search

espressif / qemu

Fork of QEMU with Espressif patches. See Wiki for details.
https://github.com/espressif/esp-toolchain-docs/blob/main/qemu/README.md
Other
230 stars 60 forks source link

ESP32 Core 1 panic when going into deep sleep (QEMU-223) #105

Open InRiPa opened 4 days ago

InRiPa commented 4 days ago

Checklist

How often does this bug occurs?

always

Expected behavior

Getting into deep sleep with a timer setup previously.

Actual behavior (suspected bug)

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

Error logs or terminal output

I (12415) boot: ESP-IDF v5.3-357-g6568f8c553 2nd stage bootloader
I (12416) boot: compile time Nov 11 2024 15:43:57
I (12416) boot: Multicore bootloader
I (12435) boot: chip revision: v0.0
I (12438) boot.esp32: SPI Speed      : 40MHz
I (12439) boot.esp32: SPI Mode       : DIO
I (12439) boot.esp32: SPI Flash Size : 2MB
I (12450) boot: Enabling RNG early entropy source...
I (12462) boot: Partition Table:
I (12462) boot: ## Label            Usage          Type ST Offset   Length
I (12463) boot:  0 nvs              WiFi data        01 02 00009000 00006000
I (12464) boot:  1 phy_init         RF data          01 01 0000f000 00001000
I (12464) boot:  2 factory          factory app      00 00 00010000 00100000
I (12470) boot: End of partition table
I (12477) esp_image: segment 0: paddr=00010020 vaddr=3f400020 size=0a584h ( 42372) map
I (12507) esp_image: segment 1: paddr=0001a5ac vaddr=3ff80063 size=00008h (     8) load
I (12526) esp_image: segment 2: paddr=0001a5bc vaddr=3ffb0000 size=023e4h (  9188) load
I (12549) esp_image: segment 3: paddr=0001c9a8 vaddr=40080000 size=03670h ( 13936) load
I (12578) esp_image: segment 4: paddr=00020020 vaddr=400d0020 size=1506ch ( 86124) map
I (12618) esp_image: segment 5: paddr=00035094 vaddr=40083670 size=0a494h ( 42132) load
I (12661) esp_image: segment 6: paddr=0003f530 vaddr=400c0000 size=00064h (   100) load
I (12679) esp_image: segment 7: paddr=0003f59c vaddr=50000000 size=00004h (     4) load
I (12700) boot: Loaded app from partition at offset 0x10000
I (12701) boot: Disabling RNG early entropy source...
I (12723) cpu_start: Multicore app
I (17231) cpu_start: Pro cpu start user code
I (17232) cpu_start: cpu freq: 160000000 Hz
I (17232) app_init: Application information:
I (17232) app_init: Project name:     esp32-idf-mastering-rtos
I (17232) app_init: App version:      a1047cf-dirty
I (17233) app_init: Compile time:     Nov 11 2024 15:43:48
I (17233) app_init: ELF file SHA256:  165488e867d81688...
I (17233) app_init: ESP-IDF:          v5.3-357-g6568f8c553
I (17234) efuse_init: Min chip rev:     v0.0
I (17234) efuse_init: Max chip rev:     v3.99 
I (17234) efuse_init: Chip rev:         v0.0
I (17235) heap_init: Initializing. RAM available for dynamic allocation:
I (17237) heap_init: At 3FFAE6E0 len 00001920 (6 KiB): DRAM
I (17237) heap_init: At 3FFB2CD8 len 0002D328 (180 KiB): DRAM
I (17237) heap_init: At 3FFE0440 len 00003AE0 (14 KiB): D/IRAM
I (17238) heap_init: At 3FFE4350 len 0001BCB0 (111 KiB): D/IRAM
I (17238) heap_init: At 4008DB04 len 000124FC (73 KiB): IRAM
I (17282) spi_flash: detected chip: gd
I (17289) spi_flash: flash io: dio
W (17300) spi_flash: Detected size(4096k) larger than the size in the binary image header(2048k). Using the size in the binary image header.
I (17309) main_task: Started on CPU0
I (17319) main_task: Calling app_main()
I (17319) gpio: GPIO[0]| InputEn: 0| OutputEn: 0| OpenDrain: 0| Pullup: 1| Pulldown: 0| Intr:0 
I (17319) gpio: GPIO[2]| InputEn: 0| OutputEn: 0| OpenDrain: 0| Pullup: 1| Pulldown: 0| Intr:0 
I (17319) gpio: GPIO[15]| InputEn: 0| OutputEn: 0| OpenDrain: 0| Pullup: 1| Pulldown: 0| Intr:0 
I (17319) gpio: GPIO[35]| InputEn: 0| OutputEn: 0| OpenDrain: 0| Pullup: 1| Pulldown: 0| Intr:0 
I (17319) gpio: GPIO[26]| InputEn: 0| OutputEn: 0| OpenDrain: 0| Pullup: 1| Pulldown: 0| Intr:0 
I (17319) gpio: GPIO[25]| InputEn: 0| OutputEn: 0| OpenDrain: 0| Pullup: 1| Pulldown: 0| Intr:0 
I (17319) gpio: GPIO[33]| InputEn: 0| OutputEn: 0| OpenDrain: 0| Pullup: 1| Pulldown: 0| Intr:0 
W (17329) SLEEP_DEEPTIMER: Wakeup was not caused by deep sleep: 0

I (17329) Main: I got called by IDF initially and now I'm done here!
I (17329) main_task: Returned from app_main()
I (20329) SLEEP_DEEPTIMER: Tick count before sleep: 302
I (20329) SLEEP_DEEPTIMER: Going to sleep for 30 Seconds
Guru Meditation Error: Core  1 panic'ed (LoadStorePIFAddrError). Exception was unhandled.

