)]}' { "log": [ { "commit": "60cf23e54e46c807513f7a36d0a7b777920b5881", "tree": "eda88d1a0bf00653c4b70b08383704fb980d1b9a", "parents": [ "eb468ca4e7b83ff77d638ac1ac849eb506b9150f" ], "author": { "name": "Sanjay Patel", "email": "spatel@rotateright.com", "time": "Fri Feb 15 16:31:55 2019 +0000" }, "committer": { "name": "Yi Kong", "email": "yikong@google.com", "time": "Tue Apr 02 01:21:59 2019 -0700" }, "message": "[InstCombine] fix crash while trying to narrow a binop of shuffles (PR40734)\n\nhttps://bugs.llvm.org/show_bug.cgi?id\u003d40734\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@354144 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "eb468ca4e7b83ff77d638ac1ac849eb506b9150f", "tree": "42decb85e9eed045b6cc24a09acca8eec803acb8", "parents": [ "06479f302e0b83afc9ab4e03cc9621f6bdc9ad78" ], "author": { "name": "Craig Topper", "email": "craig.topper@intel.com", "time": "Sun Feb 24 19:33:37 2019 +0000" }, "committer": { "name": "Yi Kong", "email": "yikong@google.com", "time": "Tue Apr 02 01:19:51 2019 -0700" }, "message": "[X86] Fix tls variable lowering issue with large code model\n\nSummary:\nThe problem here is the lowering for tls variable. Below is the DAG for the code.\nSelectionDAG has 11 nodes:\n\nt0: ch \u003d EntryToken\n t8: i64,ch \u003d load\u003c(load 8 from `i8 addrspace(257)* null`, addrspace 257)\u003e t0, Constant:i64\u003c0\u003e, undef:i64\n t10: i64 \u003d X86ISD::WrapperRIP TargetGlobalTLSAddress:i64\u003ci32* @x\u003e 0 [TF\u003d10]\n t11: i64,ch \u003d load\u003c(load 8 from got)\u003e t0, t10, undef:i64\n t12: i64 \u003d add t8, t11\n t4: i32,ch \u003d load\u003c(dereferenceable load 4 from @x)\u003e t0, t12, undef:i64\nt6: ch \u003d CopyToReg t0, Register:i32 %0, t4\nAnd when mcmodel is large, below instruction can NOT be folded.\n\n t10: i64 \u003d X86ISD::WrapperRIP TargetGlobalTLSAddress:i64\u003ci32* @x\u003e 0 [TF\u003d10]\nt11: i64,ch \u003d load\u003c(load 8 from got)\u003e t0, t10, undef:i64\nSo \"t11: i64,ch \u003d load\u003c(load 8 from got)\u003e t0, t10, undef:i64\" is lowered to \" Morphed node: t11: i64,ch \u003d MOV64rm\u003cMem:(load 8 from got)\u003e t10, TargetConstant:i8\u003c1\u003e, Register:i64 $noreg, TargetConstant:i32\u003c0\u003e, Register:i32 $noreg, t0\"\n\nWhen llvm start to lower \"t10: i64 \u003d X86ISD::WrapperRIP TargetGlobalTLSAddress:i64\u003ci32* @x\u003e 0 [TF\u003d10]\", it fails.\n\nThe patch is to fold the load and X86ISD::WrapperRIP.\n\nFixes PR26906\n\nPatch by LuoYuanke\n\nReviewers: craig.topper, rnk, annita.zhang, wxiao3\n\nReviewed By: rnk\n\nSubscribers: llvm-commits\n\nTags: #llvm\n\nDifferential Revision: https://reviews.llvm.org/D58336\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@354756 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "06479f302e0b83afc9ab4e03cc9621f6bdc9ad78", "tree": "9583601b5ff72cf3eeb05d877c60f4361a138fec", "parents": [ "45936e9d187aaac2b7c0245f6f8c10138f6a02e9" ], "author": { "name": "Yi Kong", "email": "yikong@google.com", "time": "Mon Apr 01 18:12:43 2019 +0000" }, "committer": { "name": "Yi Kong", "email": "yikong@google.com", "time": "Mon Apr 01 14:27:04 2019 -0700" }, "message": "[llvm-objcopy] Add --keep-symbols option\n\nDifferential Revision: https://reviews.llvm.org/D60054\n\nChange-Id: I8ed5491cd553d3dc1909b2c57eebbba3d3ab1c74\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@357418 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "45936e9d187aaac2b7c0245f6f8c10138f6a02e9", "tree": "d0fd858344034365751159e96bcc75dfe9bf6c31", "parents": [ "e213a798d9f836e371196276dfdb69a21ad3343b" ], "author": { "name": "Yi Kong", "email": "yikong@google.com", "time": "Tue Mar 26 12:18:08 2019 +0000" }, "committer": { "name": "Yi Kong", "email": "yikong@google.com", "time": "Tue Mar 26 05:19:06 2019 -0700" }, "message": "Fix nondeterminism introduced in r353954\n\nDenseMap iteration order is not guaranteed, use MapVector instead.\n\nFix provided by srhines.\n\nDifferential Revision: https://reviews.llvm.org/D59807\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@356988 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "e213a798d9f836e371196276dfdb69a21ad3343b", "tree": "b69abe5af0005c8d25d81384820ce85c1c735fbb", "parents": [ "c322391378a0ba8e41a2dbcad0859c78997b3fdc" ], "author": { "name": "Mitch Phillips", "email": "mitchphillips@outlook.com", "time": "Fri Mar 08 21:22:35 2019 +0000" }, "committer": { "name": "Yi Kong", "email": "yikong@google.com", "time": "Fri Mar 08 17:39:07 2019 -0800" }, "message": "[HWASan] Save + print registers when tag mismatch occurs in AArch64.\n\nSummary:\nThis change change the instrumentation to allow users to view the registers at the point at which tag mismatch occured. Most of the heavy lifting is done in the runtime library, where we save the registers to the stack and emit unwind information. This allows us to reduce the overhead, as very little additional work needs to be done in each __hwasan_check instance.\n\nIn this implementation, the fast path of __hwasan_check is unmodified. There are an additional 4 instructions (16B) emitted in the slow path in every __hwasan_check instance. This may increase binary size somewhat, but as most of the work is done in the runtime library, it\u0027s manageable.\n\nThe failure trace now contains a list of registers at the point of which the failure occured, in a format similar to that of Android\u0027s tombstones. It currently has the following format:\n\nRegisters where the failure occurred (pc 0x0055555561b4):\n x0 0000000000000014 x1 0000007ffffff6c0 x2 1100007ffffff6d0 x3 12000056ffffe025\n x4 0000007fff800000 x5 0000000000000014 x6 0000007fff800000 x7 0000000000000001\n x8 12000056ffffe020 x9 0200007700000000 x10 0200007700000000 x11 0000000000000000\n x12 0000007fffffdde0 x13 0000000000000000 x14 02b65b01f7a97490 x15 0000000000000000\n x16 0000007fb77376b8 x17 0000000000000012 x18 0000007fb7ed6000 x19 0000005555556078\n x20 0000007ffffff768 x21 0000007ffffff778 x22 0000000000000001 x23 0000000000000000\n x24 0000000000000000 x25 0000000000000000 x26 0000000000000000 x27 0000000000000000\n x28 0000000000000000 x29 0000007ffffff6f0 x30 00000055555561b4\n\n... and prints after the dump of memory tags around the buggy address.\n\nEvery register is saved exactly as it was at the point where the tag mismatch occurs, with the exception of x16/x17. These registers are used in the tag mismatch calculation as scratch registers during __hwasan_check, and cannot be saved without affecting the fast path. As these registers are designated as scratch registers for linking, there should be no important information in them that could aid in debugging.\n\nReviewers: pcc, eugenis\n\nReviewed By: pcc, eugenis\n\nSubscribers: srhines, kubamracek, mgorny, javed.absar, krytarowski, kristof.beyls, hiraditya, jdoerfert, llvm-commits, #sanitizers\n\nTags: #sanitizers, #llvm\n\nDifferential Revision: https://reviews.llvm.org/D58857\n\nChange-Id: I60f20f61194da1680cfa9968d802101622ed6e51\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@355738 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "c322391378a0ba8e41a2dbcad0859c78997b3fdc", "tree": "052a5606b6dc6d6ece3061c454277bae8406b4b9", "parents": [ "a4c8940d5c26cee0b333460c12326f716ecd9106" ], "author": { "name": "Teresa Johnson", "email": "tejohnson@google.com", "time": "Thu Feb 14 21:22:50 2019 +0000" }, "committer": { "name": "Yi Kong", "email": "yikong@google.com", "time": "Fri Mar 01 10:32:13 2019 -0800" }, "message": "[ThinLTO] Detect partially split modules during the thin link\n\nSummary:\nThe changes to disable LTO unit splitting by default (r350949) and\ndetect inconsistently split LTO units (r350948) are causing some crashes\nwhen the inconsistency is detected in multiple threads simultaneously.\nFix that by having the code always look for the inconsistently split\nLTO units during the thin link, by checking for the presence of type\ntests recorded in the summaries.\n\nModify test added in r350948 to remove single threading required to fix\na bot failure due to this issue (and some debugging options added in the\nprocess of diagnosing it).\n\nReviewers: pcc\n\nSubscribers: mehdi_amini, inglorion, eraman, steven_wu, dexonsmith, llvm-commits\n\nTags: #llvm\n\nDifferential Revision: https://reviews.llvm.org/D57561\n\nChange-Id: I1df30bf57e0fb8b23a2444e07f1e0a04eb8ef2ea\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@354062 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "a4c8940d5c26cee0b333460c12326f716ecd9106", "tree": "ba6dcf620fa54b254a8ecebe9af9a8756d6249ba", "parents": [ "8092d59b2d737dda2df591450089c14c43030191" ], "author": { "name": "Stephen Hines", "email": "srhines@google.com", "time": "Tue Sep 04 14:43:47 2018 -0700" }, "committer": { "name": "Yi Kong", "email": "yikong@google.com", "time": "Fri Mar 01 10:32:08 2019 -0800" }, "message": "Revert \"IRMover: Account for matching types present across modules\"\n\nThis reverts commit 444617b4b373eef10ffddcecbeb018b5cb84b66f.\n\nBug: 113573336\nChange-Id: Iac87da49b42dbb55cdb7c807f67fad523fc790c4\n" }, { "commit": "8092d59b2d737dda2df591450089c14c43030191", "tree": "d4bc42f1c18d707ade6454683a143ea4cb9b56d5", "parents": [ "06a18b67a0c135bfaf4e9dd604e83a63090bbe50", "5077597e0d5b86d9f9c27286d8b28f8b3645a74c" ], "author": { "name": "Yi Kong", "email": "yikong@google.com", "time": "Wed Feb 27 14:16:20 2019 -0800" }, "committer": { "name": "Yi Kong", "email": "yikong@google.com", "time": "Wed Feb 27 14:16:20 2019 -0800" }, "message": "Merge 5077597e0d5 for LLVM update to 353983\n\nChange-Id: Ifad5eec4c06bd25ca33fcd8874563a0651c1dbde\n" }, { "commit": "06a18b67a0c135bfaf4e9dd604e83a63090bbe50", "tree": "b664493264259e1ca955b139d80574fbeaa094b7", "parents": [ "4ccd15c2037ec8506aff757452131a084bc3583e" ], "author": { "name": "Yi Kong", "email": "yikong@google.com", "time": "Wed Feb 27 14:16:19 2019 -0800" }, "committer": { "name": "Yi Kong", "email": "yikong@google.com", "time": "Wed Feb 27 14:16:19 2019 -0800" }, "message": " revert to previous base llvm-svn.349610\n\nChange-Id: Ibbe1931d087763e3fd973a6266aec9b6efa083c2\n" }, { "commit": "4ccd15c2037ec8506aff757452131a084bc3583e", "tree": "5ec9bd9c60806e804e10be918b7355a52a34edf1", "parents": [ "ab552995647d484a712bf15f77e6c6f4e8ac532b" ], "author": { "name": "Eli Friedman", "email": "efriedma@codeaurora.org", "time": "Wed Dec 19 22:52:04 2018 +0000" }, "committer": { "name": "Stephen Hines", "email": "srhines@google.com", "time": "Wed Feb 20 11:36:28 2019 -0800" }, "message": "[CodeGenPrepare] Fix bad IR created by large offset GEP splitting.\n\nCreating the IR builder, then modifying the CFG, leads to an IRBuilder\nwhere the BB and insertion point are inconsistent, so new instructions\nhave the wrong parent.\n\nModified an existing test because the test wasn\u0027t covering anything\nuseful (the \"invoke\" was not actually an invoke by the time we hit the\ncode in question).\n\nDifferential Revision: https://reviews.llvm.org/D55729\n\n\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@349693 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "5077597e0d5b86d9f9c27286d8b28f8b3645a74c", "tree": "70aefdc5e3d27d97c573b6622144292ae3395a8d", "parents": [ "2d7289e2fa5a69fab523607bfa3b3e2ef2abc072" ], "author": { "name": "Thomas Lively", "email": "tlively@google.com", "time": "Wed Feb 13 22:11:16 2019 +0000" }, "committer": { "name": "Thomas Lively", "email": "tlively@google.com", "time": "Wed Feb 13 22:11:16 2019 +0000" }, "message": "[WebAssembly] Bulk memory intrinsics and builtins\n\nSummary:\nimplements llvm intrinsics and clang intrinsics for\nmemory.init and data.drop.\n\nReviewers: aheejin\n\nSubscribers: dschuff, sbc100, jgravelle-google, hiraditya, sunfish, cfe-commits\n\nTags: #clang\n\nDifferential Revision: https://reviews.llvm.org/D57736\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353983 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "2d7289e2fa5a69fab523607bfa3b3e2ef2abc072", "tree": "9db9d748d0d8c58faf51149a95d7009ac8ef8857", "parents": [ "ff64ebce03520a872f457a1d8b8263923fe1b2b5" ], "author": { "name": "Serge Guelton", "email": "sguelton@redhat.com", "time": "Wed Feb 13 22:11:09 2019 +0000" }, "committer": { "name": "Serge Guelton", "email": "sguelton@redhat.com", "time": "Wed Feb 13 22:11:09 2019 +0000" }, "message": "Revert r353962\n\nSpecialization of Optional for trivially copyable types yields failure on the buildbots I fail to reproduce locally.\nBetter safe than sorry, reverting.\n\n\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353982 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "ff64ebce03520a872f457a1d8b8263923fe1b2b5", "tree": "d73a4504a36c2e716eba8cf97b813dfe6c0bc900", "parents": [ "5a03a071de7cfde255fe8bb40aaffcef02c600c2" ], "author": { "name": "Peter Collingbourne", "email": "peter@pcc.me.uk", "time": "Wed Feb 13 21:03:23 2019 +0000" }, "committer": { "name": "Peter Collingbourne", "email": "peter@pcc.me.uk", "time": "Wed Feb 13 21:03:23 2019 +0000" }, "message": "gn build: Merge r353957.\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353980 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "5a03a071de7cfde255fe8bb40aaffcef02c600c2", "tree": "3d16c5c3eb298336cbf90e8026540197c93c701a", "parents": [ "21b311090c4f2eb5e6517c4a840d3676c298f076" ], "author": { "name": "Philip Reames", "email": "listmail@philipreames.com", "time": "Wed Feb 13 20:42:59 2019 +0000" }, "committer": { "name": "Philip Reames", "email": "listmail@philipreames.com", "time": "Wed Feb 13 20:42:59 2019 +0000" }, "message": "[SelectionDAG] Kill last uses of getAtomic w/o a MMO operand [NFC]\n\nThe helper function was used by only two callers, and largely ended up providing distinct functionality based on optional arguments and opcode. Inline and simply to make the functionality much more clear.\n\n\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353977 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "21b311090c4f2eb5e6517c4a840d3676c298f076", "tree": "8fe12f7abaf4887fcfedb6dbfe8b860c472b1a07", "parents": [ "18afa443d99c507dd60294f975510b2138a15813" ], "author": { "name": "Craig Topper", "email": "craig.topper@intel.com", "time": "Wed Feb 13 20:12:41 2019 +0000" }, "committer": { "name": "Craig Topper", "email": "craig.topper@intel.com", "time": "Wed Feb 13 20:12:41 2019 +0000" }, "message": "[X86] Add \u0027mpx\u0027 to getHostCPUFeatures.\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353974 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "18afa443d99c507dd60294f975510b2138a15813", "tree": "d0a2978db358f7e0b29dc8a00e75aa38fbc411a7", "parents": [ "4477a571f31d4a03e630c49b94ab94b807ec3288" ], "author": { "name": "Vedant Kumar", "email": "vsk@apple.com", "time": "Wed Feb 13 19:53:38 2019 +0000" }, "committer": { "name": "Vedant Kumar", "email": "vsk@apple.com", "time": "Wed Feb 13 19:53:38 2019 +0000" }, "message": "[CodeExtractor] Only lift lifetime markers present in the extraction region\n\nWhen CodeExtractor finds liftime markers referencing inputs to the\nextraction region, it lifts these markers out of the region and inserts\nthem around the call to the extracted function (see r350420, PR39671).\n\nHowever, it should *only* lift lifetime markers that are actually\npresent in the extraction region. I.e., if a start marker is present in\nthe extraction region but a corresponding end marker isn\u0027t (or vice\nversa), only the start marker (or end marker, resp.) should be lifted.\n\nDifferential Revision: https://reviews.llvm.org/D57834\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353973 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "4477a571f31d4a03e630c49b94ab94b807ec3288", "tree": "c8647bd7f2b788e3aca53b982995d8f727378e44", "parents": [ "70b3852f944ca47f4597d35f2efeef87d34ddd44" ], "author": { "name": "Philip Reames", "email": "listmail@philipreames.com", "time": "Wed Feb 13 19:49:17 2019 +0000" }, "committer": { "name": "Philip Reames", "email": "listmail@philipreames.com", "time": "Wed Feb 13 19:49:17 2019 +0000" }, "message": "[Tests] More unordered atomic lowering tests\n\nThis time, focused around narrowing and widening transformations. Also, include a few simple memory optimization tests to highlight missed oppurtunities. This is part of building up the test base for D57601.\n\n\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353972 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "70b3852f944ca47f4597d35f2efeef87d34ddd44", "tree": "8e63ef5a64c3c3671944a9ee8f070517b08b9b66", "parents": [ "011c044ce0a9b6d3ffb05f6c1500cc2fa39ad7da" ], "author": { "name": "Philip Reames", "email": "listmail@philipreames.com", "time": "Wed Feb 13 18:41:54 2019 +0000" }, "committer": { "name": "Philip Reames", "email": "listmail@philipreames.com", "time": "Wed Feb 13 18:41:54 2019 +0000" }, "message": "[Tests] RMW folding tests w/unordered atomic operations\n\nWe get a suprising number of these today actually, but some are missed. The main point of this is strengthen the test set for D57601.\n\n\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353966 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "011c044ce0a9b6d3ffb05f6c1500cc2fa39ad7da", "tree": "08797fb6331e02ae6cae7d9c9e4764c984cafe59", "parents": [ "acb7bf3cf985b38a505eb3a5be0569ecab7770ba" ], "author": { "name": "Philip Reames", "email": "listmail@philipreames.com", "time": "Wed Feb 13 18:26:01 2019 +0000" }, "committer": { "name": "Philip Reames", "email": "listmail@philipreames.com", "time": "Wed Feb 13 18:26:01 2019 +0000" }, "message": "[Tests] Add a bunch of tests for load folding w/unordered atomics\n\n\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353964 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "acb7bf3cf985b38a505eb3a5be0569ecab7770ba", "tree": "6532069bc4a84bc1f16c2f8c8fecce0f97d5f838", "parents": [ "3637405f12571ef952feb8a6084bbd95804ca9cb" ], "author": { "name": "Craig Topper", "email": "craig.topper@intel.com", "time": "Wed Feb 13 18:21:36 2019 +0000" }, "committer": { "name": "Craig Topper", "email": "craig.topper@intel.com", "time": "Wed Feb 13 18:21:36 2019 +0000" }, "message": "[X86] Add \u0027fxsr\u0027 to the getHostCPUFeatures detection code.\n\nWe implicitly mark this feature as enabled when the target is 64-bits, but our detection code for -march\u003dnative didn\u0027t support it so you can\u0027t detect it on 32-bit targets.\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353963 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "3637405f12571ef952feb8a6084bbd95804ca9cb", "tree": "2135b9ae62e153df375e9f212b021c0d04755878", "parents": [ "9483ee997b90a9be80f13bb38ee7ee29ef566546" ], "author": { "name": "Serge Guelton", "email": "sguelton@redhat.com", "time": "Wed Feb 13 18:12:04 2019 +0000" }, "committer": { "name": "Serge Guelton", "email": "sguelton@redhat.com", "time": "Wed Feb 13 18:12:04 2019 +0000" }, "message": "Re-commit rL353927, patch included\n\nMake llvm::Optional\u003cT\u003e trivially copyable when T is trivially copyable\n\nThis is an ever-recurring issue (see https://bugs.llvm.org/show_bug.cgi?id\u003d39427 and https://bugs.llvm.org/show_bug.cgi?id\u003d35978)\nbut I believe that thanks to https://reviews.llvm.org/D54472 we can now ship a decent implementation of this.\n\nBasically the fact that llvm::is_trivially_copyable has a consistent behavior across compilers should prevent any ABI issue,\nand using in-place new instead of memcpy should keep compiler bugs away.\n\nThis patch is slightly different from the original revision https://reviews.llvm.org/rL353927 but achieves the same goal. It just avoids\ngoing through std::conditional which may the code more explicit.\n\n\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353962 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "9483ee997b90a9be80f13bb38ee7ee29ef566546", "tree": "d837ef86a343b686840c83a53741cedb13add5f6", "parents": [ "3a73f3f781bc7eaef43690abbc52a4a6e5affb20" ], "author": { "name": "Philip Reames", "email": "listmail@philipreames.com", "time": "Wed Feb 13 18:00:58 2019 +0000" }, "committer": { "name": "Philip Reames", "email": "listmail@philipreames.com", "time": "Wed Feb 13 18:00:58 2019 +0000" }, "message": "[Tests] First batch of cornercase tests for unordered atomics\n\nMixture of things we legally can\u0027t do, and things we\u0027re missing. Once D57601 is in, the later will serve as a punch list.\n\n\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353959 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "3a73f3f781bc7eaef43690abbc52a4a6e5affb20", "tree": "6b0c8e2e8b5514e27063d41cf3b225746329733c", "parents": [ "a876ea7b72775027b690027eecb14b74932c274b" ], "author": { "name": "Philip Reames", "email": "listmail@philipreames.com", "time": "Wed Feb 13 17:30:03 2019 +0000" }, "committer": { "name": "Philip Reames", "email": "listmail@philipreames.com", "time": "Wed Feb 13 17:30:03 2019 +0000" }, "message": "[Tests] Auto update a test\n\n\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353958 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "a876ea7b72775027b690027eecb14b74932c274b", "tree": "13793d4b698287b506dfe6cf931fae3805d1a533", "parents": [ "5414d2957fa5bbefcdbbb3ef521c982d761c4204" ], "author": { "name": "Petr Hosek", "email": "phosek@chromium.org", "time": "Wed Feb 13 17:28:47 2019 +0000" }, "committer": { "name": "Petr Hosek", "email": "phosek@chromium.org", "time": "Wed Feb 13 17:28:47 2019 +0000" }, "message": "[AArch64] Support reserving arbitrary general purpose registers\n\nThis is a follow up to D48580 and D48581 which allows reserving\narbitrary general purpose registers with the exception of registers\nwith special purpose (X8, X16-X18, X29, X30) and registers used by LLVM\n(X0, X19). This change also generalizes some of the existing logic to\nrely entirely on values generated from tablegen.\n\nDifferential Revision: https://reviews.llvm.org/D56305\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353957 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "5414d2957fa5bbefcdbbb3ef521c982d761c4204", "tree": "f07fb954984c02ed8ad72a4e961d0fa15c47b17e", "parents": [ "826e231bb6f77b2f066d73393e9ce66733266d15" ], "author": { "name": "Philip Reames", "email": "listmail@philipreames.com", "time": "Wed Feb 13 17:23:11 2019 +0000" }, "committer": { "name": "Philip Reames", "email": "listmail@philipreames.com", "time": "Wed Feb 13 17:23:11 2019 +0000" }, "message": "[Tests] Rename some test files for consistency\n\nMost are named \"atomic-something\" so rename the few which were \"atomic_something\". I keep typing the wrong name due to the inconsistency. :)\n\n\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353956 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "826e231bb6f77b2f066d73393e9ce66733266d15", "tree": "f5a2445c3787a8c63ad7a772827a18772b6d10e9", "parents": [ "64ea45e488996e0371d10ac2b8abfe8079201b96" ], "author": { "name": "Jeremy Morse", "email": "jeremy.morse.llvm@gmail.com", "time": "Wed Feb 13 16:33:05 2019 +0000" }, "committer": { "name": "Jeremy Morse", "email": "jeremy.morse.llvm@gmail.com", "time": "Wed Feb 13 16:33:05 2019 +0000" }, "message": "[DebugInfo][DAG] Either salvage dangling debug info or emit Undef DBG_VALUEs\n\nIn this patch SelectionDAG tries to salvage any dbg.values that are going to be\ndropped, in case they can be recovered from Values in the current BB. It also\nstrengthens SelectionDAGs handling of dangling debug data, so that dbg.values\nare *always* emitted (as Undef or otherwise) instead of dangling forever.\n\nThe motivation behind this patch exists in the new test case: a memory address\n(here a bitcast and GEP) exist in one basic block, and a dbg.value referring to\nthe address is left in the \u0027next\u0027 block. The base pointer is live across all\nbasic blocks. In current llvm trunk the dbg.value cannot be encoded, and it\nisn\u0027t even emitted as an Undef DBG_VALUE.\n\nThe change is simply: if we\u0027re definitely going to drop a dbg.value, repeatedly\napply salvageDebugInfo to its operand until either we find something that can\nbe encoded, or we can\u0027t salvage any further in which case we produce an Undef\nDBG_VALUE. To know when we\u0027re \"definitely going to drop a dbg.value\",\nSelectionDAG signals SelectionDAGBuilder when all IR instructions have been\nencoded to force salvaging. This ensures that any dbg.value that\u0027s dangling\nafter DAG creation will have a corresponding DBG_VALUE encoded. \n\nDifferential Revision: https://reviews.llvm.org/D57694\n\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353954 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "64ea45e488996e0371d10ac2b8abfe8079201b96", "tree": "42dfa0167e811e1f5f014fcb4dc490d1a9abd060", "parents": [ "95d339f46f1ea45b68589b257e20a986705532b8" ], "author": { "name": "Simon Pilgrim", "email": "llvm-dev@redking.me.uk", "time": "Wed Feb 13 16:12:36 2019 +0000" }, "committer": { "name": "Simon Pilgrim", "email": "llvm-dev@redking.me.uk", "time": "Wed Feb 13 16:12:36 2019 +0000" }, "message": "[X86][AVX] Add shuffle_v8i32_0dcd3f14 shuffle test case\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353953 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "95d339f46f1ea45b68589b257e20a986705532b8", "tree": "f63ed1dc5d5bc2330b6fcb5d614469f4a17a353b", "parents": [ "7d7f9355cd87e593ac068cb73378cc3bec6c6271" ], "author": { "name": "Fangrui Song", "email": "maskray@google.com", "time": "Wed Feb 13 15:58:23 2019 +0000" }, "committer": { "name": "Fangrui Song", "email": "maskray@google.com", "time": "Wed Feb 13 15:58:23 2019 +0000" }, "message": "[llvm-readobj] Rename pr_data to PrData\n\nAs requested by grimar in D58112.\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353951 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "7d7f9355cd87e593ac068cb73378cc3bec6c6271", "tree": "3a77cff0fd52b1da96f64772fbef5f1a2589b9b3", "parents": [ "e9cd1e556a89903568e490358ae7e28874014e32" ], "author": { "name": "Jeremy Morse", "email": "jeremy.morse.llvm@gmail.com", "time": "Wed Feb 13 15:53:10 2019 +0000" }, "committer": { "name": "Jeremy Morse", "email": "jeremy.morse.llvm@gmail.com", "time": "Wed Feb 13 15:53:10 2019 +0000" }, "message": "[DebugInfo][DAG] Refactor dbg.value lowering into its own method\n\nThis is a pure copy-and-paste job, moving the logic for lowering dbg.value\nintrinsics to SDDbgValues into its own function. This is ahead of adding some\nmore users of this logic.\n\nDifferential Revision: https://reviews.llvm.org/D57697\n\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353950 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "e9cd1e556a89903568e490358ae7e28874014e32", "tree": "3219b834f8e04f802cf0619a99f42e40e3b24cde", "parents": [ "e52eea9dbbeacb024dc43e5ea6620d09a75011bb" ], "author": { "name": "Andrea Di Biagio", "email": "Andrea_DiBiagio@sn.scee.net", "time": "Wed Feb 13 14:56:06 2019 +0000" }, "committer": { "name": "Andrea Di Biagio", "email": "Andrea_DiBiagio@sn.scee.net", "time": "Wed Feb 13 14:56:06 2019 +0000" }, "message": "[MCA] Store a bitmask of used groups in the instruction descriptor.\n\nThis is to speedup \u0027checkAvailability\u0027 queries in class ResourceManager.\nNo functional change intended.\n\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353949 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "e52eea9dbbeacb024dc43e5ea6620d09a75011bb", "tree": "2a99c1a64a71e2f119cbce2d5ccc7ae4f16f8d18", "parents": [ "259523f75aad17a4d5bf93eff32c1e8dcea62e12" ], "author": { "name": "Jeremy Morse", "email": "jeremy.morse.llvm@gmail.com", "time": "Wed Feb 13 13:37:33 2019 +0000" }, "committer": { "name": "Jeremy Morse", "email": "jeremy.morse.llvm@gmail.com", "time": "Wed Feb 13 13:37:33 2019 +0000" }, "message": "[DebugInfo][DAG] Limit special-casing of dbg.values for Arguments\n\nSelectionDAGBuilder has special handling for dbg.value intrinsics that are\nunderstood to define the location of function parameters on entry to the\nfunction. To enable this, we avoid recording a dbg.value as a virtual register\nreference if it might be such a parameter, so that it later hits\nEmitFuncArgumentDbgValue.\n\nThis patch reduces the set of circumstances where we avoid recording a\ndbg.value as a virtual register reference, to allow more \"normal\" variables\nto be recorded that way. We now only bypass for potential parameters if:\n * The dbg.value operand is an Argument,\n * The Variable is a parameter, and\n * The Variable is not inlined.\nmeaning it\u0027s very likely that the dbg.value is a function-entry parameter\nlocation.\n\nDifferential Revision: https://reviews.llvm.org/D57584\n\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353948 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "259523f75aad17a4d5bf93eff32c1e8dcea62e12", "tree": "1c7bd092f120a82fb2ed16c2859ab4c33f2f0970", "parents": [ "e194414673592749c06af78585d27aafc27b3a87" ], "author": { "name": "Max Kazantsev", "email": "max.kazantsev@azul.com", "time": "Wed Feb 13 11:54:45 2019 +0000" }, "committer": { "name": "Max Kazantsev", "email": "max.kazantsev@azul.com", "time": "Wed Feb 13 11:54:45 2019 +0000" }, "message": "[NFC] Add const qualifiers where possible\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353941 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "e194414673592749c06af78585d27aafc27b3a87", "tree": "a54768a6bdf0439f91f4e631276e5080b3b786ae", "parents": [ "8810171ee8995b00d4fbb8a8cd754fbc76d1f528" ], "author": { "name": "Serge Guelton", "email": "sguelton@redhat.com", "time": "Wed Feb 13 11:35:45 2019 +0000" }, "committer": { "name": "Serge Guelton", "email": "sguelton@redhat.com", "time": "Wed Feb 13 11:35:45 2019 +0000" }, "message": "Revert r353927\n\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353940 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "8810171ee8995b00d4fbb8a8cd754fbc76d1f528", "tree": "69a5e1802ee278c8fccb6b735a55b0ceb198232d", "parents": [ "fcf88f08a21221c9fcb0fa27a7895ed3a9b30fd9" ], "author": { "name": "Diana Picus", "email": "diana.picus@linaro.org", "time": "Wed Feb 13 11:25:32 2019 +0000" }, "committer": { "name": "Diana Picus", "email": "diana.picus@linaro.org", "time": "Wed Feb 13 11:25:32 2019 +0000" }, "message": "[ARM GlobalISel] Support G_SELECT for Thumb2\n\nSame as arm mode, but slightly different opcodes.\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353938 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "fcf88f08a21221c9fcb0fa27a7895ed3a9b30fd9", "tree": "f3100be777f59bb935d0340c5c7ced4d5e415ed7", "parents": [ "b902ab171c3c46f61e4485717b046609956f386b" ], "author": { "name": "Andrea Di Biagio", "email": "Andrea_DiBiagio@sn.scee.net", "time": "Wed Feb 13 11:02:42 2019 +0000" }, "committer": { "name": "Andrea Di Biagio", "email": "Andrea_DiBiagio@sn.scee.net", "time": "Wed Feb 13 11:02:42 2019 +0000" }, "message": "[MCA][Scheduler] Use latency information to further classify busy instructions.\n\nThis patch introduces a new instruction stage named \u0027IS_PENDING\u0027.\nAn instruction transitions from the IS_DISPATCHED to the IS_PENDING stage if\ninput registers are not available, but their latency is known.\n\nThis patch also adds a new set of instructions named \u0027PendingSet\u0027 to class\nScheduler. The idea is that the PendingSet will only contain instructions that\nhave reached the IS_PENDING stage.\nBy construction, an instruction in the PendingSet is only dependent on\ninstructions that have already reached the execution stage. The plan is to use\nthis knowledge to identify bottlenecks caused by data dependencies (see\nPR37494).\n\nDifferential Revision: https://reviews.llvm.org/D58066\n\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353937 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "b902ab171c3c46f61e4485717b046609956f386b", "tree": "6800b42127ede8eead1527f03f967da87fa6971b", "parents": [ "f37f383fd3d7ee4243c66ed33344e6ee01e41d5d" ], "author": { "name": "Jeremy Morse", "email": "jeremy.morse.llvm@gmail.com", "time": "Wed Feb 13 10:54:53 2019 +0000" }, "committer": { "name": "Jeremy Morse", "email": "jeremy.morse.llvm@gmail.com", "time": "Wed Feb 13 10:54:53 2019 +0000" }, "message": "[DebugInfo][InstCombine] Prefer to salvage debuginfo over sinking it\n\nWhen instcombine sinks an instruction between two basic blocks, it sinks any\ndbg.value users in the source block with it, to prevent debug use-before-free.\nHowever we can do better by attempting to salvage the debug users, which would\navoid moving where the variable location changes. If we successfully salvage,\nstill sink a (cloned) dbg.value with the sunk instruction, as the sunk\ninstruction is more likely to be \"live\" later in the compilation process.\n\nIf we can\u0027t salvage dbg.value users of a sunk instruction, mark the dbg.values\nin the original block as being undef. This terminates any earlier variable\nlocation range, and represents the fact that we\u0027ve optimized out the variable\nlocation for a portion of the program.\n\nDifferential Revision: https://reviews.llvm.org/D56788\n\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353936 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "f37f383fd3d7ee4243c66ed33344e6ee01e41d5d", "tree": "889eaaae859542edc6949849c1406d9b47aa43b5", "parents": [ "9f583675584c9d63abc2f5037abdb75427a5219b" ], "author": { "name": "Serge Guelton", "email": "sguelton@redhat.com", "time": "Wed Feb 13 10:19:06 2019 +0000" }, "committer": { "name": "Serge Guelton", "email": "sguelton@redhat.com", "time": "Wed Feb 13 10:19:06 2019 +0000" }, "message": "Missing header\n\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353933 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "9f583675584c9d63abc2f5037abdb75427a5219b", "tree": "0ce56447b0b7c6c9cc313cc12b8f0c10b8769781", "parents": [ "ac705e8410a709d40dc2a2a19cac19ae2a9f9925" ], "author": { "name": "Max Kazantsev", "email": "max.kazantsev@azul.com", "time": "Wed Feb 13 09:56:30 2019 +0000" }, "committer": { "name": "Max Kazantsev", "email": "max.kazantsev@azul.com", "time": "Wed Feb 13 09:56:30 2019 +0000" }, "message": "[GuardWidening] Support widening of explicitly expressed guards\n\nThis patch adds support of guards expressed in explicit form via\n`widenable_condition` in Guard Widening pass.\n\nDifferential Revision: https://reviews.llvm.org/D56075\nReviewed By: reames\n\n\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353932 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "ac705e8410a709d40dc2a2a19cac19ae2a9f9925", "tree": "51ee45ac7ac564e76ecfade8f0224f5428ea40cb", "parents": [ "42f63a9ef4a41690ce523189ce5436f98d947715" ], "author": { "name": "David Stenberg", "email": "david.stenberg@ericsson.com", "time": "Wed Feb 13 09:34:07 2019 +0000" }, "committer": { "name": "David Stenberg", "email": "david.stenberg@ericsson.com", "time": "Wed Feb 13 09:34:07 2019 +0000" }, "message": "[DebugInfo] Stop changing labels for register-described parameter DBG_VALUEs\n\nSummary:\nThis is a follow-up to D57510. This patch stops DebugHandlerBase from\nchanging the starting label for the first non-overlapping,\nregister-described parameter DBG_VALUEs to the beginning of the\nfunction. That code did not consider what defined the registers, which\ncould result in the ranges for the debug values starting before their\ndefining instructions. We currently do not emit debug values for\nconstant values directly at the start of the function, so this code is\nstill useful for such values, but my intention is to remove the code\nfrom DebugHandlerBase completely when we get there. One reason for\nremoving it is that the code violates the history map\u0027s ranges, which I\nthink can make it quite confusing when troubleshooting.\n\nIn D57510, PrologEpilogInserter was amended so that parameter DBG_VALUEs\nnow are kept at the start of the entry block, even after emission of\nprologue code. That was done to reduce the degradation of debug\ncompleteness from this patch. PR40638 is another example, where the\nlexical-scope trimming that LDV does, in combination with scheduling,\nresults in instructions after the prologue being left without locations.\nThere might be other cases where the DBG_VALUEs are pushed further down,\nfor which the DebugHandlerBase code may be helpful, but as it now quite\noften result in incorrect locations, even after the prologue, it seems\nbetter to remove that code, and try to work our way up with accurate\nlocations.\n\nIn the long run we should maybe not aim to provide accurate locations\ninside the prologue. Some single location descriptions, at least those\nreferring to stack values, generate inaccurate values inside the\nepilogue, so we maybe should not aim to achieve accuracy for location\nlists. However, it seems that we now emit line number programs that can\nresult in GDB and LLDB stopping inside the prologue when doing line\nnumber stepping into functions. See PR40188 for more information.\n\nA summary of some of the changed test cases is available in PR40188#c2.\n\nReviewers: aprantl, dblaikie, rnk, jmorse\n\nReviewed By: aprantl\n\nSubscribers: jdoerfert, jholewinski, jvesely, javed.absar, llvm-commits\n\nTags: #debug-info, #llvm\n\nDifferential Revision: https://reviews.llvm.org/D57511\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353928 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "42f63a9ef4a41690ce523189ce5436f98d947715", "tree": "d962ba5feb37ab5cb765bd773355e8d3045fe48f", "parents": [ "95decc39b55517684c798b4ba4ce1e4142a25306" ], "author": { "name": "Serge Guelton", "email": "sguelton@redhat.com", "time": "Wed Feb 13 09:31:22 2019 +0000" }, "committer": { "name": "Serge Guelton", "email": "sguelton@redhat.com", "time": "Wed Feb 13 09:31:22 2019 +0000" }, "message": "Make llvm::Optional\u003cT\u003e trivially copyable when T is trivially copyable\n\nThis is an ever-recurring issue (see https://bugs.llvm.org/show_bug.cgi?id\u003d39427 and https://bugs.llvm.org/show_bug.cgi?id\u003d35978)\nbut I believe that thanks to https://reviews.llvm.org/D54472 we can now ship a decent implementation of this.\n\nBasically the fact that llvm::is_trivially_copyable has a consistent behavior across compilers should prevent any ABI issue,\nand using in-place new instead of memcpy should keep compiler bugs away.\n\nDifferential Revision: https://reviews.llvm.org/D57097\n\n\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353927 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "95decc39b55517684c798b4ba4ce1e4142a25306", "tree": "e9ba100fe7924f06152d8f13ccc9f87b678442c0", "parents": [ "74c86f74996dd5d17c8f6e8a2ee3d73e0ec9ba41" ], "author": { "name": "Michal Gorny", "email": "mgorny@gentoo.org", "time": "Wed Feb 13 08:34:40 2019 +0000" }, "committer": { "name": "Michal Gorny", "email": "mgorny@gentoo.org", "time": "Wed Feb 13 08:34:40 2019 +0000" }, "message": "[llvm] [cmake] Provide split include paths in LLVMConfig\n\nModify LLVMConfig to provide split variables for in-source and generated\ninclude paths. Currently, it uses a single value for both\nLLVM_INCLUDE_DIRS and LLVM_INCLUDE_DIR which works for install tree but\nfails hard at build tree (where LLVM_INCLUDE_DIR incorrectly contains\nmultiple values).\n\nInstead, put the generated directory in LLVM_INCLUDE_DIR, and the source\ntree in LLVM_MAIN_INCLUDE_DIR which is consistent with in-LLVM builds.\nFor install tree, both variables will have the same value.\n\nDifferential Revision: https://reviews.llvm.org/D58109\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353924 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "74c86f74996dd5d17c8f6e8a2ee3d73e0ec9ba41", "tree": "71a1bbca064d6d628bf02ff59428d40a0446a070", "parents": [ "7cbba3526136e1802a9292283248aaba85556320" ], "author": { "name": "Anton Afanasyev", "email": "anton.a.afanasyev@gmail.com", "time": "Wed Feb 13 08:26:43 2019 +0000" }, "committer": { "name": "Anton Afanasyev", "email": "anton.a.afanasyev@gmail.com", "time": "Wed Feb 13 08:26:43 2019 +0000" }, "message": "[X86][SLP] Enable SLP vectorization for 128-bit horizontal X86 instructions (add, sub)\n\nTry to use 64-bit SLP vectorization. In addition to horizontal instrs\nthis change triggers optimizations for partial vector operations (for instance,\nusing low halfs of 128-bit registers xmm0 and xmm1 to multiply \u003c2 x float\u003e by\n\u003c2 x float\u003e).\n\nFixes llvm.org/PR32433\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353923 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "7cbba3526136e1802a9292283248aaba85556320", "tree": "4e051395a970d16189b64c16aebc236b8cda6b3d", "parents": [ "0e7879be673c21afe9629400c32eab93e37bbf2c" ], "author": { "name": "Craig Topper", "email": "craig.topper@intel.com", "time": "Wed Feb 13 07:42:34 2019 +0000" }, "committer": { "name": "Craig Topper", "email": "craig.topper@intel.com", "time": "Wed Feb 13 07:42:34 2019 +0000" }, "message": "[X86] Use default expansion for (i64 fp_to_uint f80) when avx512 is enabled on 64-bit targets to match what happens without avx512.\n\nIn 64-bit mode prior to avx512 we use Expand, but with avx512 we need to make f32/f64 conversions Legal so we use Custom and then do our own expansion for f80. But this seems to produce codegen differences relative to avx2. This patch corrects this.\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353921 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "0e7879be673c21afe9629400c32eab93e37bbf2c", "tree": "0f38e0589f6805b56c28135673b74278edf576d0", "parents": [ "cb9edfda6caef0cb56568663e3dca8a9c3368f1e" ], "author": { "name": "Craig Topper", "email": "craig.topper@intel.com", "time": "Wed Feb 13 07:42:31 2019 +0000" }, "committer": { "name": "Craig Topper", "email": "craig.topper@intel.com", "time": "Wed Feb 13 07:42:31 2019 +0000" }, "message": "[X86] Refactor the FP_TO_INTHelper interface. NFCI\n\n-Pull the final stack load creation from the two callers into the helper.\n-Return a single SDValue instead of a std::pair.\n-Remove the Replace flag which isn\u0027t really needed.\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353920 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "cb9edfda6caef0cb56568663e3dca8a9c3368f1e", "tree": "425d8af044bfb198f32c477f8e0b0a7143401300", "parents": [ "9b0581974207d38a5ef880bed858a33e6cbf12db" ], "author": { "name": "Eugene Leviant", "email": "eleviant@accesssoftek.com", "time": "Wed Feb 13 07:34:54 2019 +0000" }, "committer": { "name": "Eugene Leviant", "email": "eleviant@accesssoftek.com", "time": "Wed Feb 13 07:34:54 2019 +0000" }, "message": "[llvm-objcopy] Add --strip-unneeded-symbol(s)\n\nDifferential revision: https://reviews.llvm.org/D58027\n\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353919 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "9b0581974207d38a5ef880bed858a33e6cbf12db", "tree": "010f132ab84c6d9c10d6d8218b9b27c40d7750e1", "parents": [ "45dd4f2fbc10da612dd89779d2da659aae397c25" ], "author": { "name": "Max Kazantsev", "email": "max.kazantsev@azul.com", "time": "Wed Feb 13 06:12:48 2019 +0000" }, "committer": { "name": "Max Kazantsev", "email": "max.kazantsev@azul.com", "time": "Wed Feb 13 06:12:48 2019 +0000" }, "message": "[LoopSimplifyCFG] Re-enable const branch folding by default\n\nKnown underlying bugs have been fixed, intensive fuzz testing did not\nfind any new problems. Re-enabling by default. Feel free to revert if\nit causes any functional failures.\n\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353911 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "45dd4f2fbc10da612dd89779d2da659aae397c25", "tree": "03f51bca32476b7126614aa863deea5cfa5bc762", "parents": [ "c987e861dd1299b88d66c656f5f7d136362c1466" ], "author": { "name": "Fangrui Song", "email": "maskray@google.com", "time": "Wed Feb 13 01:51:45 2019 +0000" }, "committer": { "name": "Fangrui Song", "email": "maskray@google.com", "time": "Wed Feb 13 01:51:45 2019 +0000" }, "message": "[llvm-readobj] Dump GNU_PROPERTY_X86_FEATURE_2_{NEEDED,USED} notes in .note.gnu.property\n\nSummary: And change the output (\"X86 features\" -\u003e \"x86 feature\") a bit.\n\nReviewers: grimar, xiangzhangllvm, hjl.tools, rupprecht\n\nReviewed By: rupprecht\n\nSubscribers: rupprecht, llvm-commits\n\nTags: #llvm\n\nDifferential Revision: https://reviews.llvm.org/D58112\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353908 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "c987e861dd1299b88d66c656f5f7d136362c1466", "tree": "d797771c1159a9ca267f1620253b041e24e835e6", "parents": [ "a1881f222dd5d43a992a21b3d7fa463fc190a4e7" ], "author": { "name": "Reid Kleckner", "email": "rnk@google.com", "time": "Wed Feb 13 01:39:32 2019 +0000" }, "committer": { "name": "Reid Kleckner", "email": "rnk@google.com", "time": "Wed Feb 13 01:39:32 2019 +0000" }, "message": "[MC] Make symbol version errors non-fatal\n\nWe stil don\u0027t have a source location, which is pretty lame, but at least\nwe won\u0027t tell the user to file a clang bug report anymore.