RayPan
    • Create new note
    • Create a note from template
      • Sharing URL Link copied
      • /edit
      • View mode
        • Edit mode
        • View mode
        • Book mode
        • Slide mode
        Edit mode View mode Book mode Slide mode
      • Customize slides
      • Note Permission
      • Read
        • Only me
        • Signed-in users
        • Everyone
        Only me Signed-in users Everyone
      • Write
        • Only me
        • Signed-in users
        • Everyone
        Only me Signed-in users Everyone
      • Engagement control Commenting, Suggest edit, Emoji Reply
    • Invite by email
      Invitee

      This note has no invitees

    • Publish Note

      Share your work with the world Congratulations! 🎉 Your note is out in the world Publish Note

      Your note will be visible on your profile and discoverable by anyone.
      Your note is now live.
      This note is visible on your profile and discoverable online.
      Everyone on the web can find and read all notes of this public team.
      See published notes
      Unpublish note
      Please check the box to agree to the Community Guidelines.
      View profile
    • Commenting
      Permission
      Disabled Forbidden Owners Signed-in users Everyone
    • Enable
    • Permission
      • Forbidden
      • Owners
      • Signed-in users
      • Everyone
    • Suggest edit
      Permission
      Disabled Forbidden Owners Signed-in users Everyone
    • Enable
    • Permission
      • Forbidden
      • Owners
      • Signed-in users
    • Emoji Reply
    • Enable
    • Versions and GitHub Sync
    • Note settings
    • Note Insights New
    • Engagement control
    • Transfer ownership
    • Delete this note
    • Save as template
    • Insert from template
    • Import from
      • Dropbox
      • Google Drive
      • Gist
      • Clipboard
    • Export to
      • Dropbox
      • Google Drive
      • Gist
    • Download
      • Markdown
      • HTML
      • Raw HTML
Menu Note settings Note Insights Versions and GitHub Sync Sharing URL Create Help
Create Create new note Create a note from template
Menu
Options
Engagement control Transfer ownership Delete this note
Import from
Dropbox Google Drive Gist Clipboard
Export to
Dropbox Google Drive Gist
Download
Markdown HTML Raw HTML
Back
Sharing URL Link copied
/edit
View mode
  • Edit mode
  • View mode
  • Book mode
  • Slide mode
Edit mode View mode Book mode Slide mode
Customize slides
Note Permission
Read
Only me
  • Only me
  • Signed-in users
  • Everyone
Only me Signed-in users Everyone
Write
Only me
  • Only me
  • Signed-in users
  • Everyone
Only me Signed-in users Everyone
Engagement control Commenting, Suggest edit, Emoji Reply
  • Invite by email
    Invitee

    This note has no invitees

  • Publish Note

    Share your work with the world Congratulations! 🎉 Your note is out in the world Publish Note