Core  1 register dump:
PC      : 0x400847ba  PS      : 0x00060233  A0      : 0x80084978  A1      : 0x3ffb5360  
A2      : 0x000001e3  A3      : 0x00000001  A4      : 0x00000000  A5      : 0x01c9c1f0  
A6      : 0x00000003  A7      : 0x00000000  A8      : 0x3ff80064  A9      : 0x3ffb5340  
A10     : 0x3ffb5360  A11     : 0x0000000a  A12     : 0x00029fff  A13     : 0x00000000  
A14     : 0x00000002  A15     : 0x00000001  SAR     : 0x00000009  EXCCAUSE: 0x0000000f  
EXCVADDR: 0x3ff80064  LBEG    : 0x400014fd  LEND    : 0x4000150d  LCOUNT  : 0xfffffffc  

Backtrace: 0x400847b7:0x3ffb5360 0x40084975:0x3ffb53b0 0x400849b5:0x3ffb53d0 0x400d5fac:0x3ffb53f0 0x40086ee1:0x3ffb5410

ELF file SHA256: 165488e867d81688

Rebooting...

-------------------------
-------------------------
$ ./print_trace.sh 0x400847b7:0x3ffb5360 0x40084975:0x3ffb53b0 0x400849b5:0x3ffb53d0 0x400d5fac:0x3ffb53f0 0x40086ee1:0x3ffb5410
0x400847b7: misc_modules_sleep_prepare at /opt/esp/idf/components/esp_hw_support/sleep_modes.c:670 (discriminator 1)
 (inlined by) esp_sleep_start at /opt/esp/idf/components/esp_hw_support/sleep_modes.c:846 (discriminator 1)
0x40084975: deep_sleep_start at /opt/esp/idf/components/esp_hw_support/sleep_modes.c:1163
0x400849b5: esp_deep_sleep_start at /opt/esp/idf/components/esp_hw_support/sleep_modes.c:1187
0x400d5fac: vDeepSleeperTask at /home/esp/projects/esp32-idf-mastering-rtos/components/rtostasks/task_sleep_deeptimer.c:34
0x40086ee1: vPortTaskWrapper at /opt/esp/idf/components/freertos/FreeRTOS-Kernel/portable/xtensa/port.c:134

Steps to reproduce the behavior

static const char *TAG = "SLEEP_DEEPTIMER";
#define uS_TO_S_FACTOR 1000000 /* Conversion factor for micro seconds to seconds */
#define TIME_TO_SLEEP 30       /* Time ESP32 will go to sleep (in seconds) */
RTC_DATA_ATTR int bootCount = 0;

void vDeepSleeperTask(void *pvParameters)
{
    TickType_t xTickCountBefore = xTaskGetTickCount();

    for (;;)
    {
        vTaskDelay(pdMS_TO_TICKS(3000));
        // Get and print current tick count
        xTickCountBefore = xTaskGetTickCount();
        ESP_LOGI(TAG, "Tick count before sleep: %lu", xTickCountBefore);

        // Set up the TIMER as a wake-up source
        esp_sleep_enable_timer_wakeup(TIME_TO_SLEEP * uS_TO_S_FACTOR);
        ESP_LOGI(TAG, "Going to sleep for %d Seconds", TIME_TO_SLEEP);

        /* Enter the low power state. */
        esp_deep_sleep_start();

        ESP_LOGE(TAG, "Should never get here!");
    }
}
int app_main(void)
{
     print_wakeup_reason();

    xTaskCreatePinnedToCore(vDeepSleeperTask, "sleeper", 2048, NULL, 3, NULL, portNUM_PROCESSORS - 1);

    ESP_LOGI(TAG, "I got called by IDF initially and now I'm done here!");
    /* Will not reach here. */
    return 0;
}

Project release version

ESP-IDF v5.3-357-g6568f8c553

System architecture

Intel/AMD 64-bit (modern PC, older Mac)

Operating system

Linux

Operating system version

FROM espressif/idf:release-v5.3

Shell

sh

Additional context

I'm not sure, maybe I'm using the deep sleep wrongly, but I also checked some online examples, and the setup seemed pretty much the same. Following the minimal example for a classic esp32.

  1. I checked existing and closed tickets for esp_deep_sleep_start.
  2. I checked the wiki for potential issues to respect in regards to deep sleep.
  3. Checked the version: /opt/esp/tools/qemu-xtensa/esp_develop_9.0.0_20240606/qemu/bin/qemu-system-xtensa
igrr commented 1 day ago

Hi @InRiPa, We haven't implemented any sleep-related functionality in QEMU yet. The exception (LoadStorePIFAddrError) happens because the sleep driver in IDF is trying to write to a peripheral which isn't defined in the emulator.

Since from software perspective a deep sleep is almost like a software reset (just with a variable amount of time until the next boot), I would recommend replacing the deep sleep entry with a call to esp_restart, when building for QEMU.