LoadProhibited crash on ESP32, referenced code from ESP32S2

[bertmelis]

Before filing a bug report, I'd like to check if it's a framework bug or not.

Guru Meditation Error: Core  1 panic'ed (LoadProhibited). Exception was unhandled.

Core  1 register dump:
PC      : 0x40090563  PS      : 0x00060a33  A0      : 0x80091438  A1      : 0x3ffb2050
A2      : 0x80e0b514  A3      : 0x40091470  A4      : 0x3ffb20b4  A5      : 0x7ff72964
A6      : 0xfffffffc  A7      : 0x00000001  A8      : 0x8009057d  A9      : 0x3ffb2030
A10     : 0x7ff72964  A11     : 0x00e98bb4  A12     : 0x00000000  A13     : 0x3ffb20b4  
A14     : 0x00060a20  A15     : 0x00000001  SAR     : 0x0000001f  EXCCAUSE: 0x0000001c
EXCVADDR: 0x80e0b518  LBEG    : 0x40089798  LEND    : 0x400897a3  LCOUNT  : 0x00000000


Backtrace: 0x40090560:0x3ffb2050 0x40091435:0x3ffb2070 0x4009167b:0x3ffb2090 0x4010fa7a:0x3ffb20b0 0x4010fac9:0x3ffb20f0 0x400e62f6:0x3ffb2130 0x400df5e1:0x3ffb2150 0x400df6e1:0x3ffb2180 0x400df0bb:0x3ffb21c0 0x400e068d:0x3ffb2230 0x400e0b01:0x3ffb2250 0x400d3e4a:0x3ffb2270 0x400e8619:0x3ffb2290

  #0  0x40090560:0x3ffb2050 in block_size at /Users/ficeto/Desktop/ESP32/ESP32S2/esp-idf-public/components/heap/heap_tlsf_block_functions.h:74 (discriminator 4)
      (inlined by) block_is_last at /Users/ficeto/Desktop/ESP32/ESP32S2/esp-idf-public/components/heap/heap_tlsf_block_functions.h:85 (discriminator 4)
      (inlined by) tlsf_walk_pool at /Users/ficeto/Desktop/ESP32/ESP32S2/esp-idf-public/components/heap/heap_tlsf.c:641 (discriminator 4)
  #1  0x40091435:0x3ffb2070 in multi_heap_get_info_impl at /Users/ficeto/Desktop/ESP32/ESP32S2/esp-idf-public/components/heap/multi_heap.c:413
  #2  0x4009167b:0x3ffb2090 in multi_heap_get_info at /Users/ficeto/Desktop/ESP32/ESP32S2/esp-idf-public/components/heap/multi_heap_poisoning.c:364
  #3  0x4010fa7a:0x3ffb20b0 in heap_caps_get_info at /Users/ficeto/Desktop/ESP32/ESP32S2/esp-idf-public/components/heap/heap_caps.c:535
  #4  0x4010fac9:0x3ffb20f0 in heap_caps_get_largest_free_block at /Users/ficeto/Desktop/ESP32/ESP32S2/esp-idf-public/components/heap/heap_caps.c:523
  #5  0x400e62f6:0x3ffb2130 in EspClass::getMaxAllocHeap() at C:/Users/bert_/.platformio/packages/framework-arduinoespressif32/cores/esp32/Esp.cpp:151

I call std::max(ESP.getMaxAllocHeap(), ESP.getMaxAllocPsram()) to find the max available memory on a ESP32. This crashes my application based on the trace pasted above. The strange this about the log is that although I build for ESP32, it apparently refrences to ESP32S2.

Is this some configuration error on my side? How can I check?

I'm building using latest platformio.

PS The reason I'm not sure this is a framework bug, it that it only fails when I enable debug logging. So there might be something in my code that smells.

[bertmelis]

Commenting out the call to ESP.getMaxAllocHeap() stops the crash from happening.

[me-no-dev]

I can not comment on why it is crashing, because that is IDF's internal API. I can say that it is not referencing ESP32S2, but rather the libs were build in a folder that had S2 in the path (my own computer and env), so you can disregard that. What version of Arduino are you using? 2.0.9?

[bertmelis]

2.0.9: I'm using platformio, it's latest esp32 platform pulls in 2.0.9.

[me-no-dev]

@igrr any ideas what could cause this exception?

[bertmelis]

Could it be that when I check the free heap and "at the same time" create a task, this issue presents?

The pattern in my program is:

• command via MQTT
• create and start a task
• confirm via MQTT (<-- here the free heap is checked)

Enabling debug logging obviously alters the timing and this may trigger the issue.

EDIT: check again, it happens much earlier in the program. I'm not even fully connected to MQTT. No extra tasks have been started.

[bertmelis]

Argh. found it. I compiled again but now with -Wshadow. Fixed my own mistake.
Case closed.

[diskgokey]

Hi Bert,

I also encounter this error, not immediately but eg. after 1 hour of work.
Can you shed some light what your problem fix was and maybe I can relate to this in my code?

because it's really strange that a ESP.getMaxAllocHeap() call makes it crash

[diskgokey]

    0x401ca6b4: block_size at /Users/ficeto/Desktop/ESP32/ESP32S2/esp-idf-public/components/heap/heap_tlsf_block_functions.h:74
    0x401ca6b4: block_is_last at /Users/ficeto/Desktop/ESP32/ESP32S2/esp-idf-public/components/heap/heap_tlsf_block_functions.h:85
    0x401ca6b4: tlsf_walk_pool at /Users/ficeto/Desktop/ESP32/ESP32S2/esp-idf-public/components/heap/heap_tlsf.c:641
    0x401caa34: multi_heap_get_info_impl at /Users/ficeto/Desktop/ESP32/ESP32S2/esp-idf-public/components/heap/multi_heap.c:400
    0x401caad9: multi_heap_get_info at /Users/ficeto/Desktop/ESP32/ESP32S2/esp-idf-public/components/heap/multi_heap_poisoning.c:371
    0x40158f9a: heap_caps_get_info at /Users/ficeto/Desktop/ESP32/ESP32S2/esp-idf-public/components/heap/heap_caps.c:556
    0x40158fe9: heap_caps_get_largest_free_block at /Users/ficeto/Desktop/ESP32/ESP32S2/esp-idf-public/components/heap/heap_caps.c:544
    0x401244ae: EspClass::getMaxAllocHeap() at /home/pionix/.platformio/packages/framework-arduinoespressif32/cores/esp32/Esp.cpp:151
    0x4010446b: AsyncMqttClient::publish(char const*, unsigned char, bool, char const*, unsigned int, bool, unsigned short) at /home/xxx/Documents/Github/xxx/.pio/libdeps/xxx/AsyncMqttClient/src/AsyncMqttClient.cpp:739