summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* [Inliner] Restrict soft-float inlining penalty.Eli Friedman2017-12-225-35/+143
| | | | | | | | | | | | | | The penalty is currently getting applied in a bunch of places where it doesn't make sense, like bitcasts (which are free) and calls (which were getting the call penalty applied twice). Instead, just apply the penalty to binary operators and floating-point casts. While I'm here, also fix getFPOpCost() to do the right thing in more cases, so we don't have to dig into function attributes. Differential Revision: https://reviews.llvm.org/D41522 llvm-svn: 321332
* Add hasProfileData() to check if a function has profile data. NFC.Easwaran Raman2017-12-2210-16/+21
| | | | | | | | | | | | | | | | | | | Summary: This replaces calls to getEntryCount().hasValue() with hasProfileData that does the same thing. This refactoring is useful to do before adding synthetic function entry counts but also a useful cleanup IMO even otherwise. I have used hasProfileData instead of hasRealProfileData as David had earlier suggested since I think profile implies "real" and I use the phrase "synthetic entry count" and not "synthetic profile count" but I am fine calling it hasRealProfileData if you prefer. Reviewers: davidxl, silvas Subscribers: llvm-commits Differential Revision: https://reviews.llvm.org/D41461 llvm-svn: 321331
* [DWARF] Fix formatting bug with r321295. This fixes a MIPS buildbot failure.Wolfgang Pieb2017-12-221-1/+1
| | | | llvm-svn: 321330
* Fix isl out-of-quota errors affecting later quota guards.Michael Kruse2017-12-222-0/+20
| | | | | | | | | | | | | | | | | | | If an out-of-quota error occurred, the last error would be isl_error_quota unless a different error occured. We typically check whether the max-operations occured by comparing to that error value after leaving the quota guard. This would check whether there ever was a quota-error, not just in the last quota guards. The observable bug occurred if the max-operations limit was reached in DeLICM, and if -polly-dependences-computout=0, DependenceInfo would think that the quota for computing dependencies was the reason, i.e., fail the operation even if the calculation itself was successful. Fix by reseting the last error to isl_error_none when entering a quota guard, signaling that no quota error occured unless in the guard's scope. llvm-svn: 321329
* Change the default Aarch64 ISA to be v8.2 to correctly decode newerJason Molenda2017-12-221-0/+5
| | | | | | | instructions (e.g. on the new iphones). <rdar://problem/30585124> llvm-svn: 321328
* Enable more abilities in SymbolFilePDBAaron Smith2017-12-222-4/+46
| | | | | | | | | | | | | | | | Summary: 1) Finding symbols through --symfile 2) More abilities: Functions, Blocks, GlobalVariables, LocalVariables, VariableTypes Reviewers: zturner, lldb-commits Reviewed By: zturner Subscribers: clayborg Differential Revision: https://reviews.llvm.org/D41092 llvm-svn: 321327
* [CUDA] More fixes for __shfl_* intrinsics.Artem Belevich2017-12-211-28/+49
| | | | | | | | | * __shfl_{up,down}* uses unsigned int for the third parameter. * added [unsigned] long overloads for non-sync shuffles. Differential Revision: https://reviews.llvm.org/D41521 llvm-svn: 321326
* [X86] Allow _mm_prefetch (both the header implementation and the builtin) to ↵Craig Topper2017-12-213-7/+11
| | | | | | | | accept bit 2 which is supposed to indicate the prefetched addresses will be written to Add the appropriate _MM_HINT_ET0/ET1 defines to match gcc. llvm-svn: 321325
* Correct hasFeature/isValidFeatureName's handling of shstk/adx/mwaitxErich Keane2017-12-211-2/+7
| | | | | | | | | | | | https://bugs.llvm.org/show_bug.cgi?id=35721 reports that x86intrin.h is issuing a few warnings. This is because attribute target is using isValidFeatureName for its source. It was also discovered that two of these were missing from hasFeature. Additionally, shstk is and ibu are reordered alphabetically, as came up during code review. llvm-svn: 321324
* Update test after r321312Paul Robinson2017-12-211-3/+2
| | | | llvm-svn: 321323
* Bring clang options in error messages up to date.Adrian Prantl2017-12-211-4/+4
| | | | llvm-svn: 321322
* Delete dead code.Adrian Prantl2017-12-211-1/+0
| | | | llvm-svn: 321321
* [analyzer] Add Javascript to analyzer HTML output to allow keyboard navigation.George Karpenkov2017-12-212-2/+94
| | | | | | Differential Revision: https://reviews.llvm.org/D41414 llvm-svn: 321320
* [ODRHash] Canonicalize Decl's before processing.Richard Trieu2017-12-211-0/+1
| | | | | | | Canonicalizing the Decl before processing it as part of the hash should reduce issues with non-canonical types showing up as mismatches. llvm-svn: 321319
* Suppress "redundant parens" warning for "A (::B())".Richard Smith2017-12-213-1/+39
| | | | | | | | This is a slightly odd construct (it's more common to see "A (::B)()") but can happen in friend declarations, and the parens are not redundant as they prevent the :: binding to the left. llvm-svn: 321318
* Define isUndefWeak inline.Rafael Espindola2017-12-212-7/+4
| | | | | | | This small function was showing up in the profile. Defining it inline gives about 0.3% speedup. llvm-svn: 321317
* [CMake][Fuchsia] Enable assertionsPetr Hosek2017-12-212-0/+2
| | | | | | | | | | | | | | | | Enable assertions in both stages. Release+Asserts is fast enough. No need to let insanity through. Patch By: mcgrathr Reviewers: phosek Reviewed By: phosek Differential Revision: https://reviews.llvm.org/D41471 llvm-svn: 321316
* [X86] Use SIGN_EXTEND rather than ZERO_EXTEND for lowering ↵Craig Topper2017-12-212-6/+6
| | | | | | | | extract_vector_elt from vXi1 with a non-const index. We have a better range of instructions we can use if we can fill with the value i1 value rather than zeroing. llvm-svn: 321315
* [scudo] Add -fsanitize=scudo option to FuchsiaPetr Hosek2017-12-212-0/+18
| | | | | | | | | | | | | | Apparently the -fsanitize flag hadn't been added for Scudo upstream yet. Patch By: flowerhack Reviewers: cryptoad, alekseyshl, mcgrathr, phosek Reviewed By: mcgrathr, phosek Differential Revision: https://reviews.llvm.org/D41413 llvm-svn: 321314
* Include process.h for getpid on Windows in instr profilingReid Kleckner2017-12-211-0/+1
| | | | llvm-svn: 321313
* [AST] Incorrectly qualified unscoped enumeration as template actual parameter.Paul Robinson2017-12-216-7/+36
| | | | | | | | | | | | | | | An unscoped enumeration used as template argument, should not have any qualified information about its enclosing scope, as its visibility is global. In the case of scoped enumerations, they must include information about their enclosing scope. Patch by Carlos Alberto Enciso! Differential Revision: https://reviews.llvm.org/D39239 llvm-svn: 321312
* Call isStaticLinkTimeConstant only once per relocation.Rafael Espindola2017-12-211-7/+10
| | | | | | | | | It is a pretty expensive function. Some of the speedups: clang: 1.92% chrome: 1.15% linux-kernel: 1.40% llvm-svn: 321311
* Re-commit r321223, which adds a printing policy to the ASTDumper.Aaron Ballman2017-12-2118-1998/+2016
| | | | | | | | This allows you to dump C++ code that spells bool instead of _Bool, leaves off the elaborated type specifiers when printing struct or class names, and other C-isms. Fixes the -Wreorder issue and fixes the ast-dump-color.cpp test. llvm-svn: 321310
* [ModRefInfo] Add must alias info to ModRefInfo.Alina Sbirlea2017-12-2111-64/+295
| | | | | | | | | | | | | | | | | | | | | | Summary: Add an additional bit to ModRefInfo, ModRefInfo::Must, to be cleared for known must aliases. Shift existing Mod/Ref/ModRef values to include an additional most significant bit. Update wrappers that modify ModRefInfo values to reflect the change. Notes: * ModRefInfo::Must is almost entirely cleared in the AAResults methods, the remaining changes are trying to preserve it. * Only some small changes to make custom AA passes set ModRefInfo::Must (BasicAA). * GlobalsModRef already declares a bit, who's meaning overlaps with the most significant bit in ModRefInfo (MayReadAnyGlobal). No changes to shift the value of MayReadAnyGlobal (see AlignedMap). FunctionInfo.getModRef() ajusts most significant bit so correctness is preserved, but the Must info is lost. * There are cases where the ModRefInfo::Must is not set, e.g. 2 calls that only read will return ModRefInfo::NoModRef, though they may read from exactly the same location. Reviewers: dberlin, hfinkel, george.burgess.iv Subscribers: llvm-commits, sanjoy Differential Revision: https://reviews.llvm.org/D38862 llvm-svn: 321309
* [WebAssembly] Add extra test for weak global symbolsSam Clegg2017-12-213-9/+20
| | | | | | | | | | | | | | | | | Summary: Currently the test only checks behaviour for weak function symbols. Should be good to merge straight away? Reviewers: sbc100 Reviewed By: sbc100 Subscribers: jfb, dschuff, jgravelle-google, aheejin, sunfish, llvm-commits Differential Revision: https://reviews.llvm.org/D41449 llvm-svn: 321308
* Revert "[CodeGen] Fix crash when a function taking transparent union is ↵Volodymyr Sapsai2017-12-217-91/+11
| | | | | | | | | | | | redeclared." This reverts commit r321296. It caused performance regressions FAIL: imp.execution_time FAIL: 2007-01-04-KNR-Args.execution_time FAIL: sse_expandfft.execution_time FAIL: sse_stepfft.execution_time llvm-svn: 321306
* [sanitizer] Make function declarations C-compatiblePetr Hosek2017-12-219-37/+37
| | | | | | | | | | | | | | | | | The public sanitizer headers are intended to be usable from either C++ or C, but they declare no-argument functions with the syntax that is not a proper prototype declaration in C. This goes unnoticed until someone uses -Wsystem-headers. Patch By: mcgrathr Reviewers: phosek, vitalybuka Reviewed By: vitalybuka Differential Revision: https://reviews.llvm.org/D41475 llvm-svn: 321305
* Don't produce redundant parentheses warning for "A (::B);" and the like.Richard Smith2017-12-212-5/+42
| | | | | | | The parentheses here are not redundant as they affect the binding of the '::' token. llvm-svn: 321304
* [X86] When lowering truncates to vXi1, don't sign extend i16/i8 types to ↵Craig Topper2017-12-214-22/+23
| | | | | | | | 512-bit if we have VLX. This should only affect what we do for v8i16. Previously we went to v8i64, but if we have VLX we only need v8i32. This prevents an unnecessary zmm usage. llvm-svn: 321303
* [clangd] Don't re-hash SymbolID in maps, just use the SHA1 dataSam McCall2017-12-212-2/+4
| | | | llvm-svn: 321302
* [AArch64] Enable fp16 data type for the Builtin for AArch64 only.Abderrazek Zaafrani2017-12-213-141/+160
| | | | | | Differential Revision: https:://reviews.llvm.org/D41360 llvm-svn: 321301
* [CMake] Allow passing extra CMake arguments to custom libc++Petr Hosek2017-12-211-1/+2
| | | | | | | | This can be used to customize the libc++ build. Differential Revision: https://reviews.llvm.org/D41103 llvm-svn: 321299
* Revert "Fix for PR32990"Reid Kleckner2017-12-216-138/+27
| | | | | | | | | | | | This reverts commit r321239. It broke the Chromium DLL build: [8834/50217] LINK(DLL) icui18n.dll icui18n.dll.lib icui18n.dll.pdb FAILED: icui18n.dll icui18n.dll.lib icui18n.dll.pdb zrule.obj : error LNK2001: unresolved external symbol "__declspec(dllimport) public: void __cdecl icu_60::UnicodeString::`vbase destructor'(void)" (__imp_??_DUnicodeString@icu_60@@QEAAXXZ) llvm-svn: 321298
* When instantiating a deduction guide, transform its name.Richard Smith2017-12-212-3/+20
| | | | | | Otherwise it will serve as a deduction guide for the wrong class template. llvm-svn: 321297
* [CodeGen] Fix crash when a function taking transparent union is redeclared.Volodymyr Sapsai2017-12-217-11/+91
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | When a function taking transparent union is declared as taking one of union members earlier in the translation unit, clang would hit an "Invalid cast" assertion during EmitFunctionProlog. This case corresponds to function f1 in test/CodeGen/transparent-union-redecl.c. We decided to cast i32 to union because after merging function declarations function parameter type becomes int, CGFunctionInfo::ArgInfo type matches with ABIArgInfo type, so we decide it is a trivial case. But these types should also be castable to parameter declaration type which is not the case here. The fix is in checking for the trivial case if ABIArgInfo type matches with parameter declaration type. It exposed inconsistency that we check hasScalarEvaluationKind for different types in EmitParmDecl and EmitFunctionProlog, and comment says they should match. Additional tests in Sema/transparent-union.c capture current behavior and make sure there are no regressions. rdar://problem/34949329 Reviewers: rjmccall, rafael Reviewed By: rjmccall Subscribers: aemerson, cfe-commits, kristof.beyls Differential Revision: https://reviews.llvm.org/D41311 llvm-svn: 321296
* [DWARF v5] Rework of string offsets table readerWolfgang Pieb2017-12-219-207/+497
| | | | | | | | | | | | | Reorganizes the DWARF consumer to derive the string offsets table contribution's format from the contribution header instead of (incorrectly) from the unit's format. Reviewers: JDevliegehere, aprantl Differential Revision: https://reviews.llvm.org/D41146 llvm-svn: 321295
* [AARch64] Add ARMv8.2-A FP16 vector intrinsicsAbderrazek Zaafrani2017-12-2110-363/+2348
| | | | | | | | Putting back the code that was reverted few weeks ago. Differential Revision: https://reviews.llvm.org/D34161 llvm-svn: 321294
* Revert "[mips][compiler-rt] Provide 64bit atomic add and sub"Simon Dardis2017-12-213-125/+73
| | | | | | | | | | | This reverts commit r321260. It appears to have broken the sanitizer bot sanitizer-ppc64be-linux. http://lab.llvm.org:8011/builders/sanitizer-ppc64be-linux/builds/5029 Reverting to see if the buildbot turns green. llvm-svn: 321292
* [X86] Promote v8i1 shuffles to v8i32 instead of v8i64 if we have VLX.Craig Topper2017-12-214-145/+163
| | | | | | | | We should have equally good shuffle options for v8i32 with VLX. This was spotted during my attempts to remove 512-bit vectors from SKX. We still use 512-bits for v16i1, v32i1, and v64i1. I'm less sure we can handle those well with narrower vectors. i32 and i64 element sizes get the best shuffle support. llvm-svn: 321291
* [analyzer] Fix zero-initialization of stack VLAs under ObjC ARC.Artem Dergachev2017-12-212-3/+50
