Fangrui Song [Wed, 24 Jun 2020 15:22:17 +0000 (08:22 -0700)]
[ELF] -r: don't parse @ (symbol versioning) for .symver inline asm in bitcode
Fixes PR46420
Similar to D43307 for non-LTO.
Module-level inline assembly can use .symver to create a symbol with `@` in the name.
For relocatable output, @ should be retained in the symbol name. `@ver` should
not be parsed and dropped.
Reviewed By: grimar, psmith
Differential Revision: https://reviews.llvm.org/D82433
Matt Arsenault [Wed, 24 Jun 2020 14:27:23 +0000 (10:27 -0400)]
Revert "[AMDGPU] Enable compare operations to be selected by divergence"
This reverts commit
521ac0b5cea02f629d035f807460affbb65ae7ad.
Reported to break thousands of piglit tests.
Arthur Eubanks [Tue, 23 Jun 2020 23:11:59 +0000 (16:11 -0700)]
[NewPM] Add SimpleLoopUnswitchPass to PassRegistry.def
Summary:
Seems to just be missing from PassRegistry.def.
Makes the number of check-llvm failures under new PM go from 2619 to 2581.
Reviewers: hans, ychen, asbirlea, leonardchan
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D82422
Arthur Eubanks [Tue, 23 Jun 2020 23:11:59 +0000 (16:11 -0700)]
[NewPM] Handle -simplifycfg in opt
Summary:
-simplifycfg is the legacy pass name for SimplifyCFGPass.
There is already -simplify-cfg in FUNCTION_PASS_WITH_PARAMS which
handles options for SimplifyCFGPass. Maybe that should be renamed to
-simplifycfg as well?
This reduces the number of check-llvm failures under NewPM from 2619 to 2392.
Reviewers: hans, leonardchan, asbirlea, ychen
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D82421
Mircea Trofin [Tue, 9 Jun 2020 21:50:50 +0000 (14:50 -0700)]
[llvm] Release-mode ML InlineAdvisor
Summary:
This implementation uses a pre-trained model which is statically
compiled into a native function.
RFC: http://lists.llvm.org/pipermail/llvm-dev/2020-April/140763.html
Reviewers: davidxl, jdoerfert, dblaikie
Subscribers: mgorny, eraman, hiraditya, arphaman, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D81515
Mircea Trofin [Tue, 9 Jun 2020 19:44:07 +0000 (12:44 -0700)]
[llvm] Added support for stand-alone cmake object libraries.
Summary:
Currently, add_llvm_library would create an OBJECT library alongside
of a STATIC / SHARED library, but losing the link interface (its
elements would become dependencies instead). To support scenarios
where linking an object library also brings in its usage
requirements, this patch adds support for 'stand-alone' OBJECT
libraries - i.e. without an accompanying SHARED/STATIC library, and
maintaining the link interface defined by the user.
This is useful for cases where, for example, we want to build a part
of a component separately. Using a STATIC target would incur the risk
that symbols not referenced in the consumer would be dropped (which may
be undesirable).
The current application is the ML part of Analysis. It should be part
of the Analysis component, so it may reference other analyses; and (in
upcoming changes) it has dependencies on optional libraries.
Reviewers: karies, davidxl, beanz, phosek, smeenai
Subscribers: mgorny, hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D81447
Sanjay Patel [Wed, 24 Jun 2020 14:11:18 +0000 (10:11 -0400)]
[VectorCombine] give invalid index value a name; NFC
Sanjay Patel [Tue, 23 Jun 2020 16:12:54 +0000 (12:12 -0400)]
[VectorCombine] add tests for extract + cmp + binop; NFC
Matt Arsenault [Thu, 18 Jun 2020 14:32:32 +0000 (10:32 -0400)]
WebAssembly: Don't store MachineFunction in MachineFunctionInfo
Soon it will be disallowed to depend on MachineFunction state in the
constructor. This was only being used to get the MachineRegisterInfo
for an assert, which I'm not sure is necessarily worth it. I would
think any missing defs would be caught by the verifier later instead.
Simon Pilgrim [Wed, 24 Jun 2020 14:12:11 +0000 (15:12 +0100)]
Cloning.h - reduce AliasAnalysis.h include to forward declarations. NFC.
Fix implicit include dependencies in source files.
Haojian Wu [Wed, 24 Jun 2020 14:14:34 +0000 (16:14 +0200)]
[AST][RecoveryExpr] Add error-bit TemplateArgument
Summary: We are missing the error-bit somehow if the error-bit is propagated
through the code path: "error type/expr" -> "template argument" ->
"template specialization type", which will lead to crashes.
Reviewers: sammccall
Reviewed By: sammccall
Subscribers: cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D82102
Pavel Labath [Wed, 24 Jun 2020 13:26:42 +0000 (15:26 +0200)]
[lldb] Refactor Scalar::TruncOrExtendTo
The "type" argument to the function is mostly useless -- the only
interesting aspect of it is signedness. Pass signedness directly and
compute the value of bits and signedness fields -- that's exactly
what the single caller of this function does.
Tim Corringham [Fri, 19 Jun 2020 17:23:56 +0000 (18:23 +0100)]
[AMDGPU] Avoid redundant mode register writes
Summary:
The SIModeRegister pass attempts to generate the minimal number of
writes to the mode register. However it was failing to correctly
deal with some loops, resulting in some redundant setreg instructions
being inserted.
This change amends the pass to avoid generating these redundant
instructions.
Subscribers: arsenm, kzhuravl, jvesely, wdng, nhaehnle, yaxunl, dstuttard, tpr, t-tye, hiraditya, kerbowa, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D82215
Florian Hahn [Wed, 24 Jun 2020 13:02:09 +0000 (14:02 +0100)]
[DSE,MSSA] Add missing -enable-dse-memoryssa flag to test.
Georgii Rymar [Wed, 24 Jun 2020 10:23:04 +0000 (13:23 +0300)]
[llvm-readobj] - Don't crash when a broken GNU hash table is dumped with --hash-symbols.
