Windows Build 18.1.8

Changes:

• ac2765e Workaround for cmake bug when compiling with /MT
• 7d99be9 Create test.txt

This list of changes was auto generated.

Windows Build 16.0.1

Changes:

• f7a5e52 16.0.1

This list of changes was auto generated.

Windows Build 16.0.0

Changes:

• 89d5468 *** USE llvm/llvm-project ***
• 9b52b6c Clean up and improve Zen CPU detection
• bc7b2ed Quick hack for Zen 4, more Zen 3 detection
• e8f2aa0 add raptorlake target
• 5521155 Fix register clobbering on aarch64 GHC when mixing tail/non-tail calls
• c725f49 X86: Avoid converting EVEX to VEX when disp8 would be beneficial
• eb7a5e5 Fix tail call guarantee setting for GHC on arm64 backend
• 509d31a Force ELF on macOS
• 1c0ca19 Relax volatile stores (always return)
• a670c45 Enable LLVM_USE_PERF

See More

• 318b8fe X86: fixup matchPMADDWD_3
• 1cc7bdd X86: improve (V)PMADDWD detection (2)
• 610c27a X86: disable AVX512 truncate with saturation instructions
• 747fc0e lib/Support: Detect AVX-512 on Alderlake CPUs
• 18f153b lib/Support: Add Tigerlake and Alderlake detection
• 6382230 test: Remove test files
• 05332b7 TypeSize.h: add integral constructor to ElementCount
• 902970b IRBuilder.h: remove deprecations
• c9fceef X86: fixup (V)PMADDWD detection
• f7d625e X86: improve (V)PMADDWD detection
• c36b21c X86: modify PreserveAll CC to save full AVX-512 state
• 2e65b3a Fix possible compilation issue with gcc-11
• 548daf0 X86: avoid vector-scalar shifts if splat amount is directly a vector ADD/SUB/AND op.
• d5bc359 X86: add patterns for X86ISD::VSHLV and X86ISD::VSRLV
• bed7001 X86: add pattern for X86ISD::VSRAV
• 2ffa822 X86: expand detectAVGPattern()
• 5ff8f41 X86: optimize VSELECT for v16i8 with shl + sign bit test
• 4743d02 X86: LowerShift: new algorithm for vector-vector shifts
• d5b5885 Disable GDBRegistrationListener
• 6516f56 MCJIT: don't finalize modules on symbol lookup (workaround)
• d18817d X86: Fix/workaround Small Code Model for JIT
• 8edd96e Set up CI with Azure Pipelines
• 39c406a [AArch64][GlobalISel] Use ZExtValue for zext(xor) when invert tb(n)z
• be1489a [debuginfo-test][cross-project-tests] Release note for new project name
• aed4e74 [X86] combineX86ShuffleChain - ensure we only peek through bitcasts to vectors (PR51858)
• 47e63cb [openmp] Apply code change from D109500
• fca6438 [VPlan] Fix crash caused by not updating all users properly.
• 7b93a88 Revert "[AArch64][GlobalISel] Legalize bswap <2 x i16>"
• b6614b2 Revert [MC][ELF] Emit separate unique sections for different flags
• a967752 [X86] Don't clobber EBX in stackprobes
• 7d56483 [IR] Handle constant expressions in containsUndefinedElement()
• bedf13b [SimplifyCFG] performBranchToCommonDestFolding(): require block-closed SSA form for bonus instructions (PR51125)
• 5e5115b [MemCpyOpt] Fix a variety of scalable-type crashes
• 28d7691 Workaround incorrect types when lowering fixed length gather/scatter
• b37f5f2 Inform pass manager when child loops are deleted
• 601c2dd Fine grain control over some symbol visibility
• 5f6ef6f [AArch64][SME] Fix imm bug in mov vector to tile aliases
• 921995a Revert "[HardwareLoops] Change order of SCEV expression construction for InitLoopCount."
• d21237c [SelectionDAGBuilder] Bugfix in visitInlineAsm()
• 64f4a52 [llvm-objdump] Fix 'llvm-objdump -dr' for executables with relocations
• 09c230e Add llvm-ml to LLVM_TOOLCHAIN_TOOLS (PR50536)
• e0d7c39 [AArch64][sve] Prevent incorrect function call on fixed width vector
• 3c59cf5 [SCEV] Fix applyLoopGuards() with range check idiom (PR51760)
• 8830999 [DAGCombine] Prevent the transform of combine for multi-use operand
• 6d085fa [tests] precommit tests for D107692
• bf02d31 [docs] Update release notes with items related to Flang
• 12537ac [libomptarget][amdcgn] Only add opt/llvm-link dependency if TARGET is available [ org/D107156#2969862 ]
• 4b60833 [libomptarget][amdcgn] Add build dependency for llvm-link and opt
• ba85498 [RISCV] Fix reporting of incorrect commutable operand indices
• bf7908c [Orc] Enable debug object tests only on x86_64 hosts
• 683bdb0 [Linker] Support weak symbols in nodeduplicate COMDAT group
• e415eb6 [docs] Mention that the legacy PM is deprecated and will be removed after 14
• fda0edf [NewPM] Add missing LTO ArgPromotion pass
• 72d2352 [AlignFromAssume] Bailout w/non-constant alignments (pr51680)
• 6d1749e [SelectionDAG] Fix miscompile bugs related to smul.fix.sat with scale zero
• 4ddceef [WebAssembly] Fix FastISel of condition in different block (PR51651)
• 8fbe4dd [M68k] Update pointer data layout
• de85b17 [M68k][NFC] Rename M68kOperand::Kind to KindTy
• ab550c7 [CMake] Change -DENABLE_EXPERIMENTAL_NEW_PASS_MANAGER=off to -DLLVM_ENABLE_NEW_PASS_MANAGER=off
• 79ff6e4 [LoopIdiom] Don't transform loop into memmove when load from body has more than one use
• b61f809 [NFC][LoopIdiom] Add reproducer of wrong memmove transformation
• 5aea8f0 Revert "[RISCV] Fix reporting of incorrect commutable operand indices"
• 8a19c7f ThinLTO: Fix inline assembly references to static functions with CFI
• 2424302 [RISCV] Fix reporting of incorrect commutable operand indices
• 13fbbff [Coverage][llvm-cov] Correctly export branch coverage in LCOV format
• fa3a9f0 Revert sharing subprograms across CUs
• e4e6f3e [AArch64] Fix comparison peephole opt with non-0/1 immediate (PR51476)
• 45d26b8 [X86][AVX] Extract SUBV_BROADCAST constant bits from just the lower subvector range (PR51281)
• 4d78ad4 [ARM][atomicrmw] Fix CMP_SWAP_32 expand assert
• 8202608 [PowerPC] Disable CTR Loop generate for fma with the PPC double double type.
• 22b0c02 [DAGCombiner] Stop visitEXTRACT_SUBVECTOR creating illegal BITCASTs post legalisation.
• 0bbcb19 [Attributor][FIX] Guard constant casts with type size checks
• 4345e14 [InstCombine] avoid infinite loops from min/max canonicalization
• 5a14ea1 [InstSimplify] fold min/max with limit constant
• de802b8 [InstSimplify] add tests for min/max idioms; NFC
• 9740b5c [LoopVectorize] Improve vectorisation of some intrinsics by treating them as uniform
• 0020382 [NFC] Clean up tests in test/Transforms/LoopVectorize/assume.ll
• 3a88dc8 Add release notes for things relating to MinGW in the release
• 4a38ef8 [ELF] Don't emit SHF_GNU_RETAIN on Solaris
• fd411b2 [profile] Fix profile merging with binary IDs
• 681b643 [X86][SchedModel] Add missing ReadAdvance for some arithmetic ops (PR51318 and PR51322).
• fb132cb [MCA][NFC] Add tests for PR51318 and PR51322.
• db94372 [MCA] Simplify the rounding logic used in TimelineView::printWaitTimeEntry.
• 64a6596 [Attributor][NFC] Try to make the windows build bots happy
• d2cc939 [Attributor][FIX] Handle recurrences (PHIs) in AAPointerInfo explicitly
• e9a9a80 [Attributor][FIX] Only avoid visiting PHI uses multiple times (PR51249)
• 5cf4e36 [Attributor][NFC] Precommit reproducer for PR51249
• 0c8a79e [RISCV] Add scheduling resources for V
• 11c22c9 Drop LLVM_VERSION_SUFFIX
• 13e216d [llvm] [cmake] Export LLVM_EN...

Windows Build

Changes:

• 9b52b6c Clean up and improve Zen CPU detection
• bc7b2ed Quick hack for Zen 4, more Zen 3 detection
• e8f2aa0 add raptorlake target
• 5521155 Fix register clobbering on aarch64 GHC when mixing tail/non-tail calls
• c725f49 X86: Avoid converting EVEX to VEX when disp8 would be beneficial
• eb7a5e5 Fix tail call guarantee setting for GHC on arm64 backend
• 509d31a Force ELF on macOS
• 1c0ca19 Relax volatile stores (always return)
• a670c45 Enable LLVM_USE_PERF
• 318b8fe X86: fixup matchPMADDWD_3

See More

• 1cc7bdd X86: improve (V)PMADDWD detection (2)
• 610c27a X86: disable AVX512 truncate with saturation instructions
• 747fc0e lib/Support: Detect AVX-512 on Alderlake CPUs
• 18f153b lib/Support: Add Tigerlake and Alderlake detection
• 6382230 test: Remove test files
• 05332b7 TypeSize.h: add integral constructor to ElementCount
• 902970b IRBuilder.h: remove deprecations
• c9fceef X86: fixup (V)PMADDWD detection
• f7d625e X86: improve (V)PMADDWD detection
• c36b21c X86: modify PreserveAll CC to save full AVX-512 state
• 2e65b3a Fix possible compilation issue with gcc-11
• 548daf0 X86: avoid vector-scalar shifts if splat amount is directly a vector ADD/SUB/AND op.
• d5bc359 X86: add patterns for X86ISD::VSHLV and X86ISD::VSRLV
• bed7001 X86: add pattern for X86ISD::VSRAV
• 2ffa822 X86: expand detectAVGPattern()
• 5ff8f41 X86: optimize VSELECT for v16i8 with shl + sign bit test
• 4743d02 X86: LowerShift: new algorithm for vector-vector shifts
• d5b5885 Disable GDBRegistrationListener
• 6516f56 MCJIT: don't finalize modules on symbol lookup (workaround)
• d18817d X86: Fix/workaround Small Code Model for JIT
• 8edd96e Set up CI with Azure Pipelines
• 39c406a [AArch64][GlobalISel] Use ZExtValue for zext(xor) when invert tb(n)z
• be1489a [debuginfo-test][cross-project-tests] Release note for new project name
• aed4e74 [X86] combineX86ShuffleChain - ensure we only peek through bitcasts to vectors (PR51858)
• 47e63cb [openmp] Apply code change from D109500
• fca6438 [VPlan] Fix crash caused by not updating all users properly.
• 7b93a88 Revert "[AArch64][GlobalISel] Legalize bswap <2 x i16>"
• b6614b2 Revert [MC][ELF] Emit separate unique sections for different flags
• a967752 [X86] Don't clobber EBX in stackprobes
• 7d56483 [IR] Handle constant expressions in containsUndefinedElement()
• bedf13b [SimplifyCFG] performBranchToCommonDestFolding(): require block-closed SSA form for bonus instructions (PR51125)
• 5e5115b [MemCpyOpt] Fix a variety of scalable-type crashes
• 28d7691 Workaround incorrect types when lowering fixed length gather/scatter
• b37f5f2 Inform pass manager when child loops are deleted
• 601c2dd Fine grain control over some symbol visibility
• 5f6ef6f [AArch64][SME] Fix imm bug in mov vector to tile aliases
• 921995a Revert "[HardwareLoops] Change order of SCEV expression construction for InitLoopCount."
• d21237c [SelectionDAGBuilder] Bugfix in visitInlineAsm()
• 64f4a52 [llvm-objdump] Fix 'llvm-objdump -dr' for executables with relocations
• 09c230e Add llvm-ml to LLVM_TOOLCHAIN_TOOLS (PR50536)
• e0d7c39 [AArch64][sve] Prevent incorrect function call on fixed width vector
• 3c59cf5 [SCEV] Fix applyLoopGuards() with range check idiom (PR51760)
• 8830999 [DAGCombine] Prevent the transform of combine for multi-use operand
• 6d085fa [tests] precommit tests for D107692
• bf02d31 [docs] Update release notes with items related to Flang
• 12537ac [libomptarget][amdcgn] Only add opt/llvm-link dependency if TARGET is available [ org/D107156#2969862 ]
• 4b60833 [libomptarget][amdcgn] Add build dependency for llvm-link and opt
• ba85498 [RISCV] Fix reporting of incorrect commutable operand indices
• bf7908c [Orc] Enable debug object tests only on x86_64 hosts
• 683bdb0 [Linker] Support weak symbols in nodeduplicate COMDAT group
• e415eb6 [docs] Mention that the legacy PM is deprecated and will be removed after 14
• fda0edf [NewPM] Add missing LTO ArgPromotion pass
• 72d2352 [AlignFromAssume] Bailout w/non-constant alignments (pr51680)
• 6d1749e [SelectionDAG] Fix miscompile bugs related to smul.fix.sat with scale zero
• 4ddceef [WebAssembly] Fix FastISel of condition in different block (PR51651)
• 8fbe4dd [M68k] Update pointer data layout
• de85b17 [M68k][NFC] Rename M68kOperand::Kind to KindTy
• ab550c7 [CMake] Change -DENABLE_EXPERIMENTAL_NEW_PASS_MANAGER=off to -DLLVM_ENABLE_NEW_PASS_MANAGER=off
• 79ff6e4 [LoopIdiom] Don't transform loop into memmove when load from body has more than one use
• b61f809 [NFC][LoopIdiom] Add reproducer of wrong memmove transformation
• 5aea8f0 Revert "[RISCV] Fix reporting of incorrect commutable operand indices"
• 8a19c7f ThinLTO: Fix inline assembly references to static functions with CFI
• 2424302 [RISCV] Fix reporting of incorrect commutable operand indices
• 13fbbff [Coverage][llvm-cov] Correctly export branch coverage in LCOV format
• fa3a9f0 Revert sharing subprograms across CUs
• e4e6f3e [AArch64] Fix comparison peephole opt with non-0/1 immediate (PR51476)
• 45d26b8 [X86][AVX] Extract SUBV_BROADCAST constant bits from just the lower subvector range (PR51281)
• 4d78ad4 [ARM][atomicrmw] Fix CMP_SWAP_32 expand assert
• 8202608 [PowerPC] Disable CTR Loop generate for fma with the PPC double double type.
• 22b0c02 [DAGCombiner] Stop visitEXTRACT_SUBVECTOR creating illegal BITCASTs post legalisation.
• 0bbcb19 [Attributor][FIX] Guard constant casts with type size checks
• 4345e14 [InstCombine] avoid infinite loops from min/max canonicalization
• 5a14ea1 [InstSimplify] fold min/max with limit constant
• de802b8 [InstSimplify] add tests for min/max idioms; NFC
• 9740b5c [LoopVectorize] Improve vectorisation of some intrinsics by treating them as uniform
• 0020382 [NFC] Clean up tests in test/Transforms/LoopVectorize/assume.ll
• 3a88dc8 Add release notes for things relating to MinGW in the release
• 4a38ef8 [ELF] Don't emit SHF_GNU_RETAIN on Solaris
• fd411b2 [profile] Fix profile merging with binary IDs
• 681b643 [X86][SchedModel] Add missing ReadAdvance for some arithmetic ops (PR51318 and PR51322).
• fb132cb [MCA][NFC] Add tests for PR51318 and PR51322.
• db94372 [MCA] Simplify the rounding logic used in TimelineView::printWaitTimeEntry.
• 64a6596 [Attributor][NFC] Try to make the windows build bots happy
• d2cc939 [Attributor][FIX] Handle recurrences (PHIs) in AAPointerInfo explicitly
• e9a9a80 [Attributor][FIX] Only avoid visiting PHI uses multiple times (PR51249)
• 5cf4e36 [Attributor][NFC] Precommit reproducer for PR51249
• 0c8a79e [RISCV] Add scheduling resources for V
• 11c22c9 Drop LLVM_VERSION_SUFFIX
• 13e216d [llvm] [cmake] Export LLVM_ENABLE_NEW_PASS_MANAGER into LLVMConfig.cmake
• ee15bdb...

