LLVM/project 5f991d0compiler-rt/lib/asan asan_errors.cpp asan_report.cpp, compiler-rt/test/asan/TestCases assume_dereferenceable.cpp assume_dereferenceable_pass.cpp

Revert "[compiler-rt][asan] Add asan checks for __builtin_assume_dereferencab…"

This reverts commit bf8cf4b7b31b956ea4dfec93b0b4d77b359b188b.
DeltaFile
+0-87compiler-rt/test/asan/TestCases/assume_dereferenceable.cpp
+0-42llvm/lib/Transforms/Instrumentation/AddressSanitizer.cpp
+8-30compiler-rt/lib/asan/asan_errors.cpp
+0-35compiler-rt/test/asan/TestCases/assume_dereferenceable_pass.cpp
+0-22compiler-rt/test/asan/TestCases/assume_dereferenceable_halt_on_error.cpp
+2-19compiler-rt/lib/asan/asan_report.cpp
+10-2359 files not shown
+18-31315 files

LLVM/project 61a47e2libcxx/cmake/caches Generic-llvm-libc.cmake, libcxx/test/configs llvm-libc++-llvm-libc.cfg.in

[𝘀𝗽𝗿] initial version

Created using spr 1.3.7
DeltaFile
+62-0libcxx/cmake/caches/Generic-llvm-libc.cmake
+27-0libcxxabi/test/configs/llvm-libc++abi-llvm-libc.cfg.in
+27-0libcxx/test/configs/llvm-libc++-llvm-libc.cfg.in
+20-0libcxx/utils/ci/run-buildbot
+4-0libcxx/utils/libcxx/test/features/platform.py
+3-0libcxx/test/std/numerics/complex.number/complex.transcendentals/tanh.pass.cpp
+143-076 files not shown
+364-182 files

LLVM/project 8887969libcxx/cmake/caches Generic-llvm-libc.cmake, libcxx/test/configs llvm-libc++-llvm-libc.cfg.in

[𝘀𝗽𝗿] changes to main this commit is based on

Created using spr 1.3.7

[skip ci]
DeltaFile
+62-0libcxx/cmake/caches/Generic-llvm-libc.cmake
+27-0libcxx/test/configs/llvm-libc++-llvm-libc.cfg.in
+27-0libcxxabi/test/configs/llvm-libc++abi-llvm-libc.cfg.in
+20-0libcxx/utils/ci/run-buildbot
+4-0libcxx/utils/libcxx/test/features/platform.py
+3-0libcxx/test/std/containers/container.adaptors/container.adaptors.format/format.functions.format.pass.cpp
+143-075 files not shown
+364-081 files

LLVM/project abb4ff5llvm/lib/Transforms/IPO LowerTypeTests.cpp, llvm/test/Transforms/LowerTypeTests x86-jumptable-dbg.ll aarch64-jumptable-dbg.ll

[LowerTypeTests] Add debug info to jump table entries (#192736)

When Control Flow Integrity (CFI) is enabled, jump tables are used to
redirect indirect calls. Previously, these jump table entries lacked
debug information, making it difficult for profilers and debuggers to
attribute execution time correctly.

Now stack trace, when stopped on jump table entry will looks like this:
```
#0: __ubsan_check_cfi_icall_jt at sanitizer/ubsan_interface.h:0
#1: c::c() (.cfi_jt) at sanitizer/ubsan_interface.h:0:0
#2: .cfi.jumptable.81 at sanitizer/ubsan_interface.h:0:0
```
DeltaFile
+67-4llvm/lib/Transforms/IPO/LowerTypeTests.cpp
+32-8llvm/test/Transforms/LowerTypeTests/x86-jumptable-dbg.ll
+16-4llvm/test/Transforms/LowerTypeTests/aarch64-jumptable-dbg.ll
+115-163 files

LLVM/project c74951cllvm/lib/Passes PassBuilderPipelines.cpp, llvm/test/Other new-pm-defaults.ll new-pm-thinlto-postlink-pgo-defaults.ll

Revert "Reapply "[JTS][Passes] Enable JTS By Default" (#193409)" (#193649)

Breaks asan buildbots
https://lab.llvm.org/buildbot/#/builders/52/builds/16717
This reverts commit bb762095af80273ed2db7509ebc2cae4613ab787 from
#193409.

Symbolized report
https://gist.github.com/vitalybuka/f60465c13ffad56b8870e87b0fc443ff
Probably false positive caused by pass, but can't tell for sure.
DeltaFile
+5-1llvm/test/Other/new-pm-defaults.ll
+2-2llvm/lib/Passes/PassBuilderPipelines.cpp
+1-1llvm/test/Other/new-pm-thinlto-postlink-pgo-defaults.ll
+1-1llvm/test/Other/new-pm-thinlto-postlink-samplepgo-defaults.ll
+1-1llvm/test/Other/new-pm-thinlto-prelink-defaults.ll
+1-1llvm/test/Other/new-pm-thinlto-prelink-samplepgo-defaults.ll
+11-72 files not shown
+13-98 files

LLVM/project ea97a6bllvm/docs AMDGPUUsage.rst, llvm/include/llvm/IR IntrinsicsAMDGPU.td

[AMDGPU] Add a sched group mask for LDSDMA instructions

The existing VMEM masks are not fine-grained enough for some use cases. For
example, if users want to control async loads, using VMEM may cause the compiler
to pick instructions it shouldn't.

This PR adds a new sched group mask for LDSDMA instructions. It is a subclass of
VMEM, but only targets isLDSDMA instructions.
DeltaFile
+342-0llvm/test/CodeGen/AMDGPU/sched-ldsdma-mask.mir
+29-20llvm/test/CodeGen/AMDGPU/sched.barrier.inverted.mask.ll
+20-10llvm/lib/Target/AMDGPU/AMDGPUIGroupLP.cpp
+2-0llvm/include/llvm/IR/IntrinsicsAMDGPU.td
+1-0llvm/docs/AMDGPUUsage.rst
+394-305 files

LLVM/project 4623238clang/test/CodeGen/SystemZ builtins-systemz-zvector.c, llvm/test/CodeGen/AMDGPU load-constant-i8.ll load-global-i8.ll

no llvm::

Created using spr 1.3.7
DeltaFile
+1,117-1,613llvm/test/CodeGen/RISCV/rvv/setcc-fp-vp.ll
+918-920llvm/test/CodeGen/AMDGPU/load-constant-i8.ll
+781-1,020llvm/test/CodeGen/X86/ssub_sat_vec.ll
+897-867clang/test/CodeGen/SystemZ/builtins-systemz-zvector.c
+785-891llvm/test/CodeGen/AMDGPU/GlobalISel/ssubsat.ll
+820-816llvm/test/CodeGen/AMDGPU/load-global-i8.ll
+5,318-6,1271,101 files not shown
+41,155-27,5371,107 files

LLVM/project 2855525lld/ELF ScriptParser.cpp ScriptLexer.cpp, lld/test/ELF/linkerscript include-mid-construct.s memory-include.test

[ELF] Handle INCLUDE like a call stack (#193427)

The lexer maintains a stack of buffers, which allows a construct
started in an INCLUDE'd file to be closed by the parent. This produces
spurious acceptance of malformed scripts (e.g. a bare assignment with
no trailing `;` in the include, terminated by the parent's `;` after
`INCLUDE`) and undefined-behavior span computations in
`readAssignment`'s `commandString` (issue #190376).

Force each INCLUDE to fully parse its own content, similar to a call
stack frame. `ScriptLexer::lex` no longer auto-pops on EOF; the
`buffers` member is gone. `readInclude` takes a `function_ref<void()>`
callback, and the four call sites (top-level, SECTIONS, output
section, MEMORY) pass a context-appropriate parser.

