Johannes Doerfert [Thu, 17 Nov 2016 21:41:08 +0000 (21:41 +0000)]
[DBG] Collect statistics about taken assumptions
llvm-svn: 287261
Rafael Espindola [Thu, 17 Nov 2016 21:20:16 +0000 (21:20 +0000)]
Allow use define symbols to override linker defined ones.
I hit an internal linker script that was defining _DYNAMIC instead of
letting the linker do it. It turns out that both bfd and gold allow
that.
This is pretty easy to implement, just make the linker defined symbol
weak. This should have no impact in the case where there is no user
defined symbol: The visibility is hidden, which causes the output to
still be local.
llvm-svn: 287260
Zachary Turner [Thu, 17 Nov 2016 21:15:14 +0000 (21:15 +0000)]
Convert Platform, Process, and Connection functions to StringRef.
All tests pass on Linux and Windows.
llvm-svn: 287259
Malcolm Parsons [Thu, 17 Nov 2016 21:00:09 +0000 (21:00 +0000)]
Fixes for r287241. Use placement new. Apply clang-format.
llvm-svn: 287258
Rui Ueyama [Thu, 17 Nov 2016 20:27:10 +0000 (20:27 +0000)]
Allow SIZEOF() command on nonexistent section.
Linker script doesn't create a section if it has no content. So the following
script doesn't create .norelocs section if it doesn't have any .rel* sections.
.norelocs : { *(.rel*) }
Later, if you assert that the size of .norelocs is 0, LLD printed out
an error message, because it didn't allow calling SIZEOF() on nonexistent
sections.
This patch allows SIZEOF() on nonexistent sections, so that you can do
something like this.
ASSERT(SIZEOF(.norelocs), "shouldn't contain .rel sections!")
Note that this behavior is compatible with GNU.
Differential Revision: https://reviews.llvm.org/D26810
llvm-svn: 287257
Petr Hosek [Thu, 17 Nov 2016 20:22:49 +0000 (20:22 +0000)]
[CMake] Error when LTO and lld are enabled on Darwin
lld on Darwin does not currently support LTO.
Differential Revision: https://reviews.llvm.org/D26715
llvm-svn: 287256
Eric Fiselier [Thu, 17 Nov 2016 20:08:43 +0000 (20:08 +0000)]
Workaround compilers w/o C++1z inline variables
llvm-svn: 287255
Rui Ueyama [Thu, 17 Nov 2016 19:57:47 +0000 (19:57 +0000)]
Add single quotes to error messages.
llvm-svn: 287254
Rui Ueyama [Thu, 17 Nov 2016 19:57:45 +0000 (19:57 +0000)]
Use consistent variable name.
llvm-svn: 287253
Rui Ueyama [Thu, 17 Nov 2016 19:57:43 +0000 (19:57 +0000)]
Use llvm::reverse to get a reverse range.
llvm-svn: 287252
Eric Fiselier [Thu, 17 Nov 2016 19:24:34 +0000 (19:24 +0000)]
Remove files missed in r287250
llvm-svn: 287251
Eric Fiselier [Thu, 17 Nov 2016 19:24:04 +0000 (19:24 +0000)]
Implement P0504R0: Revisiting in-place tag types for any/optional/variant
llvm-svn: 287250
Eric Fiselier [Thu, 17 Nov 2016 19:23:35 +0000 (19:23 +0000)]
Test changes for P0504R0 "Revisiting in-place tag types for any/optional/variant". Patch from Casey Carter
llvm-svn: 287249
Simon Pilgrim [Thu, 17 Nov 2016 19:21:20 +0000 (19:21 +0000)]
Fix spelling mistakes in Hexagon target comments. NFC.
Identified by Pedro Giffuni in PR27636.
llvm-svn: 287248
Simon Pilgrim [Thu, 17 Nov 2016 19:03:05 +0000 (19:03 +0000)]
Fix spelling mistakes in X86 target comments. NFC.
Identified by Pedro Giffuni in PR27636.
llvm-svn: 287247
Reid Kleckner [Thu, 17 Nov 2016 19:02:53 +0000 (19:02 +0000)]
[asan] Create a .ASAN$G(A-Z) section for global registration
Summary:
The expectation is that new instrumented code will add global variable
metadata to the .ASAN$GL section, and we will use this new code to
iterate over it.
This technique seems to break when using incremental linking, which
seems to align every global to a 256 byte boundary. Presumably this is
so that it can incrementally cope with global changing size. Clang
already passes -incremental:no as a linker flag when you invoke it to do
the link step.
