)]}' { "log": [ { "commit": "82e436c17196b649ec5278ee51b995a8b9d0d309", "tree": "aae31d8ff7aa4fa097d4104ab430fc0730be4882", "parents": [ "f7f6e6f4011252d3aacd44e87dd07c263049aad1" ], "author": { "name": "Fangrui Song", "email": "i@maskray.me", "time": "Sat Jun 06 19:41:20 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 19:41:20 2026" }, "message": "[llvm] Remove unused DenseMapInfo::getEmptyKey (#201996)\n\nAfter #201281 DenseMapInfo\u003cT\u003e::getEmptyKey() is no longer used by\nDenseMap. Remove the unused getEmptyKey definitions and dead sentinel\nuses." }, { "commit": "f7f6e6f4011252d3aacd44e87dd07c263049aad1", "tree": "ea31489a8967e16abc84d162f82aa5eff2f5c28b", "parents": [ "4f265ce6730f6b109783a56f4c9251ed537c08c4" ], "author": { "name": "Fangrui Song", "email": "i@maskray.me", "time": "Sat Jun 06 19:33:34 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 19:33:34 2026" }, "message": "[mlir] Remove unused DenseMapInfo::getEmptyKey (#201991)\n\nAfter #201281 DenseMapInfo\u003cT\u003e::getEmptyKey() is no longer used by\nDenseMap. Remove the unused getEmptyKey definitions and dead sentinel\nuses." }, { "commit": "4f265ce6730f6b109783a56f4c9251ed537c08c4", "tree": "9acf17c58dbf322264f6f914a8c769a6de5faea9", "parents": [ "2e0fd9033183952d49ce759328764c3b0c75b94f" ], "author": { "name": "Daniil Dudkin", "email": "unterumarmung@yandex.ru", "time": "Sat Jun 06 19:26:06 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 19:26:06 2026" }, "message": "[clang-tidy] Fix `altera-id-dependent-backward-branch` false positives (#200660)\n\nThe check unconditionally marked the LHS of any assignment from a\nvariable or field as ID-dependent. As a result, ordinary assignments\nlike `x \u003d y` or `i +\u003d chunk_size` could trigger false positives in later\nloop conditions.\n\nOnly propagate ID-dependency when the RHS variable or field is already\nknown ID-dependent.\n\nBased on the fix by @yeputons, with helper renaming and extra tests for\nordinary assignments, fields, macros, aliases, cv/ref cases, lambdas,\ntemplates, concepts, and the #53777 range-for case.\n\nFix #52790\nFix #53777\n\nAssisted by Codex." }, { "commit": "2e0fd9033183952d49ce759328764c3b0c75b94f", "tree": "f65623d3c988ac5e9b92e590a3f300d83b48abe8", "parents": [ "8d5b7f729e07d5394746ae868adabc8b327d94cf" ], "author": { "name": "Florian Hahn", "email": "flo@fhahn.com", "time": "Sat Jun 06 19:25:48 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 19:25:48 2026" }, "message": "[VPlan] Only print original trip count if it is used. (#202069)\n\nSimilarly to other VPlan-scope values, only print trip count when it has\nusers for consistency." }, { "commit": "8d5b7f729e07d5394746ae868adabc8b327d94cf", "tree": "d90af7231cff18727934fa88e86ba7f5b81e449d", "parents": [ "731dd05c34d9bb8d42741b67afc658d228b30e85" ], "author": { "name": "Min-Yih Hsu", "email": "min.hsu@sifive.com", "time": "Sat Jun 06 19:18:30 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 19:18:30 2026" }, "message": "[InstCombine] Fix incorrect insert point when folding zext of deinterleave (#201977)\n\nFix invalid module crash in\nhttps://github.com/llvm/llvm-project/pull/195330#issuecomment-4635245035\n\nThe problem was caused by incorrect IRBuilder insertion point. The\ninsertion point used when generating new instructions might not dominate\nall the de-interleave fields users. This patch fixes this issue by\nexplicitly setting the insertion point to either\n`llvm.vector.deinterleave2` or the earliest `shufflevector` instruction\n(in the de-interleaving group) within the same basicblock." }, { "commit": "731dd05c34d9bb8d42741b67afc658d228b30e85", "tree": "c498a8383de1975dcd503c44df083eaca172eb9c", "parents": [ "8e1afdada5d09cccb18b3ff17dac04e30645b93e" ], "author": { "name": "Sergei Druzhkov", "email": "serzhdruzhok@gmail.com", "time": "Sat Jun 06 18:48:44 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 18:48:44 2026" }, "message": "[lldb] Add nullptr check in GetBroadcaster (#201769)\n\nI recently noticed LLDB crash during execution of `script\nprint(lldb.SBDebugger().GetBroadcaster().GetName())` command:\n```\nPLEASE submit a bug report to https://github.com/llvm/llvm-project/issues/ and include the crash backtrace.\nStack dump:\n0.\tProgram arguments: /home/sergei/llvm-project/build/bin/lldb-dap\n #0 0x000062735c3403d2 llvm::sys::PrintStackTrace(llvm::raw_ostream\u0026, int) (/home/sergei/llvm-project/build/bin/lldb-dap+0x7c3d2)\n #1 0x000062735c33d7ec llvm::sys::RunSignalHandlers() (/home/sergei/llvm-project/build/bin/lldb-dap+0x797ec)\n #2 0x000062735c33d94c SignalHandler(int, siginfo_t*, void*) Signals.cpp:0:0\n #3 0x00007eaa6aa45330 (/lib/x86_64-linux-gnu/libc.so.6+0x45330)\n #4 0x00007eaa6bb0c092 lldb::SBBroadcaster::GetName() const (/home/sergei/llvm-project/build/bin/../lib/liblldb.so.23.0git+0x90c092)\n #5 0x00007eaa6bcb9a5d _wrap_SBBroadcaster_GetName LLDBWrapPython.cpp:0:0\n #6 0x00007eaa6a1df5f5 (/lib/x86_64-linux-gnu/libpython3.12.so.1.0+0x1df5f5)\n #7 0x00007eaa6a182b2c PyObject_Vectorcall (/lib/x86_64-linux-gnu/libpython3.12.so.1.0+0x182b2c)\n #8 0x00007eaa6a11d5ee _PyEval_EvalFrameDefault (/lib/x86_64-linux-gnu/libpython3.12.so.1.0+0x11d5ee)\n #9 0x00007eaa6a2a091f PyEval_EvalCode (/lib/x86_64-linux-gnu/libpython3.12.so.1.0+0x2a091f)\n#10 0x00007eaa6a29c8b0 (/lib/x86_64-linux-gnu/libpython3.12.so.1.0+0x29c8b0)\n#11 0x00007eaa6a11fbd3 _PyEval_EvalFrameDefault (/lib/x86_64-linux-gnu/libpython3.12.so.1.0+0x11fbd3)\n#12 0x00007eaa6c4891b7 lldb_private::ScriptInterpreterPythonImpl::ExecuteOneLine(llvm::StringRef, lldb_private::CommandReturnObject*, lldb_private::ExecuteScriptOptions const\u0026) (/home/sergei/llvm-project/build/bin/../lib/liblldb.so.23.0git+0x12891b7)\n#13 0x00007eaa70326ff5 CommandObjectScriptingRun::DoExecute(llvm::StringRef, lldb_private::CommandReturnObject\u0026) (/home/sergei/llvm-project/build/bin/../lib/liblldb.so.23.0git+0x5126ff5)\n#14 0x00007eaa6bee3739 lldb_private::CommandObjectRaw::Execute(char const*, lldb_private::CommandReturnObject\u0026) (/home/sergei/llvm-project/build/bin/../lib/liblldb.so.23.0git+0xce3739)\n#15 0x00007eaa6bede09a lldb_private::CommandInterpreter::HandleCommand(char const*, lldb_private::LazyBool, lldb_private::CommandReturnObject\u0026, bool) (/home/sergei/llvm-project/build/bin/../lib/liblldb.so.23.0git+0xcde09a)\n#16 0x00007eaa6bb0f0f8 lldb::SBCommandInterpreter::HandleCommand(char const*, lldb::SBExecutionContext\u0026, lldb::SBCommandReturnObject\u0026, bool) (/home/sergei/llvm-project/build/bin/../lib/liblldb.so.23.0git+0x90f0f8)\n#17 0x00007eaa6bb0f265 lldb::SBCommandInterpreter::HandleCommand(char const*, lldb::SBCommandReturnObject\u0026, bool) (/home/sergei/llvm-project/build/bin/../lib/liblldb.so.23.0git+0x90f265)\n#18 0x000062735c3707f3 lldb_dap::RunLLDBCommands[abi:cxx11](lldb::SBDebugger\u0026, lldb::SBMutex, llvm::StringRef, llvm::ArrayRef\u003clldb_dap::protocol::String\u003e const\u0026, bool\u0026, bool, bool) (/home/sergei/llvm-project/build/bin/lldb-dap+0xac7f3)\n#19 0x000062735c3a8019 lldb_dap::EvaluateRequestHandler::Run(lldb_dap::protocol::EvaluateArguments const\u0026) const (/home/sergei/llvm-project/build/bin/lldb-dap+0xe4019)\n#20 0x000062735c3aba78 lldb_dap::RequestHandler\u003clldb_dap::protocol::EvaluateArguments, llvm::Expected\u003clldb_dap::protocol::EvaluateResponseBody\u003e\u003e::operator()(lldb_dap::protocol::Request const\u0026) const (/home/sergei/llvm-project/build/bin/lldb-dap+0xe7a78)\n#21 0x000062735c3ce1bf lldb_dap::BaseRequestHandler::Run(lldb_dap::protocol::Request const\u0026) (/home/sergei/llvm-project/build/bin/lldb-dap+0x10a1bf)\n#22 0x000062735c3577e7 lldb_dap::DAP::HandleObject(std::variant\u003clldb_dap::protocol::Request, lldb_dap::protocol::Response, lldb_dap::protocol::Event\u003e const\u0026) (/home/sergei/llvm-project/build/bin/lldb-dap+0x937e7)\n#23 0x000062735c358705 lldb_dap::DAP::Loop() (/home/sergei/llvm-project/build/bin/lldb-dap+0x94705)\n#24 0x000062735c2ed0c7 main (/home/sergei/llvm-project/build/bin/lldb-dap+0x290c7)\n#25 0x00007eaa6aa2a1ca __libc_start_call_main ./csu/../sysdeps/nptl/libc_start_call_main.h:74:3\n```\nAs far as I understand default constuctors should be covered by fuzzing\ntests, so I don\u0027t know how to write test for that patch." }, { "commit": "8e1afdada5d09cccb18b3ff17dac04e30645b93e", "tree": "7f5127179384717dad21ab7d737d30d13ca38aa4", "parents": [ "7ef1b22c75da98b58e5a91cacabeac8e1356ad8f" ], "author": { "name": "Matt Turner", "email": "mattst88@gmail.com", "time": "Sat Jun 06 18:34:31 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 18:34:31 2026" }, "message": "[ASan] Skip high-shadow and gap setup when HighMem region is empty (#202037)\n\nOn targets where the shadow offset sits above all addressable user\nmemory (e.g. Alpha with ASAN_SHADOW_OFFSET\u003d0x70000000000 and a 42-bit\nuser VAS), kHighMemBeg is set above kHighMemEnd so the HighMem region is\nempty. Since MEM_TO_SHADOW is monotonically increasing, kHighMemBeg \u003e\nkHighMemEnd implies kHighShadowBeg \u003e kHighShadowEnd. Calling\nReserveShadowMemoryRange(kHighShadowBeg, kHighShadowEnd) passes size \u003d\nkHighShadowEnd - kHighShadowBeg + 1, which underflows to a large\nnegative value, and mmap() fails with ENOMEM.\n\nProtectGap is also skipped: there is no meaningful shadow gap between\nLowShadow and an empty HighShadow.\n\nGuard both operations on kHighMemBeg \u003c\u003d kHighMemEnd." }, { "commit": "7ef1b22c75da98b58e5a91cacabeac8e1356ad8f", "tree": "f55c2ccabf38534dab5b42c0aacd7baa0786bc12", "parents": [ "1ab61645d59c25699e32a09f74e3b5a7824915f1" ], "author": { "name": "Elio", "email": "xiongzile@bytedance.com", "time": "Sat Jun 06 18:29:03 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 18:29:03 2026" }, "message": "[AMDGPU][GISel] Fold \u0027min(min(x,y),z)\u0027 and \u0027max(max(x,y),z)\u0027 into min3 and max3 (#200410)\n\nOriginal PR: https://github.com/llvm/llvm-project/pull/124263\nFixes: https://github.com/llvm/llvm-project/issues/123079" }, { "commit": "1ab61645d59c25699e32a09f74e3b5a7824915f1", "tree": "a80689bfaf25ac8972300f57011fd28f1572cb8d", "parents": [ "6b5883b38e158ce4659859ffd78f6ad2869979c1" ], "author": { "name": "Fangrui Song", "email": "i@maskray.me", "time": "Sat Jun 06 18:18:22 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 18:18:22 2026" }, "message": "[BOLT] Remove unused DenseMapInfo::getEmptyKey (#201986)\n\nAfter #201281 DenseMapInfo\u003cT\u003e::getEmptyKey() is no longer used by\nDenseMap. Remove the unused getEmptyKey definitions and dead sentinel\nuses." }, { "commit": "6b5883b38e158ce4659859ffd78f6ad2869979c1", "tree": "35b54a8500f62f7fa3fff044328de6545b0b67ce", "parents": [ "eafc7b672217288c873bea6d2489e93e2e683246" ], "author": { "name": "Fangrui Song", "email": "i@maskray.me", "time": "Sat Jun 06 18:18:12 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 18:18:12 2026" }, "message": "[Polly] Remove unused DenseMapInfo::getEmptyKey (#201992)\n\nAfter #201281 DenseMapInfo\u003cT\u003e::getEmptyKey() is no longer used by\nDenseMap. Remove the unused getEmptyKey definitions and dead sentinel\nuses." }, { "commit": "eafc7b672217288c873bea6d2489e93e2e683246", "tree": "38f01dd7a1709abe782f893a00a57ff1b6205200", "parents": [ "d5a05ac792829360348f57166c0ae2f707d3ba3f" ], "author": { "name": "Fangrui Song", "email": "i@maskray.me", "time": "Sat Jun 06 18:18:02 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 18:18:02 2026" }, "message": "[lld] Remove unused DenseMapInfo::getEmptyKey (#201989)\n\nAfter #201281 DenseMapInfo\u003cT\u003e::getEmptyKey() is no longer used by\nDenseMap. Remove the unused getEmptyKey definitions and dead sentinel\nuses." }, { "commit": "d5a05ac792829360348f57166c0ae2f707d3ba3f", "tree": "403126559da81c8328d14e4f9c32357dfc98a44b", "parents": [ "4bdd4eb56704cdbd5d74c6576c06219628cf2255" ], "author": { "name": "Fangrui Song", "email": "i@maskray.me", "time": "Sat Jun 06 18:17:46 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 18:17:46 2026" }, "message": "[Target] Remove unused DenseMapInfo::getEmptyKey (#201993)\n\nAfter #201281 DenseMapInfo\u003cT\u003e::getEmptyKey() is no longer used by\nDenseMap. Remove the unused getEmptyKey definitions and dead sentinel\nuses." }, { "commit": "4bdd4eb56704cdbd5d74c6576c06219628cf2255", "tree": "61015de7706420ed0a9d619c6aef8f36d7f30bf1", "parents": [ "42ec0c66867ff23642d5745368191e9f7d2b155b" ], "author": { "name": "Fangrui Song", "email": "i@maskray.me", "time": "Sat Jun 06 18:00:24 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 18:00:24 2026" }, "message": "[ADT] Don\u0027t use getEmptyKey() when hashing std::optional (#202002)\n\nDenseMapInfo\u003cstd::optional\u003cT\u003e\u003e::getHashValue() calls T\u0027s getEmptyKey()\nfor the nullopt case. This blocks removing the now-unused getEmptyKey()\nfrom DenseMapInfo specializations: a leaf type that drops getEmptyKey()\nfails to compile wherever std::optional\u003cT\u003e is used as a DenseMap key\n(e.g. DenseMapInfo\u003cmlir::spirv::StorageClass\u003e).\n\nCompute the hash directly instead. Prerequisite for the getEmptyKey()\nremoval series; #201281 made getEmptyKey() unused by DenseMap." }, { "commit": "42ec0c66867ff23642d5745368191e9f7d2b155b", "tree": "f3a316896805d216964afafcd220ba953a0ba863", "parents": [ "3ab0c5109ed2bfe3baffbd8f62cd85de3983daab" ], "author": { "name": "Fangrui Song", "email": "i@maskray.me", "time": "Sat Jun 06 17:59:27 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 17:59:27 2026" }, "message": "[flang] Remove unused DenseMapInfo::getEmptyKey (#201988)\n\nAfter #201281 DenseMapInfo\u003cT\u003e::getEmptyKey() is no longer used by\nDenseMap. Remove the unused getEmptyKey definitions and dead sentinel\nuses." }, { "commit": "3ab0c5109ed2bfe3baffbd8f62cd85de3983daab", "tree": "e212a5d085360aba1530c355bda3a3440746d638", "parents": [ "9aef31502094e2b5c997683a6d624aaf39f5bd50" ], "author": { "name": "Nikita Taranov", "email": "nickita.taranov@gmail.com", "time": "Sat Jun 06 16:28:44 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 16:28:44 2026" }, "message": "[libc] Add missing `dl_iterate_phdr` dependencies to avoid spurious build failures (#201574)\n\nExample:\nhttps://github.com/llvm/llvm-project/actions/runs/26945498241/job/79504837451?pr\u003d201452\n\n\"Spurious\" means, in this case, that the build may succeed or fail\ndepending on whether the files were generated before the dependent is\nbuilt." }, { "commit": "9aef31502094e2b5c997683a6d624aaf39f5bd50", "tree": "d9a69b4f40ebcc34ca552e76424de629d3abf8f4", "parents": [ "c1ee07b95724d80634a399ff84e0e398021862d4" ], "author": { "name": "Jonas Devlieghere", "email": "jonas@devlieghere.com", "time": "Sat Jun 06 16:26:32 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 16:26:32 2026" }, "message": "[lldb] Add SBDebugger::SetTerminalDimensions to set width and height atomically (#201965)\n\nTerminal width and height were communicated to the debugger separately,\nvia SetTerminalWidth() and SetTerminalHeight(). Each notified the\nIOHandler and the statusline, so on a resize they recomputed their\nlayout twice: once with one dimension updated and the other still stale.\n\nAdd Debugger::SetTerminalDimensions(width, height) (exposed through\nSBDebugger) that updates both properties before notifying, and\nreimplement the single-axis setters and the driver\u0027s resize handler in\nterms of it.\n\nAlso fix SBDebugger::GetTerminalHeight(), which returned the width." }, { "commit": "c1ee07b95724d80634a399ff84e0e398021862d4", "tree": "cd6816c886b74a52347b6ce5cc8fe98b9ffd6e81", "parents": [ "e6fab598acd8a0026ddc28eab78885d6f23e08f4" ], "author": { "name": "Alexey Bataev", "email": "a.bataev@outlook.com", "time": "Sat Jun 06 16:26:14 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 16:26:14 2026" }, "message": "Revert \"[SLP]Keep reuse mask in sync when reordering split node operand\"\n\nThis reverts commit dee1687bdba79e729b4ddf3e2c37ff9b5766dc75 to pacify\nbuildbots after failures in\nhttps://lab.llvm.org/buildbot/#/builders/25/builds/18282 and others.\n\nReviewers: \n\nPull Request: https://github.com/llvm/llvm-project/pull/202057" }, { "commit": "e6fab598acd8a0026ddc28eab78885d6f23e08f4", "tree": "67b2983d8add37a794797aaa952065c857e0c18b", "parents": [ "35d520a2cb8b3b445f419fb15d674941b2da38e9" ], "author": { "name": "Dave Lee", "email": "davelee.com@gmail.com", "time": "Sat Jun 06 16:11:00 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 16:11:00 2026" }, "message": "[lldb] Change Symbol size assert to 64 bit only (#202042)\n\nFollow up to\nhttps://github.com/llvm/llvm-project/pull/200919#issuecomment-4635479078" }, { "commit": "35d520a2cb8b3b445f419fb15d674941b2da38e9", "tree": "4a59e674f20787c13abb1b733e1f4f5895cd7102", "parents": [ "79da521f4e5adba781e62cfefd673ccfd8d168b0" ], "author": { "name": "Rahul Joshi", "email": "rjoshi@nvidia.com", "time": "Sat Jun 06 16:03:21 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 16:03:21 2026" }, "message": "[NFC][LLVM] Introduce `IIT_MATCH` to represents `LLVMMatchType` (#202034)\n\nCurrently, the fully dependent identity type `LLVMMatchType` is\nrepresented in the IIT encoding table as `IIT_ANY` with `AK_MatchType`\nargument kind. Instead, add a new IIT code `IIT_MATCH` to represent such\ndependent types, so that `IIT_ANY` is used to represent just the core\noverload types." }, { "commit": "79da521f4e5adba781e62cfefd673ccfd8d168b0", "tree": "694b7adb99a3d7f9669b1906ae54310613463d52", "parents": [ "5458641e2e81dd77b7145a78b0aac7ba3f0c1bdd" ], "author": { "name": "Tom Stellard", "email": "tstellar@redhat.com", "time": "Sat Jun 06 15:42:03 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 15:42:03 2026" }, "message": "workflows/libcxx-run-benchmarks: Remove template expansion (#200282)\n\nhttps://docs.zizmor.sh/audits/#template-injection\n\nhttps://github.com/llvm/llvm-project/security/code-scanning/1648\nhttps://github.com/llvm/llvm-project/security/code-scanning/1649\nhttps://github.com/llvm/llvm-project/security/code-scanning/1650\nhttps://github.com/llvm/llvm-project/security/code-scanning/1651" }, { "commit": "5458641e2e81dd77b7145a78b0aac7ba3f0c1bdd", "tree": "ea47c1c8ff788e09c319fd8642364fc64156bee1", "parents": [ "4706906ba256bfbb3590ae1eb05ef9cd4b92421a" ], "author": { "name": "David Green", "email": "david.green@arm.com", "time": "Sat Jun 06 15:39:10 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 15:39:10 2026" }, "message": "[AArch64] Fix and vs or in tryCombineExtendRShTrunc (#202053)\n\nThis condition should be checking that both are true." }, { "commit": "4706906ba256bfbb3590ae1eb05ef9cd4b92421a", "tree": "84a1dec643b3f06a774868dfdac6851806ee3e81", "parents": [ "6808d14c92eaaa4f8e80dbc9e5786511e8eec458" ], "author": { "name": "Vassil Vassilev", "email": "v.g.vassilev@gmail.com", "time": "Sat Jun 06 15:19:13 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 15:19:13 2026" }, "message": "[clang-repl] Fix Value\u0027s move ctor releasing storage on construction (#200888)\n\nValue::Value(Value \u0026\u0026) called Release() on the just-moved-into storage,\ndecrementing the refcount to zero on the only remaining reference.\nSubsequent reads -- including ~Value() running clear(), which calls\nRelease() a second time on the now-freed allocation -- hit\nuse-after-free.\n\nThe move should transfer the existing reference: the source clears\nIsManuallyAlloc so its destructor will not Release, and *this assumes\nownership of the same refcount. Neither side needs to Retain or Release\nto keep the count correct.