platform/upstream/llvm.git
5 years agoImplement the 'sys_time' portions of the C++20 calendaring stuff. Reviewed as D56494
Marshall Clow [Fri, 11 Jan 2019 15:12:04 +0000 (15:12 +0000)]
Implement the 'sys_time' portions of the C++20 calendaring stuff. Reviewed as D56494

llvm-svn: 350929

5 years ago[x86] allow insert/extract when matching horizontal ops
Sanjay Patel [Fri, 11 Jan 2019 14:27:59 +0000 (14:27 +0000)]
[x86] allow insert/extract when matching horizontal ops

Previously, we limited this transform to cases where the
extraction into the build vector happens from vectors of
the same type as the build vector, but that's not required.

There's a slight potential regression seen in the AVX512
result for phadd -- we're using the 256-bit flavor of the
instruction now even though the 128-bit subset is sufficient.
The same problem could already be seen in the AVX2 result.
Follow-up patches will attempt to narrow that back down.

llvm-svn: 350928

5 years ago[llvm-objcopy] [COFF] Implmement --strip-unneeded and -x/--discard-all for symbols
Martin Storsjo [Fri, 11 Jan 2019 14:13:04 +0000 (14:13 +0000)]
[llvm-objcopy] [COFF] Implmement --strip-unneeded and -x/--discard-all for symbols

Differential Revision: https://reviews.llvm.org/D56480

llvm-svn: 350927

5 years ago[llvm-objcopy] [COFF] Fix writing object files without symbols/string table
Martin Storsjo [Fri, 11 Jan 2019 13:47:37 +0000 (13:47 +0000)]
[llvm-objcopy] [COFF] Fix writing object files without symbols/string table

Previously, this was broken - by setting PointerToSymbolTable to zero
but still actually writing the string table length, the object file
header was corrupted.

Differential Revision: https://reviews.llvm.org/D56584

llvm-svn: 350926

5 years ago[llvm-symbolizer] Add -exe, -e as aliases to -obj
Dmitry Venikov [Fri, 11 Jan 2019 11:51:52 +0000 (11:51 +0000)]
[llvm-symbolizer] Add -exe, -e as aliases to -obj

Summary: Provides -exe, -e as aliases to -obj. Motivation: https://bugs.llvm.org/show_bug.cgi?id=40071

Reviewers: ruiu, rnk, fjricci, jhenderson

Reviewed By: jhenderson

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D56580

llvm-svn: 350925

5 years agoIntroduce SymbolFileBreakpad and use it to fill symtab
Pavel Labath [Fri, 11 Jan 2019 11:17:51 +0000 (11:17 +0000)]
Introduce SymbolFileBreakpad and use it to fill symtab

Summary:
This commit adds the glue code necessary to integrate the
SymbolFileBreakpad into the plugin system. Most of the methods are
stubbed out. The only method implemented method is AddSymbols, which
parses the PUBLIC "section" of the breakpad "object file", and fills out
the Module's symtab.

To enable testing this, I've made two additional changes:
- dump Symtab from the SymbolVendor class. The symtab was already being
  dumped as a part of the object file dump, but that happened before
  symbol vendor kicked in, so it did not reflect any symbols added
  there.
- add ability to explicitly specify the external symbol file in
  lldb-test (so that the object file could be linked with the breakpad
  symbol file). To make things simpler, I've changed lldb-test from
  consuming multiple inputs (and dumping their symbols) to having it
  just process a single file per invocation. This was not a problem
  since everyone was using it that way already.

Reviewers: clayborg, zturner, lemo, markmentovai, amccarth

Subscribers: mgorny, lldb-commits

Differential Revision: https://reviews.llvm.org/D56173

llvm-svn: 350924

5 years agoELF: Fix base address computation code for files generated by yaml2obj
Pavel Labath [Fri, 11 Jan 2019 10:18:40 +0000 (10:18 +0000)]
ELF: Fix base address computation code for files generated by yaml2obj

The code was assuming that the elf file will have a PT_LOAD segment
starting from the first byte of the file. While this is true for files
generated by most linkers (it's a way of saving space), it is not a
requirement. And files not satisfying this constraint can still be
perfectly executable. yaml2obj is one of the tools which produces files
like this.

This patch relaxes the check in ObjectFileELF to take the address of the
first PT_LOAD segment as the base address of the object (instead of the
one with the offset 0). Since the PT_LOAD segments are supposed to be
sorted according to the VM address, this entry will also be the one with
the lowest VM address.

If we ever run into files which don't have the PT_LOAD segments sorted,
we can easily change this code to return the lowest VM address as the
base address (if that is the correct thing to do for these files).

llvm-svn: 350923

5 years ago[clang-tidy] new check 'readability-redundant-preprocessor'
Miklos Vajna [Fri, 11 Jan 2019 07:59:47 +0000 (07:59 +0000)]
[clang-tidy] new check 'readability-redundant-preprocessor'

Finds potentially redundant preprocessor directives.

Reviewed By: aaron.ballman

Differential Revision: https://reviews.llvm.org/D54349

llvm-svn: 350922

5 years agoRevert "[SelectionDAGBuilder] Refactor GetRegistersForValue. NFCI."
Martin Storsjo [Fri, 11 Jan 2019 07:31:17 +0000 (07:31 +0000)]
Revert "[SelectionDAGBuilder] Refactor GetRegistersForValue. NFCI."

This reverts commit r350841, as it actually had functional changes
and broke compilation. See PR40290.

llvm-svn: 350921

5 years ago[Sema] Make canPassInRegisters return true if the CXXRecordDecl passed
Akira Hatanaka [Fri, 11 Jan 2019 07:06:38 +0000 (07:06 +0000)]
[Sema] Make canPassInRegisters return true if the CXXRecordDecl passed
to it is a trivial_abi class.

A class that has all of its copy and move constructors deleted can still
be passed or returned in registers if the class is annotated with
trivial_abi.

This fixes PR39683.

llvm-svn: 350920

5 years ago[Sanitizer] Intercept getusershell
David Carlier [Fri, 11 Jan 2019 05:53:09 +0000 (05:53 +0000)]
[Sanitizer] Intercept getusershell

- If entries are properly copied (there were a bug in FreeBSD implementation in earlier version), or list properly reset.

Reviewers: vitalybuka, krytarowski

Reviewed By: krytarowski

Differential Revision: https://reviews.llvm.org/D56562

llvm-svn: 350919

5 years ago[X86] Change vXi1 extract_vector_elt lowering to be legal if the index is 0. Add...
Craig Topper [Fri, 11 Jan 2019 05:44:56 +0000 (05:44 +0000)]
[X86] Change vXi1 extract_vector_elt lowering to be legal if the index is 0. Add DAG combine to turn scalar_to_vector+extract_vector_elt into extract_subvector.

