oota-llvm.git
9 years agoremove names from comments; NFC
Sanjay Patel [Fri, 9 Jan 2015 16:47:20 +0000 (16:47 +0000)]
remove names from comments; NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225526 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agofix typos; NFC
Sanjay Patel [Fri, 9 Jan 2015 16:35:37 +0000 (16:35 +0000)]
fix typos; NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225525 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agofix typo; NFC
Sanjay Patel [Fri, 9 Jan 2015 16:29:50 +0000 (16:29 +0000)]
fix typo; NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225524 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agomore efficient use of a dyn_cast; no functional change intended
Sanjay Patel [Fri, 9 Jan 2015 16:28:15 +0000 (16:28 +0000)]
more efficient use of a dyn_cast; no functional change intended

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225523 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[PowerPC] Enable late partial unrolling on the POWER7
Hal Finkel [Fri, 9 Jan 2015 15:51:16 +0000 (15:51 +0000)]
[PowerPC] Enable late partial unrolling on the POWER7

The P7 benefits from not have really-small loops so that we either have
multiple dispatch groups in the loop and/or the ability to form more-full
dispatch groups during scheduling. Setting the partial unrolling threshold to
44 seems good, empirically, for the P7. Compared to using no late partial
unrolling, this yields the following test-suite speedups:

SingleSource/Benchmarks/Adobe-C++/simple_types_constant_folding
-66.3253% +/- 24.1975%
SingleSource/Benchmarks/Misc-C++/oopack_v1p8
-44.0169% +/- 29.4881%
SingleSource/Benchmarks/Misc/pi
-27.8351% +/- 12.2712%
SingleSource/Benchmarks/Stanford/Bubblesort
-30.9898% +/- 22.4647%

I've speculatively added a similar setting for the P8. Also, I've noticed that
the unroller does not quite calculate the unrolling factor correctly for really
tiny loops because it neglects to account for the fact that not every loop body
replicant contains an ending branch and counter increment. I'll fix that later.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225522 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[mips] Add comment which explains why we need to change the assembler options before...
Toma Tabacu [Fri, 9 Jan 2015 15:00:30 +0000 (15:00 +0000)]
[mips] Add comment which explains why we need to change the assembler options before and after inline asm blocks. NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225521 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoAssumption that "VectorizedValue" will always be an Instruction is not correct.
Suyog Sarda [Fri, 9 Jan 2015 10:23:48 +0000 (10:23 +0000)]
Assumption that "VectorizedValue" will always be an Instruction is not correct.
It can be a constant or a vector argument.

ex :

define i32 @hadd(<4 x i32> %a) #0 {
entry:
  %vecext = extractelement <4 x i32> %a, i32 0
  %vecext1 = extractelement <4 x i32> %a, i32 1
  %add = add i32 %vecext, %vecext1
  %vecext2 = extractelement <4 x i32> %a, i32 2
  %add3 = add i32 %add, %vecext2
  %vecext4 = extractelement <4 x i32> %a, i32 3
  %add5 = add i32 %add3, %vecext4
  ret i32 %add5
}

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225517 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoARM: add support for R_ARM_ABS16
Saleem Abdulrasool [Fri, 9 Jan 2015 06:57:24 +0000 (06:57 +0000)]
ARM: add support for R_ARM_ABS16

Add support for R_ARM_ABS16 relocation mapping.  Addresses PR22156.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225510 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agotest: add additional test for SVN r225507
Saleem Abdulrasool [Fri, 9 Jan 2015 06:57:18 +0000 (06:57 +0000)]
test: add additional test for SVN r225507

Add an additional test case to ensure that we generate the relocation even if
the thumb target is used.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225509 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoARM: add support for R_ARM_ABS8 relocations
Saleem Abdulrasool [Fri, 9 Jan 2015 05:59:12 +0000 (05:59 +0000)]
ARM: add support for R_ARM_ABS8 relocations

Add support for R_ARM_ABS8 relocation.  Addresses PR22126.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225507 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoRegisterCoalescer: Fix removeCopyByCommutingDef with subreg liveness
Matthias Braun [Fri, 9 Jan 2015 03:01:31 +0000 (03:01 +0000)]
RegisterCoalescer: Fix removeCopyByCommutingDef with subreg liveness

The code that eliminated additional coalescable copies in
removeCopyByCommutingDef() used MergeValueNumberInto() which internally
may merge A into B or B into A. In this case A and B had different Def
points, so we have to reset ValNo.Def to the intended one after merging.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225503 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoRegisterCoalescer: Some cleanup in removeCopyByCommutingDef(), NFC
Matthias Braun [Fri, 9 Jan 2015 03:01:28 +0000 (03:01 +0000)]
RegisterCoalescer: Some cleanup in removeCopyByCommutingDef(), NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225502 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoRegisterCoalescer: No need to set kill flags, they are recompute later anyway
Matthias Braun [Fri, 9 Jan 2015 03:01:26 +0000 (03:01 +0000)]
RegisterCoalescer: No need to set kill flags, they are recompute later anyway

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225501 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoRegisterCoalescer: Turn some impossible conditions into asserts
Matthias Braun [Fri, 9 Jan 2015 03:01:23 +0000 (03:01 +0000)]
RegisterCoalescer: Turn some impossible conditions into asserts

