邮箱阻塞接收设为永久阻塞却收到了RT_ETIMEOUT错误码

发布于 2020-08-20 12:01:47 浏览：1994 订阅该版

问题出现在以太网接收线程中，收到网络数据后如果转发比较慢的话跑几分钟会出现死机，看了一下log问题是由此线程中的断言`LWIP_ASSERT`引起的。
----------
调试过程发现`rt_mb_recv`返回`-2`即`-RT_ETIMEOUT`。另外在`rt_mb_recv()`内部返回错误的地方也加入了调试语句，发现`thread->error`为`-2`如图，信箱的地址也和接收线程中做了对比是一致的。
----------
另外测试过把`LWIP_ASSERT`改成普通的log后运行正常，数据也不丢失。不清楚为什么永久等待却出现了超时？还有个疑问`rt_mb_recv()`返回的错误码是`thread->error`，这个错误码似乎是`线程`而非`邮箱`的，这样是否合理呢，用4.0.2稳定版和最新的4.0.3都测出来一样的问题了

![image.png](/uploads/20200820/cff5c7a9c08f248a079a8581388173b1.png)

```c
static void eth_rx_thread_entry(void* parameter)
{
    rt_err_t ret;
    struct eth_device* device;

rt_kprintf("Should not happen! ret = %d, mb: %p\r\n", ret, ð_rx_thread_mb);

while (1)
    {
        ret = rt_mb_recv(ð_rx_thread_mb, (rt_ubase_t *)&device, RT_WAITING_FOREVER);
        if (ret == RT_EOK)
        {
            struct pbuf *p;

/* check link status */
            if (device->link_changed)
            {
                int status;
                rt_uint32_t level;

level = rt_hw_interrupt_disable();
                status = device->link_status;
                device->link_changed = 0x00;
                rt_hw_interrupt_enable(level);

if (status)
                    netifapi_netif_set_link_up(device->netif);
                else
                    netifapi_netif_set_link_down(device->netif);
            }

/* receive all of buffer */
            while (1)
            {
                if(device->eth_rx == RT_NULL) break;
                
                p = device->eth_rx(&(device->parent));
                if (p != RT_NULL)
                {
                    /* notify to upper layer */
                    if( device->netif->input(p, device->netif) != ERR_OK )
                    {
                        LWIP_DEBUGF(NETIF_DEBUG, ("ethernetif_input: Input error\n"));
                        pbuf_free(p);
                        p = NULL;
                    }
                }
                else break;
            }
        }
        else
        {
            //LWIP_ASSERT("Should not happen!\n", ret);
            rt_kprintf("Should not happen! ret = %d, mb: %p\r\n", ret, ð_rx_thread_mb);
            RT_ASSERT(0);
        }
    }
}
```

4 个回答

红枫认证专家 2020-08-20

这家伙很懒，什么也没写！

Jone 2020-08-20

写了还是懒

红枫 2020-08-20

这家伙很懒，什么也没写！

请登录后再发布评论，点击登录

追加回复

Jone 2020-08-20

写了还是懒

aozima 2020-08-20

调网络不抓包，调I2C等时序不上逻辑分析仪，就像电工不用万用表！多用整理的好的文字，比截图更省流量，还能在整理过程中思考。

Jone 2020-08-20

写了还是懒

请登录后再发布评论，点击登录

追加回复

lizhen9880 认证专家 2020-08-20

这家伙很懒，什么也没写！

Jone 2020-08-20

写了还是懒

问题出现在以太网接收线程`eth_rx_thread_entry`中
```c
Should not happen! ret = -2
thread   pri  status      sp     stack size max used left tick  error
-------- ---  ------- ---------- ----------  ------  ---------- ---
U1_Sa      9  ready   0x000000d4 0x00000800    35%   0x00000009 -02
atcmd      7  suspend 0x0000012c 0x00000600    25%   0x00000007 000
uart1      8  suspend 0x0000020c 0x00000600    74%   0x00000009 000
search    15  suspend 0x00000328 0x00000800    49%   0x00000007 000
tshell    20  suspend 0x000000f8 0x00001000    06%   0x0000000a 000
sys_work  23  suspend 0x0000005c 0x00000800    42%   0x0000000a 000
tcpip     10  ready   0x00000098 0x00000600    38%   0x0000000f -02
etx       12  suspend 0x00000081 0x00000400    21%   0x00000009 000
erx       12  running 0x000000b1 0x00000400    46%   0x00000001 -02
tidle0    31  ready   0x0000004c 0x00000100    39%   0x0000001b -02
timer      4  suspend 0x00000058 0x00000200    22%   0x00000009 000
main      10  suspend 0x000001a0 0x00000800    50%   0x00000002 000

(0) has assert failed at eth_rx_thread_entry:825.
Firmware name: rtthread, hardware version: 1.0, software version: 1.0
Assert on thread erx
===== Thread stack information =====
  addr: 24001c38    data: 0000014b
  addr: 24001c3c    data: 00000339
  addr: 24001c40    data: 9003c068
  addr: 24001c44    data: 9006a2fd
  addr: 24001c48    data: deadbeef
  addr: 24001c4c    data: 01003627
  addr: 24001c50    data: 9003c121
  addr: 24001c54    data: 900266e9
  addr: 24001c58    data: deadbeef
  addr: 24001c5c    data: 900036af
  addr: 24001c60    data: 9006a2fd
  addr: 24001c64    data: 0003c068
  addr: 24001c68    data: fffffffe
  addr: 24001c6c    data: 00000001
  addr: 24001c70    data: deadbeef
  addr: 24001c74    data: 90007721
  addr: 24001c78    data: 00000000
  addr: 24001c7c    data: 24003818
  addr: 24001c80    data: 00000000
  addr: 24001c84    data: deadbeef
  addr: 24001c88    data: deadbeef
  addr: 24001c8c    data: 900043d5
  addr: 24001c90    data: 23232323
  addr: 24001c94    data: 23232323
====================================
Show more call stack info by run: addr2line -e rtthread.elf -a -f 900036ae 90007720 
Current system tick: 98353
```