We were lowering the last step extract_vector_elt to a bitcast+truncate. Change it to use an extract_vector_elt of index 0 instead. Add isel patterns to do the equivalent of what the bitcast would have done. Plus an isel pattern for an any_extend+extract to prevent some regressions.

Finally add a DAG combine to turn v1i1 scalar_to_vector+extract_vector_elt of 0 into an extract_subvector.

This fixes some of the regressions from D350800.

llvm-svn: 350918

5 years ago[Sema] If CheckPlaceholderExpr rewrites the initializer of an auto
Akira Hatanaka [Fri, 11 Jan 2019 04:57:34 +0000 (04:57 +0000)]
[Sema] If CheckPlaceholderExpr rewrites the initializer of an auto
variable during auto type deduction, use the rewritten initializer when
performing initialization of the variable.

This silences spurious -Warc-repeated-use-of-weak warnings that are
issued when the initializer uses a weak ObjC pointer.

Differential Revision: https://reviews.llvm.org/D55662

llvm-svn: 350917

5 years agoAdapt to CXXMethodDecl::getThisType change (NFC)
Brian Gesiak [Fri, 11 Jan 2019 02:12:31 +0000 (02:12 +0000)]
Adapt to CXXMethodDecl::getThisType change (NFC)

Summary:
https://reviews.llvm.org/D56509 changed the API of the
CXXMethodDecl::getThisType method. Adapt to the change (and re-apply
clang-format) to fix the clang-tidy build.

llvm-svn: 350916

5 years agoRemember to instantiate explicit template argument lists in a friend
Richard Smith [Fri, 11 Jan 2019 01:59:33 +0000 (01:59 +0000)]
Remember to instantiate explicit template argument lists in a friend
function declaration.

We'd previously often just drop these on the floor, and friend
redeclaration matching would usually (but not always) figure out the
right redeclaration anyway.

Also, don't try to match a dependent friend function template
specialization to a template until instantiation, and don't forget to
reject qualified friend declarations in dependent contexts that don't
name an already-declared entity.

llvm-svn: 350915

5 years ago[AST] Remove ASTContext from getThisType (NFC)
Brian Gesiak [Fri, 11 Jan 2019 01:54:53 +0000 (01:54 +0000)]
[AST] Remove ASTContext from getThisType (NFC)

Summary:
https://reviews.llvm.org/D54862 removed the usages of `ASTContext&` from
within the `CXXMethodDecl::getThisType` method. Remove the parameter
altogether, as well as all usages of it. This does not result in any
functional change because the parameter was unused since
https://reviews.llvm.org/D54862.

Test Plan: check-clang

Reviewers: akyrtzi, mikael

Reviewed By: mikael

Subscribers: mehdi_amini, dexonsmith, cfe-commits

Differential Revision: https://reviews.llvm.org/D56509

llvm-svn: 350914

5 years agoFix header issues.
Richard Trieu [Fri, 11 Jan 2019 01:32:35 +0000 (01:32 +0000)]
Fix header issues.

Several headers would fail to compile if other headers were not previously
included.  The usual issue is that a class is forward declared, but the
full definition is needed.  The requirement for the definition is use of
isa/dyn_cast or calling functions of pointer-packed data types such as
DenseMap or PointerIntPair.  Add missing includes to these headers.

SVals.h required an out-of-line method definition in the .cpp file to avoid
circular inclusion of headers with BasicValueFactory.h

llvm-svn: 350913

5 years agoAdd an explicit initializer to silence a -Wmissing-field-initializers diagnostic...
Aaron Ballman [Fri, 11 Jan 2019 00:13:57 +0000 (00:13 +0000)]
Add an explicit initializer to silence a -Wmissing-field-initializers diagnostic; NFC.

llvm-svn: 350912

5 years ago[WebAssembly] Add unimplemented-simd128 feature, gate builtins
Thomas Lively [Thu, 10 Jan 2019 23:49:00 +0000 (23:49 +0000)]
[WebAssembly] Add unimplemented-simd128 feature, gate builtins

Summary: Depends on D56501. Also adds a macro define
`__wasm_unimplemented_simd128__` for feature detection of
unimplemented SIMD builtins.

Reviewers: aheejin, dschuff

Subscribers: sbc100, jgravelle-google, sunfish, kristina, cfe-commits, rrwinterton
llvm-svn: 350909

5 years ago[llvm-objdump][MachO] Disable some invalid input tests
Francis Visoiu Mistrih [Thu, 10 Jan 2019 23:46:31 +0000 (23:46 +0000)]
[llvm-objdump][MachO] Disable some invalid input tests

It causes some (but not all) bots to fail. I'll look into it tomorrow
morning. Remove the tests for now to make the bots green.

llvm-svn: 350908

5 years ago[analyzer] pr38838, pr39976: Fix crash on diagnosing before implicit destructor.
Artem Dergachev [Thu, 10 Jan 2019 23:44:44 +0000 (23:44 +0000)]
[analyzer] pr38838, pr39976: Fix crash on diagnosing before implicit destructor.

We need to be able to emit the diagnostic at PreImplicitCall,
and the patch implements this functionality.

However, for now the need for emitting such diagnostics is not all that great:
it is only necessary to not crash when emitting a false positive due to an
unrelated issue of having dead symbol collection not working properly.

Coming up with a non-false-positive test seems impossible with the current
set of checkers, though it is likely to be needed for good things as well
in the future.

Differential Revision: https://reviews.llvm.org/D56042

rdar://problem/46911462

llvm-svn: 350907

5 years ago[WebAssembly] Fix stack pointer store check in RegStackify
Heejin Ahn [Thu, 10 Jan 2019 23:12:07 +0000 (23:12 +0000)]
[WebAssembly] Fix stack pointer store check in RegStackify

Summary:
We now use __stack_pointer global and global.get/global.set instruction.
This fixes the checking routine for stack_pointer writes accordingly.

This also fixes the existing __stack_pointer test in reg-stackify.ll:
That test used to pass not because of __stack_pointer clashes but
because the function `stackpointer_callee` was not marked as `readnone`,
so it was assumed to possibly write to memory arbitraily, and
`global.set` instruction was marked as `mayStore` in the .td definition,
so they were identified as intervening writes. After we added `readnone`
to its attribute, this test fails without this patch.

