Board bring up_popy-ChinaUnix博客

博文

(2011-03-09 10:47:19)

系统分类: Linux

最近做了部分SoC的验证以及Bring up工作，对整个流程做一个简单的回顾，算是一个小结： Information to cover <ul style="font-family: Georgia;"><li>&nbsp;&nbsp;&nbsp; Schematics (检查电路图) </li></ul><ul style="font-family: Georgia;"><li>&nbsp;&nbsp;&nbsp; PCB(检查PCB板)</li></ul><ul style="font-family: Georgia;"><li>&nbsp;&nbsp;&nbsp; Assembly(检查期间的装配)</li></ul><ul style="font-family: Georgia;"><li>&nbsp;&nbsp;&nbsp; Scientific Method</li></ul>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Observe &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Theory &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; &nbsp; &nbsp; &nbsp;&nbsp; Test &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Analyze <ul style="font-family: Georgia;"><li>&nbsp;&nbsp;&nbsp; Tools</li></ul>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Logic Analyzer &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Oscilloscope &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; JTAG &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; 在<a style="font-family: Georgia;" href="http://www.elinux.org/Board_Bringup_Utilities" target="_blank">http://www.elinux.org/Board_Bringup_Utilities</a>列出了一些有用的软件工具 <ul style="font-family: Georgia;"><li>&nbsp;&nbsp;&nbsp; Test Case</li></ul>

查看全文阅读(268) ┊ 评论(0) ┊ 收藏(0) ┊举报

查看进程使用何种调度器类

(2011-02-16 12:13:46)

分类：瑞士军刀系统分类: Linux

查看全文阅读(324) ┊ 评论(1) ┊ 收藏(0) ┊举报

TDD 的基本流程

(2011-01-21 16:33:57)

系统分类: Linux

&nbsp;&nbsp;&nbsp; * Think about what you want to do. &nbsp;&nbsp;&nbsp; * Think about how to test it. &nbsp;&nbsp;&nbsp; * Write a small test. Think about the desired API. &nbsp;&nbsp;&nbsp; * Write just enough code to fail the test. &nbsp;&nbsp;&nbsp; * Run and watch the test fail. (The test-runner, if you're using something like JUnit, shows the "Red Bar"). Now you know that your test is going to be executed. &nbsp;&nbsp;&nbsp; * Write just enough code to pass the test (and pass all your previous tests). &nbsp;&nbsp;&nbsp; * Run and watch all of the tests pass. (The test-runner, if you're using JUnit, etc., shows the "Green Bar"). If it doesn't pass, you did something wrong, fix it now since it's got to be something you just wrote. &nbsp;&nbsp;&nbsp; * If you have any duplicate logic, or inexpressive code, refactor to remove duplication and increase expressiveness -- this includes reducing coupling and increasing cohesion. &nbsp;&nbsp;&nbsp; * Run the tests again, you should still have the Green Bar. If you get the Red Bar, then you made a mistake in your refactoring. Fix it now and re-run. &nbsp;&nbsp;&nbsp; * Repeat the steps above until you can't find any more tests that drive writing new code.&nbsp;

查看全文阅读(324) ┊ 评论(1) ┊ 收藏(0) ┊举报

[转载][zz]Locks, SWPs and two Smoking Barriers (Part 2)

(2010-11-02 12:26:58)

分类：驱动程序系统分类: Linux

原文地址：http://blogs.arm.com/software-enablement/locks-swp

In the <a style="font-family: Georgia;" href="http://bit.ly/d4OL8R" target="_blank">last article</a>, I
 explained how to modify SWP code to make use of compiler intrinsics. 
Using intrinsics hides the underlying detail needed to use the load and 
store exclusive instructions <a style="font-family: Georgia;" href="http://infocenter.arm.com/help/topic/com.arm.doc.ddi0406b/index.html" target="_blank">(LDREX and STREX)</a> and the use of memory barriers. In this article I look at implementing atomic memory accesses in assembler. In order to describe memory barriers, what they are and how they should be used, I need to describe two types of memory model, strongly ordered and weakly ordered. The strongly-ordered
 model is very natural for programmers. In this model, the order that a 
program writes data to memory is the order in which the data is observed
 being written into memory, that is, other programs sharing the data 
will "see" the same ordering regardless of the CPU that they are 
executing on. For example, if a CPU writes a new X then writes a 
new Y, all other CPUs that subsequently read Y then a read X, will 
access either the new Y and new X, the old Y and the new X, or old Y and
 the old X. However, because the order of the write is strongly ordered 
as write X first then write Y, n o CPU will access the new Y and the old
 X . Modern CPUs, such as ARM, optimize memory accesses via 
hardware mechanisms avoiding memory accesses when it can, for example by
 using a cache. When it has to access memory, it makes those memory 
accesses as efficient as possible. An example of efficient memory 
accesses is the speculative reading of memory into caches before the CPU
 asks for it. This could happen when a program is accessing an array in 
memory. Whilst the program is working on element N, the memory subsystem
 could be reading element N+1 (and above) into cache ready for use. 
These mechanisms can cause the memory accesses to be reordered . Systems
 that are allowed to reorder memory accesses in this way are called 
weakly ordered memory systems. The upshot of this is that, whilst a 
(program running on a) CPU has a coherent view of its own memory 
accesses, those memory accesses will be observed differently by 
(programs running on) other CPUs. To take my example above, another 
CPU might easily access the new Y and the old X. Unfortunately, 
being able to observe the same order of memory accesses is key when 
accessing memory atomically. If, in the above example, X is a spinlock 
and Y is the data structure being modified within the critical region 
protected by X, it is important that the accesses to X are observed by 
all other CPUs in the system as being before accesses to Y. What you 
need is some mechanism to make sure that all CPUs observe accesses to X 
before the accesses to Y and this mechanism is a memory barrier. Taking load and store exclusives and memory barriers into account, an assembly version of acquiring a spinlock looks like this: <div style="font-family: Georgia;" class="codetop">CODE</div><div style="font-family: Georgia;" class="codemain"><div id="codeText" class="codeText"><ol style="margin: 0pt 1px 0pt 0pt; padding: 5px 0pt;" start="1" class="dp-css"><li>ENTRY (_spin_lock)</li><li>
 mov r1,#1</li><li>
1: ldrex r2,[r0]</li><li>
 cmp r2,#0</li><li>
 strexeq r2, r1, [r0]</li><li>
 cmpeq r2, #0</li><li>
 bne 1b</li><li>
 dmb </li><li>
 mov r0,#0</li><li>
END (_spin_lock)</li></ol></div></div> As
 I described in the previous article, a pair of load and store 
exclusives are used to acquire the lock. The ‘cmpeq r2, #0’ instruction
 checks that the exclusive store to the lock location failed, which it 
will do if another CPU has accessed the lock between the LDREX and STREX
 operations. A failure will cause the code to retry the load exclusive. You
 will note the Data Memory Barrier (DMB) instruction that is issued once
 the lock has been acquired. The DMB guarantees that all memory accesses
 before the memory barrier will be observed by all of the other CPUs in 
the system before all memory accesses made after the memory barrier. 
This makes more sense if you consider that once a lock has been 
acquired, a program will then access the data structure(s) locked by the
 lock. The DMB in the lock function above ensures that accesses to the 
locked data structure are observed after accesses to the lock. The unlock code looks like: <div style="font-family: Georgia;" class="codetop">CODE</div><div style="font-family: Georgia;" class="codemain"><div id="codeText" class="codeText"><ol style="margin: 0pt 1px 0pt 0pt; padding: 5px 0pt;" start="1" class="dp-css"><li>ENTRY (_spin_unlock)</li><li>
 mov r1, #0</li><li>
 dmb</li><li>
 str r1, [r0]</li><li>
END (_spin_unlock)</li></ol></div></div> The
 write to unlock the spinlock uses a simple STR instruction (not a store
 exclusive). This is fine if there is only one owner of the lock, the 
most common case. If a critical region allows more than one CPU to 
access it, load and store exclusives would need to be used. The DMB instruction ensures that accesses to the locked data structure(s) are observed before the lock is unlocked .

