Matt Arsenault [Thu, 17 Apr 2014 17:06:37 +0000 (17:06 +0000)]
R600/SI: f64 frint is legal on CI
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206475
91177308-0d34-0410-b5e6-
96231b3b80d8
Chad Rosier [Thu, 17 Apr 2014 16:19:54 +0000 (16:19 +0000)]
[AArch64] Implement the getCSRFirstUseCost API, mirroring that in ARM64.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206473
91177308-0d34-0410-b5e6-
96231b3b80d8
NAKAMURA Takumi [Thu, 17 Apr 2014 12:22:14 +0000 (12:22 +0000)]
Inliner::OptimizationRemark: Fix crash in clang/test/Frontend/optimization-remark.c on some hosts, including --vg.
DebugLoc in Callsite would not live after Inliner. It should be copied before Inliner.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206459
91177308-0d34-0410-b5e6-
96231b3b80d8
Chandler Carruth [Thu, 17 Apr 2014 09:41:54 +0000 (09:41 +0000)]
[LCG] Remove a dead declaration. This stopped being used when I switched
to a more normal move operation on the graph itself. The definition
already got removed, but I missed the declaration.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206455
91177308-0d34-0410-b5e6-
96231b3b80d8
Chandler Carruth [Thu, 17 Apr 2014 09:40:13 +0000 (09:40 +0000)]
[LCG] Move the call graph node class into the graph class's definition.
This will become necessary to build up the SCC iterators and SCC
definitions. Moving it now so that subsequent diffs are incremental.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206454
91177308-0d34-0410-b5e6-
96231b3b80d8
Chandler Carruth [Thu, 17 Apr 2014 09:07:50 +0000 (09:07 +0000)]
Make the User::value_op_iterator a random access iterator. I had written
this code ages ago and lost track of it. Seems worth doing though --
this thing can get called from places that would benefit from knowing
that std::distance is O(1). Also add a very fledgeling unittest for
Users and make sure various aspects of this seem to work reasonably.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206453
91177308-0d34-0410-b5e6-
96231b3b80d8
Chandler Carruth [Thu, 17 Apr 2014 07:25:59 +0000 (07:25 +0000)]
[LCG] Just move the allocator (now that we can) when moving a call
graph. This simplifies the custom move constructor operation to one of
walking the graph and updating the 'up' pointers to point to the new
location of the graph. Switch the nodes from a reference to a pointer
for the 'up' edge to facilitate this.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206450
91177308-0d34-0410-b5e6-
96231b3b80d8
Chandler Carruth [Thu, 17 Apr 2014 07:22:19 +0000 (07:22 +0000)]
[LCG] Remove the Module reference member which we weren't using for
anything and doesn't make sense if assigning.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206449
91177308-0d34-0410-b5e6-
96231b3b80d8
Chandler Carruth [Thu, 17 Apr 2014 07:08:56 +0000 (07:08 +0000)]
[Allocator] Make SpecificBumpPtrAllocator also movable and move
assignable.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206448
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Thu, 17 Apr 2014 06:33:45 +0000 (06:33 +0000)]
[X86] Add disassembler support for the 0x0f 0x7f form of movq %mm, %mm.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206447
91177308-0d34-0410-b5e6-
96231b3b80d8
Saleem Abdulrasool [Thu, 17 Apr 2014 06:17:23 +0000 (06:17 +0000)]
objdump: identify WoA WinCOFF/ARM correctly
Since LLVM currently only supports WinCOFF, assume that the input is WinCOFF
rather than another type of COFF file (ECOFF/XCOFF). If the architecture is
detected as thumb (e.g. the file has a IMAGE_FILE_MACHINE_ARMNT magic) then use
a triple of thumbv7-windows.