\n\nAdd a regression test exercising move-construction, move-assignment, and\nfollow-on copy-construction on a K_PtrOrObj Value. AddressSanitizer\ncatches the bug without the fix." }, { "commit": "6808d14c92eaaa4f8e80dbc9e5786511e8eec458", "tree": "d98c6fd4fe71fa28a3a3ee43e5e39546cb5ca53b", "parents": [ "6eb47bb35bed321cbed22f46de4e301152dbe544" ], "author": { "name": "Fuad Ismail", "email": "fuad1502@gmail.com", "time": "Sat Jun 06 15:04:09 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 15:04:09 2026" }, "message": "[InstCombine] Drop `ninf` FMF when input element can be `Inf` in shuffle-select transform (#201315)\n\nSolves https://github.com/llvm/llvm-project/issues/74326\n\nWhen binary operation has `ninf` FMF, but the input does not have\n`nofpclass(inf)`, we should not propagate the `ninf` FMF. Because the\ntransformation may produce poison value when the input has an `Inf`\nelement, whereas the original code will simply pass through the `Inf`\nelement.\n\nAlive proof: https://alive2.llvm.org/ce/z/nkv-vE" }, { "commit": "6eb47bb35bed321cbed22f46de4e301152dbe544", "tree": "575fe43441bce3b110315be771d3a9f04940b40a", "parents": [ "2853284556509da5de1b52cba488e053dcf2ffc5" ], "author": { "name": "Sean Clarke", "email": "seanedwardsclarke@gmail.com", "time": "Sat Jun 06 14:55:41 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 14:55:41 2026" }, "message": "[InstCombine] Migrate undef -\u003e poison only for certain cast-related optimizations (#201631)\n\nFurther deprecate UndefValue by restricting several related\noptimizations in InstCombineCasts to PoisonValue only. Update regression\ntests to reflect these changes." }, { "commit": "2853284556509da5de1b52cba488e053dcf2ffc5", "tree": "83055cfb0a4f081163a50a6b0d9ea802ebef2fd1", "parents": [ "428a03d5ff9352a1da801bf2a6a94db982dee79c" ], "author": { "name": "mbhade-amd", "email": "mbhade@amd.com", "time": "Sat Jun 06 14:12:01 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 14:12:01 2026" }, "message": "[VectorCombine] foldShuffleOfShuffles - replace IR with poison on the !NewX path (#201826)\n\nThe function returns bool. The early return `return\nPoisonValue::get(ShuffleDstTy);` relies on implicit pointer-to-bool\nconversion of a non-null PoisonValue*, so the caller records a change\nand invalidates non-CFG analyses while the original shuffle is never\nreplaced.\n\nCall replaceValue(I, *PoisonValue::get(ShuffleDstTy)) and return true so\nthe fold actually fires. Downstream InstCombine usually cleans up the\nresidual shuffle via simplifyShuffleVectorInst, which masks the bug at\n-O2 but leaves vector-combine producing the wrong result on its own.\n\nAdds a regression test covering the case where both inner shuffles have\nall-poison masks, so the merge loop never assigns NewX.\n\nThis was found as part of jlebar\u0027s X86 LLVM bug hunt / FuzzX effort:\nhttps://github.com/SemiAnalysisAI/FuzzX/tree/master/x86 :\n[024-foldshuffleofshuffles-poison-bool-cast](https://github.com/SemiAnalysisAI/FuzzX/tree/master/x86/bugs/024-foldshuffleofshuffles-poison-bool-cast)" }, { "commit": "428a03d5ff9352a1da801bf2a6a94db982dee79c", "tree": "f0da8ca08e6f1f16b2666e4c6d2163a8ddb6afa2", "parents": [ "7564cffdad802e80a595b2b3ce032fab7656df35" ], "author": { "name": "Amr Hesham", "email": "amr96@programmer.net", "time": "Sat Jun 06 13:54:34 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 13:54:34 2026" }, "message": "[CIR] Implement CompoundAssignLValue for Atomic Complex (#201895)\n\nImplement CompoundAssignLValue support for Atomic Complex\n\nIssue https://github.com/llvm/llvm-project/issues/192331" }, { "commit": "7564cffdad802e80a595b2b3ce032fab7656df35", "tree": "dc58fe3a12d32980f44f61e97610092182d6d915", "parents": [ "3ca2f009b8d6eae9e098dbdfe9674739dc6cff96" ], "author": { "name": "Michael Kruse", "email": "llvm-project@meinersbur.de", "time": "Sat Jun 06 13:33:11 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 13:33:11 2026" }, "message": "[Flang][Driver] Fix -foffload-device misspelling (#201857)\n\n#200863 added a new `-foffload-device` argument for informing the\nfrontend that it compiling for the device-side (and as a consequence\nmust not overwrite any module files compiled for the host), but the\ndriver was mistakenly adding `-offload-device`.\n\nAlso fix the condition and add a regression test for the driver." }, { "commit": "3ca2f009b8d6eae9e098dbdfe9674739dc6cff96", "tree": "ef9ff87453b0d0cc68ac48301924724b3f9bf2f1", "parents": [ "dee1687bdba79e729b4ddf3e2c37ff9b5766dc75" ], "author": { "name": "Arseniy Obolenskiy", "email": "arseniy.obolenskiy@amd.com", "time": "Sat Jun 06 13:27:38 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 13:27:38 2026" }, "message": "[mlir][SPIR-V] Convert complex.abs (#202026)" }, { "commit": "dee1687bdba79e729b4ddf3e2c37ff9b5766dc75", "tree": "d0d78a21b553a291c410a49e55d9a3c86930b43f", "parents": [ "1f44aeb56183e87547c752b27c879fd1c0299593" ], "author": { "name": "Alexey Bataev", "email": "a.bataev@outlook.com", "time": "Sat Jun 06 13:20:42 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 13:20:42 2026" }, "message": "[SLP]Keep reuse mask in sync when reordering split node operand\n\nWhen reorderBottomToTop uses an operand order for a split vectorize\nnode and the operand has both reordered and reused scalars, only the\nreorder indices were cleared while the reuse mask was left stale. This\ndiverged the split node scalars from the operand effective order.\nFold the reorder into the reuse mask (getCommonMask), reorder it by the\nused mask, then clear the reorder indices, so the operand effective\norder stays consistent with the reordered split node.\n\nFixes #202003\n\nReviewers: \n\nPull Request: https://github.com/llvm/llvm-project/pull/202041" }, { "commit": "1f44aeb56183e87547c752b27c879fd1c0299593", "tree": "83b90aa502fef49a852519aa2f5c1f4c2a7dbc03", "parents": [ "0373a653ecfd55c2ed1c014fb07e3872583af736" ], "author": { "name": "David Green", "email": "david.green@arm.com", "time": "Sat Jun 06 13:14:56 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 13:14:56 2026" }, "message": "[AArch64] Protect against v3i64-\u003ev3i8 truncates in combineI8TruncStore (#202039)\n\nWe were previously creating invalid bitcasts, protect against that by\nmaking\nsure that the type is legal." }, { "commit": "0373a653ecfd55c2ed1c014fb07e3872583af736", "tree": "d164c650f63ddcef3d636a5bc8884cd09aa35db3", "parents": [ "d54ff4c9b91250b93cf6b05377246fbad9e028ce" ], "author": { "name": "Amr Hesham", "email": "amr96@programmer.net", "time": "Sat Jun 06 12:54:59 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 12:54:59 2026" }, "message": "[CIR][NFC] Fix converting AtomicType after RecordType modification (#202036)\n\nFix the conversion of the AtomicType after the change in the structure\nin #199790 and #200668\n\nFix #202031" }, { "commit": "d54ff4c9b91250b93cf6b05377246fbad9e028ce", "tree": "809a891811c101032f9875a249f37ce9df7687b0", "parents": [ "48b138533bd62599247d5b24e913ddfe0384cb2d" ], "author": { "name": "hpkfft.com", "email": "paul@hpkfft.com", "time": "Sat Jun 06 12:30:00 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 12:30:00 2026" }, "message": "Add _MM_FROUND_TO_NEAREST_TIES_EVEN to avx512fintrin.h (#99691)\n\nIntrinsics such as `_mm512_add_round_ps` take a rounding mode argument\nto specify the floating point rounding mode. This, and similar\ninstructions, do NOT round their result to an integer. Thus it is\ninappropriate for user code to specify the existing\n`_MM_FROUND_TO_NEAREST_INT` when desiring to round to the nearest\nfloating point number. This commit adds a suitable macro definition." }, { "commit": "48b138533bd62599247d5b24e913ddfe0384cb2d", "tree": "c8813f199bdd3167cc4dd2b7dff5fe1d25493816", "parents": [ "3a247473cf32b29d254827e1c03b88a444438a35" ], "author": { "name": "Sudharsan Veeravalli", "email": "svs@qti.qualcomm.com", "time": "Sat Jun 06 12:26:02 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 12:26:02 2026" }, "message": "[InstCombine] Fix UB in align-assume check (#201985)\n\nWhen we have a `NULL` pointer the `1ULL \u003c\u003c computeKnownBits(RK.WasOn,\nII).countMinTrailingZeros()` check becomes `1ULL \u003c\u003c 64` which is UB.\n\nWe hit the following error in our downstream sanitizer builder: \n\n\u003e llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp: runtime error:\nshift exponent 64 is too large for 64-bit type \u0027unsigned long long\u0027\n\nTests were generated using an AI." }, { "commit": "3a247473cf32b29d254827e1c03b88a444438a35", "tree": "5a44b868f1c3ec175f75910c76ee9fd346063d0d", "parents": [ "e133cb4f27525e05fbcb04994ed8767140f62db9" ], "author": { "name": "Simon Pilgrim", "email": "llvm-dev@redking.me.uk", "time": "Sat Jun 06 12:01:44 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 12:01:44 2026" }, "message": "[DAG] Use TLO.LegalTypes() instead of AfterLegalizeTypes (#201840)\n\nFix typo from #178617 - AfterLegalizeTypes is an enum constant, not an actual check for legalised types" }, { "commit": "e133cb4f27525e05fbcb04994ed8767140f62db9", "tree": "908a74e7ca6a50aacf6f9583b56a7e84d1012096", "parents": [ "6729f3b5a2d8650432e8415a5289fff42c660527" ], "author": { "name": "Alexey Bataev", "email": "a.bataev@outlook.com", "time": "Sat Jun 06 11:22:28 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 11:22:28 2026" }, "message": "[SLP] Recompute copyable operand deps for nodes sharing an instruction\n\nWhen an instruction is vectorized in several nodes and one models an operand as\ncopyable while another (built later) uses it directly, the operand\u0027s dependency\ncount missed the direct def-use edge and the scheduler decremented it more times\nthan its count, tripping the unscheduled-deps assertion. Defer such operand dep\nrecomputation unconditionally via RecalcCopyableOperandDeps instead of the narrow\nIsDuplicateCopyableNode gate.\n\nFixes #201855\n\nReviewers: \n\nPull Request: https://github.com/llvm/llvm-project/pull/202032" }, { "commit": "6729f3b5a2d8650432e8415a5289fff42c660527", "tree": "5bc4f0253c797743a4b53b3fa4cd54abd82c6a3b", "parents": [ "fbe6b5d8536181ab42b0856cf747e09c4c29a175" ], "author": { "name": "David Green", "email": "david.green@arm.com", "time": "Sat Jun 06 10:51:27 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 10:51:27 2026" }, "message": "[AArch64] Use 64bit mask size for detecting extending Ands. (#202029)" }, { "commit": "fbe6b5d8536181ab42b0856cf747e09c4c29a175", "tree": "b0815482f35c96f5067485f0d86fe12e452b0c47", "parents": [ "81e51e0aaf7219ebf51b52eae12d5f94e888b9bf" ], "author": { "name": "David Green", "email": "david.green@arm.com", "time": "Sat Jun 06 10:16:27 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 10:16:27 2026" }, "message": "[AArch64] Change postinc index types to uint64_t (#202024)\n\nThe uint32_t could overflow, make sure we do not throw away the high\nbits by\nusing a uint64_t." }, { "commit": "81e51e0aaf7219ebf51b52eae12d5f94e888b9bf", "tree": "ebd59d4e6f9019e9e39446f897b20dd1f4e96d35", "parents": [ "3fed38f4b8aa6c4a8ab1a9a864e11394eb3716d1" ], "author": { "name": "Paul Osmialowski", "email": "pawel.osmialowski@arm.com", "time": "Sat Jun 06 09:46:45 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 09:46:45 2026" }, "message": "[openmp][omptest] Include cstdlib for malloc() (#202021)\n\nThis is to address the error appearing when building this code with\nsomewhat more recent compilers:\n\n```\nUse of undeclared identifier \u0027malloc\u0027\n```\n\nSuch inclusion has already been added to the OmptTester.cpp file." }, { "commit": "3fed38f4b8aa6c4a8ab1a9a864e11394eb3716d1", "tree": "513e4b23a38e70a463e2f9fd5c00a8b571c374cd", "parents": [ "570e532d5639c5da8746ffa8551259e742b51572" ], "author": { "name": "Petar Avramovic", "email": "Petar.Avramovic@amd.com", "time": "Sat Jun 06 09:42:54 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 09:42:54 2026" }, "message": "AMDGPU/GlobalISel: Fix tail call target in vgprs (#201873)\n\nInsert readfirstlane, similar to SDAG version from #110984." }, { "commit": "570e532d5639c5da8746ffa8551259e742b51572", "tree": "c345ea519cce52501c14f61518d1dac04f03ed1d", "parents": [ "86fc55edc125fd86e2aebab278650bf724a8c1c5" ], "author": { "name": "Ralf Jung", "email": "post@ralfj.de", "time": "Sat Jun 06 08:21:30 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 08:21:30 2026" }, "message": "[llubi] explain Byte.TagValue encoding trick (#201863)\n\nThis took me a while to understand in #185977 so let\u0027s make it more\nexplicit why `TagValue` can be so small." }, { "commit": "86fc55edc125fd86e2aebab278650bf724a8c1c5", "tree": "984a50f9ae09108092ee554413cd49f83e8ecea7", "parents": [ "f8ee61b393dc5701d1556f22f28849ddfdae1495" ], "author": { "name": "Madhur Amilkanthwar", "email": "madhura@nvidia.com", "time": "Sat Jun 06 07:48:30 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 07:48:30 2026" }, "message": "[LoopFusion] Remove unused DataLayout parameter (NFC) (#202009)\n\nThe LoopFuser constructor took a DataLayout reference that was never\nstored or used, and run() computed it solely to pass it in. Drop both." }, { "commit": "f8ee61b393dc5701d1556f22f28849ddfdae1495", "tree": "1121a8342e4c2e113f8475d60353c85b456355fa", "parents": [ "cff8815ebb23354ac506e6b299cf74cd60227163" ], "author": { "name": "David Green", "email": "david.green@arm.com", "time": "Sat Jun 06 07:43:17 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 07:43:17 2026" }, "message": "[X86][GlobalISel] Remove dependency on legal ruleset (#197374)\n\nThis fills in always legal rules, to remove the dependency on the legacy\nruleset. I\u0027m not sure about the truncate rule but all tests pass. This\nis not guaranteed to be all the rules, just the ones that appear in\ntests." }, { "commit": "cff8815ebb23354ac506e6b299cf74cd60227163", "tree": "a7ba02acf5a89192002767531f9fabc84f8c6c64", "parents": [ "43df6450eb37a47b89c027828cd35a4ab399640f" ], "author": { "name": "Oleksandr Tarasiuk", "email": "oleksandr.tarasiuk@outlook.com", "time": "Sat Jun 06 07:08:47 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 07:08:47 2026" }, "message": "[Clang] support C23 printf width length modifiers (#199991)\n\nThis patch adds `-Wformat` support for the C23 `wN` and `wfN` length\nmodifiers in `printf`/`scanf` format strings. #116962" }, { "commit": "43df6450eb37a47b89c027828cd35a4ab399640f", "tree": "32b37c817ff059a0326ab08dd9ea124a08444a4b", "parents": [ "18b3f1f44d19ccef8f78d833b8cebd6ffd19f24d" ], "author": { "name": "paperchalice", "email": "liujunchang97@outlook.com", "time": "Sat Jun 06 06:40:42 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 06:40:42 2026" }, "message": "[Target] Remove `Target::resetTargetOptions` (#201825)\n\nThis function shouldn\u0027t exist, now it is an empty function, remove it." }, { "commit": "18b3f1f44d19ccef8f78d833b8cebd6ffd19f24d", "tree": "74af359039443afb1e91643db0a2e04641681acd", "parents": [ "f1267565f55ce3a57bc17a0114a07b6a22709a16" ], "author": { "name": "Ian Anderson", "email": "iana@apple.com", "time": "Sat Jun 06 06:37:50 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 06:37:50 2026" }, "message": "Revert \"[clang][driver][darwin] Hold onto full triples in Darwin SDKPlatformInfo (#200896)\" (#202010)\n\nThis doesn\u0027t work for 32 bit arm because that usually gets converted to\nthumb-apple-os, and that doesn\u0027t match arm-apple-os from\nSDKSettings.json.\n\nThis reverts commit b89bb06afd069aa1b5e9f05ab692b3e6b41318c0." }, { "commit": "f1267565f55ce3a57bc17a0114a07b6a22709a16", "tree": "1f0e22356a05fbceff72a5a985ef4861725c73e8", "parents": [ "26ffc71afa7c9a4a9904742016b343de6f2ea4e6" ], "author": { "name": "Robert Imschweiler", "email": "robert.imschweiler@amd.com", "time": "Sat Jun 06 05:50:59 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 05:50:59 2026" }, "message": "[OpenMP][offload] use per-type checks for multiple_reductions.cpp (#201045)" }, { "commit": "26ffc71afa7c9a4a9904742016b343de6f2ea4e6", "tree": "25d694d982418422f5d707596d5271d2e13732cd", "parents": [ "11d7939060e874cdc8378e2b711c719b0d4ba8e1" ], "author": { "name": "owenca", "email": "owenpiano@gmail.com", "time": "Sat Jun 06 05:09:59 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 05:09:59 2026" }, "message": "[clang-format] Disallow breaking before/after ## (#200721)\n\nFixes #199775" }, { "commit": "11d7939060e874cdc8378e2b711c719b0d4ba8e1", "tree": "e09fba2ccceef2c097bdbc9a143f6f9a88cf0a16", "parents": [ "978191848cdb929f85c76bbf54b0666971328c59" ], "author": { "name": "owenca", "email": "owenpiano@gmail.com", "time": "Sat Jun 06 05:03:18 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 05:03:18 2026" }, "message": "[clang-format] Keep C++20 module/import decls on a single line (#199459)\n\nThis patch fixes #193676.\n\n- Added `UnwrappedLineParser::parseModuleDecl()` to parse C++20 module\ndeclarations.\n- Adapted `parseCppModuleImport()` from #193834 and renamed it to\n`parseImportDecl()`.\n- Used the test cases from the same PR.\n- Removed the invalid test cases and fixed an incorrect one in\n`FormatTest.cpp`.\n\n---------\n\nCo-authored-by: Björn Schäpers \u003cgithub@hazardy.de\u003e" }, { "commit": "978191848cdb929f85c76bbf54b0666971328c59", "tree": "a3a124b0193de8caa0a7911d254e44d60a15a2a5", "parents": [ "54ed4695d46a178b80578a3114b6dd70a8c59ec2" ], "author": { "name": "Sirui Mu", "email": "msrlancern@gmail.com", "time": "Sat Jun 06 04:00:51 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 04:00:51 2026" }, "message": "[CIR] Initialization of atomic aggregates with padding (#200668)\n\nThis patch adds support for the initialization of atomic aggregates with\npadding. The changes include:\n\n- During CIRGen, the type `_Atomic(T)` is represented by a CIR struct\n`{T, sint8[padding_size]}` if the size of `_Atomic(T)` does not match\nthe size of `T`. `padding_size` is the difference between the size of\n`_Atomic(T)` and `T`.\n- CIRGen for the initialization process is updated to handle the\ninitialization of such CIR struct values." }, { "commit": "54ed4695d46a178b80578a3114b6dd70a8c59ec2", "tree": "533459aace0b79487831842d13f0b3a2e5fdef63", "parents": [ "d08a2a43abd91865002a42938b60f172527defa3" ], "author": { "name": "dyung", "email": "douglas.yung@sony.com", "time": "Sat Jun 06 03:02:24 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 03:02:24 2026" }, "message": "Revert \"[clang] Adding an Atomic Line Logger\" (#201984)\n\nReverts llvm/llvm-project#195885\n\nThis is causing a build failure on a Windows bot running VS2019:\nhttps://lab.llvm.org/buildbot/#/builders/46/builds/36187" }, { "commit": "d08a2a43abd91865002a42938b60f172527defa3", "tree": "dfa2a53dd99f3847b9db576c31dc5465f1fcb57a", "parents": [ "3e2ccf5a4d45f98bb1f8506f064e55279b244480" ], "author": { "name": "Ryosuke Niwa", "email": "rniwa@webkit.org", "time": "Sat Jun 06 02:42:35 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 02:42:35 2026" }, "message": "[WebKit checkers] Treat the return value of an instance method as an unsafe pointer origin (#160569)" }, { "commit": "3e2ccf5a4d45f98bb1f8506f064e55279b244480", "tree": "0becf26106ec6895e07ef2dca1b6bddda41b2c48", "parents": [ "7389aa2ef380ca2b64d8fa34b633e5d2a4efef0a" ], "author": { "name": "Nikolas Klauser", "email": "nikolasklauser@berlin.de", "time": "Sat Jun 06 02:34:31 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 02:34:31 2026" }, "message": "[libc++] Assume that \u003catomic\u003e is available (#199674)\n\nWe always define either `_LIBCPP_HAS_C_ATOMIC_IMP` or\n`_LIBCPP_HAS_GCC_ATOMIC_IMP`, so we can remove any special handling of\nnot having an `\u003catomic\u003e` header." }, { "commit": "7389aa2ef380ca2b64d8fa34b633e5d2a4efef0a", "tree": "86fa2b2ccdc2c1bddf61102caebee137e3b7d4cb", "parents": [ "ea6943c45e0880f384c9db374a60968349732947" ], "author": { "name": "Alexey Bader", "email": "alexey.bader@intel.com", "time": "Sat Jun 06 01:59:16 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 01:59:16 2026" }, "message": "[Frontend][Offloading] Fix GCC 7 build error in ArchiveLinker (#201978)\n\nGCC 7 cannot perform implicit move construction when converting\n`ResolvedInputs` to `Expected\u003cResolvedInputs\u003e`." }, { "commit": "ea6943c45e0880f384c9db374a60968349732947", "tree": "a3b03ef3a784e6126fa5ba6e894eabac4114ec94", "parents": [ "4113577578485d6b6e8cb398aa69637263fdb64f" ], "author": { "name": "Craig Topper", "email": "craig.topper@sifive.com", "time": "Sat Jun 06 01:56:12 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 01:56:12 2026" }, "message": "[X86] Remove stray :w from a comment. NFC (#201982)" }, { "commit": "4113577578485d6b6e8cb398aa69637263fdb64f", "tree": "ea10b2cad3f57b2752c065e0c02a5c0ff872ea5b", "parents": [ "f53c19ebd81d372429475950bf1314b9d01f2adc" ], "author": { "name": "Justin Lebar", "email": "justin.lebar@gmail.com", "time": "Sat Jun 06 01:55:05 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 01:55:05 2026" }, "message": "[TableGen] Recompute only the affected UberSet when inheriting reg units (#200962)\n\nCodeGenRegBank::computeRegUnitWeights() runs a fixpoint over all registers;\nnormalizeWeight() calls the global computeUberWeights() -- which rescans\nevery UberRegSet, every register, and all of their register units -- each time\na register inherits register units from its subregisters.\n\nMost of the time, we do better by just recomputing one register\u0027s\nUberSet.\n\nOn AMDGPU (21266 registers) with this change, the \"Compute reg unit\nweights\" phase drops from 3.19s to 0.70s (4.5x speedup) and\n-gen-register-info improves overall from ~16.4s to ~14.0s." }, { "commit": "f53c19ebd81d372429475950bf1314b9d01f2adc", "tree": "d81afc1120e9a523e7a1c92c52052984976af6f5", "parents": [ "ee5e682f4db85afca52ac6519a4b9bf0bfab4cf9" ], "author": { "name": "Vladimir Vereschaka", "email": "vvereschaka@accesssoftek.com", "time": "Sat Jun 06 01:44:47 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 01:44:47 2026" }, "message": "Revert \"[clang-cl] Add new option `/pathmap:\u003cfrom\u003e\u003d\u003cto\u003e` to replace the path prefix \u003cfrom\u003e with \u003cto\u003e.\" (#201981)\n\nReverts llvm/llvm-project#198664\n\nCauses test failures on\n[llvm-clang-aarch64-darwin](https://lab.llvm.org/buildbot/#/builders/190)\nbot." }, { "commit": "ee5e682f4db85afca52ac6519a4b9bf0bfab4cf9", "tree": "11471386c7597d675ed1aabc52146d29567ccac2", "parents": [ "7ef07bca9b05b4a3435aa857dbd417c2aec40c22" ], "author": { "name": "David Rivera", "email": "davidriverg@gmail.com", "time": "Sat Jun 06 00:52:21 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 00:52:21 2026" }, "message": "[CIR] Centralize dialect registration across CIR tools (#200266)" }, { "commit": "7ef07bca9b05b4a3435aa857dbd417c2aec40c22", "tree": "1eef59ba1ebe3dbdbf5e0f317e1cd20845e0d1ba", "parents": [ "832f4c41ae172950bbb5797803d6ec1a94edd146" ], "author": { "name": "Daniel Thornburgh", "email": "dthorn@google.com", "time": "Sat Jun 06 00:28:51 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 00:28:51 2026" }, "message": "[compiler-rt] Add missing test stdlib.h include (#201972)\n\nFixes test after libc++ PR #195509 which drops transitive includes." }, { "commit": "832f4c41ae172950bbb5797803d6ec1a94edd146", "tree": "89e606410ffc885c88c8cb68e60af61f32f69b53", "parents": [ "8fa4d4490ddaca238b5c420fbf17020e85ede2fc" ], "author": { "name": "Vladimir Vereschaka", "email": "vvereschaka@accesssoftek.com", "time": "Sat Jun 06 00:25:22 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 00:25:22 2026" }, "message": "[clang-cl] Add new option `/pathmap:\u003cfrom\u003e\u003d\u003cto\u003e` to replace the path prefix \u003cfrom\u003e with \u003cto\u003e. (#198664)\n\nThis option matches MSVC options and does the path substitution for the\nfile references in the preprocessor macros, debug and coverage information.\n\nThis option acts as a clang\u0027s ``-ffile-prefix-map\u003dvalue`` and with some\nknown differences in behaviour with original CL\u0027s option that do not affect \nthe functionality:\n* nomalizes the macro prefix map pathes -- removes `./` and uses the target\u0027s\nplatform-specific path separator character when exanding the preprocessor\nmacros -- ``-ffile-reproducible`` (but not the debug and coverage prefix maps).\n* does not require ``/experimental:deterministic`` as by MSVC. It needed for \nremoving a hostname from a mangling hash gen, but clang-cl does not use\na hostname when generates the hashes.\n\nKnown issues:\n * does not remap the pathes within PCH/PCM files." }, { "commit": "8fa4d4490ddaca238b5c420fbf17020e85ede2fc", "tree": "0b5df06578f8125496469b434322f37af82b0d3e", "parents": [ "139bcc17805161ddc168ef5363969b2b94242516" ], "author": { "name": "Andy Kaylor", "email": "akaylor@nvidia.com", "time": "Sat Jun 06 00:17:00 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 00:17:00 2026" }, "message": "[CIR] Fix CIR Test failures after unnamed_addr restored to vtables (#201962)\n\nWe had previously removed the `unnamed_addr` keyword check from our\nvtable checks for OGCG because it was temporarily not emitted in that\ncase. The OGCG output has been modified again, so we need to update our\nchecks again.\n\nThe CIR output has not changed." }, { "commit": "139bcc17805161ddc168ef5363969b2b94242516", "tree": "fcb78d0e49e57e61d480e9ef3ae7bf73fa46a86d", "parents": [ "8b46bd51399d3e49c601d9958b2e3ff5c1a706cd" ], "author": { "name": "Vladislav Dzhidzhoev", "email": "vdzhidzhoev@accesssoftek.com", "time": "Sat Jun 06 00:14:50 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 00:14:50 2026" }, "message": "[DirectX] Generate compiler version part in llc (#199699)\n\nThis change modifies DXContainerGlobals pass to generate compiler\nversion (VERS) part in DXContainer.\nVERS part allows consumers to information about compiler version used to\nbuild shader.\n\nWhen debug info PDB file creation will be implemented, VERS part should\ngo to PDB file." }, { "commit": "8b46bd51399d3e49c601d9958b2e3ff5c1a706cd", "tree": "da88e724186f82f8314abe638d4357e6429d5c6d", "parents": [ "dca9eaa98fd71535e8d283ae390c96266c5b9b8c" ], "author": { "name": "Jonas Devlieghere", "email": "jonas@devlieghere.com", "time": "Sat Jun 06 00:10:25 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Sat Jun 06 00:10:25 2026" }, "message": "[lldb] Skip plugin teardown when exiting without Terminate (#201739)\n\n`import lldb` auto-calls SBDebugger::Initialize() but never Terminate().\nThe g_debugger_lifetime is a deliberately-leaked ManagedStatic, so the\nPluginInstances containers are still populated when their static\ndestructors run at process exit. That tripped the \"forgot to unregister\nplugin?\" assert, and once the assert was gated the dynamically-loaded\nplugin map\u0027s PluginInfo terminate callbacks ran against PluginInstances\nmutexes that had already been destroyed.\n\nThis only surfaces with LLDB_ENABLE_DYNAMIC_SCRIPTINTERPRETERS, where\n`_lldb` is the script-interpreter plugin dylib and liblldb comes in as a\ndependency that exit() finalizes. A static build leaks identically but\nnever reaches those destructors, so the bug stayed latent.\n\nTrack an explicit lifecycle (Uninitialized/Initialized/Terminated) in a\nsingle never-destroyed PluginRegistry. ~PluginInstances only checks for\nleftover registrations once Terminate() has run, and the map (never torn\ndown at exit) only runs its terminate callbacks during an explicit\nclear, while every container is still alive." }, { "commit": "dca9eaa98fd71535e8d283ae390c96266c5b9b8c", "tree": "e7f8380f99217f137e3bfb687b2883b4bd6f3541", "parents": [ "d0397a6d7d78888e278823c4d32749f8c896ef9c" ], "author": { "name": "Alexey Bader", "email": "alexey.bader@intel.com", "time": "Fri Jun 05 23:42:11 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 23:42:11 2026" }, "message": "[clang-sycl-linker] Add mising dependency on BitReader (#201961)\n\nThis fixes the shared library build configuration." }, { "commit": "d0397a6d7d78888e278823c4d32749f8c896ef9c", "tree": "c032829f17958d9a4c290cc763dee84b85515121", "parents": [ "7bf820c20f08a56eeef027838b992708e5654ff7" ], "author": { "name": "Qiongsi Wu", "email": "qiongsiwu@gmail.com", "time": "Fri Jun 05 23:31:21 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 23:31:21 2026" }, "message": "[clang] Adding an Atomic Line Logger (#195885)\n\nThis PR adds an atomic line logger to `clang`. \n\nSituations have arisen where `clang` performs multi-threaded tasks (such\nas dependency scanning), and race conditions may happen. Such race\nconditions are difficult to debug using either `lldb` or with\n`llvm::errs()`.\n\nThis logger provides atomic logging per line to a file on disk with time\nstamps at each line to facilitate such investigations. Specifically, the\nlogger is designed with the following properties:\n\n1. Each line is atomically written to the backing file. This avoids\nconcurrent writes making the output text interleaving.\n2. Each line is prefixed with a timestamp, a process ID and a thread ID.\n3. `LogLine` implements a `\u003c\u003c` operator to allow arbitrary printable\ntypes to be piped into it.\n4. The `LogLine`\u0027s user does not need to check if it is setup or valid.\nA LogLine is always valid and can always accept input from `\u003c\u003c`. It\nbecomes a no-op if the `LogLine` object is returned from a default\nconstructed `AtomicLineLogger`.\n5. The write happens when a `LogLine` object goes out of scope. \n\nThe logger is inspired by the\n[OnDiskCASLogger](https://github.com/llvm/llvm-project/blob/09abee845d2136630fc3f50524148daa55a740a8/llvm/include/llvm/CAS/OnDiskCASLogger.h#L33).\nA followup PR https://github.com/llvm/llvm-project/pull/195896 wires up\nthis logger to clang\u0027s dependency scanning stack.\n\nAssisted-by: claude-opus-4.6\n\nrdar://39907408" }, { "commit": "7bf820c20f08a56eeef027838b992708e5654ff7", "tree": "11deebaec53b155503405df66a6276f117044650", "parents": [ "985e49989fa6ac24a1318969f6865bb5407574c6" ], "author": { "name": "Alexey Bataev", "email": "a.bataev@outlook.com", "time": "Fri Jun 05 23:19:36 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 23:19:36 2026" }, "message": "[SLP] Fix crash in ordered reduction with loop-carried chain root\n\ntryToReduceOrdered built the buildTree ignore list only from the current\nwindow\u0027s reduced values, omitting the chain root when its leaf is a\ntrailing scalar outside the window.\nbuildTree then vectorized the root through the phi cycle,\nvectorizeTree unlinked it, and emitReduction inserted at the dangling\ninsertion point.\nBuild the ignore list from the full ReductionOps chain, matching the\nassociative-reduction path, so the root stays scalar to fix a crash.\n\nReviewers: \n\nPull Request: https://github.com/llvm/llvm-project/pull/201963" }, { "commit": "985e49989fa6ac24a1318969f6865bb5407574c6", "tree": "1634cae246b95298fafaab3f6a4824a1276c6cf1", "parents": [ "e8753fdde70b8a2bd0d3219000228704587935bd" ], "author": { "name": "Wenju He", "email": "wenju.he@intel.com", "time": "Fri Jun 05 23:18:19 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 23:18:19 2026" }, "message": "[libclc][Test] Don\u0027t hard code `clang` path in update_libclc_tests.py (#201806)\n\nIn #201773, build folder is `redhat-linux-build` rather than `build`.\nDetect clang from PATH. Also add --clang-binary option to specify clang path." }, { "commit": "e8753fdde70b8a2bd0d3219000228704587935bd", "tree": "a3b974152911ee971a8bf7ee98663f2b1664a578", "parents": [ "517a0755f5d128d950fc8a08b2bd45e629e40a95" ], "author": { "name": "Wenju He", "email": "wenju.he@intel.com", "time": "Fri Jun 05 23:17:55 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 23:17:55 2026" }, "message": "[libclc][Test] Add -fno-discard-value-names to RUN line (#201808)\n\nEntry block label could be missing. Add -fno-discard-value-names to make\nsure it exists.\n\nResolves #201773" }, { "commit": "517a0755f5d128d950fc8a08b2bd45e629e40a95", "tree": "c989677b7dbe43c3b7ce34d1c1e47712affdb69f", "parents": [ "961d5b8ac7deb754a8b66a86ee53c9e77cb6769f" ], "author": { "name": "Wenju He", "email": "wenju.he@intel.com", "time": "Fri Jun 05 23:17:42 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 23:17:42 2026" }, "message": "[libclc] Fix LIBCLC_OUTPUT_LIBRARY_DIR when find_package(LLVM) fails (#201816)\n\nIf we configure llvm from llvm-project/build folder, find_package(LLVM)\nsucceeds, `LLVM_LIBRARY_OUTPUT_INTDIR` is `build/./lib` and\n`LLVM_BINARY_DIR` is `build`.\n\nHowever, in #201773, find_package(LLVM) fails due to\nLLVM_LIBDIR_SUFFIX\u003d64, LLVM_LIBRARY_OUTPUT_INTDIR is\n`redhat-linux-build/runtimes/runtimes-nvptx64-nvidia-cuda-bins/lib64`\nand LLVM_BINARY_DIR is `redhat-linux-build`.\n\nUse LLVM_BINARY_DIR, which is stable for both cases, for libclc output dir.\n\nFixes #201773" }, { "commit": "961d5b8ac7deb754a8b66a86ee53c9e77cb6769f", "tree": "b2d24d6629fff9e3e91a86b26e0c47beef0a9bdd", "parents": [ "b7152ff7026a05282b6ae91ccf150ede0217b08a" ], "author": { "name": "Michael Jones", "email": "michaelrj@google.com", "time": "Fri Jun 05 23:15:00 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 23:15:00 2026" }, "message": "[libc] Fix clock_gettime on arm32 (#201959)\n\nAfter removing 32 bit time_t the buildbot has been failing due to a\nsegfault. This PR fixes it. Not the cleanest fix but I don\u0027t want the\nbuildbot failing over the weekend.\n\nAssisted-by: Automated tooling, human reviewed." }, { "commit": "b7152ff7026a05282b6ae91ccf150ede0217b08a", "tree": "6533b99558be59212ae4f70753b22c69ea58ba79", "parents": [ "8aafa50c7a2dfb8ca1d5cdf8980f7f2d259779f5" ], "author": { "name": "forking-google-bazel-bot[bot]", "email": "265904573+forking-google-bazel-bot[bot]@users.noreply.github.com", "time": "Fri Jun 05 23:08:23 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 23:08:23 2026" }, "message": "[Bazel] Fixes 8aafa50 (#201958)\n\nThis fixes 8aafa50c7a2dfb8ca1d5cdf8980f7f2d259779f5.\n\nCo-authored-by: Google Bazel Bot \u003cgoogle-bazel-bot@google.com\u003e" }, { "commit": "8aafa50c7a2dfb8ca1d5cdf8980f7f2d259779f5", "tree": "8830c0fa104310b36d8db8d52480b754f611397e", "parents": [ "1dae5dee50e73d18ef42027bc3bd853a9ae740ce" ], "author": { "name": "Alexey Bader", "email": "alexey.bader@intel.com", "time": "Fri Jun 05 22:41:04 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 22:41:04 2026" }, "message": "[clang][sycl][nvlink] Share static library linking in Frontend/Offloading (#201253)\n\nMove clang-nvlink-wrapper\u0027s archive member selection engine into a new\nshared library in llvm/lib/Frontend/Offloading (ArchiveLinker.h/.cpp)\nand use it from both clang-nvlink-wrapper and clang-sycl-linker, adding\nstatic library (.a) and -l support to the SYCL linker.\n\nThe shared llvm::offloading::resolveArchiveMembers() API:\n- Searches -L paths for -l library names (lib\u003cname\u003e.a or :\u003cname\u003e)\n- Expands archives, honouring --whole-archive/--no-whole-archive\n- Runs a symbol-driven fixed-point loop to extract only the archive\n members that resolve undefined symbols\n- Returns the resolved MemoryBuffers and symbol table; the symbol table\n is consumed by clang-nvlink-wrapper\u0027s LTO resolution pass\n\nclang-sycl-linker gains -l, --whole-archive/--no-whole-archive, and -u\noptions (added to SYCLLinkOpts.td). The previous --bc-library option\nhas been removed in favor of the standard -l mechanism.\n\nBug fixes included:\n\n* Fix dangling StringRef UB: Args.getAllArgValues() returns a temporary\n vector; retain it in ForcedUndefStorage so the StringRefs\n remain valid through the resolveArchiveMembers call (both tools).\n* Fix assert crash in clang-sycl-linker when all positional inputs are\n non-existent: return a proper error instead of propagating an empty\n buffer vector to linkInputs.\n* Fixed forced undefined symbol handling: corrected -u option \n processing to properly handle symbols.\n\nCo-Authored-By: Claude" }, { "commit": "1dae5dee50e73d18ef42027bc3bd853a9ae740ce", "tree": "17496ede53969165e546797c6fe894e54152ad24", "parents": [ "488875cedfbea669afa93298a0a98594c1a9cb92" ], "author": { "name": "Daniel Thornburgh", "email": "dthorn@google.com", "time": "Fri Jun 05 22:40:19 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 22:40:19 2026" }, "message": "[compiler-rt] Add missing cstdlib include to signal_line.cpp test (#201955)\n\nFixes test after libc++ PR #195509 which drops transitive includes." }, { "commit": "488875cedfbea669afa93298a0a98594c1a9cb92", "tree": "eac03a4cd492d3d095b958bf82a44513e46e2361", "parents": [ "cab8d8c9e268d1d94c56e3068377fd7c006cdd7d" ], "author": { "name": "Paweł Bylica", "email": "pawel@hepcolgum.band", "time": "Fri Jun 05 22:37:47 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 22:37:47 2026" }, "message": "[X86] Fix crash on BT bit index narrower than log2(BW) (#201793)\n\npeekThroughBitPosExtTrunc asserted that peeling\nTRUNCATE/ZERO_EXTEND/ANY_EXTEND never drops below log2(BW) bits, but a\nBT bit index can be e.g. a zext of i1. Stop peeking there instead: below\nlog2(BW) bits a peeled value no longer determines the bit index on its\nown, so comparing peeled values would also be unsound.\n\nFixes #201444." }, { "commit": "cab8d8c9e268d1d94c56e3068377fd7c006cdd7d", "tree": "9c46ba16f5aeb7f6c22551d759bb450522961838", "parents": [ "638f3ae61de291f85b711f6e3597fae6edd29986" ], "author": { "name": "Reid Kleckner", "email": "rkleckner@nvidia.com", "time": "Fri Jun 05 22:26:28 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 22:26:28 2026" }, "message": "[docs] Migrate 20 LLVM docs to markdown (#201465)\n\nAn LLM assisted in the rewrite. I didn\u0027t use pandoc or any other tools,\nand I think there are fewer artifacts. I validated these docs by opening\nall 20 HTML files in my browser.\n\nFiles:\n- DebuggingLLVM.md\n- FAQ.md\n- CompilerWriterInfo.md\n- GettingStartedVS.md\n- NewPassManager.md\n- MemorySSA.md\n- LoopTerminology.md\n- Lexicon.md\n- MIRLangRef.md\n- CoverageMappingFormat.md\n- InstrProfileFormat.md\n- BitCodeFormat.md\n- Atomics.md\n- GarbageCollection.md\n- AliasAnalysis.md\n- LibFuzzer.md\n- Extensions.md\n- CommandLine.md\n- ExceptionHandling.md\n- Coroutines.md\n\nTracking issue: #201242" }, { "commit": "638f3ae61de291f85b711f6e3597fae6edd29986", "tree": "354b498999e366b65c676d99b1ca5257cde9bd37", "parents": [ "ba7b74ff70328b3dff4fd3a33ee513342c453ec5" ], "author": { "name": "Reid Kleckner", "email": "rkleckner@nvidia.com", "time": "Fri Jun 05 22:24:49 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 22:24:49 2026" }, "message": "[docs] Rename 20 key LLVM docs to Markdown (#201464)\n\nThis may break the docs build temporarily, but there will be a follow-up\nPR that fixes it immediately after. The rename is being committed first\nto aid review and source code archaelogy.\n\nTracking issue: #201242" }, { "commit": "ba7b74ff70328b3dff4fd3a33ee513342c453ec5", "tree": "f9229cbd11559fb372f02a6e363c1b708f07dedd", "parents": [ "4f73c8195b51042d349916c162663252c7f60eb7" ], "author": { "name": "Evgenii Kudriashov", "email": "evgenii.kudriashov@intel.com", "time": "Fri Jun 05 22:22:18 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 22:22:18 2026" }, "message": "[X86][GlobalISel] Adding missing Query to G_BUILD_VECTOR legalization (#199435)\n\nOnce G_BUILD_VECTOR was fixed, legalize-undef-vec-scaling.mir falled\ninto endless legalization loop as incoming MIR doesn\u0027t match correct\nlowering of the type." }, { "commit": "4f73c8195b51042d349916c162663252c7f60eb7", "tree": "f2d8a4a309bef7444d16ae894a2baff4d161de2e", "parents": [ "80ad2966e2902ec378050228321080eff3f4702e" ], "author": { "name": "Jan Svoboda", "email": "jan_svoboda@apple.com", "time": "Fri Jun 05 22:06:57 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 22:06:57 2026" }, "message": "[clang][lex] Store `HeaderFileInfo` in a `MapVector` (#201936)\n\nThis is a re-commit of PR #200968 (f357a47) that was reverted by PR\n#201702 (adaa9e0). One call-site of\n`HeaderSearch::forEachExistingLocalFileInfo()` in `ASTWriter` relies on\nstable order, which the `DenseMap` didn\u0027t provide. This PR switches to\n`MapVector`, which has the same performance characteristics as the\noriginal solution at the cost of a theoretical memory usage increase.\n\nOriginal PR description:\n\nCalling `FileManager::GetUniqueIDMapping()` during modular builds gets\nvery expensive if the `FileManager` has seen lots of files. This\nfunction is used in two places in the `ASTWriter` to look up\n`HeaderFileInfo` in `HeaderSearch`.\n\nThis PR changes the storage of `HeaderFileInfo` from\n`FileEntry::getUID()`-indexed `std::vector\u003cT\u003e` to\n`llvm::DenseMap\u003cFileEntryRef, T\u003e`, improving scanning performance by\n~2.5%." }, { "commit": "80ad2966e2902ec378050228321080eff3f4702e", "tree": "5ccb515086a8234d6ccfd62e8b444cca56003e45", "parents": [ "e9a4dd0b325c8d171e68662fbea554a86757a79f" ], "author": { "name": "Arseniy Obolenskiy", "email": "arseniy.obolenskiy@amd.com", "time": "Fri Jun 05 22:00:11 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 22:00:11 2026" }, "message": "Reland \"Remove the optional bitcast between a musttail call and its ret\" (#201280) (#201642)\n\nThis relands #201280, which was reverted in #201479 because it broke the\nsanitizer buildbot\n\nThe revert was caused by a heap-use-after-free in\n`BitcodeReader::materialize`\n\nFix: The CallBase branch could have erased I before we read it in the\nnew conditional" }, { "commit": "e9a4dd0b325c8d171e68662fbea554a86757a79f", "tree": "df89441801cddfa70f1ffa33e216a36337fc59e6", "parents": [ "3b89e77151214ec62546f7437b4d5e07b1c2ce1d" ], "author": { "name": "Daniel Thornburgh", "email": "dthorn@google.com", "time": "Fri Jun 05 21:52:40 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 21:52:40 2026" }, "message": "[compiler-rt] Add stdlib include to OutOfProcessFuzzTarget.cpp (#201944)\n\nFixes test after libc++ PR #195509 which drops transitive includes." }, { "commit": "3b89e77151214ec62546f7437b4d5e07b1c2ce1d", "tree": "3eaa1c61e7951e1c9efbea1cea3f6c9bc4bb9d1e", "parents": [ "de59f9ed12db9d47ad41ad44d54ec604ef8841cb" ], "author": { "name": "Florian Mayer", "email": "fmayer@google.com", "time": "Fri Jun 05 21:37:08 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 21:37:08 2026" }, "message": "[compiler-rt] [sanitizers] Add includes for symbolizer (#201929)\n\nAfter ce5b2e876494cb95f02d9f915081e2b8781e74d1 this stopped building." }, { "commit": "de59f9ed12db9d47ad41ad44d54ec604ef8841cb", "tree": "9813ee7e0c9579fc4ab3f344044a4ae8ea4cc856", "parents": [ "35901313800ea6e6cbeb9226e51c7c4b29bfc40e" ], "author": { "name": "Vladislav Dzhidzhoev", "email": "vdzhidzhoev@accesssoftek.com", "time": "Fri Jun 05 21:28:36 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 21:28:36 2026" }, "message": "[DirectX][ObjectYAML] Add SRCI part support (#199446)\n\nAdd support for DXContainer SRCI in the ObjectYAML pipeline so it can be\nrepresented in structured YAML and round-tripped through\nyaml2obj/obj2yaml.\n\nSRCI is emitted into shader PDB files. It stores shader source code that\nis displayed in PIX, and shader compilation flags." }, { "commit": "35901313800ea6e6cbeb9226e51c7c4b29bfc40e", "tree": "ff091a08734b02f4da66cc5ba08b0fb4bb75b4fb", "parents": [ "276cc2fac75a476824ea25b5349dfbac0d5b5545" ], "author": { "name": "Arseniy Obolenskiy", "email": "arseniy.obolenskiy@amd.com", "time": "Fri Jun 05 21:27:58 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 21:27:58 2026" }, "message": "[mlir][SPIR-V] Add OpenCL cbrt op (#201914)" }, { "commit": "276cc2fac75a476824ea25b5349dfbac0d5b5545", "tree": "ef2b78c247a00f4844cfc843950aae64fa4712fc", "parents": [ "c3e9d05ce8044612ecae2eb27dd5ca52b612a8ac" ], "author": { "name": "Harald van Dijk", "email": "hdijk@accesssoftek.com", "time": "Fri Jun 05 21:08:12 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 21:08:12 2026" }, "message": "[DirectX] Rewrite DIGlobalVariableExpression to DXDIGlobalVariable (#199791)\n\nDIGlobalVariable used to have an `expr` operand. Starting from\nbceaaa964360 `expr` is now an operand of DIGlobalVariableExpression,\nwhich references both the variable and the expression.\n\nBefore:\n !0 \u003d !DIGlobalVariable(expr: !1)\n !1 \u003d !DIExpression(DW_OP_plus, 4)\n\nNow:\n !2 \u003d !DIGlobalVariableExpression(var: !3, expr: !4)\n!3 \u003d distinct !DIGlobalVariable(name: \"a\", scope: null, isLocal: false,\nisDefinition: true, type: !6)\n !4 \u003d !DIExpression(DW_OP_plus, 4)\n\nDXIL does not support DIGlobalVariableExpression, so we have to fold\n`expr` back to DIGlobalVariable.\n\n---------\n\nCo-authored-by: Andrew Savonichev \u003candrew.savonichev@gmail.com\u003e" }, { "commit": "c3e9d05ce8044612ecae2eb27dd5ca52b612a8ac", "tree": "7ce394f9edd77950002b19b347fab8593960d0b4", "parents": [ "75db4ce2177b37a2ae94fdbb9cee076dde5080df" ], "author": { "name": "alx32", "email": "103613512+alx32@users.noreply.github.com", "time": "Fri Jun 05 20:38:20 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 20:38:20 2026" }, "message": "[lld-macho][LTO] Emit `__llvm_addrsig` for `--icf\u003dsafe_thunks` (#188486)\n\nLTO was emitting `__llvm_addrsig` metadata when `--icf\u003dsafe` was\nspecified, but not for `--icf\u003dsafe_thunks`. After the recent PR\nhttps://github.com/llvm/llvm-project/pull/188400 that makes safe ICF\nconservative without `__llvm_addrsig` (marking all symbols as\naddress-significant when the section is absent), this omission caused\nsafe_thunks to silently degrade for all LTO-compiled objects: every\nsymbol became `keepUnique`, preventing body folding entirely.\n\nFix this by also enabling `EmitAddrsig` when `icfLevel` is\n`safe_thunks`. This allows the LTO backend to emit precise\naddress-significance metadata, so that only truly address-significant\nfunctions get thunk treatment while non-address-significant identical\nfunctions can still be body-folded.\n\nAdd a regression test that verifies LTO + `--icf\u003dsafe_thunks` correctly\nbody-folds non-address-significant identical functions, which would fail\nwithout this fix due to missing addrsig metadata." }, { "commit": "75db4ce2177b37a2ae94fdbb9cee076dde5080df", "tree": "183f0edc4b3e3cf2aa97c1194c8b5047c5d3adc9", "parents": [ "0dcca55e4a9adb7909eacb59e56ddcb0167d0d36" ], "author": { "name": "Hans Wennborg", "email": "hans@hanshq.net", "time": "Fri Jun 05 20:34:02 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 20:34:02 2026" }, "message": "Restore unnamed_addr on vtables in -fno-rtti builds (#201846)\n\nThis is a follow-up to #200108 which dropped unnamed_addr from vtables\nsince it can break dynamic_cast under some circumstances. However, that\ncaused significant binary size regressions due to preventing ICF of\nvtables.\n\nThis patch restores unnamed_addr on vtables in -fno-rtti builds, since\nthose do not support dynamic_cast, recovering the binary size savings at\nleast for those using that configuration." }, { "commit": "0dcca55e4a9adb7909eacb59e56ddcb0167d0d36", "tree": "6c7f721f175a2df51a93ec0c0677e0f8160f1dd3", "parents": [ "6ffd9f8c5c574f20c930d1104ef7215ff20536d5" ], "author": { "name": "Arseniy Obolenskiy", "email": "arseniy.obolenskiy@amd.com", "time": "Fri Jun 05 20:26:08 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 20:26:08 2026" }, "message": "[mlir][SPIR-V] Validate float types in parseAndVerifyType (#201910)" }, { "commit": "6ffd9f8c5c574f20c930d1104ef7215ff20536d5", "tree": "d01426bd43e15b63589e73f57d25a6e5f558be7f", "parents": [ "1227b19800524b542df685282096a2fe07a0aa5d" ], "author": { "name": "Arseniy Obolenskiy", "email": "arseniy.obolenskiy@amd.com", "time": "Fri Jun 05 20:25:44 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 20:25:44 2026" }, "message": "[mlir][SPIR-V] Add OpenCL erfc op (#201906)" }, { "commit": "1227b19800524b542df685282096a2fe07a0aa5d", "tree": "8181cbf9e6c49c5b0966ffd5710755b18ebdae2d", "parents": [ "fa9e7b77b19a9d59b13a49116924b562b09a6599" ], "author": { "name": "Aiden Grossman", "email": "aidengrossman@google.com", "time": "Fri Jun 05 20:24:58 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 20:24:58 2026" }, "message": "[SelectionDAG] Fix -Wunused-variable (#201932)\n\nNumDstElts is only used in assertions so mark it [[maybe_unused]] to\nprevent any issues with non-asserts builds and -Wunused-variable." }, { "commit": "fa9e7b77b19a9d59b13a49116924b562b09a6599", "tree": "baa891eb09c76dfe075e9ed3a24b34da849ac2cf", "parents": [ "6a8608e5a75f9c4b945c3ed210dee6b20705cefe" ], "author": { "name": "macurtis-amd", "email": "macurtis@amd.com", "time": "Fri Jun 05 20:06:15 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 20:06:15 2026" }, "message": "AMDGPU: fix assert when calling LiveRegUnits::stepBackward (#201887)\n\n`stepBackward` assert added in\nhttps://github.com/llvm/llvm-project/pull/193104.