With this, each buffer contains complete parser structures by
construction, so the `[oldS, curTok)` pointer range in
`readAssignment` no longer needs a guard.
DeltaFile
+40-23lld/ELF/ScriptParser.cpp
+7-28lld/ELF/ScriptLexer.cpp
+18-0lld/test/ELF/linkerscript/include-mid-construct.s
+17-0lld/test/ELF/linkerscript/memory-include.test
+17-0lld/test/ELF/linkerscript/section-include.test
+9-0lld/test/ELF/linkerscript/output-section-include.test
+108-511 files not shown
+110-577 files

LLVM/project 96bc719flang/docs/MeetingNotes/2026 2026-04-22.md 2026-04-08.md

[flang] Add Flang Community Call notes for 4/22/2026 (#193575)

Add Flang Community Call notes for 4/22/2026 and make one correction the
4/8/2026 notes
DeltaFile
+94-0flang/docs/MeetingNotes/2026/2026-04-22.md
+1-1flang/docs/MeetingNotes/2026/2026-04-08.md
+95-12 files

LLVM/project 50916c4clang/lib/CIR/CodeGen CIRGenBuiltinRISCV.cpp, clang/test/CIR/CodeGenBuiltins/RISCV riscv-zbb.c

[CIR][RISCV] Support zbb builitin codegen (#188932)

Include 2 builtins: __builtin_riscv_orc_b_32 and
__builtin_riscv_orc_b_64.
DeltaFile
+40-0clang/test/CIR/CodeGenBuiltins/RISCV/riscv-zbb.c
+6-2clang/lib/CIR/CodeGen/CIRGenBuiltinRISCV.cpp
+46-22 files

LLVM/project d7d35e4llvm/lib/Passes PassBuilderPipelines.cpp, llvm/test/Other new-pm-defaults.ll new-pm-thinlto-prelink-defaults.ll

[𝘀𝗽𝗿] initial version

Created using spr 1.3.7
DeltaFile
+5-1llvm/test/Other/new-pm-defaults.ll
+2-2llvm/lib/Passes/PassBuilderPipelines.cpp
+1-1llvm/test/Other/new-pm-thinlto-prelink-defaults.ll
+1-1llvm/test/Other/new-pm-thinlto-prelink-pgo-defaults.ll
+1-1llvm/test/Other/new-pm-thinlto-prelink-samplepgo-defaults.ll
+1-1llvm/test/Other/new-pm-thinlto-postlink-defaults.ll
+11-72 files not shown
+13-98 files

LLVM/project b2ae992llvm/lib/Target/RISCV RISCVInstrInfoZvzip.td RISCVISelLowering.cpp, llvm/test/CodeGen/RISCV/rvv fixed-vectors-shuffle-zipeven-zipodd.ll fixed-vectors-shuffle-deinterleave2.ll

[RISCV][CodeGen] Add initial CodeGen support of vpair{e,o} (#192918)

Add initial support for vpair{e,o} instructions, which are included in
zvzip extension.
Doc:

https://github.com/ved-rivos/riscv-isa-manual/blob/zvzip/src/zvzip.adoc

https://github.com/riscv/riscv-opcodes/blob/master/extensions/unratified/rv_zvzip.
DeltaFile
+50-68llvm/test/CodeGen/RISCV/rvv/fixed-vectors-shuffle-zipeven-zipodd.ll
+18-24llvm/test/CodeGen/RISCV/rvv/fixed-vectors-shuffle-deinterleave2.ll
+20-0llvm/lib/Target/RISCV/RISCVInstrInfoZvzip.td
+11-6llvm/lib/Target/RISCV/RISCVISelLowering.cpp
+4-8llvm/test/CodeGen/RISCV/rvv/vector-deinterleave-fixed.ll
+6-2llvm/lib/Target/RISCV/RISCVInstrInfoVVLPatterns.td
+109-1086 files

LLVM/project 9f6ab04libcxx/test/std/depr/depr.c.headers math_h.pass.cpp signal_h.compile.pass.cpp, libcxx/test/std/diagnostics/syserr/syserr.errcat/syserr.errcat.objects system_category.pass.cpp

update

Created using spr 1.3.7
DeltaFile
+4-0libcxx/utils/libcxx/test/features/platform.py
+3-0libcxx/test/std/depr/depr.c.headers/math_h.pass.cpp
+3-0libcxx/test/std/depr/depr.c.headers/signal_h.compile.pass.cpp
+3-0libcxx/test/std/depr/depr.c.headers/stdio_h.compile.pass.cpp
+3-0libcxx/test/std/depr/depr.c.headers/stdlib_h.pass.cpp
+3-0libcxx/test/std/diagnostics/syserr/syserr.errcat/syserr.errcat.objects/system_category.pass.cpp
+19-071 files not shown
+226-177 files