Reviewers: dschuff, sunfish

Subscribers: jgravelle-google, sbc100, llvm-commits

Differential Revision: https://reviews.llvm.org/D56094

llvm-svn: 350906

5 years agogn build: Add a template for calling write_cmake_config.py
Nico Weber [Thu, 10 Jan 2019 23:10:04 +0000 (23:10 +0000)]
gn build: Add a template for calling write_cmake_config.py

No behavior change.

Differential Revision: https://reviews.llvm.org/D56487

llvm-svn: 350905

5 years agogn build: Merge r350852
Nico Weber [Thu, 10 Jan 2019 23:05:39 +0000 (23:05 +0000)]
gn build: Merge r350852

llvm-svn: 350904

5 years ago[MSP430] Minor fixes/improvements for assembler/disassembler
Anton Korobeynikov [Thu, 10 Jan 2019 22:59:50 +0000 (22:59 +0000)]
[MSP430] Minor fixes/improvements for assembler/disassembler

* Teach AsmParser to recognize @rn in distination operand as 0(rn).
* Do not allow Disassembler decoding instructions that have size more
  than a number of input bytes.
* Fix UB in MSP430MCCodeEmitter.

Patch by Kristina Bessonova!

Differential Revision: https://reviews.llvm.org/D56547

llvm-svn: 350903

5 years ago[MSP430] Add missing instruction forms
Anton Korobeynikov [Thu, 10 Jan 2019 22:54:53 +0000 (22:54 +0000)]
[MSP430] Add missing instruction forms

* Add missing mm, [r|m]n, [r|m]p instruction forms.
* Fix bit16mc instruction.

Patch by Kristina Bessonova!

Differential Revision: https://reviews.llvm.org/D56546

llvm-svn: 350902

5 years ago[WebAssembly] Add unimplemented-simd128 subtarget feature
Thomas Lively [Thu, 10 Jan 2019 22:32:11 +0000 (22:32 +0000)]
[WebAssembly] Add unimplemented-simd128 subtarget feature

Summary:
This is a third attempt, but this time we have vetted it on Windows
first. The previous errors were due to an uninitialized class member.

Reviewers: aheejin

Subscribers: dschuff, sbc100, jgravelle-google, sunfish, jfb, llvm-commits

Differential Revision: https://reviews.llvm.org/D56560

llvm-svn: 350901

5 years agoProperly support -shared-libgcc.
Sterling Augustine [Thu, 10 Jan 2019 22:25:58 +0000 (22:25 +0000)]
Properly support -shared-libgcc.
This revision was revied in D55016.

llvm-svn: 350900

5 years ago[llvm-objcopy] [COFF] Fix a test matching pathnames for Windows. NFC.
Martin Storsjo [Thu, 10 Jan 2019 22:05:21 +0000 (22:05 +0000)]
[llvm-objcopy] [COFF] Fix a test matching pathnames for Windows. NFC.

llvm-svn: 350899

5 years ago[llvm-objcopy] [COFF] Fix warnings abuilt missing field initialization. NFC.
Martin Storsjo [Thu, 10 Jan 2019 21:59:41 +0000 (21:59 +0000)]
[llvm-objcopy] [COFF] Fix warnings abuilt missing field initialization. NFC.

llvm-svn: 350898

5 years agogn build: Use "git rev-parse --git-dir" to discover the path to the .git directory.
Peter Collingbourne [Thu, 10 Jan 2019 21:57:07 +0000 (21:57 +0000)]
gn build: Use "git rev-parse --git-dir" to discover the path to the .git directory.

This makes it compatible with worktrees.

Differential Revision: https://reviews.llvm.org/D56422

llvm-svn: 350897

5 years ago[MachineCombiner][NFC] Prevent dereferencing past-the-end object in an MRI container
Gerolf Hoflehner [Thu, 10 Jan 2019 21:53:13 +0000 (21:53 +0000)]
[MachineCombiner][NFC] Prevent dereferencing past-the-end object in an MRI container

llvm-svn: 350896

5 years ago[MemorySSA] Disable checkClobberSanity for SkipSelfWalker.
Alina Sbirlea [Thu, 10 Jan 2019 21:47:15 +0000 (21:47 +0000)]
[MemorySSA] Disable checkClobberSanity for SkipSelfWalker.

Sanity will fail for this, since we're exploring getting a clobber
further than the sanity check expects.
Ideally we need to teach the sanity check to differentiate between the
two walkers based on the SkipSelf bool in the query.

llvm-svn: 350895

5 years agogn build: Merge r350893
Nico Weber [Thu, 10 Jan 2019 21:47:10 +0000 (21:47 +0000)]
gn build: Merge r350893

llvm-svn: 350894

5 years ago[llvm-objcopy] [COFF] Add support for removing symbols
Martin Storsjo [Thu, 10 Jan 2019 21:28:24 +0000 (21:28 +0000)]
[llvm-objcopy] [COFF] Add support for removing symbols

Differential Revision: https://reviews.llvm.org/D55881

llvm-svn: 350893

5 years agoUpdate clang-tools-extra for r350891 from Clang.
Aaron Ballman [Thu, 10 Jan 2019 21:22:28 +0000 (21:22 +0000)]
Update clang-tools-extra for r350891 from Clang.

llvm-svn: 350892

5 years agoCorrect the source range returned from preprocessor callbacks.
Aaron Ballman [Thu, 10 Jan 2019 21:22:13 +0000 (21:22 +0000)]
Correct the source range returned from preprocessor callbacks.

This adjusts the source range passed in to the preprocessor callbacks to only include the condition range itself, rather than all of the conditionally skipped tokens.

llvm-svn: 350891

5 years agoNFC: Change case of identifiers
Stephen Kelly [Thu, 10 Jan 2019 20:58:21 +0000 (20:58 +0000)]
NFC: Change case of identifiers

llvm-svn: 350890

5 years agoChange SymbolFile::ParseTypes to ParseTypesForCompileUnit.
Zachary Turner [Thu, 10 Jan 2019 20:57:50 +0000 (20:57 +0000)]
Change SymbolFile::ParseTypes to ParseTypesForCompileUnit.

The function SymbolFile::ParseTypes previously accepted a SymbolContext.
This makes it extremely difficult to implement faithfully, because you
have to account for all possible combinations of members being set in
the SymbolContext. On the other hand, no clients of this function
actually care about implementing this function to this strict of a
standard. AFAICT, there is actually only 1 client in the entire
codebase, and it is the function ParseAllDebugSymbols, which is itself
only called for testing purposes when dumping information. At this
call-site, the only field it sets is the CompileUnit, meaning that an
implementer of a SymbolFile need not worry about any examining or
handling any other fields which might be set.

