ESP32-LyraTD-MSC不能跑 ESP-ADF wwe 参考应用吗？一直死机。

[hyuan75]

## Environment

Audio development kit: ESP32-LyraTD-MSC
Audio kit version: v2.2
[Required] Module or chip used: ESP32-WROVER-E
[Required] IDF version: v5.2.1
[Required] ADF version: v2.6-108-g1f275b50
Build system: visual code/idf.py
[Required] Running log: All logs from power-on to problem recurrence
Compiler version : xtensa-esp-elf-gcc.exe (crosstool-NG esp-13.2.0_20230928) 13.2.0
Operating system: Windows
(Windows only) Environment type: PowerShell
Using an IDE?: Yes, Visual Code+ IDF extension
Power supply: USB

## Problem Description

ESP32-LyraTD-MSC v2.2板子不能跑 ESP-ADF wwe example, CPU会crash/死机。
当 ‘I (12132) main_task: Returned from app_main()“ log 打印后, 任何声音输入系统都不会有反应, 几十秒后, 系统就会死机

### Expected Behavior
按照官方说明，这个板子应该能正常跑这个应用.

### Actual Behavior
Crash.

### Steps to Reproduce

1. 按照https://github.com/espressif/esp-adf/tree/master/examples readme 的板子/参考应用对照表, wwe example(examples/speech_recognition/wwe)应该在 ESP32-LyraTD-MSC板子上能跑起来. 也试了另外一个参考应用VAD (examples/speech_recognition/vad) 在同一块板子上能成功跑起来.
2. In VSCode, 从wwe example创建一个应用，没改任何代码。
3. 根据ADF文档（https://docs.espressif.com/projects/esp ... tions.html），修改一些 memory/system 如下：
a. Enabled SPI PSRAM. The option CONFIG_SPIRAM_CACHE_WORKAROUND, set by default in the same menu, kept enabled.
b. Not enable Bluetooth.
c. CONFIG_WIFI_LWIP_ALLOCATION_FROM_SPIRAM_FIRST set as “yes”
d. Set CONFIG_WL_SECTOR_SIZE as 512 in Component config > Wear Levelling
e. Flash SPI mode as QIO
f. Flash SPI speed as 80MHz
g. CPU frequency as 240MHz
h. Set Default receive window size as 5 times(7200) greater than Maximum Segment Size(1440) in Component config > LWIP > TCP
i. Set Flash Size as 4MB
4.根据wwe参考应用说明中 "Troubleshooting" 一节（https://github.com/espressif/esp-adf/tr ... nition/wwe), 修改esp\esp-adf\components\audio_hal\driver\zl38063.c ( tw_upload_dsp_firmware(0) ===> tw_upload_dsp_firmware(-1); ）进行强制dsp firmware下载。
5. 编译并运行应该，发现会死机而且log里有警告： "W (10615) RECORDER_SR: Multinet is not enabled in SDKCONFIG" , 于是在sdkconfig设置里:
. enable "use multinet"
. Chinese Speech Commands Model: Chinese Single Recognition(mn2_cn)
. Select wake words: nihaoxiaozhi(wn5_nihaoxiaozi)
6. 重新编译并运行，还是会死机:

rst:0x1 (POWERON_RESET),boot:0x1f (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:1
load:0x3fff0030,len:7524
load:0x40078000,len:16528
ho 0 tail 12 room 4
load:0x40080400,len:4
0x40080400: _init at ??:?

load:0x40080404,len:4312
entry 0x4008065c
I (31) boot: ESP-IDF v5.2.1-dirty 2nd stage bootloader
I (31) boot: compile time May 7 2024 23:55:14
I (31) boot: Multicore bootloader
I (35) boot: chip revision: v1.0
I (39) qio_mode: Enabling default flash chip QIO
I (44) boot.esp32: SPI Speed : 80MHz
I (49) boot.esp32: SPI Mode : QIO
I (54) boot.esp32: SPI Flash Size : 4MB
I (58) boot: Enabling RNG early entropy source...
I (64) boot: Partition Table:
I (67) boot: ## Label Usage Type ST Offset Length
I (74) boot: 0 nvs WiFi data 01 02 00009000 00004000
I (82) boot: 1 otadata OTA data 01 00 0000d000 00002000
I (89) boot: 2 phy_init RF data 01 01 0000f000 00001000
I (97) boot: 3 ota_0 OTA app 00 10 00010000 00300000
I (104) boot: 4 flash_tone Unknown data 01 27 00310000 00032000
I (112) boot: End of partition table
I (116) boot: No factory image, trying OTA 0
I (121) esp_image: segment 0: paddr=00010020 vaddr=3f400020 size=193ea0h (1654432) map
I (554) esp_image: segment 1: paddr=001a3ec8 vaddr=3ffb0000 size=0285ch ( 10332) load
I (557) esp_image: segment 2: paddr=001a672c vaddr=40080000 size=098ech ( 39148) load
I (571) esp_image: segment 3: paddr=001b0020 vaddr=400d0020 size=50828h (329768) map
I (656) esp_image: segment 4: paddr=00200850 vaddr=400898ec size=0a0a4h ( 41124) load
I (679) boot: Loaded app from partition at offset 0x10000
I (717) boot: Set actual ota_seq=1 in otadata[0]
I (718) boot: Disabling RNG early entropy source...
I (728) cpu_start: Multicore app
I (728) quad_psram: This chip is ESP32-D0WD
I (729) esp_psram: Found 8MB PSRAM device
I (731) esp_psram: Speed: 80MHz
I (734) esp_psram: PSRAM initialized, cache is in low/high (2-core) mode.
W (742) esp_psram: Virtual address not enough for PSRAM, map as much as we can. 4MB is mapped
I (1214) esp_psram: SPI SRAM memory test OK
I (1222) cpu_start: Pro cpu start user code
I (1222) cpu_start: cpu freq: 240000000 Hz
I (1222) cpu_start: Application information:
I (1226) cpu_start: Project name: example_wwe
I (1231) cpu_start: App version: 1
I (1236) cpu_start: Compile time: May 8 2024 10:09:58
I (1242) cpu_start: ELF file SHA256: b05478608...
I (1247) cpu_start: ESP-IDF: v5.2.1-dirty
I (1253) cpu_start: Min chip rev: v0.0
I (1257) cpu_start: Max chip rev: v3.99
I (1262) cpu_start: Chip rev: v1.0
I (1267) heap_init: Initializing. RAM available for dynamic allocation:
I (1274) heap_init: At 3FFAE6E0 len 00001920 (6 KiB): DRAM
I (1281) heap_init: At 3FFB3938 len 0002C6C8 (177 KiB): DRAM
I (1287) heap_init: At 3FFE0440 len 00003AE0 (14 KiB): D/IRAM
I (1293) heap_init: At 3FFE4350 len 0001BCB0 (111 KiB): D/IRAM
I (1300) heap_init: At 40093990 len 0000C670 (49 KiB): IRAM
I (1306) esp_psram: Adding pool of 4096K of PSRAM memory to heap allocator
I (1314) spi_flash: detected chip: generic
I (1318) spi_flash: flash io: qio
W (1323) i2c: This driver is an old driver, please migrate your application code to adapt driver/i2c_master.h
W (1333) ADC: legacy driver is deprecated, please migrate to esp_adc/adc_oneshot.h
I (1342) main_task: Started on CPU0
I (1352) esp_psram: Reserving pool of 32K of internal memory for DMA/internal allocations
I (1352) main_task: Calling app_main()
I (2352) SPI: ** Loading DSP firmware ret:0,Status:0 **
I (2352) SPI: b=1
I (2352) SPI: 1- Firmware boot loading started ....
I (5392) SPI: prgmBase 0x00080200

I (5392) SPI: execAddr 0x00084040

I (5392) SPI: DONE

I (5402) SPI: 2- Loading the image to RAM....done
I (5402) SPI: -- Saving firmware to flash....
I (11002) SPI: -- Saving firmware to flash....done
I (11022) SPI: 3- Loading the config file into the device RAM....
I (11192) SPI: -- Saving config to flash....
I (11432) SPI: -- Saving config to flash....done
I (11632) SPI: Device boot loading completed successfully...
I (11632) gpio: GPIO[22]| InputEn: 0| OutputEn: 1| OpenDrain: 0| Pullup: 0| Pulldown: 0| Intr:0
I (11642) gpio: GPIO[21]| InputEn: 0| OutputEn: 1| OpenDrain: 0| Pullup: 0| Pulldown: 0| Intr:0
W (11652) AUDIO_BOARD: The board has already been initialized!