\nFound while building `aqlprofile` as part of TheRock." }, { "commit": "6a8608e5a75f9c4b945c3ed210dee6b20705cefe", "tree": "60f652573950df3695d3cfd6851c98dc961caa14", "parents": [ "ab9644ac1733b30e5947a65f0e11946bb9d56895" ], "author": { "name": "Daniel Thornburgh", "email": "dthorn@google.com", "time": "Fri Jun 05 19:56:27 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 19:56:27 2026" }, "message": "[clang] Disable crash-diagnostic-tar.c if backtraces are disabled (#201921)\n\nPR #201643 adds a test that fails if backtraces are disabled, e.g. in\ntests for a stripped build. This was failing the Fuchsia CI, so this\nchange disables this test in such cases. It is likely possible to make\nthis test finer-grained to allow it to succeed with backtraces disabled." }, { "commit": "ab9644ac1733b30e5947a65f0e11946bb9d56895", "tree": "afd08049630e4c7f106638d4c7412af84685255a", "parents": [ "1118fa12a9db10712b38db126807df983f4dda9a" ], "author": { "name": "Andy Kaylor", "email": "akaylor@nvidia.com", "time": "Fri Jun 05 19:46:19 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 19:46:19 2026" }, "message": "[CIR] Update CIR unary operations (#201727)\n\nThis creates a floating-point-specific FNegOp and updates the type\nconstraints on the existing unary operations to match the types that are\nactually produced by the Clang AST.\n\nThis change also moves complex types out of the unary operations. A\nComplexConjOp is added to handle what was previously being incorrectly\nreferred to as \"not\", while inc, dec, and minus are just expanded to\ntheir component representation during codegen. We may want to consider\nadding operations like cir.complex.add later, but for now I\u0027m just\nremoving this type from the unary ops. I\u0027m doing this for two reasons:\n(1) the old use was inconsistent with how we handle other arithmetic\noperations on complex types, and (2) I will soon need to attach\nFP-specific attributes to add and subtract of complex types (and so also\ninc and dec).\n\nAssisted-by: Cursor / claude-opus-4.8" }, { "commit": "1118fa12a9db10712b38db126807df983f4dda9a", "tree": "227d0518953bd62e219087a98f660900313f145b", "parents": [ "989ef9ccdf6eb327f7b358485509774f11eadbce" ], "author": { "name": "Leandro Lupori", "email": "leandro.lupori@linaro.org", "time": "Fri Jun 05 19:43:56 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 19:43:56 2026" }, "message": "[flang][OpenMP] Fix failure in linear-clause01.f90 test (#201920)\n\nApparently the order of some OpenMP checks changed since the pre-commit\nCI tested PR #189170.\nNow the error `The list item \u0027i\u0027 must be a dummy argument` occurs\ninstead of\n`\u0027cc\u0027 is a common block name and must not appear in a LINEAR clause`." }, { "commit": "989ef9ccdf6eb327f7b358485509774f11eadbce", "tree": "01c9195d2d0da11a92ede531d228e1e8b383c73d", "parents": [ "e3fee08c14eab9ee30c7d6d19c66cbd6bf1ff704" ], "author": { "name": "Krzysztof Drewniak", "email": "Krzysztof.Drewniak@amd.com", "time": "Fri Jun 05 19:43:35 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 19:43:35 2026" }, "message": "[DAG] isGuaranteedNotToBeUndefOrPoison - add ISD::BITCAST demanded elements handling (#200933)\n\nBitcasts preserve undef/poison status, but vector bitcasts can change\nwhich source lanes cover a demanded result lane. Map the demanded\nelement mask through fixed-length vector bitcasts before checking the\nsource where possible.\n\nAI note: an LLM generated the code and the test, I\u0027ve read them\n\nFixes #161512\n\nCo-Authored-By: OpenAI Codex \u003ccodex@openai.com\u003e\n\n---------\n\nCo-authored-by: OpenAI Codex \u003ccodex@openai.com\u003e" }, { "commit": "e3fee08c14eab9ee30c7d6d19c66cbd6bf1ff704", "tree": "f2d10697a7daa202cb082d4cc0e0f2f3848576ce", "parents": [ "ff65d814129d3220845ff64cbec90c1b83cf79e7" ], "author": { "name": "Florian Hahn", "email": "flo@fhahn.com", "time": "Fri Jun 05 19:43:23 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 19:43:23 2026" }, "message": "[VPlan] Remove special cost logic for stores predicated by header mask. (#196632)\n\nUpdate the predicated memop costs with uniform operands to \nmatch the code we generate.\n\nDepends on https://github.com/llvm/llvm-project/pull/196630\n\nPR: https://github.com/llvm/llvm-project/pull/196632" }, { "commit": "ff65d814129d3220845ff64cbec90c1b83cf79e7", "tree": "8c33797ee5e61c6c164a257d9be071a331e9c401", "parents": [ "4bf77a4abbeec52d2b229d36f9ae01d09d2794b8" ], "author": { "name": "Jonas Devlieghere", "email": "jonas@devlieghere.com", "time": "Fri Jun 05 19:37:36 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 19:37:36 2026" }, "message": "[lldb] Drop client-side Python pre-load and remove PythonPathSetup (#200533)\n\nPythonPathSetup was a Windows-only helper that pre-mapped python3xx.dll\ninto lldb.exe / lldb-dap.exe so liblldb.dll\u0027s direct imports from\npython3xx.dll would resolve when the delay-load thunk for liblldb fired\non the first SB call. ScriptInterpreterRuntimeLoader\u0027s Windows backend\nhas subsumed that responsibility, so this commit deletes the helper and\nrewrites the two clients on the new API.\n\nThe Windows pre-load itself stays in the clients: the central load in\nSystemInitializerFull::Initialize() lives behind the liblldb.dll thunk\nthat the import binding fires against, so the loader still has to run\nfrom the binary that statically links lldbHost (lldb.exe / lldb-dap.exe)\nbefore the first SB call lands." }, { "commit": "4bf77a4abbeec52d2b229d36f9ae01d09d2794b8", "tree": "1df8bcd5a3c590c2992a7b6ca849165b671138a3", "parents": [ "ae48a36859750575a9d9166b5ebd604656741ad1" ], "author": { "name": "Charles Zablit", "email": "c_zablit@apple.com", "time": "Fri Jun 05 19:35:39 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 19:35:39 2026" }, "message": "[lldb][windows] Fix Python library link for LLDB_ENABLE_PYTHON_LIMITED_API (#201407)\n\nWhen `LLDB_ENABLE_PYTHON_LIMITED_API` is `ON`, `liblldb` should link\nagainst `python3.dll` (stable ABI) rather than the `python310.dll`.\n\nPreviously, two issues prevented this on Windows:\n\n1. The `ScriptInterpreter` plugin explicitly linked\n`${Python3_LIBRARIES}` regardless of whether the limited API was\nenabled.\n\n2. Python\u0027s `pyconfig.h` emits a `#pragma` to select the import library\nat link time, choosing `python3.lib` or `python310.lib` depending on\nwhether `Py_LIMITED_API` is defined. However,\n`LLDB_ENABLE_PYTHON_LIMITED_API` was only surfaced through Config.h,\nwhich is not included early enough in the compilation chain to affect\npyconfig.h in all translation units.\n\nThis patch fixes both issues.\n\nThis is needed for https://github.com/llvm/llvm-project/pull/200533." }, { "commit": "ae48a36859750575a9d9166b5ebd604656741ad1", "tree": "b864a42f903e2620ef86ed4015adc57c752d9b58", "parents": [ "237644975a5d2df88f624b5e797522cc81b65aac" ], "author": { "name": "Daniel Thornburgh", "email": "dthorn@google.com", "time": "Fri Jun 05 19:33:45 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 19:33:45 2026" }, "message": "[orc] Add missing \u003citerator\u003e include (#201916)\n\nFixes build after libc++ PR #195509 which drops transitive includes." }, { "commit": "237644975a5d2df88f624b5e797522cc81b65aac", "tree": "e79d9666afb6577b14f4cffcd794c1afe9f5028f", "parents": [ "97b7e451423b8e971d6feb3369d67f659037ce83" ], "author": { "name": "Alexey Samsonov", "email": "vonosmas@gmail.com", "time": "Fri Jun 05 19:24:53 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 19:24:53 2026" }, "message": "[libc] Fix putting tuple_size into libc++ namespace. (#201905)\n\n633539bfa1516e616b798b2eae98bea689b3c410 added special-case for putting\n`tuple_size` into correct namespace when building with libc++. It uses\n`_LIBCPP_BEGIN_NAMESPACE_STD` macro, which recently started to contain\nmore logic, including the Clang pragmas to ensure correct visibility. It\nalso may also introduce warnings for `-Wpragma-clang-attribute` which\nneed to be suppressed. After fad51d3f41482e148683e12741ac9abfbe49742d,\nthis suppression only works for libc++ headers themselves (which define\n`#pragma GCC system_header`), but not for the user code.\n\nThis change bypasses explicit ABI annotation macros in libc++ by\nswitching to using `_LIBCPP_ABI_NAMESPACE` (that libc++ uses under the\nhood) if it\u0027s available." }, { "commit": "97b7e451423b8e971d6feb3369d67f659037ce83", "tree": "a3577682f3d63cf7a0e143550eb16b5be9d65dc4", "parents": [ "8dd73767aa9081cf67cd6aaa1358970034a7d697" ], "author": { "name": "Petr Hosek", "email": "phosek@google.com", "time": "Fri Jun 05 19:18:38 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 19:18:38 2026" }, "message": "[ORC] Add missing dependency on intrinsics_gen to Debugging library (#201908)\n\nThe Debugging library includes llvm/IR/Attributes.inc so it needs to\ndepend on intrinsics_gen to make sure the file is generated before it\u0027s\nused." }, { "commit": "8dd73767aa9081cf67cd6aaa1358970034a7d697", "tree": "bf40ba7cd573bed4021050526afa4e680be7d5d3", "parents": [ "0dedc9511bca8038ef0386225b97c867037f45bb" ], "author": { "name": "Craig Topper", "email": "craig.topper@sifive.com", "time": "Fri Jun 05 19:17:21 2026" }, "committer": { "name": "GitHub", "email": "noreply@github.com", "time": "Fri Jun 05 19:17:21 2026" }, "message": "[RISCV][P-ext] Support 4i16/v8i8 MULHS/MULHU/MULHSU for division by constant (#201747)\n\nUpdate the mulhs+add-\u003emulhsu combine to create a split mulhsu for 64-bit\ntypes. LegalizeVectorOps will ignore target specific opcodes so we can\u0027t\ncustom lower it later." } ], "next": "0dedc9511bca8038ef0386225b97c867037f45bb" }