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225500 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoBitcode: Share logic for last instruction, NFC
Duncan P. N. Exon Smith [Fri, 9 Jan 2015 02:51:45 +0000 (02:51 +0000)]
Bitcode: Share logic for last instruction, NFC

Share logic for getting the last instruction emitted.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225499 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoBitcode: Move the DEBUG_LOC record to DEBUG_LOC_OLD
Duncan P. N. Exon Smith [Fri, 9 Jan 2015 02:48:48 +0000 (02:48 +0000)]
Bitcode: Move the DEBUG_LOC record to DEBUG_LOC_OLD

Prepare to simplify the `DebugLoc` record.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225498 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[PowerPC] Add a flag for experimenting with subreg liveness tracking
Hal Finkel [Fri, 9 Jan 2015 02:03:11 +0000 (02:03 +0000)]
[PowerPC] Add a flag for experimenting with subreg liveness tracking

This cannot yet be enabled by default, it causes ~50 miscompiles in the test
suite.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225497 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[PowerPC] Fold [sz]ext with fp_to_int lowering where possible
Hal Finkel [Fri, 9 Jan 2015 01:34:30 +0000 (01:34 +0000)]
[PowerPC] Fold [sz]ext with fp_to_int lowering where possible

On modern cores with lfiw[az]x, we can fold a sign or zero extension from i32
to i64 into the load necessary for an i64 -> fp conversion.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225493 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[DAGCombine] Remainder of fix to r225380 (More FMA folding opportunities)
Hal Finkel [Fri, 9 Jan 2015 01:29:29 +0000 (01:29 +0000)]
[DAGCombine] Remainder of fix to r225380 (More FMA folding opportunities)

As pointed out by Aditya (and Owen), when we elide an FP extend to form an FMA,
we need to extend the incoming operands so that the resulting node will really
be legal. This is currently enabled only for PowerPC, and it happens to work
there regardless, but this should fix the functionality for everyone else
should anyone else wish to use it.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225492 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[x86] Add a flag to control the vector shuffle legality predicates that
Chandler Carruth [Fri, 9 Jan 2015 01:24:36 +0000 (01:24 +0000)]
[x86] Add a flag to control the vector shuffle legality predicates that
complements the new vector shuffle lowering code path. This flag,
naturally, is *off* because we've not tested or evaluated the results of
this at all. However, the flag will make it much easier to evaluate
whether we can be this aggressive and whether there are missing vector
shuffle lowering optimizations.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225491 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoCleaup ValueHandle to no longer keep a PointerIntPair for the Value*.
Chandler Carruth [Fri, 9 Jan 2015 00:48:47 +0000 (00:48 +0000)]
Cleaup ValueHandle to no longer keep a PointerIntPair for the Value*.
This was used previously for metadata but is no longer needed there. Not
doing this simplifies ValueHandle and will make it easier to fix things
like AssertingVH's DenseMapInfo.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225487 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoPartial fix to r225380 (More FMA folding opportunities)
Hal Finkel [Fri, 9 Jan 2015 00:45:54 +0000 (00:45 +0000)]
Partial fix to r225380 (More FMA folding opportunities)

As pointed out by Aditya (and Owen), there are two things wrong with this code.
First, it adds patterns which elide FP extends when forming FMAs, and that might
not be profitable on all targets (it belongs behind the pre-existing
aggressive-FMA-formation flag). This is fixed by this change.

Second, the resulting nodes might have operands of different types (the
extensions need to be re-added). That will be fixed in the follow-up commit.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225485 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[REFACTOR] Push logic from MemDepPrinter into getNonLocalPointerDependency
Philip Reames [Fri, 9 Jan 2015 00:26:45 +0000 (00:26 +0000)]
[REFACTOR] Push logic from MemDepPrinter into getNonLocalPointerDependency

Previously, MemDepPrinter handled volatile and unordered accesses without involving MemoryDependencyAnalysis.  By making a slight tweak to the documented interface - which is respected by both callers - we can move this responsibility to MDA for the benefit of any future callers.  This is basically just cleanup.

In the future, we may decide to extend MDA's non local dependency analysis to return useful results for ordered or volatile loads.  I believe (but have not really checked in detail) that local dependency analyis does get useful results for ordered, but not volatile, loads.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225483 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoReleaseNotes.rst: these are for 3.6
Hans Wennborg [Fri, 9 Jan 2015 00:21:26 +0000 (00:21 +0000)]
ReleaseNotes.rst: these are for 3.6

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225482 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[Refactor] Have getNonLocalPointerDependency take the query instruction
Philip Reames [Fri, 9 Jan 2015 00:04:22 +0000 (00:04 +0000)]
[Refactor] Have getNonLocalPointerDependency take the query instruction

Previously, MemoryDependenceAnalysis::getNonLocalPointerDependency was taking a list of properties about the instruction being queried. Since I'm about to need one more property to be passed down through the infrastructure - I need to know a query instruction is non-volatile in an inner helper - fix the interface once and for all.

I also added some assertions and behaviour clarifications around volatile and ordered field accesses. At the moment, this is mostly to document expected behaviour. The only non-standard instructions which can currently reach this are atomic, but unordered, loads and stores. Neither ordered or volatile accesses can reach here.

The call in GVN is protected by an isSimple check when it first considers the load. The calls in MemDepPrinter are protected by isUnordered checks. Both utilities also check isVolatile for loads and stores.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225481 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoLangRef: Add usage points for distinct MDNodes
Duncan P. N. Exon Smith [Thu, 8 Jan 2015 23:50:26 +0000 (23:50 +0000)]
LangRef: Add usage points for distinct MDNodes

Omission pointed out by Sean Silva!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225479 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoIR: Drop TODO now that PR22111 is finished
Duncan P. N. Exon Smith [Thu, 8 Jan 2015 22:43:19 +0000 (22:43 +0000)]
IR: Drop TODO now that PR22111 is finished

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225477 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoUtils: Keep distinct MDNodes distinct in MapMetadata()
Duncan P. N. Exon Smith [Thu, 8 Jan 2015 22:42:30 +0000 (22:42 +0000)]
Utils: Keep distinct MDNodes distinct in MapMetadata()

Create new copies of distinct `MDNode`s instead of following the
uniquing `MDNode` logic.

Just like self-references (or other cycles), `MapMetadata()` creates a
new node.  In practice most calls use `RF_NoModuleLevelChanges`, in
which case nothing is duplicated anyway.

Part of PR22111.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225476 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoIR: Add 'distinct' MDNodes to bitcode and assembly
Duncan P. N. Exon Smith [Thu, 8 Jan 2015 22:38:29 +0000 (22:38 +0000)]
IR: Add 'distinct' MDNodes to bitcode and assembly

Propagate whether `MDNode`s are 'distinct' through the other types of IR
(assembly and bitcode).  This adds the `distinct` keyword to assembly.

Currently, no one actually calls `MDNode::getDistinct()`, so these nodes
only get created for:

  - self-references, which are never uniqued, and
  - nodes whose operands are replaced that hit a uniquing collision.

The concept of distinct nodes is still not quite first-class, since
distinct-ness doesn't yet survive across `MapMetadata()`.

Part of PR22111.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225474 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoremove function names from comments; NFC
Sanjay Patel [Thu, 8 Jan 2015 22:36:56 +0000 (22:36 +0000)]
remove function names from comments; NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225473 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[PowerPC] Mark all instructions as non-cheap for MachineLICM
Hal Finkel [Thu, 8 Jan 2015 22:11:49 +0000 (22:11 +0000)]
[PowerPC] Mark all instructions as non-cheap for MachineLICM

MachineLICM uses a callback named hasLowDefLatency to determine if an
instruction def operand has a 'low' latency. If all relevant operands have a
'low' latency, the instruction is considered too cheap to hoist out of loops
even in low-register-pressure situations. On PowerPC cores, both the embedded
cores and the others, there is no reason to believe that this is a good choice:
all instructions have a cost inside a loop, and hoisting them when not limited
by register pressure is a reasonable default.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225471 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[MachineLICM] A command-line option to hoist even cheap instructions
Hal Finkel [Thu, 8 Jan 2015 22:10:48 +0000 (22:10 +0000)]
[MachineLICM] A command-line option to hoist even cheap instructions

Add a command-line option to enable hoisting even cheap instructions (in
low-register-pressure situations). This is turned off by default, but has
proved useful for testing purposes.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225470 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoCodeGen: Use handy new-fangled post-increment, NFC
Duncan P. N. Exon Smith [Thu, 8 Jan 2015 21:07:55 +0000 (21:07 +0000)]
CodeGen: Use handy new-fangled post-increment, NFC

Drive-by cleanup; I noticed this when reviewing the patch that became
r225466.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225468 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[ARM] Fix a bug in constant island pass that was triggering an assertion.
Akira Hatanaka [Thu, 8 Jan 2015 20:44:50 +0000 (20:44 +0000)]
[ARM] Fix a bug in constant island pass that was triggering an assertion.

The assert was being triggered when the distance between a constant pool entry
and its user exceeded the maximally allowed distance after thumb2 branch
shortening. A padding was inserted after a thumb2 branch instruction was shrunk,
which caused the user to be out of range. This is wrong as the padding should
have been inserted by the layout algorithm so that the distance between two
instructions doesn't grow later during thumb2 instruction optimization.

This commit fixes the code in ARMConstantIslands::createNewWater to call
computeBlockSize and set BasicBlock::Unalign when a branch instruction is
inserted to create new water after a basic block. A non-zero Unalign causes
the worst-case padding to be inserted when adjustBBOffsetsAfter is called to
recompute the basic block offsets.

rdar://problem/19130476

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225467 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoCodeGen: Use range-based for loops, NFC
Duncan P. N. Exon Smith [Thu, 8 Jan 2015 20:44:33 +0000 (20:44 +0000)]
CodeGen: Use range-based for loops, NFC

Patch by Ramkumar Ramachandra!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225466 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoFix fcmp + fabs instcombines when using the intrinsic
Matt Arsenault [Thu, 8 Jan 2015 20:09:34 +0000 (20:09 +0000)]
Fix fcmp + fabs instcombines when using the intrinsic