\n\nFixes PR40712\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353907 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "a1881f222dd5d43a992a21b3d7fa463fc190a4e7", "tree": "4f411c92fb8c3fd9eec710eb3de3b288edb78a52", "parents": [ "629efa3c8d69c8d504db299168c7ef0f9c73270e" ], "author": { "name": "Jonas Devlieghere", "email": "jonas@devlieghere.com", "time": "Wed Feb 13 00:32:21 2019 +0000" }, "committer": { "name": "Jonas Devlieghere", "email": "jonas@devlieghere.com", "time": "Wed Feb 13 00:32:21 2019 +0000" }, "message": "[dsymutil] Improve readability of cloneAllCompileUnits (NFC)\n\nAdd some newlines and improve consistency between the two loops.\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353904 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "629efa3c8d69c8d504db299168c7ef0f9c73270e", "tree": "47de0688a5fe7a09021d194ecf747a6dc94bde67", "parents": [ "7c0006da85146897fe2837f65f5ce501e707cd7f" ], "author": { "name": "Jonas Devlieghere", "email": "jonas@devlieghere.com", "time": "Wed Feb 13 00:32:06 2019 +0000" }, "committer": { "name": "Jonas Devlieghere", "email": "jonas@devlieghere.com", "time": "Wed Feb 13 00:32:06 2019 +0000" }, "message": "[dsymutil] Don\u0027t clone empty CUs\n\nThe DWARF standard says that an empty compile unit is not valid:\n\n\u003e Each such contribution consists of a compilation unit header (see\n\u003e Section 7.5.1.1 on page 200) followed by a single DW_TAG_compile_unit or\n\u003e DW_TAG_partial_unit debugging information entry, together with its\n\u003e children.\n\nTherefore we shouldn\u0027t clone them in dsymutil.\n\nDifferential revision: https://reviews.llvm.org/D57979\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353903 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "7c0006da85146897fe2837f65f5ce501e707cd7f", "tree": "07060447beb6f2dc5a73991f0275a2fea69a619a", "parents": [ "389e0664f8ef2dde5b03cea236ef0a15610aaadf" ], "author": { "name": "Alina Sbirlea", "email": "asbirlea@google.com", "time": "Tue Feb 12 23:48:02 2019 +0000" }, "committer": { "name": "Alina Sbirlea", "email": "asbirlea@google.com", "time": "Tue Feb 12 23:48:02 2019 +0000" }, "message": "[MemorySSA \u0026 LoopPassManager] Add remaining book keeping [NFCI].\n\nAdd plumbing to get MemorySSA in the remaining loop passes.\nAlso update unit test to add the dependency.\n[EnableMSSALoopDependency remains disabled].\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353901 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "389e0664f8ef2dde5b03cea236ef0a15610aaadf", "tree": "2743d57a780823220a391ec82eb6529a73bddeef", "parents": [ "b232056f5e991908e8ec4ae54427f91eb49f6558" ], "author": { "name": "Matt Arsenault", "email": "Matthew.Arsenault@amd.com", "time": "Tue Feb 12 23:44:13 2019 +0000" }, "committer": { "name": "Matt Arsenault", "email": "Matthew.Arsenault@amd.com", "time": "Tue Feb 12 23:44:13 2019 +0000" }, "message": "AMDGPU: Try to use function specific ST\n\nSubtargets are a function level property, so ideally we would\neliminate everywhere that needs to check the global one. Rename the\nfunction to try avoiding confusion.\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353900 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "b232056f5e991908e8ec4ae54427f91eb49f6558", "tree": "f12f499e9e62e35de381a8ccddbb14654ff51061", "parents": [ "17f2e5bd11f955eda64800b8659bd03f6bdad1bf" ], "author": { "name": "Matt Arsenault", "email": "Matthew.Arsenault@amd.com", "time": "Tue Feb 12 23:30:11 2019 +0000" }, "committer": { "name": "Matt Arsenault", "email": "Matthew.Arsenault@amd.com", "time": "Tue Feb 12 23:30:11 2019 +0000" }, "message": "AMDGPU: Ignore CodeObjectV3 when inlining\n\nThis was inhibiting inlining of library functions when clang was\ninvoking the inliner directly. This is covering a bit of a mess with\nsubtarget feature handling, and this shouldn\u0027t be a subtarget\nfeature. The behavior is different depending on whether you are using\na -mattr flag in clang, or llc, opt.\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353899 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "17f2e5bd11f955eda64800b8659bd03f6bdad1bf", "tree": "82655ec209388a6b16a49676b5328b0f0065cf82", "parents": [ "bb0402417fc1d386649870e3347225d3cf9483b9" ], "author": { "name": "Jonas Paulsson", "email": "paulsson@linux.vnet.ibm.com", "time": "Tue Feb 12 23:13:18 2019 +0000" }, "committer": { "name": "Jonas Paulsson", "email": "paulsson@linux.vnet.ibm.com", "time": "Tue Feb 12 23:13:18 2019 +0000" }, "message": "[SystemZ] Remember to cast value to void to disable warning.\n\nHopefully fixes buildbot problems.\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353898 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "bb0402417fc1d386649870e3347225d3cf9483b9", "tree": "ffb15d7c8d5e7dc9fa0829b55c247971fadd6080", "parents": [ "56bc56e24546966090a680b027d4e6b8460312d4" ], "author": { "name": "Alina Sbirlea", "email": "asbirlea@google.com", "time": "Tue Feb 12 23:05:40 2019 +0000" }, "committer": { "name": "Alina Sbirlea", "email": "asbirlea@google.com", "time": "Tue Feb 12 23:05:40 2019 +0000" }, "message": "[LICM] Cap the clobbering calls in LICM.\n\nSummary:\nUnlimitted number of calls to getClobberingAccess can lead to high\ncompile times in pathological cases.\nSwitching EnableLicmCap flag from bool to int, and enabling to default 100.\n(tested to be appropriate for current bechmarks)\nWe can revisit this value when enabling MemorySSA.\n\nReviewers: sanjoy, chandlerc, george.burgess.iv\n\nSubscribers: jlebar, llvm-commits\n\nTags: #llvm\n\nDifferential Revision: https://reviews.llvm.org/D57968\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353897 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "56bc56e24546966090a680b027d4e6b8460312d4", "tree": "25873442c0a55f3253b9195e7883f911b8cf221b", "parents": [ "9cf30e4f29260dc644de2c4a5521a0c00a0abc7a" ], "author": { "name": "Philip Reames", "email": "listmail@philipreames.com", "time": "Tue Feb 12 23:00:07 2019 +0000" }, "committer": { "name": "Philip Reames", "email": "listmail@philipreames.com", "time": "Tue Feb 12 23:00:07 2019 +0000" }, "message": "[Tests] A few more live-in deopt lowering tests\n\nNothing super interesting, just making sure obvious cases work.\n\n\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353895 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "9cf30e4f29260dc644de2c4a5521a0c00a0abc7a", "tree": "a033cb8376b8ddaa4f728b27279d581cc335680e", "parents": [ "029ba20623d31c2093e96146498e05f46222636d" ], "author": { "name": "Konstantin Zhuravlyov", "email": "kzhuravl_dev@outlook.com", "time": "Tue Feb 12 22:49:49 2019 +0000" }, "committer": { "name": "Konstantin Zhuravlyov", "email": "kzhuravl_dev@outlook.com", "time": "Tue Feb 12 22:49:49 2019 +0000" }, "message": "AMDGPU/NFC: Remove SubtargetFeatureISAVersion since it is not used anywhere\n\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353892 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "029ba20623d31c2093e96146498e05f46222636d", "tree": "61a3d469b3aa18d1600aa31da06f1f6eb6547155", "parents": [ "e126f1baea2bbf17a27e782305f0a80e66b919a9" ], "author": { "name": "Konstantin Zhuravlyov", "email": "kzhuravl_dev@outlook.com", "time": "Tue Feb 12 22:29:25 2019 +0000" }, "committer": { "name": "Konstantin Zhuravlyov", "email": "kzhuravl_dev@outlook.com", "time": "Tue Feb 12 22:29:25 2019 +0000" }, "message": "AMDGPU: Remove duplicate processor (gfx900)\n\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353889 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "e126f1baea2bbf17a27e782305f0a80e66b919a9", "tree": "7646cfaff88ade4d7f577729f197353ebd1b8709", "parents": [ "4499765b6a914aecfbe2daf0a04518afb9764c18" ], "author": { "name": "David Major", "email": "dmajor@mozilla.com", "time": "Tue Feb 12 22:24:45 2019 +0000" }, "committer": { "name": "David Major", "email": "dmajor@mozilla.com", "time": "Tue Feb 12 22:24:45 2019 +0000" }, "message": "[gn build] Separate debug and optimization settings\n\nThis patch adds an `is_optimized` variable, orthogonal to `is_debug`, to allow for a gn analogue to `RelWithDebInfo` builds.\n\nAs part of this we\u0027ll want to explicitly enable GC+ICF, for the sake of `is_debug \u0026\u0026 is_optimized` builds. The flags normally default to true except that if you pass `/DEBUG` they default to false.\n\nDifferential Revision: https://reviews.llvm.org/D58075\n\n\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353888 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "4499765b6a914aecfbe2daf0a04518afb9764c18", "tree": "e105884564ef7b43dc7255086443daa20d4c2471", "parents": [ "2d0b1b5542de55d6502f38b73575681cb337583a" ], "author": { "name": "Bjorn Pettersson", "email": "bjorn.a.pettersson@ericsson.com", "time": "Tue Feb 12 22:11:20 2019 +0000" }, "committer": { "name": "Bjorn Pettersson", "email": "bjorn.a.pettersson@ericsson.com", "time": "Tue Feb 12 22:11:20 2019 +0000" }, "message": "[SelectionDAG] Clean up comments in SelectionDAGBuilder.h. NFC\n\nRemove redundant function/variable names from doxygen\ncomments (as suggested in https://reviews.llvm.org/D57697).\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353886 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "2d0b1b5542de55d6502f38b73575681cb337583a", "tree": "ceff483fa94dcb9dbf443e97686baa4b8f3e5ef1", "parents": [ "5466d0bac4d8df259c03a6559f1d6176e116df55" ], "author": { "name": "Erik Pilkington", "email": "erik.pilkington@gmail.com", "time": "Tue Feb 12 21:55:38 2019 +0000" }, "committer": { "name": "Erik Pilkington", "email": "erik.pilkington@gmail.com", "time": "Tue Feb 12 21:55:38 2019 +0000" }, "message": "Fix auto-upgrade for the new parameter to llvm.objectsize\n\nr352664 added a \u0027dynamic\u0027 parameter to objectsize, but the AutoUpgrade\nchanges were incomplete. Also, fix an off-by-one error I made in the\nupgrade logic that is now no longer unreachable.\n\nDifferential revision: https://reviews.llvm.org/D58071\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353884 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "5466d0bac4d8df259c03a6559f1d6176e116df55", "tree": "70a1cb9ebb0e895fe6ee21386b07df7b57490f1f", "parents": [ "5833ba562fc6fda4853fa1de5fd25f39e2417989" ], "author": { "name": "Sanjay Patel", "email": "spatel@rotateright.com", "time": "Tue Feb 12 21:49:56 2019 +0000" }, "committer": { "name": "Sanjay Patel", "email": "spatel@rotateright.com", "time": "Tue Feb 12 21:49:56 2019 +0000" }, "message": "[ConstProp] add test for miscompile from bitcast transform; NFC\n\nThis problem goes with the fix in D51215.\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353883 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "5833ba562fc6fda4853fa1de5fd25f39e2417989", "tree": "52d1dc9239a4153d16a270de69fd46a66a90d94d", "parents": [ "638090e32fa4c18e1251e265e2b16d990ff96e28" ], "author": { "name": "Jordan Rupprecht", "email": "rupprecht@google.com", "time": "Tue Feb 12 20:37:33 2019 +0000" }, "committer": { "name": "Jordan Rupprecht", "email": "rupprecht@google.com", "time": "Tue Feb 12 20:37:33 2019 +0000" }, "message": "[llvm-dwp] Use color-formatted error reporting\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353876 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "638090e32fa4c18e1251e265e2b16d990ff96e28", "tree": "ba676890c8412a341296d773ede47bbea0c883f3", "parents": [ "02f8fc49ff4ad5f37a0bd7ad4264e7fb98745b99" ], "author": { "name": "Sean Fertile", "email": "sfertile@ca.ibm.com", "time": "Tue Feb 12 20:03:04 2019 +0000" }, "committer": { "name": "Sean Fertile", "email": "sfertile@ca.ibm.com", "time": "Tue Feb 12 20:03:04 2019 +0000" }, "message": "Fix undefined behaviour in PPCInstPrinter::printBranchOperand.