Start using the `checkGNUHashTable` helper which was recently introduced to report
a proper warning when a GNU hash table goes past the end of the file.
Differential revision: https://reviews.llvm.org/D82449
Simon Pilgrim [Wed, 24 Jun 2020 12:17:44 +0000 (13:17 +0100)]
Loads.h - reduce AliasAnalysis.h include to forward declarations. NFC.
Fix implicit include dependencies in source files.
Stefan Pintilie [Wed, 24 Jun 2020 12:40:35 +0000 (07:40 -0500)]
[LLD][PowerPC] Add support for R_PPC64_GOT_PCREL34
Add support for the 34bit relocation R_PPC64_GOT_PCREL34 for
PC Relative in LLD.
Reviewers: sfertile, MaskRay
Differential Revision: https://reviews.llvm.org/D81948
Florian Hahn [Wed, 24 Jun 2020 08:56:35 +0000 (09:56 +0100)]
[DSE] Eliminate stores at the end of the function.
This patch add support for eliminating MemoryDefs that do not have any
aliasing users, which indicates that there are no reads/writes to the
memory location until the end of the function.
To eliminate such defs, we have to ensure that the underlying object is
not visible in the caller and does not escape via returning. We need a
separate check for that, as InvisibleToCaller does not consider returns.
Reviewers: dmgreen, rnk, efriedma, bryant, asbirlea, Tyker, george.burgess.iv
Reviewed By: asbirlea
Differential Revision: https://reviews.llvm.org/D72631
sstefan1 [Fri, 19 Jun 2020 14:51:35 +0000 (16:51 +0200)]
[OpenMPOpt] ICV macro definitions
Summary:
This defines some basic information about ICVs in `OMPKinds.def`.
We also emit remarks with initial values for each function (which are default for now)
as a way to test this.
Reviewers: jdoerfert, JonChesterfield, hamax97, jhuber6
Subscribers: yaxunl, hiraditya, guansong, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D82193
Joachim Protze [Wed, 24 Jun 2020 11:42:00 +0000 (13:42 +0200)]
[OpenMP] NFC: Create OpenMP release notes file
Simon Pilgrim [Wed, 24 Jun 2020 11:22:07 +0000 (12:22 +0100)]
ObjCARC.h - remove unnecessary includes. NFC.
Add implicit InstIterator.h dependency in ObjCARCContract.cpp
Simon Pilgrim [Wed, 24 Jun 2020 10:48:17 +0000 (11:48 +0100)]
StackLifetime.h - remove unused AliasAnalysis.h include. NFC.
Alex Richardson [Tue, 23 Jun 2020 16:31:47 +0000 (17:31 +0100)]
Don't install clang-import-test
I have been trying to reduce the installed size of our CHERI toolchain and
noticed that this tool was being installed even with -DLLVM_INSTALL_TOOLCHAIN_ONLY=ON.
This appears to be a test binary that should not be installed.
Reviewed By: v.g.vassilev
Differential Revision: https://reviews.llvm.org/D82169
Krasimir Georgiev [Wed, 24 Jun 2020 10:55:05 +0000 (12:55 +0200)]
[clang-format] restore indent in conditionals when AlignOperands is DontAlign
Summary:
After D50078, we're experiencing unexpected un-indent using a style combining `AlignOperands: DontAlign` with `BreakBeforeTernaryOperators: false`, such as Google's JavaScript style:
```
% bin/clang-format -style=google ~/test.js
aaaaaaaaaaa =
bbbbbbbb ?
cccccccccccccccccc() :
dddddddddd ?
eeeeeeeeeeeeee :
fffff;
```
The issue lies with the interaction of `AlignOperands: DontAlign` and the edited code section in ContinuationIndenter.cpp, which de-dents the intent by `Style.ContinuationIndentWidth`. From [[ https://github.com/llvm/llvm-project/blob/
ac3e5c4d93fbe7fb2db3c745c721aff41cc1b851/clang/include/clang/Format/Format.h#L170 | the documentation ]] of AlignOperands: DontAlign:
> The wrapped lines are indented `ContinuationIndentWidth` spaces from the start of the line.
So the de-dent effectively erases the necessary `ContinuationIndentWidth` in that case.
This patch restores the `AlignOperands: DontAlign` behavior, producing:
```
% bin/clang-format -style=google ~/test.js
aaaaaaaaaaa =
bbbbbbbb ?
cccccccccccccccccc() :
dddddddddd ?
eeeeeeeeeeeeee :
fffff;
```
Reviewers: sammccall
Reviewed By: sammccall
Subscribers: cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D82199
Georgii Rymar [Tue, 23 Jun 2020 12:55:17 +0000 (15:55 +0300)]
[llvm-readelf] - Don't crash when e_shstrndx==SHN_XINDEX, but there is no section header.
Currently we crash when trying to print --sections and the SHN_XINDEX escape value
is used for the e_shstrndx field, but there is no section header at index 0 to
read the value from.
Differential revision: https://reviews.llvm.org/D82374
Christian Kandeler [Wed, 24 Jun 2020 10:56:45 +0000 (11:56 +0100)]
[libclang] Extend clang_Cursor_Evaluate().
Let this function (try to) evaluate expressions, in addition to
declarations and compound statements.
