fix(vd960DBN): 联网 SocketSend 0x11 死循环 — 发送退避 + SocketCreat 失败中止

现象: loop_data 606B 发送 0x11(ERR_MEM) → 立即重试10次 → 3次断连 →
重连 SocketCreat 0x1D(ISCONN) 仍继续 Connect → timeout 死循环永久失联

根因:
1. WCHNET_NUM_TCP_SEG=2 (瘦身改小): 重传队列占用即无发送缓冲 → 0x11 瞬时
2. iot_mqtt_send 失败处理过激: 无退避立即重试 + 3次就强制断连
3. WCHNET_CreateTcpMqttSocket: SocketCreat 失败(mStopIfError只打印)仍继续
   Connect → 无效 socket 等超时死循环; SocketId_TCP 未置 0xFF 沿用旧 id

修复:
- iot_mqtt_send: 0x11 200ms 退避重试(≤10次, 等SEG释放), 不立即断连
- 强制重连: 补 SocketId_TCP=0xFF
- WCHNET_CreateTcpMqttSocket: SocketCreat 失败置 0xFF + return 不 Connect,
  上层指数退避

验证: 语法 0 错误; 待板级确认重发自动恢复 + 重连退避
This commit is contained in:
wangfq
2026-08-20 18:16:31 +08:00
parent c305ca651f
commit 1badba3893
3 changed files with 59 additions and 12 deletions
@@ -351,7 +351,17 @@ static int iot_mqtt_send(const uint8_t *buf, uint16_t len) {
uint32_t slen = len - total_sent; uint32_t slen = len - total_sent;
uint8_t ret = WCHNET_SocketSend(g_iot_socket, (uint8_t *)(buf + total_sent), &slen); uint8_t ret = WCHNET_SocketSend(g_iot_socket, (uint8_t *)(buf + total_sent), &slen);
total_sent += (uint16_t)slen; total_sent += (uint16_t)slen;
if (slen == 0 || ret != WCHNET_ERR_SUCCESS || ++retries > 10) { if (slen == 0 || ret != WCHNET_ERR_SUCCESS) {
/* 0x11 = WCHNET_ERR_MEM: 瞬时发送缓冲不足 (WCHNET_NUM_TCP_SEG=2 极小,
重传队列占用即无缓冲)。200ms 退避重试等 WCHNET 释放 SEG —
不要立即断连: 断连重连会让 WCHNET socket 表更乱
(SocketCreat ISCONN 死循环, 2026-08-20 现场) */
if (ret == WCHNET_ERR_MEM && retries < 10) {
uint32_t w = mstick();
while ((mstick() - w) < 200) { } /* 退避 (IWDG 4s 内安全) */
}
retries++;
if (retries > 10) {
_iot_send_fail_cnt++; _iot_send_fail_cnt++;
PRINT("IOT: SocketSend FAIL#%d ret=0x%02X sent=%u/%u\n", PRINT("IOT: SocketSend FAIL#%d ret=0x%02X sent=%u/%u\n",
_iot_send_fail_cnt, ret, total_sent, len); _iot_send_fail_cnt, ret, total_sent, len);
@@ -359,12 +369,15 @@ static int iot_mqtt_send(const uint8_t *buf, uint16_t len) {
PRINT("IOT: SocketSend 3 consecutive failures, forcing reconnect\n"); PRINT("IOT: SocketSend 3 consecutive failures, forcing reconnect\n");
WCHNET_SocketClose(g_iot_socket, TCP_CLOSE_NORMAL); WCHNET_SocketClose(g_iot_socket, TCP_CLOSE_NORMAL);
g_iot_socket = 0xFF; g_iot_socket = 0xFF;
SocketId_TCP = 0xFF; /* 防重连 SocketCreat 沿用旧 id (2026-08-20) */
g_iot_state = IOT_STATE_DISCONNECTED; g_iot_state = IOT_STATE_DISCONNECTED;
_iot_send_fail_cnt = 0; _iot_send_fail_cnt = 0;
_iot_reconnect_deadline = mstick() + 3000; // 3s 后重连, 让 WCHNET 清旧 socket _iot_reconnect_deadline = mstick() + 3000; // 3s 后重连, 让 WCHNET 清旧 socket
