亚洲国产天堂久久综合,亚洲美女激情视频,亚洲理论电影在线观看

Java concurrency in practice

矛_��同学 — Sun, 18 Dec 2011 14:56:00 GMT

12�?8�?chapter 2 Thread Safe

1.what is thread safe ?
多个�U�程�q�发讉K��Ӟ��无需额外的同步或协调�Q�能保证执行的正��性，则此�c�Mؓ�U�程安全。（�l�合Java�Ҏ(gu��)��，每个thread都会有独立的stack�I�间�Q�因��Z��在存在状态的�c�，才会��D��thread safe问题。无状态的�c�，��L��U�程安全的。）

A class is thread-safe if it behaves correctly when accessed from multiple threads, regardless of the scheduling or interleaving of the execution of those threads by the runtime environment, and with no additional synchronization or other coordination on the part of the calling code.

2.Atomicity
如：servlet通过count属性记录访问数�Q�此�c�L��U�程不安全的。原因是�Q�count++�Q��ƈ非原子操作；实际�?个步骤，��d��count、加1、写回主存，其中��M��阶段其它�U�程均可�q�行操作�Q�若count++操作时原子时�Q�是保证计数功能是线�E�安全的�?br />race condition : The possibility of incorrect results in the presence of unlucky timing is so important in concurrent programming that it has a name: a race condition.

The most common type of race condition is check-then-act, where a potentially stale observation is used to make a decision on what to do next. 最常见的race condition操作�Q?check-then-act' -> '��d��-操作'�Q�lazy init�Q�，此时会导致��用一个stale状态去军_��下一步的操作。（或count++操作�Q�read-modify-write�c�d��Q?br />

Both LazyInitRace and UnsafeCountingFactorizer contained a sequence of operations that needed to be atomic, or indivisible, relative to other operations on the same state. To avoid race conditions, there must be a way to prevent other threads from using a variable while we're in the middle of modifying it, so we can ensure that other threads can observe or modify the state only before we start or after we finish, but not in the middle.
什么是原子�?�Q�一个操作，要么全部完成�Q�要么没有发生，不会存在中间状态�?br />
3.Locking

But if we want to add more state to our servlet, can we just add more thread-safe state variables ? 必须保证相关状态操作的原子性。Java provides a built-in locking mechanism for enforcing atomicity: the synchronized block.
Every Java object can implicitly act as a lock for purposes of synchronization。锁是属于实例的�Q�类�U�别对应的实例是Class对象。且在�Q意时��M��有一个线�E�持有此锁。某�U�程�q�去被synchronized保护的代码块�Q�必��首先获取到此锁�Q�否则wait直到获取此锁�?/div>

3.1 Reentrancy

But because intrinsic locks are reentrant, if a thread tries to acquire a lock that it already holds, the request succeeds.

Reentrancy is implemented by associating with each lock an acquisition count and an owning thread. 实现依赖锁被获取的次数和拥有此锁的线�E��?br />当每�ơ获取到此锁�Ӟ��acquisition count�?�Q�当退出synchronized�Ӟ��count�?�Q�直到count�?�Ӟ��释放此锁�?/div>惌��Q�若synchronized是非重入的，会有什么结�?�Q?br />
4.Guarding state with lock

sync保证compound actions原子性；对共享变量的��M��操作都需要显�C�的去sync�Q�这是一个问题；
对象内徏的lock与属性�ƈ没有固有的关�p�，内徏lock方便��d��建sync policy��M��证thread safe�Q?br />一�U�简单的�Ҏ(gu��)��Q�对所有状态统一使用内徏的lock�Q�意味着在�Q何action code path都需要去sync�Q�如�Q�Vector�Q?br />For every invariant that involves more than one variable, all the variables involved in that invariant must be guarded by the same lock.
滥用sync会导致性能的问题�?br />
5.Liveness and Performance
sync会导致性能问题。原则：sync块尽量small�Q�sync操作不能分割太细。如何做�Q�需要从safe、simple、performance考虑�?br />

Deciding how big or small to make synchronized blocks may require tradeoffs among competing design forces, including safety (which must not be compromised), simplicity, and performance.通常��单性和性能是相互牵制的�Q�不要�ؓ了性能�q�早的牺牲简单性（是潜在的对safe的妥协）�?br />Avoid holding locks during lengthy computations or operations at risk of not completing quickly such as network or console I/O.

