We are pleased to announce the release of GCC Toolchain for Renesas RX version 14.2.0.202607.

This GCC-based cross-development toolchain supports C and C++ application development for all Renesas RX devices and generates ELF-format object files. Installers are available for Ubuntu 22.04 or later and compatible Linux distributions, Windows 7 or later, and macOS Ventura 13.4 running on Apple M1.

The latest version of the toolchain is available for download here.

GCC Toolchain for Renesas RX 14.2.0.202607 is based on the following components:

– GCC 14.2.0
– GNU Binutils 2.44
– Newlib 4.4.0
– GDB 16.2

The latest applicable patches have also been incorporated into the toolchain sources.

Key improvements and fixes in this release include:

– Renesas RXv3e Protect Zone support

Support for the RXv3e Protect Zone has been added through the new -mprotect-zone compiler option. Developers working with this feature should refer to the GCC documentation supplied with the toolchain and the relevant RXv3e software and hardware manuals for configuration details.

– New TFU security attribute option

The new -mtfu-security-attribute=[secure/nonsecure] command-line option allows developers to select the memory-region address to be used as the TFU base address.

– Updated interaction between rx-elf-libgen and MTFU options

MTFU-related options are now filtered from the rx-elf-libgen --compiler-options command line by the associated makefiles. This change helps prevent potential address inconsistencies following the introduction of Protect Zone support.

– Corrected -mtfu=intrinsic,mathlib behavior in C++

When compiling C++ applications, the -mtfu=intrinsic,mathlib option did not correctly replace calls to sinf, cosf, atan2f, and hypotf with their corresponding built-in implementations. This issue has been resolved.

– Fixed a compiler failure involving certain 64-bit subtraction operations

Certain 64-bit subtraction expressions could prevent the compiler from selecting a valid instruction and cause the compiler to terminate unexpectedly. This regression affected the 202505 and 202511 toolchain releases and has been corrected in version 14.2.0.202607.

Developers upgrading existing projects should also review the following compatibility considerations:

– -Wreturn-type is enabled by default. Warnings generated for non-void functions that can reach the end without returning a value should be addressed, as such code may behave incorrectly when optimized.
– G++ applies stricter validation to malformed template definitions, including templates that cannot be valid for any possible instantiation.
– The C++ alignof operator now reports the minimum alignment required by the target ABI rather than the compiler’s preferred alignment. Code requiring the preferred alignment can use __alignof__.
– In C++17 mode, comparison objects used by associative containers such as std::map, std::multimap, std::set, and std::multiset must be callable when const-qualified.
– The Optlib library has been removed because it did not provide complete ANSI/ISO headers and definitions, implemented only part of the standard library functionality, and included mathematical functions that traded precision for speed or code size and were not IEEE 754 compliant.

For e² studio users, the installer does not provide a separate integration option because e² studio can automatically detect an installed GNURX toolchain when it starts. The toolchain can also be configured through the Toolchain Management feature in e² studio. Integration with the HEW IDE is not supported by this installer.

For complete technical information, usage details, and known issues, please refer to the release notes, available here.

We welcome your feedback on this release. Please share your comments, questions, or experiences with the new toolchain in the comments below.