How To Write Shared Libraries(6)

1.5 Startup in the Dynamic Linker

The second phase of the program startup happens in the dynamic linker. Its tasks include:
链接器启动第二阶段的任务包括：
• Determine and load dependencies;
• Relocate the application and all dependencies;
• Initialize the application and dependencies in the correct order.
判断加载依赖；
重定位应用和依赖；
正确顺序的初始化应用和依赖。

In the following we will discuss in more detail only the relocation handling. For the other two points the way for better performance is clear: have fewer dependen- cies. Each participating object is initialized exactly once but some topological sorting has to happen. The identify and load process also scales with the number dependen- cies; in most (all?) implementations this does not scale linearly.
接下来，我们更多的讨论重定位相关内容。
另外两点的优化效果很清晰：更少的依赖。
每个部分的对象在排序后可以正确的完成初始化。
识别和加载的过程随着依赖扩展。大部分这里不是线性扩展的。

The relocation process is normally the most expensive part of the dynamic linker’s work. It is a process which is asymptotically at least O(R + nr) where R is the number of relative relocations, r is the number of named reloca- tions, and n is the number of participating DSOs (plus the main executable). Deficiencies in the ELF hash ta- ble function and various ELF extensions modifying the symbol lookup functionality may well increase the factor to O(R + rn log s) where s is the number of symbols. This should make clear that for improved performance it is significant to reduce the number if relocations and sym- bols as much as possible. After explaining the relocation process we will do some estimates for actual numbers.
一般情况下重加载过程是最耗时的部分。接近O(R + nr)：R是重定位的数量，r是命名的数量，n是DSO数量。
ELF缺少hash函数，ELF扩展修改语法查找函数可以增加到O(R + rn log s)：s是语法数量。这样可以显著提高重定位效率。（待完善）
解释完重定位内容，评估一下实际数量。