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Using ARC, strong, weak, and autoreleasing stack variables are implicitly initialized with nil. This includes variable-length arrays of Objective-C object pointers. However, in the analyzer we don't zero-initialize them. We used to, but it accidentally regressed after r289618. Under ARC, the array variable's initializer within DeclStmt is an ImplicitValueInitExpr. Environment doesn't maintain any bindings for this expression kind - instead it always knows that it's a known constant (0 in our case), so it just returns the known value by calling SValBuilder::makeZeroVal() (see EnvironmentManager::getSVal(). Commit r289618 had introduced reasonable behavior of SValBuilder::makeZeroVal() for the arrays, which produces a zero-length compoundVal{}. When such value is bound to arrays, in RegionStoreManager::bindArray() "remaining" items in the array are default-initialized with zero, as in RegionStoreManager::setImplicitDefaultValue(). The similar mechanism works when an array is initialized by an initializer list that is too short, eg. int a[3] = { 1, 2 }; would result in a[2] initialized with 0. However, in case of variable-length arrays it didn't know if any more items need to be added, because, well, the length is variable. Add the default binding anyway, regardless of how many actually need to be added. We don't really care how many, because the default binding covers the whole array anyway. Differential Revision: https://reviews.llvm.org/D41478 rdar://problem/35477763 llvm-svn: 321290
* [X86][SSE] Split large PAVGB/PAVGW vectors to legal widthsSimon Pilgrim2017-12-213-2323/+318
| | | | | | | | Patch to allow detectAVGPattern handle vectors larger than the legal size (128 SSE2, 256 AVX2, 512 AVX512BW), splitting the vectors accordingly. Differential Revision: https://reviews.llvm.org/D41440 llvm-svn: 321288
* [clangd] Fix use after free.Benjamin Kramer2017-12-211-3/+4
| | | | | | Found by asan. llvm-svn: 321286
* [ASTImporterTest] Add mandatory testing with '-fdelayed-template-parsing'Aleksei Sidorin2017-12-211-384/+386
| | | | | | | | | | | | | | | | * While running ASTImporterTests, we often forget about Windows MSVC buildbots which enable '-fdelayed-template-parsing' by default. This leads to AST import errors because templates are not parsed and corresponding parts of AST are not built so importer cannot import them. It takes both reviewing time to find such issues and post-commit time to fix unexpected buildbot failures. To solve this issue, we enable testing with '-fdelayed-template-parsing' option enabled by default in addition to testing with default compiler options. This allows us to catch the problem during development. Differential Revision: https://reviews.llvm.org/D41444 llvm-svn: 321285
* [YAML] Refactor escaping unittestsFrancis Visoiu Mistrih2017-12-212-112/+25
| | | | llvm-svn: 321284
* [YAML] Fix UTF-8 handlingFrancis Visoiu Mistrih2017-12-212-1/+34
| | | | | | | | Previous YAML quoting patches broke UTF-8 printing in YAML: see https://reviews.llvm.org/D41290#961801. Differential Revision: https://reviews.llvm.org/D41490 llvm-svn: 321283
* [TableGen] Print more helpful information in case of type contradictionKrzysztof Parzyszek2017-12-212-21/+39
| | | | | | Dump the failing TreePattern. llvm-svn: 321282
* [DAGCombiner] Remove (xor (xor x, c1), c2) -> (xor x, (xor c1, c2)) fold. NFCI.Simon Pilgrim2017-12-211-15/+0
| | | | | | More general cases are already handled by constant canonicalization and then the ReassociateOps call at line 5327 llvm-svn: 321280
* [DAGCombiner] Generalize (or (and X, c1), c2) -> (and (or X, c2), c1|c2) ↵Simon Pilgrim2017-12-213-18/+20
| | | | | | | | combine to work on non-splat vectors The knownbits_mask_or_shuffle_uitofp change is interesting - shuffle combines manage to kick in, removing the AND constant mask load. For targets with fast-variable-shuffle this should reduce further to VPOR+VPSHUFB+VCVTDQ2PS. llvm-svn: 321279
* [X86] Add (or (and X, c1), c2) -> (and (or X, c2), c1|c2) non-splat vector testSimon Pilgrim2017-12-211-4/+14
| | | | llvm-svn: 321278
* Make one more test redhat-compatiblePavel Labath2017-12-211-11/+11
| | | | | | This test was also using "a" in an expression. llvm-svn: 321277
OpenPOWER on IntegriCloud