This was only handling the libcall. This is another example
of why only the intrinsic should ever be used when it exists.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225465 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoThe Kaleidoscope tutorial should be using "mcjit" for the library,
Eric Christopher [Thu, 8 Jan 2015 19:07:01 +0000 (19:07 +0000)]
The Kaleidoscope tutorial should be using "mcjit" for the library,
"jit" doesn't exist anymore.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225462 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[MCJIT] Remove a few redundant MCJIT tests, and drop the extraneous datalayout
Lang Hames [Thu, 8 Jan 2015 18:52:15 +0000 (18:52 +0000)]
[MCJIT] Remove a few redundant MCJIT tests, and drop the extraneous datalayout
strings from the copies that remain.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225460 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoMake the TargetMachine in MipsSubtarget a reference rather
Eric Christopher [Thu, 8 Jan 2015 18:18:57 +0000 (18:18 +0000)]
Make the TargetMachine in MipsSubtarget a reference rather
than a pointer to make unifying code a bit easier.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225459 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoUpdate include - this class doesn't use the target machine, but
Eric Christopher [Thu, 8 Jan 2015 18:18:54 +0000 (18:18 +0000)]
Update include - this class doesn't use the target machine, but
only the subtarget.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225458 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoFix a couple of odd formatting issues.
Eric Christopher [Thu, 8 Jan 2015 18:18:53 +0000 (18:18 +0000)]
Fix a couple of odd formatting issues.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225457 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoThis routine is in InstrInfo, there's no need to access it again.
Eric Christopher [Thu, 8 Jan 2015 18:18:50 +0000 (18:18 +0000)]
This routine is in InstrInfo, there's no need to access it again.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225456 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[X86] Reflow comment. NFC.
Ahmed Bougacha [Thu, 8 Jan 2015 17:49:48 +0000 (17:49 +0000)]
[X86] Reflow comment. NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225455 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoclang-format. NFC.
Rafael Espindola [Thu, 8 Jan 2015 16:25:01 +0000 (16:25 +0000)]
clang-format. NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225454 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoMake this test a bit stricter.
Rafael Espindola [Thu, 8 Jan 2015 16:11:18 +0000 (16:11 +0000)]
Make this test a bit stricter.

It now checks for the end of the line or the opening '{'.
While at it, remove empty comments.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225451 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoAdd saving and restoring of r30 to the prologue and epilogue, respectively
Justin Hibbits [Thu, 8 Jan 2015 15:47:19 +0000 (15:47 +0000)]
Add saving and restoring of r30 to the prologue and epilogue, respectively

Summary: The PIC additions didn't update the prologue and epilogue code to save and restore r30 (PIC base register).  This does that.

Test Plan: Tests updated.

Reviewers: hfinkel

Reviewed By: hfinkel

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D6876

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225450 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoExplicitly handle LinkOnceODRAutoHideLinkage. NFC. We already have a test.
Rafael Espindola [Thu, 8 Jan 2015 15:39:50 +0000 (15:39 +0000)]
Explicitly handle LinkOnceODRAutoHideLinkage. NFC. We already have a test.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225449 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoUpdate naming style and clang-format. NFC.
Rafael Espindola [Thu, 8 Jan 2015 15:36:32 +0000 (15:36 +0000)]
Update naming style and clang-format. NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225448 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoFix large stack alignment codegen for ARM and Thumb2 targets
Kristof Beyls [Thu, 8 Jan 2015 15:09:14 +0000 (15:09 +0000)]
Fix large stack alignment codegen for ARM and Thumb2 targets

This partially fixes PR13007 (ARM CodeGen fails with large stack
alignment): for ARM and Thumb2 targets, but not for Thumb1, as it
seems stack alignment for Thumb1 targets hasn't been supported at
all.

Producing an aligned stack pointer is done by zero-ing out the lower
bits of the stack pointer. The BIC instruction was used for this.
However, the immediate field of the BIC instruction only allows to
encode an immediate that can zero out up to a maximum of the 8 lower
bits. When a larger alignment is requested, a BIC instruction cannot
be used; llvm was silently producing incorrect code in this case.

This commit fixes code generation for large stack aligments by
using the BFC instruction instead, when the BFC instruction is
available.  When not, it uses 2 instructions: a right shift,
followed by a left shift to zero out the lower bits.

The lowering of ARM::Int_eh_sjlj_dispatchsetup still has code
that unconditionally uses BIC to realign the stack pointer, so it
very likely has the same problem. However, I wasn't able to
produce a test case for that. This commit adds an assert so that
the compiler will fail the assert instead of silently generating
wrong code if this is ever reached.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225446 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoR600/SI: Remove SIISelLowering::legalizeOperands()
Tom Stellard [Thu, 8 Jan 2015 15:08:17 +0000 (15:08 +0000)]
R600/SI: Remove SIISelLowering::legalizeOperands()

Its functionality has been replaced by calling
SIInstrInfo::legalizeOperands() from
SIISelLowering::AdjstInstrPostInstrSelection() and running the
SIFoldOperands and SIShrinkInstructions passes.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225445 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoMasked Load/Store - fixed a bug in type legalization.
Elena Demikhovsky [Thu, 8 Jan 2015 12:29:19 +0000 (12:29 +0000)]
Masked Load/Store - fixed a bug in type legalization.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225441 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoFix a think-o in the test for r225438.
Michael Kuperstein [Thu, 8 Jan 2015 12:05:02 +0000 (12:05 +0000)]
Fix a think-o in the test for r225438.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225440 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoFix include ordering, NFC.
Michael Kuperstein [Thu, 8 Jan 2015 11:59:43 +0000 (11:59 +0000)]
Fix include ordering, NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225439 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[X86] Don't try to generate direct calls to TLS globals
Michael Kuperstein [Thu, 8 Jan 2015 11:50:58 +0000 (11:50 +0000)]
[X86] Don't try to generate direct calls to TLS globals

The call lowering assumes that if the callee is a global, we want to emit a direct call.
This is correct for regular globals, but not for TLS ones.