The two tests added for this feature will fail until the LLVM
instrumentation change in D26770 lands, so they are marked XFAIL for
now.
Reviewers: pcc, kcc, mehdi_amini, kubabrecka
Subscribers: llvm-commits, mgorny
Differential Revision: https://reviews.llvm.org/D26771
llvm-svn: 287246
Michal Gorny [Thu, 17 Nov 2016 18:41:54 +0000 (18:41 +0000)]
[tests] Use __SSE2__ (rather than __i686__...) for SSE2 ASAN test
Use the __SSE2__ to determine whether SSE2 is enabled in the ASAN tests
rather than relying on either of the __i686__ and __x86_64__. The former
is only set with explicit -march=i686, and therefore misses most of
the x86 CPUs that support SSE2. __SSE2__ is in turn defined if
the current settings (-march, -msse2) indicate that SSE2 is supported
which should be more reliable.
Differential Revision: https://reviews.llvm.org/D26763
llvm-svn: 287245
Manman Ren [Thu, 17 Nov 2016 18:41:18 +0000 (18:41 +0000)]
ObjC Module: try to make objc module deterministic.
Make sure that comparing selectors in DeclarationName does its job.
rdar://problem/
28988750
llvm-svn: 287244
Eugene Zelenko [Thu, 17 Nov 2016 18:11:21 +0000 (18:11 +0000)]
[CodeView] Fix some Clang-tidy modernize-use-default, modernize-use-override and Include What You Use warnings; other minor fixes (NFC).
Per Zachary Turner and Mehdi Amini suggestion to make only post-commit reviews.
llvm-svn: 287243
Zachary Turner [Thu, 17 Nov 2016 18:08:12 +0000 (18:08 +0000)]
Rewrite all Property related functions in terms of StringRef.
This was a bit tricky, especially for things like
OptionValueArray and OptionValueDictionary since they do some
funky string parsing. Rather than try to re-write line-by-line
I tried to make the StringRef usage idiomatic, even though
it meant often re-writing from scratch large blocks of code
in a different way while keeping true to the original intent.
The finished code is a big improvement though, and often much
shorter than the original code. All tests and unit tests
pass on Windows and Linux.
llvm-svn: 287242
Malcolm Parsons [Thu, 17 Nov 2016 17:52:58 +0000 (17:52 +0000)]
Use unique_ptr for cached tokens for default arguments in C++.
Summary:
This changes pointers to cached tokens for default arguments in C++ from raw pointers to unique_ptrs. There was a fixme in the code where the cached tokens are created about using a smart pointer.
The change is straightforward, though I did have to track down and fix a memory corruption caused by the change. memcpy was being used to copy parameter information. This duplicated the unique_ptr, which led to the cached token buffer being deleted prematurely.
Patch by David Tarditi!
Reviewers: malcolm.parsons
Subscribers: arphaman, malcolm.parsons, cfe-commits
Differential Revision: https://reviews.llvm.org/D26435
llvm-svn: 287241
Kostya Serebryany [Thu, 17 Nov 2016 17:31:54 +0000 (17:31 +0000)]
[libFuzzer] better documentation for -fsanitize-coverage=trace-cmp
llvm-svn: 287240
Saleem Abdulrasool [Thu, 17 Nov 2016 17:10:54 +0000 (17:10 +0000)]
Sema: correct typo correction for ivars in @implementation
The previous typo correction handling assumed that ivars are only declared in
the interface declaration rather than as a private ivar in the implementation.
Adjust the handling to permit both interfaces. Assert earlier that the
interface has been acquired to ensure that we can identify when both possible
casts have failed.
Addresses PR31040!
llvm-svn: 287238
Rui Ueyama [Thu, 17 Nov 2016 17:06:51 +0000 (17:06 +0000)]
Enable -threads by default.
LLD supports multi-threading, and it seems to be working well as
you can see in r287140. In short, LLD runs a few percent to 30%
faster with -threads and more than 50% faster if you are using
-build-id (your mileage may vary depending on your computer).
However, I don't think most users even don't know about that because
-threads is not a default option.
This patch enables it by default.
Discussion thread:
http://lists.llvm.org/pipermail/llvm-dev/2016-November/107160.html
llvm-svn: 287237
Anna Zaks [Thu, 17 Nov 2016 16:56:11 +0000 (16:56 +0000)]
[asan] Update the tests as Mach-O metadata liveness tracking is on by default
This commit reverts r274579.
llvm-svn: 287236
Anna Zaks [Thu, 17 Nov 2016 16:55:40 +0000 (16:55 +0000)]
[asan] Turn on Mach-O global metadata liveness tracking by default
This patch turns on the metadata liveness tracking since all known issues
have been resolved. The future has been implemented in
https://reviews.llvm.org/D16737 and enables support of dead code stripping
option on Mach-O platforms.
As part of enabling the feature, I also plan on reverting the following
patch to compiler-rt:
http://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-
20160704/369910.html
Differential Revision: https://reviews.llvm.org/D26772
llvm-svn: 287235
Rui Ueyama [Thu, 17 Nov 2016 16:48:53 +0000 (16:48 +0000)]
Pass StringRefs instead of StringMatcher because it's simpler.
llvm-svn: 287234
Konstantin Zhuravlyov [Thu, 17 Nov 2016 16:41:49 +0000 (16:41 +0000)]
Revert "AMDGPU: Enable ConstrainCopy DAG mutation"
This reverts commit r287146.
This breaks few conformance tests.
llvm-svn: 287233
Daniil Fukalov [Thu, 17 Nov 2016 16:07:52 +0000 (16:07 +0000)]
[SCEV] limit recursion depth of CompareSCEVComplexity
Summary:
CompareSCEVComplexity goes too deep (50+ on a quite a big unrolled loop) and runs almost infinite time.
Added cache of "equal" SCEV pairs to earlier cutoff of further estimation. Recursion depth limit was also introduced as a parameter.
Reviewers: sanjoy
Subscribers: mzolotukhin, tstellarAMD, llvm-commits
Differential Revision: https://reviews.llvm.org/D26389
llvm-svn: 287232
Rafael Espindola [Thu, 17 Nov 2016 15:29:11 +0000 (15:29 +0000)]
Simplify. NFC.
llvm-svn: 287231
Benjamin Kramer [Thu, 17 Nov 2016 15:23:06 +0000 (15:23 +0000)]
[include-fixer] Add a test for the full round trip through libclang and the plugin.
llvm-svn: 287230
Benjamin Kramer [Thu, 17 Nov 2016 15:22:36 +0000 (15:22 +0000)]
Link include-fixer into libclang if clang-tools-extra is checked out.
include-fixer only slightly bloats the size of libclang, but since
libclang has no explicit plugin mechanism it's the only way of getting
this to work. Clang-tidy is already there and so far there weren't many
complaints ;)
This is designed to be easy to remove again if libclang ever grows
proper plugin support.
llvm-svn: 287229
Benjamin Kramer [Thu, 17 Nov 2016 15:16:05 +0000 (15:16 +0000)]
[include-fixer] Refactor include fixer to be usable as a plugin
- Refactor the external sema source into a visible class
- Add support for emitting FixIts
- Wrap up include fixer as a plugin as I did with clang-tidy
Test case will follow as soon as I wire this up in libclang.
Differential Revision: https://reviews.llvm.org/D26752
llvm-svn: 287228
Alexey Bataev [Thu, 17 Nov 2016 15:12:05 +0000 (15:12 +0000)]
[OPENMP] Fixed codegen for 'omp cancel' construct.
If 'omp cancel' construct is used in a worksharing construct it may
cause hanging of the software in case if reduction clause is used. Patch fixes this problem by avoiding extra reduction processing for branches that were canceled.
llvm-svn: 287227
Malcolm Parsons [Thu, 17 Nov 2016 14:26:45 +0000 (14:26 +0000)]
[docs] Remove doubled spaces
Reviewers: aaron.ballman
Subscribers: nemanjai, cfe-commits
Differential Revision: https://reviews.llvm.org/D26798
llvm-svn: 287226
Rafael Espindola [Thu, 17 Nov 2016 14:18:08 +0000 (14:18 +0000)]
Avoid accessing an end() iterator.
llvm-svn: 287225
Simon Pilgrim [Thu, 17 Nov 2016 12:21:45 +0000 (12:21 +0000)]
Wdocumentation fix
llvm-svn: 287224
Simon Pilgrim [Thu, 17 Nov 2016 12:14:49 +0000 (12:14 +0000)]
[X86][SSE] Improve lowering of vXi64 multiply with known zero 32-bit halves
vXi64 multiplication is lowered into 3 calls of vpmuludq with the upper/lower 32-bit halves.