    Your note will be visible on your profile and discoverable by anyone.
    Your note is now live.
    This note is visible on your profile and discoverable online.
    Everyone on the web can find and read all notes of this public team.
    See published notes
    Unpublish note
    Please check the box to agree to the Community Guidelines.
    View profile
    Engagement control
    Commenting
    Permission
    Disabled Forbidden Owners Signed-in users Everyone
    Enable
    Permission
    • Forbidden
    • Owners
    • Signed-in users
    • Everyone
    Suggest edit
    Permission
    Disabled Forbidden Owners Signed-in users Everyone
    Enable
    Permission
    • Forbidden
    • Owners
    • Signed-in users
    Emoji Reply
    Enable
    Import from Dropbox Google Drive Gist Clipboard
       Owned this note    Owned this note      
    Published Linked with GitHub
    • Any changes
      Be notified of any changes
    • Mention me
      Be notified of mention me
    • Unsubscribe
    ```cpp #include <pthread.h> #include <stdlib.h> #include <unistd.h> #include <stdio.h> int myglobal; void *thread_function(void *arg) { for (int i = 0; i < 20; i++) { int j = myglobal; j = j + 1; printf("."); fflush(stdout); sleep(1); myglobal = j; } return NULL; } int main(void) { pthread_t mythread; if (pthread_create(&mythread, NULL, thread_function, NULL)) { printf("error creating thread."); abort(); } for (int i = 0; i < 20; i++) { myglobal = myglobal + 1; printf("o"); fflush(stdout); sleep(1); } if (pthread_join(mythread, NULL)) { printf("error joining thread."); abort(); } printf("\nmyglobal equals %d\n",myglobal); exit(0); } ``` 執行結果 ``` o.o.o.o.o..o.o.o.o.o.o.o.oo..oo.o..o.oo. myglobal equals 21 ``` myglobal 起始值為 0,主執行緒與新執行緒各將 myglobal 增加 20,照理說應該是 40,但輸出 21 原因為何? 因為兩條執行緒同時執行,當 thread_fucntion 把 j 值寫回 myglobal,他會複寫主執行緒中的修改。 此便為程式同步執行所造成的競爭危害問題。 --- >> 將 concurrent / concurrency 統一翻譯為「並行」 [name=jserv] * ### 競爭危害(race conditon) 兩個工作同時存記憶體的過程中,並不會即時更新記憶體資料,造成最後用舊的資料計算,產生錯誤結果。 * ### 臨界區域(critical section) 共用資源被並行存取可能造成不可預期的錯誤,所以有些程式共用資源的存取會被保護,這些保護區域就稱為臨界區域。其只能被一個 process 執行。 * 解決 critical section 必須滿足3個條件: * #### Mutual Exclusion(互斥鎖,簡稱 Mutex) Mutex 是一種並行操作(concurrency control),用在多執行緒程式設計中,目的為避免競爭危害(race conditions),不會有兩個 processes 同時進入同一個臨界區域(critical section)。 * #### Progress (1)當沒有 process 在 critical section 內執行時,不能阻擋其他請求進入 critical section 的 processes 進入。 (2)從那些想進入 critical section 之 processes 決定一個 process 可以進入 critical section 的決定時間是有限的。(No Deadlock) * #### Bounded Waiting Process 要求進入 critical section 有個等待時間的bound、一個限制,也就是一個 process 不能一直佔住 critical section。