查看全文阅读(331) ┊ 评论(0) ┊ 收藏(0) ┊举报

[转载][zz]Locks, SWPs and two Smoking Barriers

(2010-10-29 20:57:57)

分类：驱动程序系统分类: Linux

原文地址：http://blogs.arm.com/software-enablement/locks-swp

Before <a style="font-family: Georgia;" href="http://www.arm.com/products/CPUs/architecture.html" target="_blank">ARMv6</a>, the main synchronisation mechanism was the <a style="font-family: Georgia;" href="http://infocenter.arm.com/help/topic/com.arm.doc.ddi0406b/index.html" target="_blank">SWP instruction</a>. SWP has two aspects, in a <a style="font-family: Georgia;" href="http://en.wikipedia.org/wiki/Uniprocessor_system" target="_blank">uniprocessor system</a> it allows the read and write operations not to be interrupted between them. In a <a style="font-family: Georgia;" href="http://en.wikipedia.org/wiki/MPSoC" target="_blank">multiprocessor system</a>
 it ensures that multiple masters will do the locking. For 
multiprocessor systems with complex memory hierarchies and long memory 
latencies SWP creates performance bottlenecks. This was replaced in the ARMv6 architecture by exclusive loads and stores (<a style="font-family: Georgia;" href="http://infocenter.arm.com/help/topic/com.arm.doc.ddi0406b/index.html" target="_blank">LDREX and STREX</a>).
 This works on the principle of a monitor existing for the location in
 memory. This effectively tags the memory with the identity of the 
agent(s) trying to access it. In a spinlock implementation, an 
exclusive load reads data from the memory, tagging it with its 
identifier. A short number of instructions later, it uses an 
exclusive store to write data to memory but this only works if the tag 
is still valid and the tag will only be valid if some other agent has 
not also modified that location since the exclusive load. At the 
same time that the load and store exclusives were added to the ARM 
architecture, the SWP instruction was depreciated and the architecture 
notes that use of SWP is not guaranteed to work for SMP systems. The 
load and store exclusives and the deprecation of the SWP instructions is
 described in detail in the <a style="font-family: Georgia;" href="http://blogs.arm.com/software-enablement/locks-swps-and-two-smoking-barriers#ARM">ARM ARM [1]</a>. As an aid to removing legacy SWP instructions, <a style="font-family: Georgia;" href="http://www.arm.com/products/CPUs/architecture.html" target="_blank">ARMv7</a> allows you to disable the SWP instruction. In <a style="font-family: Georgia;" href="http://www.arm.com/community/software-enablement/linux.php" target="_blank">ARM SMP Linux</a>
 to help find legacy uses of the SWP instruction, we disable SWP but 
emulate the instructions (via the undefined instruction trap) and log 
those emulations. While this generates extra instruction overhead, it 
ensures that the software operates safely. You should also be aware 
that the SWP instruction does not exist in the <a style="font-family: Georgia;" href="http://en.wikipedia.org/wiki/Thumb-2#Thumb-2" target="_blank">Thumb 2 instruction set</a> and so will see errors if you try and assemble code containing the SWP instruction into the Thumb 2 instruction set. For
 SMP performance we want to the replace the use of the SWP instruction 
with appropriate load and store exclusive instructions in libraries and 
applications. This easiest way to achieve this is to make use of the 
GCC compiler built-ins (described <a style="font-family: Georgia;" href="http://gcc.gnu.org/onlinedocs/gcc/Atomic-Builtins.html" target="_blank">here</a>) . ARM GCC will either directly generate the correct inline code or insert a call to a <a style="font-family: Georgia;" href="http://en.wikipedia.org/wiki/Linux_kernel" target="_blank">kernel</a> user helper function containing the right code. As an example, consider the following assembly code function implementing a spin lock: <div style="font-family: Georgia;" class="codetop">CODE</div><div style="font-family: Georgia;" class="codemain"><div id="codeText" class="codeText"><ol style="margin: 0pt 1px 0pt 0pt; padding: 5px 0pt;" start="1" class="dp-css"><li>ENTRY (__spin_lock)</li><li>
 mov r1,#1</li><li>
1: swp r2,r1,[r0]</li><li>
 teq r2,#0</li><li>
 bne 1b</li><li>
 mov r0,#0</li><li>
END (__spin_lock)</li></ol></div></div> This would be replaced with <div style="font-family: Georgia;" class="codetop">CODE</div><div style="font-family: Georgia;" class="codemain"><div id="codeText" class="codeText"><ol style="margin: 0pt 1px 0pt 0pt; padding: 5px 0pt;" start="1" class="dp-css"><li>typedef struct {int flag; …} spinlock_t; </li><li>
 </li><li>
int __spin_lock(spinlock_t *lock) </li><li>
{ </li><li>
&nbsp;&nbsp;while (__sync_lock_test_and_set(&amp;lock-&gt;flag, 1)); </li><li>
&nbsp;&nbsp;return 0; </li><li>
}</li></ol></div></div> To release the lock, you need to call another of the GCC builtins, in this case __sync_lock_release(): <div style="font-family: Georgia;" class="codetop">CODE</div><div style="font-family: Georgia;" class="codemain"><div id="codeText" class="codeText"><ol style="margin: 0pt 1px 0pt 0pt; padding: 5px 0pt;" start="1" class="dp-css"><li>void _spin_unlock(spinlock_t *lock) { </li><li>
&nbsp;&nbsp;__sync_lock_release(&amp;lock-&gt;flag); </li><li>
}</li></ol></div></div> Code wishing to lock a data structure would look something like this: <div style="font-family: Georgia;" class="codetop">CODE</div><div style="font-family: Georgia;" class="codemain"><div id="codeText" class="codeText"><ol style="margin: 0pt 1px 0pt 0pt; padding: 5px 0pt;" start="1" class="dp-css"><li>// grab the lock </li><li>
__spin_lock(&amp;lock); </li><li>
 </li><li>
// modify the locked data structure </li><li>
&lt;modification code&gt; </li><li>
 </li><li>
// release the lock </li><li>
__spin_unlock(&amp;lock);</li></ol></div></div> The
 built in functions take care of all of the details for you, including 
dealing with weakly ordered memory systems via memory barriers. This is 
only a brief introduction for more details I suggest that you read the 
“Barrier Litmus Tests and Cookbook” document in ARM’s Infocenter <a style="font-family: Georgia;" href="http://blogs.arm.com/software-enablement/locks-swps-and-two-smoking-barriers#BLT">[2]</a>. In the next article, I explain how to implement spin locks in assembler and describe how memory barriers should be used. <a style="font-family: Georgia;" name="ARM"> </a> References:<ol style="font-family: Georgia;" type="1"><li>ARM DDI 0406B_errata_2009_Q3 (ID100209) : <a href="http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.ddi0406b/index.html" target="_blank" target="_blank">ARM&#174; Architecture Reference Manual ARM&#174;v7-A and ARM&#174;v7-R edition</a> <a name="BLT"> </a></li><li>PRD03-GENC-007826 1.0	: <a href="http://infocenter.arm.com/help/topic/com.arm.doc.genc007826/Barrier_Litmus_Tests_and_Cookbook_A08.pdf" target="_blank" target="_blank">Barrier Litmus Tests and Cookbook</a></li></ol><a style="font-family: Georgia;" href="http://blogs.arm.com/software-enablement/locks-swps-and-two-smoking-barriers/" target="_blank">http://blogs.arm.com/software-enablement/locks-swps-and-two-smoking-barriers/</a>

查看全文阅读(344) ┊ 评论(0) ┊ 收藏(0) ┊举报

Linux 内核时钟

(2010-10-24 20:53:37)

分类：驱动程序系统分类: Linux

1. 时钟源 时钟源其实只是为了提供一个单调递增的时钟函数 <div id="codeText" class="codeText"><ol style="margin: 0pt 1px 0pt 0pt; padding: 5px 0pt;" start="1" class="dp-css"><li>struct clocksource { </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;/* </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;* First part of structure is read mostly </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;*/ </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;char *name; </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;struct list_head list; </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;int rating; // 表明时钟源的能力,值越大表示精度越高 </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;cycle_t (*read)(struct clocksource *cs); // 获取计数值 </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;int (*enable)(struct clocksource *cs); </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;void (*disable)(struct clocksource *cs); </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;cycle_t mask; // 由于cycle_t是一个64bit的值，该mask表明多少bit是有效的 </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;u32 mult; // 和下面的shift将cycle值转换为ns, 计算方式为: (cycle * mult)/2^shift = x ns </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;u32 shift; </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;u64 max_idle_ns; </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;unsigned long flags; </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;cycle_t (*vread)(void); </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;void (*suspend)(struct clocksource *cs); </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;void (*resume)(struct clocksource *cs); </li><li>
#ifdef CONFIG_IA64 </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;void *fsys_mmio; /* used by fsyscall asm code */ </li><li>
#define CLKSRC_FSYS_MMIO_SET(mmio, addr) ((mmio) = (addr)) </li><li>
#else </li><li>
#define CLKSRC_FSYS_MMIO_SET(mmio, addr) do { } while (0) </li><li>
#endif </li><li>
 </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;/* </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;* Second part is written at each timer interrupt </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;* Keep it in a different cache line to dirty no </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;* more than one cache line. </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;*/ </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;cycle_t cycle_last ____cacheline_aligned_in_smp; </li><li>
 </li><li>
#ifdef CONFIG_CLOCKSOURCE_WATCHDOG </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;/* Watchdog related data, used by the framework */ </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;struct list_head wd_list; </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;cycle_t wd_last; </li><li>
#endif </li><li>
};</li></ol></div> 2. 时钟事件设备 <div id="codeText" class="codeText"><ol style="margin: 0pt 1px 0pt 0pt; padding: 5px 0pt;" start="1" class="dp-css"><li>struct clock_event_device { </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;const char *name; </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;unsigned int features; </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;u64 max_delta_ns; </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;u64 min_delta_ns; </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;u32 mult; </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;u32 shift; </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;int rating; </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;int irq; // 对应的中断号，区分多核和单核的情况 </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;const struct cpumask *cpumask; </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;int (*set_next_event)(unsigned long evt, </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;struct clock_event_device *); </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;void (*set_mode)(enum clock_event_mode mode, </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;struct clock_event_device *); // 设置模式 </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;void (*event_handler)(struct clock_event_device *); </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;void (*broadcast)(const struct cpumask *mask); </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;struct list_head list; </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;enum clock_event_mode mode; </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;ktime_t next_event; </li><li>
&nbsp;&nbsp;&nbsp;&nbsp;unsigned long retries; </li><li>
};</li></ol></div> 需要注意的是，其实这2个概念并无联系，而是两个独立的概念，没有任何关系，只是实现时候一般一个硬件会同时提供这两种机制，内核中，对于clock_event_device的使用其实更多一些.

