search

espressif / esp-hosted

Hosted Solution (Linux/MCU) with ESP32 (Wi-Fi + BT + BLE)
Other
689 stars 160 forks source link

ESP32-P4 can not connect to ESP32-C6 on the Function_EV_Board v1.2 and v1.4 #511

Open me-no-dev opened 1 week ago

me-no-dev commented 1 week ago

Checklist

How often does this bug occurs?

always

Expected behavior

I expect to be able to connect to the C6 over SDIO

Actual behavior (suspected bug)

Instead I get E (226409) transport: Not able to connect with ESP-Hosted slave device

Error logs or terminal output

E (226409) transport: Not able to connect with ESP-Hosted slave device

Steps to reproduce the behavior

Follow your steps in https://github.com/espressif/esp-hosted/blob/feature/esp_as_mcu_host/docs/esp32_p4_function_ev_board.md

Project release version

latest

System architecture

ARM 64-bit (Apple M1/M2, Raspberry Pi 4/5)

Operating system

MacOS

Operating system version

15.0.1

Shell

Bash

Additional context

I made sure that the versions of both the P4 host and C6 slave are the same. Logs on both devices do not show any errors, other than the one above. C6 shows zero packets and is properly reset by the P4 host. Tested on my two EV boards v1.2 and v1.4. This is for adding support for P4 to the Arduino project

me-no-dev commented 1 week ago

Pin configuration on both host and slave have been verified to be consistent with schematic and documentation

me-no-dev commented 1 week ago

Forgot to mention that everything is built with the latest release/v5.3 branch of ESP-IDF

mantriyogesh commented 1 week ago

We need following info for understanding:

Sdkconfig both sides, and complete textual logs, Hosted registry component version used, any changes you have done in code or pins to be used.

The complete flashing steps and logs would be helpful.

me-no-dev commented 1 week ago

espressif/esp_hosted (0.0.22) no changes have been done. Pins are as defaults, which have been checked to match the P4 EV board schematics. C6-flash-and-monitor.txt sdkconfig.slave.txt sdkconfig.p4.txt P4-flash-and-monitor.txt

mantriyogesh commented 1 week ago

For C6, the log is the success log.

Not sure why the C6 did not bootup correctly at the end.

The end log in C6 is shows as :

ESP-ROM:esp32c6-20220919
Build:Sep 19 2022
rst:0x1 (POWERON),boot:0xc (SPI_FAST_FLASH_BOOT)
SPIWP:0xee
mode:DIO, clock div:2
load:0x40875720,len:0x1804
load:0x4086c110,len:0xe2c
load:0x4086e610,len:0x2e58
entry 0x4086c11a

repeats boot

What happens after this log?

Is the P4 sufficiently powered?

me-no-dev commented 1 week ago

this happens when the P4 reboots the C6, after which the same boot log repeats. This is what I meant by repeats boot

me-no-dev commented 1 week ago

C6 boots correctly, the P4 attempts connect, timeouts and resets the C6 (using it's IO54)

mantriyogesh commented 1 week ago

I can test this on ESP32-P4-Function-EV-Board v1.2 earliest by tomorrow. Let you know if I can reproduce with your sdkconfig.

Also, if works fine let me share the logs and observations.

me-no-dev commented 1 week ago

Sure thing. I'm waiting for @igrr to give me known working binaries to at least figure out on which side is the problem. There is nothing specific I can add from our side. It's a basic WiFi scan firmware that runs on all other chips (that have WiFi).

me-no-dev commented 1 week ago

Another unrelated issue I see is that always after the same number of unsuccessful attempts (100), the host reboots because of the following error:

E (1117493) transport: Failed to get ESP_Hosted slave transport up
ESP_ERROR_CHECK failed: esp_err_t 0xffffffff (ESP_FAIL) at 0x4003a48a
file: "./managed_components/espressif__esp_hosted/host/api/esp_hosted_api.c" line 183
func: esp_wifi_remote_init
expression: transport_drv_reconfigure()

abort() was called at PC 0x4ff0734f on core 1
mantriyogesh commented 1 week ago

c6_p4_sdio_flashing_0.0.22_idf_dbf5934b.tgz

@me-no-dev , PFA C6 - SDIO - P4 tested binaries.

