Skip to content

Latest commit

 

History

History
47 lines (27 loc) · 2.37 KB

KNOWN_ISSUES.md

File metadata and controls

47 lines (27 loc) · 2.37 KB
Raw

IIC-OSIC-TOOLS Known Issues

Container

Starting the Container

If you run into problems when starting the container (mostly in combination with SELinux), try to run the container with the following environment variable set to DOCKER_EXTRA_PARAMS='--security-opt seccomp=unconfined'. This will overwrite the Docker Default security settings so use with it care!

Example:

DOCKER_EXTRA_PARAMS='--security-opt seccomp=unconfined' ./start_x.sh

Issues with OpenGL on some environments

A few applications are using OpenGL graphics, which can lead to issues on some computing environments. A (potential) remedy is to enable SW-rendering with can be achieved by setting the following environment variable inside the Docker VM:

LIBGL_ALWAYS_INDIRECT=0

The OpenROAD Flow Scripts (ORFS)

The ORFS required a recent version of openroad. Since image tag 2024.12 a recent version is installed alongside the OpenROAD version required by openlane. In order to use the ORFS, before calling the make script make sure to set the following env vars:

export YOSYS_EXE=$TOOLS/yosys/bin/yosys
export OPENROAD_EXE=$TOOLS/openroad-latest/bin/openroad
export OPENSTA_EXE=$TOOLS/openroad-latest/bin/sta

OpenEMS

The visualization tool "AppCSXCAD" will not work in the container with our default settings (vtkXOpenGLRenderWindow (0x....): Cannot create GLX context. Aborting.). The issue has been located to be connected with the environment variable "LIBGL_ALWAYS_INDIRECT". As a workaround, we suggest either unsetting the variable or setting it to 0 (unset LIBGL_ALWAYS_INDIRECT or export LIBGL_ALWAYS_INDIRECT=0) which is persistent for the running terminal or run AppCSXCAD with the variable set to zero inline: LIBGL_ALWAYS_INDIRECT=0 AppCSXCAD.

Hdl21/Vlsirtools

We (temporarily) removed Hdl21 and Vlsirtools, as they force numpy to version 1. All other packages allow numpy 2, so this removal, until the dependencies are fixed.

PyOPUS

PyOPUS is removed, as build fails, and it forces numpy to version 1.

Build

Boost

Boost is currently installed from the package sources of Ubuntu and a manual install/build. This is currently required, as there are some dependencies from packages, but also, some manually built tools require a newer boost version. This issue will be resolved in the future when switching to a more modern Ubuntu release.