Patch by Christian Kandeler <christian.kandeler@qt.io>
Reviewers: nik, akyrtzi, arphaman, jkorous
Reviewed By: jkorous
Differential Revision: https://reviews.llvm.org/D80279
Cullen Rhodes [Fri, 19 Jun 2020 10:50:38 +0000 (10:50 +0000)]
[AArch64][SVE2] Add bfloat16 support to whilerw/whilewr intrinsics
Reviewed By: fpetrogalli
Differential Revision: https://reviews.llvm.org/D82399
Cullen Rhodes [Fri, 19 Jun 2020 09:37:14 +0000 (09:37 +0000)]
[AArch64][SVE] Add bfloat16 support to svlen intrinsic
Reviewed By: fpetrogalli
Differential Revision: https://reviews.llvm.org/D82186
Cullen Rhodes [Thu, 18 Jun 2020 17:25:18 +0000 (17:25 +0000)]
[AArch64][SVE] Add bfloat16 support to perm and select intrinsics
Summary:
Added for following intrinsics:
* zip1, zip2, zip1q, zip2q
* trn1, trn2, trn1q, trn2q
* uzp1, uzp2, uzp1q, uzp2q
* splice
* rev
* sel
Reviewed By: david-arm
Differential Revision: https://reviews.llvm.org/D82182
Kerry McLaughlin [Wed, 24 Jun 2020 08:47:49 +0000 (09:47 +0100)]
[AArch64][SVE] Add bfloat16 support to load intrinsics
Summary:
Bfloat16 support added for the following intrinsics:
- LD1
- LD1RQ
- LDNT1
- LDNF1
- LDFF1
Reviewers: sdesmalen, c-rhodes, efriedma, stuij, fpetrogalli, david-arm
Reviewed By: fpetrogalli
Subscribers: tschuett, kristof.beyls, hiraditya, rkruppe, psnobl, danielkiss, cfe-commits, llvm-commits
Tags: #clang, #llvm
Differential Revision: https://reviews.llvm.org/D82298
Victor Campos [Mon, 15 Jun 2020 14:54:15 +0000 (15:54 +0100)]
[ARM] Improve diagnostics message when Neon is unsupported
Summary:
Whenever Neon is not supported, a generic message is printed:
error: "NEON support not enabled"
Followed by a series of other error messages that are not useful once
the first one is printed.
This patch gives a more precise message in the case where Neon is
unsupported because an invalid float ABI was specified: the soft float
ABI.
error: "NEON intrinsics not available with the soft-float ABI. Please
use -mfloat-abi=softfp or -mfloat-abi=hard"
This message is the same one that GCC gives, so it is also making their
diagnostics more compatible with each other.
Also, by rearranging preprocessor directives, these "unsupported" error
messages are now the only ones printed out, which is also GCC's
behaviour.
Differential Revision: https://reviews.llvm.org/D81847
Konrad Kleine [Wed, 24 Jun 2020 09:17:36 +0000 (05:17 -0400)]
[lldb] fix typo in docs: withing -> within
Florian Hahn [Wed, 24 Jun 2020 09:16:26 +0000 (10:16 +0100)]
[DSE,MSSA] Precommit small test changes for D72631.
alex-t [Fri, 19 Jun 2020 14:51:54 +0000 (17:51 +0300)]
[AMDGPU] Enable compare operations to be selected by divergence
Summary: Details: This patch enables SETCC to be selected to S_CMP_* if uniform and V_CMP_* if divergent.
Reviewers: rampitec, arsenm
Reviewed By: rampitec
Subscribers: kzhuravl, jvesely, wdng, nhaehnle, yaxunl, dstuttard, tpr, t-tye, hiraditya, kerbowa, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D82194
Simon Tatham [Tue, 23 Jun 2020 16:29:40 +0000 (17:29 +0100)]
[ARM][BFloat] Legalize bf16 type even without fullfp16.
Summary:
This change permits scalar bfloats to be loaded, stored, moved and
used as function call arguments and return values, whenever the bf16
feature is supported by the subtarget.
Previously that was only supported in the presence of the fullfp16
feature, because the code generation strategy depended on instructions
from that extension. This change adds alternative code generation
strategies so that those operations can be done even without fullfp16.
The strategy for loads and stores is to replace VLDRH/VSTRH with
integer LDRH/STRH plus a move between register classes. I've written
isel patterns for those, conditional on //not// having the fullfp16
feature (so that in the fullfp16 case, the existing patterns will
still be used).
For function arguments and returns, instead of writing isel patterns
to match `VMOVhr` and `VMOVrh`, I've avoided generating those SDNodes
in the first place, by factoring out the code that constructs them
into helper functions `MoveToHPR` and `MoveFromHPR` which have a
fallback for non-fullfp16 subtargets.
The current output code is not especially pretty: in the new test file
you can see unnecessary store/load pairs implementing no-op bitcasts,
and lots of pointless moves back and forth between FP registers and
GPRs. But it at least works, which is an improvement on the previous
situation.
Reviewers: dmgreen, SjoerdMeijer, stuij, chill, miyuki, labrinea
Reviewed By: dmgreen, labrinea
Subscribers: labrinea, kristof.beyls, hiraditya, danielkiss, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D82372
Haojian Wu [Wed, 24 Jun 2020 08:19:36 +0000 (10:19 +0200)]
[AST][RecoveryExpr] Populate error-bit from Type to Expr.
Summary: Looks like this is a fallout when we introduce the error-bit in Type.
Reviewers: sammccall
Reviewed By: sammccall
Subscribers: cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D82099
LLVM GN Syncbot [Wed, 24 Jun 2020 08:17:48 +0000 (08:17 +0000)]
[gn build] Port
96d4ccf00c8
Haojian Wu [Wed, 24 Jun 2020 08:02:06 +0000 (10:02 +0200)]
[AST][RecoveryExpr] Fix a crash: don't attach error-type base specifiers.
Summary:
otherwise we'll run into code path which expects a good base specifiers,
and lead to crashes.
The crash only occurs in template instantiations (in non-template case,
the bad base specifiers are dropped during parsing.)
crash stacktrace:
```
clang: llvm-project/clang/lib/Sema/SemaInit.cpp:7864: clang::ExprResult clang::InitializationSequence::Perform(clang::Sema &, const clang::InitializedEntity &, const clang::InitializationKind &, clang::MultiExprArg, clang::QualType *): Assertion `Kind.getKind() == InitializationKind::IK_Copy || Kind.isExplicitCast() || Kind.getKind() == InitializationKind::IK_DirectList' failed.
PLEASE submit a bug report to https://bugs.llvm.org/ and include the crash backtrace, preprocessed source, and associated run script.