By restricting this API to accept exactly a CompileUnit& and nothing
more, we can simplify the life of new SymbolFile plugin implementers by
making it clear exactly what the necessary and sufficient set of
functionality they need to implement is, while at the same time removing
some dead code that tried to handle other types of SymbolContext fields
that were never going to be set anyway.

Differential Revision: https://reviews.llvm.org/D56462

llvm-svn: 350889

5 years ago[NativePDB] Add support for parsing typedef records.
Zachary Turner [Thu, 10 Jan 2019 20:57:32 +0000 (20:57 +0000)]
[NativePDB] Add support for parsing typedef records.

Typedefs are represented as S_UDT records in the globals stream. This
creates a strange situation where "types" are actually represented as
"symbols", so they need special handling.

In order to test this, we don't just use lldb and print out some
variables causing the AST to get created, because variables whose type
is a typedef will have debug info referencing the original type, not the
typedef. So we use lldb-test instead which will parse all debug info in
the entire file. This exposed some problems with lldb-test and the
native reader, mainly that certain types of obscure symbols which we can
find when iterating every single record would trigger crashes. These
have been fixed as well so that lldb-test can be used to test this
functionality.

Differential Revision: https://reviews.llvm.org/D56461

llvm-svn: 350888

5 years ago[Sema] Call CheckPlaceholderExpr to resolve typeof or decltype
Akira Hatanaka [Thu, 10 Jan 2019 20:12:16 +0000 (20:12 +0000)]
[Sema] Call CheckPlaceholderExpr to resolve typeof or decltype
placeholder expressions while an unevaluated context is still on the
expression evaluation context stack.

This prevents recordUseOfWeek from being called when a weak variable is
used as an operand of a decltype or a typeof expression and fixes
spurious -Warc-repeated-use-of-weak warnings.

rdar://problem/45742525

Differential Revision: https://reviews.llvm.org/D55662

llvm-svn: 350887

5 years agoAvoid use-after-free in ~LegacyRTDyldObjectLinkingLayer
Sanjoy Das [Thu, 10 Jan 2019 20:12:09 +0000 (20:12 +0000)]
Avoid use-after-free in ~LegacyRTDyldObjectLinkingLayer

Reviewers: lhames

Subscribers: mcrosier, jlebar, bixia, llvm-commits

Differential Revision: https://reviews.llvm.org/D56521

llvm-svn: 350886

5 years ago[HIP] Use nul instead of /dev/null when running on windows
Yaxun Liu [Thu, 10 Jan 2019 20:09:52 +0000 (20:09 +0000)]
[HIP] Use nul instead of /dev/null when running on windows

When clang is running on windows, /dev/null is not available. Use nul as empty input file instead.

Differential Revision: https://reviews.llvm.org/D56225

llvm-svn: 350885

5 years ago[libcxx] Reorganize tests since the application of P0602R4
Louis Dionne [Thu, 10 Jan 2019 20:06:11 +0000 (20:06 +0000)]
[libcxx] Reorganize tests since the application of P0602R4

Summary:
P0602R4 makes the special member functions of optional and variant
conditionally trivial based on the types in the optional/variant.
We already implemented that, but the tests were organized as if this
were a non-standard extension. This patch reorganizes the tests in a
way that makes more sense since this is not an extension anymore.

Reviewers: EricWF, mpark, mclow.lists

Subscribers: christof, jkorous, dexonsmith, libcxx-commits

Differential Revision: https://reviews.llvm.org/D54772

llvm-svn: 350884

5 years ago[TSan] Remove ignore_interceptors_accesses flag
Julian Lettner [Thu, 10 Jan 2019 20:04:33 +0000 (20:04 +0000)]
[TSan] Remove ignore_interceptors_accesses flag

Summary:
It has been superseded by the `ignore_noninstrumented_modules` flag and is no longer needed.

Also simplify a test that checks that `mmap_interceptor` respects ignore annotations (`thr->ignore_reads_and_writes `).
Relevant: https://reviews.llvm.org/rL269855

<rdar://problem/46263073> Remove obsolete Apple-specific suppression option

Reviewers: dcoughlin, kubamracek, dvyukov, delcypher

Reviewed By: dvyukov

Subscribers: jfb, llvm-commits, #sanitizers

Tags: #sanitizers

Differential Revision: https://reviews.llvm.org/D55075

llvm-svn: 350883

5 years ago[sanitizer_common] Define __sanitizer_FILE on NetBSD
Michal Gorny [Thu, 10 Jan 2019 19:56:57 +0000 (19:56 +0000)]
[sanitizer_common] Define __sanitizer_FILE on NetBSD

Differential Revision: https://reviews.llvm.org/D56109

llvm-svn: 350882

5 years ago[sanitizer_common] Remove support for tirpc/rpc/xdr.h
Michal Gorny [Thu, 10 Jan 2019 19:56:50 +0000 (19:56 +0000)]
[sanitizer_common] Remove support for tirpc/rpc/xdr.h

Remove the partial support for rpc/xdr.h from libtirpc.  Since it is
an entirely external library, we ought to build it sanitized separately
and not attempt to intercept like the libc implementation.  Besides,
the existing code for tirpc support was neither complete nor working.

Noted by @krytarowski.

Differential Revision: https://reviews.llvm.org/D47817

llvm-svn: 350881

5 years ago[GVN] Update BlockRPONumber prior to use.
Matt Davis [Thu, 10 Jan 2019 19:56:03 +0000 (19:56 +0000)]
[GVN] Update BlockRPONumber prior to use.

Summary:
The original patch addressed the use of BlockRPONumber by forcing a sequence point when accessing that map in a conditional.  In short we found cases where that map was being accessed with blocks that had not yet been added to that structure.  For context, I've kept the wall of text below,  to what we are trying to fix, by always ensuring a updated BlockRPONumber.

== Backstory ==

I was investigating an ICE (segfault accessing a DenseMap item).  This failure happened non-deterministically, with no apparent reason and only on a Windows build of LLVM (from October 2018).

After looking into the crashes (multiple core files) and running DynamoRio, the cores and DynamoRio (DR) log pointed to the same code in `GVN::performScalarPRE()`. The values in the map are unsigned integers, the keys are `llvm::BasicBlock*`.  Our test case that triggered this warning and periodic crash is rather involved.  But the problematic line looks to be:

GVN.cpp: Line 2197

```
     if (BlockRPONumber[P] >= BlockRPONumber[CurrentBlock] &&
```

To test things out, I cooked up a patch that accessed the items in the map outside of the condition, by forcing a sequence point between accesses. DynamoRio stopped warning of the issue, and the test didn't seem to crash after 1000+ runs.

My investigation was on an older version of LLVM, (source from October this year). What it looks like was occurring is the following, and the assembly from the latest pull of llvm in December seems to confirm this might still be an issue; however, I have not witnessed the crash on more recent builds. Of course the asm in question is generated from the host compiler on that Windows box (not clang), but it hints that we might want to consider how we access the BlockRPONumber map in this conditional (line 2197, listed above).  In any case, I don't think the host compiler is wrong, rather I think it is pointing out a possibly latent bug in llvm.

1) There is no sequence point for the `>=` operation.

2) A call to a `DenseMapBase::operator[]` can have the side effect of the map reallocating a larger store (more Buckets, via a call to `DenseMap::grow`).

3) It seems perfectly legal for a host compiler to generate assembly that stores the result of a call to `operator[]` on the stack (that's what my host compile of GVN.cpp is doing) .  A second call to `operator[]` //might// encourage the map to 'grow' thus making any pointers to the map's store invalid.  The `>=` compares the first and second values. If the first happens to be a pointer produced from operator[], it could be invalid when dereferenced at the time of comparison.

The assembly generated from the Window's host compiler does show the result of the first access to the map via `operator[]` produces a pointer to an unsigned int.  And that pointer is being stored on  the stack.  If a second call to the map (which does occur) causes the map to grow, that address (on the stack) is now invalid.

Reviewers: t.p.northover, efriedma

Reviewed By: efriedma

Subscribers: efriedma, llvm-commits

Differential Revision: https://reviews.llvm.org/D55974

llvm-svn: 350880

5 years agoUse MemorySSA in LICM to do sinking and hoisting.
Alina Sbirlea [Thu, 10 Jan 2019 19:29:04 +0000 (19:29 +0000)]
Use MemorySSA in LICM to do sinking and hoisting.

Summary:
Step 2 in using MemorySSA in LICM:
Use MemorySSA in LICM to do sinking and hoisting, all under "EnableMSSALoopDependency" flag.
Promotion is disabled.

Enable flag in LICM sink/hoist tests to test correctness of this change. Moved one test which
relied on promotion, in order to test all sinking tests.

Reviewers: sanjoy, davide, gberry, george.burgess.iv

Subscribers: llvm-commits, Prazek

Differential Revision: https://reviews.llvm.org/D40375

llvm-svn: 350879

5 years agofixup: sphinx warning
Nick Desaulniers [Thu, 10 Jan 2019 19:26:35 +0000 (19:26 +0000)]
fixup: sphinx warning

Fixes the sphinx warning:
tools/clang/docs/DiagnosticsReference.rst:7889: WARNING: Title underline
too short.

That I just introduced in r350877.

llvm-svn: 350878

5 years ago[SemaCXX] add -Woverride-init alias to -Winitializer-overrides
Nick Desaulniers [Thu, 10 Jan 2019 19:12:39 +0000 (19:12 +0000)]
[SemaCXX] add -Woverride-init alias to -Winitializer-overrides

Summary:
https://bugs.llvm.org/show_bug.cgi?id=40251
https://github.com/ClangBuiltLinux/linux/issues/307

Reviewers: rsmith

Reviewed By: rsmith

Subscribers: cfe-commits, nathanchance, srhines

Differential Revision: https://reviews.llvm.org/D56522

llvm-svn: 350877

5 years agolldbtest.py: try to fix a runtime exception
Adrian Prantl [Thu, 10 Jan 2019 19:06:46 +0000 (19:06 +0000)]
lldbtest.py: try to fix a runtime exception
found on http://green.lab.llvm.org/green/view/LLDB/job/lldb-cmake-matrix/32/consoleFull#15046896708254eaf0-7326-4999-85b0-388101f2d404

llvm-svn: 350876

5 years ago[X86] Call SimplifyDemandedBits on conditions of X86ISD::SHRUNKBLEND
Craig Topper [Thu, 10 Jan 2019 19:05:34 +0000 (19:05 +0000)]
[X86] Call SimplifyDemandedBits on conditions of X86ISD::SHRUNKBLEND

This extends to combineVSelectToShrunkBlend to be able to resimplify SHRUNKBLENDS that have already been created.

This should help some of the regressions from D56387

Differential Revision: https://reviews.llvm.org/D56421

llvm-svn: 350875

5 years agoFix false positive unsequenced access and modification warning in array subscript...
Nicolas Lesser [Thu, 10 Jan 2019 19:03:33 +0000 (19:03 +0000)]
Fix false positive unsequenced access and modification warning in array subscript expression.

Summary: In the [expr.sub] p1, we can read that for a given E1[E2], E1 is sequenced before E2.

Patch by Mateusz Janek.

Reviewers: rsmith, Rakete1111

Reviewed By: rsmith, Rakete1111

Subscribers: riccibruno, lebedev.ri, Rakete1111, hiraditya, cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D50766

llvm-svn: 350874

5 years ago[X86] Simplify the BRCOND handling for FCMP_UNE.
Craig Topper [Thu, 10 Jan 2019 19:02:14 +0000 (19:02 +0000)]
[X86] Simplify the BRCOND handling for FCMP_UNE.

Despite what the comment says, FCMP_UNE would be an OR not an AND. In the lowering code the first branch created still goes to the original destination. The second branch was exchanged to go to where the subsequent unconditional branch went. This is different than what we do for FCMP_OEQ where both branches that we create go to the original unconditional branch.

As far as I can tell, I think this means we don't need to exchange the branch target with the unconditional branch for FCMP_UNE at all.

Differential Revision: https://reviews.llvm.org/D56309

llvm-svn: 350873

5 years agoFilesystem tests: fix fs.op.relative
JF Bastien [Thu, 10 Jan 2019 18:50:34 +0000 (18:50 +0000)]
Filesystem tests: fix fs.op.relative

Summary: The test wasn't using the testing infrastructure properly.

Reviewers: ldionne, mclow.lists, EricWF

Subscribers: christof, jkorous, dexonsmith, libcxx-commits

Differential Revision: https://reviews.llvm.org/D56519

llvm-svn: 350872

5 years ago[llvm-objdump][MachO] Fix test to work on Windows
Francis Visoiu Mistrih [Thu, 10 Jan 2019 18:32:30 +0000 (18:32 +0000)]
[llvm-objdump][MachO] Fix test to work on Windows

This fails in http://lab.llvm.org:8011/builders/clang-x64-windows-msvc/builds/3208/steps/stage%201%20check/logs/stdio.

llvm-svn: 350871

5 years ago[analyzer] [hotfix] Fix the tests
George Karpenkov [Thu, 10 Jan 2019 18:28:10 +0000 (18:28 +0000)]
[analyzer] [hotfix] Fix the tests

The error must have crept during the cherry-pick.

llvm-svn: 350870

5 years ago[analyzer] Update the category name for RetainCountChecker reports
George Karpenkov [Thu, 10 Jan 2019 18:16:25 +0000 (18:16 +0000)]
[analyzer] Update the category name for RetainCountChecker reports

..now that it includes OSObjects

rdar://46509986

Differential Revision: https://reviews.llvm.org/D56404

llvm-svn: 350869

5 years ago[analyzer] [NFC] [RetainCountChecker] Remove dead unused map
George Karpenkov [Thu, 10 Jan 2019 18:16:10 +0000 (18:16 +0000)]
[analyzer] [NFC] [RetainCountChecker] Remove dead unused map

Differential Revision: https://reviews.llvm.org/D56402

llvm-svn: 350868

5 years ago[analyzer] Quote the type of the leaked/problematic object in diagnostics for readability
George Karpenkov [Thu, 10 Jan 2019 18:15:57 +0000 (18:15 +0000)]
[analyzer] Quote the type of the leaked/problematic object in diagnostics for readability

Differential Revision: https://reviews.llvm.org/D56344

llvm-svn: 350867

5 years ago[analyzer] [NFC] Reverse the argument order for "diff" in tests
George Karpenkov [Thu, 10 Jan 2019 18:15:44 +0000 (18:15 +0000)]
[analyzer] [NFC] Reverse the argument order for "diff" in tests

The current argument order has "expected" and "actual" the wrong way around,
so that the diff shows the change from expected to actual, not from actual to expected.

Namely, if the expected diagnostics contains the string "foo", but the analyzer emits "bar",
we really want to see:

```
- foo
+ bar
```

not

```
- bar
+ foo
```

since adapting to most changes would require applying that diff to the expected output.

Differential Revision: https://reviews.llvm.org/D56340

llvm-svn: 350866

5 years ago[analyzer] [NFC] Reduce redundancy in RetainSummaryManager by using a function
George Karpenkov [Thu, 10 Jan 2019 18:15:30 +0000 (18:15 +0000)]
[analyzer] [NFC] Reduce redundancy in RetainSummaryManager by using a function

Differential Revision: https://reviews.llvm.org/D56282

llvm-svn: 350865

5 years ago[analyzer] [RetainCountChecker] [NFC] Remove SummaryLog
George Karpenkov [Thu, 10 Jan 2019 18:15:17 +0000 (18:15 +0000)]
[analyzer] [RetainCountChecker] [NFC] Remove SummaryLog

The complicated machinery for passing the summary log around is actually
only used for one thing! To figure out whether the "dealloc" message was
sent.

Since I have tried to extend it for other uses and failed (it's actually
very hard to use), I think it's much better to simply use a tag and
remove the summary log altogether.

Differential Revision: https://reviews.llvm.org/D56228

llvm-svn: 350864

5 years ago[analyzer] [RetainCountChecker] [NFC] Another minor cleanup
George Karpenkov [Thu, 10 Jan 2019 18:15:04 +0000 (18:15 +0000)]
[analyzer] [RetainCountChecker] [NFC] Another minor cleanup

Differential Revision: https://reviews.llvm.org/D56224

llvm-svn: 350863

5 years ago[analyzer] [RetainCountChecker] [NFC] Refactor the way attributes are handled
George Karpenkov [Thu, 10 Jan 2019 18:14:51 +0000 (18:14 +0000)]
[analyzer] [RetainCountChecker] [NFC] Refactor the way attributes are handled

Make sure all checks for attributes go through a centralized function,
which checks whether attribute handling is enabled, and performs
validation.  The type of the attribute is returned.

Sadly, metaprogramming is required as attributes have no sensible static
getters.

Differential Revision: https://reviews.llvm.org/D56222

llvm-svn: 350862

5 years ago[analyzer] [RetainCountChecker] Remove redundant enum UnarySummaryKind
George Karpenkov [Thu, 10 Jan 2019 18:14:38 +0000 (18:14 +0000)]
[analyzer] [RetainCountChecker] Remove redundant enum UnarySummaryKind

Differential Revision: https://reviews.llvm.org/D56072

llvm-svn: 350861

5 years ago[analyzer] [RetainCountChecker] Remove obsolete "MakeCollectable" enum value
George Karpenkov [Thu, 10 Jan 2019 18:14:25 +0000 (18:14 +0000)]
[analyzer] [RetainCountChecker] Remove obsolete "MakeCollectable" enum value

Differential Revision: https://reviews.llvm.org/D56071

llvm-svn: 350860

5 years ago[analyzer] [RetainCountChecker] [NFC] Remove redundant enum items *Msg, as the object...
George Karpenkov [Thu, 10 Jan 2019 18:14:12 +0000 (18:14 +0000)]
[analyzer] [RetainCountChecker] [NFC] Remove redundant enum items *Msg, as the object type is already communicated by a separate field

Differential Revision: https://reviews.llvm.org/D56070

llvm-svn: 350859

5 years ago[analyzer] [NFC] Track object type with ArgEffect in RetainCountChecker.
George Karpenkov [Thu, 10 Jan 2019 18:13:59 +0000 (18:13 +0000)]
[analyzer] [NFC] Track object type with ArgEffect in RetainCountChecker.

This would be needed in the future.

https://reviews.llvm.org/D56040

llvm-svn: 350858

5 years ago[analyzer] [NFC] Move ObjKind into a separate top-level enum in RetainSummaryManager.
George Karpenkov [Thu, 10 Jan 2019 18:13:46 +0000 (18:13 +0000)]
[analyzer] [NFC] Move ObjKind into a separate top-level enum in RetainSummaryManager.

Allows using it in future outside of RetEffect.

Differential Revision: https://reviews.llvm.org/D56039

llvm-svn: 350857

5 years agoSplit -Wdelete-non-virtual-dtor into two groups
Erik Pilkington [Thu, 10 Jan 2019 18:03:07 +0000 (18:03 +0000)]
Split -Wdelete-non-virtual-dtor into two groups

This group controls two diagnostics: deleting an abstract class with
a non-virtual dtor, which is a guaranteed crash, and deleting a
non-abstract polymorphic class with a non-virtual dtor, which is just
suspicious.

rdar://40380564

Differential revision: https://reviews.llvm.org/D56405

llvm-svn: 350856

5 years agoRemove an external test file. NFC.
Rui Ueyama [Thu, 10 Jan 2019 17:57:30 +0000 (17:57 +0000)]
Remove an external test file. NFC.

llvm-svn: 350855

5 years ago[lit] Make it possible for the lit test suite to pass with
Dan Liew [Thu, 10 Jan 2019 17:47:44 +0000 (17:47 +0000)]
[lit] Make it possible for the lit test suite to pass with
`FILECHECK_OPTS=-v` set in the environment.

