ModbusClient TCP how to stop a running work task
Miq1 opened this issue · comments
Discussed in #277
Originally posted by dima-sz March 10, 2023
ModbusClient TCP how to stop a running worker task started by calling begin(int coreID)?
I added ModbusClientTCP::end() to the master branch. I would ask you to test it and report your findings here.
Thanks, it works.
вс, 26 мар. 2023 г. в 22:51, Michael Harwerth ***@***.***>:
… I added ModbusClientTCP::end() to the master branch. I would ask you to
test it and report your findings here.
—
Reply to this email directly, view it on GitHub
<#278 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AREL47MRGEONEJ4IYUNU2MTW6CF3NANCNFSM6AAAAAAVV7OZUY>
.
You are receiving this because you are subscribed to this thread.Message
ID: ***@***.***>
But if you call the end() command when it is not yet executing a work task,
it will cause a crash.
E (76818) task_wdt: Task watchdog got triggered. The following tasks did
not reset the watchdog in time:
E (76818) task_wdt: - async_tcp (CPU 1)
E (76818) task_wdt: Tasks currently running:
E (76818) task_wdt: CPU 0: IDLE
E (76818) task_wdt: CPU 1: loopTask
E (76818) task_wdt: Aborting.
abort() was called at PC 0x4013761c on core 0
Backtrace: 0x40083ae1:0x3ffbfe9c |<-CORRUPTED
пн, 27 мар. 2023 г. в 12:15, Дмитрий Захаренко ***@***.***>:
… Thanks, it works.
вс, 26 мар. 2023 г. в 22:51, Michael Harwerth ***@***.***>:
> I added ModbusClientTCP::end() to the master branch. I would ask you to
> test it and report your findings here.
>
> —
> Reply to this email directly, view it on GitHub
> <#278 (comment)>,
> or unsubscribe
> <https://github.com/notifications/unsubscribe-auth/AREL47MRGEONEJ4IYUNU2MTW6CF3NANCNFSM6AAAAAAVV7OZUY>
> .
> You are receiving this because you are subscribed to this thread.Message
> ID: ***@***.***>
>
Argh. Missed an if (worker) { ... } there. Will fix that asap.
Thank you for testing it!
and also, if you start an already running process, it also crashes.
пн, 27 мар. 2023 г. в 12:57, Michael Harwerth ***@***.***>:
… Argh. Missed an if (worker) { ... } there. Will fix that asap.
Thank yiu for testing it!
—
Reply to this email directly, view it on GitHub
<#278 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AREL47MWLS6737D7ABIS3EDW6FCANANCNFSM6AAAAAAVV7OZUY>
.
You are receiving this because you commented.Message ID:
***@***.***>
Thanks again - I fixed it a minute ago in master. Could you give it another try?
Thank you very much, everything works great now!
пн, 27 мар. 2023 г. в 13:45, Michael Harwerth ***@***.***>:
… Thanks again - I fixed it a minute ago in master. Could you give it
another try?
—
Reply to this email directly, view it on GitHub
<#278 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AREL47NQ2QVO3STKDEBDMZDW6FHTRANCNFSM6AAAAAAVV7OZUY>
.
You are receiving this because you commented.Message ID:
***@***.***>
Thanks for verifying!
There was another problem if you create two clients MB0 and MB1 and poll
devices that are not online (offline)
MB0.setTarget(IP-1, 502);
MB0.start();
MB1.setTarget(IP-2, 502);
MB1.start();
it will crash.
Guru Meditation Error: Core 0 panic'ed (LoadProhibited). Exception was
unhandled.
пн, 27 мар. 2023 г. в 15:18, Michael Harwerth ***@***.***>:
… Thanks for verifying!
—
Reply to this email directly, view it on GitHub
<#278 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AREL47IFIOUSOC3Z62IF2A3W6FSQ3ANCNFSM6AAAAAAVV7OZUY>
.
You are receiving this because you commented.Message ID:
***@***.***>
Sorry
MB0.setTarget(IP-1, 502);
MB0.begin();
MB1.setTarget(IP-2, 502);
MB1.begin();
пн, 27 мар. 2023 г. в 16:00, Дмитрий Захаренко ***@***.***>:
… There was another problem if you create two clients MB0 and MB1 and poll
devices that are not online (offline)
MB0.setTarget(IP-1, 502);
MB0.start();
MB1.setTarget(IP-2, 502);
MB1.start();
it will crash.
Guru Meditation Error: Core 0 panic'ed (LoadProhibited). Exception was
unhandled.
пн, 27 мар. 2023 г. в 15:18, Michael Harwerth ***@***.***>:
> Thanks for verifying!
>
> —
> Reply to this email directly, view it on GitHub
> <#278 (comment)>,
> or unsubscribe
> <https://github.com/notifications/unsubscribe-auth/AREL47IFIOUSOC3Z62IF2A3W6FSQ3ANCNFSM6AAAAAAVV7OZUY>
> .
> You are receiving this because you commented.Message ID:
> ***@***.***>
>
Can we have a full exception dump please?
The error hints f.i. to a use of a single memory area for both clients, where the second tries to access while the first already has freed it. This may happen more often because a connection is tried for 3s before determining it is impossible for offline servers. So the clients will stay in that state without doing anything else.
Do you use the same onData/onError/onResponse handlers for both clients? is there some dynamic data management involved in those functions?
