[gojimmypi]

I’ve updated the repo with two separate VisualGDB project files:

• Pure Windows-only RISC-V build (using xpack install in scripts directory)
• Mixed build using WSL RISC-V toolchain

https://github.com/ulx3s/Hazard3-Doom/tree/main/VisualGDB

There’s a complete getting started here:

https://ulx3s.github.io/ulx-doom/

 

 

 

[gojimmypi]

The gnu++2b was in my (~ 27th of) August commits  of the ESP-IDF, here in esp-idf/tools/cmake/build.cmake.

The most recent version exhibits a similar syntax highlighting problem, with a different root cause: -std=gnu23 here.

My fix above does still work, although clearly a bit of a hack, needing an additional replace:

string(REPLACE "-std=gnu2b" "-std=gnu23" CMAKE_C_FLAGS "${CMAKE_C_FLAGS}")

I’ll try to take the 6.1 for a test drive soon. Thanks for the heads up.

 

[gojimmypi]

Looking forward to v6.1!

In the meantime, I have this in a batch file to launch VS2022 for unreleased ESP-IDF v6 (latest via GitHub)

@echo off

set IDF_COMPONENT_STORAGE_URL=file:///C:/SysGCC/esp32-master/registry;default
set IDF_PYTHON_ENV_PATH=C:\SysGCC\esp32-master\python_env\
set IDF_TOOLS_PATH=C:\SysGCC\esp32-master
set CONFIG_WOLFSSL_USE_MY_PRIVATE_CONFIG=1
set CONFIG_WOLFSSL_FORCE_V6_INTELLISENSE_FIX=1

rem Launch Visual Studio without making this console wait
start "" "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\Common7\IDE\devenv.exe" %*
exit /b

see also related Intellisense for ESP-IDF v6: https://sysprogs.com/w/forums/topic/espressif-esp-idf-v6-0-intellisense-error-invalid-value-gnu2b-in-stdgnu2b/

[gojimmypi]

Heads up the above solution only worked for my local project source code.

For other source code in my component being developed, as well as the entire ESP-IDF source,  I added this to the main project cmake, just before the ending project(projectname):

if(  "$ENV{CONFIG_WOLFSSL_FORCE_V6_INTELLISENSE_FIX}" STREQUAL "1" )
    if(DEFINED IDF_VERSION_MAJOR AND IDF_VERSION_MAJOR GREATER_EQUAL 6)
        message(STATUS "-- Found CONFIG_WOLFSSL_FORCE_V6_INTELLISENSE_FIX, replacing -std=gnu2b with -std=gnu2x")
        if(CMAKE_C_COMPILER_ID MATCHES "Clang")
          string(REPLACE "-std=gnu2b" "-std=gnu2x" CMAKE_C_FLAGS "${CMAKE_C_FLAGS}")
        endif()
    else()
        message(STATUS "-- Visual Studio Intellisense Fix not needed for this ESP-IDF version=${IDF_VERSION_MAJOR}")
    endif()
else()
    message(STATUS "-- Not replacing -std=gnu2b with -std=gnu2x for Viosual Studio Intellisense fix")
    message(STATUS "-- To enable, define environment variable: CONFIG_WOLFSSL_FORCE_V6_INTELLISENSE_FIX=1")
endif()

# Once the project is loaded, next check for ESP-IDF version 6 or greater.
# Numerous "dangerous relocation: call8: call target out of range: memcpy" errors encountered
# So we'll allow long calls with the -mlongcalls compiler option for all components.
if(IDF_VERSION_MAJOR GREATER_EQUAL 6)
    if(IDF_TARGET STREQUAL "esp32" OR IDF_TARGET STREQUAL "esp32s2" OR IDF_TARGET STREQUAL "esp32s3")
        message(STATUS "mlongcalls for all components")
        idf_build_set_property(COMPILE_OPTIONS "-mlongcalls" APPEND)
    endif()
endif()