Differential Revision: http://reviews.llvm.org/D6862

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225438 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoMove SPAdj logic from PEI into the targets (NFC)
Michael Kuperstein [Thu, 8 Jan 2015 11:04:38 +0000 (11:04 +0000)]
Move SPAdj logic from PEI into the targets (NFC)

PEI tries to keep track of how much starting or ending a call sequence adjusts the stack pointer by, so that it can resolve frame-index references. Currently, it takes a very simplistic view of how SP adjustments are done - both FrameStartOpcode and FrameDestroyOpcode adjust it exactly by the amount written in its first argument.

This view is in fact incorrect for some targets (e.g. due to stack re-alignment, or because it may want to adjust the stack pointer in multiple steps). However, that doesn't cause breakage, because most targets (the only in-tree exception appears to be 32-bit ARM) rely on being able to simplify the call frame pseudo-instructions earlier, so this code is never hit.

Moving the computation into TargetInstrInfo allows targets to override the way the adjustment is computed if they need to have a non-zero SPAdj.

Differential Revision: http://reviews.llvm.org/D6863

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225437 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoFix test case I missed in r225432.
Craig Topper [Thu, 8 Jan 2015 07:57:27 +0000 (07:57 +0000)]
Fix test case I missed in r225432.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225434 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[X86] Don't print 'dword ptr' or 'qword ptr' on the operand to some of the LEA varian...
Craig Topper [Thu, 8 Jan 2015 07:41:30 +0000 (07:41 +0000)]
[X86] Don't print 'dword ptr' or 'qword ptr' on the operand to some of the LEA variants in Intel syntax. The memory operand is inherently unsized.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225432 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoRevert "Reapply: Teach SROA how to update debug info for fragmented variables."
Adrian Prantl [Thu, 8 Jan 2015 02:02:00 +0000 (02:02 +0000)]
Revert "Reapply: Teach SROA how to update debug info for fragmented variables."

This reverts commit r225379 while investigating an assertion failure reported
by Alexey.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225424 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[RegAllocGreedy] Introduce a late pass to repair broken hints.
Quentin Colombet [Thu, 8 Jan 2015 01:16:39 +0000 (01:16 +0000)]
[RegAllocGreedy] Introduce a late pass to repair broken hints.

A broken hint is a copy where both ends are assigned different colors. When a
variable gets evicted in the neighborhood of such copies, it is likely we can
reconcile some of them.

** Context **

Copies are inserted during the register allocation via splitting. These split
points are required to relax the constraints on the allocation problem. When
such a point is inserted, both ends of the copy would not share the same color
with respect to the current allocation problem. When variables get evicted,
the allocation problem becomes different and some split point may not be
required anymore. However, the related variables may already have been colored.

This usually shows up in the assembly with pattern like this:
def A
...
save A to B
def A
use A
restore A from B
...
use B

Whereas we could simply have done:
def B
...
def A
use A
...
use B

** Proposed Solution **

A variable having a broken hint is marked for late recoloring if and only if
selecting a register for it evict another variable. Indeed, if no eviction
happens this is pointless to look for recoloring opportunities as it means the
situation was the same as the initial allocation problem where we had to break
the hint.

Finally, when everything has been allocated, we look for recoloring
opportunities for all the identified candidates.
The recoloring is performed very late to rely on accurate copy cost (all
involved variables are allocated).
The recoloring is simple unlike the last change recoloring. It propagates the
color of the broken hint to all its copy-related variables. If the color is
available for them, the recoloring uses it, otherwise it gives up on that hint
even if a more complex coloring would have worked.

The recoloring happens only if it is profitable. The profitability is evaluated
using the expected frequency of the copies of the currently recolored variable
with a) its current color and b) with the target color. If a) is greater or
equal than b), then it is profitable and the recoloring happen.

** Example **

Consider the following example:
BB1:
  a =
  b =
BB2:
  ...
   = b
   = a
Let us assume b gets split:
BB1:
  a =
  b =
BB2:
  c = b
  ...
  d = c
  = d
  = a
Because of how the allocation work, b, c, and d may be assigned different
colors. Now, if a gets evicted to make room for c, assuming b and d were
assigned to something different than a.
We end up with:
BB1:
  a =
  st a, SpillSlot
  b =
BB2:
  c = b
  ...
  d = c
  = d
  e = ld SpillSlot
  = e
This is likely that we can assign the same register for b, c, and d,
getting rid of 2 copies.

** Performances **

Both ARM64 and x86_64 show performance improvements of up to 3% for the
llvm-testsuite + externals with Os and O3. There are a few regressions too that
comes from the (in)accuracy of the block frequency estimate.

<rdar://problem/18312047>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225422 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[SelectionDAG] Allow targets to specify legality of extloads' result
Ahmed Bougacha [Thu, 8 Jan 2015 00:51:32 +0000 (00:51 +0000)]
[SelectionDAG] Allow targets to specify legality of extloads' result
type (in addition to the memory type).

The *LoadExt* legalization handling used to only have one type, the
memory type.  This forced users to assume that as long as the extload
for the memory type was declared legal, and the result type was legal,
the whole extload was legal.

