Random crash when receiving a continuous stream on socket.

[karawin]

My software runs pretty well but time to time this bug happens in a random time.
I don't have timers in it apart for the sockets.
The software receives a stream of mp3 radio.

(gdb) bt
#0 0x400896d6 in invoke_abort ()
at C:/msys32/home/jp/esp/esp-idf/components/esp32/panic.c:139
#1 0x40089a9c in _esp_error_check_failed (rc=259,
file=0x3f41f914 "C:/msys32/home/jp/esp/esp-idf/components/esp32/ets_timer_legacy.c", line=89, function=0x3f41f9e0 <func$5258> "ets_timer_arm",
expression=0x3f41f974 "esp_timer_start_once(ESP_TIMER(ptimer), time_us)")
at C:/msys32/home/jp/esp/esp-idf/components/esp32/panic.c:565
Backtrace stopped: previous frame identical to this frame (corrupt stack?)

Seems to come from:
#1 0x400dfbb5 in sta_input ()
No symbol table info available.
#2 0x400e02b4 in sta_rx_cb ()
No symbol table info available.
#3 0x400e833d in ppRxPkt ()
No symbol table info available.
#4 0x400e9bf2 in ppTask ()
No symbol table info available.

But cannot investigate further.

[kglowacki]

I have very similar problem, it happens after some time of network activity (a few hours):

ESP_ERROR_CHECK failed: esp_err_t 0x103 at 0x4012a9a9
ESC[0;33m0x4012a9a9: ets_timer_arm at /opt/esp/esp-idf/components/esp32/./ets_timer_legacy.c:88 (discriminator 1)
ESC[0m

Backtrace: 0x4008781c:0x3ffcb390 0x40087b99:0x3ffcb3b0 0x4012a9a9:0x3ffcb3d0 0x400e6589:0x3ffcb3f0 0x400e67ba:0x3ffcb420
ESC[0;33m0x4008781c: invoke_abort at /opt/esp/esp-idf/components/esp32/./panic.c:553
ESC[0m
ESC[0;33m0x40087b99: _esp_error_check_failed at /opt/esp/esp-idf/components/esp32/./panic.c:565
ESC[0m
ESC[0;33m0x4012a9a9: ets_timer_arm at /opt/esp/esp-idf/components/esp32/./ets_timer_legacy.c:88 (discriminator 1)
ESC[0m
ESC[0;33m0x400e6589: pm_rx_beacon_process at ??:?
ESC[0m
ESC[0;33m0x400e67ba: pm_rx_beacon_process at ??:?

I have plenty of heap free and none of tasks run out of stack.
same code works fine in v2.1

[karawin]

Seems to be a stack corruption. Closing

[karawin]

Reopened. It appears time to time.

[karawin]

Changed in ets_timer_legacy.c
void ets_timer_arm(ETSTimer *ptimer, uint32_t time_ms, bool repeat_flag) { uint64_t time_us = 1000LL * (uint64_t) time_ms; assert(timer_initialized(ptimer)); esp_err_t err; esp_timer_stop(ESP_TIMER(ptimer)); // no error check if (!repeat_flag) { ESP_ERROR_CHECK( esp_timer_start_once(ESP_TIMER(ptimer), time_us) ); } else { ESP_ERROR_CHECK( esp_timer_start_periodic(ESP_TIMER(ptimer), time_us) ); } }
to
void ets_timer_arm(ETSTimer *ptimer, uint32_t time_ms, bool repeat_flag) { uint64_t time_us = 1000LL * (uint64_t) time_ms; assert(timer_initialized(ptimer)); esp_err_t err; esp_timer_stop(ESP_TIMER(ptimer)); // no error check if (!repeat_flag) { // ESP_ERROR_CHECK( esp_timer_start_once(ESP_TIMER(ptimer), time_us) ); err = esp_timer_start_once(ESP_TIMER(ptimer), time_us) ; if (err != ESP_OK) { printf("err: %d, timer: %x, time_us:%Ld, time_ms:%d\n",err,(int)ptimer,time_us, time_ms); abort(); } } else { ESP_ERROR_CHECK( esp_timer_start_periodic(ESP_TIMER(ptimer), time_us) ); } }
After a random time of running the application i see with gdb:
#0 0x400898e2 in invoke_abort ()
at C:/msys32/home/jp/esp/esp-idf/components/esp32/panic.c:139
#1 0x400899ae in abort ()
at C:/msys32/home/jp/esp/esp-idf/components/esp32/panic.c:148
#2 0x4012de01 in ets_timer_arm (ptimer=0x3ffb4f48, time_ms=240,
repeat_flag=false)
at C:/msys32/home/jp/esp/esp-idf/components/esp32/ets_timer_legacy.c:94
#3 0x400e76b0 in pm_rx_beacon_process ()
#4 0x400e78e2 in ?? ()

With the log:
err: 259, timer: 3ffb4f48, time_us:0, time_ms:240000
abort() was called at PC 0x4012ddfe on core 1

**That confirm the diagnostic of kglowacki.**

Something has broken the stack because
#2 0x4012de01 in ets_timer_arm (ptimer=0x3ffb4f48, time_ms=240,
repeat_flag=false)
is ok.
Error 253 = 0x103 ESP_ERR_INVALID_STATE

esp_err_t esp_timer_start_once(esp_timer_handle_t timer, uint64_t timeout_us)
{
if (!is_initialized() || timer_armed(timer)) {
return ESP_ERR_INVALID_STATE;
}
timer->alarm = esp_timer_get_time() + timeout_us;
timer->period = 0;
#if WITH_PROFILING
timer->times_armed++;
#endif
return timer_insert(timer);
}

[karawin]

Seen to be not reproductible with the last idf. Closing