i.e. 若有 n 個 processes 皆想進入 critical section,則任一 process 至多等 n-1 次後即可進入。(No Starvation) * 解決 critical section 的方法: * #### Software Based Solution - Peterson's Solution * #### Synchronization Hardware 以硬體方式解決,藉由 lock 來保護 critical section。 * #### Atomic Operation [參考:makelinux-Atomic Operation](http://www.makelinux.net/books/lkd2/ch09lev1sec1) 原子操作 (Atomic Operation) 意思是不可被中斷的操作,如同原子一般是不可再分割的。 * #### Test and Set (TAS) * #### Compare and Swap (CAS) * 為一種在多執行緒系統中實現同步的原子操作,操作過程虛擬碼如下 ```clike CAS(*位址, 舊值, 新值){ if(*位址!=舊值) return 0 //失敗 *位址 = 新值 return 1 //成功 } ``` 使用上會記錄某塊記憶體的舊值,如果這塊記憶體在期間內沒被修改過,舊值會與記憶體位址中的值相同,這時CAS操作會成功執行並將該記憶體位址存的值變為新值;反之,若在期間被修改過,則記憶體位址中的值會與舊值不同,CAS操作失敗,新值不被寫入。 * #### Semaphore * #### mutex 與 semaphore 的差異? 參考:[Mutexes VS Semaphores 大揭秘](http://swem.logdown.com/posts/177651-mutexes-and-semaphores-demystified)、[Semaphore 和 Mutex 的差異](https://hackpad.com/ep/pad/static/5TOjUJI2rKu) Mutex 和 Semphore 都是在解決同步問題的機制,用在保護 critical section,確保多個 process 同時存取資源時,執行結果不會因為執行程序的時間先後的影響而導致錯誤。然而其差異為: * semaphore 是一個 counter 的概念,可以設定有多少個 process 存取資源,如果存取的 process 數量到達上限,其他要求存取資源的 process 會被送到 wait queue 中,等待有 process 釋放資源,再繼續執行。 * Mutex 就像是一把鑰匙,會記錄擁有這把鑰匙的 owner 是哪個 process ,process 需要取得鑰匙後才能進到 critical section 存取資源,等到存取完成後才釋放出這把鑰匙的擁有權,達到 mutual exclusion。 * mutex 與 binary semapohre 差異為何? mutex 會導致 priority inversion。 * #### 優先權倒置 (Priority Inverse) [參考:Jserv's blog-Priority inversion 簡介](http://blog.linux.org.tw/~jserv/archives/001299.html) 在此情形下,高優先權的 task 會被低優先權的搶先(preempted),使得兩個 task 的相對優先權被顛倒。 如下圖所示,當具備中度優先權的 task (簡稱 M) 搶先 (preempt) 一個原本享有 resource 的低優先權的 task (簡稱 L),而該 resource 又是一個高優先權的 task (簡稱 H) 所等待。問題就在於,H 與 L 共享 resource,當 L 被 preempt 時,就該放下 resource,這是合理的行為,而原本 H 就在等待 resource 的釋放,因為隨後就會使用到。但問題是,這段 latency 中,M 把這個規則打破,先行 preempt 了 L,也就把 L 的 resource 給「搶走」,這下有趣的事情就發生了。原本 L 與 H 具備相對高低的優先權差異,但因為 M 的介入,造成延遲,這樣的過程可能就讓 H 發生超出 deadline 的情況。我們稱這樣逆轉 L 與 H 的優先權順序的現象為 priority inversion。 ![](https://i.imgur.com/Wy9kRk2.gif) * ### 經典同步問題 * #### Bounded-Buffer Problem * 描述 兩個行程 (生產者、消費者) 共享 buffer(由容量可分為 Bounded (有限的) 及 Unbounded (無限的) ,此討論有限情況),生產者負責產生資料並放入 buffer,然後重複這個過程;同時消費者從 buffer 消耗資料。本問題的關鍵在於,確保 buffer 滿時生產者不再放入資料;buffer 空時消費者不再消耗資料。 * 解決方法 生產者在 buffer 滿時進入睡眠 (sleep) 或放棄資料,當消費者消耗資料時再喚醒生產者放資料進 buffer;同理,消費者在 buffer 空時進入睡眠,當生產者放入資料時再喚醒消費者消費資料。可以透過 semapohres 達成,然而不完整解法可能造成死結 (deadlock),即兩個行程都進入睡眠等待被喚醒。 * #### Readers and Writers Problem * 描述 多個執行緒想要同時存取 (access) 同一個共用資源 (shared resource),有些想要讀 (read) 稱 reader 、有些想要寫 (write) 稱 writer,共用資源可以同時被讀但不能同時被寫。 * First readers-writers problem writers 可能 starve * Second readers-writers problem readers 可能 starve * 解決方法 Readers-Writer lock * #### Dining-Philosophers Problem ![](https://i.imgur.com/L6hsfYm.jpg) * 描述 五個哲學家只能思考和吃,他們圍著中間放有一碗飯的圓桌就座,每個哲學家左右手各有一支筷子,當一個哲學家餓了就會試著拿起左右各一支筷子吃飯,哲學家彼此無法得知其他人要在吃或在思考。問題的關鍵在於如何不讓哲學家 starve,如果哲學家都拿左邊的筷子,就會永遠等待右邊的筷子造成 Deadlock。 * 解決方法 Represent each chopstick with a semaphore --- 論文研讀:[A Pragmatic Implementation Of Non-Blocking Linked-Lists](https://timharris.uk/papers/2001-disc.pdf) ` This paper present a novel implementation of linked-lists which is non-blocking, linearizable and which is based on the compare-and-swap(CAS)operation found on contemporary processoers. ` 問題:若是在移除節點(10)的同時,插入新節點(20),可能會造成新插入節點的遺失 ![](https://i.imgur.com/MAjVTbn.png) 解決:作者提出使用兩個分開的 CAS (compare and swap) 分兩階段移除節點 1. Logically deleted : 先 mark 欲刪除節點(10)的 next 指標 2. Physically deleted : 刪除該節點(10) 所以假設要並行插入新節點(20)時,程式會觀察(10)是否已被 logically deleted,若有則嘗試 logically delete 它再插入新節點。 --- ### Thread Pool實作原理 在 Thread-Per-Message 模式中,當每次請求來到,就建立一個新的執行緒,用完就不再使用,然而執行緒的建立需要系統資源,對於一個接受許多請求的情況,不斷的建立新執行緒,會導致系統效能的降低。 >> 搭配閱讀: [Thread pools and work queues](http://www.ibm.com/developerworks/java/library/j-jtp0730/index.html) [name=jserv] 若能重複使用所建立的執行緒,而不是用完就丟,可以有效的重複利用資源。在 Worker Thread 模式的範例中,預先建立好執行緒,當請求佇列有新請求時,通知等待的執行緒取出請求進行處理,其實就是一種重複使用執行緒的方式。 示意圖如下: ```graphviz digraph hierarchy { nodesep=0.5 // increases the separation between nodes node [color=black,fontname=Courier,shape=box] //All nodes will this shape and colour edge [color=black, style=dashed] // All the lines look like this Manager->{Worker1 Worker2 Worker3} } ``` 為了分配工作,在 worker thread 實作 Task 的機制。每個主要的操作會先被放進 task queue 裡頭,空閒的 thread 再從 task queue 裡頭提取 task 執行,如下面的步驟: ```graphviz digraph { 開始Process[shape="box", style=rounded]; 是否為結束Task[shape="diamond"]; 結束Process[shape="box", style=rounded]; 開始Process->提取Task->是否為結束Task 是否為結束Task->重發結束Task[label="是"] 重發結束Task->結束Process 是否為結束Task->執行Task[label="否"] 執行Task->提取Task; } ``` 大多數 thread pool 實做都離不開 lock 的使用,如 pthread_mutex 結合 (condition variable) pthread_cond。一般來說,lock 的使用對於程式效能影響較大,雖然現有的 pthread_mutex 在 lock 的取得 (acquire) 與釋放,已在 Linux 核心和對應函式庫進行效能提昇,但我們仍會希望有不仰賴 lock 的 thread pool 的實做。 ![](https://i.imgur.com/ktljGJC.png) 如上圖所示,workqueue (工作佇列) 由 main thread (主執行緒) 和 worker thread 共享,main thread 將任務放進 workqueue,worker thread 從 workqueue 中取出任務執行。要注意到,共享 workqueue 的操作必須在 mutex 的保護下安全進行,main thread 將任務放進 workqueue 時,若偵測到目前待執行的工作數目小於 worker thread 總數,則要透過 condition variable 喚醒可能處於等待狀態的 worker thread。 ### Lock-Free Thread Pool 實作原理 ![](https://i.imgur.com/GjNMAjd.png) 為解決 lock-free 所面臨的議題,我們一定要避免共享資源的競爭 (contention),因此將共享 workqueue 加以拆分成每 worker thread 一個 workqueue 的方式,如上圖。 >> 沒有顯著進展!你們到底做了什麼? [name=jserv] #### 未修改前 * sort 1 thread `mutrace: Total runtime is 0.734 ms.` * sort 2 thread `mutrace: Total runtime is 1.097 ms.` >發現thread開越多時間越長[name=儂偉] #### 參考[mbrossard 完整的 ThreadPool](https://github.com/mbrossard/threadpool/blob/master/src/threadpool.c) 原始碼 去優化threadPool #### 未優化(4thread) ``` Mutex # Locked Changed Cont. tot.Time[ms] avg.Time[ms] max.Time[ms] Flags 1 177 43 27 0.048 0.000 0.002 Mx.--. 0 20 15 9 0.011 0.001 0.001 M-.--. 2 13 8 0 0.004 0.000 0.000 M-.--. |||||| /||||| Object: M = Mutex, W = RWLock /|||| State: x = dead, ! = inconsistent /||| Use: R = used in realtime thread /|| Mutex Type: r = RECURSIVE, e = ERRRORCHECK, a = ADAPTIVE /| Mutex Protocol: i = INHERIT, p = PROTECT / RWLock Kind: r = PREFER_READER, w = PREFER_WRITER, W = PREFER_WRITER_NONREC mutrace: Note that the flags column R is only valid in --track-rt mode! ``` >> 需要詳述優化策略,以及實作 fine-grained locks 和接著做到 lock-free [name=jserv] 著手修改 task_run ```c= pthread_mutex_lock(&(pool->mutex)); while ( pool->size == 0) { pthread_cond_wait(&(pool->cond), &(pool->mutex)); } if (pool->size == 0) break; pthread_mutex_unlock(&(pool->mutex)); task_t *_task = tqueue_pop(pool); ``` #### 優化後(thread) ``` Mutex # Locked Changed Cont. tot.Time[ms] avg.Time[ms] max.Time[ms] Flags 1 20 17 5 0.009 0.000 0.002 M-.--. 0 13 9 2 0.020 0.002 0.007 Mx.--. |||||| /||||| Object: M = Mutex, W = RWLock /|||| State: x = dead, ! = inconsistent /||| Use: R = used in realtime thread /|| Mutex Type: r = RECURSIVE, e = ERRRORCHECK, a = ADAPTIVE /| Mutex Protocol: i = INHERIT, p = PROTECT / RWLock Kind: r = PREFER_READER, w = PREFER_WRITER, W = PREFER_WRITER_NONREC mutrace: Note that the flags column R is only valid in --track-rt mode! ``` >>發現lock數量大大減少[name=儂偉] ``` ./sort 1 8 input unsorted data line-by-line sorted results: [8] [7] [6] [5] [4] [3] [2] [1] sorted results: [1] [2] [3] [4] [5] [6] [7] [8] ./sort 2 8 input unsorted data line-by-line sorted results: [8] [7] [6] [5] [4] [3] [2] [1] ``` >> 只要開超過2個thread就執行失敗,待解決[name=儂偉] >>發現原因[mbrossard 的 ThreadPool](https://github.com/mbrossard/threadpool/blob/master/src/threadpool.c) 直接在cut套用會產生鎖兩次 造成死鎖,重新動手改原本的thread_pool[name=儂偉] #### thread_pool 加入pthread_cond_t機制 參考[mbrossard 的 ThreadPool](https://github.com/mbrossard/threadpool/blob/master/src/threadpool.c) ```c= typedef struct { pthread_t *threads; uint32_t count; task_t *head, *tail; pthread_mutex_t mutex; pthread_cond_t cond; uint32_t size; } tpool_t; ``` * 將tqueue與tpool_t合併為tpool_t,使pool結構更簡單 #### 新增bench ```shell= bench:intput.txt for i in 1 2 4 8 16 32 64; do \ while read line;do \ #逐行讀取 echo $$line ; \ done < intput.txt | ./sort $$i $ 10000 ;\ done # '<':將原本需要由鍵盤輸入的資料,改由檔案內容('input.txt')來取代 # '|':管線命令。