Stack dump:
```
Reviewers: sammccall
Reviewed By: sammccall
Subscribers: cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D82086
Kazushi (Jam) Marukawa [Wed, 24 Jun 2020 08:11:59 +0000 (10:11 +0200)]
[VE] Clang toolchain for VE
Summary:
This patch enables compilation of C code for the VE target with Clang.
Differential Revision: https://reviews.llvm.org/D79411
Siva Chandra [Wed, 24 Jun 2020 07:16:23 +0000 (00:16 -0700)]
[libc] Enable copysignl, frexpl, logbl and modfl on aarch64.
Craig Topper [Wed, 24 Jun 2020 07:10:36 +0000 (00:10 -0700)]
[X86] Speculatively fix to X86AvoidStoreForwardingBlocks not deference a machine mem operand if there isn't one present.
Eric Christopher informed me that FastISel memcpy handling creates
load/store instructions without mem operands. We should fix that,
but I doubt that's the only case of missed mem operands so seems
better to be defensive here.
I don't have a test case yet, but I'll try to add one if i get a
test from Eric.
Siva Chandra Reddy [Wed, 24 Jun 2020 07:01:03 +0000 (00:01 -0700)]
[libc][Obvious] Remove a debug #include of iostream.
Siva Chandra Reddy [Fri, 19 Jun 2020 20:08:10 +0000 (13:08 -0700)]
[libc] Add long double flavors of the floating point manipulation functions.
Specifically: copysignl, frexpl, logbl and modfl have been added.
Reviewers: asteinhauser
Differential Revision: https://reviews.llvm.org/D82357
Craig Topper [Wed, 24 Jun 2020 06:40:01 +0000 (23:40 -0700)]
[X86] Add mayLoad/mayStore flags to some X87 instructions that don't have isel patterns to infer them from.
Should remove part of the differences in D81833 due to some
some of these getting isel patterns.
Jonas Devlieghere [Wed, 24 Jun 2020 05:40:26 +0000 (22:40 -0700)]
[lldb/Test] Temporarily disable TestSimulatorPlatform.py
Alex Lorenz [Wed, 24 Jun 2020 04:17:55 +0000 (21:17 -0700)]
test fix: add missing system-darwin REQUIRES
The test should only run with a Darwin driver only.
Alex Lorenz [Wed, 24 Jun 2020 04:05:11 +0000 (21:05 -0700)]
[clang][driver] set macOS as the target OS for -arch arm64 when clang
is running on an Apple Silicon mac
This change allows users to use `-arch arm64` to build for mac when
running it on Apple Silicon mac without explicit `-target` option.
Differential Revision: https://reviews.llvm.org/D82428
Alex Lorenz [Wed, 24 Jun 2020 04:01:26 +0000 (21:01 -0700)]
[cmake] configure the host triple on an Apple Silicon machine correctly
The cmake build of LLVM now uses the appropriate arm64 arch for the
host triple when building llvm-project on an Apple Silicon mac.
Differential Revision: https://reviews.llvm.org/D82428
Rahul Joshi [Wed, 24 Jun 2020 03:21:42 +0000 (20:21 -0700)]
[NFC][MLIR] Undo anonymous namespace change from https://reviews.llvm.org/D82417
Undo as it does not conform to LLVM coding style
(https://llvm.org/docs/CodingStandards.html#anonymous-namespaces)
Vitaly Buka [Wed, 24 Jun 2020 02:26:03 +0000 (19:26 -0700)]
Add option to symbolize inline frames for InternalSymbolizer
Summary:
Currently, there is no way to let the `InternalSymbolizer` implemented
functions know if inline frames should be symbolized. This patch updates
the function `__sanitizer_symbolize_code` to include a parameter for
this ASAN option and toggle between LLVM symbolization functions when
appropriate.
Fixes the following two failing tests when internal symbolization is
enabled:
```
SanitizerCommon-*-x86_64-Linux :: print-stack-trace.cpp
SanitizerCommon-*-x86_64-Linux :: symbolize_pc_inline.cpp
```
Reviewers: vitalybuka, kcc, filcab
Reviewed By: vitalybuka
Subscribers: #sanitizers
Tags: #sanitizers
Differential Revision: https://reviews.llvm.org/D79280
Eli Friedman [Wed, 20 May 2020 22:08:36 +0000 (15:08 -0700)]
[BitcodeReader] Fix DelayedShuffle handling for ConstantExpr shuffles.
The indexing was messed up, so the result was completely broken.
Shuffle constant exprs are rare in practice; without vscale types,
constant folding generally elminates them. So sort of hard to trip over.
Fixes regression from D72467.
Differential Revision: https://reviews.llvm.org/D80330
Ahsan Saghir [Mon, 15 Jun 2020 01:37:29 +0000 (20:37 -0500)]
[PowerPC] Add support for vector bool __int128 for Power10
Summary:
This patch adds support for `vector bool __int128` type for Power10.
Reviewers: #powerpc, hfinkel, lei, stefanp, amyk
Reviewed By: #powerpc, lei, amyk
Subscribers: lei, amyk, wuzish, nemanjai, shchenz, cfe-commits
Tags: #llvm, #powerpc, #clang
Differential Revision: https://reviews.llvm.org/D81816
Amara Emerson [Mon, 22 Jun 2020 23:25:49 +0000 (16:25 -0700)]
[AArch64][GlobalISel] Improve codegen for some constant vectors by using constant pool loads.
There's more smarts in AArch64ISelLowering that we don't have yet, but this
change incrementally improves some of the more common patterns. I think future
iterations will want to use some combination of PostLegalizerCombiner and the
selector to catch the other cases.
Differential Revision: https://reviews.llvm.org/D82340
Eli Friedman [Tue, 19 May 2020 03:38:13 +0000 (20:38 -0700)]
Remove GlobalValue::getAlignment().
This function is deceptive at best: it doesn't return what you'd expect.
If you have an arbitrary GlobalValue and you want to determine the
alignment of that pointer, Value::getPointerAlignment() returns the
correct value. If you want the actual declared alignment of a function
or variable, GlobalObject::getAlignment() returns that.
