Optimization of application startup speed

Hello everyone , I meet you again , I'm the king of the whole stack

Mac OS/Android Hereunder Static Initializer

Mozillaproject Through optimization Static Initializer( initiate static , Or global constructor , Global Constructor) and Binary Layout to improve FireFox Start speed article . It is of great reference value .

In the article x86 And x86-64 Platform based , Added below Mac OS And Android Upper binary Layout .

What is? Static Initializer? In short, the overall situation C++ Object initialization .

Someone laughed and said C++ programmatic main() Before the function runs , Maybe everything is done . This is it. Static Initializer Influence . Suppose there is another layer of dependent references , It will greatly affect the startup time . The following is a demo sample program :

MyClass oneClass(0x010203); const MyClass twoClass(0x010204); attribute ((constructor)) void foo(void) { printf(“foo is running and printf is available at this point\n”); } int main(int argc, const char * argv[]) { //do something here… }

Copy

 The first two objects oneClass and twoClass That is, two objects that use static initialization ,  and foo Function is forced into the initialization section of the program through compilation options (init segement) in , It will run when the program is initialized . The following is finally in Mac OS Layout on :


 stay Android ARM ELF The following layout is shown in :

Copy

 FireFox The optimization of the   stay Mozillaproject Teacher's article [ link ] in , be based on Firefox 4.0b8 stay x86 And x86-64 The test data found, for example, the following average startup time :
  
      
   
    
     
       
      Average startup time (ms) 
     Pages Read 
     Bytes Read 
     
     
     x86 
     3,228.76 ± 0.57% 
     4,787 
     19,607,552 
     
     
     x86-64 
     3,382.0 ± 0.51% 
     5,874 
     24,059,904 
     
   
  
  
 Use systemtap[ link ] You can get an access to the core library libxul.so Of access pattern:  

  
  
   
    
   
    The red dot indicates the number of pages loaded from the disk 
   
    
   
    The red line is the location in the document (seek) operation 
   
    
   
    The color patches in the background represent .rel.dyn/.rela.syn( Red )..text( Pink ),.rodata( green ),.data.rel.ro( Light green ). 
  
  

Static Initializers
  At the beginning, those vertical lines are Static Initializers Running time , It takes up a lot of time . The solution is to reduce static initializers. Pay special attention to those global variables 、 Static variables . 
 Analyzed in this way , We all have 237 individual static initializers, among 147 By cycle collection globals Introduced . After revision cycle collection The global object of is reduced to a , The whole situation has not changed much : 

  
      
   
    
     
       
      Average startup time (ms) 
     Pages Read 
     Bytes Read 
     
     
     x86 
     3,216.1 ± 0.59% 
     4,656 
     19,070,976 
     
     
     x86-64 
     3,488.14 ± 0.75% 
     5,759 
     23,588,864 
     
   
  
  
 Here are the new I/O access pattern:  
 I/O Although it has decreased , In fact, there are many other reading operations in static initialization It happened before , So there are other jobs to do .
Reordering objects
 Another job is to lay out again binary,  So that the data required by the kernel can be obtained as soon as possible . Before Taras A study found that just do something toolchain Changes on can be achieved .
 Use Taras Of icegrind After optimization . The improvement became obvious :

  
      
   
    
     
       
      Average startup time (ms) 
     Pages Read 
     Bytes Read 
     
     
     x86 
     2,939.18 ± 0.81% 
     4,129 
     16,912,384 
     
     
     x86-64 
     3,247.64 ± 0.68% 
     5,254 
     21,520,384 
     
   
  
  
I/O pattern:  Packing Relocations Last , Then we can reduce relocation paragraph . To optimize startup time . This can effectively reduce I/O, as well as dynamic relocations section, It can also reduce the package . The tools I use are here . Refer to : About adjusting ELF Optimize the startup time. The following is the final effect :
  
      
   
    
     
       
      Average startup time (ms) 
     Pages Read 
     Bytes Read 
     
     
     x86 
     3,149.32 ± 0.62% 
     4,443 
     18,198,528 
     
     
     x86-64 
     3,191.58 ± 0.62% 
     4,733 
     19,386,368 
     
   
  
  
I/O Pattern For example, the following :   This is an obscure topic , Non Party organizations deserve in-depth study . Start with the links provided by the author . My level is limited . throw away a brick in order to get a gem , Looking forward to more in-depth elaboration . Reprint please indicate the source : http://blog.csdn.net/horkychen Refer to   1. How to Make Startup Suck Less (Also Reduce Memory Usage!)  2. Death by static initialization  3. icegrind - Valgrind Plugin for optimizing Cold Startup  4. Resolving ELF Relocation Name / Symbols  5. Static initializers  6. ELF for ARM Architecture

Copy

Copyright notice ： This article is the original article of the blogger , Blog , Do not reprint without permission .

Publisher ： Full stack programmer stack length , Reprint please indicate the source ：https://javaforall.cn/116786.html Link to the original text ：https://javaforall.cn