僅能處理經由前面一個指令傳來的正確資訊,也就是 standard output 的資訊,對於 stdandard error 並沒有直接處理的能力。 ``` #### 產生測資 * makefile ```shell= random_num: random_num.c $(CC) $(CFLAGS) $^ -o $@ intput.txt: random_num ./random_num $ 10000 ``` * random_num.c 利用洗牌來達到均勻分布的目的 ```c= /*random_num.c*/ srand(time(NULL)); int num_count, i; FILE *input = fopen("input.txt", "w"); num_count = atoi(argv[1]); int *arr; int pos , tmp; arr = (int *) malloc(sizeof(int) * num_count); for (i = 0; i < num_count; i++) //產生num_count筆數據 arr[i] = i + 1; for (i = 0; i < num_count; i++) { //進行洗牌 pos = rand() % num_count; tmp = arr[i]; arr[i] = arr [pos]; arr[pos] = tmp; //進行arr[i]及arr[pos]交換 } for (i = 0; i < num_count; i++) fprintf(input, "%d\n", arr[i]); free(arr); fclose(input); ``` >> 程式碼縮排為 4 個空白,不是 Tab,請修正 [name=jserv] >> 已修正[name=儂偉] #### 比較圖 ![](https://i.imgur.com/k5e2FsT.png) >>由於實驗用CPU為4thread,所以4thread最快,並且超過4的時間就會越長[name=儂偉] --- ## Lock contention * 閱讀[Lock contention](https://en.wikipedia.org/wiki/Lock_(computer_science)#Granularity) * lock overhead : 使用 lock 的額外資源,像是記憶體空間、初始化和摧毀 lock 的 CPU 時間等等。 * lock contention : 這會發生在一個程序或是執行緒要去請求一個正在被使用中的 lock 時。 * deadlock : 當兩個 task 正在搶奪一個正被另一個 task 使用爭的 lock。除非有什麼完成了,否則這兩個 task 會一直等待。 * granularity : 指的是 lock 中所保護的 data 數量。如果選擇少量的 lock 去保護大量的資料的話,會有較少的 lock overhead 但是較差的 concurrency 表現,這是因為會有 lock contention 的問題產生。 The more coarse the lock (這個不知道怎麼翻), lock 越容易去停止不相關的多工請求。 >> coarse 和 fine-grained 互為反義字,前者表示較為粗糙的切割和處理方式,而後者則較為細緻,以 lock 來說,就是「大而粗略」和「細緻且多處」的差異。 [name=jserv] * 閱讀[Measuring Lock Contention](http://0pointer.de/blog/projects/mutrace.html) >> 做實驗了嗎?你們設計哪些實驗?不要只是「看」,這樣無從驗證自己的認知 [name=jserv] >> 提醒中英文關鍵字間要以空白區隔喔![color=red][name=課程助教] #### task_run未加入pthread_cond_t機制 ``` Mutex # Locked Changed Cont. tot.Time[ms] avg.Time[ms] max.Time[ms] Flags 1 177 43 27 0.048 0.000 0.002 Mx.--. 0 20 15 9 0.011 0.001 0.001 M-.--. 2 13 8 0 0.004 0.000 0.000 M-.--. |||||| /||||| Object: M = Mutex, W = RWLock /|||| State: x = dead, ! = inconsistent /||| Use: R = used in realtime thread /|| Mutex Type: r = RECURSIVE, e = ERRRORCHECK, a = ADAPTIVE /| Mutex Protocol: i = INHERIT, p = PROTECT / RWLock Kind: r = PREFER_READER, w = PREFER_WRITER, W = PREFER_WRITER_NONREC mutrace: Note that the flags column R is only valid in --track-rt mode! ``` #### task_run加入pthread_cond_t機制 ``` Mutex # Locked Changed Cont. tot.Time[ms] avg.Time[ms] max.Time[ms] Flags 1 53 30 11 0.070 0.001 0.016 M-.