Bootloader problem for RP Pico W

[exor2008]

Hi, I'am trying to implement firmware update for my RP Pi Pico W via Wi-Fi and I ran into a problem. So I tried to run the boot example from https://github.com/embassy-rs/embassy/tree/main/examples/boot/bootloader/rp and https://github.com/embassy-rs/embassy/tree/main/examples/boot/application/rp

This example compiling and flashing just fine, but I can't see the result, because https://github.com/embassy-rs/embassy/blob/main/examples/boot/application/rp/src/bin/b.rs is designed for RP Pico, not RP Pico W.

Commands I'am using to flash bootloader example:

# Flash bootloader
cargo flash --manifest-path ../../bootloader/rp/Cargo.toml --release --chip RP2040

# Build 'b'
cargo build --release --bin b

# Generate binary for 'b'
cargo objcopy --release --bin b -- -O binary b.bin

# Flash `a` (which includes b.bin)
cargo flash --release --bin a --chip RP2040

So I built https://github.com/embassy-rs/embassy/blob/main/examples/rp/src/bin/wifi_blinky.rs to binary and used it as b.bin for bootloading example. I used the same memory layout as in examples:

MEMORY
{
  /* NOTE 1 K = 1 KiBi = 1024 bytes */
  BOOT2 : ORIGIN = 0x10000000, LENGTH = 0x100
  BOOTLOADER_STATE : ORIGIN = 0x10006000, LENGTH = 4K
  FLASH : ORIGIN = 0x10007000, LENGTH = 512K
  DFU : ORIGIN = 0x10087000, LENGTH = 516K
  RAM : ORIGIN = 0x20000000, LENGTH = 256K
}

__bootloader_state_start = ORIGIN(BOOTLOADER_STATE) - ORIGIN(BOOT2);
__bootloader_state_end = ORIGIN(BOOTLOADER_STATE) + LENGTH(BOOTLOADER_STATE) - ORIGIN(BOOT2);

__bootloader_dfu_start = ORIGIN(DFU) - ORIGIN(BOOT2);
__bootloader_dfu_end = ORIGIN(DFU) + LENGTH(DFU) - ORIGIN(BOOT2);

After cargo flash --release --bin a --chip RP2040 I start to getting errors:

Erasing sectors ✔ [00:00:06] [#####################################################################################] 312.00 KiB/312.00 KiB @ 51.35 KiB/s (eta 0s ) Programming pages ✔ [00:00:00] [#########----------------------------------------------------------------------------] 32.00 KiB/312.00 KiB @ 331.17 KiB/s (eta 0s )
Error Something during the interaction with the core went wrong

or

Erasing sectors ✔ [00:00:03] [############################################################-------------------------] 220.00 KiB/312.00 KiB @ 82.59 KiB/s (eta 0s ) Programming pages ✔ [00:00:03] [---------------------------------------------------------------------------------------] 0B/312.00 KiB @ 81.89 KiB/s (eta 0s )
Error Failed to erase flash sector at address 0x1003e000.

[Athorus]

Hi, I had the same issue, it is caused by the watchdog, if you flash many data, as you have to do with wifi-blinky sample, the watchdog will reboot the pico as it takes too much time. So remove watchdog code from all the samples and all your flashes with probe-run will work fine !

Use Flash instead of WatchdogFlash in the bootloader and remove Watchdog in a.rs

It is a real issue as probe-run does not disable Watchdog.

[CBJamo]

The watchdog is also not enabled when in USB bootloader mode, so if your application needs the watchdog and is too big to flash before it resets, this is a viable, if annoying, workaround.

[lulf]

I'm wondering: would it be better if we disable the watchdog and instead make it some opt-in compile option? Although it's generally a 'good thing to do', I've seen multiple users hitting this.

[Athorus]

Watchdog is only useful when the pico is not plugged to a debug probe. With a debug probe, I can restart anytime I want.

If you know nothing about the watchdog, you think it is a random bug while flashing, as I did. This stucks many beginners.

[exor2008]

@Athorus @lulf @CBJamo thanks for your responses.

But I thought I figured it out with watchdog. I made it 8 sec and was feeding him during the flashing. And according to my observations an execution was never stuck for at least a single second. But I still getting those errors.

pub async fn update_firmware<'a>(
    firmware: &'static Channel<ThreadModeRawMutex, Vec<u8, 4096>, 1>,
    watchdog: WATCHDOG,
    flash: FLASH,
) {
    //Override bootloader watchdog
    let mut watchdog = Watchdog::new(watchdog);
    watchdog.start(Duration::from_secs(8));

    let flash: Flash<_, FLASH_SIZE> = Flash::new(flash);
    let flash = Mutex::new(RefCell::new(flash));

    let config = FirmwareUpdaterConfig::from_linkerfile_blocking(&flash);
    let mut updater = BlockingFirmwareUpdater::new(config);

    watchdog.feed();

    let mut offset = 0;
    let mut buf: AlignedBuffer<4096> = AlignedBuffer([0; 4096]);
    defmt::info!("preparing update");

    let state = updater
        .get_state(&mut buf.0[..1])
        .map_err(|e| defmt::warn!("E: {:?}", defmt::Debug2Format(&e)))
        .unwrap();

    let writer = updater
        .prepare_update(&mut buf.0[..1])
        .map_err(|e| defmt::warn!("E: {:?}", defmt::Debug2Format(&e)))
        .unwrap();
    defmt::info!("writer created, starting write");

    let mut chunk = firmware.recv().await;

    while !chunk.is_empty() {
        buf.0[..chunk.len()].copy_from_slice(&chunk.as_slice());
        defmt::info!("writing block at offset {}", offset);
        writer.write(offset, &buf.0[..chunk.len()]).unwrap();
        offset += chunk.len() as u32;
        watchdog.feed();
    }

    watchdog.feed();
    defmt::info!("firmware written ({} bytes), marking update", offset);

    updater.mark_updated(&mut buf.0[..1]).unwrap();
    Timer::after(Duration::from_secs(2)).await;
    defmt::info!("update marked, resetting");
    cortex_m::peripheral::SCB::sys_reset();
    loop {}
}

What confuses me is that sometimes an error occurs during the erasure of memory, and sometimes during the writing of a new one.

[Athorus]

I’m not sure of what I’m saying, but I think during flashing your soft is stopped but not the watchdog, that is was why you don’t know when the reset will happen. As I said, without watchdog in all the code (bootloader + soft), no issue while flashing.