-
Notifications
You must be signed in to change notification settings - Fork 6.8k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Build fails with gcc-arm-none-eabi-9-2019-q4-major #22060
Comments
Could not reproduce. My env:
v2.0.1, is that supposed to be a Zephyr revision? I can't find a git tag by that name. |
Please post full build logs. |
sorry my bad v2.1.0 (latest release) |
Also, try copying the cache directory to a different directory in case the cache directory is corrupted. CMake spits out in the logs where the cache directory is on every run. EDIT: Sorry, the cache directory should not be copied, it should be moved (to clear it without losing it permanently). |
Could not reproduce on v2.0.1 either. |
Closing as could-not-reproduce. Feel free to re-open after further investigation is done. |
Worked on another device so probably was a cmake cache problem sorry for the false alert |
I'm facing this issue with this toolchain version on Linux. I actually see that I downloaded tarball some time last year, but didn't upgrade our CI, quite possible because I saw this or similar issue. We're now forced to upgrade due to #25804 (comment) . The situation doesn't seem right - arm-none-eabi-gcc, when passed no options, is unable to link final executable (note that source file, CMakeCCompilerId.c, doesn't have explicit invocation of exit(), which then leads to error of missing _exit()). But still, the question why CMake is doing all this, why CMake tries to invoke compiler in random ways, e.g. Zephyr isn't going to invoke compiler without options, it's going to invoke it with very specific options regarding include paths and libraries, and chances are that with them, everything will work. @SebastianBoe, @tejlmand, can CMake work in "hands off" approach, where it just does what user tells (e.g., use a toolchain at specific path), instead of trying to outsmart a user and find "errors"? What's the option to tell CMake to do that now? @galak: FYI |
…019-q4-major Locally, this toolchain doesn't work for me due to zephyrproject-rtos/zephyr#22060 . Trying to upgrade CI, as there's report in that issue that "it worked on another machine". We need upgrade due to zephyrproject-rtos/zephyr#25804 (comment) Change-Id: I26477ed4705d6916f244465405309776eb06f98e Signed-off-by: Paul Sokolovsky <paul.sokolovsky@linaro.org>
So, despite failure on my system, I decided to upgrade our CI anyway, given that original submitter commented "Worked on another device". And weird as it is, the build went thru: https://ci.linaro.org/view/lite-iot-ci/job/zephyr-upstream/5258/ . So, I don't know, let me close this again. |
I'm facing the same issue
CMakeError.log
|
Reinstalling the tool chain fixed the issue. |
I'm having the same issue. It also appears to come up in https://devzone.nordicsemi.com/f/nordic-q-a/55475/the-toolchain-is-unable-to-build-a-dummy-c-file/225042 and platformio/platformio-core#3389. Based on this thread and the linked threads, I've tried:
Environment:
CMakeError.log
|
@cf1331 This looks suspicious to me:
Mixing of path separators is usually a bad sign. |
Seeing the same issue when running from within PlatformIO@Windows. Fixed by |
I am facing the same issue as indicated by @cf1331 but with ubuntu 14.04 environment. Reinstalling the tool chain didn't solved the problem. |
I've encountered this issue when I use the latest version of CMake (v3.19.0), once I downgrade CMake to v3.16.2, the build process is finished without any errors. BTW, I work on windows 10. |
After downgrading CMake, build finishes for me as well. But... While I get an elf file, the CMakeError.log file still complains about _exit. I have not yet tried to flash to a board. I'm on arc linux |
Can reproduce on Linux by deleting build folder and The third point seems relevant: After downgrading CMake to 3.18.5 the error is resolved. |
I had a similar issue, what worked for me was installing CMake release version 3.17.3 from https://cmake.org/files/ and then created a symlink to the executable. What did not work was |
I'm having the same issue. in window10.when I delete the contents of the "zephyr\.cache\ToolchainCapabilityDatabase" folder, the problem is solved |
This is something that should be improved, but without sacrifying the benefits of the toolchain capabilities database. |
cmake 3.19 and 3.19.1 currently don't work with Zephyr. This appears to be a regression with upstream CMake and I've opened the following issue to track things: |
Like last comment, this worked for me on Windows 10 (Powershell):
|
Removing the zephyr cache before west build fixed the probem |
Describe the bug
There is a Toolchain issue with gnuarmemb 9.2 (current version) under Windows
To Reproduce
Steps to reproduce the behaviour:
Just follow getting started and 3rd party toolchain
I installed to
c:/gnuarmemb
but I don't think this should matterExpected behavior
west -b nucleo_f401re samples\hello_world
should work but I get error (see below)Impact
Showstopper for new developers should be easy fixable for people with knowledge to cross compile on windows
Screenshots or console output
Environment (please complete the following information):
The text was updated successfully, but these errors were encountered: