summaryrefslogtreecommitdiffstats
path: root/clang/lib
Commit message (Collapse)AuthorAgeFilesLines
* Revert "[clang][IFS] Driver pipeline: generate interface stubs after ↵Puyan Lotfi2019-11-193-99/+30
| | | | | | | | | standard pipeline." This reverts commit 58ea00b51fe9b011301484957556872fced7dd08. Test for .o + .ifs sidecar files is brittle and failing on bots. Reverting to unblock.
* [clang][IFS] Driver pipeline: generate interface stubs after standard pipeline.Puyan Lotfi2019-11-193-30/+99
| | | | | | | | | | | | | | | | | Up until now, clang interface stubs has replaced the standard PP -> C -> BE -> ASM -> LNK pipeline. With this change, it will happen in conjunction with it. So what when you build your code you will get an a.out or lib.so as well as an interface stub file. Example: clang -shared -o libfoo.so -emit-interface-stubs ... will generate both a libfoo.so and a libfoo.ifso. The .so file will contain the code from the standard compilation pipeline and the .ifso file will contain the ELF stub library. Differential Revision: https://reviews.llvm.org/D70274
* [CodeGen] Assign locations to calls to special struct helpersVedant Kumar2019-11-181-0/+1
| | | | | | | | | | | | | | Assign artificial locations to calls to special struct-related helper functions. Such calls may not inherit a location if emitted within FinishFunction, at which point the lexical scope stack may be empty, causing CGDebugInfo to report the current DebugLoc as empty. Fixes an IR verifier complaint about a call to '__destructor_8_s0' not having a !dbg location attached. rdar://57293361
* [NFC] Fix 'target' condition in checkTargetFeaturesErich Keane2019-11-181-2/+2
| | | | | | | | | | | | | | | | checkTargetFeatures was incorrectly checking for cpu_specific instead of just 'target'. While this function was never called in that situation, it seemed correct to fix the condition. Additionally, multiversion functions can never be always_inline, but if any function accidentially ended up here we shouldn't diagnose. Note that the adding of target-features to the list is unnecessary since the getFunctionFeatureMap actually considers attribute target, however adding it results in significantly better error messages by putting the 'target' features first (and thus first to fail). Otherwise, the error message would be the first feature 'implied' by the target attribute, and not necessarily the feature listed in the attribute itself.
* Populate CUDA flags on FreeBSD too, as many other toolchains do.Dimitry Andric2019-11-182-0/+7
| | | | | | | | | | | | | | | | Summary: This allows `clang` to be used to compile CUDA programs. Compiled simple helloworld.cu with this. Reviewers: dim, emaste, tra, yaxunl, ABataev Reviewed By: tra Subscribers: dim, emaste, cfe-commits Tags: #clang Differential Revision: https://reviews.llvm.org/D69990
* [Sema] Fix a -Wobjc-signed-char-bool false-positiveErik Pilkington2019-11-181-0/+6
| | | | | | | Unsigned bit-field flags can only have boolean values, so handle that case in Expr::isKnownToHaveBooleanValue. rdar://56256999
* Implement __attribute__((objc_direct)), __attribute__((objc_direct_members))Pierre Habouzit2019-11-1813-45/+408
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | __attribute__((objc_direct)) is an attribute on methods declaration, and __attribute__((objc_direct_members)) on implementation, categories or extensions. A `direct` property specifier is added (@property(direct) type name) These attributes / specifiers cause the method to have no associated Objective-C metadata (for the property or the method itself), and the calling convention to be a direct C function call. The symbol for the method has enforced hidden visibility and such direct calls are hence unreachable cross image. An explicit C function must be made if so desired to wrap them. The implicit `self` and `_cmd` arguments are preserved, however to maintain compatibility with the usual `objc_msgSend` semantics, 3 fundamental precautions are taken: 1) for instance methods, `self` is nil-checked. On arm64 backends this typically adds a single instruction (cbz x0, <closest-ret>) to the codegen, for the vast majority of the cases when the return type is a scalar. 2) for class methods, because the class may not be realized/initialized yet, a call to `[self self]` is emitted. When the proper deployment target is used, this is optimized to `objc_opt_self(self)`. However, long term we might want to emit something better that the optimizer can reason about. When inlining kicks in, these calls aren't optimized away as the optimizer has no idea that a single call is really necessary. 3) the calling convention for the `_cmd` argument is changed: the caller leaves the second argument to the call undefined, and the selector is loaded inside the body when it's referenced only. As far as error reporting goes, the compiler refuses: - making any overloads direct, - making an overload of a direct method, - implementations marked as direct when the declaration in the interface isn't (the other way around is allowed, as the direct attribute is inherited from the declaration), - marking methods required for protocol conformance as direct, - messaging an unqualified `id` with a direct method, - forming any @selector() expression with only direct selectors. As warnings: - any inconsistency of direct-related calling convention when @selector() or messaging is used, - forming any @selector() expression with a possibly direct selector. Lastly an `objc_direct_members` attribute is added that can decorate `@implementation` blocks and causes methods only declared there (and in no `@interface`) to be automatically direct. When decorating an `@interface` then all methods and properties declared in this block are marked direct. Radar-ID: rdar://problem/2684889 Differential Revision: https://reviews.llvm.org/D69991 Reviewed-By: John McCall
* Reland: [Remarks][Driver] Use different remark files when targeting multiple ↵Francis Visoiu Mistrih2019-11-181-0/+19
| | | | | | | | | | | | | | | | | | architectures When the driver is targeting multiple architectures at once, for things like Universal Mach-Os, we need to emit different remark files for each cc1 invocation to avoid overwriting the files from a different invocation. For example: $ clang -c -o foo.o -fsave-optimization-record -arch x86_64 -arch x86_64h will create two remark files: * foo-x86_64.opt.yaml * foo-x86_64h.opt.yaml
* Revert "[Remarks][Driver] Use different remark files when targeting multiple ↵Reid Kleckner2019-11-181-18/+0
| | | | | | | | architectures" This reverts commit b4e2b112b58154a89171df39dae80044865ff4ff. Test doesn't appear to pass on Windows, maybe all non-Mac.
* Temporarily Revert "Add support for options -frounding-math, ftrapping-math, ↵Eric Christopher2019-11-184-275/+9
| | | | | | | | -ffp-model=, and -ffp-exception-behavior=" and a follow-up NFC rearrangement as it's causing a crash on valid. Testcase is on the original review thread. This reverts commits af57dbf12e54f3a8ff48534bf1078f4de104c1cd and e6584b2b7b2de06f1e59aac41971760cac1e1b79
* Allocate builtins table earlier to fix bug found by ubsanReid Kleckner2019-11-181-1/+2
| | | | Follow up to 979da9a4c3ba
* [Remarks][Driver] Use different remark files when targeting multiple ↵Francis Visoiu Mistrih2019-11-181-0/+18
| | | | | | | | | | | | | | | | | | architectures When the driver is targeting multiple architectures at once, for things like Universal Mach-Os, we need to emit different remark files for each cc1 invocation to avoid overwriting the files from a different invocation. For example: $ clang -c -o foo.o -fsave-optimization-record -arch x86_64 -arch x86_64h will create two remark files: * foo-x86_64.opt.yaml * foo-x86_64h.opt.yaml
* [Attr] Fix `-ast-print` for `asm` attributeJoel E. Denny2019-11-181-5/+6
| | | | | | | | | | | | | | | | Without this fix, the tests introduced here produce the following assert fail: ``` clang: /home/jdenny/llvm/clang/include/clang/Basic/AttributeCommonInfo.h:163: unsigned int clang::AttributeCommonInfo::getAttributeSpellingListIndex() const: Assertion `(isAttributeSpellingListCalculated() || AttrName) && "Spelling cannot be found"' failed. ``` The bug was introduced by D67368, which caused `AsmLabelAttr`'s spelling index to be set to `SpellingNotCalculated`. Reviewed By: aaron.ballman Differential Revision: https://reviews.llvm.org/D70349
* [OPENMP50]Fix PR44024: runtime assert in distribute construct.Alexey Bataev2019-11-181-1/+3
| | | | | If the code is emitted for distribute construct, the nonmonotonic modifier should not be added.
* Revert "[DWARF5]Addition of alignment atrribute in typedef DIE."Sam McCall2019-11-181-8/+5
| | | | | This reverts commit 423f541c1a322963cf482683fe9777ef0692082d, which breaks llvm-c ABI.
* Fix "not all control paths return a value" warning. NFCI.Simon Pilgrim2019-11-181-2/+3
|
* [RISCV] Set triple based on -march flagSimon Cook2019-11-181-0/+11
| | | | | | | | | For RISC-V the value provided to -march should determine whether to compile for 32- or 64-bit RISC-V irrespective of the target provided to the Clang driver. This adds a test for this flag for RISC-V and sets the Target architecture correctly in these cases. Differential Revision: https://reviews.llvm.org/D54214
* [ARM,MVE] Add intrinsics for vector comparisons.Simon Tatham2019-11-181-0/+8
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This adds the `vcmp` family of ACLE MVE intrinsics: vector/vector, vector/scalar, and the predicated forms of both. All are represented using standard existing IR: vector/scalar comparisons are represented by making a vector out of the scalar first, and predicated forms are represented by taking the bitwise AND of the input predicate and the output of the comparison. Existing LLVM-side tests demonstrate that ISel will pattern-match all of that back down to single MVE VCMPs. The idiom of handling a vector/scalar operation by generating IR to expand the scalar into a second vector is going to be needed for a lot of MVE intrinsics, so to make that easy, I've provided a helper function that automatically works out the element count. The comparison intrinsics are the first ones that have to //return// a predicate, in the user-facing `mve_pred16_t` format. This means we have to use the `arm_mve_pred_v2i` low-level intrinsic to convert it back from the logical `<n x i1>` form used in IR. I've done that explicitly in the code gen specification for the builtins, because it happens much more rarely in the ACLE API than passing a Predicate as input, so it didn't seem worth automating in MveEmitter. Reviewers: ostannard, MarkMurrayARM, dmgreen Reviewed By: dmgreen Subscribers: kristof.beyls, hiraditya, cfe-commits, llvm-commits Tags: #clang, #llvm Differential Revision: https://reviews.llvm.org/D70297
* [AST] Attach comment in `/** doc */ typedef struct A {} B` to B as well as A.Sam McCall2019-11-181-1/+3
| | | | | | | | | | | | | | | | | | | | | | Summary: Semantically they're the same thing, and it's important when the underlying struct is anonymous. There doesn't seem to be a problem attaching the same comment to multiple things as it already happens with `/** doc */ int a, b;` This affects an Index test but the results look better (name present, USR points to the typedef). Fixes https://github.com/clangd/clangd/issues/189 Reviewers: kadircet, lh123 Subscribers: ilya-biryukov, jkorous, arphaman, usaxena95, cfe-commits Tags: #clang Differential Revision: https://reviews.llvm.org/D70203
* [ARM] Allocatable Global Register Variables for ARMAnna Welker2019-11-184-5/+70
| | | | | | | | | | | | Provides support for using r6-r11 as globally scoped register variables. This requires a -ffixed-rN flag in order to reserve rN against general allocation. If for a given GRV declaration the corresponding flag is not found, or the the register in question is the target's FP, we fail with a diagnostic. Differential Revision: https://reviews.llvm.org/D68862
* Revert "[NFC] Refactor representation of materialized temporaries"Nico Weber2019-11-1732-184/+99
| | | | | | This reverts commit 08ea1ee2db5f9d6460fef1d79d0d1d1a5eb78982. It broke ./ClangdTests/FindExplicitReferencesTest.All on the bots, see comments on https://reviews.llvm.org/D69360
* [NFC] Refactor representation of materialized temporariesTyker2019-11-1632-99/+184
| | | | | | | | | | | | | | | Summary: this patch refactor representation of materialized temporaries to prevent an issue raised by rsmith in https://reviews.llvm.org/D63640#inline-612718 Reviewers: rsmith, martong, shafik Reviewed By: rsmith Subscribers: rnkovacs, arphaman, cfe-commits Tags: #clang Differential Revision: https://reviews.llvm.org/D69360
* [DWARF5]Addition of alignment atrribute in typedef DIE.Sourabh Singh Tomar2019-11-161-5/+8
| | | | | | | | | | | | This patch, adds support for DW_AT_alignment[DWARF5] attribute, to be emitted with typdef DIE. When explicit alignment is specified. Patch by Awanish Pandey <Awanish.Pandey@amd.com> Reviewers: aprantl, dblaikie, jini.susan.george, SouraVX, alok, deadalinx Differential Revision: https://reviews.llvm.org/D70111
* [clang-format] fix regression in middle pointer alignmentmydeveloperday2019-11-161-1/+1
