Bad restart with esp_restart()

[kostyan5]

When calling esp_restart() most of the time CPU reboots with SW_CPU_RESET. However, sometimes before booting up it restarts an extra time with RTCWDT_RTC_RESET. It doesn't happen every time, maybe 1 out of 5. Log below shows 1 "good" restart followed by
a "bad" restart. This presents a problem because RTCWDT_RTC_RESET resets real time clock. I'm using ESP32 Thing from sparkfun.
The code used to test this is simple:

Code: Select all

void app_main(void)
{
    esp_restart();
}

Code: Select all

rst:0xc (SW_CPU_RESET),boot:0x13 (SPI_FAST_FLASH_BOOT)
configsip: 0, SPIWP:0x00
clk_drv:0x00,q_drv:0x00,d_drv:0x00,cs0_drv:0x00,hd_drv:0x00,wp_drv:0x00
mode:DIO, clock div:2
load:0x3fff0008,len:8
load:0x3fff0010,len:3448
load:0x40078000,len:10388
load:0x40080000,len:252
0x40080000: _iram_start at ??:?

entry 0x40080034
0x40080034: _iram_start at ??:?

I (550) cpu_start: Pro cpu up.
I (551) cpu_start: Single core mode
I (552) heap_alloc_caps: Initializing. RAM available for dynamic allocation:
I (564) heap_alloc_caps: At 3FFAFF10 len 000000F0 (0 KiB): DRAM
I (584) heap_alloc_caps: At 3FFC5880 len 0001A780 (105 KiB): DRAM
I (605) heap_alloc_caps: At 3FFE0440 len 00003BC0 (14 KiB): D/IRAM
I (626) heap_alloc_caps: At 3FFE4350 len 0001BCB0 (111 KiB): D/IRAM
I (647) heap_alloc_caps: At 400929D8 len 0000D628 (53 KiB): IRAM
I (668) cpu_start: Pro cpu start user code
ets Jun  8 2016 00:22:57

rst:0xc (SW_CPU_RESET),boot:0x13 (SPI_FAST_FLASH_BOOT)
configsip: 0, SPIWP:0x00
clk_drv:0x00,q_drv:0x00,d_drv:0x00,cs0_drv:0x00,hd_drv:0x00,wp_drv:0x00
mode:DIO, clock div:2
load:0x3fff0008,len:8
load:0x3fff0010,len:3448
load:0x40078000,len:10388
load:0x40080000,len:252
0x40080000: _iram_start at ??:?

entry 0x40080034
0x40080034: _iram_start at ??:?

ets Jun  8 2016 00:22:57

rst:0x10 (RTCWDT_RTC_RESET),boot:0x13 (SPI_FAST_FLASH_BOOT)
configsip: 0, SPIWP:0x00
clk_drv:0x00,q_drv:0x00,d_drv:0x00,cs0_drv:0x00,hd_drv:0x00,wp_drv:0x00
mode:DIO, clock div:2
load:0x3fff0008,len:8
load:0x3fff0010,len:3448
load:0x40078000,len:10388
load:0x40080000,len:252
0x40080000: _iram_start at ??:?

entry 0x40080034
0x40080034: _iram_start at ??:?

I (550) cpu_start: Pro cpu up.
I (551) cpu_start: Single core mode
I (552) heap_alloc_caps: Initializing. RAM available for dynamic allocation:
I (564) heap_alloc_caps: At 3FFAFF10 len 000000F0 (0 KiB): DRAM
I (584) heap_alloc_caps: At 3FFC5880 len 0001A780 (105 KiB): DRAM
I (605) heap_alloc_caps: At 3FFE0440 len 00003BC0 (14 KiB): D/IRAM
I (626) heap_alloc_caps: At 3FFE4350 len 0001BCB0 (111 KiB): D/IRAM
I (647) heap_alloc_caps: At 400929D8 len 0000D628 (53 KiB): IRAM
I (668) cpu_start: Pro cpu start user code

[martinayotte]

I think it is related to the v0 silicon bug that Igrr is describing in this comment.
He said that it should be fixed in v1 silicon ...
https://github.com/espressif/esp-idf/is ... -291916959

[kostyan5]

Thanks! That sounds like my problem. When I source these parts, how do I make sure I get v1? What's the specific part #?

[WiFive]

I think it is related to the v0 silicon bug that Igrr is describing in this comment.
He said that it should be fixed in v1 silicon ...
https://github.com/espressif/esp-idf/is ... -291916959

No I dont think this should happen with a software reset.

Possibly increase the timeout here: https://github.com/espressif/esp-idf/bl ... api.c#L250

Or reset by deep sleep wakeup.

[ESP_igrr]

This doesn't look like the ECO bug related thing, as the startup process goes all the way to the 2nd bootloader entry, according to the log. Most likely we are having an issue switching clock to PLL. Curious thing however is that i can't reproduce this on the test board I have here. Will try a few other boards on Monday.