多线程使用libcurl的坑

用libcurl一段时间遇到莫名其妙的程序崩溃的情况，开始觉得是线程栈溢出导致的段错误，专门增加了线程栈的大小，还是有问题。线程也是分离的。用valgrind定位到问题可能出现在curl的调用上。

排查的时候也发现了libcurl一些额外的坑，现做个总结笔记。

1.、

线程使用libcurl访问时，设置了超时时间，而libcurl库不会为这个超时信号做任何处理，信号产生而没有信号句柄处理，可能导致程序退出。用以下选项禁止访问超时的时候抛出超时信号。

curl_setopt(curl, CURLOPT_NOSIGNAL,1L);

2、

正常使用流程是先调用curl_global_init初始化资源，而这个函数不是线程安全的。curl_easy_init发现没有做初始化时会自动初始化，多个线程同时进行curl_easy_init时会导致异常。curl官方建议在主线程中先做初始化。

curl_global_init(CURL_GLOBAL_ALL); 

3、

用valgrind调试，做压力测试多线程运行时，valgrind弹出类似访问冲突的错误，把错误定位到libcurl的互斥访问的处理上。

大致是进行访问设计到ssl的东西的时候，需要做相应的加锁处理，libcurl源码没有加这块处理，不过给了个针对这种情况加锁的例子，总的来说curl的示例代码很重要很有用。

https://curl.haxx.se/libcurl/c/example.html

#define USE_OPENSSL  
 
#include <stdio.h>
#include <pthread.h>
#include <curl/curl.h>
 
#define NUMT 4
 
/* we have this global to let the callback get easy access to it */ 
static pthread_mutex_t *lockarray;
 
#ifdef USE_OPENSSL
#include <openssl/crypto.h>
static void lock_callback(int mode, int type, char *file, int line)
{
  (void)file;
  (void)line;
  if(mode & CRYPTO_LOCK) {
    pthread_mutex_lock(&(lockarray[type]));
  }
  else {
    pthread_mutex_unlock(&(lockarray[type]));
  }
}
 
static unsigned long thread_id(void)
{
  unsigned long ret;
 
  ret=(unsigned long)pthread_self();
  return ret;
}
 
static void init_locks(void)
{
  int i;
 
  lockarray=(pthread_mutex_t *)OPENSSL_malloc(CRYPTO_num_locks() *
                                            sizeof(pthread_mutex_t));
  for(i=0; i<CRYPTO_num_locks(); i++) {
    pthread_mutex_init(&(lockarray[i]), NULL);
  }
 
  CRYPTO_set_id_callback((unsigned long (*)())thread_id);
  CRYPTO_set_locking_callback((void (*)())lock_callback);
}
 
static void kill_locks(void)
{
  int i;
 
  CRYPTO_set_locking_callback(NULL);
  for(i=0; i<CRYPTO_num_locks(); i++)
    pthread_mutex_destroy(&(lockarray[i]));
 
  OPENSSL_free(lockarray);
}
#endif
 
#ifdef USE_GNUTLS
#include <gcrypt.h>
#include <errno.h>
 
GCRY_THREAD_OPTION_PTHREAD_IMPL;
 
void init_locks(void)
{
  gcry_control(GCRYCTL_SET_THREAD_CBS);
}
 
#define kill_locks()
#endif
 
/* List of URLs to fetch.*/ 
const char * const urls[]= {
  "https://www.example.com/",
  "https://www2.example.com/",
  "https://www3.example.com/",
  "https://www4.example.com/",
};
 
static void *pull_one_url(void *url)
{
  CURL *curl;
 
  curl = curl_easy_init();
  curl_easy_setopt(curl, CURLOPT_URL, url);
  /* this example doesn't verify the server's certificate, which means we
     might be downloading stuff from an impostor */ 
  curl_easy_setopt(curl, CURLOPT_SSL_VERIFYPEER, 0L);
  curl_easy_setopt(curl, CURLOPT_SSL_VERIFYHOST, 0L);
  curl_easy_perform(curl); /* ignores error */ 
  curl_easy_cleanup(curl);
 
  return NULL;
}
 
int main(int argc, char **argv)
{
  pthread_t tid[NUMT];
  int i;
  int error;
  (void)argc; /* we don't use any arguments in this example */ 
  (void)argv;
 
  /* Must initialize libcurl before any threads are started */ 
  curl_global_init(CURL_GLOBAL_ALL);
 
  init_locks();
 
  for(i=0; i< NUMT; i++) {
    error = pthread_create(&tid[i],
                           NULL, /* default attributes please */ 
                           pull_one_url,
                           (void *)urls[i]);
    if(0 != error)
      fprintf(stderr, "Couldn't run thread number %d, errno %d\n", i, error);
    else
      fprintf(stderr, "Thread %d, gets %s\n", i, urls[i]);
  }
 
  /* now wait for all threads to terminate */ 
  for(i=0; i< NUMT; i++) {
    error = pthread_join(tid[i], NULL);
    fprintf(stderr, "Thread %d terminated\n", i);
  }
 
  kill_locks();
 
  return 0;
}

未完待续。。。