在Linux应用程序上使用release有很大概率出现core dump

[Yangyuanxin]

杰神，请看这是由内存检测工具打出来的log： 执行以下命令: valgrind --tool=memcheck --leak-check=yes --show-reachable=yes ./my_app 在做这件事情之前，我已经修复了内部线程退出时由linux pthread模块引发的内存泄漏问题，文件：linux\platform_thread.c的platform_thread_destroy函数：

void platform_thread_destroy(platform_thread_t* thread)
{
    if (NULL != thread)
    {
        pthread_detach(thread->thread);
        //内存释放
        platform_memory_free(thread);
        thread = NULL;
    }
}

mqtt_release.c文件：int mqtt_release(mqtt_client_t* c)函数:

    /* wait for the clean session to complete */
    while ((CLIENT_STATE_INVALID != mqtt_get_client_state(c))) {
        // platform_timer_usleep(1000);            // 1ms avoid compiler optimization.
        if (platform_timer_is_expired(&timer)) {
            MQTT_LOG_E("%s:%d %s()... mqtt release failed...", __FILE__, __LINE__, __FUNCTION__);
            //RETURN_ERROR(MQTT_FAILED_ERROR)
            //退出后才会清除没有释放的内存，销毁互斥锁
            break;
        }    
    }

然后，我的应用逻辑如下： 初始化->连接->主题订阅->进入循环->发布两次消息->退出循环->断开连接->执行mqtt_release。

正常情况下是这样的，运行不会产生错误，也没有任何内存泄漏，程序样例正常进行

mqttclient.c:946 mqtt_yield_thread()..., mqtt clean session....
mqttclient.c:1241 mqtt_release()... mqtt release failed...
MqttTest-308::MqttUnInit success!
LogDestroy Success : 0
==46068== 
==46068== HEAP SUMMARY:
==46068==     in use at exit: 0 bytes in 0 blocks
==46068==   total heap usage: 26 allocs, 26 frees, 42,573 bytes allocated
==46068== 
==46068== All heap blocks were freed -- no leaks are possible
==46068== 
==46068== For counts of detected and suppressed errors, rerun with: -v
==46068== ERROR SUMMARY: 0 errors from 0 contexts (suppressed: 0 from 0)