查看全文阅读(348) ┊ 评论(1) ┊ 收藏(0) ┊举报

If & When

(2010-10-18 21:36:45)

系统分类: 外语沙龙

We use the following structure to talk about actions which are very likely to take place in the future: If it rains, I'll use my umbrella. (It rains sometimes). If he goes to the station, he'll pick you up. (He can easily go to the station). This type of conditional sentence is formed by IF + SUBJECT + PRESENT TENSE + FUTURE TENSE. First conditional sentences are normally reversible. When the if-clause comes at the end, a comma is normally not required between the two clauses. The following forms mean exactly the same as their counterparts above. I'll use my umbrella if it rains. He'll pick you up if he goes to the station. We use the first conditional for events that are very likely to happen in the future. In the examples above, it is very possible that it will rain or that he will go to the station. We can also use this structure with the imperative form. This type of sentence is formed by IF + SUBJECT + PRESENT TENSE + IMPERATIVE CLAUSE. If it rains, use your umbrella. This form is also reversible and again a comma is not normally required between clauses: Use your umbrella if it rains. We use when instead of if to say that the action or event is 100% certain to happen. These types of sentences are formed by WHEN + SUBJECT + PRESENT TENSE + FUTURE TENSE. When I buy a new car, I'll give you my old one. As with the forms above, this one is also reversible and a comma is not normally required: I'll give you my old car when I buy a new one.

查看全文阅读(283) ┊ 评论(0) ┊ 收藏(0) ┊举报

中国Linux内核开发者大会(2010)小记

(2010-10-18 12:50:07)

分类：驱动程序系统分类: Linux

<pre style="font-family: Georgia;" id="article_html_Linux_M.1287377288.A"><pre>中国Linux内核开发者大会终于在上海了，更为关键的是：能亲眼见见Herbert Xu,机会不
能错过:)

这是介绍的网址:<a href="http://www.aka-kernel.org/news/" target="_blank">http://www.aka-kernel.org/news/</a>

主持整个会议的MM似乎对于Linux没有过任何兴趣，于是，对各个演讲者的背景只能用一个
词形容--无知:(,还好不是为了看主持人的:)

1. 到来的人士据我所知，有Intel开源技术中心的各位，有Redhat，有Ubuntu的开发团队
，有HP，有南大富士通，居然没有IBM，奇怪，开源界三巨头，居然少了其一，不知道组织
者是否有邀请IBM的开发人员的意愿

2. 上午的亮点在于吴峰光博士的讲演&lt;针对脏内存的重重限制&gt;，主要是针对dirty page的
处理算法的讲解，偶功力有限，只能听听，连提个问题的能力都没有，不过会后去查了一
下吴博士的一些经历，原来还在Linux Symposium 2007就因为文件预取算法而做过演讲，
而查阅其公开发表的文章，全是于此相关，Herbert Xu在会中问道吴博士怎么进入内核开
发，吴博士回答是因为要做一个视频服务器的IO优化而最终进入内核开发，可以看到2003
的样子吴博士就开始发表IO优化相关论文，可能从那个时期开始进入内核吧，恩<a href="http://h/" target="_blank">http://h</a>
ome.ustc.edu.cn/~wfg/是吴博士在学校期间的主页，博士论文是篇非常不错的文章，这周
希望能看一遍:)

3. 下午其实最为期待Herbert Xu的演讲，之前还分析过他提出的GRO/GSO，但Herbert Xu
作为压轴，只能在最后，倒是Mark Smith(惠普大中国区Linux组织领导人 大连自由软件
协会主席)的演讲Topic Visual comparison of linux and other Unix--like systems让
我有了一些兴趣，一个是内容中涉及的操作系统之多，超出我的所知，二则是表现内核代
码的图表的方式非常的明了，怪不得标题为可视比较(Visual comparison)，而对于linux
的一个基本结论也非常有意思，现在的Linux内核代码已经超过500M,接近900万行，而Mar
k Smith列出，对于他的notebook,所需的代码其实不过50M左右，这其实也说明了一个问题
，如果要研究内核，不要刻意求全，而是要根据自己的需要选择一个方向研究下去，吴博
士的经历其实也说明了这点

4. Herbert Xu作为压轴上台，居然和我一样，超级佩服为Linux网络协议栈奠定基础的Al
exey Kuznetsov :)，对Alexey Kuznetsov的评价非常之高，Herbert Xu认为正是Alexey 
Kuznetsov在10年前重写了网络协议栈，才使得Linux的优势凸显出来，而现在的很多工作
，只能算是对Alexey Kuznetsov工作的一个补充。Herbert Xu这次的演讲其实不是讲怎么
解决了问题，而更像说明在多核情况下，网络协议栈所面临的挑战，从而希望有人能一起
去解决这个问题，期间讲到一个笑话，说的是网络协议栈所面临的重重困难，他说，现在
的境况，就像是一个精神病院，有2个人想逃出去，但需要翻越100座围墙，2个人辛辛苦苦
翻越了99座，觉得实在太累，好像没有什么希望出去，于是又翻回去了；而Linux协议栈在
多核时代的挑战，就像是这100座围墙，而现在，已经翻阅了几座，但更多的围墙，需要有
人来协助翻越，因此，如果你有兴趣，不妨来帮助Linux来翻越这些围墙，Herbert Xu的演
讲，指明对队列对于多核情况下是种解决方法，但还没有完全的解决，而问题的实质，则
在于“在有全局资源需要更新的情况下，在多核时代必然成为一个瓶颈”

5. 除去演讲者，参与者中对内核有一定认识的人不多，比较遗憾:(,更遗憾的是，我都没
有提问的机会:(,Herbert Xu因为时间关系，跳过了讲解为什么多核时代，对netfilt
er也有巨大的影响
</pre></pre>

查看全文阅读(1008) ┊ 评论(1) ┊ 收藏(0) ┊举报

SDL添加一个新的底层支持 [2]

(2010-10-08 10:39:56)