LLVM/project 93bed18libcxx/test/libcxx/fuzzing random.pass.cpp, libcxx/test/libcxx/language.support/support.dynamic assert.nothrow_new_not_overridden_fno_exceptions.pass.cpp new_dont_return_nullptr.pass.cpp

[𝘀𝗽𝗿] changes introduced through rebase

Created using spr 1.3.7

[skip ci]
DeltaFile
+4-0libcxx/utils/libcxx/test/features/platform.py
+3-0libcxx/test/std/utilities/format/format.range/format.range.fmtset/format.functions.format.pass.cpp
+3-0libcxx/test/libcxx/fuzzing/random.pass.cpp
+3-0libcxx/test/libcxx/language.support/support.dynamic/assert.nothrow_new_not_overridden_fno_exceptions.pass.cpp
+3-0libcxx/test/libcxx/language.support/support.dynamic/new_dont_return_nullptr.pass.cpp
+3-0libcxx/test/libcxx/numerics/c.math/constexpr-cxx23-clang.pass.cpp
+19-071 files not shown
+226-177 files

LLVM/project e288c2clibcxx/cmake/caches Generic-llvm-libc.cmake, libcxx/test/configs llvm-libc++-llvm-libc.cfg.in

[𝘀𝗽𝗿] initial version

Created using spr 1.3.7
DeltaFile
+62-0libcxx/cmake/caches/Generic-llvm-libc.cmake
+27-0libcxx/test/configs/llvm-libc++-llvm-libc.cfg.in
+27-0libcxxabi/test/configs/llvm-libc++abi-llvm-libc.cfg.in
+20-0libcxx/utils/ci/run-buildbot
+4-0libcxx/utils/libcxx/test/features/platform.py
+3-0libcxx/test/std/containers/container.adaptors/container.adaptors.format/format.functions.vformat.pass.cpp
+143-074 files not shown
+361-080 files

LLVM/project 103dcc2libcxx/cmake/caches Generic-llvm-libc.cmake, libcxx/test/configs llvm-libc++-llvm-libc.cfg.in

[𝘀𝗽𝗿] changes to main this commit is based on

Created using spr 1.3.7

[skip ci]
DeltaFile
+62-0libcxx/cmake/caches/Generic-llvm-libc.cmake
+27-0libcxx/test/configs/llvm-libc++-llvm-libc.cfg.in
+27-0libcxxabi/test/configs/llvm-libc++abi-llvm-libc.cfg.in
+20-0libcxx/utils/ci/run-buildbot
+136-04 files

LLVM/project e030a42libcxx/test/configs llvm-libc++-llvm-libc.cfg.in

update

Created using spr 1.3.7
DeltaFile
+1-1libcxx/test/configs/llvm-libc++-llvm-libc.cfg.in
+1-11 files

LLVM/project 06a7d41flang/lib/Evaluate check-expression.cpp, flang/test/Lower call-copy-in-out.f90

[flang] Disable copy-out to INTENT(IN) args (#192382)

Don't copy out to actual args that themselves happen to be INTENT(IN)
dummy args.
```
    subroutine test(a)
      real, intent(in) :: a(:)
      call require_contiguous_arg(a(1:n:2))  ! copy-in only, no copy-out
    end
```

---------

Co-authored-by: Claude Sonnet 4.6 <noreply at anthropic.com>
DeltaFile
+44-0flang/test/Lower/call-copy-in-out.f90
+15-0flang/lib/Evaluate/check-expression.cpp
+59-02 files

LLVM/project 1a772bcllvm/lib/Target/X86 X86ISelLowering.cpp, llvm/test/CodeGen/X86 setcc-poison.ll setcc-carry.ll

[X86] Improve FREEZE node elimination for SETCC operations (#192362)

This improves FREEZE node handling around SETCC and SETCC_CARRY
operations to enable better optimization, particularly for APX
CCMP/CTEST
pattern matching with fastmath comparisons.

