reduce size of dummy task

1 files changed, 27 insertions, 17 deletions

diff --git a/docs/application/thread.md b/docs/application/thread.md
index b3f8d21..59654e7 100644
--- a/docs/application/thread.md
+++ b/docs/application/thread.md
@@ -5,18 +5,18 @@
 ## Object Definitions
 
 ```c
-typedef struct _kpl_task kpl_task;
+typedef struct __kpl_task kpl_task;
 
 typedef void kpl_task_fn(kpl_task *t);
 
 typedef struct _kpl_task {
-    _Atomic bool join_ready;
-    int32_t thread_id;
-    kpl_group *state;
-    kpl_result ret;
     kpl_task *_Atomic next, *join;
-    task_fn *fn;
-} kpl_task;
+    kpl_group *state;
+    kpl_result return_value;
+    kpl_task_fn *fn;
+    int32_t thread_id;
+    _Atomic bool join_ready;
+} task;
 
 #define KPL_TASK_SLAB_SIZE 50
 
@@ -31,7 +31,12 @@ typedef struct _kpl_task_slab {
 #define KPL_MAIN_THREAD 0
 
 typedef struct {
-    kpl_task *_Atomic head, *_Atomic tail, dummy;
+    kpl_task *_Atomic next;
+} kpl_task_dummy;
+
+typedef struct {
+    kpl_task *_Atomic head, *_Atomic tail;
+    kpl_task_dummy dummy;
 } kpl_atomic_queue;
 
 typedef struct {
@@ -96,12 +101,12 @@ typedef void task_fn(task *t);
 #define TASK_STATE_SIZE 55
 
 typedef struct _task {
-    _Atomic bool join_ready;
-    int32_t thread_id;
+    task *_Atomic next, *join;
     void *state[TASK_STATE_SIZE];
     void *return_value;
-    task *_Atomic next, *join;
     task_fn *fn;
+    int32_t thread_id;
+    _Atomic bool join_ready;
 } task;
 
 #define TASK_SLAB_SIZE 50
@@ -117,7 +122,12 @@ typedef struct _task_slab {
 #define MAIN_THREAD 0
 
 typedef struct {
-    task *_Atomic head, *_Atomic tail, dummy;
+    task *_Atomic next;
+} task_dummy;
+
+typedef struct {
+    task *_Atomic head, *_Atomic tail;
+    task_dummy dummy;
 } atomic_queue;
 
 typedef struct {
@@ -170,8 +180,8 @@ void task_slab_free(int32_t thread_id) {
 }
 
 void task_queue_init(int32_t thread_id) {
-    threads[thread_id].queue.head = &threads[thread_id].queue.dummy;
-    threads[thread_id].queue.tail = &threads[thread_id].queue.dummy;
+    threads[thread_id].queue.head = (task*) &threads[thread_id].queue.dummy;
+    threads[thread_id].queue.tail = (task*) &threads[thread_id].queue.dummy;
     threads[thread_id].queue.dummy.next = NULL;
     threads[thread_id].priority = 1;
     threads[thread_id].pool = NULL;
@@ -181,7 +191,7 @@ void task_queue_add(int32_t thread_id, task *t) {
     t->next = NULL;
     task *head = __atomic_exchange_n(&threads[thread_id].queue.head, t, __ATOMIC_SEQ_CST);
     head->next = t;
-    if (t != &threads[thread_id].queue.dummy)
+    if (t != (task*) &threads[thread_id].queue.dummy)
         priority_increment(thread_id);
 }
 
@@ -189,7 +199,7 @@ task *task_queue_next(int32_t thread_id) {
     task *t = NULL;
     for (;;) {
         task *tail = threads[thread_id].queue.tail, *next = tail->next;
-        if (tail == &threads[thread_id].queue.dummy) {
+        if (tail == (task*) &threads[thread_id].queue.dummy) {
             if (!next)
                 break;
             threads[thread_id].queue.tail = next;
@@ -204,7 +214,7 @@ task *task_queue_next(int32_t thread_id) {
         task *head = threads[thread_id].queue.head;
         if (tail != head)
             continue;
-        task_queue_add(thread_id, &threads[thread_id].queue.dummy);
+        task_queue_add(thread_id, (task*) &threads[thread_id].queue.dummy);
         next = tail->next;
         if (next) {
             threads[thread_id].queue.tail = next;