WifiClient exception 29

[s0170071]

Basic Infos

• [x ] This issue complies with the issue POLICY doc.

• [x ] I have read the documentation at readthedocs and the issue is not addressed there.

• [x ] I have tested that the issue is present in current master branch (aka latest git).

• [x ] I have searched the issue tracker for a similar issue.

• [ x] If there is a stack dump, I have decoded it.

• [x ] I have filled out all fields below.

Platform

• Hardware: [ESP-12]
• Core Version: latest
• Development Env: [Arduino IDE]
• Operating System: [Ubuntu]

Settings in IDE

• Module: [Wemos D1 mini r2]
• Flash Mode: [dio]
• Flash Size: [4MB]
• lwip Variant: [v2 Lower Memory]
• Reset Method: [ck|nodemcu]
• Flash Frequency: [40Mhz]
• CPU Frequency: [80Mhz]
• Upload Using: [SERIAL]
• Upload Speed: [115200] (serial upload only)

Problem Description

This issue is related to #4487.
What changed is that I now have two instances of the SMA class which means I do have two instances of the WiFiClient too.

Sketch difference to #4487 is pretty much just the fact that I instantiated the class twice :

  mySMA = new SMA;
  mySMA2 = new SMA;
  dummy[1]=0;
}

void loop ( void ) {
 
  delay(100); 
  Serial.println(ESP.getFreeHeap());
  mySMA->handleTXRX();
  mySMA2->handleTXRX();
}

It does drain the memory twice as fast but instead of stalling, it is throwing an exception:

connect 192.168.1.41 OK, sending: 0 1 0 0 0 6 126 4 157 7 0 1
connect 192.168.1.41 fail.
1328
1160
reading bytes: 0 1 0 0 0 5 126 4 2 16 16 pow: -32768
1328
1328
1328
1328
1328
1328
1328
1328
1328
connect 192.168.1.41 OK, sending: 0 1 0 0 0 6 126 4 157 7 0 1

Exception (29):
epc1=0x40202e7b epc2=0x00000000 epc3=0x00000000 excvaddr=0x00000000 depc=0x00000000

ctx: cont 
sp: 3fff8330 end: 3fff8590 offset: 01a0

>>>stack>>>
3fff84d0:  7a797877 00000001 3ffe8b65 40202e79  
3fff84e0:  2901a8c0 00000001 3fff8521 3fff9209  
3fff84f0:  c02363d8 0000003e 3fff92d4 3ffe84dc  
3fff8500:  3fff92bc 000001f6 3fff92bc 40202a20  
3fff8510:  401070b0 2901a8c0 401070b0 2901a8c0  
3fff8520:  3ffe88a3 00000004 3fff91fc 3fff755c  
3fff8530:  3fff92bc 3fff92d4 3fff9294 402022d0  
3fff8540:  00000000 00000000 3fff749c 40203670  
3fff8550:  402033f0 00000064 00000064 3fff755c  
3fff8560:  3fffdad0 00000000 3fff7554 402024fc  
3fff8570:  3fffdad0 00000000 3fff7554 40203cec  
3fff8580:  feefeffe feefeffe 3fff7570 401006ed  
<<<stack<<<

 ets Jan  8 2013,rst cause:2, boot mode:(3,6)

load 0x4010f000, len 1384, room 16 
tail 8
chksum 0x2d
csum 0x2d
v3b269c4a
~ld

Exception 29: StoreProhibited: A store referenced a page mapped with an attribute that does not permit stores
Decoding 12 results
0x40202e7b: ClientContext at /home/john/ArduinoPortable/arduino-1.8.5_ESPgit/hardware/esp8266com/esp8266/libraries/ESP8266WiFi/src/WiFiClient.cpp line 371
:  (inlined by) WiFiClient::connect(IPAddress, unsigned short) at /home/john/ArduinoPortable/arduino-1.8.5_ESPgit/hardware/esp8266com/esp8266/libraries/ESP8266WiFi/src/WiFiClient.cpp line 137
0x40202e79: WiFiClient::connect(IPAddress, unsigned short) at /home/john/ArduinoPortable/arduino-1.8.5_ESPgit/hardware/esp8266com/esp8266/libraries/ESP8266WiFi/src/WiFiClient.cpp line 371
0x40202a20: WiFiClient::connect(char const*, unsigned short) at /home/john/ArduinoPortable/arduino-1.8.5_ESPgit/hardware/esp8266com/esp8266/libraries/ESP8266WiFi/src/WiFiClient.cpp line 371
0x401070b0: gdb_init at ?? line ?
0x401070b0: gdb_init at ?? line ?
0x402022d0: SMA::handleTXRX() at /home/john/Arduino/scetchbooks/clientmemloss/clientmemloss.ino line 71
0x40203670: Print::println() at /home/john/ArduinoPortable/arduino-1.8.5_ESPgit/hardware/esp8266com/esp8266/cores/esp8266/Print.cpp line 99
0x402033f0: EspClass::getFreeHeap() at /home/john/ArduinoPortable/arduino-1.8.5_ESPgit/hardware/esp8266com/esp8266/cores/esp8266/Esp.cpp line 165
0x402024fc: loop at /home/john/Arduino/scetchbooks/clientmemloss/clientmemloss.ino line 183
0x40203cec: loop_wrapper at /home/john/ArduinoPortable/arduino-1.8.5_ESPgit/hardware/esp8266com/esp8266/cores/esp8266/core_esp8266_main.cpp line 58
0x401006ed: cont_wrapper at /home/john/ArduinoPortable/arduino-1.8.5_ESPgit/hardware/esp8266com/esp8266/cores/esp8266/cont.S line 81

[s0170071]

It appears that the connection is not closed after the communication is done. client.stop() closes it, however, the issue remains.