\n\nFix the undefined behaviour introduced by my previous patch r353865 (left\nshifting a potentially negative value), which was caught by the bots that run\nUBSan.\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353874 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "02f8fc49ff4ad5f37a0bd7ad4264e7fb98745b99", "tree": "ff99562bd13966fbbc2c0c3a965e903c8d818e56", "parents": [ "80d4df92c75843bafe6610ce5877413f2510a600" ], "author": { "name": "Jordan Rupprecht", "email": "rupprecht@google.com", "time": "Tue Feb 12 20:00:51 2019 +0000" }, "committer": { "name": "Jordan Rupprecht", "email": "rupprecht@google.com", "time": "Tue Feb 12 20:00:51 2019 +0000" }, "message": "[llvm-dwp] Avoid writing the output dwp file when there is an error\n\nSummary: Use ToolOutputFile to clean up the output file unless dwp actually finishes successfully.\n\nReviewers: dblaikie\n\nReviewed By: dblaikie\n\nSubscribers: jdoerfert, llvm-commits\n\nTags: #llvm\n\nDifferential Revision: https://reviews.llvm.org/D58130\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353873 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "80d4df92c75843bafe6610ce5877413f2510a600", "tree": "4a61d9e96c16ba6d213afc29846009a3be00daa5", "parents": [ "9becf5911e89bbe440fb70080bbed42735bf1900" ], "author": { "name": "Nikita Popov", "email": "nikita.ppv@gmail.com", "time": "Tue Feb 12 18:55:53 2019 +0000" }, "committer": { "name": "Nikita Popov", "email": "nikita.ppv@gmail.com", "time": "Tue Feb 12 18:55:53 2019 +0000" }, "message": "[AArch64] Expand v8i8 cttz (PR39729)\n\nFix for https://bugs.llvm.org/show_bug.cgi?id\u003d39729.\n\nRather than adding just a case for v8i8 I\u0027m setting cttz to expand\nfor all vector types.\n\nDifferential Revision: https://reviews.llvm.org/D58008\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353872 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "9becf5911e89bbe440fb70080bbed42735bf1900", "tree": "6c3a65ec954ab37feb6d17ae9aa1ea301174ca8e", "parents": [ "3044368de258cd87b7a2e40be70212b2d2d35208" ], "author": { "name": "Philip Reames", "email": "listmail@philipreames.com", "time": "Tue Feb 12 18:33:01 2019 +0000" }, "committer": { "name": "Philip Reames", "email": "listmail@philipreames.com", "time": "Tue Feb 12 18:33:01 2019 +0000" }, "message": "[InlineSpiller] Fix a crash due to lack of forward progress from remat (try 2)\n\nThis is a recommit of r335091 Add more test cases for deopt-operands via regalloc, and r335077 [InlineSpiller] Fix a crash due to lack of forward progress from remat specifically for STATEPOINT. They were reverted due to a crash.\n\nThis change includes the text of both original changes, but also includes three aditional pieces:\n1) A bug fix for the observed crash. I had failed to record the failed remat value as live which resulted in an instruction being deleted which still had uses. With the machine verifier, this is caught quickly. Without it, we fail in StackSlotColoring due to an empty live interval from LiveStack.\n2) A test case which demonstrates the fix for (1). See @test11.\n3) A control flag which defaults to disabling this for the moment. Once I\u0027ve run more extensive validaton, I will switch the default and then remove this flag. \n\n\n\n\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353871 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "3044368de258cd87b7a2e40be70212b2d2d35208", "tree": "e8a1715a05635f64e9a519003b17bd32800a5de4", "parents": [ "bdb2519a0ccb9b3ef4fd0a4ab275fad3ce76b139" ], "author": { "name": "Jonas Paulsson", "email": "paulsson@linux.vnet.ibm.com", "time": "Tue Feb 12 18:06:06 2019 +0000" }, "committer": { "name": "Jonas Paulsson", "email": "paulsson@linux.vnet.ibm.com", "time": "Tue Feb 12 18:06:06 2019 +0000" }, "message": "[SystemZ] Use VGM whenever possible to load FP immediates.\n\nisFPImmLegal() has been extended to recognize certain FP immediates that can\nbe built with VGM (Vector Generate Mask).\n\nThese scalar FP immediates (that were previously loaded from the constant\npool) are now selected as VGMF/VGMG in Select().\n\nReview: Ulrich Weigand\nhttps://reviews.llvm.org/D58003\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353867 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "bdb2519a0ccb9b3ef4fd0a4ab275fad3ce76b139", "tree": "dd9cfeec6217de3a6b8d16b115b15caf447b4c0c", "parents": [ "96db5f238ebb4254fca19c8b734d76a0442c0293" ], "author": { "name": "Sean Fertile", "email": "sfertile@ca.ibm.com", "time": "Tue Feb 12 17:48:22 2019 +0000" }, "committer": { "name": "Sean Fertile", "email": "sfertile@ca.ibm.com", "time": "Tue Feb 12 17:48:22 2019 +0000" }, "message": "[PowerPC] Fix printing of negative offsets in call instruction dissasembly.\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353865 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "96db5f238ebb4254fca19c8b734d76a0442c0293", "tree": "8266ef5226ad7d3992dd807abeaa1c4fb631f154", "parents": [ "d5af09c2145807ff2c85f81c7c2cf688d5e91a60" ], "author": { "name": "Jessica Paquette", "email": "jpaquette@apple.com", "time": "Tue Feb 12 17:38:34 2019 +0000" }, "committer": { "name": "Jessica Paquette", "email": "jpaquette@apple.com", "time": "Tue Feb 12 17:38:34 2019 +0000" }, "message": "[GlobalISel][NFC] Gardening: Make translateSimpleUnaryIntrinsic general\n\nInstead of only having this code work for unary intrinsics, have it work for\nan arbitrary number of parameters.\n\nFactor out the cases that fall under this (fma, pow).\n\nThis makes it a bit easier to add more intrinsics which don\u0027t require any\nspecial work.\n\nDifferential Revision: https://reviews.llvm.org/D58079\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353863 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "d5af09c2145807ff2c85f81c7c2cf688d5e91a60", "tree": "906ae47f01d222dd21721f96da55a74f44396191", "parents": [ "d1ae68f0324b7f1b2d7a9cabcb954bce427cb64a" ], "author": { "name": "Daniel Sanders", "email": "daniel_l_sanders@apple.com", "time": "Tue Feb 12 17:36:57 2019 +0000" }, "committer": { "name": "Daniel Sanders", "email": "daniel_l_sanders@apple.com", "time": "Tue Feb 12 17:36:57 2019 +0000" }, "message": "[tablegen] Add locations to many PrintFatalError() calls\n\nSummary:\nWhile working on the GISel Combiner, I noticed I was producing location-less\nerror messages fairly often and set about fixing this. In the process, I\nnoticed quite a few places elsewhere in TableGen that also neglected to include\na relevant location.\n\nThis patch adds locations to errors that relate to a specific record (or a\nfield within it) and also have easy access to the relevant location. This is\nparticularly useful when multiclasses are involved as many of these errors\nrefer to the full name of a record and it\u0027s difficult to guess which substring\nis grep-able.\n\nUnfortunately, tablegen currently only supports Record granularity so it\u0027s not\ncurrently possible to point at a specific Init so these sometimes point at the\nrecord that caused the error rather than the precise origin of the error.\n\nReviewers: bogner, aditya_nandakumar, volkan, aemerson, paquette, nhaehnle\n\nReviewed By: nhaehnle\n\nSubscribers: jdoerfert, nhaehnle, asb, rbar, johnrusso, simoncook, apazos, sabuasal, niosHD, jrtc27, zzheng, edward-jones, rogfer01, MartinMosbeck, brucehoult, the_o, PkmX, jocewei, llvm-commits\n\nTags: #llvm\n\nDifferential Revision: https://reviews.llvm.org/D58077\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353862 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "d1ae68f0324b7f1b2d7a9cabcb954bce427cb64a", "tree": "03ffbb47120aa6034d4f74a8ca8a36f8c7c477a2", "parents": [ "1fcc05880eb106562929885857f55f558dc2a36e" ], "author": { "name": "Jessica Paquette", "email": "jpaquette@apple.com", "time": "Tue Feb 12 17:28:17 2019 +0000" }, "committer": { "name": "Jessica Paquette", "email": "jpaquette@apple.com", "time": "Tue Feb 12 17:28:17 2019 +0000" }, "message": "[GlobalISel][AArch64] Select llvm.bswap* for non-vector types\n\nThis teaches the IRTranslator to emit G_BSWAP when it runs into\nIntrinsic::bswap. This allows us to select G_BSWAP for non-vector types in\nAArch64.\n\nAdd a select-bswap.mir test, and add global isel checks to a couple existing\ntests in test/CodeGen/AArch64.\n\nThis doesn\u0027t handle every bswap case, since some of these rely on known bits\nstuff. This just lets us handle the naive case.\n\nDifferential Revision: https://reviews.llvm.org/D58081\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353861 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "1fcc05880eb106562929885857f55f558dc2a36e", "tree": "0133d7fc467608d6e348fc04bf27e09f989b9494", "parents": [ "51dad635adaf53f3e5d2d563fccbe786244de7f7" ], "author": { "name": "Simon Pilgrim", "email": "llvm-dev@redking.me.uk", "time": "Tue Feb 12 17:22:35 2019 +0000" }, "committer": { "name": "Simon Pilgrim", "email": "llvm-dev@redking.me.uk", "time": "Tue Feb 12 17:22:35 2019 +0000" }, "message": "[X86][AVX] Enable shuffle combining support for zero_extend \n\nA more limited version of rL352997 that had to be disabled in rL353198 - allow extension of any 128/256/512 bit vector that at least uses byte sized scalars.\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353860 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "51dad635adaf53f3e5d2d563fccbe786244de7f7", "tree": "1c18c0433edf3848b4badac19a692994e598eceb", "parents": [ "9d61e28a96cd0918a6f5fe7e6fc262bcb939277e" ], "author": { "name": "Sanjay Patel", "email": "spatel@rotateright.com", "time": "Tue Feb 12 17:07:47 2019 +0000" }, "committer": { "name": "Sanjay Patel", "email": "spatel@rotateright.com", "time": "Tue Feb 12 17:07:47 2019 +0000" }, "message": "[DAGCombiner] convert logic-of-setcc into bit magic (PR40611)\n\nIf we\u0027re comparing some value for equality against 2 constants\nand those constants have an absolute difference of just 1 bit,\nthen we can offset and mask off that 1 bit and reduce to a single\ncompare against zero:\n and/or (setcc X, C0, ne), (setcc X, C1, ne/eq) --\u003e\n setcc ((add X, -C1), ~(C0 - C1)), 0, ne/eq\n\nhttps://rise4fun.com/Alive/XslKj\n\nThis transform is disabled by default using a TLI hook\n(\"convertSetCCLogicToBitwiseLogic()\").