C, C ++ language compilation process summary

In C language [Yu Yan], the compilation of [Bian Yi] process is summarized in detail (for conversion). In this case, we intend to summarize the content of [Bian Yi, I accidentally saw a netizen write a very comprehensive article, so I made a little effort to make a memo.

 

The C language [Yu Yan] compilation [Bian Yi] links to [Lian Jie] to compile a C program (source code [Dai Ma] [Yuan Dai Ma]). convert [Zhuan Huan] to a program that can run [Yun hang] on the hardware [Ying Jian] (the [Zhi hang] Code [Dai Ma] can be executed), you need to compile [Bian Yi] and link [Lian Jie]. Compiling [Bian Yi] is to translate the source code [Dai Ma] [Yuan Dai Ma] of the text [Wen Ben] form into the machine language [Yu Yan] [Ji Qi Yu Yan] form. [mu Biao] File [Wen Jian] process. The link [Lian Jie] is to start the [Qi Dong] Code [Dai Ma] of the [mu Biao] File [Wen Jian], the operating system [XI Tong] [CaO Zuo Xi Tong]] and the library file [Wen Jian] used to organize and form the final process of generating the executable [Zhi hang] Code [Dai Ma. The process is illustrated as follows:

 

 

As shown in the figure, the [Bian Yi] compilation process of the entire code [Dai Ma] is divided into two processes: compiling [Bian Yi] and linking [Lian Jie, compile the [Bian Yi] section enclosed by braces in the corresponding graph, and the rest is the link [Lian Jie] process.

 

[Bian Yi] compilation process

The [Bian Yi] compilation process can be divided into two phases: compiling [Bian Yi] and compiling [Hui Bian].

 

Compile [Bian Yi]

Compile [Bian Yi] to read the [Du Qu] source program [Yuan Cheng Xu] (character [Zi Fu] stream ), perform lexical [ci fa] and syntax [Yu fa] analysis, convert the [Yu Yan] [Gao Ji Yu Yan] command [Zhi Ling] to [Zhuan Huan] for the compilation of [Hui Bian] Code [Dai Ma], the compilation of the source file [Wen Jian] [Bian Yi] contains two main stages:

 

The first phase is the pre-processing phase, which is carried out before the official compilation of the [Bian Yi] phase. The pre-processing phase will modify the content of the [Xiu Gai] source file [Wen Jian] based on the pre-processing instruction [Zhi Ling] that has been placed in the [Wen Jian] file. For example, the # include command [Zhi Ling] is a preprocessing Command [Zhi Ling], which adds the content of the header file [Wen Jian] To the. cpp file [Wen Jian. This method of modifying the [Xiu Gai] source file [Wen Jian] Before compiling [Bian Yi] provides great flexibility [Ling Huo Xing], to adapt to the [XI Tong] [CaO Zuo Xi Tong] environment [XI Tong Huan Jing] restrictions of different computers and operating systems. The Code [Dai Ma] required by one environment may be different from the code [Dai Ma] required by another environment, because the available hardware [Ying Jian] or the operating system [XI Tong] [CaO Zuo Xi Tong] is different. In many cases [Qing Kuang], [Dai Ma] codes for [Yong Yu] different environments can be placed in the same file [Wen Jian, then, modify the [Xiu Gai] Code [Dai Ma] In the preprocessing phase to adapt it to the current environment.

It mainly deals with the following aspects:

 

(1) macro definition Command [Zhi Ling], for example, # define a B
For this pseudo Command [Zhi Ling] [Wei Zhi Ling], to pre-compile [Bian Yi] [Yu Bian Yi], replace [Ti Huan] with B for all a in the program. however, [Ti Huan] is not replaced with a of the [Zi Fu] string [Zi Fu Chuan] constant [chang Liang]. In addition, # UNDEF will cancel [Qu Xiao]'s definition of a macro so that the appearance of this string will not be replaced by [Ti Huan].

 

(2) condition [Tiao Jian] compile the [Bian Yi] command [Zhi Ling], for example, # ifdef, # ifndef, # else, # Elif, # endif.
The introduction of these pseudo commands [Zhi Ling] [Wei Zhi Ling] allows programmers [Cheng Xu yuan] to determine how to compile the [Bian Yi] program [Bian Yi Cheng Xu] which codes are processed by [Dai Ma. Pre-compile the [Bian Yi] [Yu Bian Yi] program [Bian Yi Cheng Xu] [Yu Bian Yi Cheng Xu] based on the relevant files [Wen Jian], filter out unnecessary code [Dai Ma] [Guo LV.

(3) the header file [Wen Jian] contains the command [Zhi Ling], such as # include "FILENAME" or # include <FILENAME>.