----------------------------- ESP Audio Platform -----------------------------
| |
| ESP_AUDIO-v1.7.3-40-g00ffda8-dbab925-9815982 |
| Compile date: Apr 12 2024-02:32:01 |
I (11692) wwe_example: Func:setup_player, Line:140, MEM Total:4434152 Bytes, Inter:332075 Bytes, Dram:282103 Bytes

I (11702) wwe_example: esp_audio instance is:0x3f801d10

I (11822) wwe_example: Recorder has been created
I (11822) RECORDER_SR: The first wakenet model: wn5_nihaoxiaozhi

I (11822) AFE_SR: afe interface for speech recognition

I (11832) AFE_SR: AFE version: SR_V220727

I (11832) AFE_SR: Initial auido front-end, total channel: 2, mic num: 1, ref num: 1

I (11842) AFE_SR: aec_init: 0, se_init: 1, vad_init: 1

I (11842) AFE_SR: wakenet_init: 1

Deprecated Quantized wakenet5: wakeNet5_v1_nihaoxiaozhi_6_0.983_0.95, mode:0 (Nov 21 2023 19:15:51)
Please switch to ESP32-S3 and wakenet9 ! // 请问这个需要处理吗？按照说明ESP32-LyraTD-MSC是可以支持的啊？
I (11962) AFE_SR: wake num: 1, mode: 1, (Nov 21 2023 19:15:51)

SINGLE_RECOGNITION: v2; (Nov 21 2023 19:40:21)
W (11962) MN: MultiNet may cause WatchDog on ESP32 sometimes! //同问，既然ESP32-LyraTD-MSC可以支持，为啥还会watchdog超时？官方可以fix这个问题吗？
esp_mn_commands_update_from_sdkconfig
I (11972) MN_COMMAND: ---------------------SPEECH COMMANDS---------------------
I (11972) MN_COMMAND: Command ID0, phrase ID0: da kai kong tiao
I (11982) MN_COMMAND: Command ID1, phrase ID1: guan bi kong tiao
I (11992) MN_COMMAND: Command ID2, phrase ID2: zeng da feng su
I (11992) MN_COMMAND: Command ID3, phrase ID3: jian xiao feng su
I (12002) MN_COMMAND: Command ID4, phrase ID4: sheng gao yi du
I (12012) MN_COMMAND: Command ID5, phrase ID5: jiang di yi du
I (12012) MN_COMMAND: Command ID6, phrase ID6: zhi re mo shi
I (12022) MN_COMMAND: Command ID7, phrase ID7: zhi leng mo shi
I (12032) MN_COMMAND: Command ID8, phrase ID8: song feng mo shi
I (12032) MN_COMMAND: Command ID9, phrase ID9: jie neng mo shi
I (12042) MN_COMMAND: Command ID10, phrase ID10: chu shi mo shi
I (12042) MN_COMMAND: Command ID11, phrase ID11: jian kang mo shi
I (12052) MN_COMMAND: Command ID12, phrase ID12: shui mian mo shi
I (12062) MN_COMMAND: Command ID13, phrase ID13: da kai lan ya
I (12062) MN_COMMAND: Command ID14, phrase ID14: guan bi lan ya
I (12072) MN_COMMAND: Command ID15, phrase ID15: kai shi bo fang
I (12082) MN_COMMAND: Command ID16, phrase ID16: zan ting bo fang
I (12092) MN_COMMAND: Command ID17, phrase ID17: ding shi yi xiao shi
I (12092) MN_COMMAND: Command ID18, phrase ID18: da kai dian deng
I (12102) MN_COMMAND: Command ID19, phrase ID19: guan bi dian deng
I (12112) MN_COMMAND: ---------------------------------------------------------

I (12132) main_task: Returned from app_main() // 这条log打印出来后系统会停在这里，无论如何说提示词毫无反应，过了几十秒，系统就崩了，崩的log如下
Guru Meditation Error: Core 0 panic'ed (LoadStoreAlignment). Exception was unhandled.