\n\nThat should be overridden for AArch64, MIPS, Sparc and possibly\nothers based on the asm shown in:\nhttps://bugs.llvm.org/show_bug.cgi?id\u003d40611\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353859 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "9d61e28a96cd0918a6f5fe7e6fc262bcb939277e", "tree": "b9e7cf1811368cb7ac8349120732ae666239e731", "parents": [ "040509be6c2ad9f4b8dffebecad2a2a656d6c9db" ], "author": { "name": "Sanjay Patel", "email": "spatel@rotateright.com", "time": "Tue Feb 12 16:44:37 2019 +0000" }, "committer": { "name": "Sanjay Patel", "email": "spatel@rotateright.com", "time": "Tue Feb 12 16:44:37 2019 +0000" }, "message": "[x86] add negative tests for setcc folds; NFC\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353855 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "040509be6c2ad9f4b8dffebecad2a2a656d6c9db", "tree": "1a0e0d478bf67585edbc849ffa7c0dbda8fd98f6", "parents": [ "bcc80f35660542635f5a17200d20ad6c4367dd29" ], "author": { "name": "whitequark", "email": "whitequark@whitequark.org", "time": "Tue Feb 12 16:41:50 2019 +0000" }, "committer": { "name": "whitequark", "email": "whitequark@whitequark.org", "time": "Tue Feb 12 16:41:50 2019 +0000" }, "message": "[SelectionDAG] Fix return calling convention in expansion of ?MULO\n\nSummary:\nThe SMULO/UMULO DAG nodes, when not directly supported by the target,\nexpand to a multiplication twice as wide. In case that the resulting\ntype is not legal, the legalizer cannot directly call the intrinsic\nwith the wide arguments; instead, it \"pre-lowers\" them by splitting\nthem in halves.\n\nrL283203 made sure that on big endian targets, the legalizer passes\nthe argument halves in the correct order. It did not do the same\nfor the return value halves because the existing code used a hack;\nit put an illegal type into DAG and hoped that nothing would break\nand it would be correctly lowered elsewhere.\n\nrL307207 fixed this, handling return value halves similar to how\nargument handles are handled, but did not take big-endian targets\ninto account.\n\nThis commit fixes the expansion on big-endian targets, such as\nthe out-of-tree OR1K target.\n\nReviewers: eli.friedman, vadimcn\n\nSubscribers: george-hopkins, efriedma, llvm-commits\n\nDifferential Revision: https://reviews.llvm.org/D45355\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353854 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "bcc80f35660542635f5a17200d20ad6c4367dd29", "tree": "2f4bfd2ea1fec4bc5a159288b4629103efb31a5f", "parents": [ "02b7d8fb42db96a5e7f6d4c2b6eb789fab4d26de" ], "author": { "name": "Andrea Di Biagio", "email": "Andrea_DiBiagio@sn.scee.net", "time": "Tue Feb 12 16:18:57 2019 +0000" }, "committer": { "name": "Andrea Di Biagio", "email": "Andrea_DiBiagio@sn.scee.net", "time": "Tue Feb 12 16:18:57 2019 +0000" }, "message": "[MCA] Improved debug prints. NFC\n\n\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353852 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "02b7d8fb42db96a5e7f6d4c2b6eb789fab4d26de", "tree": "552744c9f590a19fe2092cd95999227624276aa7", "parents": [ "d135f2406f9432310af24066448ed626abf62909" ], "author": { "name": "Simon Pilgrim", "email": "llvm-dev@redking.me.uk", "time": "Tue Feb 12 16:10:50 2019 +0000" }, "committer": { "name": "Simon Pilgrim", "email": "llvm-dev@redking.me.uk", "time": "Tue Feb 12 16:10:50 2019 +0000" }, "message": "[PowerPC] Regenerate test\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353851 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "d135f2406f9432310af24066448ed626abf62909", "tree": "f272dce5300c73f7bc4fde69ada8b01d7c2e27a5", "parents": [ "48fad281443b350a984e6d694b18caa84baec781" ], "author": { "name": "Matt Arsenault", "email": "Matthew.Arsenault@amd.com", "time": "Tue Feb 12 15:04:03 2019 +0000" }, "committer": { "name": "Matt Arsenault", "email": "Matthew.Arsenault@amd.com", "time": "Tue Feb 12 15:04:03 2019 +0000" }, "message": "AMDGPU/GlobalISel: Add more insert/extract testcases\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353848 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "48fad281443b350a984e6d694b18caa84baec781", "tree": "c0fc733cfb37b9c4dd777a840f9da26f06df6d7c", "parents": [ "19350f7dddc1b086a215743efab24b86fab6afbc" ], "author": { "name": "David Green", "email": "david.green@arm.com", "time": "Tue Feb 12 15:02:57 2019 +0000" }, "committer": { "name": "David Green", "email": "david.green@arm.com", "time": "Tue Feb 12 15:02:57 2019 +0000" }, "message": "[Codegen] Make sure kill flags are not incorrect from removed machine phi\u0027s\n\nWe need to clear the kill flags on both SingleValReg and OldReg, to ensure they remain\nconservatively correct.\n\nDifferential Revision: https://reviews.llvm.org/D58114\n\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353847 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "19350f7dddc1b086a215743efab24b86fab6afbc", "tree": "2758537c89a3e5d28bfdc36723e87ced1328f596", "parents": [ "bb8d9165a419a4281d34391148296a069fab5fe1" ], "author": { "name": "Jordan Rupprecht", "email": "rupprecht@google.com", "time": "Tue Feb 12 15:01:07 2019 +0000" }, "committer": { "name": "Jordan Rupprecht", "email": "rupprecht@google.com", "time": "Tue Feb 12 15:01:07 2019 +0000" }, "message": "[llvm-dwp] Abort when dwo_id is unset\n\nSummary:\nAn empty dwo_id indicates a degenerate .dwo file that should not have been generated in the first place. Instead of discovering this error later when merging with another degenerate .dwo file, print an error immediately when noticing an unset dwo_id, including the filename of the offending file.\n\nTest case created by compiling a trivial file w/ `-fno-split-dwarf-inlining -gmlt -gsplit-dwarf -c` prior to r353771\n\nReviewers: dblaikie\n\nReviewed By: dblaikie\n\nSubscribers: jdoerfert, aprantl, llvm-commits\n\nTags: #llvm\n\nDifferential Revision: https://reviews.llvm.org/D58085\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353846 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "bb8d9165a419a4281d34391148296a069fab5fe1", "tree": "a43f8d5c009b2bfca98b334c3fd222efd7ac4615", "parents": [ "bfe35f840da65d23ecb2fce2cc93143013d169f8" ], "author": { "name": "Matt Arsenault", "email": "Matthew.Arsenault@amd.com", "time": "Tue Feb 12 14:54:55 2019 +0000" }, "committer": { "name": "Matt Arsenault", "email": "Matthew.Arsenault@amd.com", "time": "Tue Feb 12 14:54:55 2019 +0000" }, "message": "AMDGPU/GlobalISel: Only make f16 constants legal on f16 targets\n\nWe could deal with it, but there\u0027s no real point.\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353845 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "bfe35f840da65d23ecb2fce2cc93143013d169f8", "tree": "b1c3e2a9fbf6a921d0312418e34028e6aca9fabd", "parents": [ "31cd6654ed7caefcb84f35450c2bc77f263bf780" ], "author": { "name": "Matt Arsenault", "email": "Matthew.Arsenault@amd.com", "time": "Tue Feb 12 14:54:54 2019 +0000" }, "committer": { "name": "Matt Arsenault", "email": "Matthew.Arsenault@amd.com", "time": "Tue Feb 12 14:54:54 2019 +0000" }, "message": "GlobalISel: Use default rounding mode when extending fconstant\n\nI don\u0027t think this matters since the values should all be exactly\nrepresentable.\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353844 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "31cd6654ed7caefcb84f35450c2bc77f263bf780", "tree": "c8a5e1bdba7914f39fd134f10b6e3a4ed4b71d3a", "parents": [ "aaee83c287d971d7e537c7f46c555158aee01d0f" ], "author": { "name": "Matt Arsenault", "email": "Matthew.Arsenault@amd.com", "time": "Tue Feb 12 14:54:52 2019 +0000" }, "committer": { "name": "Matt Arsenault", "email": "Matthew.Arsenault@amd.com", "time": "Tue Feb 12 14:54:52 2019 +0000" }, "message": "GlobalISel: Move some more legalize cases into functions\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353843 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "aaee83c287d971d7e537c7f46c555158aee01d0f", "tree": "3d117110d86f7f99b9deffc2043790c7d67500eb", "parents": [ "185dbb6e1b17c81f35766849780be763ea3835be" ], "author": { "name": "Sam McCall", "email": "sam.mccall@gmail.com", "time": "Tue Feb 12 14:43:30 2019 +0000" }, "committer": { "name": "Sam McCall", "email": "sam.mccall@gmail.com", "time": "Tue Feb 12 14:43:30 2019 +0000" }, "message": "[LoopSimplifyCFG] Fix test broken in release mode in r353813\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353842 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "185dbb6e1b17c81f35766849780be763ea3835be", "tree": "b306f1dba9fe4ae78a7419136740d49829e88e0d", "parents": [ "814924db549473262fb9a7b93ce431e95583b1ca" ], "author": { "name": "Max Kazantsev", "email": "max.kazantsev@azul.com", "time": "Tue Feb 12 11:31:46 2019 +0000" }, "committer": { "name": "Max Kazantsev", "email": "max.kazantsev@azul.com", "time": "Tue Feb 12 11:31:46 2019 +0000" }, "message": "[NFC] Simplify code \u0026 reduce nest slightly\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353832 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "814924db549473262fb9a7b93ce431e95583b1ca", "tree": "0b875a80e5c4df73d6619d28c4b474d34e5aceb1", "parents": [ "bbe8122a3ca229c5b4f5c46d5423dcb638f0f37f" ], "author": { "name": "Jeremy Morse", "email": "jeremy.morse.llvm@gmail.com", "time": "Tue Feb 12 10:54:30 2019 +0000" }, "committer": { "name": "Jeremy Morse", "email": "jeremy.morse.llvm@gmail.com", "time": "Tue Feb 12 10:54:30 2019 +0000" }, "message": "[DebugInfo] Don\u0027t salvage load operations (PR40628).\n\nSalvaging a redundant load instruction into a debug expression hides a\nmemory read from optimisation passes. Passes that alter memory behaviour\n(such as LICM promoting memory to a register) aren\u0027t aware of these debug\nmemory reads and leave them unaltered, making the debug variable location\npoint somewhere unsafe.\n\nTeaching passes to know about these debug memory reads would be challenging\nand probably incomplete. Finding dbg.value instructions that need to be fixed\nwould likely be computationally expensive too, as more analysis would be\nrequired. It\u0027s better to not generate debug-memory-reads instead, alas.\n\nChanged tests:\n * DeadStoreElim: test for salvaging of intermediate operations contributing\n to the dead store, instead of salvaging of the redundant load,\n * GVN: remove debuginfo behaviour checks completely, this behaviour is still\n covered by other tests,\n * InstCombine: don\u0027t test for salvaged loads, we\u0027re removing that behaviour.\n\nDifferential Revision: https://reviews.llvm.org/D57962\n\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353824 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "bbe8122a3ca229c5b4f5c46d5423dcb638f0f37f", "tree": "172f6d2eb3aaffcfd755b2de6da7a00f1d27680e", "parents": [ "9d1a27061ccc93b141c4e06f03f3aa52084743c4" ], "author": { "name": "David Stenberg", "email": "david.stenberg@ericsson.com", "time": "Tue Feb 12 10:51:27 2019 +0000" }, "committer": { "name": "David Stenberg", "email": "david.stenberg@ericsson.com", "time": "Tue Feb 12 10:51:27 2019 +0000" }, "message": "[DebugInfo] Keep parameter DBG_VALUEs before prologue code\n\nSummary:\nThis is a preparatory change for removing the code from\nDebugHandlerBase::beginFunction() which changes the starting label for\nthe first non-overlapping DBG_VALUEs of parameters to the beginning of\nthe function. It does that to be able to show parameters when entering a\nfunction. However, that code does not consider what defines the values,\nwhich can result in the ranges for the debug values starting before\ntheir defining instructions. That code is removed in a follow-up patch.