This patch switches all the users of GlobalValue::getAlignment to an
appropriate alternative.
Differential Revision: https://reviews.llvm.org/D80368
Vedant Kumar [Tue, 23 Jun 2020 22:53:09 +0000 (15:53 -0700)]
[SimplifyCFG] Drop debug loc in SpeculativelyExecuteBB
Summary:
According to HowToUpdateDebugInfo.rst:
```
Preserving the debug locations of speculated instructions can make
it seem like a condition is true when it's not (or vice versa), which
leads to a confusing single-stepping experience
```
This patch follows the recommendation to drop debug locations on
speculated instructions.
Reviewers: aprantl, davide
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D82420
Matt Arsenault [Wed, 24 Jun 2020 01:06:11 +0000 (21:06 -0400)]
AMDGPU/GlobalISel: Fix fixed ABI special VGPR function arguments
I forgot to copy the new fixed function ABI into GlobalISel, so this
was mismatched with the DAG compiled calling function. This was
allocating part of the argument list to v31, which was supposed to be
reserved for the workitem IDs.
Sam Clegg [Tue, 23 Jun 2020 03:30:40 +0000 (20:30 -0700)]
[WebAssebmly] Fully disable 'protected' visibility
Emscripten doesn't use protected visibility either.
Differential Revision: https://reviews.llvm.org/D82346
Rahul Joshi [Wed, 24 Jun 2020 00:23:35 +0000 (17:23 -0700)]
[MLIR] [NFC] Add new line and empty line before printing modified loop
to make the debug output readable.
Differential Revision: https://reviews.llvm.org/D82417
Eli Friedman [Tue, 28 Apr 2020 23:04:26 +0000 (16:04 -0700)]
[clang codegen] Fix alignment of "Address" for incomplete array pointer.
The code was assuming all incomplete types don't have meaningful
alignment, but incomplete arrays do have meaningful alignment.
Fixes https://bugs.llvm.org/show_bug.cgi?id=45710
Differential Revision: https://reviews.llvm.org/D79052
Richard Smith [Tue, 16 Jun 2020 23:55:56 +0000 (16:55 -0700)]
DR458: Search template parameter scopes in the right order.
C++ unqualified name lookup searches template parameter scopes
immediately after finishing searching the entity the parameters belong
to. (Eg, for a class template, you search the template parameter scope
after looking in that class template and its base classes and before
looking in the scope containing the class template.) This is complicated
by the fact that scope lookup within a template parameter scope looks in
a different sequence of places prior to reaching the end of the
declarator-id in the template declaration.
We used to approximate the proper lookup rule with a hack in the scope /
decl context walk inside name lookup. Now we instead compute the lookup
parent for each template parameter scope.
In order to get this right, we now make sure to enter a distinct Scope
for each template parameter scope, and make sure to re-enter the
enclosing class scopes properly when handling delay-parsed regions
within a class.
Amy Huang [Wed, 24 Jun 2020 00:10:34 +0000 (17:10 -0700)]
[NFC] Remove outdated comment in llvm-symbolizer test case.
Eli Friedman [Mon, 8 Jun 2020 23:34:15 +0000 (16:34 -0700)]
[AArch64][SVE] Add legalization support for i32/i64 vector srem/urem
Implement them on top of sdiv/udiv, similar to what we do for integer
types.
Potential future work: implementing i8/i16 srem/urem, optimizations for
constant divisors, optimizing the mul+sub to mls.
Differential Revision: https://reviews.llvm.org/D81511
Eli Friedman [Wed, 17 Jun 2020 23:35:35 +0000 (16:35 -0700)]
[IR] Prefer scalar type for struct indexes in GEP constant expressions.
This has two advantages: one, it's simpler, and two, it doesn't require
heroic pattern matching with scalable vectors.
Also includes a small fix to DataLayout to allow the scalable vector
testcase to work correctly.
Differential Revision: https://reviews.llvm.org/D82061
Leonard Chan [Tue, 23 Jun 2020 23:10:07 +0000 (16:10 -0700)]
[lld][ELF][AArch64] Handle R_AARCH64_PLT32 relocation
This is the followup to D77647 which implements handling for the new
R_AARCH64_PLT32 relocation type in lld. This relocation would benefit the
PIC-friendly vtables feature described in D72959.
Differential Revision: https://reviews.llvm.org/D81184
Petr Hosek [Wed, 17 Jun 2020 21:10:02 +0000 (14:10 -0700)]
[ELF] Add -z start-stop-visibility= to set __start_/__stop_ symbol visibility
This matches the equivalent flag implemented in GNU linkers, see
https://sourceware.org/pipermail/binutils/2020-June/111685.html for
the associated discussion.
Differential Revision: https://reviews.llvm.org/D55682
Tony [Fri, 19 Jun 2020 19:01:37 +0000 (15:01 -0400)]
[AMDGPU] Update AMD GPU processor information
Summary:
- Add product names for some processors.
- Correct XNACK support for a processor.
Subscribers: kzhuravl, jvesely, wdng, nhaehnle, yaxunl, dstuttard, tpr, kerbowa, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D82348
Richard Smith [Wed, 17 Jun 2020 23:02:06 +0000 (16:02 -0700)]
Move late-parsed class member attribute handling adjacent to all the
other late-parsed class component handling.
No functionality change intended.
Sam Clegg [Tue, 23 Jun 2020 21:59:18 +0000 (14:59 -0700)]
[WebAssembly] Fix for use of uninitialized member in WasmObjectWriter.cpp
Currently, section indices may be passed uninitialized by value if
writing the section fails. Removes section indices form class
initialization and returns them from the write{Code,Data}Section
function calls instead.
Patch by Gui Andrade!