I want to say right away that this problem was not in the version before
the addition of the end() function, it was tested on 5-6 clients.
I use one onResponse handler.
exception dump is different, here are some options
assert failed: lwip_unlink_select_cb
IDF/components/lwip/lwip/src/api/sockets.c:1865 (select_cb->prev == NULL)
Backtrace: 0x40083a49:0x3ffd9140 0x4008bcfd:0x3ffd9160
0x40091919:0x3ffd9180 0x400fd8ef:0x3ffd92b0 0x400ff28b:0x3ffd92d0
0x400f5645:0x3ffd9370 0x400d9cc3:0x3ffd93e0 0x400d956e:0x3ffd9430
0x400e49ab:0x3ffd9460
==================================================
Guru Meditation Error: Core 0 panic'ed (LoadProhibited). Exception was
unhandled.
Core 0 register dump:
PC : 0x400fd7c9 PS : 0x00060330 A0 : 0x80111a88 A1 :
0x3ffb2f60
A2 : 0x06141150 A3 : 0x0000001d A4 : 0x00000030 A5 :
0x00000001
A6 : 0x00010000 A7 : 0x3ffc61a4 A8 : 0x3ffc61a0 A9 :
0x0000001d
A10 : 0x00000001 A11 : 0x00000001 A12 : 0x00000001 A13 :
0x00000001
A14 : 0x00060b20 A15 : 0x00000001 SAR : 0x00000010 EXCCAUSE:
0x0000001c
EXCVADDR: 0x0614116c LBEG : 0x40089660 LEND : 0x40089676 LCOUNT :
0xffffffff
Backtrace: 0x400fd7c6:0x3ffb2f60 0x40111a85:0x3ffb2f90
0x40103b61:0x3ffb2fb0 0x40103c6f:0x3ffb2ff0 0x40107053:0x3ffb3010
0x40107111:0x3ffb3030 0x400ff6f5:0x3ffb3050
================================================================
assert failed: lwip_unlink_select_cb
IDF/components/lwip/lwip/src/api/sockets.c:1865 (select_cb->prev == NULL)
Backtrace: 0x40083a49:0x3ffd9580 0x4008bcfd:0x3ffd95a0
0x40091919:0x3ffd95c0 0x400fd8ef:0x3ffd96f0 0x400ff28b:0x3ffd9710
0x400f5645:0x3ffd97b0 0x400d9cc3:0x3ffd9820 0x400d956e:0x3ffd9870
0x400e49ab:0x3ffd98a0
===========================================================
пн, 27 мар. 2023 г. в 18:00, Michael Harwerth ***@***.***>:
… Do you use the same onData/onError/onResponse handlers for both clients?
is there some dynamic data management involved in those functions?
—
Reply to this email directly, view it on GitHub
<#278 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AREL47KOUEJPCNT6OL2JZRTW6GFQNANCNFSM6AAAAAAVV7OZUY>
.
You are receiving this because you commented.Message ID:
***@***.***>
Do you happen to issue the end() while the connection is still hanging for the offline server? There have been reports that killing a TCP client while the polling is still going on results in a dangling pointer in lwIP, hence crashing on the next access to it (see here for an example).
Very strange, I wanted to reproduce this problem with a simple example,
without my main code, but I did not succeed, this error does not occur.
Perhaps in my code there is incompatibility of different libraries, I use
#include <Arduino.h>
#include <SimpleKalmanFilter.h>
#include "fs.h"
#include <LittleFS.h>
#include <wifi.h>
#include <AsyncTCP.h>
#include <ESPAsyncWebServer.h>
#include <WebAuthentication.h>
#include <DNSServer.h>
#include <AsyncElegantOTA.h>
#include "eprom.h"
#include <OneWire.h>
#include <DallasTemperaturemy.h>
#include "soc/soc.h"
#include "soc/rtc_cntl_reg.h"
#include "Button2.h"
#include "ModbusClientTCP.h"
#include "CoilData.h"
but, as I said earlier, this was not there before the changes were made,
now I have checked it again.
пн, 27 мар. 2023 г. в 18:51, Michael Harwerth ***@***.***>:
… Do you happen to issue the end() while the connection is still hanging
for the offline server? There have been reports that killing a TCP client
while the polling is still going on results in a dangling pointer in lwIP,
hence crashing on the next access to it (see here
<espressif/esp-idf#6205 (comment)>
for an example).
—
Reply to this email directly, view it on GitHub
<#278 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AREL47JFGDB3M6LUIUERVWTW6GLN7ANCNFSM6AAAAAAVV7OZUY>
.
You are receiving this because you commented.Message ID:
***@***.***>
You are using a heck of different TCP services here, both async and synchronous - may be there is a collision sometimes. I have no real idea what could cause the issue...
Any news? Else I would like to close here.
Unfortunately, I'm busy with other important things right now and can't
continue checking for now, so close this one for now.
ср, 29 мар. 2023 г. в 18:13, Michael Harwerth ***@***.***>:
… Any news? Else I would like to close here.
—
Reply to this email directly, view it on GitHub
<#278 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AREL47N7OWAWTKANGKVSKQDW6QYQ7ANCNFSM6AAAAAAVV7OZUY>
.
You are receiving this because you commented.Message ID:
***@***.***>
No worries, I will leave it open then. Whenever you find the time I will be eager to learn the outcome.
I am assuming nothing more will be posted here.