However, this isn't always the case.  For instance, on X86, with AVX,
this is legal:
    v4i32 load, zext from v4i8
but this isn't:
    v4i64 load, zext from v4i8
Whereas v4i64 is (arguably) legal, even without AVX2.

Note that the same thing was done a while ago for truncstores (r46140),
but I assume no one needed it yet for extloads, so here we go.

Calls to getLoadExtAction were changed to add the value type, found
manually in the surrounding code.

Calls to setLoadExtAction were mechanically changed, by wrapping the
call in a loop, to match previous behavior.  The loop iterates over
the MVT subrange corresponding to the memory type (FP vectors, etc...).
I also pulled neighboring setTruncStoreActions into some of the loops;
those shouldn't make a difference, as the additional types are illegal.
(e.g., i128->i1 truncstores on PPC.)

No functional change intended.

Differential Revision: http://reviews.llvm.org/D6532

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225421 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoRemove empty statement. No functionality change.
Nick Lewycky [Thu, 8 Jan 2015 00:47:03 +0000 (00:47 +0000)]
Remove empty statement. No functionality change.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225420 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoX86: VZeroUpperInserter: shortcut should not trigger if we have any function live...
Matthias Braun [Thu, 8 Jan 2015 00:33:48 +0000 (00:33 +0000)]
X86: VZeroUpperInserter: shortcut should not trigger if we have any function live-ins.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225419 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoRun clang-format on tools/llvm-objdump/MachODump.cpp again as some of my
Kevin Enderby [Thu, 8 Jan 2015 00:25:24 +0000 (00:25 +0000)]
Run clang-format on tools/llvm-objdump/MachODump.cpp again as some of my
previous changes got in with incorrect formatting. No functional change.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225417 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoRegisterCoalescer: Do not remove IMPLICIT_DEFS if they are required for subranges.
Matthias Braun [Thu, 8 Jan 2015 00:21:23 +0000 (00:21 +0000)]
RegisterCoalescer: Do not remove IMPLICIT_DEFS if they are required for subranges.

The register coalescer used to remove implicit_defs when they are
covered by the main range anyway. With subreg liveness tracking we can't
do that anymore in places where the IMPLICIT_DEF is required as begin of
a subregister liverange.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225416 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoRegisterCoalescer: Fix valuesIdentical() in some subrange merge cases.
Matthias Braun [Wed, 7 Jan 2015 23:58:38 +0000 (23:58 +0000)]
RegisterCoalescer: Fix valuesIdentical() in some subrange merge cases.

I got confused and assumed SrcIdx/DstIdx of the CoalescerPair is a
subregister index in SrcReg/DstReg, but they are actually subregister
indices of the coalesced register that get you back to SrcReg/DstReg
when applied.

Fixed the bug, improved comments and simplified code accordingly.

Testcase by Tom Stellard!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225415 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoLiveInterval: Implement feedback by Quentin Colombet.
Matthias Braun [Wed, 7 Jan 2015 23:35:11 +0000 (23:35 +0000)]
LiveInterval: Implement feedback by Quentin Colombet.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225413 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[GC] improve testing around gc.relocate and fix a test
Philip Reames [Wed, 7 Jan 2015 22:48:01 +0000 (22:48 +0000)]
[GC] improve testing around gc.relocate and fix a test

Patch by: Ramkumar Ramachandra <artagnon@gmail.com>

"This patch started out as an exploration of gc.relocate, and an attempt
to write a simple test in call-lowering. I then noticed that the
arguments of gc.relocate were not checked fully, so I went in and fixed
a few things. Finally, the most important outcome of this patch is that
my new error handling code caught a bug in a callsite in
stackmap-format."

Differential Revision: http://reviews.llvm.org/D6824

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225412 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[CodeGen] Add MVT::isValid to replace manual validity checks. NFC.
Ahmed Bougacha [Wed, 7 Jan 2015 22:47:46 +0000 (22:47 +0000)]
[CodeGen] Add MVT::isValid to replace manual validity checks. NFC.

Now that we have MVT::FIRST_VALUETYPE (r225362), we can provide a method
checking that the MVT is valid, that is, it's in
  [FIRST_VALUETYPE, LAST_VALUETYPE[.
This commit also uses it in a few asserts, that would previously accept
invalid MVTs, such as the default constructed -1.  In that case,
the code following those asserts would do an out-of-bounds array access.
Using MVT::isValid, those assertions fail as expected when passed
invalid MVTs.
It feels clunky to have such a validity checking function, but it's
at least better than the alternative of broken manual checks.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225411 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoR600/SI: Commute instructions to enable more folding opportunities
Tom Stellard [Wed, 7 Jan 2015 22:44:19 +0000 (22:44 +0000)]
R600/SI: Commute instructions to enable more folding opportunities

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225410 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoIR: Add MDNode::getDistinct()
Duncan P. N. Exon Smith [Wed, 7 Jan 2015 22:24:46 +0000 (22:24 +0000)]
IR: Add MDNode::getDistinct()

Allow distinct `MDNode`s to be explicitly created.  There's no way (yet)
of representing their distinctness in assembly/bitcode, however, so this
still isn't first-class.

Part of PR22111.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225406 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoR600/SI: Only fold immediates that have one use
Tom Stellard [Wed, 7 Jan 2015 22:18:27 +0000 (22:18 +0000)]
R600/SI: Only fold immediates that have one use

Folding the same immediate into multiple instruction will increase
program size, which can hurt performance.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225405 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoTest commit.
Sean Silva [Wed, 7 Jan 2015 22:07:33 +0000 (22:07 +0000)]
Test commit.

