LLVM 21.0.0git Release Notes

Warning

These are in-progress notes for the upcoming LLVM 21.0.0git release. Release notes for previous releases can be found on the Download Page.

Introduction

This document contains the release notes for the LLVM Compiler Infrastructure, release 21.0.0git. Here we describe the status of LLVM, including major improvements from the previous release, improvements in various subprojects of LLVM, and some of the current users of the code. All LLVM releases may be downloaded from the LLVM releases web site.

For more information about LLVM, including information about the latest release, please check out the main LLVM web site. If you have questions or comments, the Discourse forums is a good place to ask them.

Note that if you are reading this file from a Git checkout or the main LLVM web page, this document applies to the next release, not the current one. To see the release notes for a specific release, please see the releases page.

Non-comprehensive list of changes in this release

  • …

Changes to the LLVM IR

  • The nocapture attribute has been replaced by captures(none).

  • The constant expression variants of the following instructions have been removed:

    • mul

  • Updated semantics of llvm.type.checked.load.relative to match that of llvm.load.relative.

Changes to LLVM infrastructure

  • Removed support for target intrinsics being defined in the target directories themselves (i.e., the TargetIntrinsicInfo class).

  • Fix Microsoft demangling of string literals to be stricter (#GH129970))

Changes to building LLVM

Changes to TableGen

Changes to Interprocedural Optimizations

Changes to the AArch64 Backend

  • Added the execute-only target feature, which indicates that the generated program code doesn’t contain any inline data, and there are no data accesses to code sections. On ELF targets this property is indicated by the SHF_AARCH64_PURECODE section flag. (#125687, #132196, #133084)

Changes to the AMDGPU Backend

  • Enabled the FWD_PROGRESS bit for all GFX ISAs greater or equal to 10, for the AMDHSA OS.

  • Bump the default .amdhsa_code_object_version to 6. ROCm 6.3 is required to run any program compiled with COV6.

Changes to the ARM Backend

Changes to the AVR Backend

Changes to the DirectX Backend

Changes to the Hexagon Backend

  • The default Hexagon architecture version in ELF object files produced by the tools such as llvm-mc is changed to v68. This version will be set if the user does not provide the CPU version in the command line.

Changes to the LoongArch Backend

  • Changing the default code model from small to medium for 64-bit.

Changes to the MIPS Backend

  • -mcpu=i6400 and -mcpu=i6500 were added.

Changes to the PowerPC Backend

Changes to the RISC-V Backend

  • Adds experimental assembler support for the Qualcomm uC ‘Xqcilb` (Long Branch) extension.

  • Adds experimental assembler support for the Qualcomm uC ‘Xqcili` (Load Large Immediate) extension.

  • Adds experimental assembler support for the Qualcomm uC ‘Xqcilia` (Large Immediate Arithmetic) extension.

  • Adds experimental assembler support for the Qualcomm uC ‘Xqcibm` (Bit Manipulation) extension.

  • Adds experimental assembler support for the Qualcomm uC ‘Xqcibi` (Branch Immediate) extension.

  • Adds experimental assembler and code generation support for the Qualcomm ‘Xqccmp’ extension, which is a frame-pointer convention compatible version of Zcmp.

  • Added non-quadratic log-vrgather cost model for vrgather.vv instruction

  • Adds experimental assembler support for the Qualcomm uC ‘Xqcisim` (Simulation Hint) extension.

  • Adds experimental assembler support for the Qualcomm uC ‘Xqcisync` (Sync Delay) extension.

  • Adds experimental assembler support for the Qualcomm uC ‘Xqciio` (External Input Output) extension.

  • Adds assembler support for the ‘Zilsd` (Load/Store Pair Instructions) extension.

  • Adds assembler support for the ‘Zclsd` (Compressed Load/Store Pair Instructions) extension.

  • Adds experimental assembler support for Zvqdotq.

  • Adds Support for Qualcomm’s qci-nest and qci-nonest interrupt types, which use instructions from Xqciint to save and restore some GPRs during interrupt handlers.

  • When the experimental extension Xqcili is enabled, qc.e.li and qc.li may now be used to materialize immediates.

  • Adds assembler support for .option exact, which disables automatic compression, and branch and linker relaxation. This can be disabled with .option noexact, which is also the default.

  • -mcpu=xiangshan-kunminghu was added.

  • -mcpu=andes-n45 and -mcpu=andes-nx45 were added.

  • -mcpu=andes-a45 and -mcpu=andes-ax45 were added.

  • Adds support for the ‘Ziccamoc` (Main Memory Supports Atomics in Zacas) extension, which was introduced as an optional extension of the RISC-V Profiles specification.

  • Adds experimental assembler support for SiFive CLIC CSRs, under the names Zsfmclic for the M-mode registers and Zsfsclic for the S-mode registers.

  • Adds Support for SiFive CLIC interrupt attributes, which automate writing CLIC interrupt handlers without using inline assembly.

Changes to the WebAssembly Backend

Changes to the Windows Target

  • fp128 is now passed indirectly, meaning it uses the same calling convention as i128.

Changes to the X86 Backend

Changes to the OCaml bindings

Changes to the Python bindings

Changes to the C API

  • The following functions for creating constant expressions have been removed, because the underlying constant expressions are no longer supported. Instead, an instruction should be created using the LLVMBuildXYZ APIs, which will constant fold the operands if possible and create an instruction otherwise:

    • LLVMConstMul

    • LLVMConstNUWMul

    • LLVMConstNSWMul

  • Added LLVMConstDataArray and LLVMGetRawDataValues to allow creating and reading ConstantDataArray values without needing extra LLVMValueRefs for individual elements.

Changes to the CodeGen infrastructure

Changes to the Metadata Info

Changes to the Debug Info

Changes to the LLVM tools

  • llvm-objcopy now supports the --update-section flag for intermediate Mach-O object files.

  • llvm-strip now supports continuing to process files on encountering an error.

  • In llvm-objcopy/llvm-strip’s ELF port, --discard-locals and --discard-all now allow and preserve symbols referenced by relocations. (#47468)

Changes to LLDB

  • When building LLDB with Python support, the minimum version of Python is now 3.8.

  • LLDB now supports hardware watchpoints for AArch64 Windows targets. Windows does not provide API to query the number of supported hardware watchpoints. Therefore current implementation allows only 1 watchpoint, as tested with Windows 11 on the Microsoft SQ2 and Snapdragon Elite X platforms.

  • LLDB now steps through C++ thunks. This fixes an issue where previously, it wouldn’t step into multiple inheritance virtual functions.

  • A statusline was added to command-line LLDB to show progress events and information about the current state of the debugger at the bottom of the terminal. This is on by default and can be configured using the show-statusline and statusline-format settings.

  • The min-gdbserver-port and max-gdbserver-port options have been removed from lldb-server’s platform mode. Since the changes to lldb-server’s port handling in LLDB 20, these options have had no effect.

  • LLDB now supports process continue --reverse when used with debug servers supporting reverse execution, such as rr. When using reverse execution, process continue --forward returns to the forward execution.

Changes to lldb-dap

  • Breakpoints can now be set for specific columns within a line.

  • Function return value is now displayed on step-out.

Changes to BOLT

Changes to Sanitizers

Other Changes

External Open Source Projects Using LLVM 21.0.0git

  • A project…

Additional Information

A wide variety of additional information is available on the LLVM web page, in particular in the documentation section. The web page also contains versions of the API documentation which is up-to-date with the Git version of the source code. You can access versions of these documents specific to this release by going into the llvm/docs/ directory in the LLVM tree.

If you have any questions or comments about LLVM, please feel free to contact us via the Discourse forums.