Skip to content

Conversation

@jhol
Copy link
Contributor

@jhol jhol commented Feb 18, 2025

No description provided.

@jhol
Copy link
Contributor Author

jhol commented Feb 18, 2025

A proposed architecture port for OpenRISC 1000 (or1k) is being discussed in this Pull Request: zephyrproject-rtos/zephyr#83933

This PR adds the requisite tools to the tool-chain.

@keith-packard
Copy link
Contributor

Yup, picolibc has upstream or1k support in 1.8.9

@jhol
Copy link
Contributor Author

jhol commented Feb 18, 2025

v2 of the patch-set removes the line-break added to QEMUS_BUILT which was not parsed properly.

@jhol
Copy link
Contributor Author

jhol commented Oct 23, 2025

Just rebased on main.

Copy link
Member

@stephanosio stephanosio left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I have created topic-or1k branch for testing since the CI workflow changes do not apply in PRs.

Here is the result of a CI run:
https://github.com/zephyrproject-rtos/sdk-ng/actions/runs/18764695298/job/53537210188

There is an issue building GCC:

[ALL  ]    Error calling /__w/_temp/workspace/build/.build/or1k-zephyr-elf/src/gcc/gcc/genmultilib: No dirname for option: msoft-mul
[ERROR]    make[1]: *** [Makefile:2359: s-mlib] Error 1
[ERROR]    make[1]: *** Waiting for unfinished jobs....

See log_toolchain_gnu_linux-x86_64_or1k-zephyr-elf at the bottom of the run summary page for the full log.

@stephanosio stephanosio changed the base branch from main to topic-or1k October 23, 2025 23:50
@stephanosio
Copy link
Member

I have changed the base branch of this PR from main to topic-or1k with the CI workflow changes in this PR so that the CI workflow actually builds the or1k toolchain.

Signed-off-by: Joel Holdsworth <jholdsworth@nvidia.com>
@jhol
Copy link
Contributor Author

jhol commented Oct 25, 2025

I have changed the base branch of this PR from main to topic-or1k with the CI workflow changes in this PR so that the CI workflow actually builds the or1k toolchain.

@stephanosio Thanks. This will be very helpful.

To fix the issue, I have been running the build locally in the crosstool-ng Ubuntu 22.04 docker image.

The solution is to disable CT_MULTILIB and friends, and enable CT_DEMULTILIB which is the setting provided by the default ct-ng template for or1k-unknown-elf.

Because OpenRISC has no need to support multiple floating-point or ABI variants at present (no FPU/hard-float vs. soft-float distinction), a demultilib build is the only valid configuration. The or1k GCC configuration assumes a single ABI and option set; using multilib is redundant and unsupported.

@stephanosio stephanosio changed the base branch from topic-or1k to main October 25, 2025 10:11
@stephanosio stephanosio merged commit 07df5a6 into zephyrproject-rtos:main Oct 25, 2025
44 of 47 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants