opencl/llvm.git
7 years agoRewrite calculateDbgValueHistory to make it (hopefully) more transparent.
Alexey Samsonov [Tue, 20 May 2014 18:34:54 +0000 (18:34 +0000)]
Rewrite calculateDbgValueHistory to make it (hopefully) more transparent.

This change preserves the original algorithm of generating history
for user variables, but makes it more clear.

High-level description of algorithm:
Scan all the machine basic blocks and machine instructions in the order
they are emitted to the object file. Do the following:
1) If we see a DBG_VALUE instruction, add it to the history of the
corresponding user variable. Keep track of all user variables, whose
locations are described by a register.
2) If we see a regular instruction, look at all the registers it clobbers,
and terminate the location range for all variables described by these registers.
3) At the end of the basic block, terminate location ranges for all
user variables described by some register.

Although this change shouldn't be user-visible (the contents of .debug_loc section
should be the same), it changes some internal assumptions about the set
of instructions used to track the variable locations. Watching the bots.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209225 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoPR19767: DebugInfo emission of pointer constants.
David Blaikie [Tue, 20 May 2014 18:21:51 +0000 (18:21 +0000)]
PR19767: DebugInfo emission of pointer constants.

In refactoring DwarfUnit::isUnsignedDIType I restricted it to only work
on values with signedness (unsigned or signed), asserting on anything
else (which did uncover some bugs). But it turns out that we do need to
emit constants of signless data, such as pointer constants - only null
pointer constants are known to need this so far, but it's conceivable
that there might be non-null pointer constants at some point (hardcoded
address offsets for device drivers?).

This patch just uses 'unsigned' for signless data such as pointer
constants. Arguably we could use signless representations
(DW_FORM_dataN) instead, allowing a trinary result from isUnsignedDIType
(signed, unsigned, signless), but this seems reasonable for now.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209223 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAvoids DCE on write_register
Renato Golin [Tue, 20 May 2014 17:40:03 +0000 (17:40 +0000)]
Avoids DCE on write_register

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209222 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[PowerPC] PR19796: Also match ISD::TargetConstant in isIntS16Immediate
Adam Nemet [Tue, 20 May 2014 17:20:34 +0000 (17:20 +0000)]
[PowerPC] PR19796: Also match ISD::TargetConstant in isIntS16Immediate

The SplitIndexingFromLoad changes exposed a latent isel bug in the PowerPC64
backend.  We matched an immediate offset with STWX8 even though it only
supports register offset.

The culprit is the complex-pattern predicate, SelectAddrIdx, which decides
that if the offset is not ISD::Constant it must be a register.

Many thanks to Bill Schmidt for testing this.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209219 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoClean up language and grammar.
Eric Christopher [Tue, 20 May 2014 17:11:11 +0000 (17:11 +0000)]
Clean up language and grammar.

Based on a patch by jfcaron3@gmail.com!
PR19806

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209216 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoTemporarily revert: r209129 - [mips][mips64r6] Sorted *_ENC, *_DESC classes and tests
Daniel Sanders [Tue, 20 May 2014 14:46:24 +0000 (14:46 +0000)]
Temporarily revert: r209129 - [mips][mips64r6] Sorted *_ENC, *_DESC classes and tests

After discussion with Zoran, we have decided to temporarily revert this commit.
It's causing some difficult to resolve conflicts and we are under time pressure
to deliver an initial MIPS64r6 compiler.

We will re-apply an equivalent patch once the time pressure has passed.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209211 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[ARM64] Port basic-a64-diagnostics.s over to ARM64
Bradley Smith [Tue, 20 May 2014 13:33:41 +0000 (13:33 +0000)]
[ARM64] Port basic-a64-diagnostics.s over to ARM64

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209207 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoTableGen: permit non-leaf ComplexPattern uses
Tim Northover [Tue, 20 May 2014 11:52:46 +0000 (11:52 +0000)]
TableGen: permit non-leaf ComplexPattern uses

This allows the results of a ComplexPattern check to be distributed to separate
named Operands, instead of the current system where all results must apply (and
match perfectly) with a single Operand.

For example, if "some_addrmode" is a ComplexPattern producing two results, you
can write:

   def : Pat<(load (some_addrmode GPR64:$base, imm:$offset)),
             (INST GPR64:$base, imm:$offset)>;

This should allow neater instruction definitions in TableGen that don't put all
possible aspects of addressing into a single operand, but are still usable with
relatively simple C++ CodeGen idioms.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209206 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAdd parentheses to suppress the gcc warning '-Wparentheses'.
Simon Atanasyan [Tue, 20 May 2014 10:23:04 +0000 (10:23 +0000)]
Add parentheses to suppress the gcc warning '-Wparentheses'.

No functional changes.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209203 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoLegalizer: Make bswap promotion safe for vectors.
Benjamin Kramer [Tue, 20 May 2014 09:42:31 +0000 (09:42 +0000)]
Legalizer: Make bswap promotion safe for vectors.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209202 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[Mips] Add more relocation types and MIPS specific e_flags constants.
Simon Atanasyan [Tue, 20 May 2014 09:27:49 +0000 (09:27 +0000)]
[Mips] Add more relocation types and MIPS specific e_flags constants.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209201 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoARMEB: Additional test files for ARM fixups
Christian Pirker [Tue, 20 May 2014 09:24:37 +0000 (09:24 +0000)]
ARMEB: Additional test files for ARM fixups

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209200 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoTableGen: convert InstAlias's Emit bit to an int.
Tim Northover [Tue, 20 May 2014 09:17:16 +0000 (09:17 +0000)]
TableGen: convert InstAlias's Emit bit to an int.

When multiple aliases overlap, the correct string to print can often be
determined purely by considering the InstAlias declarations in some particular
order. This allows the user to specify that order manually when desired,
without resorting to hacking around with the default lexicographical order on
Record instantiation, which is error-prone and ugly.

I was also mistaken about "add w2, w3, w4" being the same as "add w2, w3, w4,
uxtw". That's only true if Rn is the stack pointer.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209199 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86] Tune LEA usage for Silvermont
Alexey Volkov [Tue, 20 May 2014 08:55:50 +0000 (08:55 +0000)]
[X86] Tune LEA usage for Silvermont

According to Intel Software Optimization Manual on Silvermont in some cases LEA
is better to be replaced with ADD instructions:
"The rule of thumb for ADDs and LEAs is that it is justified to use LEA
with a valid index and/or displacement for non-destructive destination purposes
(especially useful for stack offset cases), or to use a SCALE.
Otherwise, ADD(s) are preferable."

Differential Revision: http://reviews.llvm.org/D3826

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209198 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[LV][REFACTOR] One more tiny fix for printing debug locations in loop vectorizer...
Zinovy Nis [Tue, 20 May 2014 08:26:20 +0000 (08:26 +0000)]
[LV][REFACTOR] One more tiny fix for printing debug locations in loop vectorizer. Now consistent with the remarks emitter.

Differential Revision: http://reviews.llvm.org/D3821

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209197 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agollvm-readobj: use range-based for loop
Saleem Abdulrasool [Tue, 20 May 2014 05:18:06 +0000 (05:18 +0000)]
llvm-readobj: use range-based for loop

Convert an additional site to a range based for loop.  NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209194 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoTeach isKnownNonNull that a nonnull return is not null. Add a test for this case...
Nick Lewycky [Tue, 20 May 2014 05:13:21 +0000 (05:13 +0000)]
Teach isKnownNonNull that a nonnull return is not null. Add a test for this case as well as the case of a nonnull attribute (already handled but not tested).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209193 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoDebugInfo: Emit function definitions within their namespace scope.
David Blaikie [Tue, 20 May 2014 03:23:24 +0000 (03:23 +0000)]
DebugInfo: Emit function definitions within their namespace scope.

This workaround (presumably for ancient GDB) doesn't appear to be
required (GDB 7.5 seems to tolerate function definition DIEs in
namespace scope just fine).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209189 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoFollow up to 209187, updating a test to use FileCheck. Needed to ignore an extra...
David Blaikie [Tue, 20 May 2014 02:40:34 +0000 (02:40 +0000)]
Follow up to 209187, updating a test to use FileCheck. Needed to ignore an extra DW_TAG_class_type

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209188 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoUpdate test/DebugInfo/2010-04-06-NestedFnDbgInfo.ll to use FileCheck.
David Blaikie [Tue, 20 May 2014 02:19:15 +0000 (02:19 +0000)]
Update test/DebugInfo/2010-04-06-NestedFnDbgInfo.ll to use FileCheck.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209187 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAdd 'nonnull', a new parameter and return attribute which indicates that the pointer...
Nick Lewycky [Tue, 20 May 2014 01:23:40 +0000 (01:23 +0000)]
Add 'nonnull', a new parameter and return attribute which indicates that the pointer is not null. Instcombine will elide comparisons between these and null. Patch by Luqman Aden!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209185 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoLTO: Add a testcase for linking modules with incompatible Debug Info
Adrian Prantl [Mon, 19 May 2014 23:41:25 +0000 (23:41 +0000)]
LTO: Add a testcase for linking modules with incompatible Debug Info
Versions.
rdar://problem/16926122

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209183 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoOK, NAKAMURA Takumi beat me to this change. So backing out my addition of
Kevin Enderby [Mon, 19 May 2014 23:26:51 +0000 (23:26 +0000)]
OK, NAKAMURA Takumi beat me to this change.  So backing out my addition of
llvm-size to the list so it only has one.  Sorry for the noise.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209181 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoDavid Blaikie pointed out that the test added in r209158 fails in the
Kevin Enderby [Mon, 19 May 2014 23:22:58 +0000 (23:22 +0000)]
David Blaikie pointed out that the test added in r209158 fails in the
check-llvm target under CMake.  And the test/CMakeFiles.txt needs to
have llvm-size added to the list.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209179 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoDebugInfo: Assume all subprogram DIEs have been created before any abstract subprogra...
David Blaikie [Mon, 19 May 2014 23:16:19 +0000 (23:16 +0000)]
DebugInfo: Assume all subprogram DIEs have been created before any abstract subprograms are constructed.

