OTA升级失败的原因？

硬件:esp32 dwod

idf v4.4

参照例子:native_ota_example写的ota程序，升级完成后，重启进入:

    const esp_partition_t *running = esp_ota_get_running_partition();

    esp_ota_img_states_t ota_state;

    if (esp_ota_get_state_partition(running, &ota_state) == ESP_OK) {

        if (ota_state == ESP_OTA_IMG_PENDING_VERIFY) {

            // run diagnostic function ...

            bool diagnostic_is_ok = diagnostic();

            if (diagnostic_is_ok) {

                ESP_LOGI(TAG, "Diagnostics completed successfully! Continuing execution ...");

                esp_ota_mark_app_valid_cancel_rollback();

            } else {

                ESP_LOGE(TAG, "Diagnostics failed! Start rollback to the previous version ...");

                esp_ota_mark_app_invalid_rollback_and_reboot();

            }

        }

    }

校验成功，执行：esp_ota_mark_app_valid_cancel_rollback() 时，esp32直接重启：

I (1302) ESP32: SHA-256 for the partition table: : fbd17bd5dfe30a5b41cf748df6b67972d33699e18ba40e10005d93b46fa471bb

I (1304) ESP32: SHA-256 for bootloader: : 153f40bd798f490dc753ef5ddbe383adb144b40ea7af2af5cc96c413327099c3

I (1314) ESP32: Running partion:ota_0

I (1320) ESP32: SHA-256 for current firmware: : f2bfa8da66d1ddc01adfc6cd55194908894d4c6ead21fd598fa1e313e738d340

I (1339) ESP32: Running partition is verfying ...

I (1340) ESP32: Diagnostics completed successfully! Continuing execution ...

ets Jul 29 2019 12:21:46



rst:0x1 (POWERON_RESET),boot:0x13 (SPI_FAST_FLASH_BOOT)

configsip: 0, SPIWP:0xee

clk_drv:0x00,q_drv:0x00,d_drv:0x00,cs0_drv:0x00,hd_drv:0x00,wp_drv:0x00

mode:DIO, clock div:2

load:0x3fff0030,len:6612

load:0x40078000,len:14880

ho 0 tail 12 room 4

load:0x40080400,len:3792

entry 0x40080694

I (29) boot: ESP-IDF v4.4-dirty 2nd stage bootloader

I (29) boot: compile time 09:33:57

I (29) boot: chip revision: 3

I (32) boot_comm: chip revision: 3, min. bootloader chip revision: 0

I (39) boot.esp32: SPI Speed      : 40MHz

I (44) boot.esp32: SPI Mode       : DIO

I (49) boot.esp32: SPI Flash Size : 8MB

I (53) boot: Enabling RNG early entropy source...

然后继续使用升级前的分区，升级失败。

什么原因导致毫无征兆的重启呢?