me-no-dev commented 6 days ago

@mantriyogesh the problem seems to be in my p4 bin. I see that you compile with IDF 5.4. Can you try with 5.3? We can not yet update to 5.4 in Arduino

mantriyogesh commented 6 days ago

Any specific idf commit in 5.3? or v5.3 in https://github.com/espressif/esp-idf/tags should be fine?

me-no-dev commented 6 days ago

we use the head of release/v5.3 branch

me-no-dev commented 6 days ago

v5.3.1-395-g707d097b017 is the version printed. The C6 firmware I compiled is working fine with your P4 bin. Only the P4 binary is wrong on our end

mantriyogesh commented 6 days ago

I observed, by default esp_wifi_remote getting checked-out as 

[1/10] cmd_system (*) (/Users/yogesh/code/idf6/examples/system/console/advanced/components/cmd_system)
[2/10] esp-qa/ping-cmd (0.0.1)
[3/10] esp-qa/wifi-cmd (0.1.8)
[4/10] espressif/eppp_link (0.2.0)
[5/10] espressif/esp_hosted (0.0.22)
[6/10] espressif/esp_serial_slave_link (1.1.0)
[7/10] espressif/esp_wifi_remote (0.3.0)
[8/10] espressif/iperf (0.1.3)
[9/10] espressif/iperf-cmd (0.1.3)
[10/10] idf (5.3.1)

When I change, main/idf_component.yml so that 

  espressif/esp_wifi_remote:
    version: "~0.4.1"

It picked correct esp_wifi_remote (0.4.x needed for use hosted by default, else manually need to update from Wifi Remote menu)

[1/10] cmd_system (*) (/Users/yogesh/code/idf6/examples/system/console/advanced/components/cmd_system)
[2/10] esp-qa/ping-cmd (0.0.1)
[3/10] esp-qa/wifi-cmd (0.1.8)
[4/10] espressif/eppp_link (0.2.0)
[5/10] espressif/esp_hosted (0.0.22)
[6/10] espressif/esp_serial_slave_link (1.1.0)
[7/10] espressif/esp_wifi_remote (0.4.1)
[8/10] espressif/iperf (0.1.3)
[9/10] espressif/iperf-cmd (0.1.3)
[10/10] idf (5.3.1)
- [7/10] espressif/esp_wifi_remote (0.3.0)
+ [7/10] espressif/esp_wifi_remote (0.4.1)

c6_p4_bins_2.tgz

me-no-dev commented 6 days ago

it's espressif/esp_wifi_remote (0.4.1) on our end too

mantriyogesh commented 6 days ago

https://github.com/espressif/esp-hosted/issues/511#issuecomment-2413750873

Binaries should work anyway, I built through release/v5.3

me-no-dev commented 6 days ago

what commit is that, because I can not check it out. is it public 5.3 or gitlab?

me-no-dev commented 6 days ago

I also see a very different output when starting hosted that I do not see on our end:

[0;32mI (383) host_init: ESP Hosted : Host chip_ip[18]
I (411) H_API: ESP-Hosted starting. Hosted_Tasks: prio:23, stack: 5120 RPC_task_stack: 5120
sdio_mempool_create free:172624 min-free:172624 lfb-def:131072 lfb-8bit:131072

I (417) gpio: GPIO[18]| InputEn: 0| OutputEn: 0| OpenDrain: 0| Pullup: 1| Pulldown: 0| Intr:0 
I (426) gpio: GPIO[19]| InputEn: 0| OutputEn: 0| OpenDrain: 0| Pullup: 1| Pulldown: 0| Intr:0 
I (435) gpio: GPIO[14]| InputEn: 0| OutputEn: 0| OpenDrain: 0| Pullup: 1| Pulldown: 0| Intr:0 
I (444) gpio: GPIO[15]| InputEn: 0| OutputEn: 0| OpenDrain: 0| Pullup: 1| Pulldown: 0| Intr:0 
I (454) gpio: GPIO[16]| InputEn: 0| OutputEn: 0| OpenDrain: 0| Pullup: 1| Pulldown: 0| Intr:0 
I (463) gpio: GPIO[17]| InputEn: 0| OutputEn: 1| OpenDrain: 0| Pullup: 0| Pulldown: 0| Intr:0 
I (472) H_API: ** add_esp_wifi_remote_channels **
I (478) transport: Add ESP-Hosted channel IF[1]: S[0] Tx[0x4000caee] Rx[0x4001a5ec]
I (486) transport: Add ESP-Hosted channel IF[2]: S[0] Tx[0x4000ca34] Rx[0x4001a5ec]
me-no-dev commented 6 days ago

Is it possible that we need to call some hosted API to properly init and not just call WiFi APIs?

mantriyogesh commented 6 days ago

esp_hosted_host_init() is hooked through the component init. See here: https://github.com/espressif/esp-hosted/blob/83efce638beb0d59b641399df862b20e3aa12f36/host/esp_hosted_host_init.c#L10

This is hooked to component constructor. So do you mean you do not get to call this api automatically?

If you are using esp-idf, ideally this should have been called.

You case is interesting, let me know how can I reproduce this easily..

mantriyogesh commented 6 days ago 
![Screenshot 2024-10-15 at 8 51 04 PM](https://github.com/user-attachments/assets/673238f2-3171-46a4-9bd6-c3b79130bda5)

 I  ~/c/idf6   release/v5.3 *$…  examples/wifi/iperf  git remote -v  
origin  ssh://git@gitlab.espressif.cn:27227/espressif/esp-idf.git (fetch)
origin  ssh://git@gitlab.espressif.cn:27227/espressif/esp-idf.git (push)

This is indeed gitlab. But shouldn't matter, it was working on GitHub idf 5.3 also.

me-no-dev commented 6 days ago

we use ESP-IDF in a slightly different way. I do expect this should have worked and indeed it's strange it did not. in essence we build a very basic Arduino project in ESP-IDF (Arduino as component) and then extract all necessary libraries, headers, linker scripts and flags that were used by ESP-IDF to compile the project, then in Arduino we build the "Arduino component" again with the user code and link against everything extracted in the previous step. We also have some "constructor" code, but it is called in a slightly different way https://github.com/espressif/arduino-esp32/blob/idf-release/v5.3/cores/esp32/esp32-hal-misc.c#L256-L259

me-no-dev commented 6 days ago

is there any other way that we can manually call esp_hosted_host_init?

mantriyogesh commented 6 days ago

You can in app_main() or wherever you like, just before using esp_wifi_init()

Alternative 1) with extern, call this function.

