Qt6 on QNX

WIP

This page covers the process of building and deploying Qt6 for QNX 7.1.

QNX SDP

In order to build applications for a QNX target, one needs to download and install the QNX SDP, which contains the QNX toolchain for x86_64, armle-v7 and aarch64le, as well as a collection of tools to aid the creation of QNX system images for select targets and virtual machines. A myQNX account and a valid license are required in order to obtain the QNX SDP. You can create a new account here.

Once an account has been created, you'll need to download the QNX Software Center, from which the QNX SDP can be downloaded.

Installing the SDP

From within QNX Software Center, select the following packages:

• QNX Software Development Platform 7.1 [com.qnx.qnx710] - this may appear as a baseline inside the QNX Software Center
• QNX SDP 7.1 DejaVu Fonts [com.qnx.qnx710.target.screen.fonts.dejavu]
• QNX SDP 7.1 Font Engine [com.qnx.qnx710.target.screen.fonts.engine]
• QNX SDP 7.1 Time Zone Database [com.qnx.qnx710.osr.zoneinfo]
• QNX SDP 7.1 Image Codecs [com.qnx.qnx710.target.screen.img_codecs]

Using the SDP

The QNX SDP ships with a GCC based compiler called QCC. Before this compiler can be used, we need to setup the environment. This is done by sourcing

qnx710/qnxsdp-env.sh

on Linux or invoking

qnx710/qnxsdp-env.bat

on Windows. These scripts set a few important environment variable, including:

• QNX_HOST - the base path for the QNX tools and compiler for the host operating system
• QNX_TARGET - the base path for the source, binary and configuration files that are used (deployed) to the target QNX system

Once the environment has been setup, you can invoke the QNX compiler driver to produce a binary (qcc for C, q++ for C++), specifying the target architecture using the -V command line switch, for instance:

qcc -Vgcc_ntoaarch64le foo foo.c # builds an aarch64le binary
q++ -Vgcc ntox86_64 bar bar.cpp  # builds a x86_64 binary (the default when -V is omitted)

IMPORTANT INFORMATION: QCC does not handle paths containing whitespace well. It's best to avoid them altogether.

Building Qt6/QNX on Linux Hosts

The process to cross-compile Qt for QNX is divided in two parts. First, we need to build and install Qt6 for Linux. This Qt installation is then used to support the build of Qt6 binaries targeting QNX.

Clone and initialize qt5.git

$ git clone git://code.qt.io/qt/qt5.git
$ cd qt5
$ perl init-repository

Create a build directory, configure, build and install

$ mkdir build_host
$ cd build_host
$../configure -prefix $HOME/dev/Qt6_host -nomake tests
$ cmake --build . --parallel
$ cmake --install .

Assuming everything went well, we can now build Qt for QNX. In the example below, we will build for aarch64le targets. You can obtain the relevant toolchain files in

qt5.git/coin/provisioning/common/linux/cmake_toolchain_files/

.

$ mkdir build_qnx
$ cd build_qnx
$ source <path_to_qnx_sdp>/qnx710/qnxsdp-env.sh
$ cmake -DQT_HOST_PATH=$HOME/dev/Qt6_host \ # this is the path of the host Qt we've just built
  -DCMAKE_TOOLCHAIN_FILE=../coin/provisioning/common/linux/cmake_toolchain_files/qnx-toolchain-aarch64le.cmake \
  -G Ninja \
  -DCMAKE_STAGING_PREFIX=$HOME/dev/Qt6_qnx \ # the prefix where Qt will be installed on the host
  -DCMAKE_INSTALL_PREFIX=/qt \ # the prefix where Qt will be installed on the target, used to populate RPATH on binaries built against this Qt version
  ..
$ cmake --build . --parallel
$ cmake --install .

Building Qt6/QNX on Windows Hosts

Building Qt6/QNX on Windows hosts is analogous to the process to build it on Linux hosts, except that a little more is required to prepare the environment.

Setting up the environment

Apart from Visual Studio, it's important that the other dependencies are installed to paths that do not contain spaces. As mentioned before, the QNX tools have issues dealing with paths containing whitespace (e.g. it doesn't know how to properly escape the path strings when passing then to cc1plus). The safest way is to simply install Perl, Python, Ninja and CMake directly under C:\

1. Install Microsoft Visual Studio 2022 (Community Edition will do just fine)
2. Install Strawberry Perl - DO NOT AUTOMATICALLY ADD IT TO THE SYSTEM PATH
3. Install Python for Windows - DO NOT AUTOMATICALLY ADD IT TO THE SYSTEM PATH
4. Install CMake >= 3.22 - it's recommended to simply use the Windows x64 ZIP to prevent the installer from messing with the system, also DO NOT AUTOMATICALLY ADD IT TO THE SYSTEM PATH
5. Install Ninja