This allows for objdump to properly handle WoA object files without having to
specify the target triple manually.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206446
91177308-0d34-0410-b5e6-
96231b3b80d8
Saleem Abdulrasool [Thu, 17 Apr 2014 06:17:20 +0000 (06:17 +0000)]
MC: rework static_assert to be MSVC compatible
Visual Studio does not permit referencing a structure member as a static field
for sizeof calculations. Resort to a pointer cast which is compatible across
Visual Studio and other compilers.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206445
91177308-0d34-0410-b5e6-
96231b3b80d8
Justin Bogner [Thu, 17 Apr 2014 02:16:53 +0000 (02:16 +0000)]
Support: Move OnDiskHashTable from clang to llvm
This introduces clang's Basic/OnDiskHashTable.h into llvm as
Support/OnDiskHashTable.h. I've taken the opportunity to add doxygen
comments and run the file through clang-format, but other than the
namespace changing from clang:: to llvm:: the API is identical.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206438
91177308-0d34-0410-b5e6-
96231b3b80d8
Matt Arsenault [Thu, 17 Apr 2014 02:03:08 +0000 (02:03 +0000)]
R600/SI: Fix zext from i1 to i64
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206437
91177308-0d34-0410-b5e6-
96231b3b80d8
Adam Nemet [Thu, 17 Apr 2014 01:01:37 +0000 (01:01 +0000)]
[ARM64] Fix "Cannot select" for vector ctpop
The commit of r205855:
Author: Arnold Schwaighofer <aschwaighofer@apple.com>
Date: Wed Apr 9 14:20:47 2014 +0000
SLPVectorizer: Only vectorize intrinsics whose operands are widened equally
The vectorizer only knows how to vectorize intrinics by widening all operands by
the same factor.
Patch by Tyler Nowicki!
exposed a backend bug causing a regression (Cannot select ctpop).
The commit msg is a bit confusing because the patch actually changes the
behavior for the loop-vectorizer as well. As things got refactored into a
helper ctpop got snuck in to the trivially-vectorizable helper which is now
used by both vectorizers. In other words, we started seeing vector-ctpops in
the backend.
This change makes ctpop LegalizeAction::Expand for the types not supported by
the byte-only CNT instruction. We may be able to custom-lower these later to
a single CNT but this is to fix the compiler crash first.
Fixes <rdar://problem/
16578951>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206433
91177308-0d34-0410-b5e6-
96231b3b80d8
Gerolf Hoflehner [Thu, 17 Apr 2014 00:21:52 +0000 (00:21 +0000)]
Inline a function when the always_inline attribute
is set even when it contains a indirect branch.
The attribute overrules correctness concerns
like the escape of a local block address.
This is for rdar://
16501761
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206429
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Wed, 16 Apr 2014 23:15:31 +0000 (23:15 +0000)]
Teach LLVMConfigVersion.cmake to behave as find_package() expects.
Patch by Brad King
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206426
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Wed, 16 Apr 2014 23:15:28 +0000 (23:15 +0000)]
Add support for a patch version to the cmake system.
Patch by Brad King
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206425
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Wed, 16 Apr 2014 22:38:02 +0000 (22:38 +0000)]
[c++11] Tidy up AsmPrinter.cpp.
Range'ify loops and tidy up some by-reference handling. No functional
change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206422
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Wed, 16 Apr 2014 22:37:58 +0000 (22:37 +0000)]
iterator_range for machine block terminators.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206421
91177308-0d34-0410-b5e6-
96231b3b80d8
Tom Stellard [Wed, 16 Apr 2014 17:45:04 +0000 (17:45 +0000)]
Added new functionality to LLVM C API to use DiagnosticInfo to handle errors
Patch by: Darren Powell
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206407
91177308-0d34-0410-b5e6-
96231b3b80d8
Aaron Ballman [Wed, 16 Apr 2014 17:09:20 +0000 (17:09 +0000)]
Replacing a non-ASCII character in a comment with an ASCII character. Fixes a C4819 warning in MSVC.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206403
91177308-0d34-0410-b5e6-
96231b3b80d8
Diego Novillo [Wed, 16 Apr 2014 16:53:41 +0000 (16:53 +0000)]
Allow diagnostic handlers to check for optimization remarks.
Summary:
When optimization remarks are enabled via the driver flag -Rpass, we
should allow the FE diagnostic handler to check if the given pass name
needs a diagnostic.
We were unconditionally checking the pattern defined in opt's
-pass-remarks flag. This was causing the FE to not emit any diagnostics.
Reviewers: qcolombet
CC: llvm-commits
Differential Revision: http://reviews.llvm.org/D3362
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206400
91177308-0d34-0410-b5e6-
96231b3b80d8
Konrad Anheim [Wed, 16 Apr 2014 16:45:18 +0000 (16:45 +0000)]
Test commit - Added a new line
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206399
91177308-0d34-0410-b5e6-
96231b3b80d8
Matheus Almeida [Wed, 16 Apr 2014 16:28:59 +0000 (16:28 +0000)]
[mips] Use TwoOperandAliasConstraint for shift instructions.
This enables TableGen to generate an additional two operand
matcher for our shift_rotate_imm and shift_rotate_reg class of instructions.
The tests were also updated so that they include now encoding information
for all affected instructions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206398
91177308-0d34-0410-b5e6-
96231b3b80d8
Matheus Almeida [Wed, 16 Apr 2014 15:48:55 +0000 (15:48 +0000)]
[mips] Add initial support for NaN2008 in the back-end.
This is so that EF_MIPS_NAN2008 is set if we are using IEEE 754-2008
NaN encoding (-mnan=2008). This patch also adds support for parsing
'.nan legacy' and '.nan 2008' assembly directives. The handling of
these directives should match GAS' behaviour i.e., the last directive
in use sets the ELF header bit (EF_MIPS_NAN2008).
Differential Revision: http://reviews.llvm.org/D3346
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206396
91177308-0d34-0410-b5e6-
96231b3b80d8
Tim Northover [Wed, 16 Apr 2014 15:28:06 +0000 (15:28 +0000)]
ARM64: silence sign-comparison warning.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206393
91177308-0d34-0410-b5e6-
96231b3b80d8
Tim Northover [Wed, 16 Apr 2014 15:28:02 +0000 (15:28 +0000)]
AArch64/ARM64: port some NEON tests to ARM64
These ones used completely different sets of intrinsics, so the only way to do
it is create a separate ARM64 copy and change them all.
Other than that, CodeGen was straightforward, no deficiencies detected here.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206392
91177308-0d34-0410-b5e6-
96231b3b80d8
Tim Northover [Wed, 16 Apr 2014 15:27:52 +0000 (15:27 +0000)]
AArch64/ARM64: produce correct relocation for conditional branches.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206391
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Sanders [Wed, 16 Apr 2014 14:38:27 +0000 (14:38 +0000)]
[mips] Indentation
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206389
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Sanders [Wed, 16 Apr 2014 13:58:57 +0000 (13:58 +0000)]
[mips] Fix emission of '.option pic0' for MIPS-IV.
Summary: This was a case of incorrect usage of hasMips64() vs isABI_N64()
Reviewers: matheusalmeida, dsanders
Reviewed By: dsanders
Differential Revision: http://reviews.llvm.org/D3398
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206388
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Sanders [Wed, 16 Apr 2014 12:29:08 +0000 (12:29 +0000)]
[mips] Correct r206370 to account for non-Linux targets using the small data section.
This should fix the ninja-x64-msvc-RA-centos6 builder.
I suspect the check in MipsSubtarget.cpp is incorrect and is really trying to
check for a bare-metal target rather and anything other than linux. I'll
investigate this.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206385
91177308-0d34-0410-b5e6-
96231b3b80d8
Kostya Serebryany [Wed, 16 Apr 2014 12:12:19 +0000 (12:12 +0000)]
[asan] add two new hidden compile-time flags for asan: asan-instrumentation-with-call-threshold and asan-memory-access-callback-prefix. This is part of the workaround for PR17409 (instrument huge functions with callbacks instead of inlined code). These flags will also help us experiment with kasan (kernel-asan) and clang
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206383
91177308-0d34-0410-b5e6-
96231b3b80d8
Tim Northover [Wed, 16 Apr 2014 12:03:56 +0000 (12:03 +0000)]
ARM64: specify triple so that Linux tests pass
Now that Linux is trying to reparse all inline asm it chokes on the different
comment character in this test.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206382
91177308-0d34-0410-b5e6-
96231b3b80d8
Tim Northover [Wed, 16 Apr 2014 11:53:07 +0000 (11:53 +0000)]
AArch64/ARM64: add another set of tests from AArch64
Another batch with no code changes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206381
91177308-0d34-0410-b5e6-
96231b3b80d8
Tim Northover [Wed, 16 Apr 2014 11:52:55 +0000 (11:52 +0000)]
AArch64/ARM64: port across stub handling for ELF C++ exceptions.
The most important part here is that we should actuall emit the stubs we refer
to in the exception table, but as a side issue this uses more sensible & GCC
compatible representations for some of the bits of information.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206380
91177308-0d34-0410-b5e6-
96231b3b80d8
Tim Northover [Wed, 16 Apr 2014 11:52:51 +0000 (11:52 +0000)]
ARM64: use 32-bit moves for constants where possible.
If we know that a particular 64-bit constant has all high bits zero, then we
can rely on the fact that 32-bit ARM64 instructions automatically zero out the
high bits of an x-register. This gives the expansion logic less constraints to
satisfy and so sometimes allows it to pick better sequences.
Came up while porting test/CodeGen/AArch64/movw-consts.ll: this will allow a
32-bit MOVN to be used in @test8 soon.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206379
91177308-0d34-0410-b5e6-
96231b3b80d8
Tim Northover [Wed, 16 Apr 2014 11:52:40 +0000 (11:52 +0000)]
ARM64: use the integrated assembler on ELF.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206378
91177308-0d34-0410-b5e6-
96231b3b80d8
Matheus Almeida [Wed, 16 Apr 2014 11:46:59 +0000 (11:46 +0000)]
[mips] Emit '.set nomicromips' before a function's entry label
if not in micromips mode.
The test (elf_st_other.ll) was renamed as the name and description didn't
make sense as the test wasn't checking any symbol table entry.
Differential Revision: http://reviews.llvm.org/D3346
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206377
91177308-0d34-0410-b5e6-
96231b3b80d8
Aaron Ballman [Wed, 16 Apr 2014 11:15:57 +0000 (11:15 +0000)]
Fixing a compile error in debug versions of MSVC. It seems that the range-based for loop is confused by the DEBUG macro expansion unless a compound statement is used.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206376
91177308-0d34-0410-b5e6-
96231b3b80d8
Chandler Carruth [Wed, 16 Apr 2014 11:14:28 +0000 (11:14 +0000)]
[LCG] Stop playing fast and loose with reference members and assignment.
It doesn't work. I'm still cleaning up all the places where I blindly
followed this pattern. There are more to come in this code too.
As a benefit, this lets the default copy and move operations Just Work.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206375
91177308-0d34-0410-b5e6-
96231b3b80d8
Chandler Carruth [Wed, 16 Apr 2014 10:48:27 +0000 (10:48 +0000)]
[Allocator] Make BumpPtrAllocator movable and move assignable.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206372
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Sanders [Wed, 16 Apr 2014 10:23:37 +0000 (10:23 +0000)]
[mips] Correct callee saved list for the N32 ABI and enable test
Summary: Depends on D3339
Reviewers: matheusalmeida, vmedic
Reviewed By: matheusalmeida
Differential Revision: http://reviews.llvm.org/D3340
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206371
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Sanders [Wed, 16 Apr 2014 09:59:46 +0000 (09:59 +0000)]
[mips] Add calling convention tests covering O32, N32, and N64.
Summary:
I had difficulty finding tests for the N32 and N64 ABI so I've added a
collection of calling convention tests based on the document MIPS ABIs
Described (MD00305), the MIPSpro N32 Handbook, and the SYSV ABI. Where the
documents/implementations disagree, I've used GCC to resolve the conflict.
A few interesting details:
* For N32, LLVM uses 64-bit pointers when saving $ra despite pointers being
32-bit. I've yet to find a supporting statement in the ABI documentation but
the current behaviour matches GCC.
* For O32, the non-variable portion of a varargs argument list is also subject
to the rule that floating-point is passed via GPR's (on N32/N64 only the
variable portion is subject to this rule). This agrees with GCC's behaviour
and the SYSV ABI but contradicts part of the MIPSpro N32 Handbook which talks about O32's behaviour.
* The N32 implementation has the wrong callee-saved register list.
(I already have a fix for this but will commit it as a follow-up).
I've left RUN-TODO lines in for O32 on MIPS64. I don't plan to support this case
for now but we should revisit it.
Reviewers: matheusalmeida, vmedic
Reviewed By: matheusalmeida
Differential Revision: http://reviews.llvm.org/D3339
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206370
91177308-0d34-0410-b5e6-
96231b3b80d8
Chandler Carruth [Wed, 16 Apr 2014 09:21:29 +0000 (09:21 +0000)]
[Allocator] Nuke to useless functions. The implicit ones are sufficient
here (obviously).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206369
91177308-0d34-0410-b5e6-
96231b3b80d8
Tim Northover [Wed, 16 Apr 2014 09:13:44 +0000 (09:13 +0000)]
ARM64: explicitly ask for Apple NEON syntax so test passes on Linux
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206368
91177308-0d34-0410-b5e6-
96231b3b80d8
Tim Northover [Wed, 16 Apr 2014 09:03:25 +0000 (09:03 +0000)]
ARM64: mark x7 as used when an i128 gets shunted onto the stack.
The second half of a split i128 was ending up in x7, which is not a good thing.
This is another part of PR19432.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206366
91177308-0d34-0410-b5e6-
96231b3b80d8
Tim Northover [Wed, 16 Apr 2014 09:03:09 +0000 (09:03 +0000)]
DAGCombiner: don't optimise non-existant litpool load
This particular DAG combine is designed to kick in when both ConstantFPs will
end up being loaded via a litpool, however those nodes have a semi-legal
status, dictated by isFPImmLegal so in some cases there wouldn't have been a
litpool in the first place. Don't try to be clever in those circumstances.