Since we visit the whole list of subprograms for each CU at module
start, this is clearly true - don't test for the case, just assert it.

A few old test cases seemed to have incomplete subprogram lists, but any
attempt to reproduce them shows full subprogram lists that even include
entities that have been completely inlined and the out of line
definition removed.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209178 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[CMake] Add llvm-size to check-llvm, to fix build since r209158.
NAKAMURA Takumi [Mon, 19 May 2014 23:12:43 +0000 (23:12 +0000)]
[CMake] Add llvm-size to check-llvm, to fix build since r209158.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209177 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[ARM64] Adds Cortex-A53 scheduling support for vector load/store post.
Chad Rosier [Mon, 19 May 2014 22:59:51 +0000 (22:59 +0000)]
[ARM64] Adds Cortex-A53 scheduling support for vector load/store post.
Patch by Dave Estes<cestes@codeaurora.org>!
PR19761 http://reviews.llvm.org/D3829

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209176 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRemove unused method declaration
Matt Arsenault [Mon, 19 May 2014 22:55:35 +0000 (22:55 +0000)]
Remove unused method declaration

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209174 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAdd documentation for llvm-dwarfdump tool
Alexey Samsonov [Mon, 19 May 2014 22:53:29 +0000 (22:53 +0000)]
Add documentation for llvm-dwarfdump tool

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209173 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoDebugInfo: Don't include DW_AT_inline on each abstract definition multiple times.
David Blaikie [Mon, 19 May 2014 22:07:16 +0000 (22:07 +0000)]
DebugInfo: Don't include DW_AT_inline on each abstract definition multiple times.

When I refactored this in r208636 I accidentally caused this to be added
multiple times to each abstract subprogram (not accounting for the
deduplicating effect of the InlinedSubprogramDIEs set).

This got better in r208798 when the abstract definitions got the
attribute added to them at construction time, but still had the
redundant copies introduced in r208636.

This commit removes those excess DW_AT_inlines and relies solely on the
insertion in r208798.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209166 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoDebugInfo: Fix missing inlined_subroutines caused by r208748.
David Blaikie [Mon, 19 May 2014 21:54:31 +0000 (21:54 +0000)]
DebugInfo: Fix missing inlined_subroutines caused by r208748.

The check in DwarfDebug::constructScopeDIE was meant to consider inlined
subroutines as any non-top-level scope that was a subprogram. Instead of
checking "not top level scope" it was checking if the /subprogram's/
scope was non-top-level.

Fix this and beef up a test case to demonstrate some of the missing
inlined_subroutines are no longer missing.

In the course of fixing this I also found that r208748 (with this fix)
found one /extra/ inlined_subroutine in concrete_out_of_line.ll due to
two inlined_subroutines having the same inlinedAt location. The previous
implementation was collapsing these into a single inlined subroutine.

I'm not sure what the original code was that created this .ll file so
I'm not sure if this actually happens in practice today. Since we
deliberately include column information to disambiguate two calls on the
same line, that may've addressed this bug in the frontend, but it's good
to know that workaround isn't necessary for this particular case
anymore.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209165 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoFix typos.
Eric Christopher [Mon, 19 May 2014 21:18:47 +0000 (21:18 +0000)]
Fix typos.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209164 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agocmake: Remove -D NDEBUG from CFLAGS as well as CXXFLAGS
Reid Kleckner [Mon, 19 May 2014 21:13:41 +0000 (21:13 +0000)]
cmake: Remove -D NDEBUG from CFLAGS as well as CXXFLAGS

This silences ~7 warnings on .c files in the LLVM build.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209163 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[ConstantHoisting][X86] Change the cost model to never hoist constants for types...
Juergen Ributzka [Mon, 19 May 2014 21:00:53 +0000 (21:00 +0000)]
[ConstantHoisting][X86] Change the cost model to never hoist constants for types larger than i128.

Currently the X86 backend doesn't support types larger than i128 very well. For
example an i192 multiply will assert in codegen when the 2nd argument is a constant and the constant got hoisted.

This fix changes the cost model to never hoist constants for types larger than
i128. Once the codegen issues have been resolved, the cost model can be updated
to allow also larger types.

This is related to <rdar://problem/16954938>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209162 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86] Add ISel patterns to improve the selection of TZCNT and LZCNT.
Andrea Di Biagio [Mon, 19 May 2014 20:38:59 +0000 (20:38 +0000)]
[X86] Add ISel patterns to improve the selection of TZCNT and LZCNT.

Instructions TZCNT (requires BMI1) and LZCNT (requires LZCNT), always
provide the operand size as output if the input operand is zero.

We can take advantage of this knowledge during instruction selection
stage in order to simplify a few corner case.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209159 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoImplement MachOObjectFile::isSectionData() and MachOObjectFile::isSectionBSS
Kevin Enderby [Mon, 19 May 2014 20:36:02 +0000 (20:36 +0000)]
Implement MachOObjectFile::isSectionData() and MachOObjectFile::isSectionBSS
so that llvm-size will total up all the sections in the Berkeley format.  This
allows for rough categorizations for Mach-O sections.  And allows the total of
llvm-size’s Berkeley and System V formats to be the same.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209158 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAdded more insertps optimizations
Filipe Cabecinhas [Mon, 19 May 2014 19:45:57 +0000 (19:45 +0000)]
Added more insertps optimizations

Summary:
When inserting an element that's coming from a vector load or a broadcast
of a vector (or scalar) load, combine the load into the insertps
instruction.
Added PerformINSERTPSCombine for the case where we need to fix the load
(load of a vector + insertps with a non-zero CountS).
Added patterns for the broadcasts.

Also added tests for SSE4.1, AVX, and AVX2.

Reviewers: delena, nadav, craig.topper

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D3581

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209156 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoFix testers by removing dubious testcase for r209154.
Lang Hames [Mon, 19 May 2014 19:38:48 +0000 (19:38 +0000)]
Fix testers by removing dubious testcase for r209154.

It turns out that not all the world is x86-64. Who knew?

I'll get to work on a more appropriate test case for this patch.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209155 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[RuntimeDyld] Fix x86-64 MachO GOT relocation handling.
Lang Hames [Mon, 19 May 2014 19:21:25 +0000 (19:21 +0000)]
[RuntimeDyld] Fix x86-64 MachO GOT relocation handling.

For GOT relocations the addend should modify the offset to the
GOT entry, not the value of the entry itself. Teach RuntimeDyldMachO
to do The Right Thing here.

Fixes <rdar://problem/16961886>.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209154 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoKill symbolization functionality in llvm-dwarfdump. We have llvm-symbolizer for that.
Alexey Samsonov [Mon, 19 May 2014 18:45:32 +0000 (18:45 +0000)]
Kill symbolization functionality in llvm-dwarfdump. We have llvm-symbolizer for that.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209152 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoCheck the alwaysinline attribute on the call as well as on the caller.
Peter Collingbourne [Mon, 19 May 2014 18:25:54 +0000 (18:25 +0000)]
Check the alwaysinline attribute on the call as well as on the caller.

Differential Revision: http://reviews.llvm.org/D3815

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209150 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoUse range for
Matt Arsenault [Mon, 19 May 2014 17:52:48 +0000 (17:52 +0000)]
Use range for

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209147 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoreverting r209132
Jyotsna Verma [Mon, 19 May 2014 16:22:11 +0000 (16:22 +0000)]
reverting r209132

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209139 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoMemoryBuffer: Use GetNativeSystemInfo()
Alp Toker [Mon, 19 May 2014 16:13:28 +0000 (16:13 +0000)]
MemoryBuffer: Use GetNativeSystemInfo()

Removes old 4096 byte workaround. This functionality has been available since
Windows XP.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209137 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRevert "Patch for function cloning to inline all blocks whose address is taken"
Eric Christopher [Mon, 19 May 2014 16:04:10 +0000 (16:04 +0000)]
Revert "Patch for function cloning to inline all blocks whose address is taken"
as it was causing build failures in ruby.

This reverts commit r207713.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209135 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[ARM64] Split tbz/tbnz into W/X register variant
Bradley Smith [Mon, 19 May 2014 15:58:15 +0000 (15:58 +0000)]
[ARM64] Split tbz/tbnz into W/X register variant

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209134 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoHexagon: Add encoding bits to the mpy instructions.
Jyotsna Verma [Mon, 19 May 2014 15:32:07 +0000 (15:32 +0000)]
Hexagon: Add encoding bits to the mpy instructions.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209132 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[mips][mips64r6] Sorted *_ENC, *_DESC classes and tests
Zoran Jovanovic [Mon, 19 May 2014 14:57:46 +0000 (14:57 +0000)]
[mips][mips64r6] Sorted *_ENC, *_DESC classes and tests

Differential Revision: http://reviews.llvm.org/D3808

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209129 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoResolving MSVC warnings about switch statements with a default label, but no case...
Aaron Ballman [Mon, 19 May 2014 14:29:04 +0000 (14:29 +0000)]
Resolving MSVC warnings about switch statements with a default label, but no case labels. No functional changes intended.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209126 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoFlip on vectorization of bswap intrinsics.
Benjamin Kramer [Mon, 19 May 2014 13:48:08 +0000 (13:48 +0000)]
Flip on vectorization of bswap intrinsics.

