FCGI single-threaded environment and multithreaded environment examples

0. Background

The content that waits for communication is added to the project, so a single request processing time has increased. Single-threaded fcgi waste CPU and user time, so multithreading is required to reduce user queuing time.

Changing the processing of user requests from single-threaded to multithreaded requires a general understanding of whether the changes will affect performance.

The

concludes that multithreading and single-threaded processes are almost the same as those used in a single thread, so multithreading does not add up to additional burdens. 1. single-threaded Processing Steps 1.1 A simple single-threaded fcgi request C code    #include  <fcgi_stdio.h>      Void main (void)    {       int count = 0;       while (Fcgi_accept ()  >= 0)  {            printf ("content-type: text/html\r\n");            printf ("\ r \ n");           printf ("Hello  world!<br>\r\n ");           printf (" Request  number %d. ",  count++);       }       exit ( 0);  }   1.2 into fcgi_accept.

Enter this  fcgi_accept ()   method inside the file FCGI_STDIO.C. The C code    int fcgi_accept (void)    {       //variable Indicates whether the request is received. Default is Fasle, no request        if (!acceptcalled)  {            //judge whether it is CGI, the variable is global static, the next time will be used.            iscgi = fcgx_iscgi ();     The        //state is changed to receive requests.            acceptCalled = TRUE;            //the end of the request, empty the value to the initial value.            atexit (&fcgi_finish);        } else if (iscgi)  {           // is not the first request, and is a CGI program.            return (EOF);       }       if (iscgi)  {            //cgi initial assignment operation, does not care.            ...       } else  {           fcgx_stream *in, *out, * error;           //char**  string array.            FCGX_ParamArray envp;            //accept the request, this method is described in the following             int acceptresult = fcgx_accept (&AMP;IN,&NBSP;&AMP;OUT,&NBSP;&AMP;ERROR,&NBSP;&AMP;ENVP);            //receives failure, returns &LT;0, which is why the judgment on the loop is  while (fcgi_ Accept ()  >= 0)            if (acceptresult  < 0)  {               return acceptResult;            }            //assigns the resulting data to the corresponding output, input, data.            FCGI_stdin->stdio_stream = NULL;            FCGI_stdin->fcgx_stream = in;            FCGI_stdout->stdio_stream = NULL;            FCGI_stdout->fcgx_stream = out;           FCGI_stderr->stdio_stream = NULL;           FCGI_stderr->fcgx_stream = error;           environ = envp;   &NBSP;&NBsp;  }       //End        return 0;   }   1.3  fcgx_accept  (&in, &out, &error, &ENVP)

Wait for the method to receive the request, in the FCGIAOO.C. C code    static fcgx_request the_request;      int fcgx_accept (fcgx_ STREAM&NBSP;**IN,FCGX_STREAM&NBSP;**OUT,FCGX_STREAM&NBSP;**ERR,FCGX_PARAMARRAY&NBSP;*ENVP)    {    &NBSP;&NBSP;&NBSP;&NBSP;INT&NBSP;RC;//defines the returned variable.    whether the     //has been initialized.        if  (! libinitialized)  {            rc = fcgx_init ();            if  (RC)  {                return rc;           }        }       //receive data, the following describes        rc = fcgx_ Accept_r (&the_request);       //assigns values to corresponding streams and data.    &NBSP;&NBSP;&NBSP;&Nbsp;*in = the_request.in;       *out = the_request.out;        *err = the_request.err;       *envp  = the_request.envp;          return rc;  }    1.4 fcgx_accept_r  (), in fcgiapp.c;