Resolve https://github.com/llvm/llvm-project/issues/191716.
DeltaFile
+149-0llvm/test/CodeGen/X86/setcc-poison.ll
+112-0llvm/test/CodeGen/X86/setcc-carry.ll
+37-0llvm/test/CodeGen/X86/apx/ctest.ll
+37-0llvm/test/CodeGen/X86/apx/ccmp.ll
+10-10llvm/test/CodeGen/X86/freeze-binary.ll
+4-0llvm/lib/Target/X86/X86ISelLowering.cpp
+349-106 files

LLVM/project a1a40cblldb/packages/Python/lldbsuite/test lldbtest.py, lldb/packages/Python/lldbsuite/test/make Makefile.rules

[lldb/test] Fix shared library symlinks for remote testing (#189177)

When running tests on a remote device, framework convenience symlinks
created by test Makefiles (e.g. `$(BUILDDIR)/Framework` pointing to
`$(BUILDDIR)/Framework.framework/Framework`) cause launch failures.

`Platform::Install` recreates these as symlinks on the remote device
pointing to host build paths that don't exist, resulting in "No such
file or directory" from dyld.

This patch changes `LN_SF` in Makefile.rules to strip the common
directory prefix from the symlink source using `patsubst` so it produces
relative symlinks instead of absolute ones.

It also resolve symlinks with `os.path.realpath()` in
`registerSharedLibrariesWithTarget` before registering modules so that
`Platform::Install` sees a regular file and transfers the actual binary
content.


    [2 lines not shown]
DeltaFile
+7-1lldb/packages/Python/lldbsuite/test/make/Makefile.rules
+1-0lldb/packages/Python/lldbsuite/test/lldbtest.py
+8-12 files

LLVM/project 0d0595bclang/lib/CodeGen/Targets SPIR.cpp, clang/test/CodeGen/AMDGPU amdgpu-atomic-float.c

[SPIR-V] Encode Atomic metadata as UserSemantic string decoration (#193019)

AMDGPU uses metadata to guide atomic related optimisations. SPIR-V was
not handling it, which led to significant and spurious performance
differences. This patch fixes this oversight by encoding the metadata as
UserSemantic string decorations applied to the atomic instructions.
DeltaFile
+304-18clang/test/CodeGenCUDA/atomic-options.hip
+194-25clang/test/CodeGenHIP/amdgpu-flat-atomic-fadd.hip
+113-16clang/test/CodeGen/AMDGPU/amdgpu-atomic-float.c
+70-9clang/test/CodeGenHIP/amdgpu-global-atomic-fadd.hip
+25-0clang/lib/CodeGen/Targets/SPIR.cpp
+22-0llvm/lib/Target/SPIRV/SPIRVEmitIntrinsics.cpp
+728-683 files not shown
+751-699 files

LLVM/project 46e09c5llvm/lib/Transforms/Scalar ExpandMemCmp.cpp

[ExpandMemCmp] Pre-collect memcmp calls to improve compile time (#193415)

Avoid restarting the basic block iteration from the beginning of the
function every time a memcmp/bcmp is expanded. Instead, pre-collect all
memcmp/bcmp calls and process them in a single pass.
DeltaFile
+17-40llvm/lib/Transforms/Scalar/ExpandMemCmp.cpp
+17-401 files

LLVM/project 793bdd8libc/src/__support/threads CndVar.h CMakeLists.txt, libc/src/__support/threads/linux CndVar.cpp CMakeLists.txt

[libc][CndVar] reimplmement conditional variable with FIFO ordering (#192748)

This PR reimplements conditional variable with two different variants:
- futex-based shared condvar with atomic counter for waiters
- queue-based private condvar

Notice that thread-local queue node cannot be reliably accessed in
shared processes, so we cannot use a unified implementation in this
case.

POSIX.1-2024 (Issue 8) added atomicity conditions to conditional
variable:

- The `pthread_cond_broadcast()` function shall, **as a single atomic
operation**, determine which threads, if any, are blocked on the
specified condition variable cond and unblock all of these threads.