| | | | | | | | | | | | | | | | | | Summary: a75f8d98d7ac introduced a regression with Middle pointer alignment, which this patch fixes. Reviewers: MyDeveloperDay, klimek, sammccall Reviewed By: MyDeveloperDay, sammccall Subscribers: cfe-commits, merge_guards_bot Patch by: Typz Tags: #clang Differential Revision: https://reviews.llvm.org/D70305
* Remove +x permission on some filesSylvestre Ledru2019-11-161-0/+0
|
* Remove +x permission on clang/lib/Format/Format.cppSylvestre Ledru2019-11-161-0/+0
|
* [clang-format] Add SpaceBeforeBracketsmydeveloperday2019-11-162-5/+11
| | | | | | | | | | | | | | | | Summary: Adds a new option SpaceBeforeBrackets to add spaces before brackets (i.e. int a[23]; -> int a [23];) This is present as an option in the Visual Studio C++ code formatting settings, but there was no matching setting in clang-format. Reviewers: djasper, MyDeveloperDay, mitchell-stellar Reviewed By: MyDeveloperDay Subscribers: llvm-commits, cfe-commits, klimek Patch by: Anteru Tags: #clang, #clang-format, #llvm Differential Revision: https://reviews.llvm.org/D6920
* Add support of the next Ubuntu (Ubuntu 20.04 - Focal Fossa)Sylvestre Ledru2019-11-161-0/+1
|
* Avoid including Builtins.h in Preprocessor.hReid Kleckner2019-11-158-4/+13
| | | | | | Builtins are rarely if ever accessed via the Preprocessor. They are typically found on the ASTContext, so there should be no performance penalty to using a pointer indirection to store the builtin context.
* Don't add optnone or noinline if the function is already marked asAkira Hatanaka2019-11-151-5/+5
| | | | | | | | | | | | always_inline. The assertion in SetLLVMFunctionAttributesForDefinition used to fail when there was attribute OptimizeNone on the AST function and attribute always_inline on the IR function. This happens because base destructors are annotated with always_inline when the code is compiled with -fapple-kext (see r124757). rdar://problem/57169694
* Revert "re-land [DebugInfo] Add debug location to stubs generated by ↵Alexandre Ganea2019-11-152-14/+5
| | | | | | CGDeclCXX and mark them as artificial" This reverts commit 9c1baa23526c6d7d06eafefbf82d73bfe9bb3aaf.
* re-land [DebugInfo] Add debug location to stubs generated by CGDeclCXX and ↵Alexandre Ganea2019-11-152-5/+14
| | | | | | mark them as artificial Differential Revision: https://reviews.llvm.org/D66328
* [OpenMP 5.0] - Extend defaultmap, by Chi Chun Chen.cchen2019-11-153-67/+275
| | | | | | | | | | | | | | | | | Summary: For the extended defaultmap, most of the work is inside sema. The only difference for codegen is to set different initial maptype for different implicit-behavior. Reviewers: jdoerfert, ABataev Reviewed By: ABataev Subscribers: dreachem, sandoval, cfe-commits Tags: #clang, #openmp Differential Revision: https://reviews.llvm.org/D69204
* [AST] Let DeclarationNameInfo printing use PrintingPolicy, and fix ↵Sam McCall2019-11-153-13/+18
| | | | const-correctness
* [clang-format] Flexible line endingsCameron Desrochers2019-11-151-3/+12
| | | | | | | | | Line ending detection is now set with the `DeriveLineEnding` option. CRLF can now be used as the default line ending by setting `UseCRLF`. When line ending detection is disabled, all line endings are converted according to the `UseCRLF` option. Differential Revision: https://reviews.llvm.org/D19031
* Implement target(branch-protection) attribute for AArch64Momchil Velikov2019-11-155-59/+71
| | | | | | | | | This patch implements `__attribute__((target("branch-protection=...")))` in a manner, compatible with the analogous GCC feature: https://gcc.gnu.org/onlinedocs/gcc-9.2.0/gcc/AArch64-Function-Attributes.html#AArch64-Function-Attributes Differential Revision: https://reviews.llvm.org/D68711
* clang: enable unwind tables on FreeBSD !amd64Ed Maste2019-11-152-0/+3
| | | | | | | | | | | | | | | | | | | There doesn't seem to be much sense in defaulting "on" unwind tables on amd64 and not on other arches. It causes surprising differences between platforms, such as the PR below[1]. Prior to this change, FreeBSD inherited the default implementation of the method from the Gnu.h Generic_Elf => Generic_GCC parent class, which returned true only for amd64 targets. Override that and opt on always, similar to, e.g., NetBSD's driver. [1] https://bugs.freebsd.org/241562 Patch by cem (Conrad Meyer). Reviewed By: dim Differential Revision: https://reviews.llvm.org/D70110