[debsahu]

I am facing similar issues. Developing a library for esp8266 to talk to Twitter https://github.com/debsahu/TwitterWebAPI . Unfortunately, library works fine on 2.4.0 but after the upgrade to 2.4.1, I also get exception 29. I use WiFiSecureClient to send GET data and close the connection after every request. @igrr recommends to have only one client open due to the memory size. The recommendation is to use client.flush() and client.stop(), which I do and still get exception 29. Please help.

[s0170071]

I got it to work on 2.3.0 when I call tcpCleanup regularly.

struct tcp_pcb;
extern struct tcp_pcb* tcp_tw_pcbs;
extern "C" void tcp_abort (struct tcp_pcb* pcb);

void tcpCleanup()
{
         while(tcp_tw_pcbs!=NULL)
    {
      tcp_abort(tcp_tw_pcbs);
    }
 }

It does however not work with the current core and LWIP 1.4 (compile from source)

[igrr]

@s0170071 could you please upload a reproducer sketch? Thank you.

[s0170071]

@igrr: please see #4487. Its the same sketch. This issue uses two instances of the SMA class which gives a different kind of error. The underlying problem seems to be a memory leak if the connection is not closed by the peer.
Beware, for the sketch to work you would need a responding peer. If the client can't connect or doesn't receive anything nothing happens.

[igrr]

Well, that was my point — I am not sure how exactly to modify the sketch to reproduce this issue. I would be grateful if you give instructions to reproduce, including how to implement the peer side.

[s0170071]

ok, I just used another esp to respond with 11 bytes when contacted at port 502. They talk but I cannot reproduce the issue. But I observed that the connection is closed when the communication is done.
My SMA peer leaves the connection open.
Whats needed is a way to force the connection closure an freeing the allocated memory in the process.

[s0170071]

Is there a way to make the server not close the connection? That could help to reproduce.

[igrr]

Maybe connect to an HTTP server and send a Connect: keep-alive header?

[s0170071]

There is no such thing as headers when you play with modbus. And the client class has no method for headers :-)
I found a workaround that may work for me. Since I communicate cyclically I just check if the connection is still alive before calling connect() again.
Doesn't fix the issue though.

[s0170071]

Nope. No real workaround. The leak is smaller but now and then it drops a few bytes... Slowly. Tested with latest git.

[debsahu]

Try adding client.setTimeout(time_ms) before client.connect(). Seems to solve my issues with exception 29.

EDIT: nevermind. It was a bit stable for ~10min.

[s0170071]

@igrr Here is a server sketch that leaves the connection open. The trick is to just not send anything ;-)
Stupid but effective. You would have to match the IP in the client sketch to that server ESP. The client connects, sends and times out.

#include <ESP8266WiFi.h>
#include <WiFiClient.h>
#include <WiFiServer.h>
  
 const char *ssid = "...";
const char *password = "...";
WiFiServer server(502);
  

void setup ( void ) {
  Serial.begin ( 115200 );
  WiFi.mode ( WIFI_STA );
  WiFi.begin ( ssid, password );
  Serial.println ( "" );

  // Wait for connection
  while ( WiFi.status() != WL_CONNECTED ) {
    delay ( 500 );
    Serial.print ( "." );
  }

  Serial.println ( "" );
  Serial.print ( "Connected to " );
  Serial.println ( ssid );
  Serial.print ( "IP address: " );
  Serial.println ( WiFi.localIP() );

  server.begin();
  
 
}

void loop ( void ) {
 char dummy[12];
  WiFiClient client = server.available();
  if (client) {
    while (client.available()) client.read();
    Serial.println("new client");
    //client.write(dummy,11); // don't write to leave the connection open.
    //client.stop();
  
  }
  delay(100); 

}

Debug output of the client is:

.....
Connected to MNET
IP address: 192.168.1.36
45976
connect 192.168.1.9 OK, sending: 0 1 0 0 0 6 126 4 157 7 0 1 
44896
connect 192.168.1.9 OK, sending: 0 1 0 0 0 6 126 4 157 7 0 1 
timeout while receiving or too many bytes. 0
timeout while receiving or too many bytes. 0
45768
connect 192.168.1.9 OK, sending: 0 1 0 0 0 6 126 4 157 7 0 1 
44736
connect 192.168.1.9 OK, sending: 0 1 0 0 0 6 126 4 157 7 0 1 
timeout while receiving or too many bytes. 0
timeout while receiving or too many bytes. 0
45656
connect 192.168.1.9 OK, sending: 0 1 0 0 0 6 126 4 157 7 0 1 
44624
connect 192.168.1.9 OK, sending: 0 1 0 0 0 6 126 4 157 7 0 1 
timeout while receiving or too many bytes. 0
timeout while receiving or too many bytes. 0
45544
connect 192.168.1.9 OK, sending: 0 1 0 0 0 6 126 4 157 7 0 1 
44512
connect 192.168.1.9 OK, sending: 0 1 0 0 0 6 126 4 157 7 0 1 
timeout while receiving or too many bytes. 0
timeout while receiving or too many bytes. 0
45432


... until it stalls.

 
connect 192.168.1.9 fail. 
2312
connect 192.168.1.9 fail. 
2288
connect 192.168.1.9 fail. 
2312
connect 192.168.1.9 fail. 
2288
connect 192.168.1.9 fail. 
2312
connect 192.168.1.9 fail. 

@debsahu I tried that and also client.disableKeepAlive() but with little success

[igrr]

Closing, will track in #4497. Please follow the link to that issue and click "subscribe" to receive notifications.