_x + node->max_x)/2;
+		node->children[0].min_y = node->min_y; node->children[0].max_y = (node->min_y + node->max_y)/2;
+
+		node->children[1].min_x = (node->min_x + node->max_x)/2; node->children[1].max_x = node->max_x;
+		node->children[1].min_y = node->min_y; node->children[1].max_y = (node->min_y + node->max_y)/2;
+
+		node->children[2].min_x = node->min_x; node->children[2].max_x = (node->min_x + node->max_x)/2;
+		node->children[2].min_y = (node->min_y + node->max_y)/2; node->children[2].max_y = node->max_y;
+
+		node->children[3].min_x = (node->min_x + node->max_x)/2; node->children[3].max_x = node->max_x;
+		node->children[3].min_y = (node->min_y + node->max_y)/2; node->children[3].max_y = node->max_y;
+
+		//Put old particle into the appropriate child
+		place_particle(node->particle, node);
+
+		//Put new particle into the appropriate child
+		place_particle(new_particle, node);
+
+		//It now has children
+		node->has_children = 1;
+	}
+	//Add the new particle to the appropriate children
+	else {
+		//Put new particle into the appropriate child
+		place_particle(new_particle, node);
+	}
+}
+
+
+/*
+ * Puts a particle in the right child of a node with children.
+ */
+void place_particle(int particle, struct node_t __capability *node) {
+	if(x[particle] <= (node->min_x + node->max_x)/2 && y[particle] <= (node->min_y + node->max_y)/2) {
+		put_particle_in_tree(particle, &node->children[0]);
+	} else if(x[particle] > (node->min_x + node->max_x)/2 && y[particle] < (node->min_y + node->max_y)/2) {
+		put_particle_in_tree(particle, &node->children[1]);
+	} else if(x[particle] < (node->min_x + node->max_x)/2 && y[particle] > (node->min_y + node->max_y)/2) {
+		put_particle_in_tree(particle, &node->children[2]);
+	} else {
+		put_particle_in_tree(particle, &node->children[3]);
+	}
+}
+
+/*
+ * Sets initial values for a new node
+ */
+void set_node(struct node_t __capability *node) {
+	node->has_particle = 0;
+	node->has_children = 0;
+}
+
+/*
+ * Frees memory for a node and its children recursively.
+ */
+void free_node(struct node_t __capability *node){
+	if(node->has_children){
+		free_node(&node->children[0]);
+		free_node(&node->children[1]);
+		free_node(&node->children[2]);
+		free_node(&node->children[3]);
+		free(node->children);
+	}
+}
+
+/*
+ * Displays the boundaries of the tree using OpenGL and GLUT.
+ * Used for debugging purposes.
+ */
+// void display_tree(struct node_t *node) {
+// 	drawRect(node->min_x, node->max_x, node->min_y, node->max_y);
+// 	if(node->has_children == 1) {
+// 		for(int i = 0; i < 4; i++) {
+// 			display_tree(&node->children[i]);
+// 		}
+// 	}
+// }
+
+/*
+ * Calculates the total mass for the node. It recursively updates the mass
+ * of itself and all of its children.
+ */
+double calculate_mass(struct node_t __capability *node) {
+	if(!node->has_particle) {
+		node->total_mass = 0;
+	} else if(!node->has_children) {
+		node->total_mass = mass[node->particle];
+	} else {
+		node->total_mass = 0;
+		for(int i = 0; i < 4; i++) {
+			node->total_mass += calculate_mass(&node->children[i]);
+		}
+	}
+	return node->total_mass;
+}
+
+/*
+ * Calculates the x-position of the centre of mass for the
+ * node. It recursively updates the position of itself and
+ * all of its children.
+ */
+double calculate_center_of_mass_x(struct node_t __capability *node) {
+	if(!node->has_children) {
+		node->c_x = x[node->particle];
+	} else {
+		node->c_x = 0;
+		double m_tot = 0;
+		for(int i = 0; i < 4; i++) {
+			if(node->children[i].has_particle) {
+				node->c_x += node->children[i].total_mass*calculate_center_of_mass_x(&node->children[i]);
+				m_tot += node->children[i].total_mass;
+			}
+		}
+		node->c_x /= m_tot;
+	}
+	return node->c_x;
+}
+
+/*
+ * Calculates the y-position of the centre of mass for the
+ * node. It recursively updates the position of itself and
+ * all of its children.
+ */
+double calculate_center_of_mass_y(struct node_t __capability *node) {
+	if(!node->has_children) {
+		node->c_y = y[node->particle];
+	} else {
+		node->c_y = 0;
+		double m_tot = 0;
+		for(int i = 0; i < 4; i++) {
+			if(node->children[i].has_particle) {
+				node->c_y += node->children[i].total_mass*calculate_center_of_mass_y(&node->children[i]);
+				m_tot += node->children[i].total_mass;
+			}
+		}
+		node->c_y /= m_tot;
+	}
+	return node->c_y;
+}
+
+/*
+ * Calculates the forces in a time step of all particles in