The cost model conservatively assumes that it will always get scalarized and
that's about as good as we can get with the generic TTI; reasoning whether a
shuffle with an efficient lowering is available is hard. We can override that
conservative estimate for some targets in the future.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209125 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoSDAG: Legalize vector BSWAP into a shuffle if the shuffle is legal but the bswap...
Benjamin Kramer [Mon, 19 May 2014 13:12:38 +0000 (13:12 +0000)]
SDAG: Legalize vector BSWAP into a shuffle if the shuffle is legal but the bswap not.

- On ARM/ARM64 we get a vrev because the shuffle matching code is really smart. We still unroll anything that's not v4i32 though.
- On X86 we get a pshufb with SSSE3. Required more cleverness in isShuffleMaskLegal.
- On PPC we get a vperm for v8i16 and v4i32. v2i64 is unrolled.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209123 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAdded inst-combine for 'MIN(MIN(A, 97), 23)' and 'MAX(MAX(A, 23), 97)'
Dinesh Dwivedi [Mon, 19 May 2014 07:08:32 +0000 (07:08 +0000)]
Added inst-combine for 'MIN(MIN(A, 97), 23)' and 'MAX(MAX(A, 23), 97)'

This removes TODO added in r208849 [http://reviews.llvm.org/D3629]

MIN(MIN(A, 97), 23) -> MIN(A, 23)
MAX(MAX(A, 23), 97) -> MAX(A, 97)

Differential Revision: http://reviews.llvm.org/D3785

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209110 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoChange the blend tests to AVX, not AVX2.
Filipe Cabecinhas [Mon, 19 May 2014 04:47:12 +0000 (04:47 +0000)]
Change the blend tests to AVX, not AVX2.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209107 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoReformat blank lines.
NAKAMURA Takumi [Mon, 19 May 2014 04:43:26 +0000 (04:43 +0000)]
Reformat blank lines.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209106 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoWhitespace.
NAKAMURA Takumi [Mon, 19 May 2014 04:43:03 +0000 (04:43 +0000)]
Whitespace.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209105 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRemove last uses of OwningPtr from llvm. As far as I can tell these method versions...
Craig Topper [Sun, 18 May 2014 21:55:38 +0000 (21:55 +0000)]
Remove last uses of OwningPtr from llvm. As far as I can tell these method versions are not used by lldb, lld, or clang.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209103 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoUse std::unique_ptr instead of OwningPtr in the MemoryBuffer unittests.
Craig Topper [Sun, 18 May 2014 21:01:46 +0000 (21:01 +0000)]
Use std::unique_ptr instead of OwningPtr in the MemoryBuffer unittests.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209102 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoARM: make libcall setup more table driven
Saleem Abdulrasool [Sun, 18 May 2014 16:39:11 +0000 (16:39 +0000)]
ARM: make libcall setup more table driven

Rather than create a series of function calls to setup the library calls, create
a table with the information and just use the table to drive the configuration
of the library calls.  This makes it easier to both inspect the list as well as
to modify it.  NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209089 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoOptions: Use erase_if to remove Args from the list.
Benjamin Kramer [Sun, 18 May 2014 15:14:13 +0000 (15:14 +0000)]
Options: Use erase_if to remove Args from the list.

While there make getOption return a const reference so we don't have to put it
on the stack when calling methods on it. No functionality change.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209088 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoARM: improve WoA ABI conformance for frame register
Saleem Abdulrasool [Sun, 18 May 2014 04:12:52 +0000 (04:12 +0000)]
ARM: improve WoA ABI conformance for frame register

Windows on ARM uses R11 for the frame pointer even though the environment is a
pure Thumb-2, thumb-only environment.  Replicate this behaviour to improve
Windows ABI compatibility.  This register is used for fast stack walking, and
thus is part of the Windows ABI.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209085 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoARM: consolidate frame pointer register knowledge
Saleem Abdulrasool [Sun, 18 May 2014 03:18:09 +0000 (03:18 +0000)]
ARM: consolidate frame pointer register knowledge

Use the ARMBaseRegisterInfo to query the frame register.  The base register info
is aware of the frame register that is used for the frame pointer.  Use that to
determine the frame register rather than duplicating the knowledge.  Although,
the code path is slightly different in that it may return SP, that can only
occur if the frame pointer has been omitted in the machine function, which is
supposed to contain the desired value in that case.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209084 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoTarget: remove old constructors for CallLoweringInfo
Saleem Abdulrasool [Sat, 17 May 2014 21:50:17 +0000 (21:50 +0000)]
Target: remove old constructors for CallLoweringInfo

This is mostly a mechanical change changing all the call sites to the newer
chained-function construction pattern.  This removes the horrible 15-parameter
constructor for the CallLoweringInfo in favour of setting properties of the call
via chained functions.  No functional change beyond the removal of the old
constructors are intended.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209082 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoTarget: add support to build CallLoweringInfo
Saleem Abdulrasool [Sat, 17 May 2014 21:50:06 +0000 (21:50 +0000)]
Target: add support to build CallLoweringInfo

Rather than introducing an auxiliary CallLoweringInfoBuilder, add the methods to
do chained function construction directly to CallLoweringInfo.  This reduces the
monstrous 15-parameter constructor into a series of simpler (for some definition
of simpler) functions that control particular aspects of the call.  The old
interfaces can be completely removed once callers are moved to the new chained
constructor pattern.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209081 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoTarget: change member from reference to pointer
Saleem Abdulrasool [Sat, 17 May 2014 21:50:01 +0000 (21:50 +0000)]
Target: change member from reference to pointer

This is a preliminary step to help ease the construction of CallLoweringInfo.
Changing the construction to a chained function pattern requires that the
parameter be nullable.  However, rather than copying the vector, save a pointer
rather than the reference to permit a late binding of the arguments.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209080 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoARM: whitespace
Saleem Abdulrasool [Sat, 17 May 2014 21:49:54 +0000 (21:49 +0000)]
ARM: whitespace

Remove some whitespace.  NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209079 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoUse create methods since msvc doesn't handle delegating constructors.
Rafael Espindola [Sat, 17 May 2014 21:29:57 +0000 (21:29 +0000)]
Use create methods since msvc doesn't handle delegating constructors.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209076 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoHandle ConstantAggregateZero when upgrading global_ctors.
Rafael Espindola [Sat, 17 May 2014 21:00:22 +0000 (21:00 +0000)]
Handle ConstantAggregateZero when upgrading global_ctors.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209075 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoReduce abuse of default values in the GlobalAlias constructor.
Rafael Espindola [Sat, 17 May 2014 19:57:46 +0000 (19:57 +0000)]
Reduce abuse of default values in the GlobalAlias constructor.

This is in preparation for adding an optional offset.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209073 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRevert r209049 and r209065, "Add support for combining GEPs across PHI nodes"
NAKAMURA Takumi [Sat, 17 May 2014 14:39:21 +0000 (14:39 +0000)]
Revert r209049 and r209065, "Add support for combining GEPs across PHI nodes"

It broke clang selfhosting even after r209065.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209067 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoMachO: Add comments describing section flags/attrs
David Majnemer [Sat, 17 May 2014 07:36:36 +0000 (07:36 +0000)]
MachO: Add comments describing section flags/attrs

When pruning superfluous MachO structure definitions, I chose to keep
the most generically useful which is why Support's definitions won over
the ones in MC.

However, the MC copy had some useful comments describing some of the
field values.

Bring these back to the copy in Support.  While doing this, fill in some
of the underdocumented definitions as well.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209066 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoFix for sanitizer crash introduced in r209049
Louis Gerbarg [Sat, 17 May 2014 06:51:36 +0000 (06:51 +0000)]
Fix for sanitizer crash introduced in r209049

This patch fixes 3 issues introduced by r209049 that only showed up in on
the sanitizer buildbots. One was a typo in a compare. The other is a check to
confirm that the single differing value in the two incoming GEPs is the same
type. The final issue was the the IRBuilder under some circumstances would
build PHIs in the middle of the block.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209065 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoTarget: Replace getSection().empty() with hasSection()
David Majnemer [Sat, 17 May 2014 05:18:40 +0000 (05:18 +0000)]
Target: Replace getSection().empty() with hasSection()

No functional change, just a small cleanup.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209064 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agotest: fix copy-paste mistake
Saleem Abdulrasool [Sat, 17 May 2014 04:32:38 +0000 (04:32 +0000)]
test: fix copy-paste mistake

Accidental over-quoting of the match string.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209058 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoARM: use the proper target object format for WoA
Saleem Abdulrasool [Sat, 17 May 2014 04:28:08 +0000 (04:28 +0000)]
ARM: use the proper target object format for WoA

WoA uses COFF, not ELF.  ARMISelLowering::createTLOF would previously return ELF
for any non-MachO platform.  This was a missed site when the original change for
target format support for Windows on ARM was done.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209057 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[x86] Fix a bad predicate I spotted by inspection -- pshufhw and pshuflw
Chandler Carruth [Sat, 17 May 2014 03:29:20 +0000 (03:29 +0000)]
[x86] Fix a bad predicate I spotted by inspection -- pshufhw and pshuflw
were added in SSE2, no SSSE3. Found this while auditing all uses of
SSSE3 in the X86 target. I don't actually expect this to make
a significant difference on anything and I don't have any detailed test
cases but I updated the existing test cases that already covered some of
this code path.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209056 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[DWARF parser] Teach DIContext to fetch short (non-linkage) function names for a...
Alexey Samsonov [Sat, 17 May 2014 00:07:48 +0000 (00:07 +0000)]
[DWARF parser] Teach DIContext to fetch short (non-linkage) function names for a given address.