Picked up while merging some AArch64 tests.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206365
91177308-0d34-0410-b5e6-
96231b3b80d8
Timur Iskhodzhanov [Wed, 16 Apr 2014 08:30:32 +0000 (08:30 +0000)]
Simplify a static_assert so VS2013 can build it
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206363
91177308-0d34-0410-b5e6-
96231b3b80d8
Saleem Abdulrasool [Wed, 16 Apr 2014 06:22:53 +0000 (06:22 +0000)]
COFF: fix an off by one error
Adjust the tests to validate the number of auxiliary entries used to store the
filename.
Thanks to majnemer's sharp eye for catching the missing - 1 in the round up
calculation.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206359
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Wed, 16 Apr 2014 06:10:51 +0000 (06:10 +0000)]
Convert SelectionDAG::getVTList to use ArrayRef
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206357
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Wed, 16 Apr 2014 04:21:27 +0000 (04:21 +0000)]
[C++11] More 'nullptr' conversion. In some cases just using a boolean check instead of comparing to nullptr.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206356
91177308-0d34-0410-b5e6-
96231b3b80d8
Saleem Abdulrasool [Wed, 16 Apr 2014 04:15:32 +0000 (04:15 +0000)]
COFF: add support for .file symbols
Add support for emitting .file records. This is mostly a quality of
implementation change (more complete support for COFF file emission) that was
noticed while working on COFF file emission for Windows on ARM.
A .file record is emitted as a symbol with storage class FILE (103) and the name
".file". A series of auxiliary format 4 records follow which contain the file
name. The filename is stored as an ANSI string and is padded with NULL if the
length is not a multiple of COFF::SymbolSize (18).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206355
91177308-0d34-0410-b5e6-
96231b3b80d8
Saleem Abdulrasool [Wed, 16 Apr 2014 04:15:29 +0000 (04:15 +0000)]
tools: fix invalid printing, buffer overrun in llvm-readobj
All auxiliary records are consumed when accessing a File record.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206354
91177308-0d34-0410-b5e6-
96231b3b80d8
Saleem Abdulrasool [Wed, 16 Apr 2014 04:15:25 +0000 (04:15 +0000)]
Target: whitespace
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206353
91177308-0d34-0410-b5e6-
96231b3b80d8
Matt Arsenault [Wed, 16 Apr 2014 01:41:34 +0000 (01:41 +0000)]
R600: Extend r600 sign_extend_inreg tests for EG
Patch by: Jan Vesely <jan.vesely@rutgers.edu>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206349
91177308-0d34-0410-b5e6-
96231b3b80d8
Matt Arsenault [Wed, 16 Apr 2014 01:41:30 +0000 (01:41 +0000)]
R600: Expand sign extension of vectors.
Setting vector types to expand will result in scalarization on pre SI hw,
as those gpus don't have vector shifts either.
Expand also i32 vectors, this helps llvm make the correct decision
about scalarizing the vector ops.
v2: move setOperation() calls to R600ISelLowering.cpp.
cleanup the SI code to make it obvious that this patch does is nop for SI
Patch by: Jan Vesely <jan.vesely@rutgers.edu>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206348
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Wed, 16 Apr 2014 00:42:46 +0000 (00:42 +0000)]
[ARM64,C++11] Tidy up branch relaxation a bit w/ c++11.
No functional change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206344
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Wed, 16 Apr 2014 00:42:43 +0000 (00:42 +0000)]
ARM64: Nuke some dead code.
Missed in previous commit.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206343
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Tue, 15 Apr 2014 22:57:02 +0000 (22:57 +0000)]
[ARM64,C++11] Clean up the ARM64 LOH collection pass.
Range'ify a bunch of loops, mainly. As a result, we have a variety
of objects via reference rather than by pointer, so propogate that
through the various helper functions where it makes sense.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206337
91177308-0d34-0410-b5e6-
96231b3b80d8
Matt Arsenault [Tue, 15 Apr 2014 22:40:47 +0000 (22:40 +0000)]
R600/SI: Print code size along with used registers
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206336
91177308-0d34-0410-b5e6-
96231b3b80d8
Matt Arsenault [Tue, 15 Apr 2014 22:32:49 +0000 (22:32 +0000)]
R600/SI: Print more immediates in hex format
Print in decimal for inline immediates, and hex otherwise. Use hex
always for offsets in addressing offsets.
This approximately matches what the shader compiler does.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206335
91177308-0d34-0410-b5e6-
96231b3b80d8
Matt Arsenault [Tue, 15 Apr 2014 22:32:42 +0000 (22:32 +0000)]
R600/SI: Cleanup parsing of register names.