In the header file [Wen Jian], the pseudo Command [Zhi Ling] [Wei Zhi Ling] # define defines a large number of macros (the most common character is the [Zi Fu] constant [chang Liang ]), it also contains the declaration of various external symbols [Sheng Ming]. The purpose of using the header file [Wen Jian] is to make some definitions available for multiple different C source programs [Yuan Cheng Xu. Because in the C source program [Yuan Cheng Xu] That needs to use these definitions, you only need to add the # include statement [Yu Ju, instead of repeating these definitions in this file [Wen Jian. Pre-compiled [Bian Yi] [Yu Bian Yi] program [Bian Yi Cheng Xu] [Yu Bian Yi Cheng Xu] adds all the definitions in the header file [Wen Jian] to output [Shu Chu] File [Wen Jian, for compilation of [Bian Yi] program [Bian Yi Cheng Xu. The header file [Wen Jian] contained in the C source [Yuan Cheng Xu] can be provided by the system [XI Tong, these header files [Wen Jian] are generally stored in the/usr/include directory. # Include them in the program using angle brackets (<> ). In addition, developers can also define their own header file [Wen Jian]. These files [Wen Jian] are generally placed in the same directory as the C source program [Yuan Cheng Xu, double quotation marks ("") are used in # include ("").

(4) special symbols, pre-compiled [Bian Yi] [Yu Bian Yi] program [Bian Yi Cheng Xu] [Yu Bian Yi Cheng Xu] can recognize some special symbols of [Shi BIE.
For example, the line Identification [Biao Shi] that appears in the source [Yuan Cheng Xu] will be interpreted as [Jie Shi] as the current row number (in decimal format [shi jin Zhi ), file is interpreted as the name of the C source program [Yuan Cheng Xu] currently compiled by [Bian Yi. Pre-compiled [Bian Yi] [Yu Bian Yi] program [Bian Yi Cheng Xu] [Yu Bian Yi Cheng Xu] replace [Ti Huan] with the appropriate value.

After pre-Compilation of the [Bian Yi] [Yu Bian Yi] program [Bian Yi Cheng Xu] [Yu Bian Yi Cheng Xu], basically the" replace "work. After this replacement, generate an output [Shu Chu] File [Wen Jian] with no macro definition, no condition [Tiao Jian] to compile the [Bian Yi] command [Zhi Ling], and no special symbols. The meaning of this file [Wen Jian] is the same as that of the source file [Wen Jian] without preprocessing, but the content is different. Next, this output [Shu Chu] File [Wen Jian] will be translated into machine commands [Zhi Ling] as the output [Shu Chu] of the [Bian Yi Cheng Xu] program.] [Ji Qi Zhi Ling].

In the second phase, compile [Bian Yi] and optimize [You Hua, only [chang Liang] is the constant in the output [Shu Chu] File [Wen Jian] obtained by pre-compiling [Bian Yi] [Yu Bian Yi]. for example, the number [Shu Zi], the character [Zi Fu] string [Zi Fu Chuan], and the variable [Bian Liang] are defined, and the keywords [Guan Jian Zi] of the C language [Yu Yan], such as main, if, else, for, while, {,}, +,-, *,/, and so on.

To compile the [Bian Yi] program [Bian Yi Cheng Xu], you must analyze [CI Fa Fen XI] and [Yu fa] Through lexical [ci fa ]. fen XI], after confirming that all [Zhi Ling] commands of [Que Ren] comply with the syntax [Yu fa] rules, translate it into equivalent intermediate code [Dai Ma] [Zhong Jian Dai Ma] to represent or compile [Hui Bian] Code [Dai Ma].

Optimization of [You Hua] processing is a difficult technology in compiling [XI Tong] [Bian Yi Xi Tong] system. [Wen ti] is related not only to the compilation of [Bian Yi] technology, but also to the hardware [Ying Jian] environment of the machine. Part of optimization [You Hua] is the optimization of the intermediate code [Dai Ma] [Zhong Jian Dai Ma] [You Hua]. This optimization [You Hua] does not depend on a specific computer. Another optimization [You Hua] is mainly used to generate the target [mu Biao] Code [Dai Ma.

For the previous optimization [You Hua], the main task is to delete the [Shan Chu] public expression [Biao Da shi], cycle [Xun Huan] to optimize [You Hua] (code [Dai Ma] out-of-context, and weaken the intensity of [Qiang du Xue RuO], changing [Bian Huan] loops [Xun Huan] to control [Kong Zhi] condition [Tiao Jian], merging of known quantities, etc), duplication propagation, and useless value assignment [Fu Zhi] to delete [Shan Chu.