Differential Revision: https://reviews.llvm.org/D81702
aartbik [Tue, 23 Jun 2020 22:22:37 +0000 (15:22 -0700)]
[mlir] [integration-test] Let target check-mlir imply target check-mlir-integration too
Note that this does not mean that check-mlir will run check-mlir-integration
tests for all configurations. You still need to do a set up with the flag
MLIR_INCLUDE_INTEGRATION_TESTS set to ON in order to activate the integration test.
Reviewed By: jpienaar
Differential Revision: https://reviews.llvm.org/D82413
David Blaikie [Tue, 23 Jun 2020 22:09:40 +0000 (15:09 -0700)]
Remove clang::Codegen::EHPadEndScope as unused
Unused since r255423 / D15140 /
4e52d6f811a2269e946c19e77245148bd9221f99
Found indirectly by assessing -debug-info-kind=constructors and
observing the EHPadEndScope type was never emitted because the
constructor is never called. (all credit to Amy Huang for identifying
this issue)
Alex Lorenz [Tue, 23 Jun 2020 21:42:32 +0000 (14:42 -0700)]
[clang][driver] allow macOS 11 OS version in the driver
Jonas Devlieghere [Tue, 23 Jun 2020 22:13:14 +0000 (15:13 -0700)]
[lldb] Fix the modules build
Fixes error: invalid operands to binary expression ('llvm::StringRef'
and 'const char [6]')
Luís Marques [Tue, 23 Jun 2020 20:21:12 +0000 (21:21 +0100)]
[RISCV][NFC] Add tests for folds of ADDIs into load/stores
This patch adds tests for folds of ADDIs into load/stores, focusing on
load/stores with nonzero offsets. When the offset is nonzero we currently
don't do the fold. A follow-up patch will improve on that.
Differential Revision: https://reviews.llvm.org/D79689
Eric Schweitz [Tue, 23 Jun 2020 15:15:09 +0000 (08:15 -0700)]
[flang] add RTBuilder.h
This is a set of type building models that is specific to the lowering
process. It provides the mechanism of mapping C(++) header file interfaces
to the MLIR+FIR type system.
It also provides some macros to build a constexpr evaluated table to
runtime functions. This code is used to build the interface tables to
various runtime support libraries.
Differential revision: https://reviews.llvm.org/D82387
David Green [Tue, 23 Jun 2020 21:24:58 +0000 (22:24 +0100)]
[ARM] Mark more integer instructions as not having side effects.
LDRD and STRD along with UBFX and SBFX are selected from DAGToDAG
transforms, so do not have tblgen patterns. They don't get marked as
having side effects so cannot be scheduled as efficiently as you would
like.
This specifically marks then as not having side effects.
Differential Revision: https://reviews.llvm.org/D82358
Richard Smith [Wed, 17 Jun 2020 21:49:03 +0000 (14:49 -0700)]
When performing a substitution into a dependent alias template, mark the
outer levels as retained rather than omitting their arguments.
This better reflects what's going on (we're performing a substitution
while still inside a template), and in theory is more correct, but I've
not found a testcase where it matters in practice (largely because we
don't allow alias templates to be declared inside a function).
Fixed AST dumping of SubstNonTypeTemplateParm[Pack]Expr to demonstrate
that we're properly substituting through dependent alias templates. (We
can't deduce properly through these yet, but we can at least produce the
right input to template argument deduction.)
No functionality change intended.
aartbik [Tue, 23 Jun 2020 21:33:38 +0000 (14:33 -0700)]
[mlir] [VectorOps] Improve vector.create_mask lowering
Use vector compares for the 1-D case. This approach scales much better
than generating insertion operations, and exposes SIMD directly to backend.
Reviewed By: ftynse
Differential Revision: https://reviews.llvm.org/D82402
Christopher Tetreault [Tue, 23 Jun 2020 20:28:55 +0000 (13:28 -0700)]
[SVE] Remove calls to VectorType::getNumElements from AsmParser
Reviewers: efriedma, RKSimon, c-rhodes, fpetrogalli
Reviewed By: fpetrogalli
Subscribers: tschuett, hiraditya, rkruppe, psnobl, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D82208
David Green [Tue, 23 Jun 2020 21:03:30 +0000 (22:03 +0100)]
[ARM] Cortex-M4 integer instructions scheduler info test. NFC
Most useful at the moment for showing where unpredicatable instructions are.
Zequan Wu [Mon, 22 Jun 2020 18:43:52 +0000 (11:43 -0700)]
[ASan][MSan] Remove EmptyAsm and set the CallInst to nomerge to avoid from merging.
Summary: `nomerge` attribute was added at D78659. So, we can remove the EmptyAsm workaround in ASan the MSan and use this attribute.
Reviewers: vitalybuka
Reviewed By: vitalybuka
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D82322
Jonas Devlieghere [Tue, 23 Jun 2020 21:17:08 +0000 (14:17 -0700)]
[lldb/Lua] Fix typo: s/stdout/stderr/
This wasn't caught by the existing test, but will be covered by the
extended test that's part of D82412.
Gui Andrade [Tue, 23 Jun 2020 21:10:31 +0000 (14:10 -0700)]
test/msan/sigwait: Don't silently ignore assertion failures
Summary: As the parent process would return 0 independent of whether the child succeeded, assertions in the child would be ignored.
Reviewers: eugenis
Reviewed By: eugenis
Subscribers: #sanitizers
Tags: #sanitizers
Differential Revision: https://reviews.llvm.org/D82400
Ryan Santhiraraja [Tue, 23 Jun 2020 21:04:36 +0000 (22:04 +0100)]
Preserve GlobalsAA analysis result in InjectTLIMappings
InjectTLIMappings fails to preserve the analysis result of GlobalsAA. Not preserving the analysis might affect benchmark performance. This change fixes this issue.
Patch by: Ryan Santhiraraja <rsanthir@quicinc.com>
Reviewers: fpetrogalli, joerg, fhahn
Reviewed By: fhahn
Differential Revision: https://reviews.llvm.org/D82343
Adrian Prantl [Tue, 23 Jun 2020 20:36:06 +0000 (13:36 -0700)]
Add missing string conversions to fix a compile error in Local.h
Zhi Zhuang [Tue, 23 Jun 2020 20:30:23 +0000 (13:30 -0700)]
fix test failure for clang/test/CodeGen/builtin-expect-with-probability.cpp
Fix test case added by D79830
Rewrite the test case, which did similar thing as builtin-expect.c
does(test generated llvm intrinsic instead of test branch weights).