+ * the simulation using the Barnes Hut quad tree.
+ */
+void update_forces(){
+	for(int i = 0; i < N; i++) {
+		force_x[i] = 0;
+		force_y[i] = 0;
+		update_forces_help(i, root);
+	}
+}
+
+/*
+ * Help function for calculating the forces recursively
+ * using the Barnes Hut quad tree.
+ */
+void update_forces_help(int particle, struct node_t __capability *node) {
+    //The node is a leaf node with a particle and not the particle itself
+    if(!node->has_children && node->has_particle && node->particle != particle) {
+        double r = my_sqrt((x[particle] - node->c_x)*(x[particle] - node->c_x) + (y[particle] - node->c_y)*(y[particle] - node->c_y));
+        calculate_force(particle, node, r);
+    }
+    //The node has children
+    else if(node->has_children) {
+        //Calculate r and theta
+        double r = my_sqrt((x[particle] - node->c_x)*(x[particle] - node->c_x) + (y[particle] - node->c_y)*(y[particle] - node->c_y));
+        double theta = (node->max_x - node->min_x)/r;
+
+        /* If the distance to the node's centre of mass is far enough, calculate the force,
+         * otherwise traverse further down the tree
+         */
+        if(theta < 0.5){
+            calculate_force(particle, node, r);
+        } else {
+            update_forces_help(particle, &node->children[0]);
+            update_forces_help(particle, &node->children[1]);
+            update_forces_help(particle, &node->children[2]);
+            update_forces_help(particle, &node->children[3]);
+        }
+    }
+}
+
+/*
+ * Calculates and updates the force of a particle from a node.
+ */
+void calculate_force(int particle, struct node_t __capability *node, double r){
+    double temp = -grav*mass[particle]*node->total_mass/((r + epsilon)*(r + epsilon)*(r + epsilon));
+	force_x[particle] += (x[particle] - node->c_x)*temp;
+	force_y[particle] += (y[particle] - node->c_y)*temp;
+}
+
+/*
+ * Main function.
+ */
+int main(void) {
+
+	//The first arguments sets if the graphics should be used
+	// int graphics = 1;
+	// if (argc > 1) {
+	// 	graphics = atoi(argv[1]);
+	// }
+
+	//The second argument sets the number of time steps
+	// int time_steps = 1;
+	// if (argc > 2) {
+	// 	time_steps = atoi(argv[2]);
+	// }
+
+	//Initiate memory for the vectors
+	// x = (double __capability *)malloc(N*sizeof(double));
+	// y = (double __capability *)malloc(N*sizeof(double));
+	// u = (double __capability *)malloc(N*sizeof(double));
+	// v = (double __capability *)malloc(N*sizeof(double));
+	// // -------- Modified by Akilan ----------
+	// force_x = (double __capability *)malloc(N*sizeof(double));
+	// force_y = (double __capability *)malloc(N*sizeof(double));
+	// // --------------------------------------
+    // mass = (double __capability *)malloc(N*sizeof(double));
+
+	//Set the initial values
+	// for(int i = 0; i < N; i++) {
+	// 	mass[i] = 1;
+	// 	double R = frand(0, L/4);
+	// 	double theta = frand(0, 2*M_PI);
+	// 	x[i] = L/2 + R*my_cos(theta);
+	// 	y[i] = W/2 + alpha*R*my_sin(theta);
+	// 	double R_prim = my_sqrt(pow(x[i] - L/2, 2) + pow(y[i] - W/2, 2));
+	// 	u[i] = -V*R_prim*my_sin(theta);
+	// 	v[i] = V*R_prim*my_cos(theta);
+	// }
+	// for (int i = 0; i < N; i++) {
+    // 	mass[i] = 1;
+
+    // 	double R = frand(0, L/4);
+    // 	double theta = frand(0, 2*PI);
+
+    // 	double c = my_cos(theta);
+    // 	double s = my_sin(theta);
+
+    // 	x[i] = L/2 + R * c;
+    // 	y[i] = W/2 + alpha * R * s;
+
+    // 	double dx = x[i] - L/2;
+    // 	double dy = y[i] - W/2;
+
+    // 	double R_prim = my_sqrt(dx*dx + dy*dy);
+
+    // 	u[i] = -V * R_prim * s;
+    // 	v[i] =  V * R_prim * c;
+	// }
+
+	/* Run the GLUT display function if the graphics mode is on.
+     * Otherwise just run the simulations without graphics
+     */
+	// if (graphics) {
+	// 	// Initialize the graphics routines
+	// 	graphicsInit(&argc, argv, display);
+
+	// 	//Used for the display window
+	// 	glutMainLoop();
+	// } else {
+		//Begin taking time
+		// long start = clock();
+
+		//The main loop
+		// for(int i = 0; i < time_steps; i++) {
+		// 	time_step();
+		// }
+
+		//Stop taking time and print elapsed time
+		// long stop = clock();
+		// print_time(start, stop);
+	// }
+
+	//Free memory
+	// free(x);
+	// free(y);
+	// free(u);
+	// free(v);
+	// free(force_x);
+	// free(force_y);
+    // free(mass);
+
+	return 0;
+}