- The `pthread_cond_signal()` function shall, as a **single atomic
operation**, determine which threads, if any, are blocked on the

    [41 lines not shown]
DeltaFile
+324-25libc/src/__support/threads/CndVar.h
+128-0libc/test/integration/src/__support/threads/cndvar_test.cpp
+0-106libc/src/__support/threads/linux/CndVar.cpp
+16-7libc/src/__support/threads/CMakeLists.txt
+0-23libc/src/__support/threads/linux/CMakeLists.txt
+18-0libc/test/integration/src/__support/threads/CMakeLists.txt
+486-16113 files not shown
+523-18919 files

LLVM/project ccc608fllvm/lib/Target/DirectX DXILOpLowering.cpp DXILResourceAccess.cpp, llvm/test/CodeGen/DirectX TextureLoad.ll

[DirectX] Implement lowering of Texture Load and Texture .operator[] (#193343)

Fixes https://github.com/llvm/llvm-project/issues/192546 and
https://github.com/llvm/llvm-project/issues/192558

This PR defines the TextureLoad DXIL Op (opcode 66), and implements
lowering of the texture load (dx_resource_load_level) intrinsic to the
DXIL op.

This PR also implements the transformation of loads from texture
resources (via dx_resource_getpointer) into dx_resource_load_level
intrinsics.

Assisted-by: Claude Opus 4.7
DeltaFile
+223-0llvm/test/CodeGen/DirectX/TextureLoad.ll
+97-0llvm/test/CodeGen/DirectX/ResourceAccess/load_texture.ll
+57-0llvm/lib/Target/DirectX/DXILOpLowering.cpp
+46-1llvm/lib/Target/DirectX/DXILResourceAccess.cpp
+14-1llvm/lib/Target/DirectX/DXIL.td
+437-25 files

LLVM/project 417f5bccompiler-rt/test/sanitizer_common/TestCases/Posix getpw_getgr.cpp

[NFC][sanitizer_common] Fix getpw_getgr.cpp test for large groups (#193625)

On my desktop buffer needs to be about 1MB.
DeltaFile
+13-2compiler-rt/test/sanitizer_common/TestCases/Posix/getpw_getgr.cpp
+13-21 files

LLVM/project 6fa0a86compiler-rt/test/sanitizer_common/TestCases/Posix getpw_getgr.cpp

simplify

Created using spr 1.3.7
DeltaFile
+1-4compiler-rt/test/sanitizer_common/TestCases/Posix/getpw_getgr.cpp
+1-41 files

LLVM/project 50b94b6llvm/docs AMDGPUUsage.rst, llvm/include/llvm/IR IntrinsicsAMDGPU.td

[AMDGPU] Add a sched group mask for LDSDMA instructions

The existing VMEM masks are not fine-grained enough for some use cases. For
example, if users want to control async loads, using VMEM may cause the compiler
to pick instructions it shouldn't.

This PR adds a new sched group mask for LDSDMA instructions. It is a subclass of
VMEM, but only targets isLDSDMA instructions.
DeltaFile
+342-0llvm/test/CodeGen/AMDGPU/sched-ldsdma-mask.mir
+21-21llvm/test/CodeGen/AMDGPU/sched.barrier.inverted.mask.ll
+15-7llvm/lib/Target/AMDGPU/AMDGPUIGroupLP.cpp
+2-0llvm/include/llvm/IR/IntrinsicsAMDGPU.td
+1-0llvm/docs/AMDGPUUsage.rst
+381-285 files

LLVM/project 779dfb8compiler-rt/test/sanitizer_common/TestCases/Posix getpw_getgr.cpp

format

Created using spr 1.3.7
DeltaFile
+1-1compiler-rt/test/sanitizer_common/TestCases/Posix/getpw_getgr.cpp
+1-11 files

LLVM/project 95aceabcompiler-rt/test/sanitizer_common/TestCases/Posix getpw_getgr.cpp

size

Created using spr 1.3.7
DeltaFile
+1-1compiler-rt/test/sanitizer_common/TestCases/Posix/getpw_getgr.cpp
+1-11 files

LLVM/project 79c85c8compiler-rt/test/sanitizer_common/TestCases/Posix getpw_getgr.cpp

[𝘀𝗽𝗿] initial version

Created using spr 1.3.7
DeltaFile
+16-2compiler-rt/test/sanitizer_common/TestCases/Posix/getpw_getgr.cpp
+16-21 files