矛_��同学 2011-12-18 22:56 发表评论

信息的表�C�和处理

矛_��同学 — Sat, 24 Apr 2010 13:39:00 GMT

计算��Z��二进制表�C�和存储信息�?/span>

计算机内部数据的存储�Q�根据数据的�Ҏ(gu��)��分为：

1�?nbsp;无符��h��?/span>

2�?nbsp;有符��h��敎ͼ�补码�Q?/span>

3�?nbsp;��点�?/span>

明确的关键点�Q?/span>

1�?nbsp;计算机内部是通过有限的位�~�码一个数字，因此存在“溢出”现象�Q�因��C�的范围是有限的�?/span>

2�?nbsp;大多数计��机使用8位的块，做�ؓ最��的可寻址的存储器单位�?/span>

3、不同的数据�c�d��Q�数据大��是不一��L(f��ng)��。（准确的要依赖机器和编译器�Q��?/span>

4、对象的地址使用字节序列中的最��的地址。（�|�络传输、跨�q�_��时考虑字节序的问题�Q?br />

一、无�W�号整数

基本�?#8220;二进�?#8221;表示�?/span>

二、有�W�号整数 --- 补码 �?span style="color:#993300">同余�?/span>

使用补码�Q�减法也可按加法来处理（ALU不用��d��现减法运��）。（两个用补码表�C�的数据相加时候，��符号位和其它位�l�一处理�Q�如果最高位(�W�号�?/span>)有进位，则进位被舍弃。）

最高有效位为符号位�Q?/span>1负数�Q?/span>0正数�?Java只支持有�W�号整数)

三、��Q�Ҏ(gu��)��

划分��Z��个域�Q?/span>1.�W�号 2.有效位（二进制小敎ͼ� 3. 指数位（2的幂�Q�对��点数加权）

单精度：1 = 23 =8 �Q?/span> 双精�?/span> �Q?/span>1 = 52 = 11�?/span>

�Ҏ(gu��)��：�_�ֺ�有限�Q��Q点运��是不可�l�合和交换的�?/span>

四�?#8220;�?#8221;概念

“�?/span>”是指一个计量系�l�的计数范围.如时钟等.计算��Z��可以看成一个计量机�?/span>,它也有一个计量范�?/span>,即都存在一�?/span>“�?/span>”.例如�Q?/span>
旉��的计量范围是0�?/span>11,�?/span>=12.

   表示n位的计算��量范围是0�?/span>2(n)-1,�?/span>=2�Q?/span>n�Q?/span>.【注�Q?/span>n表示指数�?/span>
        “�?/span>”实质上是计量器��?/span>“溢出”的量,它的值在计量器上表示不出�?/span>,计量器上只能表示出模的余�?/span>.��M��有模的计量器,均可化减法�ؓ加法�q�算. 例如�Q?/span> 假设当前旉��指向10�?/span>,而准��时间是6�?/span>,调整旉��可有以下两种拨法�Q?/span>
              1、倒拨4��时,卻I��10-4=6
              2、顺�?/span>8��时�Q?/span>10+8=12+6=6
        在以12模的�pȝ��?/span>,�?/span>8和减4效果是一��L(f��ng)��,因此凡是�?/span>4�q�算,都可以用�?/span>8来代�?/span>.
        �?/span>“�?/span>”而言,8�?/span>4互�ؓ补数.实际上以12模的�pȝ��?/span>,11�?/span>1,10�?/span>2,9�?/span>3,7�?/span>5,6�?/span>6都有�q�个�Ҏ(gu��)�?/span>.共同的特�Ҏ(gu��)��两者相加等于模.
   对于计算�?/span>,其概念和�Ҏ(gu��)��完全一�?/span>.n位计��机,�?/span>n=8, 所能表�C�的最大数�?/span>11111111,若再�?/span>1�U�Cؓ100000000(9�?/span>),但因只有8�?/span>,最高位1自然丢失.又回�?/span>00000000,所�?/span>8位二�q�制�pȝ��的模�?/span>2(8). 在这��L(f��ng)��pȝ��中减法问题也可以化成加法问题,只需把减数用相应的补数表�C�就可以�?/span>.把补数用到计��机�Ҏ(gu��)��的处理上,��是补码�?br />

矛_��同学 2010-04-24 21:39 发表评论