Currently pass by "-disable-llvm-passes" option.
Differential Revision: https://reviews.llvm.org/D82403
Nikita Popov [Tue, 23 Jun 2020 20:25:04 +0000 (22:25 +0200)]
[IR] Remove MSVC warning workaround (NFC)
While LLVM does fold this to x+1, GCC does not. As this is hot
code, let's try to avoid that.
According to
https://developercommunity.visualstudio.com/content/problem/211134/unsigned-integer-overflows-in-constexpr-functionsa.html
this spurious warning in MSVC has been fixed in Visual Studio 2019
Version 16.4. Let's see if there are any build bots running old
MSVC versions with warnings treated as errors...
Christopher Tetreault [Tue, 23 Jun 2020 19:47:37 +0000 (12:47 -0700)]
[SVE] Remove calls to VectorType::getNumElements from Bitcode
Reviewers: efriedma, evgeny777, tejohnson, david-arm, kmclaughlin
Reviewed By: david-arm
Subscribers: tschuett, hiraditya, rkruppe, psnobl, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D82209
Nikita Popov [Tue, 23 Jun 2020 20:18:56 +0000 (22:18 +0200)]
[IR] Remove unnecessary uint64_t casts (NFC)
As pointed out by foad, it's not necessary to work on uint64_t
here. The values used here fit uint8_t.
Siva Chandra Reddy [Tue, 23 Jun 2020 20:11:07 +0000 (13:11 -0700)]
[libc][Obvious] Fix few typos in tests.
Florian Hahn [Tue, 23 Jun 2020 18:26:31 +0000 (19:26 +0100)]
[DSE,MSSA] Treat `store 0` after calloc as noop stores.
This patch extends storeIsNoop to also detect stores of 0 to an calloced
object. This basically ports the logic from legacy DSE to the MemorySSA
backed version.
It triggers in a few cases on MultiSource, SPEC2000, SPEC2006 with -O3
LTO:
Same hash: 218 (filtered out)
Remaining: 19
Metric: dse.NumNoopStores
Program base patch2 diff
test-suite...CFP2000/177.mesa/177.mesa.test 1.00 15.00 1400.0%
test-suite...6/482.sphinx3/482.sphinx3.test 1.00 14.00 1300.0%
test-suite...lications/ClamAV/clamscan.test 2.00 28.00 1300.0%
test-suite...CFP2006/433.milc/433.milc.test 1.00 8.00 700.0%
test-suite...pplications/oggenc/oggenc.test 2.00 9.00 350.0%
test-suite.../CINT2000/176.gcc/176.gcc.test 6.00 6.00 0.0%
test-suite.../CINT2006/403.gcc/403.gcc.test NaN 137.00 nan%
test-suite...libquantum/462.libquantum.test NaN 3.00 nan%
test-suite...6/464.h264ref/464.h264ref.test NaN 7.00 nan%
test-suite...decode/alacconvert-decode.test NaN 2.00 nan%
test-suite...encode/alacconvert-encode.test NaN 2.00 nan%
test-suite...ications/JM/ldecod/ldecod.test NaN 9.00 nan%
test-suite...ications/JM/lencod/lencod.test NaN 39.00 nan%
test-suite.../Applications/lemon/lemon.test NaN 2.00 nan%
test-suite...pplications/treecc/treecc.test NaN 4.00 nan%
test-suite...hmarks/McCat/08-main/main.test NaN 4.00 nan%
test-suite...nsumer-lame/consumer-lame.test NaN 3.00 nan%
test-suite.../Prolangs-C/bison/mybison.test NaN 1.00 nan%
test-suite...arks/mafft/pairlocalalign.test NaN 30.00 nan%
Reviewers: efriedma, zoecarver, asbirlea
Reviewed By: asbirlea
Differential Revision: https://reviews.llvm.org/D82204
Stefan Pintilie [Tue, 23 Jun 2020 19:59:19 +0000 (14:59 -0500)]
[LLD][PowerPC] Add support for R_PPC64_PCREL34
Add support for the 34bit relocation R_PPC64_PCREL34 for PC Relative in LLD.
Walter Erquinigo [Mon, 15 Jun 2020 21:08:52 +0000 (14:08 -0700)]
Redo of Add terminateCommands to lldb-vscode protocol
Summary:
This redoes https://reviews.llvm.org/D79726 and fixes two things.
- The logic that determines whether to automatically disconnect during the tear down is not very dumb compared to the original implementation. Each test will determine whether to do that or not.
- The terminate commands and terminate event were being sent after the disconnect response was sent to the IDE. That was not good, as VSCode stops the debug session as soon as it receives a disconnect response. Now, the terminate event and terminateEvents are being executed before the disconnect response is sent. This ensures that any connection between the IDE and lldb-vscode is alive while the terminate commands are executed. Besides, it also allows displaying the output of the terminate commands on the debug console, as it's still alive.
Reviewers: clayborg, aadsm, kusmour, labath
Subscribers: lldb-commits
Tags: #lldb
Differential Revision: https://reviews.llvm.org/D81978
Davide Italiano [Tue, 23 Jun 2020 19:45:31 +0000 (12:45 -0700)]
Generalize TestFormattersBoolRefPtr to work on Apple Silicon.
Davide Italiano [Tue, 23 Jun 2020 19:37:45 +0000 (12:37 -0700)]
[ObjectFileMachO] Check for TARGET_EMBEDDED instead of listing architectures.
Now that Apple Silicon is a thing, we need to generalize the check.
Davide Italiano [Tue, 23 Jun 2020 19:24:53 +0000 (12:24 -0700)]
[Host] Check for TARGET_OS_EMBEDDED instead of listing architectures.