* [RISCV] Match GCC `-march`/`-mabi` driver defaultsSam Elliott2019-11-153-17/+132
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Summary: Clang/LLVM is a cross-compiler, and so we don't have to make a choice about `-march`/`-mabi` at build-time, but we may have to compute a default `-march`/`-mabi` when compiling a program. Until now, each place that has needed a default `-march` has calculated one itself. This patch adds a single place where a default `-march` is calculated, in order to avoid calculating different defaults in different places. This patch adds a new function `riscv::getRISCVArch` which encapsulates this logic based on GCC's for computing a default `-march` value when none is provided. This patch also updates the logic in `riscv::getRISCVABI` to match the logic in GCC's build system for computing a default `-mabi`. This patch also updates anywhere that `-march` is used to now use the new function which can compute a default. In particular, we now explicitly pass a `-march` value down to the gnu assembler. GCC has convoluted logic in its build system to choose a default `-march`/`-mabi` based on build options, which would be good to match. This patch is based on the logic in GCC 9.2.0. This commit's logic is different to GCC's only for baremetal targets, where we default to rv32imac/ilp32 or rv64imac/lp64 depending on the target triple. Tests have been updated to match the new logic. Reviewers: asb, luismarques, rogfer01, kito-cheng, khchen Reviewed By: asb, luismarques Subscribers: sameer.abuasal, rbar, johnrusso, simoncook, apazos, sabuasal, niosHD, shiva0217, jrtc27, MaskRay, zzheng, edward-jones, MartinMosbeck, brucehoult, the_o, rkruppe, PkmX, jocewei, psnobl, benna, Jim, s.egerton, pzheng, cfe-commits Tags: #clang Differential Revision: https://reviews.llvm.org/D69383
* [CodeComplete] Constructor overload candidates report as vector(int) instead ↵Sam McCall2019-11-151-0/+4
| | | | | | | | | | | | | | | | | | | of vector<string>(int) Summary: This is shorter, shouldn't be confusing (is consistent with how they're declared), and avoids messy cases that are printed as myclass<type-param-0-0>(int) in the case of partial specialization. Fixes part of https://github.com/clangd/clangd/issues/76 Reviewers: hokein, lh123 Subscribers: ilya-biryukov, kadircet, usaxena95, cfe-commits Tags: #clang Differential Revision: https://reviews.llvm.org/D70307
* [Clang][ASTImporter] Added visibility check for ClassTemplateDecl.Balázs Kéri2019-11-151-0/+2
| | | | | | | | | | | | | | | | Summary: ASTImporter makes now difference between class templates with same name in different translation units if these are not visible outside. Reviewers: martong, a.sidorin, shafik Reviewed By: martong Subscribers: rnkovacs, dkrupp, Szelethus, gamesh411, teemperor, cfe-commits Tags: #clang Differential Revision: https://reviews.llvm.org/D67543
* Move floating point related entities to namespace levelSerge Pavlov2019-11-151-20/+13
| | | | | | | | | | | | | | Enumerations that describe rounding mode and exception behavior were defined inside ConstrainedFPIntrinsic. It makes sense to use the same definitions to represent the same properties in other cases, not only in constrained intrinsics. It was however inconvenient as required to include constrained intrinsics definitions even if they were not needed. Also using long scope prefix reduced readability. This change moves these definitioins to the namespace llvm::fp. No functional changes. Differential Revision: https://reviews.llvm.org/D69552
* Include the mangled name in -ast-dump=jsonAlex Richardson2019-11-151-1/+5
| | | | | | | | | | | | | | | | | | | | I am planning to use this feature to make update_cc_test_checks.py less fragile by obtaining the mangled names directly from -ast-dump=json. Currently, it uses c-index-test which ignores the -triple=, etc. arguments that are in the RUN: line and therefore does not generate checks for some targets. The AST dump tests were updated using the following command: `python $LLVM_BINDIR/gen_ast_dump_json_test.py --update --source $LLVM_SRC/clang/test/AST/*-json.*` Reviewers: aaron.ballman Reviewed By: aaron.ballman Subscribers: rsmith, MaskRay, cfe-commits Tags: #clang Differential Revision: https://reviews.llvm.org/D69564
* Revert "[clang] Remove the DIFlagArgumentNotModified debug info flag"Djordje Todorovic2019-11-152-0/+46
| | | | This reverts commit rG1643734741d2 due to LLDB test failure.