Tests/isa/CPrograms/Benchmarks/barnes/barnes_hut.h

@@ -0,0 +1,92 @@
+/*
+ * File: barnes_hut.h
+ * --------------
+ * Header file for barnes_hut.c
+ * Written by:
+ * Joel Backsell, Kim Torberntsson & Alexander Bilock.
+ */
+
+#ifndef Barnes_Hut_barnes_hut_h
+#define Barnes_Hut_barnes_hut_h
+
+/*
+ * Struct that represents a node of the Barnes Hut quad tree.
+ */
+struct node_t {
+    int particle;
+    int has_particle;
+    int has_children;
+    double min_x, max_x, min_y, max_y, total_mass, c_x, c_y;
+
+    struct node_t __capability *children;
+};
+
+//Function for producing a random number between two double values.
+double frand(double xmin, double xmax);
+
+//Prints the time between two clocks in seconds
+// void print_time(clock_t s, clock_t e);
+
+//OpenGL method for displaying the particles
+// void display(void);
+
+//Calculates the positions of a time step
+// void time_step(void);
+
+//If a particle moves beyond any of the boundaries then bounce it back
+// void bounce(double *x, double *y, double *u, double *v);
+
+// //Functions for handling the placement of particles in the tree
+// void put_particle_in_tree(int new_particle, struct node_t *node);
+// void place_particle(int particle, struct node_t *node);
+// void set_node(struct node_t *node);
+// void free_node(struct node_t *node);
+// void display_tree(struct node_t *node);
+
+// //Functions for calculating the mass and centre of mass of the tree
+// double calculate_mass(struct node_t *node);
+// double calculate_center_of_mass_x(struct node_t *node);
+// double calculate_center_of_mass_y(struct node_t *node);
+
+// //Functions for the force calculations
+// void update_forces();
+// void update_forces_help(int particle, struct node_t *node);
+// void calculate_force(int particle, struct node_t *node, double r);
+
+void bounce(double __capability *x,
+            double __capability *y,
+            double __capability *u,
+            double __capability *v);
+
+/* Tree functions */
+void put_particle_in_tree(int new_particle,
+                          struct node_t __capability *node);
+
+void place_particle(int particle,
+                    struct node_t __capability *node);
+
+void set_node(struct node_t __capability *node);
+
+void free_node(struct node_t __capability *node);
+
+void display_tree(struct node_t __capability *node);
+
+/* Mass / COM */
+double calculate_mass(struct node_t __capability *node);
+
+double calculate_center_of_mass_x(struct node_t __capability *node);
+
+double calculate_center_of_mass_y(struct node_t __capability *node);
+
+/* Forces */
+void update_forces(void);
+
+void update_forces_help(int particle,
+                        struct node_t __capability *node);
+
+void calculate_force(int particle,
+                     struct node_t __capability *node,
+                     double r);
+
+
+#endif