Try to figure out the class and number of subregisters.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206334
91177308-0d34-0410-b5e6-
96231b3b80d8
Nick Lewycky [Tue, 15 Apr 2014 22:29:32 +0000 (22:29 +0000)]
Make this test not match its own filename, when being run from a path that includes the string 'add'.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206331
91177308-0d34-0410-b5e6-
96231b3b80d8
Matt Arsenault [Tue, 15 Apr 2014 22:28:39 +0000 (22:28 +0000)]
R600/SI: Fix loads of i1
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206330
91177308-0d34-0410-b5e6-
96231b3b80d8
Tobias Grosser [Tue, 15 Apr 2014 22:09:36 +0000 (22:09 +0000)]
RegionInfo: Do not access a value that was just moved away
This fixes a regression introduced in r206310.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206328
91177308-0d34-0410-b5e6-
96231b3b80d8
Chandler Carruth [Tue, 15 Apr 2014 21:51:14 +0000 (21:51 +0000)]
[Allocator] Fold the two templated overloads into a single one with
a default argument. The allocator interface we're modeling doesn't
distinguish between array and non-array allocation.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206327
91177308-0d34-0410-b5e6-
96231b3b80d8
Chandler Carruth [Tue, 15 Apr 2014 21:36:02 +0000 (21:36 +0000)]
[Allocator] Remove a really problematic overload. This is very confusing
because there is another (size_t, size_t) overload of Allocator, and the
only distinguishing factor is that one is a tempalte and the other
isn't. There was only one usage of this and that one was easily
converted to carry the alignment constraint in the type itself.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206325
91177308-0d34-0410-b5e6-
96231b3b80d8
Akira Hatanaka [Tue, 15 Apr 2014 21:30:06 +0000 (21:30 +0000)]
Make FastISel::SelectInstruction return before target specific fast-isel code
handles Intrinsic::trap if TargetOptions::TrapFuncName is set.
This fixes a bug in which the trap function was not taken into consideration
when a program was compiled without optimization (at -O0).
<rdar://problem/
16291933>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206323
91177308-0d34-0410-b5e6-
96231b3b80d8
Andrea Di Biagio [Tue, 15 Apr 2014 19:30:48 +0000 (19:30 +0000)]
[X86] Improve the lowering of packed shifts by constant build_vector.
This patch teaches the backend how to efficiently lower logical and
arithmetic packed shifts on both SSE and AVX/AVX2 machines.
When possible, instead of scalarizing a vector shift, the backend should try
to expand the shift into a sequence of two packed shifts by immedate count
followed by a MOVSS/MOVSD.
Example
(v4i32 (srl A, (build_vector < X, Y, Y, Y>)))
Can be rewritten as:
(v4i32 (MOVSS (srl A, <Y,Y,Y,Y>), (srl A, <X,X,X,X>)))
[with X and Y ConstantInt]
The advantage is that the two new shifts from the example would be lowered into
X86ISD::VSRLI nodes. This is always cheaper than scalarizing the vector into
four scalar shifts plus four pairs of vector insert/extract.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206316
91177308-0d34-0410-b5e6-
96231b3b80d8
Quentin Colombet [Tue, 15 Apr 2014 19:08:46 +0000 (19:08 +0000)]
[ARM64] Set default CPU to generic instead of cyclone.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206313
91177308-0d34-0410-b5e6-
96231b3b80d8
Robert Lougher [Tue, 15 Apr 2014 18:34:24 +0000 (18:34 +0000)]
Revert r191049/r191059 as it can produce wrong code (see PR17975).
It has already been reverted on the 3.4 branch in r196521.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206311
91177308-0d34-0410-b5e6-
96231b3b80d8
David Blaikie [Tue, 15 Apr 2014 18:32:43 +0000 (18:32 +0000)]
Use unique_ptr to manage ownership of child Regions within llvm::Region
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206310
91177308-0d34-0410-b5e6-
96231b3b80d8
Julien Lerouge [Tue, 15 Apr 2014 18:06:46 +0000 (18:06 +0000)]
Add lifetime markers for allocas created to hold byval arguments, make them
appear in the InlineFunctionInfo.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206308
91177308-0d34-0410-b5e6-
96231b3b80d8
Julien Lerouge [Tue, 15 Apr 2014 18:01:54 +0000 (18:01 +0000)]
Split byval argument initialization so the memcpy(s) are injected at the
beginning of the first new block after inlining.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206307
91177308-0d34-0410-b5e6-
96231b3b80d8
Duncan P. N. Exon Smith [Tue, 15 Apr 2014 17:48:15 +0000 (17:48 +0000)]
LTO: Add more loop simplification passes to LTO
Similar to r202051, add missing loop simplification passes to the LTO
optimization pipeline.