* [clang] Remove the DIFlagArgumentNotModified debug info flagDjordje Todorovic2019-11-152-46/+0
| | | | | | | | | | It turns out that the ExprMutationAnalyzer can be very slow when AST gets huge in some cases. The idea is to move this analysis to the LLVM back-end level (more precisely, in the LiveDebugValues pass). The new approach will remove the performance regression, simplify the implementation and give us front-end independent implementation. Differential Revision: https://reviews.llvm.org/D68206
* [AST] Use written names instead of `type-param-0-0` in more cases when ↵Sam McCall2019-11-151-38/+30
| | | | | | | | | | | | | | | | | | | printing types and decls. Summary: This doesn't cover decls in diagnostics, which use NamedDecl::getNameForDiagnostic(). (That should also be fixed later I think). This covers some cases of https://github.com/clangd/clangd/issues/76 (hover, but not outline or sighelp) Reviewers: hokein Subscribers: ilya-biryukov, jkorous, arphaman, kadircet, usaxena95, cfe-commits Tags: #clang Differential Revision: https://reviews.llvm.org/D70236
* Add missing includes needed to prune LLVMContext.h include, NFCReid Kleckner2019-11-142-0/+2
| | | | | These are a pre-requisite to removing #include "llvm/Support/Options.h" from LLVMContext.h: https://reviews.llvm.org/D70280
* [BPF] Restrict preserve_access_index attribute to C onlyYonghong Song2019-11-141-31/+12
| | | | | | | | | | | | This patch is a follow-up for commit 4e2ce228ae79 [BPF] Add preserve_access_index attribute for record definition to restrict attribute for C only. A new test case is added to check for this restriction. Additional code polishing is done based on Aaron Ballman's suggestion in https://reviews.llvm.org/D69759/new/. Differential Revision: https://reviews.llvm.org/D70257
* [X86] Fix the implementation of __readcr3/__writecr3 to work in 64-bit modeCraig Topper2019-11-141-9/+16
| | | | | | | | | | | | We need to use a 64-bit type in 64-bit mode so a 64-bit register will get used in the generated assembly. I've also changed the constraints to just use "r" intead of "q". "q" forces to a only an a/b/c/d register in 32-bit mode, but I see no reason that would matter here. Fixes Nico's note in PR19301 over 4 years ago. Differential Revision: https://reviews.llvm.org/D70101
* [Hexagon] Fix clang driver to parse -mcpu/-mvXX and -march properly.Sumanth Gundapaneni2019-11-141-1/+1
| | | | | | Before this patch if we pass "-mcpu=hexagonv65 -march=hexagon" in this order, the driver fails to figure out the correct cpu version. This patch fixed this issue.
* [clang-format] Fixed edge-case with SpacesInSquareBrackets with trailing ↵Mitchell Balan2019-11-141-3/+7
| | | | | | | | | | | | | | | | | | | bare "&" lambda capture. Summary: Lambda captures allow for a lone `&` capture, so `&]` needs to be properly handled. `int foo = [& ]() {}` is fixed to give `int foo = [ & ]() {}` Reviewers: MyDeveloperDay Reviewed by: MyDeveloperDay Subscribers: cfe-commits Tags: #clang, #clang-format Differential Revision: https://reviews.llvm.org/D70249
OpenPOWER on IntegriCloud