} }
return -1; return -1;
} }
continue; /* 重试 (MEM 已退避 200ms, 其他错误立即重试) */
}
} }
_iot_send_fail_cnt = 0; // 成功一次就清零 _iot_send_fail_cnt = 0; // 成功一次就清零
return 0; return 0;
@@ -346,6 +346,13 @@ void WCHNET_CreateTcpMqttSocket(void)
{ {
PRINT("WCHNET_MQTT_SocketCreate %d,desIP:%d.%d.%d.%d, des_port:%d, srcport:%d\r\n", SocketId_TCP, RemoteIP[0],RemoteIP[1],RemoteIP[2],RemoteIP[3], iot_net_info.mqtt_port, TmpSocketInf.SourPort); PRINT("WCHNET_MQTT_SocketCreate %d,desIP:%d.%d.%d.%d, des_port:%d, srcport:%d\r\n", SocketId_TCP, RemoteIP[0],RemoteIP[1],RemoteIP[2],RemoteIP[3], iot_net_info.mqtt_port, TmpSocketInf.SourPort);
} }
if (i != WCHNET_ERR_SUCCESS) {
/* 2026-08-20: SocketCreat 失败 (0x1D ISCONN 等, 旧 socket 未清干净)
置无效并返回, 由上层退避重试 — 继续 Connect 只会等超时死循环 */
PRINT("MQTT: SocketCreat failed 0x%02X, abort connect\r\n", i);
SocketId_TCP = 0xFF;
return;
}
mStopIfError(i); mStopIfError(i);
i = WCHNET_SocketConnect(SocketId_TCP); //make a TCP connection i = WCHNET_SocketConnect(SocketId_TCP); //make a TCP connection
if(i == WCHNET_ERR_SUCCESS) if(i == WCHNET_ERR_SUCCESS)
+27
View File
@@ -4,6 +4,33 @@
> >
> 项目定位: DLD960 通信板 — BLE 配网、TCP JSON 协议服务、Loop MCU 串口桥接 > 项目定位: DLD960 通信板 — BLE 配网、TCP JSON 协议服务、Loop MCU 串口桥接
## 2026-08-20 — 联网 SocketSend 0x11 死循环修复:发送退避 + SocketCreat 失败中止
### 现象(板级)
收 report_config 后 loop_data606B)发送失败:`SocketSend FAIL ret=0x11 sent=0/606` → 立即重试 10 次无效 → 3 次强制断连 → 重连 SocketCreat 返回 **0x1D(ISCONN)** + Connect 0x17(CLSD) → TCP connect timeout 死循环,永久失联。
### 根因
1. **WCHNET_NUM_TCP_SEG = WCHNET_NUM_TCP×2 = 2**(RAM 瘦身改小):TCP 发送段缓冲仅 2 个,重传队列占用即无缓冲 → 606B 发送 `WCHNET_ERR_MEM(0x11)`
2. **iot_mqtt_send 失败处理过激**:0x11 是瞬时(SEG 被占),却立即重试 10 次(无退避,WCHNET 没时间释放)+ 3 次就强制断连
3. **重连缺陷**`WCHNET_CreateTcpMqttSocket` 里 SocketCreat 失败(0x1D ISCONN,旧 socket 未清干净)后**仍继续 Connect**mStopIfError 只打印不中止)→ 无效 socket 等超时 → 死循环;且强制重连时 `SocketId_TCP` 未置 0xFFSocketCreat 沿用旧 id
### 修复
| 位置 | 改动 |
|------|------|
| `iot_mqtt_srv.c` iot_mqtt_send | 0x11 时 **200ms 退避重试**(等 WCHNET 释放 SEG,最多 10 次×200ms≈2s),不立即断连;非 MEM 错误立即重试 |
| `iot_mqtt_srv.c` 强制重连 | 补 `SocketId_TCP = 0xFF`(防 SocketCreat 沿用旧 id |
| `net_srv.c` WCHNET_CreateTcpMqttSocket | SocketCreat 失败 → 打印 + `SocketId_TCP=0xFF` + **return 不 Connect**,由上层指数退避重试 |
### 验证
- 语法 0 错误
- **待板级确认**:重发场景(瞬时 0x11)应自动恢复不断连;真断连重连失败应退避而非死循环
---
## 2026-08-20 — OTA 下载失败修复:hex 提取不兼容 json.dumps 空格(单片 CRC 全错) ## 2026-08-20 — OTA 下载失败修复:hex 提取不兼容 json.dumps 空格(单片 CRC 全错)
### 现象(板级) ### 现象(板级)