1 files changed, 50 insertions, 22 deletions

diff --git a/libar2simplified_init_context.c b/libar2simplified_init_context.c
index 9dcbfb0..814e182 100644
--- a/libar2simplified_init_context.c
+++ b/libar2simplified_init_context.c
@@ -2,6 +2,7 @@
 #include "common.h"
 #include <pthread.h>
 #include <semaphore.h>
+#include <stdatomic.h>
 
 
 struct user_data;
@@ -11,7 +12,7 @@ struct thread_data {
 	struct user_data *master;
 	pthread_t thread;
 	sem_t semaphore;
-	int error;
+	_Atomic int error;
 	void (*function)(void *data);
 	void *function_input;
 };
@@ -50,14 +51,15 @@ alignedalloc(size_t num, size_t size, size_t extra, size_t alignment)
 static void *
 allocate(size_t num, size_t size, size_t alignment, struct libar2_context *ctx)
 {
-	size_t pad = (alignment - ((2 * sizeof(size_t)) & (alignment - 1))) & (alignment - 1);
-	char *ptr = alignedalloc(num, size, pad + 2 * sizeof(size_t), alignment);
+	size_t extra = 2 * sizeof(size_t);
+	size_t pad = -extra & (alignment - 1);
+	char *ptr = alignedalloc(num, size, pad + extra, alignment);
 	if (ptr) {
-		ptr = &ptr[pad];
-		*(size_t *)ptr = pad;
-		ptr = &ptr[sizeof(size_t)];
-		*(size_t *)ptr = num * size;
-		ptr = &ptr[sizeof(size_t)];
+		ptr += pad;
+		memcpy(ptr, &pad, sizeof(size_t));
+		ptr += sizeof(size_t);
+		memcpy(ptr, &(size_t){num * size}, sizeof(size_t));
+		ptr += sizeof(size_t);
 	}
 	(void) ctx;
 	return ptr;
@@ -68,10 +70,13 @@ static void
 deallocate(void *ptr, struct libar2_context *ctx)
 {
 	char *p = ptr;
+	size_t size, pad;
 	p -= sizeof(size_t);
-	libar2_erase(ptr, *(size_t *)p);
+	memcpy(&size, p, sizeof(size_t));
 	p -= sizeof(size_t);
-	p -= *(size_t *)p;
+	memcpy(&pad, p, sizeof(size_t));
+	p -= pad;
+	libar2_erase(p, size + pad + 2u * sizeof(size_t));
 	free(p);
 	(void) ctx;
 }
@@ -81,31 +86,44 @@ static void *
 thread_loop(void *data_)
 {
 	struct thread_data *data = data_;
+	void (*function)(void *data);
+	void *function_input;
 	int err;
 
 	for (;;) {
 		if (sem_wait(&data->semaphore)) {
 			if (errno == EINTR)
 				continue;
-			data->error = errno;
+			atomic_store(&data->error, errno);
 			return NULL;
 		}
 
-		if (!data->function) {
-			data->error = ENOTRECOVERABLE;
+		err = pthread_mutex_lock(&data->master->mutex);
+		if (err) {
+			atomic_store(&data->error, err);
+			return NULL;
+		}
+		function = data->function;
+		function_input = data->function_input;
+		if (!function) {
+			atomic_store(&data->error, ENOTRECOVERABLE);
+			pthread_mutex_unlock(&data->master->mutex);
 			return NULL;
 		}
-		data->function(data->function_input);
+		pthread_mutex_unlock(&data->master->mutex);
+
+		function(function_input);
 
 		err = pthread_mutex_lock(&data->master->mutex);
 		if (err) {
-			data->error = err;
+			atomic_store(&data->error, err);
 			return NULL;
 		}
 		data->master->resting[data->index / 64] |= (uint_least64_t)1 << (data->index % 64);
+
 		pthread_mutex_unlock(&data->master->mutex);
 		if (sem_post(&data->master->semaphore)) {
-			data->error = errno;
+			atomic_store(&data->error, errno);
 			return NULL;
 		}
 	}
@@ -124,15 +142,17 @@ run_thread(size_t index, void (*function)(void *arg), void *arg, struct libar2_c
 		return -1;
 	}
 	data->resting[index / 64] ^= (uint_least64_t)1 << (index % 64);
-	pthread_mutex_unlock(&data->mutex);
 
-	if (data->threads[index].error) {
-		errno = data->threads[index].error;
+	err = atomic_load(&data->threads[index].error);
+	if (err) {
+		pthread_mutex_unlock(&data->mutex);
+		errno = err;
 		return -1;
 	}
 
 	data->threads[index].function = function;
 	data->threads[index].function_input = arg;
+	pthread_mutex_unlock(&data->mutex);
 	if (sem_post(&data->threads[index].semaphore))
 		return -1;
 
@@ -145,15 +165,16 @@ destroy_thread_pool(struct libar2_context *ctx)
 {
 	struct user_data *data = ctx->user_data;
 	size_t i;
-	int ret = 0;
+	int err, ret = 0;
 	for (i = data->nthreads; i--;)
 		if (run_thread(i, pthread_exit, NULL, ctx))
 			return -1;
 	for (i = data->nthreads; i--;) {
 		pthread_join(data->threads[i].thread, NULL);
 		sem_destroy(&data->threads[i].semaphore);
-		if (data->threads[i].error)
-			ret = data->threads[i].error;
+		err = atomic_load(&data->threads[i].error);
+		if (err)
+			ret = err;
 	}
 	sem_destroy(&data->semaphore);
 	pthread_mutex_destroy(&data->mutex);
@@ -244,6 +265,7 @@ init_thread_pool(size_t desired, size_t *createdp, struct libar2_context *ctx)
 
 	for (i = 0; i < data->nthreads; i++) {
 		memset(&data->threads[i], 0, sizeof(data->threads[i]));
+		atomic_init(&data->threads[i].error, 0);
 		data->threads[i].master = data;
 		data->threads[i].index = i;
 		data->resting[i / 64] |= (uint_least64_t)1 << (i % 64);
@@ -290,6 +312,11 @@ await_threads(size_t *indices, size_t n, size_t require, struct libar2_context *
 
 	memset(data->joined, 0, (data->nthreads + 63) / 64 * sizeof(*data->joined));
 
+	err = pthread_mutex_lock(&data->mutex);
+	if (err) {
+		errno = err;
+		return 0;
+	}
 	for (i = 0; i < data->nthreads; i += 64) {
 		for (;;) {
 			one = data->resting[i / 64];
@@ -302,6 +329,7 @@ await_threads(size_t *indices, size_t n, size_t require, struct libar2_context *
 				indices[ret - 1] = i + lb(one);
 		}
 	}
+	pthread_mutex_unlock(&data->mutex);
 
 	for (;;) {
 		if (ret < require) {