If any of these halves are zero then we can remove individual calls. Although there was isBuildVectorAllZeros code to do this I don't think it ever worked (maybe just for constant folded cases that don't seem to be tested for any longer).
This requires additional X86ISD support for computeKnownBitsForTargetNode, so far I've just added support for X86ISD::VZEXT (VPMOVZX* - helping the AVX2+ cases).
Partial fix for PR30845
Differential Revision: https://reviews.llvm.org/D26590
llvm-svn: 287223
Simon Pilgrim [Thu, 17 Nov 2016 12:03:05 +0000 (12:03 +0000)]
Fix spelling in comment. NFC.
llvm-svn: 287222
Malcolm Parsons [Thu, 17 Nov 2016 11:40:02 +0000 (11:40 +0000)]
[clang-tidy] Ignore template instantiations in modernize-use-equals-delete check
Summary: Template instantiations were causing misplaced fixits.
Reviewers: aaron.ballman, alexfh, hokein
Subscribers: hokein, cfe-commits
Differential Revision: https://reviews.llvm.org/D26751
llvm-svn: 287221
Pavel Labath [Thu, 17 Nov 2016 11:22:23 +0000 (11:22 +0000)]
[cmake] Move LLVM_BUILD_STATIC check to an earlier point
Summary:
The motivation for this is to enable correct detection of dlopen() on Android.
Android does not provide a static version of libdl, so if we add the -static flag
after performing the check, it will succeed even though subsequent link steps
will fail. With this change we correctly detect the absence of libdl in a
LLVM_BUILD_STATIC build on Android.
The link itself still does not succeed because the code does not check the result
of this check properly, but I plan to fix that in a separate change.
Reviewers: beanz
Subscribers: danalbert, mgorny, srhines, tberghammer, llvm-commits
Differential Revision: https://reviews.llvm.org/D26463
llvm-svn: 287220
Pablo Barrio [Thu, 17 Nov 2016 10:56:58 +0000 (10:56 +0000)]
[ARM] Relax restriction on variadic functions for tailcall optimization
Summary:
Variadic functions can be treated in the same way as normal functions
with respect to the number and types of parameters.
Reviewers: grosbach, olista01, t.p.northover, rengolin
Subscribers: javed.absar, aemerson, llvm-commits
Differential Revision: https://reviews.llvm.org/D26748
llvm-svn: 287219
Eugene Leviant [Thu, 17 Nov 2016 10:34:05 +0000 (10:34 +0000)]
[ELF] Remove unneeded forward declarations. NFC.
llvm-svn: 287218
Oren Ben Simhon [Thu, 17 Nov 2016 09:59:40 +0000 (09:59 +0000)]
[X86] RegCall - Handling v64i1 in 32/64 bit target
Register Calling Convention defines a new behavior for v64i1 types.
This type should be saved in GPR.
However for 32 bit machine we need to split the value into 2 GPRs (because each is 32 bit).
Differential Revision: https://reviews.llvm.org/D26181
llvm-svn: 287217
Eugene Leviant [Thu, 17 Nov 2016 09:16:34 +0000 (09:16 +0000)]
[ELF] Convert SymbolTableSection to input section
Differential revision: https://reviews.llvm.org/D26740
llvm-svn: 287216
Malcolm Parsons [Thu, 17 Nov 2016 09:14:04 +0000 (09:14 +0000)]
[clang-tidy] Changes to modernize-use-default check
Summary:
Warn about special member functions that only contain a comment.
Report the location of the special member function, unless it is
defined in a macro. Reporting the location of the body in a macro is
more helpful as it causes the macro expansion location to be reported too.
Fixes PR30920.
Reviewers: alexfh, aaron.ballman
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D26741
llvm-svn: 287215
Sanjoy Das [Thu, 17 Nov 2016 07:29:43 +0000 (07:29 +0000)]
Delete dead code and add asserts instead; NFC
llvm-svn: 287214
Sanjoy Das [Thu, 17 Nov 2016 07:29:40 +0000 (07:29 +0000)]
[ImplicitNullCheck] Fix an edge case where we were hoisting incorrectly
ImplicitNullCheck keeps track of one instruction that the memory
operation depends on that it also hoists with the memory operation.
When hoisting this dependency, it would sometimes clobber a live-in
value to the basic block we were hoisting the two things out of. Fix
this by explicitly looking for such dependencies.
I also noticed two redundant checks on `MO.isDef()` in IsMIOperandSafe.
They're redundant since register MachineOperands are either Defs or Uses
-- there is no third kind. I'll change the checks to asserts in a later
commit.
llvm-svn: 287213
Zachary Turner [Thu, 17 Nov 2016 06:13:54 +0000 (06:13 +0000)]
Fix one more build error with lldb-server.
llvm-svn: 287212
Craig Topper [Thu, 17 Nov 2016 05:59:55 +0000 (05:59 +0000)]
[X86] Fix formatting. NFC
llvm-svn: 287211
Craig Topper [Thu, 17 Nov 2016 05:37:39 +0000 (05:37 +0000)]
[X86] Add a test case where, due to a bug in selectScalarSSELoad, we fold the same load twice.
llvm-svn: 287210
Dean Michael Berris [Thu, 17 Nov 2016 05:15:37 +0000 (05:15 +0000)]
[XRay] Support AArch64 in LLVM
This patch adds XRay support in LLVM for AArch64 targets.
This patch is one of a series:
Clang: https://reviews.llvm.org/D26415
compiler-rt: https://reviews.llvm.org/D26413
Author: rSerge
Reviewers: rengolin, dberris
Subscribers: amehsan, aemerson, llvm-commits, iid_iunknown
Differential Revision: https://reviews.llvm.org/D26412
llvm-svn: 287209
Zachary Turner [Thu, 17 Nov 2016 05:14:32 +0000 (05:14 +0000)]
Fix warnings and errors introduced with UUID changes.
llvm-svn: 287208
Chris Bieneman [Thu, 17 Nov 2016 04:36:59 +0000 (04:36 +0000)]
[CMake] [Darwin] Add support for debugging tablegen dependencies
This patch adds an option to the build system LLVM_DEPENDENCY_DEBUGGING. Over time I plan to extend this to do more complex verifications, but the initial patch causes compile errors wherever there is missing a dependency on intrinsics_gen.
Because intrinsics_gen is a compile-time dependency not a link-time dependency, everything that relies on the headers generated in intrinsics_gen needs an explicit dependency.
llvm-svn: 287207
Chris Bieneman [Thu, 17 Nov 2016 04:36:50 +0000 (04:36 +0000)]
[CMake] NFC. Updating CMake dependency specifications
This patch updates a bunch of places where add_dependencies was being explicitly called to add dependencies on intrinsics_gen to instead use the DEPENDS named parameter. This cleanup is needed for a patch I'm working on to add a dependency debugging mode to the build system.
llvm-svn: 287206
Chris Bieneman [Thu, 17 Nov 2016 04:36:35 +0000 (04:36 +0000)]
[CMake] NFC. Updating CMake dependency specifications
This patch updates a couple places where add_dependencies was being explicitly called to add dependencies on intrinsics_gen to instead use the DEPENDS named parameter. This cleanup is needed for a patch I'm working on to add a dependency debugging mode to the build system.
llvm-svn: 287205
Konstantin Zhuravlyov [Thu, 17 Nov 2016 04:33:20 +0000 (04:33 +0000)]
[AMDGPU] Add missing test for rL287203
llvm-svn: 287204
Konstantin Zhuravlyov [Thu, 17 Nov 2016 04:28:37 +0000 (04:28 +0000)]
[AMDGPU] Custom lower f16 = fp_round f64
llvm-svn: 287203
Rui Ueyama [Thu, 17 Nov 2016 04:10:09 +0000 (04:10 +0000)]
Rename a function so that that starts with a lowercase letter.
llvm-svn: 287202
Konstantin Zhuravlyov [Thu, 17 Nov 2016 04:00:46 +0000 (04:00 +0000)]
[AMDGPU] Promote f16/i16 conversions to f32/i32
llvm-svn: 287201
Rui Ueyama [Thu, 17 Nov 2016 03:52:14 +0000 (03:52 +0000)]
Simplify and use consistent variable name. NFC.
llvm-svn: 287200
Konstantin Zhuravlyov [Thu, 17 Nov 2016 03:49:01 +0000 (03:49 +0000)]
[AMDGPU] Expand `br_cc` for f16
Differential Revision: https://reviews.llvm.org/D26732
llvm-svn: 287199
Rui Ueyama [Thu, 17 Nov 2016 03:39:21 +0000 (03:39 +0000)]
Use uint16_t instead of size_t for symbol version ID.