Follow up to r350850 as requested by Joel E. Denny in
https://reviews.llvm.org/D56541 .

llvm-svn: 350854

5 years agoTrapInstr must be 4 bytes long. Currently we write two zeros on every two bytes.
Rui Ueyama [Thu, 10 Jan 2019 17:45:56 +0000 (17:45 +0000)]
TrapInstr must be 4 bytes long. Currently we write two zeros on every two bytes.

llvm-svn: 350853

5 years ago[analyzer][CrossTU][NFC] Generalize to external definitions instead of external functions
Rafael Stahl [Thu, 10 Jan 2019 17:44:04 +0000 (17:44 +0000)]
[analyzer][CrossTU][NFC] Generalize to external definitions instead of external functions

Summary: This is just changing naming and documentation to be general about external definitions that can be imported for cross translation unit analysis. There is at least a plan to add VarDecls: D46421

Reviewers: NoQ, xazax.hun, martong, a.sidorin, george.karpenkov, serge-sans-paille

Reviewed By: xazax.hun, martong

Subscribers: mgorny, whisperity, baloghadamsoftware, szepet, rnkovacs, mikhail.ramalho, Szelethus, donat.nagy, dkrupp, cfe-commits

Differential Revision: https://reviews.llvm.org/D56441

llvm-svn: 350852

5 years ago[llvm-objdump][MachO] Fix error reporting after r350848 and r350849
Francis Visoiu Mistrih [Thu, 10 Jan 2019 17:36:54 +0000 (17:36 +0000)]
[llvm-objdump][MachO] Fix error reporting after r350848 and r350849

llvm-svn: 350851

5 years ago[FileCheck] Don't propagate `FILECHECK_DUMP_INPUT_ON_FAILURE` and
Dan Liew [Thu, 10 Jan 2019 17:24:06 +0000 (17:24 +0000)]
[FileCheck] Don't propagate `FILECHECK_DUMP_INPUT_ON_FAILURE` and
`FILECHECK_OPTS` into environment for FileCheck tests.

Summary:

This fixes the following FileCheck tests:

* FileCheck/dump-input-enable.txt
* FileCheck/match-full-lines.txt

when `FILECHECK_DUMP_INPUT_ON_FAILURE` is set in the environment.

By default llvm-lit propagates `FILECHECK_DUMP_INPUT_ON_FAILURE` and
`FILECHECK_OPTS` from llvm-lit's environment into the test environment.
Unfortunately this can break FileCheck's tests because they expect that
these environment variables not to be set.

rdar://problem/47176262

Reviewers: jdenny, probinson, george.karpenkov

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D56541

llvm-svn: 350850

5 years ago[llvm-objdump][MachO] Use the -dsym file name when reporting errors
Francis Visoiu Mistrih [Thu, 10 Jan 2019 17:16:42 +0000 (17:16 +0000)]
[llvm-objdump][MachO] Use the -dsym file name when reporting errors

Instead of using the binary filename.

llvm-svn: 350849

5 years ago[llvm-objdump][MachO] Correctly handle the llvm::Error when -dsym has errors
Francis Visoiu Mistrih [Thu, 10 Jan 2019 17:16:37 +0000 (17:16 +0000)]
[llvm-objdump][MachO] Correctly handle the llvm::Error when -dsym has errors

In an assert build, the Error gets destroyed and we get "Program aborted
due to an unhandled Error:".

In release, we get an empty message.

llvm-svn: 350848

5 years ago[clangd] Introduce loading of shards within auto-index
Kadir Cetinkaya [Thu, 10 Jan 2019 17:03:04 +0000 (17:03 +0000)]
[clangd] Introduce loading of shards within auto-index

Summary:
Whenever a change happens on a CDB, load shards associated with that
CDB before issuing re-index actions.

Reviewers: ilya-biryukov

Reviewed By: ilya-biryukov

Subscribers: ioeric, MaskRay, jkorous, arphaman, cfe-commits

Differential Revision: https://reviews.llvm.org/D55224

llvm-svn: 350847

5 years ago[Docs] fix typo, adjust text order
Sanjay Patel [Thu, 10 Jan 2019 17:02:55 +0000 (17:02 +0000)]
[Docs] fix typo, adjust text order

llvm-svn: 350846

5 years ago[Docs] add note to avoid 'errno' for better vectorization (PR40265)
Sanjay Patel [Thu, 10 Jan 2019 16:57:28 +0000 (16:57 +0000)]
[Docs] add note to avoid 'errno' for better vectorization (PR40265)

This is a partial fix for the documentation improvements requested in:
https://bugs.llvm.org/show_bug.cgi?id=40265

llvm-svn: 350845

5 years ago[DAGCombiner] simplify code; NFC
Sanjay Patel [Thu, 10 Jan 2019 16:47:42 +0000 (16:47 +0000)]
[DAGCombiner] simplify code; NFC

llvm-svn: 350844

5 years agoRepair compilation of llvm-stress after r350835.
James Y Knight [Thu, 10 Jan 2019 16:43:26 +0000 (16:43 +0000)]
Repair compilation of llvm-stress after r350835.

Apparently it doesn't get built by 'ninja check'. :(

llvm-svn: 350843

5 years ago[LLD][ELF] - Fix tests after r350840.
George Rimar [Thu, 10 Jan 2019 16:26:05 +0000 (16:26 +0000)]
[LLD][ELF] - Fix tests after r350840.

D56076 (r350840) changed the llvm-objdump output.

This is a follow up commit to fix LLD test cases.

llvm-svn: 350842

5 years ago[SelectionDAGBuilder] Refactor GetRegistersForValue. NFCI.
Nirav Dave [Thu, 10 Jan 2019 16:25:47 +0000 (16:25 +0000)]
[SelectionDAGBuilder] Refactor GetRegistersForValue. NFCI.

llvm-svn: 350841

5 years ago[llvm-objdump] - Do not include reserved undefined symbol in -t output.
George Rimar [Thu, 10 Jan 2019 16:24:10 +0000 (16:24 +0000)]
[llvm-objdump] - Do not include reserved undefined symbol in -t output.

