Core changes while in cyw43_driver using FreeRTOS SMP #1478

New issue

Closed

opened 2023-08-19 06:07:58 -05:00 by ghost · 2 comments

ghost commented

2023-08-19 06:07:58 -05:00

(Migrated from github.com)

Sorry, I don't have a debugger at hand, so all is done via debug-print.

I try to run my Pico_W with FreeRTOS-SMP and WIFI and follow loosely the example. Everything is fine, except that the core switches while within the cyw43_driver. A small change in cyw43_driver.c to see what's going on:

// ...
static void cyw43_do_poll(async_context_t *context, __unused async_when_pending_worker_t *worker) {
#ifndef NDEBUG
    printf("--------------------------->4<---------------------\n");
    printf("async_context_core_num(cyw43_async_context) = %d\n", async_context_core_num(cyw43_async_context));
    printf("get_core_num() = %d\n", get_core_num());
    printf("get_core_num() = %d\n", get_core_num());
    printf("get_core_num() = %d\n", get_core_num());
    printf("get_core_num() = %d\n", get_core_num());
    assert(get_core_num() == async_context_core_num(cyw43_async_context));
#endif
// ---

Produces the following outputs:

Version: 7.95.49 (2271bb6 CY) CRC: b7a28ef3 Date: Mon 2021-11-29 22:50:27 PST Ucode Ver: 1043.2162 FWID 01-c51d9400
cyw43 loaded ok, mac xx:xx:xx:xx:xx:xx 
API: 12.2
Data: RaspberryPi.PicoW
Compiler: 1.29.4 
ClmImport: 1.47.1
Customization: v5 22/06/24 
Creation: 2022-06-24 06:55:08
[...]
--------------------------->4<---------------------
async_context_core_num(cyw43_async_context) = 0
get_core_num() = 1 
get_core_num() = 1 
get_core_num() = 0 
get_core_num() = 0 
--------------------------->3<---------------------
--------------------------->4<---------------------
async_context_core_num(cyw43_async_context) = 0
get_core_num() = 1 
get_core_num() = 1 
get_core_num() = 1 
get_core_num() = 1
assertion "get_core_num() == async_context_core_num(cyw43_async_context)" failed: file "..."

First thought was I missed the vTaskCoreAffinitySet() but it is there:

#define WLAN_PRIORITY (tskIDLE_PRIORITY + 1UL)

void WlanRtosInit(void)
{
    TaskHandle_t tTaskHandle;

    xTaskCreate(vTaskFunction,
                "WLAN",
                configMINIMAL_STACK_SIZE,
                NULL,
                WLAN_PRIORITY,
                &tTaskHandle);

    vTaskCoreAffinitySet(tTaskHandle, 1);  /* <---- Here the affinity is set to core 0  */
}

void vTaskFunction(void *pvParameters)
{
    if (cyw43_arch_init_with_country(CYW43_COUNTRY_GERMANY))
    {
        printf("failed to initialise\n");
        eRetVal = err_error;
    }
    else
    {
        cyw43_arch_enable_sta_mode();
        if (cyw43_arch_wifi_connect_timeout_ms(WLAN_SSID, WLAN_PASSWORD, CYW43_AUTH_WPA2_AES_PSK, 30000))
        // ...
    }
}

Note that I see the assert fires at all four places within cyw43_driver.c at random. And the vTaskFunction() runs reliable on core 0.

Sorry, I don't have a debugger at hand, so all is done via debug-print. I try to run my Pico_W with FreeRTOS-SMP and WIFI and follow loosely the example. Everything is fine, except that the core switches while within the cyw43_driver. A small change in `cyw43_driver.c` to see what's going on: ```c // ... static void cyw43_do_poll(async_context_t *context, __unused async_when_pending_worker_t *worker) { #ifndef NDEBUG printf("--------------------------->4<---------------------\n"); printf("async_context_core_num(cyw43_async_context) = %d\n", async_context_core_num(cyw43_async_context)); printf("get_core_num() = %d\n", get_core_num()); printf("get_core_num() = %d\n", get_core_num()); printf("get_core_num() = %d\n", get_core_num()); printf("get_core_num() = %d\n", get_core_num()); assert(get_core_num() == async_context_core_num(cyw43_async_context)); #endif // --- ``` Produces the following outputs: ``` Version: 7.95.49 (2271bb6 CY) CRC: b7a28ef3 Date: Mon 2021-11-29 22:50:27 PST Ucode Ver: 1043.2162 FWID 01-c51d9400 cyw43 loaded ok, mac xx:xx:xx:xx:xx:xx API: 12.2 Data: RaspberryPi.PicoW Compiler: 1.29.4 ClmImport: 1.47.1 Customization: v5 22/06/24 Creation: 2022-06-24 06:55:08 [...] --------------------------->4<--------------------- async_context_core_num(cyw43_async_context) = 0 get_core_num() = 1 get_core_num() = 1 get_core_num() = 0 get_core_num() = 0 --------------------------->3<--------------------- --------------------------->4<--------------------- async_context_core_num(cyw43_async_context) = 0 get_core_num() = 1 get_core_num() = 1 get_core_num() = 1 get_core_num() = 1 assertion "get_core_num() == async_context_core_num(cyw43_async_context)" failed: file "..." ``` First thought was I missed the `vTaskCoreAffinitySet()` but it is there: ```c #define WLAN_PRIORITY (tskIDLE_PRIORITY + 1UL) void WlanRtosInit(void) { TaskHandle_t tTaskHandle; xTaskCreate(vTaskFunction, "WLAN", configMINIMAL_STACK_SIZE, NULL, WLAN_PRIORITY, &tTaskHandle); vTaskCoreAffinitySet(tTaskHandle, 1); /* <---- Here the affinity is set to core 0 */ } void vTaskFunction(void *pvParameters) { if (cyw43_arch_init_with_country(CYW43_COUNTRY_GERMANY)) { printf("failed to initialise\n"); eRetVal = err_error; } else { cyw43_arch_enable_sta_mode(); if (cyw43_arch_wifi_connect_timeout_ms(WLAN_SSID, WLAN_PASSWORD, CYW43_AUTH_WPA2_AES_PSK, 30000)) // ... } } ``` Note that I see the assert fires at all four places within `cyw43_driver.c` at random. And the `vTaskFunction()` runs reliable on core 0.

kilograham commented

2023-08-19 19:57:14 -05:00

(Migrated from github.com)

follow loosely the example

you'll have to be a bit more specific... what libraries are you linking in your CMakeLists.txt, particularly are you sure you're using the FreeRTOS version of pico_cyw43 stuff

> follow loosely the example you'll have to be a bit more specific... what libraries are you linking in your `CMakeLists.txt`, particularly are you sure you're using the FreeRTOS version of pico_cyw43 stuff

ghost commented

2023-08-20 04:35:53 -05:00

(Migrated from github.com)

TL;DR: Everything works fine now.

Sorry for not provide all info from the beginning. But by collecting all information I freshly updated the local FreeRTOS and pico-sdk repositories to make sure that I have the right versions. And the problem is gone. What the actual problem was is unknown and probably not reproducible.

Anyway, thanks for your support! 👍

TL;DR: Everything works fine now. Sorry for not provide all info from the beginning. But by collecting all information I freshly updated the local FreeRTOS and pico-sdk repositories to make sure that I have the right versions. And the problem is gone. What the actual problem was is unknown and probably not reproducible. Anyway, thanks for your support! :+1: