7 Releases 8 Tags

RSS Feed

• custom-build-win-21.1.8 af199a4717

  Compare

  Stable

  yindo released this 2025-12-22 02:42:31 -05:00 | 0 commits to master since this release

  📅 Originally published on GitHub: Sat, 24 Jan 2026 11:07:22 GMT
  🏷️ Git tag created: Mon, 22 Dec 2025 07:42:31 GMT

  What's Changed

  • Bump llvm to 21.1.8
  • This release is generated by https://github.com/Megamouse/rpcs3/actions/runs/21312728363
  • We now use the official llvm repository and only use our llvm-mirror for publishing the releases
  • The source code in this release does belong to the llvm-mirror and does not match the actual release. See https://github.com/llvm/llvm-project/releases/tag/llvmorg-21.1.8 for the actual source code

  Downloads

  • llvmlibs_mt.7z
    46 MiB

    2026-02-18 19:21:00 -05:00
  • llvmlibs_mt.7z.sha256
    65 B

    2026-02-18 19:21:00 -05:00
  • Source Code (ZIP)
  • Source Code (TAR.GZ)
• custom-build-win-19.1.7 a538f027f9

  Compare

  Stable

  yindo released this 2025-02-05 12:46:23 -05:00 | 3 commits to master since this release

  📅 Originally published on GitHub: Thu, 06 Feb 2025 00:12:15 GMT
  🏷️ Git tag created: Wed, 05 Feb 2025 17:46:23 GMT

  What's Changed

  • Bump to LLVM 19.1.7 by @kd-11 in https://github.com/RPCS3/llvm-mirror/pull/19
  • Export token from pipeline by @kd-11 in https://github.com/RPCS3/llvm-mirror/pull/21

  Full Changelog: https://github.com/RPCS3/llvm-mirror/compare/custom-build-win-18.1.8...custom-build-win-19.1.7

  Downloads

  • llvmlibs_mt.7z
    46 MiB

    2026-02-18 19:20:51 -05:00
  • llvmlibs_mt.7z.sha256
    66 B

    2026-02-18 19:20:51 -05:00
  • Source Code (ZIP)
  • Source Code (TAR.GZ)
• custom-build-win-18.1.8 1e1a7a3772

  Compare

  Stable

  yindo released this 2024-09-01 18:51:23 -04:00 | 9 commits to master since this release

  📅 Originally published on GitHub: Tue, 10 Sep 2024 06:41:46 GMT
  🏷️ Git tag created: Sun, 01 Sep 2024 22:51:23 GMT

  Changes:

  • ac2765e624 Workaround for cmake bug when compiling with /MT
  • 7d99be9394 Create test.txt

  This list of changes was auto generated.

  Downloads

  • llvmlibs_mt.7z
    56 MiB

    2026-02-18 19:20:43 -05:00
  • llvmlibs_mt.7z.sha256
    65 B

    2026-02-18 19:20:43 -05:00
  • Source Code (ZIP)
  • Source Code (TAR.GZ)
• custom-build-win-16.0.1 f7a5e52b33

  Compare

  Stable

  yindo released this 2023-04-08 06:50:41 -04:00 | 10 commits to master since this release

  📅 Originally published on GitHub: Sat, 08 Apr 2023 12:17:01 GMT
  🏷️ Git tag created: Sat, 08 Apr 2023 10:50:41 GMT

  Changes:

  • f7a5e52b33 16.0.1

  This list of changes was auto generated.

  Downloads

  • llvmlibs_mt.7z
    52 MiB

    2026-02-18 19:20:33 -05:00
  • llvmlibs_mt.7z.sha256
    65 B

    2026-02-18 19:20:33 -05:00
  • Source Code (ZIP)
  • Source Code (TAR.GZ)
• custom-build-win-16.0.0 89d5468e95

  Compare

  Stable

  yindo released this 2023-03-21 07:41:10 -04:00 | 11 commits to master since this release

  📅 Originally published on GitHub: Tue, 21 Mar 2023 13:20:15 GMT
  🏷️ Git tag created: Tue, 21 Mar 2023 11:41:10 GMT

  Changes:

  • 89d5468e95 *** USE llvm/llvm-project ***
  • 9b52b6c39a Clean up and improve Zen CPU detection
  • bc7b2ed172 Quick hack for Zen 4, more Zen 3 detection
  • e8f2aa0ae4 add raptorlake target
  • 5521155be5 Fix register clobbering on aarch64 GHC when mixing tail/non-tail calls
  • c725f494c9 X86: Avoid converting EVEX to VEX when disp8 would be beneficial
  • eb7a5e5301 Fix tail call guarantee setting for GHC on arm64 backend
  • 509d31ad89 Force ELF on macOS
  • 1c0ca194dc Relax volatile stores (always return)
  • a670c459ea Enable LLVM_USE_PERF

  See More

  • 318b8fe374 X86: fixup matchPMADDWD_3
  • 1cc7bdd501 X86: improve (V)PMADDWD detection (2)
  • 610c27aa1c X86: disable AVX512 truncate with saturation instructions
  • 747fc0efaa lib/Support: Detect AVX-512 on Alderlake CPUs
  • 18f153b33b lib/Support: Add Tigerlake and Alderlake detection
  • 6382230f0e test: Remove test files
  • 05332b78c7 TypeSize.h: add integral constructor to ElementCount
  • 902970b075 IRBuilder.h: remove deprecations
  • c9fceef173 X86: fixup (V)PMADDWD detection
  • f7d625e31a X86: improve (V)PMADDWD detection
  • c36b21c023 X86: modify PreserveAll CC to save full AVX-512 state
  • 2e65b3a418 Fix possible compilation issue with gcc-11
  • 548daf04b5 X86: avoid vector-scalar shifts if splat amount is directly a vector ADD/SUB/AND op.
  • d5bc359dfd X86: add patterns for X86ISD::VSHLV and X86ISD::VSRLV
  • bed700114e X86: add pattern for X86ISD::VSRAV
  • 2ffa82223f X86: expand detectAVGPattern()
  • 5ff8f4151c X86: optimize VSELECT for v16i8 with shl + sign bit test
  • 4743d020ce X86: LowerShift: new algorithm for vector-vector shifts
  • d5b5885c23 Disable GDBRegistrationListener
  • 6516f565ed MCJIT: don't finalize modules on symbol lookup (workaround)
  • d18817ded9 X86: Fix/workaround Small Code Model for JIT
  • 8edd96e498 Set up CI with Azure Pipelines
  • 39c406a58f [AArch64][GlobalISel] Use ZExtValue for zext(xor) when invert tb(n)z
  • be1489a181 [debuginfo-test][cross-project-tests] Release note for new project name
  • aed4e7449f [X86] combineX86ShuffleChain - ensure we only peek through bitcasts to vectors (PR51858)
  • 47e63cb60e [openmp] Apply code change from D109500
  • fca643847d [VPlan] Fix crash caused by not updating all users properly.
  • 7b93a88a1e Revert "[AArch64][GlobalISel] Legalize bswap <2 x i16>"
  • b6614b2353 Revert [MC][ELF] Emit separate unique sections for different flags
  • a967752a95 [X86] Don't clobber EBX in stackprobes
  • 7d56483207 [IR] Handle constant expressions in containsUndefinedElement()
  • bedf13b1a7 [SimplifyCFG] performBranchToCommonDestFolding(): require block-closed SSA form for bonus instructions (PR51125)
  • 5e5115bbd0 [MemCpyOpt] Fix a variety of scalable-type crashes
  • 28d769100d Workaround incorrect types when lowering fixed length gather/scatter
  • b37f5f2114 Inform pass manager when child loops are deleted
  • 601c2dd9dd Fine grain control over some symbol visibility
  • 5f6ef6fbfd [AArch64][SME] Fix imm bug in mov vector to tile aliases
  • 921995afd5 Revert "[HardwareLoops] Change order of SCEV expression construction for InitLoopCount."
  • d21237cb11 [SelectionDAGBuilder] Bugfix in visitInlineAsm()
  • 64f4a52b52 [llvm-objdump] Fix 'llvm-objdump -dr' for executables with relocations
  • 09c230ecec Add llvm-ml to LLVM_TOOLCHAIN_TOOLS (PR50536)
  • e0d7c39869 [AArch64][sve] Prevent incorrect function call on fixed width vector
  • 3c59cf5aa7 [SCEV] Fix applyLoopGuards() with range check idiom (PR51760)
  • 8830999b87 [DAGCombine] Prevent the transform of combine for multi-use operand
  • 6d085fa2f5 [tests] precommit tests for D107692
  • bf02d31357 [docs] Update release notes with items related to Flang
  • 12537ac90c [libomptarget][amdcgn] Only add opt/llvm-link dependency if TARGET is available [ org/D107156#2969862 ]
  • 4b60833c5b [libomptarget][amdcgn] Add build dependency for llvm-link and opt
  • ba85498148 [RISCV] Fix reporting of incorrect commutable operand indices
  • bf7908ca9b [Orc] Enable debug object tests only on x86_64 hosts
  • 683bdb08a0 [Linker] Support weak symbols in nodeduplicate COMDAT group
  • e415eb692a [docs] Mention that the legacy PM is deprecated and will be removed after 14
  • fda0edff63 [NewPM] Add missing LTO ArgPromotion pass
  • 72d2352801 [AlignFromAssume] Bailout w/non-constant alignments (pr51680)
  • 6d1749e6c0 [SelectionDAG] Fix miscompile bugs related to smul.fix.sat with scale zero
  • 4ddceef928 [WebAssembly] Fix FastISel of condition in different block (PR51651)
  • 8fbe4ddc7c [M68k] Update pointer data layout
  • de85b171b7 [M68k][NFC] Rename M68kOperand::Kind to KindTy
  • ab550c798f [CMake] Change -DENABLE_EXPERIMENTAL_NEW_PASS_MANAGER=off to -DLLVM_ENABLE_NEW_PASS_MANAGER=off
  • 79ff6e4b2d [LoopIdiom] Don't transform loop into memmove when load from body has more than one use
  • b61f80921f [NFC][LoopIdiom] Add reproducer of wrong memmove transformation
  • 5aea8f0472 Revert "[RISCV] Fix reporting of incorrect commutable operand indices"
  • 8a19c7f52b ThinLTO: Fix inline assembly references to static functions with CFI
  • 2424302e99 [RISCV] Fix reporting of incorrect commutable operand indices
  • 13fbbfff3a [Coverage][llvm-cov] Correctly export branch coverage in LCOV format
  • fa3a9f0fc7 Revert sharing subprograms across CUs
  • e4e6f3eeff [AArch64] Fix comparison peephole opt with non-0/1 immediate (PR51476)
  • 45d26b8826 [X86][AVX] Extract SUBV_BROADCAST constant bits from just the lower subvector range (PR51281)
  • 4d78ad44fb [ARM][atomicrmw] Fix CMP_SWAP_32 expand assert
  • 8202608068 [PowerPC] Disable CTR Loop generate for fma with the PPC double double type.
  • 22b0c02125 [DAGCombiner] Stop visitEXTRACT_SUBVECTOR creating illegal BITCASTs post legalisation.
  • 0bbcb191e3 [Attributor][FIX] Guard constant casts with type size checks
  • 4345e14522 [InstCombine] avoid infinite loops from min/max canonicalization
  • 5a14ea148e [InstSimplify] fold min/max with limit constant
  • de802b8e6e [InstSimplify] add tests for min/max idioms; NFC
  • 9740b5c5ef [LoopVectorize] Improve vectorisation of some intrinsics by treating them as uniform
  • 00203829b4 [NFC] Clean up tests in test/Transforms/LoopVectorize/assume.ll
  • 3a88dc8338 Add release notes for things relating to MinGW in the release
  • 4a38ef8718 [ELF] Don't emit SHF_GNU_RETAIN on Solaris
  • fd411b2b5d [profile] Fix profile merging with binary IDs
  • 681b643c07 [X86][SchedModel] Add missing ReadAdvance for some arithmetic ops (PR51318 and PR51322).
  • fb132cb74b [MCA][NFC] Add tests for PR51318 and PR51322.
  • db94372a40 [MCA] Simplify the rounding logic used in TimelineView::printWaitTimeEntry.
  • 64a6596867 [Attributor][NFC] Try to make the windows build bots happy
  • d2cc939747 [Attributor][FIX] Handle recurrences (PHIs) in AAPointerInfo explicitly
  • e9a9a807dd [Attributor][FIX] Only avoid visiting PHI uses multiple times (PR51249)
  • 5cf4e36ce1 [Attributor][NFC] Precommit reproducer for PR51249
  • 0c8a79e78d [RISCV] Add scheduling resources for V
  • 11c22c929d Drop LLVM_VERSION_SUFFIX
  • 13e216d251 [llvm] [cmake] Export LLVM_ENABLE_NEW_PASS_MANAGER into LLVMConfig.cmake
  • ee15bdbb06 [AArch64][SVE] Fix assertion failure when lowering fixed length gather/scatter
  • 7d89a50287 [llvm] [lit] Fix inconsistent test order in shtest-keyword-parse-errors
  • f6a86e448a BPF: avoid NE/EQ loop exit condition
  • a558de374f [InstCombine] Fixed select + masked load fold failure
  • 7563d8dfe5 [llvm-rc] Allow specifying language with a leading 0x prefix
  • 910de616a4 [DebugInfo][LSR] Avoid crashes on large integer inputs
  • 46ad88f625 Follow-up to D105207, only salvage affine SCEVs to avoid a crash
  • ff86d9e5f0 [DebugInfo][LoopStrengthReduction] SCEV-based salvaging for LSR
  • a75903d41d [ConstantFold] Get rid of special cases for sizeof etc.
  • f9a2d3e277 [test] Fix tools/gold/X86/comdat-nodeduplicate.ll on non-X86 hosts
  • 8be74d232c Fixing an infinite loop problem in InstCombine
  • 3851813b89 [PredicateInfo] Use Intrinsic::getDeclaration now that it handles unnamed types. [ #2875179 ]
  • e20914df74 Revert "Revert of D49126 [PredicateInfo] Use custom mangling to support ssa_copy with unnamed types."
  • 2b94ecbbe0 [SROA] prevent crash on large memset length (PR50910)
  • 30f0ebbbe2 [Attributor] Don't test internalization in the CGSCC pass.
  • 4f1fd1c209 [Attributor] Change function internalization to not replace uses in internalized callers
  • 710ae2bfd3 [ReleaseNotes] Add scalable matrix extension support to AArch64 changes
  • bafc7f359f [llvm][Release notes] Add AArch64 SVE, PAC and LLDB prebuilt binary
  • aa2a6b072f [llvm][Release notes] Add memory tagging support to lldb changes
  • d4eedb2312 [Analysis] improve function signature checking for snprintf
  • 9770d34891 [OpenMP] Fixing llvm-omp-device-info compilation with runtimes
  • 70f5e23577 [X86][AVX] Add test case for PR51281
  • df3286259b [DAGCombiner] don't try to partially reduce add-with-overflow ops
  • a0686462c3 [AArch64][x86] add tests for add-with-overflow folds; NFC
  • b92c9f9565 [DivRemPairs] make sure we have a valid CFG for hoisting division
  • 7c9c296915 [RISCV] Restrict performANY_EXTENDCombine to prevent an infinite loop.
  • 276fcebbe0 [AArch64] Legalize MVT::i64x8 in DAG isel lowering
  • a50e569197 [AArch64] Add a Machine Value Type for 8 consecutive registers
  • cd0096f439 [DebugInfo][InstrRef] Don't break up ret-sequences on debug-info instrs
  • 183b0c7c98 [AArch64][SVE] Fix incorrect mask type when lowering fixed type SVE gather/scatter
  • 923213f844 test-release.sh: Kill python2
  • 9a10dd5b1c Revert "[DebugInfo][LoopStrengthReduction] SCEV-based salvaging for LSR"
  • c463fa6cad [mlir][openacc] Initial translation for DataOp to LLVM IR
  • 5b7208da36 [OpenMP] Folding threadLimit and numThreads when single value in kernels
  • 66a78bc217 llvm/utils: guarantee revert_checker's revert ordering
  • bab6d38daf [DAGCombiner] Fold SETCC(FREEZE(x),const) to FREEZE(SETCC(x,const)) if SETCC is used by BRCOND
  • 2e57fe1d7d Precommit test files for D105344 (NFC)
  • a6d5003afd [X86] Fix lowering to illegal type in LowerINSERT_VECTOR_ELT
  • 9b53e594b4 Reapply "[Attributor] Disable simplification AAs if a callback is present""
  • 5d447ad589 Revert "[X86] Fix lowering to illegal type in LowerINSERT_VECTOR_ELT"
  • 115164b68b Add llvm::equal convenient wrapper for ranges around std::equal
  • 409f0eedd6 [X86] Fix lowering to illegal type in LowerINSERT_VECTOR_ELT
  • 60850cdc6a [Hexagon] Fix resetting dead registers in DBG_VALUE_LISTs
  • 803630dbd1 [gn build] Port 8a48e6dda9f7
  • 96f821bd7b Revert "[Attributor] Disable simplification AAs if a callback is present"
  • 0042081f58 [Attributor] Verify checkForAllUses return value properly
  • 0b6fb7b1ef [Attributor] Disable simplification AAs if a callback is present
  • 9aacda281c Fix test/Transforms/LoopVectorize/AArch64/strict-fadd-vf1.ll.
  • 091e720912 [gn build] manually port 71909de37495
  • 506c974562 [MLGO] fix silly LLVM_DEBUG misuse
  • f4b486c976 [NFC][MLGO] Debug messages for what inline advisor is selected
  • 8b3f85a32c [PowerPC] Turn deprecated altivec prefetch instrs to nops on AIX
  • 11aa71a71d [x86] update stale code comment; NFC
  • d571039392 [x86] add more tests for cmov and lea; NFC
  • ece3299a71 AMDGPU/GlobalISel: Fix selecting G_SEXTLOAD/G_ZEXTLOAD pre-gfx9
  • 17f139f330 AMDGPU/GlobalISel: Fix wrong addrspace in test MMOs
  • cfbd77a7f1 AMDGPU/GlobalISel: Add a few tests for unaligned truncating stores
  • c0d40a1ca3 Remove unused include that's also a layering violation. NFC.
  • 0a69d4f31a Add test update for a11d9a1f480f which disables fallbacks.
  • 6ce8f2f7c1 [AArch64][GlobalISel] Fix constraining LDXPX intrinsic selection.
  • 2a88f80da4 [ASAN] NFC: Remove redundant variable
  • c563d2cbcd [gn build] Port 02077da7e7a8
  • d9613eb43c [Matrix] Fix shape for factored transpose
  • add64be20a [Matrix] RAUW should only replace an instruction in ShapeMap if supportsShapeInfo
  • 6516543c4b Add jump-threading optimization for deterministic finite automata
  • 6bfc6b8665 [RISCV] Select vector shl by 1 to a vector add.
  • 0f86a54b34 [AArch64] Update and expand min-max cost model test. NFC
  • fd18a762a0 Enabling the copy-constant-to-alloca optimization in more instances
  • cdd50ed2ff [LoopVectorize] Don't interleave scalar ordered reductions for inner loops
  • 3ad8a0b712 Update reduction test. Remove standalone test file
  • 3b4453b968 AMDGPU: Update tests for lower i1 change
  • 8979bda8e8 AMDGPU: Treat IMPLICIT_DEF like a constant lanemask source
  • bb4e957ebf [WebAssembly] Codegen for extmul SIMD instructions
  • 2b967460b4 [SystemZ][z/OS] Initial code to generate assembly files on z/OS
  • 23f3f90569 Strip undef implying attributes when moving calls
  • 05308b27ea Revert "[X86][AVX] Add getBROADCAST_LOAD helper function. NFCI."
  • d809395fde Revert "Revert "[X86][AVX] Add getBROADCAST_LOAD helper function. NFCI.""
  • 9d32182a3a Revert "[X86][AVX] Add getBROADCAST_LOAD helper function. NFCI."
  • adf33c8470 [DebugInfo][InstrRef] Correctly update DBG_PHIs during instr scheduling
  • 64d9d45951 Handle unused variable when assertions are disabled
  • b8af1d6561 [IVDescriptors] Fix bug in checkOrderedReduction
  • fe743a5b25 [DebugInfo][LoopStrengthReduction] SCEV-based salvaging for LSR
  • 47a51741f3 [DebugInfo][InstrRef] Handle llvm.frameaddress intrinsics gracefully
  • e13e00f7b3 [DebugInfo][LoopStrengthReduction] SCEV-based salvaging for LSR
  • 2e863e863d [DebugInfo][LoopStrengthReduction] SCEV-based salvaging for LSR
  • b5cbfeed77 [PowerPC] add more testcases for ld_splat; nfc
  • ac97ba67e8 [LangRef][NFC] Fix variable name in llvm.maxnum docs
  • a99adc793b [X86] Add PR37025 test coverage
  • e1c8650040 [LV] Disable Scalable VFs when tail folding is enabled b/c of low tripcount.
  • 3bc8cd6a0b [GlobalISel] Constant fold G_SITOFP and G_UITOFP in CSEMIRBuilder
  • cf6bdfc026 [SelectionDAG] Support scalable splats in U(ADD|SUB)SAT combines
  • 5afacc5171 [RISCV] Add support for vector saturating add/sub operations
  • c2bd628550 [NFC] Reflow some debug messages.
  • 1ae2c0fb16 [AArch64][SME] Add zero instruction
  • b7fc1f8c1c [RISCV] Add tests showing missed vector saturating add/sub combines
  • 4d9d8789ad [ARM] Implement isLoad/StoreFromStackSlot for MVE stack stores accesses
  • 565fcd6a48 [LoopFlatten] Use SCEV and Loop APIs to identify increment and trip count
  • ee2584c4c2 [docs] Update release notes with all LLVM-C API changes
  • e8ead0fbb2 [Debug-Info][llvm-dwarfdump] Don't try to dump location
  • 10658f9d9e [gn build] Port 2487db1f2862
  • d08c2fabb8 [ORC] Require ExecutorProcessControl when constructing an ExecutionSession.
  • e9073c9b78 [OpenMP] Try to simplify all loads in device code
  • 756885168c [Attributor][FIX] Copy all members in the assignment operator
  • 7e9917e379 [Attributor] Utilize the InstSimplify interface to simplify instructions
  • 86cba7bf5e [InstSimplify] Expose generic interface for replaced operand simplification
  • 35ac57a3f1 [Attributor] Update check lines for all AMDGPU attributor tests
  • 2b8eaac9b5 [Attributor][FIX] Update AMDGPU attributor test
  • 700878ee36 [Coroutine] Record the elided coroutines
  • ffb4271d86 Merge all the llvm-exegesis unit tests into a single binary
  • eca1a76d00 [CSSPGO] Tweak ICP threshold in top-down inliner
  • 4ce16022d3 [Local] Do not introduce a new llvm.trap before unreachable
  • 95f415bd34 [Attributor] Delete dead stores
  • 5ee3942ae7 [Attributor] Introduce getPotentialCopiesOfStoredValue and use it
  • 60670bde11 Build libSupport with -Werror=global-constructors (NFC)
  • eb0ba2928d [AArch64] Fix -Wparentheses warning with gcc 5.4. NFC
  • ce23772b94 [NFC][InstCombine] Fix typo
  • 52c4e0dd10 [llvm-jitlink] Don't hardcode LLVM version number into the runtime path.
  • fcd0279b17 Revert "[GlobalISel] Add scalar widening for G_MERGE_VALUES destination"
  • 14491b35e0 [AbstractAttributor] Fold __kmpc_parallel_level if possible
  • 6a849a320d [OpenMP] Run rewriteDeviceCodeStateMachine in the Module not CGSCC pass
  • d40841959b [Attributor][FIX] Do not return CHANGED unconditionally
  • 04130ef1fa [Attributor][FIX] Track change status for AAIsDead properly
  • 8a88013900 [gn build] Kind of port c7b3a91017d2 (libclang version script)
  • 5997b61680 [AMDGPU] Add SelectionDAG support for insert_subvector on v4f64
  • ac19c5a542 Revert "Build libSupport with -Werror=global-constructors (NFC)"
  • 64c9fc2e9c [PowerPC] Fix materialization of SP float values on Power10
  • 69e5ee4faa Revert "[AArch64][GlobalISel] Legalize ctpop s128"
  • 1c2cdfcee3 [GlobalISel] Add scalar widening for G_MERGE_VALUES destination
  • b21820b66f [SCEV] Add a comment about invariant in howManyLessThans
  • 18108c50ec [AArch64][GlobalISel] Legalize ctpop s128
  • 8bfb4cf745 [PowerPC] Add pwr7 and pwr10 support to IBM MASSV pass on AIX
  • 07a9552b3b Build libSupport with -Werror=global-constructors (NFC)
  • 6634a71bd1 [AArch64][GlobalISel] Add identity combines to post-legal combiner.
  • 3698f6f3f3 [llvm-objcopy] Fix section group flag read/write when operating on a cross-endian object file
  • 5e3045ed0c [GlobalISel] Add a constant folding combine.
  • 5b24282a50 [WebAssembly] Remove dominator dependency in WasmEHPrepare (NFC)
  • 9b48b7e7be [WebAssembly] Make Emscripten EH work with Emscripten SjLj
  • cc27d61007 [SimplifyCFG] SwitchToLookupTable(): don't increase ret count
  • 5b8d8bfd1b [SimplifyCFG] Drop support for simplifying cond branch to two (different) ret's
  • c0c15a814e [SimplifyCFG] Drop support for duplicating ret's into uncond predecessors
  • 33d35b0a79 [CodeView] Saturate values bigger than supported by APInt.
  • e4830a62b3 [ARM] Fixup vst4 test. NFC
  • bbc51b9f17 [PowerPC]Add addex instruction definition and MC tests
  • 4f1b8e266e [LV] Don't let ForceTargetInstructionCost override Invalid cost.
  • bb39cd345a [AArch64] NFC: Make some AArch64-SVE LoopVectorize tests generic.
  • 2e5bfee63f [SimplifyCFG] Remove stale comment after d7378259aa, NFC
  • a85a7951e1 Fix clang debug info irgen of i128 enums
  • c4acdbbb3c [PowerPC] Add implicit-def RM to instructions mtfsb[01]
  • 71556de73f [OpenMP][NFC] Remove unncessary capture in RAII struct
  • b2d24acf01 [amdgpu] Add 64-bit PC support when expanding unconditional branches.

  This list of changes was auto generated.

  Downloads

  • llvmlibs_mt.7z
    52 MiB

    2026-02-18 19:20:22 -05:00
  • llvmlibs_mt.7z.sha256
    65 B

    2026-02-18 19:20:23 -05:00
  • Source Code (ZIP)
  • Source Code (TAR.GZ)
• custom-build-win f286257383

  Compare

  Stable

  yindo released this 2020-03-03 02:50:32 -05:00 | 27058 commits to master since this release

  📅 Originally published on GitHub: Tue, 03 Mar 2020 09:49:47 GMT
  🏷️ Git tag created: Tue, 03 Mar 2020 07:50:32 GMT

  Changes:

  • 9b52b6c39a Clean up and improve Zen CPU detection
  • bc7b2ed172 Quick hack for Zen 4, more Zen 3 detection
  • e8f2aa0ae4 add raptorlake target
  • 5521155be5 Fix register clobbering on aarch64 GHC when mixing tail/non-tail calls
  • c725f494c9 X86: Avoid converting EVEX to VEX when disp8 would be beneficial
  • eb7a5e5301 Fix tail call guarantee setting for GHC on arm64 backend
  • 509d31ad89 Force ELF on macOS
  • 1c0ca194dc Relax volatile stores (always return)
  • a670c459ea Enable LLVM_USE_PERF
  • 318b8fe374 X86: fixup matchPMADDWD_3

  See More

  • 1cc7bdd501 X86: improve (V)PMADDWD detection (2)
  • 610c27aa1c X86: disable AVX512 truncate with saturation instructions
  • 747fc0efaa lib/Support: Detect AVX-512 on Alderlake CPUs
  • 18f153b33b lib/Support: Add Tigerlake and Alderlake detection
  • 6382230f0e test: Remove test files
  • 05332b78c7 TypeSize.h: add integral constructor to ElementCount
  • 902970b075 IRBuilder.h: remove deprecations
  • c9fceef173 X86: fixup (V)PMADDWD detection
  • f7d625e31a X86: improve (V)PMADDWD detection
  • c36b21c023 X86: modify PreserveAll CC to save full AVX-512 state
  • 2e65b3a418 Fix possible compilation issue with gcc-11
  • 548daf04b5 X86: avoid vector-scalar shifts if splat amount is directly a vector ADD/SUB/AND op.
  • d5bc359dfd X86: add patterns for X86ISD::VSHLV and X86ISD::VSRLV
  • bed700114e X86: add pattern for X86ISD::VSRAV
  • 2ffa82223f X86: expand detectAVGPattern()
  • 5ff8f4151c X86: optimize VSELECT for v16i8 with shl + sign bit test
  • 4743d020ce X86: LowerShift: new algorithm for vector-vector shifts
  • d5b5885c23 Disable GDBRegistrationListener
  • 6516f565ed MCJIT: don't finalize modules on symbol lookup (workaround)
  • d18817ded9 X86: Fix/workaround Small Code Model for JIT
  • 8edd96e498 Set up CI with Azure Pipelines
  • 39c406a58f [AArch64][GlobalISel] Use ZExtValue for zext(xor) when invert tb(n)z
  • be1489a181 [debuginfo-test][cross-project-tests] Release note for new project name
  • aed4e7449f [X86] combineX86ShuffleChain - ensure we only peek through bitcasts to vectors (PR51858)
  • 47e63cb60e [openmp] Apply code change from D109500
  • fca643847d [VPlan] Fix crash caused by not updating all users properly.
  • 7b93a88a1e Revert "[AArch64][GlobalISel] Legalize bswap <2 x i16>"
  • b6614b2353 Revert [MC][ELF] Emit separate unique sections for different flags
  • a967752a95 [X86] Don't clobber EBX in stackprobes
  • 7d56483207 [IR] Handle constant expressions in containsUndefinedElement()
  • bedf13b1a7 [SimplifyCFG] performBranchToCommonDestFolding(): require block-closed SSA form for bonus instructions (PR51125)
  • 5e5115bbd0 [MemCpyOpt] Fix a variety of scalable-type crashes
  • 28d769100d Workaround incorrect types when lowering fixed length gather/scatter
  • b37f5f2114 Inform pass manager when child loops are deleted
  • 601c2dd9dd Fine grain control over some symbol visibility
  • 5f6ef6fbfd [AArch64][SME] Fix imm bug in mov vector to tile aliases
  • 921995afd5 Revert "[HardwareLoops] Change order of SCEV expression construction for InitLoopCount."
  • d21237cb11 [SelectionDAGBuilder] Bugfix in visitInlineAsm()
  • 64f4a52b52 [llvm-objdump] Fix 'llvm-objdump -dr' for executables with relocations
  • 09c230ecec Add llvm-ml to LLVM_TOOLCHAIN_TOOLS (PR50536)
  • e0d7c39869 [AArch64][sve] Prevent incorrect function call on fixed width vector
  • 3c59cf5aa7 [SCEV] Fix applyLoopGuards() with range check idiom (PR51760)
  • 8830999b87 [DAGCombine] Prevent the transform of combine for multi-use operand
  • 6d085fa2f5 [tests] precommit tests for D107692
  • bf02d31357 [docs] Update release notes with items related to Flang
  • 12537ac90c [libomptarget][amdcgn] Only add opt/llvm-link dependency if TARGET is available [ org/D107156#2969862 ]
  • 4b60833c5b [libomptarget][amdcgn] Add build dependency for llvm-link and opt
  • ba85498148 [RISCV] Fix reporting of incorrect commutable operand indices
  • bf7908ca9b [Orc] Enable debug object tests only on x86_64 hosts
  • 683bdb08a0 [Linker] Support weak symbols in nodeduplicate COMDAT group
  • e415eb692a [docs] Mention that the legacy PM is deprecated and will be removed after 14
  • fda0edff63 [NewPM] Add missing LTO ArgPromotion pass
  • 72d2352801 [AlignFromAssume] Bailout w/non-constant alignments (pr51680)
  • 6d1749e6c0 [SelectionDAG] Fix miscompile bugs related to smul.fix.sat with scale zero
  • 4ddceef928 [WebAssembly] Fix FastISel of condition in different block (PR51651)
  • 8fbe4ddc7c [M68k] Update pointer data layout
  • de85b171b7 [M68k][NFC] Rename M68kOperand::Kind to KindTy
  • ab550c798f [CMake] Change -DENABLE_EXPERIMENTAL_NEW_PASS_MANAGER=off to -DLLVM_ENABLE_NEW_PASS_MANAGER=off
  • 79ff6e4b2d [LoopIdiom] Don't transform loop into memmove when load from body has more than one use
  • b61f80921f [NFC][LoopIdiom] Add reproducer of wrong memmove transformation
  • 5aea8f0472 Revert "[RISCV] Fix reporting of incorrect commutable operand indices"
  • 8a19c7f52b ThinLTO: Fix inline assembly references to static functions with CFI
  • 2424302e99 [RISCV] Fix reporting of incorrect commutable operand indices
  • 13fbbfff3a [Coverage][llvm-cov] Correctly export branch coverage in LCOV format
  • fa3a9f0fc7 Revert sharing subprograms across CUs
  • e4e6f3eeff [AArch64] Fix comparison peephole opt with non-0/1 immediate (PR51476)
  • 45d26b8826 [X86][AVX] Extract SUBV_BROADCAST constant bits from just the lower subvector range (PR51281)
  • 4d78ad44fb [ARM][atomicrmw] Fix CMP_SWAP_32 expand assert
  • 8202608068 [PowerPC] Disable CTR Loop generate for fma with the PPC double double type.
  • 22b0c02125 [DAGCombiner] Stop visitEXTRACT_SUBVECTOR creating illegal BITCASTs post legalisation.
  • 0bbcb191e3 [Attributor][FIX] Guard constant casts with type size checks
  • 4345e14522 [InstCombine] avoid infinite loops from min/max canonicalization
  • 5a14ea148e [InstSimplify] fold min/max with limit constant
  • de802b8e6e [InstSimplify] add tests for min/max idioms; NFC
  • 9740b5c5ef [LoopVectorize] Improve vectorisation of some intrinsics by treating them as uniform
  • 00203829b4 [NFC] Clean up tests in test/Transforms/LoopVectorize/assume.ll
  • 3a88dc8338 Add release notes for things relating to MinGW in the release
  • 4a38ef8718 [ELF] Don't emit SHF_GNU_RETAIN on Solaris
  • fd411b2b5d [profile] Fix profile merging with binary IDs
  • 681b643c07 [X86][SchedModel] Add missing ReadAdvance for some arithmetic ops (PR51318 and PR51322).
  • fb132cb74b [MCA][NFC] Add tests for PR51318 and PR51322.
  • db94372a40 [MCA] Simplify the rounding logic used in TimelineView::printWaitTimeEntry.
  • 64a6596867 [Attributor][NFC] Try to make the windows build bots happy
  • d2cc939747 [Attributor][FIX] Handle recurrences (PHIs) in AAPointerInfo explicitly
  • e9a9a807dd [Attributor][FIX] Only avoid visiting PHI uses multiple times (PR51249)
  • 5cf4e36ce1 [Attributor][NFC] Precommit reproducer for PR51249
  • 0c8a79e78d [RISCV] Add scheduling resources for V
  • 11c22c929d Drop LLVM_VERSION_SUFFIX
  • 13e216d251 [llvm] [cmake] Export LLVM_ENABLE_NEW_PASS_MANAGER into LLVMConfig.cmake
  • ee15bdbb06 [AArch64][SVE] Fix assertion failure when lowering fixed length gather/scatter
  • 7d89a50287 [llvm] [lit] Fix inconsistent test order in shtest-keyword-parse-errors
  • f6a86e448a BPF: avoid NE/EQ loop exit condition
  • a558de374f [InstCombine] Fixed select + masked load fold failure
  • 7563d8dfe5 [llvm-rc] Allow specifying language with a leading 0x prefix
  • 910de616a4 [DebugInfo][LSR] Avoid crashes on large integer inputs
  • 46ad88f625 Follow-up to D105207, only salvage affine SCEVs to avoid a crash
  • ff86d9e5f0 [DebugInfo][LoopStrengthReduction] SCEV-based salvaging for LSR
  • a75903d41d [ConstantFold] Get rid of special cases for sizeof etc.
  • f9a2d3e277 [test] Fix tools/gold/X86/comdat-nodeduplicate.ll on non-X86 hosts
  • 8be74d232c Fixing an infinite loop problem in InstCombine
  • 3851813b89 [PredicateInfo] Use Intrinsic::getDeclaration now that it handles unnamed types. [ #2875179 ]
  • e20914df74 Revert "Revert of D49126 [PredicateInfo] Use custom mangling to support ssa_copy with unnamed types."
  • 2b94ecbbe0 [SROA] prevent crash on large memset length (PR50910)
  • 30f0ebbbe2 [Attributor] Don't test internalization in the CGSCC pass.
  • 4f1fd1c209 [Attributor] Change function internalization to not replace uses in internalized callers
  • 710ae2bfd3 [ReleaseNotes] Add scalable matrix extension support to AArch64 changes
  • bafc7f359f [llvm][Release notes] Add AArch64 SVE, PAC and LLDB prebuilt binary
  • aa2a6b072f [llvm][Release notes] Add memory tagging support to lldb changes
  • d4eedb2312 [Analysis] improve function signature checking for snprintf
  • 9770d34891 [OpenMP] Fixing llvm-omp-device-info compilation with runtimes
  • 70f5e23577 [X86][AVX] Add test case for PR51281
  • df3286259b [DAGCombiner] don't try to partially reduce add-with-overflow ops
  • a0686462c3 [AArch64][x86] add tests for add-with-overflow folds; NFC
  • b92c9f9565 [DivRemPairs] make sure we have a valid CFG for hoisting division
  • 7c9c296915 [RISCV] Restrict performANY_EXTENDCombine to prevent an infinite loop.
  • 276fcebbe0 [AArch64] Legalize MVT::i64x8 in DAG isel lowering
  • a50e569197 [AArch64] Add a Machine Value Type for 8 consecutive registers
  • cd0096f439 [DebugInfo][InstrRef] Don't break up ret-sequences on debug-info instrs
  • 183b0c7c98 [AArch64][SVE] Fix incorrect mask type when lowering fixed type SVE gather/scatter
  • 923213f844 test-release.sh: Kill python2
  • 9a10dd5b1c Revert "[DebugInfo][LoopStrengthReduction] SCEV-based salvaging for LSR"
  • c463fa6cad [mlir][openacc] Initial translation for DataOp to LLVM IR
  • 5b7208da36 [OpenMP] Folding threadLimit and numThreads when single value in kernels
  • 66a78bc217 llvm/utils: guarantee revert_checker's revert ordering
  • bab6d38daf [DAGCombiner] Fold SETCC(FREEZE(x),const) to FREEZE(SETCC(x,const)) if SETCC is used by BRCOND
  • 2e57fe1d7d Precommit test files for D105344 (NFC)
  • a6d5003afd [X86] Fix lowering to illegal type in LowerINSERT_VECTOR_ELT
  • 9b53e594b4 Reapply "[Attributor] Disable simplification AAs if a callback is present""
  • 5d447ad589 Revert "[X86] Fix lowering to illegal type in LowerINSERT_VECTOR_ELT"
  • 115164b68b Add llvm::equal convenient wrapper for ranges around std::equal
  • 409f0eedd6 [X86] Fix lowering to illegal type in LowerINSERT_VECTOR_ELT
  • 60850cdc6a [Hexagon] Fix resetting dead registers in DBG_VALUE_LISTs
  • 803630dbd1 [gn build] Port 8a48e6dda9f7
  • 96f821bd7b Revert "[Attributor] Disable simplification AAs if a callback is present"
  • 0042081f58 [Attributor] Verify checkForAllUses return value properly
  • 0b6fb7b1ef [Attributor] Disable simplification AAs if a callback is present
  • 9aacda281c Fix test/Transforms/LoopVectorize/AArch64/strict-fadd-vf1.ll.
  • 091e720912 [gn build] manually port 71909de37495
  • 506c974562 [MLGO] fix silly LLVM_DEBUG misuse
  • f4b486c976 [NFC][MLGO] Debug messages for what inline advisor is selected
  • 8b3f85a32c [PowerPC] Turn deprecated altivec prefetch instrs to nops on AIX
  • 11aa71a71d [x86] update stale code comment; NFC
  • d571039392 [x86] add more tests for cmov and lea; NFC
  • ece3299a71 AMDGPU/GlobalISel: Fix selecting G_SEXTLOAD/G_ZEXTLOAD pre-gfx9
  • 17f139f330 AMDGPU/GlobalISel: Fix wrong addrspace in test MMOs
  • cfbd77a7f1 AMDGPU/GlobalISel: Add a few tests for unaligned truncating stores
  • c0d40a1ca3 Remove unused include that's also a layering violation. NFC.
  • 0a69d4f31a Add test update for a11d9a1f480f which disables fallbacks.
  • 6ce8f2f7c1 [AArch64][GlobalISel] Fix constraining LDXPX intrinsic selection.
  • 2a88f80da4 [ASAN] NFC: Remove redundant variable
  • c563d2cbcd [gn build] Port 02077da7e7a8
  • d9613eb43c [Matrix] Fix shape for factored transpose
  • add64be20a [Matrix] RAUW should only replace an instruction in ShapeMap if supportsShapeInfo
  • 6516543c4b Add jump-threading optimization for deterministic finite automata
  • 6bfc6b8665 [RISCV] Select vector shl by 1 to a vector add.
  • 0f86a54b34 [AArch64] Update and expand min-max cost model test. NFC
  • fd18a762a0 Enabling the copy-constant-to-alloca optimization in more instances
  • cdd50ed2ff [LoopVectorize] Don't interleave scalar ordered reductions for inner loops
  • 3ad8a0b712 Update reduction test. Remove standalone test file
  • 3b4453b968 AMDGPU: Update tests for lower i1 change
  • 8979bda8e8 AMDGPU: Treat IMPLICIT_DEF like a constant lanemask source
  • bb4e957ebf [WebAssembly] Codegen for extmul SIMD instructions
  • 2b967460b4 [SystemZ][z/OS] Initial code to generate assembly files on z/OS
  • 23f3f90569 Strip undef implying attributes when moving calls
  • 05308b27ea Revert "[X86][AVX] Add getBROADCAST_LOAD helper function. NFCI."
  • d809395fde Revert "Revert "[X86][AVX] Add getBROADCAST_LOAD helper function. NFCI.""
  • 9d32182a3a Revert "[X86][AVX] Add getBROADCAST_LOAD helper function. NFCI."
  • adf33c8470 [DebugInfo][InstrRef] Correctly update DBG_PHIs during instr scheduling
  • 64d9d45951 Handle unused variable when assertions are disabled
  • b8af1d6561 [IVDescriptors] Fix bug in checkOrderedReduction
  • fe743a5b25 [DebugInfo][LoopStrengthReduction] SCEV-based salvaging for LSR
  • 47a51741f3 [DebugInfo][InstrRef] Handle llvm.frameaddress intrinsics gracefully
  • e13e00f7b3 [DebugInfo][LoopStrengthReduction] SCEV-based salvaging for LSR
  • 2e863e863d [DebugInfo][LoopStrengthReduction] SCEV-based salvaging for LSR
  • b5cbfeed77 [PowerPC] add more testcases for ld_splat; nfc
  • ac97ba67e8 [LangRef][NFC] Fix variable name in llvm.maxnum docs
  • a99adc793b [X86] Add PR37025 test coverage
  • e1c8650040 [LV] Disable Scalable VFs when tail folding is enabled b/c of low tripcount.
  • 3bc8cd6a0b [GlobalISel] Constant fold G_SITOFP and G_UITOFP in CSEMIRBuilder
  • cf6bdfc026 [SelectionDAG] Support scalable splats in U(ADD|SUB)SAT combines
  • 5afacc5171 [RISCV] Add support for vector saturating add/sub operations
  • c2bd628550 [NFC] Reflow some debug messages.
  • 1ae2c0fb16 [AArch64][SME] Add zero instruction
  • b7fc1f8c1c [RISCV] Add tests showing missed vector saturating add/sub combines
  • 4d9d8789ad [ARM] Implement isLoad/StoreFromStackSlot for MVE stack stores accesses
  • 565fcd6a48 [LoopFlatten] Use SCEV and Loop APIs to identify increment and trip count
  • ee2584c4c2 [docs] Update release notes with all LLVM-C API changes
  • e8ead0fbb2 [Debug-Info][llvm-dwarfdump] Don't try to dump location
  • 10658f9d9e [gn build] Port 2487db1f2862
  • d08c2fabb8 [ORC] Require ExecutorProcessControl when constructing an ExecutionSession.
  • e9073c9b78 [OpenMP] Try to simplify all loads in device code
  • 756885168c [Attributor][FIX] Copy all members in the assignment operator
  • 7e9917e379 [Attributor] Utilize the InstSimplify interface to simplify instructions
  • 86cba7bf5e [InstSimplify] Expose generic interface for replaced operand simplification
  • 35ac57a3f1 [Attributor] Update check lines for all AMDGPU attributor tests
  • 2b8eaac9b5 [Attributor][FIX] Update AMDGPU attributor test
  • 700878ee36 [Coroutine] Record the elided coroutines
  • ffb4271d86 Merge all the llvm-exegesis unit tests into a single binary
  • eca1a76d00 [CSSPGO] Tweak ICP threshold in top-down inliner
  • 4ce16022d3 [Local] Do not introduce a new llvm.trap before unreachable
  • 95f415bd34 [Attributor] Delete dead stores
  • 5ee3942ae7 [Attributor] Introduce getPotentialCopiesOfStoredValue and use it
  • 60670bde11 Build libSupport with -Werror=global-constructors (NFC)
  • eb0ba2928d [AArch64] Fix -Wparentheses warning with gcc 5.4. NFC
  • ce23772b94 [NFC][InstCombine] Fix typo
  • 52c4e0dd10 [llvm-jitlink] Don't hardcode LLVM version number into the runtime path.
  • fcd0279b17 Revert "[GlobalISel] Add scalar widening for G_MERGE_VALUES destination"
  • 14491b35e0 [AbstractAttributor] Fold __kmpc_parallel_level if possible
  • 6a849a320d [OpenMP] Run rewriteDeviceCodeStateMachine in the Module not CGSCC pass
  • d40841959b [Attributor][FIX] Do not return CHANGED unconditionally
  • 04130ef1fa [Attributor][FIX] Track change status for AAIsDead properly
  • 8a88013900 [gn build] Kind of port c7b3a91017d2 (libclang version script)
  • 5997b61680 [AMDGPU] Add SelectionDAG support for insert_subvector on v4f64
  • ac19c5a542 Revert "Build libSupport with -Werror=global-constructors (NFC)"
  • 64c9fc2e9c [PowerPC] Fix materialization of SP float values on Power10
  • 69e5ee4faa Revert "[AArch64][GlobalISel] Legalize ctpop s128"
  • 1c2cdfcee3 [GlobalISel] Add scalar widening for G_MERGE_VALUES destination
  • b21820b66f [SCEV] Add a comment about invariant in howManyLessThans
  • 18108c50ec [AArch64][GlobalISel] Legalize ctpop s128
  • 8bfb4cf745 [PowerPC] Add pwr7 and pwr10 support to IBM MASSV pass on AIX
  • 07a9552b3b Build libSupport with -Werror=global-constructors (NFC)
  • 6634a71bd1 [AArch64][GlobalISel] Add identity combines to post-legal combiner.
  • 3698f6f3f3 [llvm-objcopy] Fix section group flag read/write when operating on a cross-endian object file
  • 5e3045ed0c [GlobalISel] Add a constant folding combine.
  • 5b24282a50 [WebAssembly] Remove dominator dependency in WasmEHPrepare (NFC)
  • 9b48b7e7be [WebAssembly] Make Emscripten EH work with Emscripten SjLj
  • cc27d61007 [SimplifyCFG] SwitchToLookupTable(): don't increase ret count
  • 5b8d8bfd1b [SimplifyCFG] Drop support for simplifying cond branch to two (different) ret's
  • c0c15a814e [SimplifyCFG] Drop support for duplicating ret's into uncond predecessors
  • 33d35b0a79 [CodeView] Saturate values bigger than supported by APInt.
  • e4830a62b3 [ARM] Fixup vst4 test. NFC
  • bbc51b9f17 [PowerPC]Add addex instruction definition and MC tests
  • 4f1b8e266e [LV] Don't let ForceTargetInstructionCost override Invalid cost.
  • bb39cd345a [AArch64] NFC: Make some AArch64-SVE LoopVectorize tests generic.
  • 2e5bfee63f [SimplifyCFG] Remove stale comment after d7378259aa, NFC
  • a85a7951e1 Fix clang debug info irgen of i128 enums
  • c4acdbbb3c [PowerPC] Add implicit-def RM to instructions mtfsb[01]
  • 71556de73f [OpenMP][NFC] Remove unncessary capture in RAII struct
  • b2d24acf01 [amdgpu] Add 64-bit PC support when expanding unconditional branches.
  • 837c44ce91 [TypePromotion] Remove redundant if. NFC

  This list of changes was auto generated.

  Downloads

  • llvmlibs_mt.7z
    52 MiB

    2026-02-18 19:20:13 -05:00
  • llvmlibs_mt.7z.sha256
    65 B

    2026-02-18 19:20:13 -05:00
  • Source Code (ZIP)
  • Source Code (TAR.GZ)
• custom-build f286257383

  Compare

  Stable

  yindo released this 2020-03-03 02:50:32 -05:00 | 27058 commits to master since this release

  📅 Originally published on GitHub: Tue, 03 Mar 2020 09:48:39 GMT
  🏷️ Git tag created: Tue, 03 Mar 2020 07:50:32 GMT

  Changes:

  • 9b52b6c39a Clean up and improve Zen CPU detection
  • bc7b2ed172 Quick hack for Zen 4, more Zen 3 detection
  • e8f2aa0ae4 add raptorlake target
  • 5521155be5 Fix register clobbering on aarch64 GHC when mixing tail/non-tail calls
  • c725f494c9 X86: Avoid converting EVEX to VEX when disp8 would be beneficial
  • eb7a5e5301 Fix tail call guarantee setting for GHC on arm64 backend
  • 509d31ad89 Force ELF on macOS
  • 1c0ca194dc Relax volatile stores (always return)
  • a670c459ea Enable LLVM_USE_PERF
  • 318b8fe374 X86: fixup matchPMADDWD_3

  See More

  • 1cc7bdd501 X86: improve (V)PMADDWD detection (2)
  • 610c27aa1c X86: disable AVX512 truncate with saturation instructions
  • 747fc0efaa lib/Support: Detect AVX-512 on Alderlake CPUs
  • 18f153b33b lib/Support: Add Tigerlake and Alderlake detection
  • 6382230f0e test: Remove test files
  • 05332b78c7 TypeSize.h: add integral constructor to ElementCount
  • 902970b075 IRBuilder.h: remove deprecations
  • c9fceef173 X86: fixup (V)PMADDWD detection
  • f7d625e31a X86: improve (V)PMADDWD detection
  • c36b21c023 X86: modify PreserveAll CC to save full AVX-512 state
  • 2e65b3a418 Fix possible compilation issue with gcc-11
  • 548daf04b5 X86: avoid vector-scalar shifts if splat amount is directly a vector ADD/SUB/AND op.
  • d5bc359dfd X86: add patterns for X86ISD::VSHLV and X86ISD::VSRLV
  • bed700114e X86: add pattern for X86ISD::VSRAV
  • 2ffa82223f X86: expand detectAVGPattern()
  • 5ff8f4151c X86: optimize VSELECT for v16i8 with shl + sign bit test
  • 4743d020ce X86: LowerShift: new algorithm for vector-vector shifts
  • d5b5885c23 Disable GDBRegistrationListener
  • 6516f565ed MCJIT: don't finalize modules on symbol lookup (workaround)
  • d18817ded9 X86: Fix/workaround Small Code Model for JIT
  • 8edd96e498 Set up CI with Azure Pipelines
  • 39c406a58f [AArch64][GlobalISel] Use ZExtValue for zext(xor) when invert tb(n)z
  • be1489a181 [debuginfo-test][cross-project-tests] Release note for new project name
  • aed4e7449f [X86] combineX86ShuffleChain - ensure we only peek through bitcasts to vectors (PR51858)
  • 47e63cb60e [openmp] Apply code change from D109500
  • fca643847d [VPlan] Fix crash caused by not updating all users properly.
  • 7b93a88a1e Revert "[AArch64][GlobalISel] Legalize bswap <2 x i16>"
  • b6614b2353 Revert [MC][ELF] Emit separate unique sections for different flags
  • a967752a95 [X86] Don't clobber EBX in stackprobes
  • 7d56483207 [IR] Handle constant expressions in containsUndefinedElement()
  • bedf13b1a7 [SimplifyCFG] performBranchToCommonDestFolding(): require block-closed SSA form for bonus instructions (PR51125)
  • 5e5115bbd0 [MemCpyOpt] Fix a variety of scalable-type crashes
  • 28d769100d Workaround incorrect types when lowering fixed length gather/scatter
  • b37f5f2114 Inform pass manager when child loops are deleted
  • 601c2dd9dd Fine grain control over some symbol visibility
  • 5f6ef6fbfd [AArch64][SME] Fix imm bug in mov vector to tile aliases
  • 921995afd5 Revert "[HardwareLoops] Change order of SCEV expression construction for InitLoopCount."
  • d21237cb11 [SelectionDAGBuilder] Bugfix in visitInlineAsm()
  • 64f4a52b52 [llvm-objdump] Fix 'llvm-objdump -dr' for executables with relocations
  • 09c230ecec Add llvm-ml to LLVM_TOOLCHAIN_TOOLS (PR50536)
  • e0d7c39869 [AArch64][sve] Prevent incorrect function call on fixed width vector
  • 3c59cf5aa7 [SCEV] Fix applyLoopGuards() with range check idiom (PR51760)
  • 8830999b87 [DAGCombine] Prevent the transform of combine for multi-use operand
  • 6d085fa2f5 [tests] precommit tests for D107692
  • bf02d31357 [docs] Update release notes with items related to Flang
  • 12537ac90c [libomptarget][amdcgn] Only add opt/llvm-link dependency if TARGET is available [ org/D107156#2969862 ]
  • 4b60833c5b [libomptarget][amdcgn] Add build dependency for llvm-link and opt
  • ba85498148 [RISCV] Fix reporting of incorrect commutable operand indices
  • bf7908ca9b [Orc] Enable debug object tests only on x86_64 hosts
  • 683bdb08a0 [Linker] Support weak symbols in nodeduplicate COMDAT group
  • e415eb692a [docs] Mention that the legacy PM is deprecated and will be removed after 14
  • fda0edff63 [NewPM] Add missing LTO ArgPromotion pass
  • 72d2352801 [AlignFromAssume] Bailout w/non-constant alignments (pr51680)
  • 6d1749e6c0 [SelectionDAG] Fix miscompile bugs related to smul.fix.sat with scale zero
  • 4ddceef928 [WebAssembly] Fix FastISel of condition in different block (PR51651)
  • 8fbe4ddc7c [M68k] Update pointer data layout
  • de85b171b7 [M68k][NFC] Rename M68kOperand::Kind to KindTy
  • ab550c798f [CMake] Change -DENABLE_EXPERIMENTAL_NEW_PASS_MANAGER=off to -DLLVM_ENABLE_NEW_PASS_MANAGER=off
  • 79ff6e4b2d [LoopIdiom] Don't transform loop into memmove when load from body has more than one use
  • b61f80921f [NFC][LoopIdiom] Add reproducer of wrong memmove transformation
  • 5aea8f0472 Revert "[RISCV] Fix reporting of incorrect commutable operand indices"
  • 8a19c7f52b ThinLTO: Fix inline assembly references to static functions with CFI
  • 2424302e99 [RISCV] Fix reporting of incorrect commutable operand indices
  • 13fbbfff3a [Coverage][llvm-cov] Correctly export branch coverage in LCOV format
  • fa3a9f0fc7 Revert sharing subprograms across CUs
  • e4e6f3eeff [AArch64] Fix comparison peephole opt with non-0/1 immediate (PR51476)
  • 45d26b8826 [X86][AVX] Extract SUBV_BROADCAST constant bits from just the lower subvector range (PR51281)
  • 4d78ad44fb [ARM][atomicrmw] Fix CMP_SWAP_32 expand assert
  • 8202608068 [PowerPC] Disable CTR Loop generate for fma with the PPC double double type.
  • 22b0c02125 [DAGCombiner] Stop visitEXTRACT_SUBVECTOR creating illegal BITCASTs post legalisation.
  • 0bbcb191e3 [Attributor][FIX] Guard constant casts with type size checks
  • 4345e14522 [InstCombine] avoid infinite loops from min/max canonicalization
  • 5a14ea148e [InstSimplify] fold min/max with limit constant
  • de802b8e6e [InstSimplify] add tests for min/max idioms; NFC
  • 9740b5c5ef [LoopVectorize] Improve vectorisation of some intrinsics by treating them as uniform
  • 00203829b4 [NFC] Clean up tests in test/Transforms/LoopVectorize/assume.ll
  • 3a88dc8338 Add release notes for things relating to MinGW in the release
  • 4a38ef8718 [ELF] Don't emit SHF_GNU_RETAIN on Solaris
  • fd411b2b5d [profile] Fix profile merging with binary IDs
  • 681b643c07 [X86][SchedModel] Add missing ReadAdvance for some arithmetic ops (PR51318 and PR51322).
  • fb132cb74b [MCA][NFC] Add tests for PR51318 and PR51322.
  • db94372a40 [MCA] Simplify the rounding logic used in TimelineView::printWaitTimeEntry.
  • 64a6596867 [Attributor][NFC] Try to make the windows build bots happy
  • d2cc939747 [Attributor][FIX] Handle recurrences (PHIs) in AAPointerInfo explicitly
  • e9a9a807dd [Attributor][FIX] Only avoid visiting PHI uses multiple times (PR51249)
  • 5cf4e36ce1 [Attributor][NFC] Precommit reproducer for PR51249
  • 0c8a79e78d [RISCV] Add scheduling resources for V
  • 11c22c929d Drop LLVM_VERSION_SUFFIX
  • 13e216d251 [llvm] [cmake] Export LLVM_ENABLE_NEW_PASS_MANAGER into LLVMConfig.cmake
  • ee15bdbb06 [AArch64][SVE] Fix assertion failure when lowering fixed length gather/scatter
  • 7d89a50287 [llvm] [lit] Fix inconsistent test order in shtest-keyword-parse-errors
  • f6a86e448a BPF: avoid NE/EQ loop exit condition
  • a558de374f [InstCombine] Fixed select + masked load fold failure
  • 7563d8dfe5 [llvm-rc] Allow specifying language with a leading 0x prefix
  • 910de616a4 [DebugInfo][LSR] Avoid crashes on large integer inputs
  • 46ad88f625 Follow-up to D105207, only salvage affine SCEVs to avoid a crash
  • ff86d9e5f0 [DebugInfo][LoopStrengthReduction] SCEV-based salvaging for LSR
  • a75903d41d [ConstantFold] Get rid of special cases for sizeof etc.
  • f9a2d3e277 [test] Fix tools/gold/X86/comdat-nodeduplicate.ll on non-X86 hosts
  • 8be74d232c Fixing an infinite loop problem in InstCombine
  • 3851813b89 [PredicateInfo] Use Intrinsic::getDeclaration now that it handles unnamed types. [ #2875179 ]
  • e20914df74 Revert "Revert of D49126 [PredicateInfo] Use custom mangling to support ssa_copy with unnamed types."
  • 2b94ecbbe0 [SROA] prevent crash on large memset length (PR50910)
  • 30f0ebbbe2 [Attributor] Don't test internalization in the CGSCC pass.
  • 4f1fd1c209 [Attributor] Change function internalization to not replace uses in internalized callers
  • 710ae2bfd3 [ReleaseNotes] Add scalable matrix extension support to AArch64 changes
  • bafc7f359f [llvm][Release notes] Add AArch64 SVE, PAC and LLDB prebuilt binary
  • aa2a6b072f [llvm][Release notes] Add memory tagging support to lldb changes
  • d4eedb2312 [Analysis] improve function signature checking for snprintf
  • 9770d34891 [OpenMP] Fixing llvm-omp-device-info compilation with runtimes
  • 70f5e23577 [X86][AVX] Add test case for PR51281
  • df3286259b [DAGCombiner] don't try to partially reduce add-with-overflow ops
  • a0686462c3 [AArch64][x86] add tests for add-with-overflow folds; NFC
  • b92c9f9565 [DivRemPairs] make sure we have a valid CFG for hoisting division
  • 7c9c296915 [RISCV] Restrict performANY_EXTENDCombine to prevent an infinite loop.
  • 276fcebbe0 [AArch64] Legalize MVT::i64x8 in DAG isel lowering
  • a50e569197 [AArch64] Add a Machine Value Type for 8 consecutive registers
  • cd0096f439 [DebugInfo][InstrRef] Don't break up ret-sequences on debug-info instrs
  • 183b0c7c98 [AArch64][SVE] Fix incorrect mask type when lowering fixed type SVE gather/scatter
  • 923213f844 test-release.sh: Kill python2
  • 9a10dd5b1c Revert "[DebugInfo][LoopStrengthReduction] SCEV-based salvaging for LSR"
  • c463fa6cad [mlir][openacc] Initial translation for DataOp to LLVM IR
  • 5b7208da36 [OpenMP] Folding threadLimit and numThreads when single value in kernels
  • 66a78bc217 llvm/utils: guarantee revert_checker's revert ordering
  • bab6d38daf [DAGCombiner] Fold SETCC(FREEZE(x),const) to FREEZE(SETCC(x,const)) if SETCC is used by BRCOND
  • 2e57fe1d7d Precommit test files for D105344 (NFC)
  • a6d5003afd [X86] Fix lowering to illegal type in LowerINSERT_VECTOR_ELT
  • 9b53e594b4 Reapply "[Attributor] Disable simplification AAs if a callback is present""
  • 5d447ad589 Revert "[X86] Fix lowering to illegal type in LowerINSERT_VECTOR_ELT"
  • 115164b68b Add llvm::equal convenient wrapper for ranges around std::equal
  • 409f0eedd6 [X86] Fix lowering to illegal type in LowerINSERT_VECTOR_ELT
  • 60850cdc6a [Hexagon] Fix resetting dead registers in DBG_VALUE_LISTs
  • 803630dbd1 [gn build] Port 8a48e6dda9f7
  • 96f821bd7b Revert "[Attributor] Disable simplification AAs if a callback is present"
  • 0042081f58 [Attributor] Verify checkForAllUses return value properly
  • 0b6fb7b1ef [Attributor] Disable simplification AAs if a callback is present
  • 9aacda281c Fix test/Transforms/LoopVectorize/AArch64/strict-fadd-vf1.ll.
  • 091e720912 [gn build] manually port 71909de37495
  • 506c974562 [MLGO] fix silly LLVM_DEBUG misuse
  • f4b486c976 [NFC][MLGO] Debug messages for what inline advisor is selected
  • 8b3f85a32c [PowerPC] Turn deprecated altivec prefetch instrs to nops on AIX
  • 11aa71a71d [x86] update stale code comment; NFC
  • d571039392 [x86] add more tests for cmov and lea; NFC
  • ece3299a71 AMDGPU/GlobalISel: Fix selecting G_SEXTLOAD/G_ZEXTLOAD pre-gfx9
  • 17f139f330 AMDGPU/GlobalISel: Fix wrong addrspace in test MMOs
  • cfbd77a7f1 AMDGPU/GlobalISel: Add a few tests for unaligned truncating stores
  • c0d40a1ca3 Remove unused include that's also a layering violation. NFC.
  • 0a69d4f31a Add test update for a11d9a1f480f which disables fallbacks.
  • 6ce8f2f7c1 [AArch64][GlobalISel] Fix constraining LDXPX intrinsic selection.
  • 2a88f80da4 [ASAN] NFC: Remove redundant variable
  • c563d2cbcd [gn build] Port 02077da7e7a8
  • d9613eb43c [Matrix] Fix shape for factored transpose
  • add64be20a [Matrix] RAUW should only replace an instruction in ShapeMap if supportsShapeInfo
  • 6516543c4b Add jump-threading optimization for deterministic finite automata
  • 6bfc6b8665 [RISCV] Select vector shl by 1 to a vector add.
  • 0f86a54b34 [AArch64] Update and expand min-max cost model test. NFC
  • fd18a762a0 Enabling the copy-constant-to-alloca optimization in more instances
  • cdd50ed2ff [LoopVectorize] Don't interleave scalar ordered reductions for inner loops
  • 3ad8a0b712 Update reduction test. Remove standalone test file
  • 3b4453b968 AMDGPU: Update tests for lower i1 change
  • 8979bda8e8 AMDGPU: Treat IMPLICIT_DEF like a constant lanemask source
  • bb4e957ebf [WebAssembly] Codegen for extmul SIMD instructions
  • 2b967460b4 [SystemZ][z/OS] Initial code to generate assembly files on z/OS
  • 23f3f90569 Strip undef implying attributes when moving calls
  • 05308b27ea Revert "[X86][AVX] Add getBROADCAST_LOAD helper function. NFCI."
  • d809395fde Revert "Revert "[X86][AVX] Add getBROADCAST_LOAD helper function. NFCI.""
  • 9d32182a3a Revert "[X86][AVX] Add getBROADCAST_LOAD helper function. NFCI."
  • adf33c8470 [DebugInfo][InstrRef] Correctly update DBG_PHIs during instr scheduling
  • 64d9d45951 Handle unused variable when assertions are disabled
  • b8af1d6561 [IVDescriptors] Fix bug in checkOrderedReduction
  • fe743a5b25 [DebugInfo][LoopStrengthReduction] SCEV-based salvaging for LSR
  • 47a51741f3 [DebugInfo][InstrRef] Handle llvm.frameaddress intrinsics gracefully
  • e13e00f7b3 [DebugInfo][LoopStrengthReduction] SCEV-based salvaging for LSR
  • 2e863e863d [DebugInfo][LoopStrengthReduction] SCEV-based salvaging for LSR
  • b5cbfeed77 [PowerPC] add more testcases for ld_splat; nfc
  • ac97ba67e8 [LangRef][NFC] Fix variable name in llvm.maxnum docs
  • a99adc793b [X86] Add PR37025 test coverage
  • e1c8650040 [LV] Disable Scalable VFs when tail folding is enabled b/c of low tripcount.
  • 3bc8cd6a0b [GlobalISel] Constant fold G_SITOFP and G_UITOFP in CSEMIRBuilder
  • cf6bdfc026 [SelectionDAG] Support scalable splats in U(ADD|SUB)SAT combines
  • 5afacc5171 [RISCV] Add support for vector saturating add/sub operations
  • c2bd628550 [NFC] Reflow some debug messages.
  • 1ae2c0fb16 [AArch64][SME] Add zero instruction
  • b7fc1f8c1c [RISCV] Add tests showing missed vector saturating add/sub combines
  • 4d9d8789ad [ARM] Implement isLoad/StoreFromStackSlot for MVE stack stores accesses
  • 565fcd6a48 [LoopFlatten] Use SCEV and Loop APIs to identify increment and trip count
  • ee2584c4c2 [docs] Update release notes with all LLVM-C API changes
  • e8ead0fbb2 [Debug-Info][llvm-dwarfdump] Don't try to dump location
  • 10658f9d9e [gn build] Port 2487db1f2862
  • d08c2fabb8 [ORC] Require ExecutorProcessControl when constructing an ExecutionSession.
  • e9073c9b78 [OpenMP] Try to simplify all loads in device code
  • 756885168c [Attributor][FIX] Copy all members in the assignment operator
  • 7e9917e379 [Attributor] Utilize the InstSimplify interface to simplify instructions
  • 86cba7bf5e [InstSimplify] Expose generic interface for replaced operand simplification
  • 35ac57a3f1 [Attributor] Update check lines for all AMDGPU attributor tests
  • 2b8eaac9b5 [Attributor][FIX] Update AMDGPU attributor test
  • 700878ee36 [Coroutine] Record the elided coroutines
  • ffb4271d86 Merge all the llvm-exegesis unit tests into a single binary
  • eca1a76d00 [CSSPGO] Tweak ICP threshold in top-down inliner
  • 4ce16022d3 [Local] Do not introduce a new llvm.trap before unreachable
  • 95f415bd34 [Attributor] Delete dead stores
  • 5ee3942ae7 [Attributor] Introduce getPotentialCopiesOfStoredValue and use it
  • 60670bde11 Build libSupport with -Werror=global-constructors (NFC)
  • eb0ba2928d [AArch64] Fix -Wparentheses warning with gcc 5.4. NFC
  • ce23772b94 [NFC][InstCombine] Fix typo
  • 52c4e0dd10 [llvm-jitlink] Don't hardcode LLVM version number into the runtime path.
  • fcd0279b17 Revert "[GlobalISel] Add scalar widening for G_MERGE_VALUES destination"
  • 14491b35e0 [AbstractAttributor] Fold __kmpc_parallel_level if possible
  • 6a849a320d [OpenMP] Run rewriteDeviceCodeStateMachine in the Module not CGSCC pass
  • d40841959b [Attributor][FIX] Do not return CHANGED unconditionally
  • 04130ef1fa [Attributor][FIX] Track change status for AAIsDead properly
  • 8a88013900 [gn build] Kind of port c7b3a91017d2 (libclang version script)
  • 5997b61680 [AMDGPU] Add SelectionDAG support for insert_subvector on v4f64
  • ac19c5a542 Revert "Build libSupport with -Werror=global-constructors (NFC)"
  • 64c9fc2e9c [PowerPC] Fix materialization of SP float values on Power10
  • 69e5ee4faa Revert "[AArch64][GlobalISel] Legalize ctpop s128"
  • 1c2cdfcee3 [GlobalISel] Add scalar widening for G_MERGE_VALUES destination
  • b21820b66f [SCEV] Add a comment about invariant in howManyLessThans
  • 18108c50ec [AArch64][GlobalISel] Legalize ctpop s128
  • 8bfb4cf745 [PowerPC] Add pwr7 and pwr10 support to IBM MASSV pass on AIX
  • 07a9552b3b Build libSupport with -Werror=global-constructors (NFC)
  • 6634a71bd1 [AArch64][GlobalISel] Add identity combines to post-legal combiner.
  • 3698f6f3f3 [llvm-objcopy] Fix section group flag read/write when operating on a cross-endian object file
  • 5e3045ed0c [GlobalISel] Add a constant folding combine.
  • 5b24282a50 [WebAssembly] Remove dominator dependency in WasmEHPrepare (NFC)
  • 9b48b7e7be [WebAssembly] Make Emscripten EH work with Emscripten SjLj
  • cc27d61007 [SimplifyCFG] SwitchToLookupTable(): don't increase ret count
  • 5b8d8bfd1b [SimplifyCFG] Drop support for simplifying cond branch to two (different) ret's
  • c0c15a814e [SimplifyCFG] Drop support for duplicating ret's into uncond predecessors
  • 33d35b0a79 [CodeView] Saturate values bigger than supported by APInt.
  • e4830a62b3 [ARM] Fixup vst4 test. NFC
  • bbc51b9f17 [PowerPC]Add addex instruction definition and MC tests
  • 4f1b8e266e [LV] Don't let ForceTargetInstructionCost override Invalid cost.
  • bb39cd345a [AArch64] NFC: Make some AArch64-SVE LoopVectorize tests generic.
  • 2e5bfee63f [SimplifyCFG] Remove stale comment after d7378259aa, NFC
  • a85a7951e1 Fix clang debug info irgen of i128 enums
  • c4acdbbb3c [PowerPC] Add implicit-def RM to instructions mtfsb[01]
  • 71556de73f [OpenMP][NFC] Remove unncessary capture in RAII struct
  • b2d24acf01 [amdgpu] Add 64-bit PC support when expanding unconditional branches.
  • 837c44ce91 [TypePromotion] Remove redundant if. NFC

  This list of changes was auto generated.

  Downloads

  • llvmlibs-linux.tar.gz
    34 MiB

    2026-02-18 19:20:04 -05:00
  • llvmlibs-linux.tar.gz.sha256
    65 B

    2026-02-18 19:20:04 -05:00
  • Source Code (ZIP)
  • Source Code (TAR.GZ)