Also added this to the local component in the project directory cmake, just after the idf_component_register():

        if(  "$ENV{CONFIG_WOLFSSL_FORCE_V6_INTELLISENSE_FIX}" STREQUAL "1" )
            if(DEFINED IDF_VERSION_MAJOR AND IDF_VERSION_MAJOR GREATER_EQUAL 6)
                message(STATUS "-- Setting -std=gnu17 with target_compile_options $<$:-std=gnu17>")
                target_compile_options(${COMPONENT_LIB} PRIVATE $<$:-std=gnu17>)
            else()
                message(STATUS "-- Visual Studio Intellisense Fix not needed for this ESP-IDF version=${IDF_VERSION_MAJOR}")
            endif()
        else()
            message(STATUS "-- Not setting -std=gnu17 with target_compile_options for Visual Studio Intellisense Fix")
            message(STATUS "-- To enable, define environment variable: CONFIG_WOLFSSL_FORCE_V6_INTELLISENSE_FIX=1")
        endif()

Note that it is optional code, based on the environment variable: CONFIG_WOLFSSL_FORCE_V6_INTELLISENSE_FIX=1

 

I’m note sure exactly which are needed, but in order for syntax highlighting to be reliably restored for me:

• Exit Visual Studio
• Delete ./build directory
• Delete ./visualgdb directory
• Delete ./vs directory
• Restart Visual Studio

[gojimmypi]

fwiw, still occurring with the latest update today (ESP32  3.3.1) See attached

VisualGDB 6.0R8 (build 5338)

Looks like of the 170+ problems, included is at least the ESP Matter Connected Home stuff. Undesired to just ignore.

Perhaps an alternative install location can be used?

There’s a Windows fix that I personally choose not to use, as I also test my own distributions for other things:

https://learn.microsoft.com/en-us/windows/win32/fileio/maximum-file-path-limitation?tabs=registry

Attachments:

You must be logged in to view attached files.

[gojimmypi]

Heads up ESP-IDF v5.5 has been released.

https://github.com/espressif/esp-idf/releases/tag/v5.5

Beware:  “Release v5.5 is mostly compatible with apps written for ESP-IDF v5.4.”

See the section: “breaking changes”

• This reply was modified 1 year, 1 month ago by gojimmypi.

[gojimmypi]

Heads up the Espressif ESP-IDF v5.5 is now in beta:

https://github.com/espressif/esp-idf/releases/tag/v5.5-beta1

I was initially unable to get VisualGDB to use the new toolchain for the same cmake version issue described in my original message.

However, I found this topic from last year: “Problem on new ESP32 toolchain”:

https://sysprogs.com/w/forums/topic/problem-on-new-esp32-toolchain/

Specifically the link to “Troubleshooting”:

https://visualgdb.com/documentation/espidf/#troubleshooting

To resolve, I set my Tools – Options – VisualGDB – General – Python Directory (ESP-IDF) to

C:\Users\gojimmypi\AppData\Local\VisualGDB\Python-3.11.5

 

 

 

• This reply was modified 1 year, 2 months ago by gojimmypi.

[gojimmypi]

That seemed like a reasonable approach, although I was left wondering why system-wide environment value would not be “seen” by cmake?

In any case, it didn’t work.  (see attached). I still see the error:

Requirement files:
- C:\SysGCC\esp32\esp-idf\v5.2\tools\requirements\requirements.core.txt
Python being checked: C:\SysGCC\esp32-master\python_env\\idf5.2_py3.8_env\Scripts\python.exe
Manifest files have changed, solving dependencies.
..-- Configuring incomplete, errors occurred!
CMake Error at C:/SysGCC/esp32/esp-idf/v5.2/tools/cmake/build.cmake:544 (message):
WARNING: Component "gojimmypi/mywolfssl" not found

ERROR: Because project depends on gojimmypi/mywolfssl (^5.7.1-preview2f1)
which doesn't match any versions, version solving failed.

Call Stack (most recent call first):

Here it is in WSL using the same toolchain; first, the error is expected since there’s no release called mywolfssl

$ idf.py add-dependency "gojimmypi/mywolfssl^5.7.1-preview2f1"
Executing action: add-dependency
ERROR: Component "gojimmypi/mywolfssl" not found

Here it is setting the staging site environment variable:

gojimmypi:/mnt/c/workspace/esp32-homekit-demo-gojimmypi-pr/examples/led
$ export IDF_COMPONENT_REGISTRY_URL=https://components-staging.espressif.com
gojimmypi:/mnt/c/workspace/esp32-homekit-demo-gojimmypi-pr/examples/led
$ idf.py add-dependency "gojimmypi/mywolfssl^5.7.1-preview2f1"
Executing action: add-dependency
Successfully added dependency "gojimmypi/mywolfssl^5.7.1-preview2f1" to component "main"

It then successfully builds using the staging instance of gojimmypi/mywolfssl.