The Optimization of the latter type [Lei Xing] [You Hua] is closely related to the hardware [Ying Jian] structure [Jie Gou] of the machine, the main consideration is how to make full use of the value of the variable [Bian Liang] stored in the [Ying Jian] register [Ji Cun Qi] of each hardware of the machine, to reduce the number of accesses to the memory [Nei Cun. In addition, how to execute the [Zhi hang] command [Zhi Ling] Based on Machine Hardware [Ying Jian] (for example, pipeline [Liu Shui Xian], Proteus, CISC, VLIW, etc) [Tiao Zheng] makes some adjustments to the Command [Zhi Ling] to make the target [mu Biao] Code [Dai Ma] relatively short, and the efficiency of executing [Zhi hang] is relatively high, it is also an important research topic.

Compile [Hui Bian]

Compilation [Hui Bian] actually refers to the compilation of [Hui Bian] language [Yu Yan] [Hui Bian Yu Yan] Code [Dai Ma] into the target [mu Biao] Machine [mu Biao ji] indicates the process of [Zhi Ling] [Ji Qi Zhi Ling] commands. For each C language [Yu Yan] source program [Yuan Cheng Xu] processed by the Translation System [XI Tong], the target [mu Biao] File [Wen Jian] will be obtained after this processing. What is stored in the [mu Biao] File [Wen Jian] is the machine language [Yu Yan] [Ji Qi Yu] of the target [mu Biao] equivalent to the source [Yuan Cheng Xu ]. yan] Code [Dai Ma]. The [mu Biao] File [Wen Jian] consists of segments. Generally, a target [mu Biao] File [Wen Jian] contains at least two segments:

 

Code [Dai Ma] section: This section mainly contains the program instructions [Zhi Ling]. This section is generally readable and executable [Zhi hang], but generally cannot be written.

Data [Shu Ju] section: Mainly stores various global variables used in the program [Bian Liang] [Quan Ju Bian Liang] or static data [Shu Ju]. Generally, the [Shu Ju] segment of the data is readable, writable, and can be executed in [Zhi hang.

 

In UNIX environments, there are three main types of [Lei Xing] target [mu Biao] File [Wen Jian]:

(1) relocated [Ding Wei] File [Wen Jian]
This includes the [mu Biao] File [Wen Jian] link [Lian Jie] suitable for other targets to create a [Zhi hang] executable or shared target [mu Biao] File [wen Jian] Code [Dai Ma] and data [Shu Ju].

(2) shared destination [mu Biao] File [Wen Jian]

This file [Wen Jian] stores the Code [Dai Ma] and data [Shu Ju] suitable for linking [Lian Jie] in two contexts [Shang Xia Wen]. The first method is to link the [Lian Jie] program to process it together with other [Ding Wei] File [Wen Jian] and the shared target [mu Biao] File [Wen Jian ]. to create another [mu Biao] File [Wen Jian]; the second type is dynamic [Dong Tai] link [Lian Jie] program to associate it with another executable [Zhi hang] File [Wen Jian] [ke Zhi hang Wen Jian] and other sharing the target [mu Biao] File [Wen Jian] is combined, create a process [Jin Cheng] image.

(3) executable [Zhi hang] File [Wen Jian] [ke Zhi hang Wen Jian]
It contains a file [Wen Jian] that can be used by the operating system [XI Tong] [CaO Zuo Xi Tong] to create a process [Jin Cheng] to execute [Zhi hang]. Compile the [Hui Bian] program [Hui Bian Cheng Xu] to generate the [mu Biao] File [Wen Jian] of the first [Lei Xing] type. For the latter two types, some other processing is required. This is the work of linking the [Lian Jie] program.

 

Link to the [Lian Jie] Process

The target [mu Biao] File [Wen Jian] generated by the [Hui Bian] program [Hui Bian Cheng Xu] cannot be executed immediately [Zhi hang], there may be many unsolved problems [wen ti].

For example, the function [Han Shu] In a source file [Wen Jian] may reference a symbol defined in [Yin Yong] in another source file [Wen Jian] (for example, the variable [Bian Liang] or the function [han Shu] calls [Tiao Yong] [Han Shu Tiao Yong] and so on ); in the program, the [Tiao Yong] function [Han Shu] In a library file [Wen Jian] may be called, and so on. All these problems [wen ti] can be solved only after being processed by the [Lian Jie] program.

The main task of linking the [Lian Jie] program is to connect the [mu Biao] File [Wen Jian] to each other [Lian Jie], the [Yin Yong] symbol will be referenced in a file [Wen Jian] To connect [Lian Jie] with the definition of this symbol in another file [Wen Jian, make all these target [mu Biao] files [Wen Jian] An operating system [XI Tong] [CaO Zuo Xi Tong] loaded into [Zhuang RU] And run [Zhi hang].