当多次执行该程序后出现Core dump的日志如下：

mqttclient.c:946 mqtt_yield_thread()..., mqtt clean session....
==46058== Thread 2:
==46058== Invalid read of size 8
==46058==    at 0x1237EE: mqtt_yield_thread (mqttclient.c:947)
==46058==    by 0x4E456DA: start_thread (pthread_create.c:463)
==46058==    by 0x517E61E: clone (clone.S:95)
==46058==  Address 0x5457898 is 248 bytes inside a block of size 312 free'd
==46058==    at 0x4C32D3B: free (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==46058==    by 0x11AB5F: AcMqttUnInit (AcMqtt.c:51)
==46058==    by 0x11B4DC: AcMqttTest (AcMqtt.c:302)
==46058==    by 0x1103A9: main (main.c:64)
==46058==  Block was alloc'd at
==46058==    at 0x4C31B0F: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==46058==    by 0x1226BA: mqtt_lease (mqttclient.c:1214)
==46058==    by 0x11AA09: AcMqttInit (AcMqtt.c:18)
==46058==    by 0x11B1AB: AcMqttTest (AcMqtt.c:259)
==46058==    by 0x1103A9: main (main.c:64)
==46058== 
==46058== Invalid read of size 4
==46058==    at 0x1248D0: network_disconnect (network.c:47)
==46058==    by 0x1237F9: mqtt_yield_thread (mqttclient.c:947)
==46058==    by 0x4E456DA: start_thread (pthread_create.c:463)
==46058==    by 0x517E61E: clone (clone.S:95)
==46058==  Address 0x14 is not stack'd, malloc'd or (recently) free'd
==46058== 
==46058== 
==46058== Process terminating with default action of signal 11 (SIGSEGV)
==46058==  Access not within mapped region at address 0x14
==46058==    at 0x1248D0: network_disconnect (network.c:47)
==46058==    by 0x1237F9: mqtt_yield_thread (mqttclient.c:947)
==46058==    by 0x4E456DA: start_thread (pthread_create.c:463)
==46058==    by 0x517E61E: clone (clone.S:95)
==46058==  If you believe this happened as a result of a stack
==46058==  overflow in your program's main thread (unlikely but
==46058==  possible), you can try to increase the size of the
==46058==  main thread stack using the --main-stacksize= flag.
==46058==  The main thread stack size used in this run was 8388608.
==46058== 
==46058== HEAP SUMMARY:
==46058==     in use at exit: 512 bytes in 4 blocks
==46058==   total heap usage: 26 allocs, 22 frees, 42,573 bytes allocated
==46058== 
==46058== Thread 1:
==46058== 40 bytes in 1 blocks are indirectly lost in loss record 1 of 4
==46058==    at 0x4C31B0F: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==46058==    by 0x12315B: mqtt_msg_handler_create (mqttclient.c:428)
==46058==    by 0x12315B: mqtt_subscribe (mqttclient.c:1326)
==46058==    by 0x11B28C: AcMqttSubscribe (AcMqtt.c:122)
==46058==    by 0x11B28C: AcMqttTest (AcMqtt.c:277)
==46058==    by 0x1103A9: main (main.c:64)
==46058== 
==46058== 96 bytes in 1 blocks are definitely lost in loss record 2 of 4
==46058==    at 0x4C31B0F: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==46058==    by 0x182873: platform_thread_init (platform_thread.c:23)
==46058==    by 0x122DAE: mqtt_connect_with_results (mqttclient.c:1033)
==46058==    by 0x122DAE: mqtt_connect (mqttclient.c:1271)
==46058==    by 0x11AC6C: AcMqttConnect (AcMqtt.c:63)
==46058==    by 0x11B1C3: AcMqttTest (AcMqtt.c:266)
==46058==    by 0x1103A9: main (main.c:64)
==46058== 
==46058== 144 (104 direct, 40 indirect) bytes in 1 blocks are definitely lost in loss record 3 of 4
==46058==    at 0x4C31B0F: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==46058==    by 0x120ED8: mqtt_ack_handler_create (mqttclient.c:317)
==46058==    by 0x120ED8: mqtt_ack_list_record (mqttclient.c:388)
==46058==    by 0x123BB3: mqtt_unsubscribe (mqttclient.c:1371)
==46058==    by 0x11AF63: AcMqttUnsubscribe (AcMqtt.c:140)
==46058==    by 0x11B99A: AcMqttTest (AcMqtt.c:291)
==46058==    by 0x1103A9: main (main.c:64)
==46058== 
==46058== 272 bytes in 1 blocks are possibly lost in loss record 4 of 4
==46058==    at 0x4C33B25: calloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==46058==    by 0x4013646: allocate_dtv (dl-tls.c:286)
==46058==    by 0x4013646: _dl_allocate_tls (dl-tls.c:530)
==46058==    by 0x4E46227: allocate_stack (allocatestack.c:627)
==46058==    by 0x4E46227: pthread_create@@GLIBC_2.2.5 (pthread_create.c:644)
==46058==    by 0x182886: platform_thread_init (platform_thread.c:25)
==46058==    by 0x122DAE: mqtt_connect_with_results (mqttclient.c:1033)
==46058==    by 0x122DAE: mqtt_connect (mqttclient.c:1271)
==46058==    by 0x11AC6C: AcMqttConnect (AcMqtt.c:63)
==46058==    by 0x11B1C3: AcMqttTest (AcMqtt.c:266)
==46058==    by 0x1103A9: main (main.c:64)
==46058== 
==46058== LEAK SUMMARY:
==46058==    definitely lost: 200 bytes in 2 blocks
==46058==    indirectly lost: 40 bytes in 1 blocks
==46058==      possibly lost: 272 bytes in 1 blocks
==46058==    still reachable: 0 bytes in 0 blocks
==46058==         suppressed: 0 bytes in 0 blocks
==46058== 
==46058== For counts of detected and suppressed errors, rerun with: -v
==46058== ERROR SUMMARY: 5 errors from 5 contexts (suppressed: 0 from 0)
Segmentation fault (core dumped)

杰神有空可以看看是由哪个模块引起的bug。