--. 0 20 12 3 0.009 0.000 0.002 M-.--. 2 13 5 0 0.004 0.000 0.000 M-.--. |||||| /||||| Object: M = Mutex, W = RWLock /|||| State: x = dead, ! = inconsistent /||| Use: R = used in realtime thread /|| Mutex Type: r = RECURSIVE, e = ERRRORCHECK, a = ADAPTIVE /| Mutex Protocol: i = INHERIT, p = PROTECT / RWLock Kind: r = PREFER_READER, w = PREFER_WRITER, W = PREFER_WRITER_NONREC mutrace: Note that the flags column R is only valid in --track-rt mode! ``` >> 發現lock數量跟contend都大幅降低[name=儂偉] --- ### Lock-free * 嘗試透過 atomic 操作來實作 >> 說好的程式碼呢? [name=jserv] ```c= static void *task_run(void *data) { (void) data; while (1) { task_t *ret , *ret_last, *ret_next, *the_pool_head, *tmp; ret = tmp = pool->tail; ret_last = ret->last; ret_next = ret->next; the_pool_head = pool->head; if (ret != NULL && !is_marked_ref(ret)) { if (CAS_PTR(&pool->tail, ret, ret_last) == ret) { ret = get_unmarked_ref(ret_last); ret_last = ret->last; ret_next = ret->next; if (ret && !is_marked_ref(ret_next)) { if (CAS_PTR(&ret->next, ret_next, NULL) == ret_next) { ret->next = NULL; FAD_U32(&(pool->size)); task_t *_task = tmp; if (_task) { if (!_task->func) { tqueue_push(pool, _task); break; } else { _task->func(_task->arg); free(_task); } } } } else if (!ret && !is_marked_ref(the_pool_head)) { if (CAS_PTR(&pool->head, the_pool_head, NULL) == the_pool_head) { FAD_U32(&(pool->size)); task_t *_task = tmp; if (_task) { if (!_task->func) { tqueue_push(pool, _task); break; } else { _task->func(_task->arg); free(_task); } } } } } } else if (ret == NULL) { task_t *_task = tmp; if (_task) { if (!_task->func) { tqueue_push(pool, _task); break; } else { _task->func(_task->arg); free(_task); } } } } pthread_exit(NULL); } ``` >> 目前 mark 部份仍有問題 解決中... --- 讀書: * [Common threads: POSIX threads explained](http://www.ibm.com/developerworks/library/l-posix2/) * [Measuring Lock Contention](http://0pointer.de/blog/projects/mutrace.html) * [More Mutrace](http://0pointer.net/blog/projects/mutrace2.html) * [POSIX Threads for Windows](https://www.sourceware.org/pthreads-win32/manual/pthread_mutex_init.html) --- 參考資料 * [A07: mergesort-concurrent](https://hackmd.io/s/rJ-GWtJ0) * [宅學習-作業系統](http://sls.weco.net/node/21326) * [Thread Pool 模式](http://openhome.cc/Gossip/DesignPattern/ThreadPool.htm) * [Toward Concurrency](https://hackmd.io/s/H10MXXoT) * [Wikipedia-CAS](https://en.wikipedia.org/wiki/Compare-and-swap) * [Wikipedia-FAI](https://en.wikipedia.org/wiki/Fetch-and-add) * [gcc Atomic-Builtins doc](https://gcc.gnu.org/onlinedocs/gcc-4.1.2/gcc/Atomic-Builtins.html)