Based on the [Lian Jie] Link Method of the same library function [Han Shu] specified by the developer, the link [Lian Jie] processing can be divided into two types:

 

(1) Static link [Lian Jie]

In this connection [Lian Jie] mode, the Code [Dai Ma] of the function [Han Shu] will be copied from its static link [Lian Jie] Library [Kao bei] to the final executable [Zhi hang] program [zhi hang Cheng Xu] [ke Zhi hang Cheng Xu. In this way, when the program is executed [Zhi hang], the Code [Dai Ma] will be loaded into [Zhuang RU] to the virtual [Xu NI] address of the Process [Jin Cheng] [di zhi] [Xu Ni di Zhi] space [Kong Jian] [di Zhi Kong Jian. The static link [Lian Jie] library is actually a collection of the [mu Biao] File [Wen Jian, each file [Wen Jian] contains the Code [Dai Ma] of one or a group of related functions [Han Shu] in the library.

 

(2) Dynamic [Dong Tai] link [Lian Jie]

In this way, the Code [Dai Ma] of the function [Han Shu] is put in a dynamic [Dong Tai] link [Lian Jie] Library [Dong Tai lian jie Ku] or shared object [DUI Xiang]. in the [mu Biao] File [Wen Jian. Link to the [Lian Jie] program. At this time, only the shared object [DUI] is recorded in the final executable [Zhi hang] program [Zhi hang Cheng Xu] [ke Zhi hang Cheng Xu ]. xiang] Name and a small amount of registration information [Xin XI]. When the [Zhi hang] File [Wen Jian] [ke Zhi hang Wen Jian] is executed, dynamic [Dong Tai] link to [Lian Jie] Library [Dong Tai lian jie Ku] will be mapped to [Ying she] to [Yun hang Shi] when running [Yun hang] the virtual address of the corresponding process [Jin Cheng] [di Zhi] space [Kong Jian] [di Zhi Kong Jian]. The dynamic [Dong Tai] link to the [Lian Jie] program will be based on the information recorded in the executable [Zhi hang] program [Zhi hang Cheng Xu] [ke Zhi hang Cheng Xu] [Xin XI] Find the corresponding function [Han Shu] Code [Dai Ma].

For the function [Han Shu] In the executable [Zhi hang] File [Wen Jian] [ke Zhi hang Wen Jian], call [Tiao Yong] [Han Shu Tiao Yong], you can use the dynamic [Dong Tai] link [Lian Jie] or static link [Lian Jie] method [Fang fa]. Using Dynamic [Dong Tai] to link [Lian Jie] makes the final executable [Zhi hang] File [Wen Jian] [ke Zhi hang Wen Jian] short, in addition, when the shared object [DUI Xiang] is used by multiple processes [Jin Cheng], it can save some memory [Nei Cun]. because in the memory [Nei Cun], you only need to save the [Dai Ma] Code of the shared object [DUI Xiang]. However, instead of using dynamic [Dong Tai] to link [Lian Jie], it must be superior to using static link [Lian Jie. In some cases, the dynamic [Dong Tai] link [Lian Jie] in [Qing Kuang] may cause some performance [Xing Neng] damages.

The GCC compiler [Bian Yi] [Bian Yi Qi] used in Linux binds the above processes, [Yong HU] can use the [Ming Ling] command only once to compile [Bian Yi], which makes it easier to compile [Bian Yi, however, it is very unfavorable for beginners to understand the compilation process of [Bian Yi], that is, the compilation process of [Bian Yi] by the GCC agent:

 

 

We can see that:

 

Pre-compile [Bian Yi] [Yu Bian Yi]
Convert the. c file [Wen Jian] To the. I file [Wen Jian]
The GCC Command [Ming Ling] used is: gcc-e
Corresponding to the pre-processing command [Ming Ling] CPP

 

Compile [Bian Yi]
Convert the. c/. h file [Wen Jian] into the. s file [Wen Jian]
The GCC Command [Ming Ling] used is: gcc-S
Corresponding to compiling [Bian Yi] command [Ming Ling] CC-S

 

Compile [Hui Bian]
Convert the. s file [Wen Jian] To the. o file [Wen Jian]
The GCC Command [Ming Ling] used is: gcc-C
The [Ming Ling] command for compiling [Hui Bian] is

 

Link [Lian Jie]
Convert the. o file [Wen Jian] to an executable [Zhi hang] program [Zhi hang Cheng Xu] [ke Zhi hang Cheng Xu]
The GCC Command [Ming Ling] used is: gcc
The connection [Lian Jie] command [Ming Ling] Is LD

 

To sum up, the compilation of [Bian Yi] involves the above four processes: pre-compile [Bian Yi] [Yu Bian Yi], compile [Bian Yi], compile [Hui Bian], and link [Lian Jie]. Lia understands the work in these four processes and helps us understand the working processes of header files [Wen Jian] and libraries, in addition, the [Bian Yi] link [Lian Jie] process is clearly understood. We also locate the [Ding Wei] error [CuO Wu] when programming [Bian Cheng]. when programming [Bian Cheng], it is very helpful to mobilize and compile [Bian Yi] [Bian Yi Qi] to detect [CuO Wu] errors.

• 0
• Top