```{warning} These are in-progress notes for the upcoming LLVM {{env.config.release}} release. Release notes for previous releases can be found on [the Download Page](https://releases.llvm.org/download.html). ```
This document contains the release notes for the LLVM Compiler Infrastructure, release {{env.config.release}}. 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.
Removed llvm.convert.to.fp16 and llvm.convert.from.fp16 intrinsics. These are equivalent to fptrunc and fpext with half with a bitcast.
“denormal-fp-math” and “denormal-fp-math-f32” string attributes were migrated to first-class denormal_fpenv attribute.
The "nooutline" attribute is now writen as nooutline. Existing IR and bitcode will be automatically updated.
Removed Constant::isZeroValue. It was functionally identical to Constant::isNullValue for all types except floating-point negative zero. All callers should use isNullValue instead. isZeroValue will be reintroduced in the future with bitwise-all-zeros semantics to support non-zero null pointers.
Removed TypePromoteFloat legalization from SelectionDAG
Removed bugpoint. Usage has been replaced by llvm-reduce and llvm/utils/reduce_pipeline.py.
The Br opcode was split into two opcodes separating unconditional (UncondBr) and conditional (CondBr) branches.
BranchInst was deprecated in favor of UncondBrInst and CondBrInst.
The operand order of CondBr instructions was adjusted to match the successor order. This can cause subtle breakage when using getOperand or setOperand to access successors.
The llvm::sys::fs link creation API has been refactored:
create_link now tries to create a symbolic link first, falling back to a hard link if that fails (previously it created a symlink on Unix and a hard link on Windows).create_symlink, which always creates a symbolic link. On windows this may fail if symlink permissions are not available.readlink, which reads the target of a symbolic link.ID{n-m} instead of ID<n-m> to be consistent with inner let statements.sysp, mrrs, and msrr instructions are now accepted without requiring the +d128 feature gating.r14 register can now be used as an alias for the link register lr in inline assembly. Clang always canonicalizes the name to lr, but other frontends may not.-gsplit-dwarf and -mrelax to be used together when building for the LoongArch platform.sm_30 to sm_75. sm_75 is the oldest GPU variant compatible with the widest range of recent major CUDA Toolkit versions (11/12/13).llvm-objdump now has support for --symbolize-operands with RISC-V.-mcpu=spacemit-x100 was added.lpad, pause, ntl.*, c.ntl.*, sspush, sspopchk, ssrdp, c.sspush, c.sspopchk) are now always available in the assembler and disassembler without requiring their respective extensions.-mcpu=spacemit-a100 was added.-riscv-enable-p-ext-simd-codegen flag has been removed. P extension SIMD code generation is now enabled automatically if the P extension is supported.-mcpu=xt-c910v2 and -mcpu=xt-c920v2 were added.-mcpu=sifive-x160 and -mcpu=sifive-x180 were added.XRivosVisni vendor extension has been removed..seh_startchained and .seh_endchained assembly instructions have been removed and replaced with a new .seh_splitchained instruction..att_syntax directive is now emitted for assembly files when AT&T syntax is in use. This matches the behaviour of Intel syntax and aids with compatibility when changing the default Clang syntax to the Intel syntax.Replaced opcode LLVMBr with LLVMUncondBr and LLVMCondBr.
The operand order of CondBr instructions was adjusted to match the successor order. This can cause subtle breakage when using LLVMGetOperand or LLVMSetOperand to access successors.
llvm-objcopy no longer corrupts the symbol table when --update-section is called for ELF files.FileCheck option -check-prefix now accepts a comma-separated list of prefixes, making it an alias of the existing -check-prefixes option.-mtune option to llc.SBTarget::GetDataByteSize(), SBTarget::GetCodeByteSize(), and SBSection::GetTargetByteSize() have been deprecated. They always return 1, as before.freebsd-kernel to freebsd-kernel-core. Remote kernel debugging is still handled by the gdb-remote plugin.plugin.process.freebsd-kernel-core.read-only must be set to false. This setting is available when using /dev/mem or a kernel dump. However, since kvm_write() does not support writing to kernel dumps, writes to a kernel dump will still fail when the setting is false.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.