Change --functions option in llvm-symbolizer tool to accept
values "none", "short" or "linkage". Update the tests and docs
accordingly.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209050 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAdd support for combining GEPs across PHI nodes
Louis Gerbarg [Fri, 16 May 2014 23:47:24 +0000 (23:47 +0000)]
Add support for combining GEPs across PHI nodes

Currently LLVM will generally merge GEPs. This allows backends to use more
complex addressing modes. In some cases this is not happening because there
is PHI inbetween the two GEPs:

  GEP1--\
        |-->PHI1-->GEP3
  GEP2--/

This patch checks to see if GEP1 and GEP2 are similiar enough that they can be
cloned (GEP12) in GEP3's BB, allowing GEP->GEP merging (GEP123):

  GEP1--\                     --\                           --\
        |-->PHI1-->GEP3  ==>    |-->PHI2->GEP12->GEP3 == >    |-->PHI2->GEP123
  GEP2--/                     --/                           --/

This also breaks certain use chains that are preventing GEP->GEP merges that the
the existing instcombine would merge otherwise.

Tests included.

rdar://15547484

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209049 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoUse a sized enum for MachineOperandType. No functionality change
Pete Cooper [Fri, 16 May 2014 23:28:17 +0000 (23:28 +0000)]
Use a sized enum for MachineOperandType.  No functionality change

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209048 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAdded tests for the cost of lowering VSELECT instructions.
Filipe Cabecinhas [Fri, 16 May 2014 22:47:58 +0000 (22:47 +0000)]
Added tests for the cost of lowering VSELECT instructions.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209045 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoImplemented special cases for PerformVSELECTCombine.
Filipe Cabecinhas [Fri, 16 May 2014 22:47:54 +0000 (22:47 +0000)]
Implemented special cases for PerformVSELECTCombine.

vselects with constant masks, after legalization, will get turned into
specialized shuffle_vectors so they can be matched to blend+imm
instructions.

Fixed some tests.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209044 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoLower vselects into X86ISD::BLENDI when appropriate.
Filipe Cabecinhas [Fri, 16 May 2014 22:47:49 +0000 (22:47 +0000)]
Lower vselects into X86ISD::BLENDI when appropriate.

LowerVSELECT will, if possible, generate a X86ISD::BLENDI DAG node if the
condition is constant and we can emit that instruction, given the
subtarget.

This is not enough for all cases. An additional SELECTCombine optimization
will be committed.

Fixed tests that were expecting variable blends but where a blend+imm can
be generated.
Added test where we can't emit blend+immediate.
Added avx2 blend+imm tests.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209043 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoImplemented LowerVSELECT to custom lower some instructions.
Filipe Cabecinhas [Fri, 16 May 2014 22:47:43 +0000 (22:47 +0000)]
Implemented LowerVSELECT to custom lower some instructions.

No functionality change intended. The types that previously were set to
lower as Expand or Legal are doing the same thing with this lowering
function.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209042 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoDelete getAliasedGlobal.
Rafael Espindola [Fri, 16 May 2014 22:37:03 +0000 (22:37 +0000)]
Delete getAliasedGlobal.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209040 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoDebugInfo: Assert rather than conditionalizing when a CU's subprogram list contains...
David Blaikie [Fri, 16 May 2014 22:21:45 +0000 (22:21 +0000)]
DebugInfo: Assert rather than conditionalizing when a CU's subprogram list contains declarations.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209039 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoDebugInfo: Handle emitting constants of C++ unicode character type.
David Blaikie [Fri, 16 May 2014 21:53:09 +0000 (21:53 +0000)]
DebugInfo: Handle emitting constants of C++ unicode character type.

Patch by Stephan Tolksdorf! (with some test case stuff by me)

Differential Revision: http://reviews.llvm.org/D3810

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209037 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoR600/SI: Refactor the VOP3_32 tablegen class
Tom Stellard [Fri, 16 May 2014 20:56:47 +0000 (20:56 +0000)]
R600/SI: Refactor the VOP3_32 tablegen class

This will allow us to use a single MachineInstr to represent
instructions which behave the same but have different encodings
on some subtargets.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209028 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoR600/SI: Add a PredicateControl class for managing TableGen predicates
Tom Stellard [Fri, 16 May 2014 20:56:45 +0000 (20:56 +0000)]
R600/SI: Add a PredicateControl class for managing TableGen predicates

This was inspired by the PredicateControl class in the MIPS backend.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209027 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoR600/SI: Move tablegen patterns away from instruction defs
Tom Stellard [Fri, 16 May 2014 20:56:44 +0000 (20:56 +0000)]
R600/SI: Move tablegen patterns away from instruction defs

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209026 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoR600/SI: Remove unused instruction
Tom Stellard [Fri, 16 May 2014 20:56:43 +0000 (20:56 +0000)]
R600/SI: Remove unused instruction

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209025 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoR600/SI: Promote f32 SELECT to i32
Tom Stellard [Fri, 16 May 2014 20:56:41 +0000 (20:56 +0000)]
R600/SI: Promote f32 SELECT to i32