Alternative 2) you can include header #include esp_hosted_api.h and call 

ESP_ERROR_CHECK(esp_hosted_init(NULL))
me-no-dev commented 6 days ago

option 2 throws

In file included from /Users/ficeto/Documents/Arduino/hardware/espressif/esp32/tools/esp32-arduino-libs/esp32p4/include/espressif__esp_hosted/host/api/esp_hosted_api.h:27,
                 from /var/folders/74/n3f3k71s7lv_snk832kfgjfw0000gn/T/arduino_modified_sketch_53374/WiFiScan.ino:8:
/Users/ficeto/Documents/Arduino/hardware/espressif/esp32/tools/esp32-arduino-libs/esp32p4/include/espressif__esp_hosted/host/drivers/transport/transport_drv.h:106:6: error: variable or field 'process_priv_communication' declared void
  106 | void process_priv_communication(interface_buffer_handle_t *buf_handle);
      |      ^~~~~~~~~~~~~~~~~~~~~~~~~~
/Users/ficeto/Documents/Arduino/hardware/espressif/esp32/tools/esp32-arduino-libs/esp32p4/include/espressif__esp_hosted/host/drivers/transport/transport_drv.h:106:33: error: 'interface_buffer_handle_t' was not declared in this scope; did you mean 'StreamBufferHandle_t'?
  106 | void process_priv_communication(interface_buffer_handle_t *buf_handle);
      |                                 ^~~~~~~~~~~~~~~~~~~~~~~~~
      |                                 StreamBufferHandle_t
/Users/ficeto/Documents/Arduino/hardware/espressif/esp32/tools/esp32-arduino-libs/esp32p4/include/espressif__esp_hosted/host/drivers/transport/transport_drv.h:106:60: error: 'buf_handle' was not declared in this scope
  106 | void process_priv_communication(interface_buffer_handle_t *buf_handle);
      |                                                            ^~~~~~~~~~
/Users/ficeto/Documents/Arduino/hardware/espressif/esp32/tools/esp32-arduino-libs/esp32p4/include/espressif__esp_hosted/host/drivers/transport/transport_drv.h:129:23: error: 'interface_buffer_handle_t' was not declared in this scope; did you mean 'StreamBufferHandle_t'?
  129 | int serial_rx_handler(interface_buffer_handle_t * buf_handle);
      |                       ^~~~~~~~~~~~~~~~~~~~~~~~~
      |                       StreamBufferHandle_t
/Users/ficeto/Documents/Arduino/hardware/espressif/esp32/tools/esp32-arduino-libs/esp32p4/include/espressif__esp_hosted/host/drivers/transport/transport_drv.h:129:51: error: 'buf_handle' was not declared in this scope
  129 | int serial_rx_handler(interface_buffer_handle_t * buf_handle);
      |                                                   ^~~~~~~~~~
me-no-dev commented 6 days ago

option 1 I doubt it will work, because the function is static

me-no-dev commented 6 days ago 
WiFiScan:8:8: error: 'static' specifier conflicts with 'extern'
    8 | extern static void esp_hosted_host_init(void);
      | ~~~~~~ ^~~~~~
me-no-dev commented 6 days ago

this worked

extern "C" {
esp_err_t esp_hosted_init(void *);
}
me-no-dev commented 6 days ago

small nitpick. The following text is printed regardless of the log level. Can it please be changed to INFO or lower?

sdio_mempool_create free:34125312 min-free:34124992 lfb-def:33030132 lfb-8bit:33030132

Name: 
Type: SDIO
Speed: 40.00 MHz (limit: 40.00 MHz)
Size: 0MB
CSD: ver=1, sector_size=0, capacity=0 read_bl_len=0
SCR: sd_spec=0, bus_width=0
TUPLE: DEVICE, size: 3: D9 01 FF 
TUPLE: MANFID, size: 4
  MANF: 0092, CARD: 6666
TUPLE: FUNCID, size: 2: 0C 00 
TUPLE: FUNCE, size: 4: 00 00 02 32 
TUPLE: CONFIG, size: 5: 01 01 00 02 07 
TUPLE: CFTABLE_ENTRY, size: 8
  INDX: C1, Intface: 1, Default: 1, Conf-Entry-Num: 1
  IF: 41
  FS: 30, misc: 0, mem_space: 1, irq: 1, io_space: 0, timing: 0, power: 0
  IR: 30, mask: 1,   IRQ: FF FF
  LEN: FFFF
TUPLE: END
mantriyogesh commented 6 days ago

great thanks for the feedback, will add this in. btw, is everything running fine now?

me-no-dev commented 6 days ago

yes! WiFi is starting, all events are received. Everything seems OK with this workaround

mantriyogesh commented 6 days ago

While adding, this was a design choice we made, instead of hooking to the other way (which you already integrated). As this current way was very private and only would be integrated using constructor, only on loading of the component.

This way it was very much abstracted but only usable when needed.

I will give thoughts later on this, any other way possible to handle.

me-no-dev commented 6 days ago

Please do let me know if you change the behavior so I can make sure that everything on our side is OK. I am fine with this as-is, because it allows me to init SDIO only if WiFi is used on the P4 in Arduino (thus allow for boards without C6 to use the pins as they wish)

mantriyogesh commented 6 days ago

Sure. Your username is very much thoughtful and easy to remember! I will notify once I get around this.