Setting AP/STA mode resets PicoW during debug

[HouseSimOn]

Description

Last time I decided to try use PicoProbe to have a debugger. I'm using Pico as PicoProbe connected to a PicoW. After some troubles (for example with sleep_ms function in debug https://github.com/raspberrypi/picoprobe/issues/45) it works. Then I decided to load simple example with cyw43 driver to setup AP or to set STA mode. Unfortunately, call of the cyw43_arch_enable_ap_mode or cyw43_arch_enable_sta_mode functions cause reset of the Pico and debug session is interrupted. Program loaded to the PicoW without debugger works.

First I've used pico_cyw43_arch_lwip_threadsafe_background but I thought that maybe multi-core under debug is a problem but no... pico_cyw43_arch_lwip_poll also fails.

Reproduce

You can find my test example here: https://github.com/HouseSimOn/PicoSdkStartTemplate/blob/test/cyw43debug/Source/main.c - It's ready to compile and test.

Output

Debug session

1. cyw43_arch_enable_ap_mode("MyAPTest", "12345678", CYW43_AUTH_WPA2_AES_PSK); is called:

2. cyw43_arch_enable_ap_mode is commented out:

No debug

cyw43_arch_enable_ap_mode("MyAPTest", "12345678", CYW43_AUTH_WPA2_AES_PSK); is called - it works as expected. AP is created.

[peterharperuk]

I debug pico-w all the time with a debug probe. I can't imagine how/why that call would cause a problem for the debugger.

[peterharperuk]

Cannot reproduce. I ran your code via the debugger and nothing "reset"

[HouseSimOn]

You used DebugProbe, I'm using PicoProbe (the latest version: https://github.com/raspberrypi/picoprobe/releases/tag/picoprobe-cmsis-v1.0.3). Do you have possibility to test it using Pico as PicoProbe?

Have you added set USE_CORE 0 to the target/rp2040.cfg file? Have you used my launch.json file to test it?

launch.json

I'm on macOs M2 so I can't install gdb. I use arm-none-eabi-gdb.

{
    "version": "0.2.0",
    "configurations": [
        {
            "name": "Pico Debug",
            "cwd": "${workspaceRoot}",
            "executable": "${command:cmake.launchTargetPath}",
            "request": "launch",
            "type": "cortex-debug",
            "servertype": "openocd",
            // This may need to be "arm-none-eabi-gdb" for some previous builds
            "gdbPath" : "arm-none-eabi-gdb",
            "device": "RP2040",
            "configFiles": [
                // This may need to be "interface/picoprobe.cfg" for some previous builds
                "interface/cmsis-dap.cfg",
                "target/rp2040.cfg"
            ],
            "svdFile": "./Sdk/pico-sdk/src/rp2040/hardware_regs/rp2040.svd",
            "runToEntryPoint": "main",
            // Work around for stopping at main on restart
            "postRestartCommands": [
                "break main",
                "continue"
            ],
            "liveWatch": {
                "enabled": true,
                "samplesPerSecond": 1
            },
            "openOCDLaunchCommands": [
                "adapter speed 5000"
            ]
        }
    ]
}

[peterharperuk]

I used linux I don't have a mac. Yes I used a Pico to do a debug. I suggest you ask on the forum, maybe they'll have better luck guessing what you're doing wrong.

[HouseSimOn]

@peterharperuk I've performed test on Ubuntu 22.04.3 LTS. I've used gdb-multiarch so I had to change one line in launch.json to:

"gdbPath" : "gdb-multiarch"

I've also added set USE_CORE 0 to the target/rp2040.cfg file. And there is the same problem as on macOS. Reset after cyw43_arch_enable_ap_mode call. Everything work ok when this line is commented out.

Questions:

1. Have you added set USE_CORE 0 to the target/rp2040.cfg file?
2. Do you use official openocd or from this repo: https://github.com/raspberrypi/openocd.git?
3. Is there anything else that should be configured/installed than:
   • install openocd,
   • install gdb-multiarch or use arm-none-eabi-gdb,
   • change rp2040.cfg?

I've created post on forum: https://forums.raspberrypi.com/viewtopic.php?t=361958 as you suggested.

[peterharperuk]

Why are you setting USE_CORE? Does it work without that?

[HouseSimOn]

See this issue: https://github.com/raspberrypi/picoprobe/issues/45 I had to setup USE_CORE because sleep_ms function doesn't work in debug - or maybe I mixed something?

[lurch]

Based on the comments in https://github.com/raspberrypi/picoprobe/issues/45#issuecomment-1332610961 (and having a quick search in the RP2040 datasheet) I guess setting TIMER: DBGPAUSE.DBG1 to 0 (in your code) might be an alternative to telling OpenOCD to only use core 0? (see https://github.com/raspberrypi/pico-sdk/blob/master/src/rp2040/hardware_regs/include/hardware/regs/timer.h#L157 and https://github.com/raspberrypi/pico-sdk/blob/master/src/rp2040/hardware_structs/include/hardware/structs/timer.h#L63 )

[HouseSimOn]

@lurch Adding timer_hw->dbgpause = 0; to the code doesn't help, but it's true that it helps with sleep_ms issue. In this case we don't have to modify target/rp2040.cfg file.

I connected PicoProbe with PicoW as on the picture below (source: https://datasheets.raspberrypi.com/pico/getting-started-with-pico.pdf):

I solved issue when I connected to the PicoW board USB cable (and disconnected VSYS between boards). It looks like there is no enough power on VSYS pin to power PicoW when AP/STA mode starts (any current-peak?). I have to investigate it. But maybe this information should be added to the Picoprobe wiring subsection in Getting started with Raspberry Pi Pico document?

[peterharperuk]

I couldn't seem to reproduce this. I had the usb connected to the pico debugger which is powering the pico_w as shown above. I could connect to wifi just fine when debugging.