Core 0 register dump:
PC : 0x4011e70f PS : 0x00060730 A0 : 0x800f69c5 A1 : 0x3f826950
0x4011e70f: bitrv2 at /home/sunxiangyu/workspace/esp_sr_lib/components/esp_audio_processor/noise_suppression/webrtc_ns.c:234

A2 : 0x00000100 A3 : 0x00000000 A4 : 0x3f826df0 A5 : 0x4015c205
A6 : 0x00000000 A7 : 0x00000000 A8 : 0x00000000 A9 : 0x7f7fffff
A10 : 0xb8bd1c6d A11 : 0xbffe49e8 A12 : 0xbb360b61 A13 : 0x3f826df0
A14 : 0x003fffff A15 : 0x3ffb8470 SAR : 0x00000004 EXCCAUSE: 0x00000009
EXCVADDR: 0x4015c205 LBEG : 0x400f8841 LEND : 0x400f8866 LCOUNT : 0x00000000
0x400f8841: WindowingEnergy at /home/sunxiangyu/workspace/esp_sr_lib/components/esp_audio_processor/noise_suppression/webrtc_ns.c:1645
(inlined by) WebRtcNs_ProcessCore at /home/sunxiangyu/workspace/esp_sr_lib/components/esp_audio_processor/noise_suppression/webrtc_ns.c:1916
0x400f8866: WebRtcNs_ProcessCore at /home/sunxiangyu/workspace/esp_sr_lib/components/esp_audio_processor/noise_suppression/webrtc_ns.c:1917

Backtrace: 0x4011e70c:0x3f826950 0x400f69c2:0x3f8269a0 0x400f8a39:0x3f8269f0 0x400f95be:0x3f827b10 0x400f4b89:0x3f827b30 0x400f36da:0x3f827b60 0x400f3898:0x3f827b80 0x400dd015:0x3f827ba0
0x4011e70c: bitrv2 at /home/sunxiangyu/workspace/esp_sr_lib/components/esp_audio_processor/noise_suppression/webrtc_ns.c:230
0x400f69c2: WebRtc_rdft at /home/sunxiangyu/workspace/esp_sr_lib/components/esp_audio_processor/noise_suppression/webrtc_ns.c:155
0x400f8a39: FFT at /home/sunxiangyu/workspace/esp_sr_lib/components/esp_audio_processor/noise_suppression/webrtc_ns.c:1564
(inlined by) WebRtcNs_ProcessCore at /home/sunxiangyu/workspace/esp_sr_lib/components/esp_audio_processor/noise_suppression/webrtc_ns.c:1944
0x400f95be: WebRtcNs_Process at /home/sunxiangyu/workspace/esp_sr_lib/components/esp_audio_processor/noise_suppression/webrtc_ns.c:2126
0x400f4b89: ns_process at /home/sunxiangyu/workspace/esp_sr_lib/components/esp_audio_processor/esp_ns.c:16
0x400f36da: afe_feed_aec_init_false at /home/sunxiangyu/workspace/esp_sr_lib/components/esp_audio_front_end/esp_afe_sr.c:546
0x400f3898: afe_feed at /home/sunxiangyu/workspace/esp_sr_lib/components/esp_audio_front_end/esp_afe_sr.c:623
0x400dd015: feed_task at C:/Users/hyuan/esp/esp-adf/components/audio_recorder/recorder_sr.c:248

ELF file SHA256: b05478608

CPU halted.
==================
sdkconfig file attached

sdkconfig.txt

(68.59 KiB) Downloaded 450 times

[Linc08]

我这两天也在看这个问题。
首先是没反应。我甚至将idf版本搞到v4.4，然后再尝试。
大家知道，现在docker都封了，简直生不如死。
我对esp32的开发也有点心灰意冷。
问题太多，没人理。

[ESP_William]

您是否在第一次烧录后执行了

Code: Select all

tw_upload_dsp_firmware(-1) ===> tw_upload_dsp_firmware(0)

？
这一步是为了更新 `zl38063.c` 的固件，更新后需要将其恢复回去。

另外，默认对于 ESP32-LyraTD-MSC ，WWE不再被支持，您可以用其他开发板进行测试。