分类：驱动程序系统分类: Linux

3. 上层API和底层代码的联系 &nbsp;&nbsp;&nbsp; 3.1). General &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_Init -- Initializes SDL &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_InitSubSystem -- Initialize subsystems &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_QuitSubSystem -- Shut down a subsystem &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_Quit -- Shut down SDL &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_WasInit -- Check which subsystems are initialized &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_GetError -- Get SDL error string &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_envvars -- SDL environment variables &nbsp;&nbsp;&nbsp; 3.2). Video &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; 3类操作: &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; 1). 矩形填充[简单，没什么好说的] &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; 2). 像素操作 &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; 基本过程: &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; a). 侦测Surface是否锁住 &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; b). 锁住Surface &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; 注意锁的过程是可递归的，因此，锁和解锁需要严格匹配 &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; c). 操作像素数据 &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; 下一行的计算方式: y * pitch + x &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; d). 解锁Surface &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; 3). Blit &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; int SDL_BlitSurface(SDL_Surface *src, SDL_Rect *srcrect, SDL_Surface *dst, SDL_Rect *dstrect); &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; a). srcrect中的w,h决定矩形大小，dstrect决定拷贝的坐标，其中w,h被忽视掉 &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; b). srcrect为NULL代表源Surface全部拷贝，而dstrect为NULL代表目的坐标为(0,0) &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; c). 不能在锁住的Surface调用blit &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; d). The final blit rectangle is saved in dstrect after all clipping is performed (srcrect is not modified). &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; e). SDL_SRCAPLHA/SDL_SRCCOLORKEY对该函数的影响: if (source surface has SDL_SRCALPHA set) { &nbsp;&nbsp;&nbsp; if (source surface has alpha channel (that is, format-&gt;Amask != 0)) &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; blit using per-pixel alpha, ignoring any colour key &nbsp;&nbsp;&nbsp; else { &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; if (source surface has SDL_SRCCOLORKEY set) &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; blit using the colour key AND the per-surface alpha value &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; else &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; blit using the per-surface alpha value &nbsp;&nbsp;&nbsp; } } else { &nbsp;&nbsp;&nbsp; if (source surface has SDL_SRCCOLORKEY set) &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; blit using the colour key &nbsp;&nbsp;&nbsp; else &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; ordinary opaque rectangular blit } &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; &nbsp; &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; 更新与显示，有2种方式： &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; 1). 直接矩形更新显示 &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; a). void SDL_UpdateRect(SDL_Surface *screen, Sint32 x, Sint32 y, Sint32 w, Sint32 h); &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; b). void SDL_UpdateRects(SDL_Surface *screen, int numrects, SDL_Rect *rects); &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; 2). 双缓冲 &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; a). int SDL_Flip(SDL_Surface *screen); &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; 只能在支持SDL_DOUBLEBUF的情况使用，在vertical retrace时候交换video buffer &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; &nbsp; &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; alpha混合,2种方式: &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; 1). per-surface混合 &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; &nbsp; &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; 2). per-pixel混合 &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; 在Blit时，a混合的具体行为： &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; a). RGBA-&gt;RGB with SDL_SRCALPHA &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; 使用源Surface的a通道,忽略SDL_SRCCOLORKEY和per-surface alpha &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; b). RGBA-&gt;RGB without SDL_SRCALPHA &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; 直接拷贝源Surface的RGB,忽略source alpha channel and the per-surface alpha &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; c). RGB-&gt;RGBA with SDL_SRCALPHA &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; 使用per-surface alpha,如果存在SDL_SRCCOLORKEY,仅不是该color key的颜色被拷贝,a通道的像素被当成不透明处理 &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; d). RGB-&gt;RGBA without SDL_SRCALPHA&nbsp;&nbsp; &nbsp; &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; RGB直接覆盖,如果存在SDL_SRCCOLORKEY,仅不是该color key的颜色被拷贝 &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; e). RGBA-&gt;RGBA with SDL_SRCALPHA &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; 使用源Surface的a值混合,目的Surface的a值不作改变,且SDL_SRCCOLORKEY被忽视 &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; f). RGBA-&gt;RGBA without SDL_SRCALPHA &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; 直接copy源Surface的RGBA，如果存在SDL_SRCCOLORKEY,则只有不是color key的值被拷贝 &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; g). RGB-&gt;RGB with SDL_SRCALPHA &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; 使用per-surface alpha,如果存在SDL_SRCCOLORKEY,仅不是该color key的颜色被拷贝 &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; h). RGB-&gt;RGB without SDL_SRCALPHA &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; RGB直接覆盖,如果存在SDL_SRCCOLORKEY,则只有不是color key的值被拷贝 &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Note: Note that RGBA-&gt;RGBA blits (with SDL_SRCALPHA set) keep the alpha of the destination surface. This means that you cannot compose two arbitrary RGBA surfaces this way and get the result you would expect from "overlaying" them; the destination alpha will work as a mask. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; per-pixel优先于per-surface &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; &nbsp; &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_GetVideoSurface -- returns a pointer to the current display surface &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_GetVideoInfo -- returns a pointer to information about the video hardware &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_VideoDriverName -- Obtain the name of the video driver &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_ListModes -- Returns a pointer to an array of available screen dimensions for the given format and video flags &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_VideoModeOK -- Check to see if a particular video mode is supported. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_SetVideoMode -- Set up a video mode with the specified width, height and bits-per-pixel. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_UpdateRect -- Makes sure the given area is updated on the given screen. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_UpdateRects -- Makes sure the given list of rectangles is updated on the given screen. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_Flip -- Swaps screen buffers &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_SetColors -- Sets a portion of the colormap for the given 8-bit surface. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_SetPalette -- Sets the colors in the palette of an 8-bit surface. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_SetGamma -- Sets the color gamma function for the display &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_GetGammaRamp -- Gets the color gamma lookup tables for the display &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_SetGammaRamp -- Sets the color gamma lookup tables for the display &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_MapRGB -- Map a RGB color value to a pixel format. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_MapRGBA -- Map a RGBA color value to a pixel format. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_GetRGB -- Get RGB values from a pixel in the specified pixel format. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_GetRGBA -- Get RGBA values from a pixel in the specified pixel format. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_CreateRGBSurface -- Create an empty SDL_Surface &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_CreateRGBSurfaceFrom -- Create an SDL_Surface from pixel data &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_FreeSurface -- Frees (deletes) a SDL_Surface &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_LockSurface -- Lock a surface for directly access. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_UnlockSurface -- Unlocks a previously locked surface. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_LoadBMP -- Load a Windows BMP file into an SDL_Surface. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_SaveBMP -- Save an SDL_Surface as a Windows BMP file. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_SetColorKey -- Sets the color key (transparent pixel) in a blittable surface and RLE acceleration. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; 如果设置了ColorKey，在Blit时，碰到该颜色则不做任何copy，即当作透明色 &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_SetAlpha -- Adjust the alpha properties of a surface &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_SetClipRect -- Sets the clipping rectangle for a surface. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; 设置剪切矩形 &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_GetClipRect -- Gets the clipping rectangle for a surface. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_ConvertSurface -- Converts a surface to the same format as another surface. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_BlitSurface -- This performs a fast blit from the source surface to the destination surface. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_FillRect -- This function performs a fast fill of the given rectangle with some color &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_DisplayFormat -- Convert a surface to the display format &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_DisplayFormatAlpha -- Convert a surface to the display format &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_WarpMouse -- Set the position of the mouse cursor. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_CreateCursor -- Creates a new mouse cursor. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_FreeCursor -- Frees a cursor created with SDL_CreateCursor. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_SetCursor -- Set the currently active mouse cursor. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_GetCursor -- Get the currently active mouse cursor. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_ShowCursor -- Toggle whether or not the cursor is shown on the screen. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_GL_LoadLibrary -- Specify an OpenGL library &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_GL_GetProcAddress -- Get the address of a GL function &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_GL_GetAttribute -- Get the value of a special SDL/OpenGL attribute &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_GL_SetAttribute -- Set a special SDL/OpenGL attribute &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_GL_SwapBuffers -- Swap OpenGL framebuffers/Update Display &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_CreateYUVOverlay -- Create a YUV video overlay &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_LockYUVOverlay -- Lock an overlay &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_UnlockYUVOverlay -- Unlock an overlay &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_DisplayYUVOverlay -- Blit the overlay to the display &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_FreeYUVOverlay -- Free a YUV video overlay &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_GLattr -- SDL GL Attributes &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_Rect -- Defines a rectangular area &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_Color -- Format independent color description &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_Palette -- Color palette for 8-bit pixel formats &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_PixelFormat -- Stores surface format information &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_Surface -- Graphical Surface Structure &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_VideoInfo -- Video Target information &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_Overlay -- YUV video overlay &nbsp;&nbsp;&nbsp; 3.3). Window Management[FIXME:不用实现] &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_WM_SetCaption -- Sets the window tile and icon name. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_WM_GetCaption -- Gets the window title and icon name. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_WM_SetIcon -- Sets the icon for the display window. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_WM_IconifyWindow -- Iconify/Minimise the window &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_WM_ToggleFullScreen -- Toggles fullscreen mode &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_WM_GrabInput -- Grabs mouse and keyboard input. &nbsp;&nbsp;&nbsp; 3.4). Events &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Introduction &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL Event Structures. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Event Functions. &nbsp;&nbsp;&nbsp; 3.5). Joystick[FIXME:不用实现] &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_NumJoysticks -- Count available joysticks. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_JoystickName -- Get joystick name. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_JoystickOpen -- Opens a joystick for use. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_JoystickOpened -- Determine if a joystick has been opened &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_JoystickIndex -- Get the index of an SDL_Joystick. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_JoystickNumAxes -- Get the number of joystick axes &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_JoystickNumBalls -- Get the number of joystick trackballs &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_JoystickNumHats -- Get the number of joystick hats &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_JoystickNumButtons -- Get the number of joysitck buttons &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_JoystickUpdate -- Updates the state of all joysticks &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_JoystickGetAxis -- Get the current state of an axis &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_JoystickGetHat -- Get the current state of a joystick hat &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_JoystickGetButton -- Get the current state of a given button on a given joystick &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_JoystickGetBall -- Get relative trackball motion &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_JoystickClose -- Closes a previously opened joystick &nbsp;&nbsp;&nbsp; 3.6). Audio &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_AudioSpec -- Audio Specification Structure &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_OpenAudio -- Opens the audio device with the desired parameters. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_PauseAudio -- Pauses and unpauses the audio callback processing &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_GetAudioStatus -- Get the current audio state &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_LoadWAV -- Load a WAVE file &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_FreeWAV -- Frees previously opened WAV data &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_AudioCVT -- Audio Conversion Structure &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_BuildAudioCVT -- Initializes a SDL_AudioCVT structure for conversion &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_ConvertAudio -- Convert audio data to a desired audio format. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_MixAudio -- Mix audio data &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_LockAudio -- Lock out the callback function &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_UnlockAudio -- Unlock the callback function &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_CloseAudio -- Shuts down audio processing and closes the audio device. &nbsp;&nbsp;&nbsp; 3.7). CD-ROM[FIXME:不用实现] &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_CDNumDrives -- Returns the number of CD-ROM drives on the system. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_CDName -- Returns a human-readable, system-dependent identifier for the CD-ROM. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_CDOpen -- Opens a CD-ROM drive for access. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_CDStatus -- Returns the current status of the given drive. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_CDPlay -- Play a CD &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_CDPlayTracks -- Play the given CD track(s) &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_CDPause -- Pauses a CDROM &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_CDResume -- Resumes a CDROM &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_CDStop -- Stops a CDROM &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_CDEject -- Ejects a CDROM &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_CDClose -- Closes a SDL_CD handle &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_CD -- CDROM Drive Information &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_CDtrack -- CD Track Information Structure &nbsp;&nbsp;&nbsp; 3.8). Multi-threaded Programming[语义和pthread基本类似，不用更改和重新实现] &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_CreateThread -- Creates a new thread of execution that shares its parent's properties. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_ThreadID -- Get the 32-bit thread identifier for the current thread. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_GetThreadID -- Get the SDL thread ID of a SDL_Thread &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_WaitThread -- Wait for a thread to finish. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_KillThread -- Gracelessly terminates the thread. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_CreateMutex -- Create a mutex &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_DestroyMutex -- Destroy a mutex &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_mutexP -- Lock a mutex &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_mutexV -- Unlock a mutex &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_CreateSemaphore -- Creates a new semaphore and assigns an initial value to it. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_DestroySemaphore -- Destroys a semaphore that was created by SDL_CreateSemaphore. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_SemWait -- Lock a semaphore and suspend the thread if the semaphore value is zero. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_SemTryWait -- Attempt to lock a semaphore but don't suspend the thread. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_SemWaitTimeout -- Lock a semaphore, but only wait up to a specified maximum time. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_SemPost -- Unlock a semaphore. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_SemValue -- Return the current value of a semaphore. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_CreateCond -- Create a condition variable &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_DestroyCond -- Destroy a condition variable &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_CondSignal -- Restart a thread wait on a condition variable &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_CondBroadcast -- Restart all threads waiting on a condition variable &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_CondWait -- Wait on a condition variable &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_CondWaitTimeout -- Wait on a condition variable, with timeout &nbsp;&nbsp;&nbsp; 3.9). Time[不用重新实现] &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_GetTicks -- Get the number of milliseconds since the SDL library initialization. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_Delay -- Wait a specified number of milliseconds before returning. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_AddTimer -- Add a timer which will call a callback after the specified number of milliseconds has elapsed. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_RemoveTimer -- Remove a timer which was added with SDL_AddTimer. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_SetTimer -- Set a callback to run after the specified number of milliseconds has elapsed. 4. SDL_ttf的字体渲染 是对FreeType的一个封装 font family: font faces: For example, "Palatino Regular" and "Palatino Italic" are two distinct faces from the same famous family, called "Palatino" itself. #A font file contains a set of glyphs; each one can be stored as a bitmap, a vector representation or any other scheme (most scalable formats use a combination of mathematical representation and control data/programs). These glyphs can be stored in any order in the font file, and is typically accessed through a simple glyph index. #The font file contains one or more tables, called a character map (or charmap in short), which is used to convert character codes for a given encoding (e.g. ASCII, Unicode, DBCS, Big5, etc..) into glyph indices relative to the font file. A single font face may contain several charmaps. For example, most TrueType fonts contain an Apple-specific charmap as well as a Unicode charmap, which makes them usable on both Mac and Windows platforms. 基本流程: 1). 获取字体文件 2). Include和库 &nbsp;&nbsp;&nbsp; 编码基本流程: &nbsp;&nbsp;&nbsp; a). 初始化 &nbsp;&nbsp;&nbsp; TTF_Init(); &nbsp;&nbsp; &nbsp; &nbsp;&nbsp;&nbsp; b). Loading font 文件 &nbsp;&nbsp;&nbsp; TTF_OpenFont(file, ptsize); &nbsp;&nbsp; &nbsp; &nbsp;&nbsp;&nbsp; c). 渲染文字 &nbsp;&nbsp;&nbsp; TTF_RenderText_XXX(); &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; 3种渲染方式: &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Solid&nbsp; : Use this mode for FPS and other fast changing updating text displays. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Shaded : Use this when you need nice text, and can live with a box &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Blended: Use this when you want high quality, and the text isn't changing too fast.