Because it is uint16_t in the ELF spec. Using size_t was confusing.
llvm-svn: 287198
Rui Ueyama [Thu, 17 Nov 2016 03:32:41 +0000 (03:32 +0000)]
Add single quotes to a warning message for consistency.
llvm-svn: 287197
Rui Ueyama [Thu, 17 Nov 2016 03:19:34 +0000 (03:19 +0000)]
Simplify handleAnonymousVersion even more.
We used to create a vector contantaining all version definitions
with wildcards because doing that was efficient. All patterns were
compiled to a regexp and matched against symbol names. Because
a regexp can be converted to a DFA, matching against union of patterns
is as cheap as matching against one patter.
We are no longer converting them to regexp. Our own glob pattern
handler doesn't do such optimization. Therefore, creating a vector
no longer makes sense.
llvm-svn: 287196
Lang Hames [Thu, 17 Nov 2016 02:33:47 +0000 (02:33 +0000)]
[Orc] Clang-format the recent RPC update (r286620 and related).
llvm-svn: 287195
Richard Smith [Thu, 17 Nov 2016 02:16:09 +0000 (02:16 +0000)]
Remove -Wsigned-enum-bitfield from -Wmost. On a wide set of ABIs, this warning
is completely irrelevant, producing (effectively) false positives, and -Wmost
is used pretty widely. We should somehow turn it back on by default when
targeting the MS ABI, however, since it indicates the program will not do as
intended in those cases.
(Or perhaps we should just treat enum bitfields as having the signedness of the
enum, even when targeting the MS ABI...)
llvm-svn: 287193
Rui Ueyama [Thu, 17 Nov 2016 02:16:06 +0000 (02:16 +0000)]
Simplify. NFC.
llvm-svn: 287192
Rui Ueyama [Thu, 17 Nov 2016 02:09:42 +0000 (02:09 +0000)]
Split scanVersionScript. NFC.
llvm-svn: 287191
Zachary Turner [Thu, 17 Nov 2016 01:38:02 +0000 (01:38 +0000)]
Convert UriParser to use StringRef.
llvm-svn: 287190
Zachary Turner [Thu, 17 Nov 2016 01:37:52 +0000 (01:37 +0000)]
Make GetValueForVariableExpression use StringRef.
Also significantly reduced the indentation level by use of
early returns, and simplified some of the logic by using
StringRef functions such as consumeInteger() and getAsInteger()
instead of strtoll, etc.
llvm-svn: 287189
Zachary Turner [Thu, 17 Nov 2016 01:37:42 +0000 (01:37 +0000)]
Convert AutoComplete related code to StringRef.
Differential Revision: https://reviews.llvm.org/D26721
llvm-svn: 287188
George Burgess IV [Thu, 17 Nov 2016 01:33:54 +0000 (01:33 +0000)]
[Sema] Fix a bug in enable_if condition instantiation.
During template instantiation, we currently fall back to just calling
Sema::SubstExpr for enable_if attributes that aren't value-dependent or
type-dependent. Since Sema::SubstExpr strips off any implicit casts
we've added to an expression, it's possible that this behavior will
leave us with an enable_if condition that's just a DeclRefExpr.
Conditions like that deeply confuse Sema::CheckEnableIf.
llvm-svn: 287187
Dehao Chen [Thu, 17 Nov 2016 01:17:02 +0000 (01:17 +0000)]
Use profile info to adjust loop unroll threshold.
Summary:
For flat loop, even if it is hot, it is not a good idea to unroll in runtime, thus we set a lower partial unroll threshold.
For hot loop, we set a higher unroll threshold and allows expensive tripcount computation to allow more aggressive unrolling.
Reviewers: davidxl, mzolotukhin
Subscribers: sanjoy, mehdi_amini, llvm-commits
Differential Revision: https://reviews.llvm.org/D26527
llvm-svn: 287186
Ivan Krasin [Thu, 17 Nov 2016 01:09:04 +0000 (01:09 +0000)]
Explicitly specify that ubsan-vtable-checks is x86-64.
This should fix a failure on PowerPC introduced by r287181.
llvm-svn: 287185
Justin Lebar [Thu, 17 Nov 2016 01:03:42 +0000 (01:03 +0000)]
[CUDA] Update docs to indicate that clang now supports std::complex in CUDA mode.
The last remaining necessary change was D25403, landed as r287012.
llvm-svn: 287184
Lang Hames [Thu, 17 Nov 2016 01:02:52 +0000 (01:02 +0000)]
Remove a stale test case.
llvm-svn: 287183
Peter Collingbourne [Thu, 17 Nov 2016 00:42:08 +0000 (00:42 +0000)]
llvm-dis: Remove dead code.
llvm-svn: 287182
Ivan Krasin [Thu, 17 Nov 2016 00:39:48 +0000 (00:39 +0000)]
Insert a type check before reading vtable.
Summary:
this is to prevent a situation when a pointer is invalid or null,
but we get to reading from vtable before we can check that
(possibly causing a segfault without a good diagnostics).
Reviewers: pcc
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D26559
llvm-svn: 287181
Dylan McKay [Thu, 17 Nov 2016 00:09:38 +0000 (00:09 +0000)]
[AVR] Remove some accidentally-commited code that broke the bots
This is a remnant of an on-chip unit testing tool that has since been
moved out-of-tree.
It was accidentally committed in r287162.
llvm-svn: 287180
Petr Hosek [Wed, 16 Nov 2016 23:59:06 +0000 (23:59 +0000)]
[CMake] Support lld with LTO bootstrap
lld has LTO support, if requested we should add a dependency on lld
rather than LLVMgold when doing LTO bootstrap build.
Differential Revision: https://reviews.llvm.org/D26649
llvm-svn: 287179
Peter Collingbourne [Wed, 16 Nov 2016 23:40:26 +0000 (23:40 +0000)]
Introduce GlobalSplit pass.
This pass splits globals into elements using inrange annotations on
getelementptr indices.
Differential Revision: https://reviews.llvm.org/D22295
llvm-svn: 287178
Reid Kleckner [Wed, 16 Nov 2016 23:40:00 +0000 (23:40 +0000)]
Add warning when assigning enums to bitfields without an explicit unsigned underlying type
Summary:
Add a warning when assigning enums to bitfields without an explicit
unsigned underlying type. This is to prevent problems with MSVC
compatibility, since the Microsoft ABI defaults to storing enums with a
signed type, causing inconsistencies with saving to/reading from
bitfields.
Also disabled the warning in the dr0xx.cpp test which throws the error,
and added a test for the warning.
The warning can be disabled with -Wno-signed-enum-bitfield.
Patch by Sasha Bermeister!
Reviewers: rnk, aaron.ballman
Subscribers: mehdi_amini, aaron.ballman, cfe-commits, thakis, dcheng
Differential Revision: https://reviews.llvm.org/D24289
llvm-svn: 287177
Dylan McKay [Wed, 16 Nov 2016 23:06:14 +0000 (23:06 +0000)]
[AVR] Wrap all methods in the pseudo expansion pass in an anon namespace
The '-fpermissive' compiler flag complains if the template
specializations used in the class are used in a different namespace.
llvm-svn: 287176
Anna Zaks [Wed, 16 Nov 2016 22:59:01 +0000 (22:59 +0000)]
[analyzer] Remove unused check::RegionChanges::wantsRegionChangeUpdate callback
Remove the check::RegionChanges::wantsRegionChangeUpdate callback as it is no
longer used (since checkPointerEscape has been added).
A patch by Krzysztof Wiśniewski!
Differential Revision: https://reviews.llvm.org/D26759
llvm-svn: 287175
Dylan McKay [Wed, 16 Nov 2016 22:48:38 +0000 (22:48 +0000)]
[AVR] Fix basic block naming in ctlz and cttz tests
The branch selector would change the names.
llvm-svn: 287174
Dylan McKay [Wed, 16 Nov 2016 22:48:30 +0000 (22:48 +0000)]
[AVR] Remove unused method from AVRTargetMachine
llvm-svn: 287173
Dylan McKay [Wed, 16 Nov 2016 22:38:43 +0000 (22:38 +0000)]
[AVR] Add tests for counting leading/trailing zeros
This adds two test files that verify the 'cttz' and 'ctlz' operations.
llvm-svn: 287172
Sanjay Patel [Wed, 16 Nov 2016 22:34:05 +0000 (22:34 +0000)]
[x86] allow FP-logic ops when one operand is FP and result is FP
We save an inter-register file move this way. If there's any CPU where
the FP logic is slower, we could transform this back to int-logic in
MachineCombiner.
This helps, but doesn't solve, PR6137:
https://llvm.org/bugs/show_bug.cgi?id=6137
The 'andn' test shows that we're missing a pattern match to
recognize the xor with -1 constant as a 'not' op.
llvm-svn: 287171
Ahmed Bougacha [Wed, 16 Nov 2016 22:25:05 +0000 (22:25 +0000)]
[AsmParser] Avoid recursing when lexing ';'. NFC.