With the advent of Apple Silicon, checking for the architectures
specifically is not correct anymore. This code is only supposed to
run on embedded devices (iPhones et similia), so mark it accordingly.
Jean-Michel Gorius [Tue, 23 Jun 2020 19:24:04 +0000 (21:24 +0200)]
[mlir] Avoid pontentially ambiguous class name
Summary: The Pass class exists in both the mlir and the llvm namespaces. Use the fully qualified class name to avoid any ambiguities.
Reviewers: rriddle
Reviewed By: rriddle
Subscribers: mehdi_amini, jpienaar, shauheen, antiagainst, nicolasvasilache, arpith-jacob, mgester, lucyrfox, aartbik, liufengdb, stephenneuendorffer, Joonsoo, grosul1, jurahul, msifontes
Tags: #mlir
Differential Revision: https://reviews.llvm.org/D82371
Your Name [Tue, 23 Jun 2020 19:08:56 +0000 (00:38 +0530)]
[AMDGPU/MemOpsCluster] Implement new heuristic for computing max mem ops cluster size
Summary:
Make use of both the - (1) clustered bytes and (2) cluster length, to decide on
the max number of mem ops that can be clustered. On an average, when loads
are dword or smaller, consider `5` as max threshold, otherwise `4`. This
heuristic is purely based on different experimentation conducted, and there is
no analytical logic here.
Reviewers: foad, rampitec, arsenm, vpykhtin
Reviewed By: rampitec
Subscribers: llvm-commits, kerbowa, hiraditya, t-tye, Anastasia, tpr, dstuttard, yaxunl, nhaehnle, wdng, jvesely, kzhuravl, thakis
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D82393
Zion Nimchuk [Sat, 6 Jun 2020 20:18:52 +0000 (16:18 -0400)]
Change CMake so that we only look for Z3 when LLVM_ENABLE_Z3_SOLVER is enabled
Reviewers: mikhail.ramalho
Reviewed By: mikhail.ramalho
Subscribers: mehdi_amini, mgorny, mikhail.ramalho, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D75544
Fangrui Song [Tue, 23 Jun 2020 18:06:39 +0000 (11:06 -0700)]
[ELF] Resolve relocations in .debug_* referencing (discarded symbols or ICF folded section symbols) to tombstone values
See D59553, https://lists.llvm.org/pipermail/llvm-dev/2020-May/141885.html and
https://sourceware.org/pipermail/binutils/2020-May/111357.html for
extensive discussions on a tombstone value.
See http://www.dwarfstd.org/ShowIssue.php?issue=200609.1
(Reserve an address value for "not present") for a DWARF enhancement proposal.
We resolve such relocations to a tombstone value to indicate that the address is invalid.
This solves several problems (the normal behavior is to resolve the relocation to the addend):
* For an empty function in a collected section, a pair of (0,0) can
terminate .debug_loc and .debug_ranges (as of binutils 2.34, GNU ld
resolves such a relocation to 1 to avoid the .debug_ranges issue)
* If DW_AT_high_pc is sufficiently large, the address range can collide
with a regular code range of low address (https://bugs.llvm.org/show_bug.cgi?id=41124 )
* If a text section is folded into another by ICF, we may leave entries
in multiple CUs claiming ownership of the same range of code, which can
confuse consumers.
* Debug information associated with COMDAT sections can have problems
similar to ICF, but is more complex - thus not addressed by this patch.
For pre-DWARF-v5 .debug_loc and .debug_ranges, a pair of 0 can terminate
entries (invalidating subsequent ranges).
-1 is a reserved value with special meaning (base address selection entry) which can't be used either.
Use -2 instead.
For all other .debug_*, use UINT32_MAX for 32-bit targets and UINT64_MAX
for 64-bit targets. In the code, we intentionally use
`uint64_t tombstone = UINT64_MAX` for 32-bit targets as well: this matches
SignExtend64 as used in `relocateAlloc`. (Actually UINT32_MAX does not work for R_386_32)
Note 0, we only special case `target->symbolicRel` (R_X86_64_64, R_AARCH64_ABS64, R_PPC64_ADDR64), not
short-range absolute relocations (e.g. R_X86_64_32). Only forms like DW_FORM_addr need to be special cased.
They can hold an arbitrary address (must be 64-bit on a 64-bit target). (In theory,
producers can make use of small code model to emit 32-bit relocations. This doesn't seem to be leveraged.)
Note 1, we have to ignore the addend, because we don't want to resolve
DW_AT_low_pc (which may have a non-zero addend) to -1+addend (wrap
around to a low address):
__attribute__((section(".text.x"))) void f1() { }
__attribute__((section(".text.x"))) void f2() { } // DW_AT_low_pc has a non-zero addend
Note 2, if the prevailing copy does not have debugging information while
a non-prevailing copy has (partial debug build), we don't do extra work
to attach debugging information to the prevailing definition. (clang
has a lot of debug info optimizations that are on-by-default that assume
the whole program is built with debug info).
clang -c -ffunction-sections a.cc # prevailing copy has no debug info
clang -c -ffunction-sections -g b.cc
Reviewed By: dblaikie, avl, jhenderson
Differential Revision: https://reviews.llvm.org/D81784
Walter Erquinigo [Thu, 4 Jun 2020 22:55:58 +0000 (15:55 -0700)]
[vscode] set default values for terminateDebuggee for the disconnect request
Summary:
Recently I've noticed that VSCode sometimes doesn't send the terminateDebuggee flag within the disconnectRequest,
even though lldb-vscode sets the terminateDebuggee capability correctly.
This has been causing that inferiors don't die after the debug session ends, and many users have reported issues because of this.
An easy way to mitigate this is to set better default values for the terminateDebuggee field in the disconnect request.
I'm assuming that for a launch request, the default will be true, and for attach it'll be false.
Reviewers: clayborg, labath, aadsm
Subscribers: lldb-commits
Tags: #lldb
Differential Revision: https://reviews.llvm.org/D81200