Linux Build

Changes:

• 9b52b6c Clean up and improve Zen CPU detection
• bc7b2ed Quick hack for Zen 4, more Zen 3 detection
• e8f2aa0 add raptorlake target
• 5521155 Fix register clobbering on aarch64 GHC when mixing tail/non-tail calls
• c725f49 X86: Avoid converting EVEX to VEX when disp8 would be beneficial
• eb7a5e5 Fix tail call guarantee setting for GHC on arm64 backend
• 509d31a Force ELF on macOS
• 1c0ca19 Relax volatile stores (always return)
• a670c45 Enable LLVM_USE_PERF
• 318b8fe X86: fixup matchPMADDWD_3

See More

• 1cc7bdd X86: improve (V)PMADDWD detection (2)
• 610c27a X86: disable AVX512 truncate with saturation instructions
• 747fc0e lib/Support: Detect AVX-512 on Alderlake CPUs
• 18f153b lib/Support: Add Tigerlake and Alderlake detection
• 6382230 test: Remove test files
• 05332b7 TypeSize.h: add integral constructor to ElementCount
• 902970b IRBuilder.h: remove deprecations
• c9fceef X86: fixup (V)PMADDWD detection
• f7d625e X86: improve (V)PMADDWD detection
• c36b21c X86: modify PreserveAll CC to save full AVX-512 state
• 2e65b3a Fix possible compilation issue with gcc-11
• 548daf0 X86: avoid vector-scalar shifts if splat amount is directly a vector ADD/SUB/AND op.
• d5bc359 X86: add patterns for X86ISD::VSHLV and X86ISD::VSRLV
• bed7001 X86: add pattern for X86ISD::VSRAV
• 2ffa822 X86: expand detectAVGPattern()
• 5ff8f41 X86: optimize VSELECT for v16i8 with shl + sign bit test
• 4743d02 X86: LowerShift: new algorithm for vector-vector shifts
• d5b5885 Disable GDBRegistrationListener
• 6516f56 MCJIT: don't finalize modules on symbol lookup (workaround)
• d18817d X86: Fix/workaround Small Code Model for JIT
• 8edd96e Set up CI with Azure Pipelines
• 39c406a [AArch64][GlobalISel] Use ZExtValue for zext(xor) when invert tb(n)z
• be1489a [debuginfo-test][cross-project-tests] Release note for new project name
• aed4e74 [X86] combineX86ShuffleChain - ensure we only peek through bitcasts to vectors (PR51858)
• 47e63cb [openmp] Apply code change from D109500
• fca6438 [VPlan] Fix crash caused by not updating all users properly.
• 7b93a88 Revert "[AArch64][GlobalISel] Legalize bswap <2 x i16>"
• b6614b2 Revert [MC][ELF] Emit separate unique sections for different flags
• a967752 [X86] Don't clobber EBX in stackprobes
• 7d56483 [IR] Handle constant expressions in containsUndefinedElement()
• bedf13b [SimplifyCFG] performBranchToCommonDestFolding(): require block-closed SSA form for bonus instructions (PR51125)
• 5e5115b [MemCpyOpt] Fix a variety of scalable-type crashes
• 28d7691 Workaround incorrect types when lowering fixed length gather/scatter
• b37f5f2 Inform pass manager when child loops are deleted
• 601c2dd Fine grain control over some symbol visibility
• 5f6ef6f [AArch64][SME] Fix imm bug in mov vector to tile aliases
• 921995a Revert "[HardwareLoops] Change order of SCEV expression construction for InitLoopCount."
• d21237c [SelectionDAGBuilder] Bugfix in visitInlineAsm()
• 64f4a52 [llvm-objdump] Fix 'llvm-objdump -dr' for executables with relocations
• 09c230e Add llvm-ml to LLVM_TOOLCHAIN_TOOLS (PR50536)
• e0d7c39 [AArch64][sve] Prevent incorrect function call on fixed width vector
• 3c59cf5 [SCEV] Fix applyLoopGuards() with range check idiom (PR51760)
• 8830999 [DAGCombine] Prevent the transform of combine for multi-use operand
• 6d085fa [tests] precommit tests for D107692
• bf02d31 [docs] Update release notes with items related to Flang
• 12537ac [libomptarget][amdcgn] Only add opt/llvm-link dependency if TARGET is available [ org/D107156#2969862 ]
• 4b60833 [libomptarget][amdcgn] Add build dependency for llvm-link and opt
• ba85498 [RISCV] Fix reporting of incorrect commutable operand indices
• bf7908c [Orc] Enable debug object tests only on x86_64 hosts
• 683bdb0 [Linker] Support weak symbols in nodeduplicate COMDAT group
• e415eb6 [docs] Mention that the legacy PM is deprecated and will be removed after 14
• fda0edf [NewPM] Add missing LTO ArgPromotion pass
• 72d2352 [AlignFromAssume] Bailout w/non-constant alignments (pr51680)
• 6d1749e [SelectionDAG] Fix miscompile bugs related to smul.fix.sat with scale zero
• 4ddceef [WebAssembly] Fix FastISel of condition in different block (PR51651)
• 8fbe4dd [M68k] Update pointer data layout
• de85b17 [M68k][NFC] Rename M68kOperand::Kind to KindTy
• ab550c7 [CMake] Change -DENABLE_EXPERIMENTAL_NEW_PASS_MANAGER=off to -DLLVM_ENABLE_NEW_PASS_MANAGER=off
• 79ff6e4 [LoopIdiom] Don't transform loop into memmove when load from body has more than one use
• b61f809 [NFC][LoopIdiom] Add reproducer of wrong memmove transformation
• 5aea8f0 Revert "[RISCV] Fix reporting of incorrect commutable operand indices"
• 8a19c7f ThinLTO: Fix inline assembly references to static functions with CFI
• 2424302 [RISCV] Fix reporting of incorrect commutable operand indices
• 13fbbff [Coverage][llvm-cov] Correctly export branch coverage in LCOV format
• fa3a9f0 Revert sharing subprograms across CUs
• e4e6f3e [AArch64] Fix comparison peephole opt with non-0/1 immediate (PR51476)
• 45d26b8 [X86][AVX] Extract SUBV_BROADCAST constant bits from just the lower subvector range (PR51281)
• 4d78ad4 [ARM][atomicrmw] Fix CMP_SWAP_32 expand assert
• 8202608 [PowerPC] Disable CTR Loop generate for fma with the PPC double double type.
• 22b0c02 [DAGCombiner] Stop visitEXTRACT_SUBVECTOR creating illegal BITCASTs post legalisation.
• 0bbcb19 [Attributor][FIX] Guard constant casts with type size checks
• 4345e14 [InstCombine] avoid infinite loops from min/max canonicalization
• 5a14ea1 [InstSimplify] fold min/max with limit constant
• de802b8 [InstSimplify] add tests for min/max idioms; NFC
• 9740b5c [LoopVectorize] Improve vectorisation of some intrinsics by treating them as uniform
• 0020382 [NFC] Clean up tests in test/Transforms/LoopVectorize/assume.ll
• 3a88dc8 Add release notes for things relating to MinGW in the release
• 4a38ef8 [ELF] Don't emit SHF_GNU_RETAIN on Solaris
• fd411b2 [profile] Fix profile merging with binary IDs
• 681b643 [X86][SchedModel] Add missing ReadAdvance for some arithmetic ops (PR51318 and PR51322).
• fb132cb [MCA][NFC] Add tests for PR51318 and PR51322.
• db94372 [MCA] Simplify the rounding logic used in TimelineView::printWaitTimeEntry.
• 64a6596 [Attributor][NFC] Try to make the windows build bots happy
• d2cc939 [Attributor][FIX] Handle recurrences (PHIs) in AAPointerInfo explicitly
• e9a9a80 [Attributor][FIX] Only avoid visiting PHI uses multiple times (PR51249)
• 5cf4e36 [Attributor][NFC] Precommit reproducer for PR51249
• 0c8a79e [RISCV] Add scheduling resources for V
• 11c22c9 Drop LLVM_VERSION_SUFFIX
• 13e216d [llvm] [cmake] Export LLVM_ENABLE_NEW_PASS_MANAGER into LLVMConfig.cmake
• ee15bdb...