Chapter7 Linking(2)——CSAPP讀書筆記

How linkers resolve Multiply defined global symbols 

Rules1. Multiply strong symbols are not allowed

2. Given a strong symbol and multiple weak symbols, choose the strong one.

3. Given multiple weak symbols, choose any of them.

考慮下面一段程序：

在IA32 linux上進行編譯連接時不會報錯，但是實際上是有潛在問題的。由於32位linux上int是4字節而double是8個字節，此時bar5中會將x提升到double類型，此時會造成x在內存中地址的改變‼這點是我們通常會忽視的狀況，很有可能很多莫名其妙的錯誤來源於此…以後需要注意。

GCC -fno-common

加上此標誌後會就重複聲明的全局變量進行提示，實際上，我感覺最好是避免全局變量的重複定義，感覺純粹的OO語言裏面這種現象出現的概率會比較小，但是還是需要注意的地方啊。

此時我們可能會比較疑惑的是…

Overload的function腫麼辦？因爲名字是一樣的嘛。因爲它們的arg list是不一樣的，因此編譯器會根名字和arg list重新打包命名交給linker處理。這個重新命名的過程就是mangling,反之稱爲demangling。

mangling的機制java和c++是互相兼容的。首先類名由原類名和類名的字符個數組成。foo就是foo3。

例如

Foo:: bar(int,long)

變成了 bar__3Fooil

也就是說，mangling的方法是：

OriginalMethodName__MangledClassNameAndLettersOfEachArguments“The general rules for libraries is to place them at end of command line”

也就是說，靜態鏈接庫最好放在編譯命令行的最末，不然的話，若是首先對.a文件進行編譯連接，則此時集合U（unresolved symbols）中沒有任何元素，連接器不會將該靜態庫添加到集合E（生成可執行文件需要的文件）中，接着若是鏈接一個object文件中引用到了靜態庫中的文件時，連接器就會報告unresolved symbol這種錯誤。

所以在編譯連接時，越是基本的庫約要放在較後的位置進行處理。若是各個庫之間沒有相互引用，則無所謂放置的位置。

Libraries can be repeated on the command line if necessary to satisfy the dependence requirements

也就是說…要是你願意…同一個庫你可以在命令行重複輸入多遍，但是最好不要這麼做…這樣的話建議將能合併的庫都合併了（ーー；）

例題：寫個最短的命令行解決編譯問題，要解決各種依賴關係。a-＞b表示a依賴b

p.o->libx.a->liby.a and liby.a->libx.a->p.o

solution:

 gcc p.o libx.a liby.a libx.a

7.7 Relocation

Relocation consists of two steps 1. Relocating sections and symbol definitions. After this step, every instruction and global variable in the program has a unique run-time memory address. 2. Relocating symbol reference within sections. To perform this step, the linker relies on data structures in the relocatable object module known as the relocatable entries.

根據deassembler課題看到，重新定位需要相應的重定位類型，地址以及symbol。

之後成爲一個可執行文件，unix shell會根據其類型調用系統駐留程序，即loader，進行加載運行。

Disadvantage :

Static libraries, like all software, need to be maintained and updated periodically

7.7 Relocation

Relocation consists of two steps 1. Relocating sections and symbol definitions. After this step, every instruction and global variable in the program has a unique run-time memory address. 2. Relocating symbol reference within sections. To perform this step, the linker relies on data structures in the relocatable object module known as the relocatable entries.

根據deassembler課題看到，重新定位需要相應的重定位類型，地址以及symbol。

之後成爲一個可執行文件，unix shell會根據其類型調用系統駐留程序，即loader，進行加載運行。

Disadvantage :

Static libraries, like all software, need to be maintained and updated periodically

7.10 Dynamic linking with shared libraries

Shared libraries are shared in 2 different ways:

First, in any given file system, there is exactly one .so file for a particular library. Second, a single copy of the .text section of a shared library in memory can be shared by different running processes.

Linux系統下允許應用程序調用外部的動態鏈接庫的接口是：

#include <dlfcn.h>

void *dlopen(const char *filename, int flag);

該函數用於加載動態鏈接庫filename

#include <dlfcn.h>

void *dlsym(void *handle, char *symbol)

該函數通過前一個打開的動態鏈接庫的句柄，以及symbol名，返回symbol的地址，沒有的話返回NULL

#include <dlfcn.h>

int dlclose(void *handle);

若無其他動態鏈接庫佔用，可對當前動態鏈接庫進行卸載操作。

#include <dlfcn.h>

const char *dlerror(void);

對於上三個函數調用過程中錯誤信息的返回。

7.12 Position Independent Code (PIC)

How can multiple process share a single copy of code? Compile library code that it can be loaded and executed at any address. Such code is known as position-independent code(PIC)

不管在何處加載何種object模塊，data段永遠緊跟在code段之後進行地址的分配。

這樣保證了地址的差值是一定的，也就是說實際上使用的是相對地址，這樣就不會有絕對地址會造成的問題了。所以編譯器在段一開始處便會有一個偏移地址對照表，方便之後根據對照表進行地址的查找。

Tools for manipulating object files AR: Creates static libraries, and inserts, deletes, lists, and extracts members. STRINGS: Lists all of the printable strings contained in an object file. STRIP: Delete symbol table information from an object file. NM: Lists the names and sizes of the sections in an object file. SIZE: Lists the name and sections in an object file. READELF: Displays the complete structure of an object file, including all of the information encoded in the ELF header, subsumed the functionality of SIZE and NM. OBJDUMP: The mother of all library tools. Can display all the information in an object file. It's most useful function is disassembling the binary instructions in the .text section. LDD: list the shared libraries that is needed at run time.