Also, after attempting to build and revisiting the IDF_COMPONENT_REGISTRY_URL=https://components-staging.espressif.com shown in the attached screen snip, when clicking on the newly created IDF_COMPONENT_REGISTRY_URL value, Visual Studio became unresponsive and restarted.

Other than this anomaly, VisualGDB is working great.

Attachments:

You must be logged in to view attached files.

[gojimmypi]

Aha! Great! Thanks for the update.

I should add this is also quite helpful for installing multiple versions of the ESP-IDF toolchain:

https://visualgdb.com/documentation/espidf/#toolchains

[gojimmypi]

In my C:\SysGCC I have manually installed C:\SysGCC\esp32-11.2 and C:\SysGCC\esp32-8.4 for the older toolchains, in addition to the default install of C:\SysGCC\esp32 which has version 12.1 of the toolchain.

Inside each of *those* directories, I have different versions of the ESP-IDF SDK, for instance:

In my C:\SysGCC\esp32\esp-idf, I performed “git clone [repo] [sdk name]” for these:

esp-idf-v5.2-beta1
master
v5.0
v5.1

That allows relatively each changing of SDK and toolchains. Be sure to delete the build directory and sdkconfig file when changing.

I’ve found that simply keeping different versions of the project file much easier than manually changing them to different SDK & toolchain versions, for example:

https://github.com/gojimmypi/wolfssl/tree/ED25519_SHA2_fix/IDE/Espressif/ESP-IDF/examples/template/VisualGDB

 

 

• This reply was modified 2 years, 6 months ago by gojimmypi.

[gojimmypi]

Hello,

I recently installed the 20231123 ESP32 Debug Methods update to VisualGDB.

The JTAG programming and debugging capabilities (ESP-IDF v5.1) seem to be working well for the ESP32-H2 now! Thank you.

There is however an oddity when pressing the “Debug – Test” button, indicating that the test failed:

Error: Error on socket 'GDB': WSAGetLastError==10054, message: An existing connection was forcibly closed by the remote host.

I’ve attached the full log for reference. So far, it seems the error can be ignored.

Also, note that unlike other devices when doing idf.py erase-flash , the ESP32-H2 is not in a happy state when freshly erased. Error “invalid header” scrolling endlessly. In this state I am unable to JTAG debug.

Here’s an example of the console output after erase & JTAG will not work:

invalid header: 0xffffffff
invalid header: 0xfffff▒ESP-ROM:esp32h2-20221101
Build:Nov 1 2022
rst:0x7 (TG0_WDT_HPSYS),boot:0xc (SPI_FAST_FLASH_BOOT)
Saved PC:0x40011bdc
invalid header: 0xffffffff
invalid header: 0xffffffff

Upon flashing a clean, operational “hello world” (or any other app) – then the JTAG capabilities work fine.

Thank you again for this fix.

Cheers

Attachments:

You must be logged in to view attached files.

[gojimmypi]

There’s a recording of the “Getting Started with wolfSSL on the Espressif ESP32” here:

https://www.youtube.com/watch?v=CzwA3ZBZBZ8

[gojimmypi]

awesome, thank you 🙂

[gojimmypi]

ESP-IDF 5.1 has been released: https://github.com/espressif/esp-idf/releases/tag/v5.1

[gojimmypi]

I’ve setup wolfSSL as a managed component.

I’m hoping to use other components such as the esp-cryptoauthlib.

Components are added to a project like this:

idf.py add-dependency "espressif/esp-cryptoauthlib^3.5.1~1"

It appears the only thing that command does is to add/update the project idf_component.yml file.

Subsequent calls with idf.py build appear to recognize that component file, download required files to the managed_components directory, and compile everything.

However, from within the VisualGDB environment, nothing new seems to happen at compile time: The IDE does not seem to recognize the ESP Registry components in the managed_components directory.

I wasn’t able to find any VisualGDB documentation on this. It this capability supported? I’m using EDP-IDF v5.

Thank you

 

[Author]

Posts