Hopefully this one won't kill the git mirror...

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225404 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoIR: Add MDNode::isDistinct()
Duncan P. N. Exon Smith [Wed, 7 Jan 2015 21:35:38 +0000 (21:35 +0000)]
IR: Add MDNode::isDistinct()

Add API to indicate whether an `MDNode` is distinct.  A distinct node is
not stored in the MDNode uniquing tables, and will never be returned by
`MDNode::get()`.

Although distinct nodes are only currently created by uniquing
collisions (when operands change), PR22111 will allow these nodes to be
explicitly created.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225401 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[LangRef] PR22118: Hyphen is allowed in IR identifiers.
Sean Silva [Wed, 7 Jan 2015 21:35:14 +0000 (21:35 +0000)]
[LangRef] PR22118: Hyphen is allowed in IR identifiers.

E.g. %-foo and %fo-o.

Thanks to eagle-eyed reporter Tomas Brukner.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225400 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoUpdate a comment.
Adrian Prantl [Wed, 7 Jan 2015 21:35:13 +0000 (21:35 +0000)]
Update a comment.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225399 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoLinker: Don't use MDNode::replaceOperandWith()
Duncan P. N. Exon Smith [Wed, 7 Jan 2015 21:32:27 +0000 (21:32 +0000)]
Linker: Don't use MDNode::replaceOperandWith()

`MDNode::replaceOperandWith()` changes all instances of metadata.  Stop
using it when linking module flags, since (due to uniquing) the flag
values could be used by other metadata.

Instead, use new API `NamedMDNode::setOperand()` to update the reference
directly.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225397 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoXFAIL several MCJIT EH tests under ASan and MSan bootstrap.
Alexey Samsonov [Wed, 7 Jan 2015 21:27:26 +0000 (21:27 +0000)]
XFAIL several MCJIT EH tests under ASan and MSan bootstrap.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225393 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[CodeGen] Use MVT iterator_ranges in legality loops. NFC intended.
Ahmed Bougacha [Wed, 7 Jan 2015 21:27:10 +0000 (21:27 +0000)]
[CodeGen] Use MVT iterator_ranges in legality loops. NFC intended.

A few loops do trickier things than just iterating on an MVT subset,
so I'll leave them be for now.
Follow-up of r225387.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225392 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[CodeGen] Add iterator_range for the MVT::SimpleValueType enum.
Ahmed Bougacha [Wed, 7 Jan 2015 21:15:38 +0000 (21:15 +0000)]
[CodeGen] Add iterator_range for the MVT::SimpleValueType enum.

This commit adds a simple iterator over that enum, and a few
functions to create iterator ranges over the most common types.

Differential Revision: http://reviews.llvm.org/D6537

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225387 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoFix uninitialized memory read in llvm-dsymutil for the second time.
Alexey Samsonov [Wed, 7 Jan 2015 21:13:30 +0000 (21:13 +0000)]
Fix uninitialized memory read in llvm-dsymutil for the second time.

This was already fixed by r224481, but apparently was accidentally
reverted in r225207.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225386 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoAdd a test that would have found the issue in r224935.
Rafael Espindola [Wed, 7 Jan 2015 21:10:25 +0000 (21:10 +0000)]
Add a test that would have found the issue in r224935.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225385 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[git] Mark the llgo directory in the LLVM gitignore.
Chandler Carruth [Wed, 7 Jan 2015 21:08:54 +0000 (21:08 +0000)]
[git] Mark the llgo directory in the LLVM gitignore.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225384 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoSlightly refactor things for llvm-objdump and the -macho option so it can be used...
Kevin Enderby [Wed, 7 Jan 2015 21:02:18 +0000 (21:02 +0000)]
Slightly refactor things for llvm-objdump and the -macho option so it can be used with
options other than just -disassemble so that universal files can be used with other
options combined with -arch options.

No functional change to existing options and use.  One test case added for the
additional functionality with a universal file an a -arch option.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225383 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoR600/SI: Remove VReg_32 register class
Tom Stellard [Wed, 7 Jan 2015 20:59:25 +0000 (20:59 +0000)]
R600/SI: Remove VReg_32 register class

Use VGPR_32 register class instead.  These two register classes were
identical and having separate classes was causing
SIInstrInfo::isLegalOperands() to be overly conservative in some cases.

This change is necessary to prevent future paches from missing a folding
opportunity in fneg-fabs.ll.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225382 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoMore FMA folding opportunities.
Olivier Sallenave [Wed, 7 Jan 2015 20:54:17 +0000 (20:54 +0000)]
More FMA folding opportunities.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225380 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoReapply: Teach SROA how to update debug info for fragmented variables.
Adrian Prantl [Wed, 7 Jan 2015 20:52:22 +0000 (20:52 +0000)]
Reapply: Teach SROA how to update debug info for fragmented variables.

The two buildbot failures were addressed in LLVM r225378 and CFE r225359.