This should prevent stack overflows in non-optimized builds on
.ll files with lots of consecutive commented-out lines.
Instead of recursing into LexToken(), continue into a 'while (true)'.
llvm-svn: 287170
Ahmed Bougacha [Wed, 16 Nov 2016 22:25:03 +0000 (22:25 +0000)]
[CodeGen] Pass references, not pointers, to MMI helpers. NFC.
While there, rename them to follow the coding style.
llvm-svn: 287169
Ahmed Bougacha [Wed, 16 Nov 2016 22:24:59 +0000 (22:24 +0000)]
Revert "Get GlobalISel to build on Linux after r286407"
This reverts commit r286962.
We want to avoid depending on SelectionDAG, and AddLandingPadInfo
lives in CodeGen now.
llvm-svn: 287168
Ahmed Bougacha [Wed, 16 Nov 2016 22:24:56 +0000 (22:24 +0000)]
[CodeGen] Pull MMI helpers from FunctionLoweringInfo to MMI. NFC.
They're not SelectionDAG- or FunctionLoweringInfo-specific. They
are, however, specific to building MMI from IR.
We could make them members, but it's nice having MMI be a "simple" data
structure and this logic kept separate.
This also lets us reuse them from GlobalISel.
llvm-svn: 287167
Ahmed Bougacha [Wed, 16 Nov 2016 22:24:53 +0000 (22:24 +0000)]
[CodeGen] Cleanup MachineModuleInfo doxygen comments. NFC.
Remove redundant names and only keep header comments.
llvm-svn: 287166
Ahmed Bougacha [Wed, 16 Nov 2016 22:24:46 +0000 (22:24 +0000)]
[CodeGen] Sort MMI forward declarations. NFC.
llvm-svn: 287165
Shoaib Meenai [Wed, 16 Nov 2016 22:18:10 +0000 (22:18 +0000)]
[libc++] Introduce `_LIBCPP_OVERRIDABLE_FUNC_VIS`
This is a generalization of `_LIBCPP_NEW_DELETE_VIS`; the new macro name
captures the semantics better, and also allows us to get rid of the
`_WIN32` check in `include/new`. No functional change.
Differential Revision: https://reviews.llvm.org/D26702
llvm-svn: 287164
Kevin Enderby [Wed, 16 Nov 2016 22:17:38 +0000 (22:17 +0000)]
General clean up of error handling in llvm-objdump to remove its use of report_fatal_error().
No real functional change with this commit.
The problem with report_fatal_error() is it does not include the tool name
and the file name the for which the error message was generated.
Uses of report_fatal_error() were change to report_error() or error()
to get a better error and to make the code smaller and cleaner.
Also changed things like error(errorToErrorCode(SOrErr.takeError())) to
use report_error() with a file name and the llvm::Error (as well as the
ArchitectureName if available) so the error message is printed.
llvm-svn: 287163
Dylan McKay [Wed, 16 Nov 2016 21:58:04 +0000 (21:58 +0000)]
[AVR] Add the pseudo instruction expansion pass
Summary:
A lot of the pseudo instructions are required because LLVM assumes that
all integers of the same size as the pointer size are legal. This means
that it will not currently expand 16-bit instructions to their 8-bit
variants because it thinks 16-bit types are legal for the operations.
This also adds all of the CodeGen tests that required the pass to run.
Reviewers: arsenm, kparzysz
Subscribers: wdng, mgorny, modocache, llvm-commits
Differential Revision: https://reviews.llvm.org/D26577
llvm-svn: 287162
Vitaly Buka [Wed, 16 Nov 2016 21:51:39 +0000 (21:51 +0000)]
Fix "isn't a prototype" warning
llvm-svn: 287161
Peter Collingbourne [Wed, 16 Nov 2016 21:48:59 +0000 (21:48 +0000)]
X86: Simplify X86ISD::Wrapper operand checks. NFCI.
We only ever create TargetConstantPool, TargetJumpTable, TargetExternalSymbol,
TargetGlobalAddress, TargetGlobalTLSAddress, MCSymbol and TargetBlockAddress
nodes as operands of X86ISD::Wrapper nodes, so we can remove one check and
invert the other.
Also update the documentation comment for X86ISD::Wrapper.
Differential Revision: https://reviews.llvm.org/D26731
llvm-svn: 287160