查看全文阅读(439) ┊ 评论(1) ┊ 收藏(0) ┊举报

SDL添加一个新的底层支持

(2010-09-30 14:07:36)

分类：驱动程序系统分类: Linux

1. 注册新的硬件底层 注意:所有代码基于1.2版本 1). VideoBootStrap 在初始化的时候被直接调用 typedef struct VideoBootStrap { &nbsp;&nbsp; &nbsp;const char *name; &nbsp;&nbsp; &nbsp;const char *desc; &nbsp;&nbsp; &nbsp;int (*available)(void);&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; // 判断设备是否可用 &nbsp;&nbsp; &nbsp;SDL_VideoDevice *(*create)(int devindex); // 创建一个实例 } VideoBootStrap; 2). SDL_VideoDevice 创建一个Video实例， 具体使用的实例为SDL_VideoDevice *current_video = NULL; 依次为： a). Video b). 硬件加速 c). gamma支持 d). Opengl支持 e). 窗口管理 f). 光标管理 g). 事件管理 h). 通用数据 i). Opengl的数据 j). 私有数据 k). Free函数 /* The SDL video driver */ typedef struct SDL_VideoDevice SDL_VideoDevice; #define SDL_VideoSurface&nbsp;&nbsp; &nbsp;(current_video-&gt;screen) #define SDL_ShadowSurface&nbsp;&nbsp; &nbsp;(current_video-&gt;shadow) #define SDL_PublicSurface&nbsp;&nbsp; &nbsp;(current_video-&gt;visible) /* Create a shadow surface if necessary */ /* There are three conditions under which we create a shadow surface: &nbsp;&nbsp; &nbsp;1.&nbsp; We need a particular bits-per-pixel that we didn't get. &nbsp;&nbsp; &nbsp;2.&nbsp; We need a hardware palette and didn't get one. &nbsp;&nbsp; &nbsp;3.&nbsp; We need a software surface and got a hardware surface. */ /* Define the SDL video driver structure */ #define _THIS&nbsp;&nbsp; &nbsp;SDL_VideoDevice *_this #ifndef _STATUS #define _STATUS&nbsp;&nbsp; &nbsp;SDL_status *status #endif struct SDL_VideoDevice { &nbsp;&nbsp; &nbsp;/* * * */ &nbsp;&nbsp; &nbsp;/* The name of this video driver */ &nbsp;&nbsp; &nbsp;const char *name; &nbsp;&nbsp; &nbsp;/* * * */ &nbsp;&nbsp; &nbsp;/* Initialization/Query functions */ &nbsp;&nbsp; &nbsp;/* Initialize the native video subsystem, filling 'vformat' with the &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; "best" display pixel format, returning 0 or -1 if there's an error. &nbsp;&nbsp; &nbsp; */ &nbsp;&nbsp; &nbsp;int (*VideoInit)(_THIS, SDL_PixelFormat *vformat); &nbsp;&nbsp; &nbsp;/* List the available video modes for the given pixel format, sorted &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; from largest to smallest. &nbsp;&nbsp; &nbsp; */ &nbsp;&nbsp; &nbsp;SDL_Rect **(*ListModes)(_THIS, SDL_PixelFormat *format, Uint32 flags); &nbsp;&nbsp; &nbsp;/* Set the requested video mode, returning a surface which will be &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; set to the SDL_VideoSurface.&nbsp; The width and height will already &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; be verified by ListModes(), and the video subsystem is free to &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; set the mode to a supported bit depth different from the one &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; specified -- the desired bpp will be emulated with a shadow &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; surface if necessary.&nbsp; If a new mode is returned, this function &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; should take care of cleaning up the current mode. &nbsp;&nbsp; &nbsp; */ &nbsp;&nbsp; &nbsp;SDL_Surface *(*SetVideoMode)(_THIS, SDL_Surface *current, &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;int width, int height, int bpp, Uint32 flags); &nbsp;&nbsp; &nbsp;/* Toggle the fullscreen mode */ &nbsp;&nbsp; &nbsp;int (*ToggleFullScreen)(_THIS, int on); &nbsp;&nbsp; &nbsp;/* This is called after the video mode has been set, to get the &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; initial mouse state.&nbsp; It should queue events as necessary to &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; properly represent the current mouse focus and position. &nbsp;&nbsp; &nbsp; */ &nbsp;&nbsp; &nbsp;void (*UpdateMouse)(_THIS); &nbsp;&nbsp; &nbsp;/* Create a YUV video surface (possibly overlay) of the given &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; format.&nbsp; The hardware should be able to perform at least 2x &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; scaling on display. &nbsp;&nbsp; &nbsp; */ &nbsp;&nbsp; &nbsp;SDL_Overlay *(*CreateYUVOverlay)(_THIS, int width, int height, &nbsp;&nbsp; &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Uint32 format, SDL_Surface *display); &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; /* Sets the color entries { firstcolor .. (firstcolor+ncolors-1) } &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; of the physical palette to those in 'colors'. If the device is &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; using a software palette (SDL_HWPALETTE not set), then the &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; changes are reflected in the logical palette of the screen &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; as well. &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; The return value is 1 if all entries could be set properly &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; or 0 otherwise. &nbsp;&nbsp; &nbsp;*/ &nbsp;&nbsp; &nbsp;int (*SetColors)(_THIS, int firstcolor, int ncolors, &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp; SDL_Color *colors); &nbsp;&nbsp; &nbsp;/* This pointer should exist in the native video subsystem and should &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; point to an appropriate update function for the current video mode &nbsp;&nbsp; &nbsp; */ &nbsp;&nbsp; &nbsp;void (*UpdateRects)(_THIS, int numrects, SDL_Rect *rects); &nbsp;&nbsp; &nbsp;/* Reverse the effects VideoInit() -- called if VideoInit() fails &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; or if the application is shutting down the video subsystem. &nbsp;&nbsp; &nbsp;*/ &nbsp;&nbsp; &nbsp;void (*VideoQuit)(_THIS); &nbsp;&nbsp; &nbsp;/* * * */ &nbsp;&nbsp; &nbsp;/* Hardware acceleration functions */ &nbsp;&nbsp; &nbsp;/* Information about the video hardware */ &nbsp;&nbsp; &nbsp;SDL_VideoInfo info; &nbsp;&nbsp; &nbsp;/* The pixel format used when SDL_CreateRGBSurface creates SDL_HWSURFACEs with alpha */ &nbsp;&nbsp; &nbsp;SDL_PixelFormat* displayformatalphapixel; &nbsp;&nbsp; &nbsp; &nbsp;&nbsp; &nbsp;/* Allocates a surface in video memory */ &nbsp;&nbsp; &nbsp;int (*AllocHWSurface)(_THIS, SDL_Surface *surface); &nbsp;&nbsp; &nbsp;/* Sets the hardware accelerated blit function, if any, based &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; on the current flags of the surface (colorkey, alpha, etc.) &nbsp;&nbsp; &nbsp; */ &nbsp;&nbsp; &nbsp;int (*CheckHWBlit)(_THIS, SDL_Surface *src, SDL_Surface *dst); &nbsp;&nbsp; &nbsp;/* Fills a surface rectangle with the given color */ &nbsp;&nbsp; &nbsp;int (*FillHWRect)(_THIS, SDL_Surface *dst, SDL_Rect *rect, Uint32 color); &nbsp;&nbsp; &nbsp;/* Sets video mem colorkey and accelerated blit function */ &nbsp;&nbsp; &nbsp;int (*SetHWColorKey)(_THIS, SDL_Surface *surface, Uint32 key); &nbsp;&nbsp; &nbsp;/* Sets per surface hardware alpha value */ &nbsp;&nbsp; &nbsp;int (*SetHWAlpha)(_THIS, SDL_Surface *surface, Uint8 value); &nbsp;&nbsp; &nbsp;/* Returns a readable/writable surface */ &nbsp;&nbsp; &nbsp;int (*LockHWSurface)(_THIS, SDL_Surface *surface); &nbsp;&nbsp; &nbsp;void (*UnlockHWSurface)(_THIS, SDL_Surface *surface); &nbsp;&nbsp; &nbsp;/* Performs hardware flipping */ &nbsp;&nbsp; &nbsp;int (*FlipHWSurface)(_THIS, SDL_Surface *surface); &nbsp;&nbsp; &nbsp;/* Frees a previously allocated video surface */ &nbsp;&nbsp; &nbsp;void (*FreeHWSurface)(_THIS, SDL_Surface *surface); &nbsp;&nbsp; &nbsp;/* * * */ &nbsp;&nbsp; &nbsp;/* Gamma support */ &nbsp;&nbsp; &nbsp;Uint16 *gamma; &nbsp;&nbsp; &nbsp;/* Set the gamma correction directly (emulated with gamma ramps) */ &nbsp;&nbsp; &nbsp;int (*SetGamma)(_THIS, float red, float green, float blue); &nbsp;&nbsp; &nbsp;/* Get the gamma correction directly (emulated with gamma ramps) */ &nbsp;&nbsp; &nbsp;int (*GetGamma)(_THIS, float *red, float *green, float *blue); &nbsp;&nbsp; &nbsp;/* Set the gamma ramp */ &nbsp;&nbsp; &nbsp;int (*SetGammaRamp)(_THIS, Uint16 *ramp); &nbsp;&nbsp; &nbsp;/* Get the gamma ramp */ &nbsp;&nbsp; &nbsp;int (*GetGammaRamp)(_THIS, Uint16 *ramp); &nbsp;&nbsp; &nbsp;/* * * */ &nbsp;&nbsp; &nbsp;/* OpenGL support */ &nbsp;&nbsp; &nbsp;/* Sets the dll to use for OpenGL and loads it */ &nbsp;&nbsp; &nbsp;int (*GL_LoadLibrary)(_THIS, const char *path); &nbsp;&nbsp; &nbsp;/* Retrieves the address of a function in the gl library */ &nbsp;&nbsp; &nbsp;void* (*GL_GetProcAddress)(_THIS, const char *proc); &nbsp;&nbsp; &nbsp;/* Get attribute information from the windowing system. */ &nbsp;&nbsp; &nbsp;int (*GL_GetAttribute)(_THIS, SDL_GLattr attrib, int* value); &nbsp;&nbsp; &nbsp;/* Make the context