Tests/isa/CPrograms/Benchmarks/barnes/build.sh

@@ -0,0 +1,17 @@
+scp ../../start.S home-1:/home/akilan/cheri/output/sdk/bin/
+# scp main.c home-1:/home/akilan/cheri/output/sdk/bin/
+scp ../../malloc.c home-1:/home/akilan/cheri/output/sdk/bin/
+scp ../../malloc_test.c home-1:/home/akilan/cheri/output/sdk/bin/
+scp ../../link.ld home-1:/home/akilan/cheri/output/sdk/bin/
+
+scp barnes.c home-1:/home/akilan/cheri/output/sdk/bin/
+scp barnes_hut.h home-1:/home/akilan/cheri/output/sdk/bin/
+
+# ssh home-1 'cd /home/akilan/cheri/output/sdk/bin/ && ./clang   --target=riscv64-unknown-elf   -march=rv64gcxcheri -mabi=lp64d -nostdlib -nostartfiles -fno-builtin-malloc -mcmodel=medany   -Wl,-Ttext=0x80000000 -o testC start.S malloc.c memaccess.c'
+ssh home-1 'cd /home/akilan/cheri/output/sdk/bin/ && ./clang   --target=riscv64-unknown-elf   -march=rv64gcxcheri -mabi=lp64d -DDEFINE_MALLOC -DDEFINE_FREE -nostdlib -nostartfiles -fno-builtin-malloc -mcmodel=medany  -T link.ld -o testC start.S barnes.c malloc.c malloc_test.c'
+
+# ssh home-1 'cd /home/akilan/cheri/output/sdk/bin/ && ./llvm-objdump -d testC'
+
+scp home-1:/home/akilan/cheri/output/sdk/bin/testC ../../../
+
+

Tests/isa/CPrograms/link.ld

@@ -15,5 +15,8 @@ SECTIONS
   _end = .;
 
   __malloc_start = .;
-  . = . + 512;
+  . = . + 0x100000;
+
+  /* End of uninitalized data segement */
+  _end = .;
 }

Tests/isa/CPrograms/malloc.c

@@ -53,11 +53,13 @@ static inline void * __capability add_delta(void * __capability cap, int offset)
 // Malloc wrapper
 void * __capability malloc(size_t size) {
 
-    void *raw = tiny_malloc(size);
+    uintptr_t raw = (uintptr_t)tiny_malloc(size);
 
-    void *__capability cap = (void *__capability)raw;
+    // void *__capability cap = (void *__capability)raw;
+    void *__capability cap = cheri_ddc_get();
 
-    cap = cheri_bounds_set(cap, size); 
+    // // Set address from raw pointer
+    cap = cheri_address_set(cap, raw);
 
     int delta = 12;
 
@@ -65,6 +67,10 @@ void * __capability malloc(size_t size) {
 
     cap = add_delta(cap, delta);
 
+    size_t aligned = cheri_representable_length(size);
+
+    cap = cheri_bounds_set(cap, aligned); 
+
     // // Align to 8 bytes (important for capability safety)
     // size = (size + 7) & ~7;