Patch by Rafael Espindola.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206306
91177308-0d34-0410-b5e6-
96231b3b80d8
Duncan P. N. Exon Smith [Tue, 15 Apr 2014 17:28:26 +0000 (17:28 +0000)]
verify-di: Add back braces for MSVC compatability
Fixup after r206300.
<rdar://problem/
15500563>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206305
91177308-0d34-0410-b5e6-
96231b3b80d8
Duncan P. N. Exon Smith [Tue, 15 Apr 2014 16:27:38 +0000 (16:27 +0000)]
verify-di: Implement DebugInfoVerifier
Implement DebugInfoVerifier, which steals verification relying on
DebugInfoFinder from Verifier.
- Adds LegacyDebugInfoVerifierPassPass, a ModulePass which wraps
DebugInfoVerifier. Uses -verify-di command-line flag.
- Change verifyModule() to invoke DebugInfoVerifier as well as
Verifier.
- Add a call to createDebugInfoVerifierPass() wherever there was a
call to createVerifierPass().
This implementation as a module pass should sidestep efficiency issues,
allowing us to turn debug info verification back on.
<rdar://problem/
15500563>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206300
91177308-0d34-0410-b5e6-
96231b3b80d8
Duncan P. N. Exon Smith [Tue, 15 Apr 2014 16:27:32 +0000 (16:27 +0000)]
verify-di: split out VerifierSupport
Split out assertion and output helpers from Verifier in preparation for
writing the DebugInfoVerifier.
<rdar://problem/
15500563>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206299
91177308-0d34-0410-b5e6-
96231b3b80d8
David Blaikie [Tue, 15 Apr 2014 15:17:14 +0000 (15:17 +0000)]
Use unique_ptr to manage PassInfo instances in the PassRegistry
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206297
91177308-0d34-0410-b5e6-
96231b3b80d8
NAKAMURA Takumi [Tue, 15 Apr 2014 14:13:21 +0000 (14:13 +0000)]
MipsAsmParser.cpp: Fix vg_leak in MipsOperand::CreateMem(). Mem.Base is managed by k_Memory itself.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206293
91177308-0d34-0410-b5e6-
96231b3b80d8
NAKAMURA Takumi [Tue, 15 Apr 2014 14:06:27 +0000 (14:06 +0000)]
MipsAsmParser::ParseRegister(): Be responsible to delete an Operand on a temporary Operands.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206292
91177308-0d34-0410-b5e6-
96231b3b80d8
Tim Northover [Tue, 15 Apr 2014 14:00:29 +0000 (14:00 +0000)]
AArch64/ARM64: enable more AArch64 tests on ARM64.
No code changes for this bunch, just some test rejigs.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206291
91177308-0d34-0410-b5e6-
96231b3b80d8
Tim Northover [Tue, 15 Apr 2014 14:00:19 +0000 (14:00 +0000)]
AArch64/ARM64: add missing pattern for extending load.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206290
91177308-0d34-0410-b5e6-
96231b3b80d8
Tim Northover [Tue, 15 Apr 2014 14:00:15 +0000 (14:00 +0000)]
AArch64/ARM64: only mangle MOVZ/MOVN during encoding when needed
Sometimes we need emit the bits that would actually be a MOVN when producing a
relocated MOVZ instruction (don't ask). But not always, a check which ARM64 got
wrong until now.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206289
91177308-0d34-0410-b5e6-
96231b3b80d8
Tim Northover [Tue, 15 Apr 2014 14:00:11 +0000 (14:00 +0000)]
AArch64/ARM64: add support for large code-model jump tables.
I've left the MachO CodeGen as it is, there's a reasonable chance it should use
the GOT like ConstPools, but I'm not certain.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206288
91177308-0d34-0410-b5e6-
96231b3b80d8
Tim Northover [Tue, 15 Apr 2014 14:00:06 +0000 (14:00 +0000)]
AArch64/ARM64: add patterns for various commutations of FNMADD.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206287
91177308-0d34-0410-b5e6-
96231b3b80d8
Tim Northover [Tue, 15 Apr 2014 14:00:03 +0000 (14:00 +0000)]
AArch64/ARM64: add half as a storage type on ARM64.