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209024 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoR600/SI: Remove duplicate pattern
Tom Stellard [Fri, 16 May 2014 20:56:37 +0000 (20:56 +0000)]
R600/SI: Remove duplicate pattern

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209023 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAdd comdat key field to llvm.global_ctors and llvm.global_dtors
Reid Kleckner [Fri, 16 May 2014 20:39:27 +0000 (20:39 +0000)]
Add comdat key field to llvm.global_ctors and llvm.global_dtors

This allows us to put dynamic initializers for weak data into the same
comdat group as the data being initialized.  This is necessary for MSVC
ABI compatibility.  Once we have comdats for guard variables, we can use
the combination to help GlobalOpt fire more often for weak data with
guarded initialization on other platforms.

Reviewers: nlewycky

Differential Revision: http://reviews.llvm.org/D3499

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209015 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoFix a warning in builds without asserts.
Rafael Espindola [Fri, 16 May 2014 20:05:08 +0000 (20:05 +0000)]
Fix a warning in builds without asserts.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209012 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoDebugInfo: Add an assert regarding the subprogram in the subprogram map matching...
David Blaikie [Fri, 16 May 2014 19:42:10 +0000 (19:42 +0000)]
DebugInfo: Add an assert regarding the subprogram in the subprogram map matching the abstract subprogram.

I'm not sure this is how it'll be going forward (I'd rather prefer the
definition to be in the main SP mapping, for various reasons) but this
helps me understand how it is today.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209009 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoFix most of PR10367.
Rafael Espindola [Fri, 16 May 2014 19:35:39 +0000 (19:35 +0000)]
Fix most of PR10367.

This patch changes the design of GlobalAlias so that it doesn't take a
ConstantExpr anymore. It now points directly to a GlobalObject, but its type is
independent of the aliasee type.

To avoid changing all alias related tests in this patches, I kept the common
syntax

@foo = alias i32* @bar

to mean the same as now. The cases that used to use cast now use the more
general syntax

@foo = alias i16, i32* @bar.

Note that GlobalAlias now behaves a bit more like GlobalVariable. We
know that its type is always a pointer, so we omit the '*'.

For the bitcode, a nice surprise is that we were writing both identical types
already, so the format change is minimal. Auto upgrade is handled by looking
through the casts and no new fields are needed for now. New bitcode will
simply have different types for Alias and Aliasee.

One last interesting point in the patch is that replaceAllUsesWith becomes
smart enough to avoid putting a ConstantExpr in the aliasee. This seems better
than checking and updating every caller.

A followup patch will delete getAliasedGlobal now that it is redundant. Another
patch will add support for an explicit offset.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209007 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoDebugInfo: Assume the CU's Subprogram list only contains definitions.
David Blaikie [Fri, 16 May 2014 18:26:53 +0000 (18:26 +0000)]
DebugInfo: Assume the CU's Subprogram list only contains definitions.

DIBuilder maintains this invariant and the current DwarfDebug code could
end up doing weird things if it contained declarations (such as putting
the definition DIE inside a CU that contained the declaration - this
doesn't seem like a good idea, so rather than adding logic to handle
this case we'll just ban in for now & cross that bridge if we come to
it later).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209004 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[ARM64] Increases the Sched Model accuracy for Cortex-A53.
Chad Rosier [Fri, 16 May 2014 17:15:33 +0000 (17:15 +0000)]
[ARM64] Increases the Sched Model accuracy for Cortex-A53.
Patch by Dave Estes <cestes@codeaurora.org>
http://reviews.llvm.org/D3769

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209001 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoInstSimplify: Improve handling of ashr/lshr
David Majnemer [Fri, 16 May 2014 17:14:03 +0000 (17:14 +0000)]
InstSimplify: Improve handling of ashr/lshr

Summary:
Analyze the range of values produced by ashr/lshr cst, %V when it is
being used in an icmp.

Reviewers: nicholas

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D3774

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209000 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoInstSimplify: Optimize using dividend in sdiv
David Majnemer [Fri, 16 May 2014 16:57:04 +0000 (16:57 +0000)]
InstSimplify: Optimize using dividend in sdiv

Summary:
The dividend in an sdiv tells us the largest and smallest possible
results.  Use this fact to optimize comparisons against an sdiv with a
constant dividend.

Reviewers: nicholas

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D3795

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208999 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[ARM64] Fix wrong comment in load/store optimization pass.
Tilmann Scheller [Fri, 16 May 2014 16:50:13 +0000 (16:50 +0000)]
[ARM64] Fix wrong comment in load/store optimization pass.

ldr x1, [x0, #64]
add x0, x0, #64
 ->
ldr x1, [x0], #64

is not a valid transformation, the correct transformation (and what the code actually does) is:

ldr x1, [x0, #64]
add x0, x0, #64
 ->
ldr x1, [x0, #64]!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208998 91177308-0d34-0410-b5e6-96231b3b80d8