associated with this driver current */ &nbsp;&nbsp; &nbsp;int (*GL_MakeCurrent)(_THIS); &nbsp;&nbsp; &nbsp;/* Swap the current buffers in double buffer mode. */ &nbsp;&nbsp; &nbsp;void (*GL_SwapBuffers)(_THIS); &nbsp; &nbsp;&nbsp; &nbsp;/* OpenGL functions for SDL_OPENGLBLIT */ #if SDL_VIDEO_OPENGL #if !defined(__WIN32__) #define WINAPI #endif #define SDL_PROC(ret,func,params) ret (WINAPI *func) params; #include "SDL_glfuncs.h" #undef SDL_PROC &nbsp;&nbsp; &nbsp;/* Texture id */ &nbsp;&nbsp; &nbsp;GLuint texture; #endif &nbsp;&nbsp; &nbsp;int is_32bit; &nbsp; &nbsp;&nbsp; &nbsp;/* * * */ &nbsp;&nbsp; &nbsp;/* Window manager functions */ &nbsp;&nbsp; &nbsp;/* Set the title and icon text */ &nbsp;&nbsp; &nbsp;void (*SetCaption)(_THIS, const char *title, const char *icon); &nbsp;&nbsp; &nbsp;/* Set the window icon image */ &nbsp;&nbsp; &nbsp;void (*SetIcon)(_THIS, SDL_Surface *icon, Uint8 *mask); &nbsp;&nbsp; &nbsp;/* Iconify the window. &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; This function returns 1 if there is a window manager and the &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; window was actually iconified, it returns 0 otherwise. &nbsp;&nbsp; &nbsp;*/ &nbsp;&nbsp; &nbsp;int (*IconifyWindow)(_THIS); &nbsp;&nbsp; &nbsp;/* Grab or ungrab keyboard and mouse input */ &nbsp;&nbsp; &nbsp;SDL_GrabMode (*GrabInput)(_THIS, SDL_GrabMode mode); &nbsp;&nbsp; &nbsp;/* Get some platform dependent window information */ &nbsp;&nbsp; &nbsp;int (*GetWMInfo)(_THIS, SDL_SysWMinfo *info); &nbsp;&nbsp; &nbsp;/* * * */ &nbsp;&nbsp; &nbsp;/* Cursor manager functions */ &nbsp;&nbsp; &nbsp;/* Free a window manager cursor &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; This function can be NULL if CreateWMCursor is also NULL. &nbsp;&nbsp; &nbsp; */ &nbsp;&nbsp; &nbsp;void (*FreeWMCursor)(_THIS, WMcursor *cursor); &nbsp;&nbsp; &nbsp;/* If not NULL, create a black/white window manager cursor */ &nbsp;&nbsp; &nbsp;WMcursor *(*CreateWMCursor)(_THIS, &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;Uint8 *data, Uint8 *mask, int w, int h, int hot_x, int hot_y); &nbsp;&nbsp; &nbsp;/* Show the specified cursor, or hide if cursor is NULL */ &nbsp;&nbsp; &nbsp;int (*ShowWMCursor)(_THIS, WMcursor *cursor); &nbsp;&nbsp; &nbsp;/* Warp the window manager cursor to (x,y) &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; If NULL, a mouse motion event is posted internally. &nbsp;&nbsp; &nbsp; */ &nbsp;&nbsp; &nbsp;void (*WarpWMCursor)(_THIS, Uint16 x, Uint16 y); &nbsp;&nbsp; &nbsp;/* If not NULL, this is called when a mouse motion event occurs */ &nbsp;&nbsp; &nbsp;void (*MoveWMCursor)(_THIS, int x, int y); &nbsp;&nbsp; &nbsp;/* Determine whether the mouse should be in relative mode or not. &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; This function is called when the input grab state or cursor &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; visibility state changes. &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; If the cursor is not visible, and the input is grabbed, the &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; driver can place the mouse in relative mode, which may result &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; in higher accuracy sampling of the pointer motion. &nbsp;&nbsp; &nbsp;*/ &nbsp;&nbsp; &nbsp;void (*CheckMouseMode)(_THIS); &nbsp;&nbsp; &nbsp;/* * * */ &nbsp;&nbsp; &nbsp;/* Event manager functions */ &nbsp;&nbsp; &nbsp;/* Initialize keyboard mapping for this driver */ &nbsp;&nbsp; &nbsp;void (*InitOSKeymap)(_THIS); &nbsp;&nbsp; &nbsp;/* Handle any queued OS events */ &nbsp;&nbsp; &nbsp;void (*PumpEvents)(_THIS); &nbsp;&nbsp; &nbsp;/* * * */ &nbsp;&nbsp; &nbsp;/* Data common to all drivers */ &nbsp;&nbsp; &nbsp;SDL_Surface *screen;&nbsp; //SDL_VideoSurface, 等于在私有的XXX_SetVideoMode的返回值 &nbsp;&nbsp; &nbsp;SDL_Surface *shadow;&nbsp; //SDL_ShadowSurface &nbsp;&nbsp; &nbsp;SDL_Surface *visible; //SDL_PublicSurface 指向SDL_ShadowSurface或者SDL_VideoSurface &nbsp;&nbsp; &nbsp; &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_Palette *physpal;&nbsp;&nbsp; &nbsp;/* physical palette, if != logical palette */ &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; SDL_Color *gammacols;&nbsp;&nbsp; &nbsp;/* gamma-corrected colours, or NULL */ &nbsp;&nbsp; &nbsp;char *wm_title; &nbsp;&nbsp; &nbsp;char *wm_icon; &nbsp;&nbsp; &nbsp;int offset_x; &nbsp;&nbsp; &nbsp;int offset_y; &nbsp;&nbsp; &nbsp;SDL_GrabMode input_grab; &nbsp;&nbsp; &nbsp;/* Driver information flags */ &nbsp;&nbsp; &nbsp;int handles_any_size;&nbsp;&nbsp; &nbsp;/* Driver handles any size video mode */ &nbsp;&nbsp; &nbsp;/* * * */ &nbsp;&nbsp; &nbsp;/* Data used by the GL drivers */ &nbsp;&nbsp; &nbsp;struct { &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;int red_size; &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;int green_size; &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;int blue_size; &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;int alpha_size; &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;int depth_size; &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;int buffer_size; &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;int stencil_size; &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;int double_buffer; &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;int accum_red_size; &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;int accum_green_size; &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;int accum_blue_size; &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;int accum_alpha_size; &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;int stereo; &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;int multisamplebuffers; &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;int multisamplesamples; &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;int accelerated; &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;int swap_control; &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;int driver_loaded; &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;char driver_path[256]; &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;void* dll_handle; &nbsp;&nbsp; &nbsp;} gl_config; &nbsp;&nbsp; &nbsp;/* * * */ &nbsp;&nbsp; &nbsp;/* Data private to this driver */ &nbsp;&nbsp; &nbsp;struct SDL_PrivateVideoData *hidden; &nbsp;&nbsp; &nbsp;struct SDL_PrivateGLData *gl_data; &nbsp;&nbsp; &nbsp;/* * * */ &nbsp;&nbsp; &nbsp;/* The function used to dispose of this structure */ &nbsp;&nbsp; &nbsp;void (*free)(_THIS); }; #undef _THIS 其中的Videoinfo /** Useful for determining the video hardware capabilities */ typedef struct SDL_VideoInfo { &nbsp;&nbsp; &nbsp;Uint32 hw_available :1;&nbsp;&nbsp; &nbsp;/**&lt; Flag: Can you create hardware surfaces? */ &nbsp;&nbsp; &nbsp;Uint32 wm_available :1;&nbsp;&nbsp; &nbsp;/**&lt; Flag: Can you talk to a window manager? */ &nbsp;&nbsp; &nbsp;Uint32 UnusedBits1&nbsp; :6; &nbsp;&nbsp; &nbsp;Uint32 UnusedBits2&nbsp; :1; &nbsp;&nbsp; &nbsp;Uint32 blit_hw&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; :1;&nbsp;&nbsp; &nbsp;/**&lt; Flag: Accelerated blits HW --&gt; HW */ &nbsp;&nbsp; &nbsp;Uint32 blit_hw_CC&nbsp;&nbsp; :1;&nbsp;&nbsp; &nbsp;/**&lt; Flag: Accelerated blits with Colorkey */ &nbsp;&nbsp; &nbsp;Uint32 blit_hw_A&nbsp;&nbsp;&nbsp; :1;&nbsp;&nbsp; &nbsp;/**&lt; Flag: Accelerated blits with Alpha */ &nbsp;&nbsp; &nbsp;Uint32 blit_sw&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; :1;&nbsp;&nbsp; &nbsp;/**&lt; Flag: Accelerated blits SW --&gt; HW */ &nbsp;&nbsp; &nbsp;Uint32 blit_sw_CC&nbsp;&nbsp; :1;&nbsp;&nbsp; &nbsp;/**&lt; Flag: Accelerated blits with Colorkey */ &nbsp;&nbsp; &nbsp;Uint32 blit_sw_A&nbsp;&nbsp;&nbsp; :1;&nbsp;&nbsp; &nbsp;/**&lt; Flag: Accelerated blits with Alpha */ &nbsp;&nbsp; &nbsp;Uint32 blit_fill&nbsp;&nbsp;&nbsp; :1;&nbsp;&nbsp; &nbsp;/**&lt; Flag: Accelerated color fill */ &nbsp;&nbsp; &nbsp;Uint32 UnusedBits3&nbsp; :16; &nbsp;&nbsp; &nbsp;Uint32 video_mem;&nbsp;&nbsp; &nbsp;/**&lt; The total amount of video memory (in K) */ &nbsp;&nbsp; &nbsp;SDL_PixelFormat *vfmt;&nbsp;&nbsp; &nbsp;/**&lt; Value: The format of the video surface */ &nbsp;&nbsp; &nbsp; &nbsp;&nbsp; &nbsp;int&nbsp;&nbsp;&nbsp; current_w;&nbsp;&nbsp; &nbsp;/**&lt; Value: The current video mode width */ &nbsp;&nbsp; &nbsp;int&nbsp;&nbsp;&nbsp; current_h;&nbsp;&nbsp; &nbsp;/**&lt; Value: The current video mode height */ } SDL_VideoInfo; 2. 需要实现的接口 1). /* Initialize the native video subsystem, filling 'vformat' with the &nbsp; "best" display pixel format, returning 0 or -1 if there's an error. &nbsp;*/ int (*VideoInit)(_THIS, SDL_PixelFormat *vformat); 初始化哪些资源，将路径设置放在此处？ a). 