This rapplies commit 225272 without modifications.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225379 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoDebug info: Allow aggregate types to be described by constants.
Adrian Prantl [Wed, 7 Jan 2015 20:48:58 +0000 (20:48 +0000)]
Debug info: Allow aggregate types to be described by constants.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225378 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[Hexagon] Fix 225372 USR register is not fully complete. Removing Uses = [USR] maint...
Colin LeMahieu [Wed, 7 Jan 2015 20:43:38 +0000 (20:43 +0000)]
[Hexagon] Fix 225372 USR register is not fully complete.  Removing Uses = [USR] maintains existing functionality to old instructions without encodings.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225377 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[Hexagon] Adding floating point classification and creation.
Colin LeMahieu [Wed, 7 Jan 2015 20:28:57 +0000 (20:28 +0000)]
[Hexagon] Adding floating point classification and creation.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225374 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoR600/SI: Add a V_MOV_B64 pseudo instruction
Tom Stellard [Wed, 7 Jan 2015 20:27:25 +0000 (20:27 +0000)]
R600/SI: Add a V_MOV_B64 pseudo instruction

This is used to simplify the SIFoldOperands pass and make it easier to
fold immediates.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225373 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[Hexagon] Adding encodings for v5 floating point instructions.
Colin LeMahieu [Wed, 7 Jan 2015 20:24:09 +0000 (20:24 +0000)]
[Hexagon] Adding encodings for v5 floating point instructions.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225372 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[Hexagon] Adding encoding for popcount, fastcorner, dword asr with rounding.
Colin LeMahieu [Wed, 7 Jan 2015 20:07:28 +0000 (20:07 +0000)]
[Hexagon] Adding encoding for popcount, fastcorner, dword asr with rounding.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225371 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoR600/SI: Teach SIFoldOperands to split 64-bit constants when folding
Tom Stellard [Wed, 7 Jan 2015 19:56:17 +0000 (19:56 +0000)]
R600/SI: Teach SIFoldOperands to split 64-bit constants when folding

This allows folding of sequences like:

s[0:1] = s_mov_b64 4
v_add_i32 v0, s0, v0
v_addc_u32 v1, s1, v1

into

v_add_i32 v0, 4, v0
v_add_i32 v1, 0, v1

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225369 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoTest commit
Olivier Sallenave [Wed, 7 Jan 2015 19:45:17 +0000 (19:45 +0000)]
Test commit

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225368 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[X86] Fix 512->256 typo in comments. NFC.
Ahmed Bougacha [Wed, 7 Jan 2015 19:38:50 +0000 (19:38 +0000)]
[X86] Fix 512->256 typo in comments. NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225367 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoAdd a missing file from 225365
Philip Reames [Wed, 7 Jan 2015 19:13:28 +0000 (19:13 +0000)]
Add a missing file from 225365

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225366 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoIntroduce an example statepoint GC strategy
Philip Reames [Wed, 7 Jan 2015 19:07:50 +0000 (19:07 +0000)]
Introduce an example statepoint GC strategy

This change includes the most basic possible GCStrategy for a GC which is using the statepoint lowering code. At the moment, this GCStrategy doesn't really do much - aside from actually generate correct stackmaps that is - but I went ahead and added a few extra correctness checks as proof of concept. It's mostly here to provide documentation on how to do one, and to provide a point for various optimization legality hooks I'd like to add going forward. (For context, see the TODOs in InstCombine around gc.relocate.)

Most of the validation logic added here as proof of concept will soon move in to the Verifier.  That move is dependent on http://reviews.llvm.org/D6811

There was discussion in the review thread about addrspace(1) being reserved for something.  I'm going to follow up on a seperate llvmdev thread.  If needed, I'll update all the code at once.

Note that I am deliberately not making a GCStrategy required to use gc.statepoints with this change. I want to give folks out of tree - including myself - a chance to migrate. In a week or two, I'll make having a GCStrategy be required for gc.statepoints. To this end, I added the gc tag to one of the test cases but not others.

Differential Revision: http://reviews.llvm.org/D6808

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225365 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[CodeGen] Add MVT::FIRST_VALUETYPE to avoid explicit 0. NFC.
Ahmed Bougacha [Wed, 7 Jan 2015 18:39:00 +0000 (18:39 +0000)]
[CodeGen] Add MVT::FIRST_VALUETYPE to avoid explicit 0. NFC.

Many places reference MVT::LAST_VALUETYPE when iterating over all
valid MVTs, but they usually start with 0.
With FIRST_VALUETYPE, we can avoid explicit constants when we really
should be using MVT::SimpleValueType.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225362 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoX86: Allow the stack probe size to be configurable per function
David Majnemer [Wed, 7 Jan 2015 18:14:07 +0000 (18:14 +0000)]
X86: Allow the stack probe size to be configurable per function

LLVM emits stack probes on Windows targets to ensure that the stack is
correctly accessed.  However, the amount of stack allocated before
emitting such a probe is hardcoded to 4096.

It is desirable to have this be configurable so that a function might
opt-out of stack probes.  Our level of granularity is at the function
level instead of, say, the module level to permit proper generation of
code after LTO.

Patch by Andrew H!

N.B.  The inliner needs to be updated to properly consider what happens
after inlining a function with a specific stack-probe-size into another
function with a different stack-probe-size.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225360 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoR600/SI: Refactor SIFoldOperands to simplify immediate folding
Tom Stellard [Wed, 7 Jan 2015 17:42:16 +0000 (17:42 +0000)]
R600/SI: Refactor SIFoldOperands to simplify immediate folding

This will make a future patch much less intrusive.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225358 91177308-0d34-0410-b5e6-96231b3b80d8