    Import from clipboard

    Paste your markdown or webpage here...

    Advanced permission required

    Your current role can only read. Ask the system administrator to acquire write and comment permission.

    This team is disabled

    Sorry, this team is disabled. You can't edit this note.

    This note is locked

    Sorry, only owner can edit this note.

    Reach the limit

    Sorry, you've reached the max length this note can be.
    Please reduce the content or divide it to more notes, thank you!

    Import from Gist

    Import from Snippet

    or

    Export to Snippet

    Are you sure?

    Do you really want to delete this note?
    All users will lose their connection.

    Create a note from template

    Create a note from template

    Oops...
    This template has been removed or transferred.
    Upgrade
    All
    • All
    • Team
    No template.

    Create a template

    Upgrade

    Delete template

    Do you really want to delete this template?
    Turn this template into a regular note and keep its content, versions, and comments.

    This page need refresh

    You have an incompatible client version.
    Refresh to update.
    New version available!
    See releases notes here
    Refresh to enjoy new features.
    Your user state has changed.
    Refresh to load new user state.

    Sign in

    Forgot password

    or

    By clicking below, you agree to our terms of service.

    Sign in via Facebook Sign in via Twitter Sign in via GitHub Sign in via Dropbox Sign in with Wallet
    Wallet ( )
    Connect another wallet

    New to HackMD? Sign up

    Help

    • English
    • 中文
    • Français
    • Deutsch
    • 日本語
    • Español
    • Català
    • Ελληνικά
    • Português
    • italiano
    • Türkçe
    • Русский
    • Nederlands
    • hrvatski jezik
    • język polski
    • Українська
    • हिन्दी
    • svenska
    • Esperanto
    • dansk

    Documents

    Help & Tutorial

    How to use Book mode

    Slide Example

    API Docs

    Edit in VSCode

    Install browser extension

    Contacts

    Feedback

    Discord

    Send us email

    Resources

    Releases

    Pricing

    Blog

    Policy

    Terms

    Privacy

    Cheatsheet

    Syntax Example Reference
    # Header Header 基本排版
    - Unordered List
    • Unordered List
    1. Ordered List
    1. Ordered List
    - [ ] Todo List
    • Todo List
    > Blockquote
    Blockquote
    **Bold font** Bold font
    *Italics font* Italics font
    ~~Strikethrough~~ Strikethrough
    19^th^ 19th
    H~2~O H2O
    ++Inserted text++ Inserted text
    ==Marked text== Marked text
    [link text](https:// "title") Link
    ![image alt](https:// "title") Image
    `Code` Code 在筆記中貼入程式碼
    ```javascript
    var i = 0;
    ```
    var i = 0;
    :smile: :smile: Emoji list
    {%youtube youtube_id %} Externals
    $L^aT_eX$ LaTeX
    :::info
    This is a alert area.
    :::

    This is a alert area.

    Versions and GitHub Sync
    Get Full History Access

    • Edit version name
    • Delete

    revision author avatar     named on  

    More Less

    Note content is identical to the latest version.
    Compare
      Choose a version
      No search result
      Version not found
    Sign in to link this note to GitHub
    Learn more
    This note is not linked with GitHub
     

    Feedback

    Submission failed, please try again

    Thanks for your support.

    On a scale of 0-10, how likely is it that you would recommend HackMD to your friends, family or business associates?

    Please give us some advice and help us improve HackMD.

     

    Thanks for your feedback

    Remove version name

    Do you want to remove this version name and description?

    Transfer ownership

    Transfer to
      Warning: is a public team. If you transfer note to this team, everyone on the web can find and read this note.

        Link with GitHub

        Please authorize HackMD on GitHub
        • Please sign in to GitHub and install the HackMD app on your GitHub repo.
        • HackMD links with GitHub through a GitHub App. You can choose which repo to install our App.
        Learn more  Sign in to GitHub

        Push the note to GitHub Push to GitHub Pull a file from GitHub

          Authorize again
         

        Choose which file to push to

        Select repo
        Refresh Authorize more repos
        Select branch
        Select file
        Select branch
        Choose version(s) to push
        • Save a new version and push
        • Choose from existing versions
        Include title and tags
        Available push count

        Pull from GitHub

         
        File from GitHub
        File from HackMD

        GitHub Link Settings

        File linked

        Linked by
        File path
        Last synced branch
        Available push count

        Danger Zone

        Unlink
        You will no longer receive notification when GitHub file changes after unlink.

        Syncing

        Push failed

        Push successfully