能将路径配置放在此函数中吗?似乎不合适,仅仅被调用一次 /* Reverse the effects VideoInit() -- called if VideoInit() fails &nbsp;&nbsp; or if the application is shutting down the video subsystem. */ void (*VideoQuit)(_THIS); 2). /* List the available video modes for the given pixel format, sorted &nbsp;&nbsp; from largest to smallest. &nbsp;*/ SDL_Rect **(*ListModes)(_THIS, SDL_PixelFormat *format, Uint32 flags); 不用实现[FIXME] 3). /* Set the requested video mode, returning a surface which will be &nbsp;&nbsp; set to the SDL_VideoSurface.&nbsp; The width and height will already &nbsp;&nbsp; be verified by ListModes(), and the video subsystem is free to &nbsp;&nbsp; set the mode to a supported bit depth different from the one &nbsp;&nbsp; specified -- the desired bpp will be emulated with a shadow &nbsp;&nbsp; surface if necessary.&nbsp; If a new mode is returned, this function &nbsp;&nbsp; should take care of cleaning up the current mode. &nbsp;*/ SDL_Surface *(*SetVideoMode)(_THIS, SDL_Surface *current, &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;int width, int height, int bpp, Uint32 flags); 基本可以保持不变 &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp; 4). /* Create a YUV video surface (possibly overlay) of the given &nbsp;&nbsp; format.&nbsp; The hardware should be able to perform at least 2x &nbsp;&nbsp; scaling on display. &nbsp;*/ SDL_Overlay *(*CreateYUVOverlay)(_THIS, int width, int height, &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Uint32 format, SDL_Surface *display); 创建一个Overlay,并分配私有数据,指定数据格式,在内存中的排布等 SDL_Overlay *SDL_CreateYUVOverlay(int width, int height, Uint32 format, SDL_Surface *display); int SDL_LockYUVOverlay(SDL_Overlay *overlay); void SDL_UnlockYUVOverlay(SDL_Overlay *overlay); int SDL_DisplayYUVOverlay(SDL_Overlay *overlay, SDL_Rect *dstrect); Blit overlay到Surface,dstrect指定目标矩形大小和位置,如果dstrect不等于Overlay大小,则会 使用缩放功能,内置2X的优化 void SDL_FreeYUVOverlay(SDL_Overlay *overlay); /** @name Overlay Formats &nbsp;*&nbsp; The most common video overlay formats. &nbsp;*&nbsp; For an explanation of these pixel formats, see: &nbsp;*&nbsp;&nbsp; &nbsp;http://www.webartz.com/fourcc/indexyuv.htm &nbsp;* &nbsp;*&nbsp; For information on the relationship between color spaces, see: &nbsp;*&nbsp; http://www.neuro.sfc.keio.ac.jp/~aly/polygon/info/color-space-faq.html &nbsp;*/ /*@{*/ #define SDL_YV12_OVERLAY&nbsp; 0x32315659&nbsp;&nbsp; &nbsp;/**&lt; Planar mode: Y + V + U&nbsp; (3 planes) */ #define SDL_IYUV_OVERLAY&nbsp; 0x56555949&nbsp;&nbsp; &nbsp;/**&lt; Planar mode: Y + U + V&nbsp; (3 planes) */ #define SDL_YUY2_OVERLAY&nbsp; 0x32595559&nbsp;&nbsp; &nbsp;/**&lt; Packed mode: Y0+U0+Y1+V0 (1 plane) */ #define SDL_UYVY_OVERLAY&nbsp; 0x59565955&nbsp;&nbsp; &nbsp;/**&lt; Packed mode: U0+Y0+V0+Y1 (1 plane) */ #define SDL_YVYU_OVERLAY&nbsp; 0x55595659&nbsp;&nbsp; &nbsp;/**&lt; Packed mode: Y0+V0+Y1+U0 (1 plane) */ /*@}*/ /** The YUV hardware video overlay */ typedef struct SDL_Overlay { &nbsp;&nbsp; &nbsp;Uint32 format;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;/**&lt; Read-only */ &nbsp;&nbsp; &nbsp;int w, h;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;/**&lt; Read-only */ &nbsp;&nbsp; &nbsp;int planes;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;/**&lt; Read-only *///表明是打包模式还是平面模式 &nbsp;&nbsp; &nbsp;Uint16 *pitches;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;/**&lt; Read-only */ &nbsp;&nbsp; &nbsp;Uint8 **pixels;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;/**&lt; Read-write */ &nbsp;&nbsp; &nbsp;/** @name Hardware-specific surface info */ &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; /*@{*/ &nbsp;&nbsp; &nbsp;struct private_yuvhwfuncs *hwfuncs; //函数指针 &nbsp;&nbsp; &nbsp;struct private_yuvhwdata *hwdata;&nbsp;&nbsp; //私有数据 &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; /*@{*/ &nbsp;&nbsp; &nbsp;/** @name Special flags */ &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; /*@{*/ &nbsp;&nbsp; &nbsp;Uint32 hw_overlay :1;&nbsp;&nbsp; &nbsp;/**&lt; Flag: This overlay hardware accelerated? */ &nbsp;&nbsp; &nbsp;Uint32 UnusedBits :31; &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; /*@}*/ } SDL_Overlay; 其中,需要实现的函数指针包括如下: 可以参考DirectFB的实现: struct private_yuvhwfuncs { &nbsp;&nbsp; &nbsp;int (*Lock)(_THIS, SDL_Overlay *overlay);&nbsp;&nbsp;&nbsp; // 可以忽略实现 &nbsp;&nbsp; &nbsp;void (*Unlock)(_THIS, SDL_Overlay *overlay); // 可以忽略实现 &nbsp;&nbsp; &nbsp;int (*Display)(_THIS, SDL_Overlay *overlay, SDL_Rect *src, SDL_Rect *dst); //主要实现了Merge操作,为了显示 &nbsp;&nbsp; &nbsp;void (*FreeHW)(_THIS, SDL_Overlay *overlay); // 非常简单 }; 在文件SDL_nullvideo_yuv.c 5). /* This pointer should exist in the native video subsystem and should &nbsp;&nbsp; point to an appropriate update function for the current video mode &nbsp;*/ void (*UpdateRects)(_THIS, int numrects, SDL_Rect *rects); 关键函数 6). /* Allocates a surface in video memory */ int (*AllocHWSurface)(_THIS, SDL_Surface *surface); 原来的代码中，新加入了两种标记: SDL_PWSWSURFACE SDL_PWHWSURFACE 考虑合并这2个标记 /* Frees a previously allocated video surface */ void (*FreeHWSurface)(_THIS, SDL_Surface *surface); 释放资源，实现起来也非常方便 7). /* Sets the hardware accelerated blit function, if any, based &nbsp;&nbsp; on the current flags of the surface (colorkey, alpha, etc.) &nbsp;*/ int (*CheckHWBlit)(_THIS, SDL_Surface *src, SDL_Surface *dst); 设置硬件加速Blit功能函数 其中的SDL_Surface的概念是核心 /** This structure should be treated as read-only, except for 'pixels', &nbsp;*&nbsp; which, if not NULL, contains the raw pixel data for the surface. &nbsp;*/ typedef struct SDL_Surface { &nbsp;&nbsp; &nbsp;Uint32 flags;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;/**&lt; Read-only */ &nbsp;&nbsp; &nbsp;SDL_PixelFormat *format;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;/**&lt; Read-only */ &nbsp;&nbsp; &nbsp;int w, h;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;/**&lt; Read-only */ &nbsp;&nbsp; &nbsp;Uint16 pitch;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;/**&lt; Read-only */ &nbsp;&nbsp; &nbsp;void *pixels;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;/**&lt; Read-write *///指向像素数据 &nbsp;&nbsp; &nbsp;int offset;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;/**&lt; Private */ &nbsp;&nbsp; &nbsp;/** Hardware-specific surface info */ &nbsp;&nbsp; &nbsp;struct private_hwdata *hwdata; // 硬件特定信息,由底层driver来实现 &nbsp;&nbsp; &nbsp;/** clipping information */ &nbsp;&nbsp; &nbsp;SDL_Rect clip_rect;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;/**&lt; Read-only */ &nbsp;&nbsp; &nbsp;Uint32 unused1;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;/**&lt; for binary compatibility */ &nbsp;&nbsp; &nbsp;/** Allow recursive locks */ &nbsp;&nbsp; &nbsp;Uint32 locked;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;/**&lt; Private */ &nbsp;&nbsp; &nbsp;/** info for fast blit mapping to other surfaces */ &nbsp;&nbsp; &nbsp;struct SDL_BlitMap *map;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;/**&lt; Private */ &nbsp;&nbsp; &nbsp;/** format version, bumped at every change to invalidate blit maps */ &nbsp;&nbsp; &nbsp;unsigned int format_version;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;/**&lt; Private */ &nbsp;&nbsp; &nbsp;/** Reference count -- used when freeing surface */ &nbsp;&nbsp; &nbsp;int refcount;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;/**&lt; Read-mostly */ } SDL_Surface; /* Blit mapping definition */ typedef struct SDL_BlitMap { &nbsp;&nbsp; &nbsp;SDL_Surface *dst; &nbsp;&nbsp; &nbsp;int identity; &nbsp;&nbsp; &nbsp;Uint8 *table; &nbsp;&nbsp; &nbsp;SDL_blit hw_blit;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp; &nbsp;// 在CheckHWBlit函数中设置DUMMY_HWAccelBlit函数 &nbsp;&nbsp; &nbsp;SDL_blit sw_blit; &nbsp;&nbsp; &nbsp;struct private_hwaccel *hw_data; &nbsp;&nbsp; &nbsp;struct private_swaccel *sw_data; &nbsp;&nbsp; &nbsp;/* the version count matches the destination; mismatch indicates &nbsp;&nbsp; &nbsp;&nbsp;&nbsp; an invalid mapping */ &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; unsigned int format_version; } SDL_BlitMap; 设置好后，会在硬件加速Blit的时候使用 8). /* Returns a readable/writable surface */ int (*LockHWSurface)(_THIS, SDL_Surface *surface); void (*UnlockHWSurface)(_THIS, SDL_Surface *surface); 判断是否可以读写 9). /* Performs hardware flipping */ int (*FlipHWSurface)(_THIS, SDL_Surface *surface); 在SDL_Flip()中被调用，如果支持SDL_DOUBLEBUF才会被使用，在SDL_Flip()中使用 SDL_UpdateRect()函数 10). /* Initialize keyboard mapping for this driver */ void (*InitOSKeymap)(_THIS); 将driver的键盘事件映射到SDL支持的键盘事件[FIXME: 我们真的需要这个功能吗], 基本不用改变该函数的实现 11). /* Handle any queued OS events */ void (*PumpEvents)(_THIS); 消息队列，这里是消息的源头，所有消息从这里开始排队,在SDL的内部其实是起了一个专门的线程 用于消息的接收,因此，用loop像消息队列里面塞消息即可

查看全文阅读(382) ┊ 评论(0) ┊ 收藏(0) ┊举报

更多>>

共 5 页

1 2 3 4 5 下一页