This is https://bugs.llvm.org/show_bug.cgi?id=26892,

GNU objdump hides the special symbol entry:

SYMBOL TABLE:
000000000000a7e0 l     F .text 00000000000003f9 bi_copymodules
while llvm-objdump does not:

SYMBOL TABLE:
0000000000000000         *UND*  00000000
000000000000a7e0 l     F .text  000003f9 bi_copymodules

Patch makes the behavior of the llvm-objdump to be consistent with the GNU objdump.

Differential revision: https://reviews.llvm.org/D56076

llvm-svn: 350840

5 years ago[SelectionDAGBuilder] Fix formatting. NFC.
Nirav Dave [Thu, 10 Jan 2019 16:22:19 +0000 (16:22 +0000)]
[SelectionDAGBuilder] Fix formatting. NFC.

llvm-svn: 350839

5 years ago[AMDGPU] Fix dwordx3/southern-islands failures.
Neil Henning [Thu, 10 Jan 2019 16:21:08 +0000 (16:21 +0000)]
[AMDGPU] Fix dwordx3/southern-islands failures.

This commit fixes the dwordx3/southern-islands failures that were found
in bugzilla https://bugs.llvm.org/show_bug.cgi?id=40129, by not
generating the dwordx3 variants of load/store instructions that were
added to the ISA after southern islands.

Differential Revision: https://reviews.llvm.org/D56434

llvm-svn: 350838

5 years ago[SelectionDAGBuilder] Refactor visitInlineAsm. NFC.
Nirav Dave [Thu, 10 Jan 2019 16:18:18 +0000 (16:18 +0000)]
[SelectionDAGBuilder] Refactor visitInlineAsm. NFC.

llvm-svn: 350837

5 years ago[ELF] Fix ARM and Thumb V7PILongThunk overflow behavior.
Peter Smith [Thu, 10 Jan 2019 16:08:23 +0000 (16:08 +0000)]
[ELF] Fix ARM and Thumb V7PILongThunk overflow behavior.

When the range between the source and target of a V7PILongThunk exceeded an
int32 we would trigger a relocation out of range error for the
R_ARM_MOVT_PREL or R_ARM_THM_MOVT_PREL relocation. This case can happen when
linking the linux kernel as it is loaded above 0xf0000000.

There are two parts to the fix.
- Remove the overflow check for R_ARM_MOVT_PREL or R_ARM_THM_MOVT_PREL. The
ELF for the ARM Architecture document defines these relocations as having no
overflow checking so the check was spurious.
- Use int64_t for the offset calculation, in line with similar thunks so
that PC + (S - P) < 32-bits. This results in less surprising disassembly.

Differential Revision: https://reviews.llvm.org/D56396

llvm-svn: 350836

5 years ago[opaque pointer types] Remove some calls to generic Type subtype accessors.
James Y Knight [Thu, 10 Jan 2019 16:07:20 +0000 (16:07 +0000)]
[opaque pointer types] Remove some calls to generic Type subtype accessors.

That is, remove many of the calls to Type::getNumContainedTypes(),
Type::subtypes(), and Type::getContainedType(N).

I'm not intending to remove these accessors -- they are
useful/necessary in some cases. However, removing the pointee type
from pointers would potentially break some uses, and reducing the
number of calls makes it easier to audit.

llvm-svn: 350835

5 years agoFix compilation error on 32-bit architectures introduced in r350511
Pavel Labath [Thu, 10 Jan 2019 15:53:20 +0000 (15:53 +0000)]
Fix compilation error on 32-bit architectures introduced in r350511

The issue was a narrowing conversion when converting from uint64_t to a
size_t.

llvm-svn: 350834

5 years ago[LLD][ELF] - A follow up for r350819 ("Support MSP430") : add a test case missing.
George Rimar [Thu, 10 Jan 2019 15:34:33 +0000 (15:34 +0000)]
[LLD][ELF] - A follow up for r350819 ("Support MSP430") : add a test case missing.

It got lost for some reason.

llvm-svn: 350833

5 years ago[llvm-symbolizer] Add -p as alias to -pretty-print
Dmitry Venikov [Thu, 10 Jan 2019 15:33:35 +0000 (15:33 +0000)]
[llvm-symbolizer] Add -p as alias to -pretty-print

Summary: Provides -p as a short alias for -pretty-print. Motivation: https://bugs.llvm.org/show_bug.cgi?id=40076

Reviewers: samsonov, khemant, ruiu, rnk, fjricci, jhenderson

Reviewed By: jhenderson

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D56542

llvm-svn: 350832

5 years ago[RISCV][MC] Add support for evaluating constant symbols as immediates
Alex Bradbury [Thu, 10 Jan 2019 15:33:17 +0000 (15:33 +0000)]
[RISCV][MC] Add support for evaluating constant symbols as immediates

This further improves compatibility with GNU as, allowing input such as the
following to be assembled:

.equ CONST, 0x123456
li a0, CONST
addi a0, a0, %lo(CONST)

.equ CONST, 1
slli a0, a0, CONST

Note that we don't have perfect compatibility with gas, as it will avoid
emitting a relocation in this case:

addi a0, a0, %lo(CONST2)
.equ CONST2, 0x123456

Thanks to Shiva Chen for suggesting a better way to approach this during review.

Differential Revision: https://reviews.llvm.org/D52298

llvm-svn: 350831

5 years ago[x86] fix remaining miscompile bug in horizontal binop matching (PR40243)
Sanjay Patel [Thu, 10 Jan 2019 15:27:23 +0000 (15:27 +0000)]
[x86] fix remaining miscompile bug in horizontal binop matching (PR40243)

When we use the partial-matching function on a 128-bit chunk, we must
account for the possibility that we've matched undef halves of the
original source vectors, so the outputs may need to be reset.

This should allow closing PR40243:
https://bugs.llvm.org/show_bug.cgi?id=40243

llvm-svn: 350830

5 years agogn build: Merge r350819
Nico Weber [Thu, 10 Jan 2019 15:16:32 +0000 (15:16 +0000)]
gn build: Merge r350819

llvm-svn: 350829

5 years agoModify InputSectionBase::getLocation to add section and offset to every loc.
Sean Fertile [Thu, 10 Jan 2019 15:08:06 +0000 (15:08 +0000)]
Modify InputSectionBase::getLocation to add section and offset to every loc.

The section and offset can be very helpful in diagnosing certian errors.
For example on a relocation overflow or misalignment diagnostic:

test.c:(function  foo): relocation R_PPC64_ADDR16_DS out of range: ...

The function foo can have many R_PPC64_ADDR16_DS relocations. Adding the offset
and section will identify exactly which relocation is causing the failure.

Differential Revision: https://reviews.llvm.org/D56453

llvm-svn: 350828