This brings it into line with the AArch64 behaviour and should open the way for
certain OpenCL features.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206286
91177308-0d34-0410-b5e6-
96231b3b80d8
Tim Northover [Tue, 15 Apr 2014 13:59:57 +0000 (13:59 +0000)]
AArch64/ARM64: copy patterns for fixed-point conversions
Code is mostly copied directly across, with a slight extension of the
ISelDAGToDAG function so that it can cope with the floating-point constants
being behind a litpool.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206285
91177308-0d34-0410-b5e6-
96231b3b80d8
Tim Northover [Tue, 15 Apr 2014 13:59:53 +0000 (13:59 +0000)]
ARM64: add constraints to various FastISel operations
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206284
91177308-0d34-0410-b5e6-
96231b3b80d8
Tim Northover [Tue, 15 Apr 2014 13:59:49 +0000 (13:59 +0000)]
FastISel: constrain the RegClass of operands when emitting instructions.
ARM64 suffered multiple -verify-machineinstr failures (principally over the
xsp/xzr issue) because FastISel was completely ignoring which subset of the
general-purpose registers each instruction required.
More fixes are coming in ARM64 specific FastISel, but this should cover the
generic problems.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206283
91177308-0d34-0410-b5e6-
96231b3b80d8
Tim Northover [Tue, 15 Apr 2014 13:59:44 +0000 (13:59 +0000)]
AArch64/ARM64: add more arm64 lines to AArch64 regression tests
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206282
91177308-0d34-0410-b5e6-
96231b3b80d8
Tim Northover [Tue, 15 Apr 2014 13:59:40 +0000 (13:59 +0000)]
AArch64/ARM64: add dp tests from AArch64
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206281
91177308-0d34-0410-b5e6-
96231b3b80d8
NAKAMURA Takumi [Tue, 15 Apr 2014 13:22:11 +0000 (13:22 +0000)]
ARM64AsmParser.cpp: Fix vg_leak in MC/ARM64/fp-encoding.s.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206279
91177308-0d34-0410-b5e6-
96231b3b80d8
Stepan Dyatkovskiy [Tue, 15 Apr 2014 11:43:09 +0000 (11:43 +0000)]
Optional hash symbol feature support for ARM64
http://reviews.llvm.org/D3328
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206276
91177308-0d34-0410-b5e6-
96231b3b80d8
Vladimir Medic [Tue, 15 Apr 2014 10:14:49 +0000 (10:14 +0000)]
Current definition of subtract with immediate instruction aliases uses CodeGenOnly defined instructions and post matcher expansion methods to emit real instructions add with immediate. However, they can directly alias add with immediate instruction and remove unnecessary definitions and code in MipsAsmParser.cpp. This patch makes no change in functionality, just removes unnecessary definitions and code.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206272
91177308-0d34-0410-b5e6-
96231b3b80d8
NAKAMURA Takumi [Tue, 15 Apr 2014 10:12:47 +0000 (10:12 +0000)]
vect.omp.persistence.ll REQUIRES asserts due to -debug-only.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206271
91177308-0d34-0410-b5e6-
96231b3b80d8
Chandler Carruth [Tue, 15 Apr 2014 09:44:09 +0000 (09:44 +0000)]
[Allocator] Finally, finish nuking the redundant code that led me here
by removing the MallocSlabAllocator entirely and just using
MallocAllocator directly. This makes all off these allocators expose and
utilize the same core interface.
The only ugly part of this is that it exposes the fact that the JIT
allocator has no real handling of alignment, any more than the malloc
allocator does. =/ It would be nice to fix both of these to support
alignments, and then to leverage that in the BumpPtrAllocator to do less
over allocation in order to manually align pointers. But, that's another
patch for another day. This patch has no functional impact, it just
removes the somewhat meaningless wrapper around MallocAllocator.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206267
91177308-0d34-0410-b5e6-
96231b3b80d8
Alexey Bataev [Tue, 15 Apr 2014 09:37:30 +0000 (09:37 +0000)]
D3348 - [BUG] "Rotate Loop" pass kills "llvm.vectorizer.enable" metadata
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206266
91177308-0d34-0410-b5e6-
96231b3b80d8
Chandler Carruth [Tue, 15 Apr 2014 08:59:52 +0000 (08:59 +0000)]
[Allocator] Pass the size to the deallocation function. This, on some
allocation libraries, may allow more efficient allocation and
deallocation. It at least makes the interface implementable by the JIT
memory manager.
However, this highlights problematic overloading between the void* and
the T* deallocation functions. I'm looking into a better way to do this,
but as it happens, it comes up rarely in the codebase.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206265
91177308-0d34-0410-b5e6-
96231b3b80d8