\n\nWhen prologue code is inserted, it leads to DBG_VALUEs that start\ndirectly in the entry block being moved down after the prologue\ninstructions. This patch fixes that by stashing away DBG_VALUEs for\nparameters before emitting the prologue, and then reinserts them at the\nstart of the block. This assumes that there is no target that somehow\nclobbers parameter registers in the frame setup; there is no such case\nin the lit tests at least.\n\nSee PR40188 for more information.\n\nReviewers: aprantl, dblaikie, rnk, jmorse\n\nReviewed By: aprantl\n\nSubscribers: bjope, llvm-commits\n\nTags: #debug-info\n\nDifferential Revision: https://reviews.llvm.org/D57510\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353823 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "9d1a27061ccc93b141c4e06f03f3aa52084743c4", "tree": "58cb2d17c22b9c5dd84c8200d1e564d73f1fc6be", "parents": [ "610cf8006d6d53675d0499126be1e455828ddd48" ], "author": { "name": "Max Kazantsev", "email": "max.kazantsev@azul.com", "time": "Tue Feb 12 09:59:44 2019 +0000" }, "committer": { "name": "Max Kazantsev", "email": "max.kazantsev@azul.com", "time": "Tue Feb 12 09:59:44 2019 +0000" }, "message": "[IndVars] Fix corner case with unreachable Phi inputs. PR40454\n\nLogic in `getInsertPointForUses` doesn\u0027t account for a corner case when `Def`\nonly comes to a Phi user from unreachable blocks. In this case, the incoming\nvalue may be arbitrary (and not even available in the input block) and break\nthe loop-related invariants that are asserted below.\n\nIn fact, if we encounter this situation, no IR modification is needed. This\nPhi will be simplified away with nearest cleanup.\n\nDifferential Revision: https://reviews.llvm.org/D58045\nReviewed By: spatel\n\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353816 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "610cf8006d6d53675d0499126be1e455828ddd48", "tree": "95d25b03f3132fb6c112ec84d57e8a6fe48e7dc5", "parents": [ "16007333e98d27bfe3886a3ee80aa89b94b047ca" ], "author": { "name": "Fangrui Song", "email": "maskray@google.com", "time": "Tue Feb 12 09:56:01 2019 +0000" }, "committer": { "name": "Fangrui Song", "email": "maskray@google.com", "time": "Tue Feb 12 09:56:01 2019 +0000" }, "message": "[llvm-readobj] Only allow 4-byte pr_data\n\nSummary: AMD64 psABI says: \"The pr_data field of each property contains a 4-byte unsigned integer.\" Thus we don\u0027t need to handle 8-byte pr_data.\n\nReviewers: mike.dvoretsky, grimar, craig.topper, xiangzhangllvm, hjl.tools\n\nReviewed By: grimar\n\nSubscribers: rupprecht, llvm-commits\n\nTags: #llvm\n\nDifferential Revision: https://reviews.llvm.org/D58103\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353815 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "16007333e98d27bfe3886a3ee80aa89b94b047ca", "tree": "796ed2461c180b6dec56c0415d215e8bb4a50c49", "parents": [ "cfa6c5f00aa546799cdf839cd4d4ebcc4da3829e" ], "author": { "name": "George Rimar", "email": "grimar@accesssoftek.com", "time": "Tue Feb 12 09:50:04 2019 +0000" }, "committer": { "name": "George Rimar", "email": "grimar@accesssoftek.com", "time": "Tue Feb 12 09:50:04 2019 +0000" }, "message": "[llvm-readobj] - Simplify .gnu.version_r dumping a bit.\n\nCurrent implementation takes \"Number of needed versions\" from DT_VERNEEDNUM\ndynamic tag entry. Though it would be a bit simpler to take it from sh_info\nsection header field directly:\nhttps://docs.oracle.com/cd/E19683-01/816-1386/chapter6-94076/index.html\n\nDifferential revision: https://reviews.llvm.org/D58048\n\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353814 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "cfa6c5f00aa546799cdf839cd4d4ebcc4da3829e", "tree": "fe6f8972c67959c9e3630d9908b7b0ea6549730e", "parents": [ "1ee69f0dfeab734abf0c67b45939286c0a6c2001" ], "author": { "name": "Max Kazantsev", "email": "max.kazantsev@azul.com", "time": "Tue Feb 12 09:37:00 2019 +0000" }, "committer": { "name": "Max Kazantsev", "email": "max.kazantsev@azul.com", "time": "Tue Feb 12 09:37:00 2019 +0000" }, "message": "[LoopSimplifyCFG] Change logic of dead loops removal to avoid hitting asserts\n\nThe function `LI.erase` has some invariants that need to be preserved when it\ntries to remove a loop which is not the top-level loop. In particular, it\nrequires loop\u0027s preheader to be strictly in loop\u0027s parent. Our current logic\nof deletion of dead blocks may erase the information about preheader before we\nhandle the loop, and therefore we may hit this assertion.\n\nThis patch changes the logic of loop deletion: we make them top-level loops\nbefore we actually erase them. This allows us to trigger the simple branch of\n`erase` logic which just detatches blocks from the loop and does not try to do\nsome complex stuff that need this invariant.\n\nThanks to @uabelho for reporting this!\n\nDifferential Revision: https://reviews.llvm.org/D57221\nReviewed By: fedor.sergeev\n\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353813 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "1ee69f0dfeab734abf0c67b45939286c0a6c2001", "tree": "450ae7da3b24febb635dc77af5bbb2e6588b0184", "parents": [ "a43cecd3b42dc0d4cc15a7d2acd4c650cee64909" ], "author": { "name": "George Rimar", "email": "grimar@accesssoftek.com", "time": "Tue Feb 12 09:08:59 2019 +0000" }, "committer": { "name": "George Rimar", "email": "grimar@accesssoftek.com", "time": "Tue Feb 12 09:08:59 2019 +0000" }, "message": "[yaml2obj/obj2yaml] - Move `Info` field out from `Section` class.\n\nELFYAML.h contains a `Section` class which is a base for a few other\nsections classes that are used for mapping different section types.\n`Section` has a `StringRef Info` field used for storing sh_info.\n\nAt the same time, sh_info has very different meanings for sections and\ncannot be processed in a similar way generally,\nfor example ELFDumper does not handle it in `dumpCommonSection`\nbut do that in `dumpGroup` and `dumpCommonRelocationSection` respectively.\n\nAt this moment, we have and handle it as a string, because that was possible for\nthe current use case. But also it can simply be a number:\nFor SHT_GNU_verdef is \"The number of version definitions within the section.\"\n\nThe patch moves `Info` field out to be able to have it as a number. \nWith that change, each class will be able to decide what type and purpose\nof the sh_info field it wants to use.\n\nI also had to edit 2 test cases. This is because patch fixes a bug. Previously we \naccepted yaml files with Info fields for all sections (for example, for SHT_DYNSYM too).\nBut we do not handle it and the resulting objects had zero sh_info fields set for\nsuch sections. Now it is accepted only for sections that supports it.\n\nDifferential revision: https://reviews.llvm.org/D58054\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353810 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "a43cecd3b42dc0d4cc15a7d2acd4c650cee64909", "tree": "e0e70d137c894cfad78760fa45dec25ec8fe44c6", "parents": [ "2841bd0ff50e8285a47cf33e61b11934e3000d18" ], "author": { "name": "Hans Wennborg", "email": "hans@hanshq.net", "time": "Tue Feb 12 09:08:52 2019 +0000" }, "committer": { "name": "Hans Wennborg", "email": "hans@hanshq.net", "time": "Tue Feb 12 09:08:52 2019 +0000" }, "message": "LibFuzzer.rst: double backticks\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353809 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "2841bd0ff50e8285a47cf33e61b11934e3000d18", "tree": "46211a481dc6aeb42a493470e9166173472f5873", "parents": [ "ce2d739dc13f65a54460e326c97c3c86717647b5" ], "author": { "name": "Max Kazantsev", "email": "max.kazantsev@azul.com", "time": "Tue Feb 12 08:10:29 2019 +0000" }, "committer": { "name": "Max Kazantsev", "email": "max.kazantsev@azul.com", "time": "Tue Feb 12 08:10:29 2019 +0000" }, "message": "Delete blocks from DTU to avoid dangling pointers\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353804 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "ce2d739dc13f65a54460e326c97c3c86717647b5", "tree": "ee2296229a42558e0ffa84917aaae0e369aa0f1b", "parents": [ "056e9a4188c8d635832fa0538a7a7a3daee6546a" ], "author": { "name": "Max Kazantsev", "email": "max.kazantsev@azul.com", "time": "Tue Feb 12 07:48:07 2019 +0000" }, "committer": { "name": "Max Kazantsev", "email": "max.kazantsev@azul.com", "time": "Tue Feb 12 07:48:07 2019 +0000" }, "message": "[LoopSimplifyCFG] Pay respect to LCSSA when removing dead blocks\n\nUtility function that we use for blocks deletion always unconditionally removes\none-input Phis. In LoopSimplifyCFG, it can lead to breach of LCSSA form.\nThis patch alters this function to keep them if needed.\n\nDifferential Revision: https://reviews.llvm.org/D57231\nReviewed By: fedor.sergeev\n\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353803 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "056e9a4188c8d635832fa0538a7a7a3daee6546a", "tree": "94b29fbf6b78316be2a8cb56d17314d2043d5cac", "parents": [ "0307be3cc16ffb07a0e195673bd0c4dada59a07c" ], "author": { "name": "Max Kazantsev", "email": "max.kazantsev@azul.com", "time": "Tue Feb 12 07:09:29 2019 +0000" }, "committer": { "name": "Max Kazantsev", "email": "max.kazantsev@azul.com", "time": "Tue Feb 12 07:09:29 2019 +0000" }, "message": "[NFC] Rename DontDeleteUselessPHIs --\u003e KeepOneInputPHIs\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353801 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "0307be3cc16ffb07a0e195673bd0c4dada59a07c", "tree": "fcf1cb66c4683355d05918ac82c9c6cb8d89a9b8", "parents": [ "d268c90aa57629d11294e684f4e003499085a396" ], "author": { "name": "Philip Reames", "email": "listmail@philipreames.com", "time": "Tue Feb 12 06:25:58 2019 +0000" }, "committer": { "name": "Philip Reames", "email": "listmail@philipreames.com", "time": "Tue Feb 12 06:25:58 2019 +0000" }, "message": "[Statepoint Lowering] Update misleading comments about chains\n\n\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353800 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "d268c90aa57629d11294e684f4e003499085a396", "tree": "11f40b0697cbab604ff41cdd39ef8a3bc82a0f4a", "parents": [ "ea98e2107dc30ad8b30a594879eafe40b5aa5a4f" ], "author": { "name": "Max Kazantsev", "email": "max.kazantsev@azul.com", "time": "Tue Feb 12 06:14:27 2019 +0000" }, "committer": { "name": "Max Kazantsev", "email": "max.kazantsev@azul.com", "time": "Tue Feb 12 06:14:27 2019 +0000" }, "message": "[NFC] Add parameter for keeping one-input Phis in DeleteDeadBlock(s)\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353799 91177308-0d34-0410-b5e6-96231b3b80d8\n" }, { "commit": "ea98e2107dc30ad8b30a594879eafe40b5aa5a4f", "tree": "ccb8a7ac7813e303ede55456951cb4474c5b6464", "parents": [ "6fe96f2c964097a204f58510392a1e7c84cc5fed" ], "author": { "name": "Craig Topper", "email": "craig.topper@intel.com", "time": "Tue Feb 12 06:14:18 2019 +0000" }, "committer": { "name": "Craig Topper", "email": "craig.topper@intel.com", "time": "Tue Feb 12 06:14:18 2019 +0000" }, "message": "[X86] Collapse FP_TO_INT16_IN_MEM/FP_TO_INT32_IN_MEM/FP_TO_INT64_IN_MEM into a single opcode using memory VT to distinquish. NFC\n\ngit-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353798 91177308-0d34-0410-b5e6-96231b3b80d8\n" } ], "next": "6fe96f2c964097a204f58510392a1e7c84cc5fed" }