oota-llvm.git
10 years agoR600/SI: Fix 64-bit bit ops that require the VALU.
Matt Arsenault [Mon, 24 Mar 2014 20:08:05 +0000 (20:08 +0000)]
R600/SI: Fix 64-bit bit ops that require the VALU.

Try to match scalar and first like the other instructions.
Expand 64-bit ands to a pair of 32-bit ands since that is not
available on the VALU.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204660 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoIn Release modes, Visual Studio complains that the Operator destructor in User.cpp
Yaron Keren [Mon, 24 Mar 2014 19:48:13 +0000 (19:48 +0000)]
In Release modes, Visual Studio complains that the Operator destructor in User.cpp
never returns, which is true by design.

Initially assumed that the reason is llvm_unreachable being dependent on NDEBUG.

However, even if llvm_unreachable is replaced by __assume(false), VC still warns in
Release modes but not in Debug modes...

The real reason turned out to be optimization flags.
With /Od in Debug modes the warning is not issued whereas with /O1 it is.

I could not find any documentation to this effect, but it is reproducable:

Try compiling http://msdn.microsoft.com/en-us/library/khwfyc5d(v=vs.90).aspx
with /O1 and then with /Od.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204659 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600: Implement isNarrowingProfitable.
Matt Arsenault [Mon, 24 Mar 2014 19:43:31 +0000 (19:43 +0000)]
R600: Implement isNarrowingProfitable.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204658 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600/SI: Move splitting 64-bit immediates to separate function.
Matt Arsenault [Mon, 24 Mar 2014 18:26:52 +0000 (18:26 +0000)]
R600/SI: Move splitting 64-bit immediates to separate function.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204651 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdding some very nascent information about the clang tablegen backends, with a promis...
Aaron Ballman [Mon, 24 Mar 2014 18:18:31 +0000 (18:18 +0000)]
Adding some very nascent information about the clang tablegen backends, with a promise to add more information later.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204635 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[PowerPC] Generate little-endian object files
Ulrich Weigand [Mon, 24 Mar 2014 18:16:09 +0000 (18:16 +0000)]
[PowerPC] Generate little-endian object files

As a first step towards real little-endian code generation, this patch
changes the PowerPC MC layer to actually generate little-endian object
files.  This involves passing the little-endian flag through the various
layers, including down to createELFObjectWriter so we actually get basic
little-endian ELF objects, emitting instructions in little-endian order,
and handling fixups and relocations as appropriate for little-endian.

The bulk of the patch is to update most test cases in test/MC/PowerPC
to verify both big- and little-endian encodings.  (The only test cases
*not* updated are those that create actual big-endian ABI code, like
the TLS tests.)

Note that while the object files are now little-endian, the generated
code itself is not yet updated, in particular, it still does not adhere
to the ELFv2 ABI.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204634 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[X86][ISelDAG] Add missing fallback patterns for avx2 broadcast instructions.
Quentin Colombet [Mon, 24 Mar 2014 17:54:19 +0000 (17:54 +0000)]
[X86][ISelDAG] Add missing fallback patterns for avx2 broadcast instructions.
Those patterns are used when the load cannot be folded into the related broadcast
during the select phase.
This happens when the load gets additional uses that were not anticipated during
the previous lowering phases (constant vector to constant load, then constant
load reused) or when selection DAG is not able to prove that folding the load
will not create a cycle in the DAG.

<rdar://problem/16074331>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204631 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600/SI: Fix 64-bit private loads.
Matt Arsenault [Mon, 24 Mar 2014 17:50:46 +0000 (17:50 +0000)]
R600/SI: Fix 64-bit private loads.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204630 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoVS integration installer: set SUCCESS=1 if we find VS 2013
Hans Wennborg [Mon, 24 Mar 2014 17:33:22 +0000 (17:33 +0000)]
VS integration installer: set SUCCESS=1 if we find VS 2013

Previously we would print an error message on machines where the only VS
version we find is 2013, even though we successfully install the integration
files for it.

Also, we shouldn't have two END labels.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204629 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdd test to test/CodeGen/NVPTX for "alloca buffer" arguments.
Eli Bendersky [Mon, 24 Mar 2014 16:52:30 +0000 (16:52 +0000)]
Add test to test/CodeGen/NVPTX for "alloca buffer" arguments.

Make sure such IR gets properly lowered to PTX.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204624 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[X86] Fix non-determinism in LowerVectorAllZeroTest
Adam Nemet [Mon, 24 Mar 2014 16:52:08 +0000 (16:52 +0000)]
[X86] Fix non-determinism in LowerVectorAllZeroTest

This can be observed with the old testcase of CodeGen/X86/pr12312.ll:

47c47
<       vorps   %ymm0, %ymm1, %ymm0
---
>       vorps   %ymm1, %ymm0, %ymm0
97c97
<       vorps   %ymm1, %ymm0, %ymm0
---
>       vorps   %ymm0, %ymm1, %ymm0

The vector VecIns is populated with all the values from VecInMap. This is done
while iterating VecInMap.  VecInMap uses a hash of pointer values so the
resulting order can vary depending on the memory layout.

The fix is to populate the vector VecIns earlier as VecInMap is populated.
This is done in DAG traversal order.

Fixes <rdar://problem/16398806>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204623 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[mips] Add error message when trying to use $at in '.set noat' mode.
Daniel Sanders [Mon, 24 Mar 2014 16:48:01 +0000 (16:48 +0000)]
[mips] Add error message when trying to use $at in '.set noat' mode.

Summary:
Patch by David Chisnall
His work was sponsored by: DARPA, AFRL

Differential Revision: http://llvm-reviews.chandlerc.com/D3158

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204621 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRemoves the NVPTXSplitBBatBar pass.
Eli Bendersky [Mon, 24 Mar 2014 16:36:39 +0000 (16:36 +0000)]
Removes the NVPTXSplitBBatBar pass.

This pass is a historic remnant and actually causes less efficient code to be
generated in some cases.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204620 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600/SI: Fix warning with gcc 4.8.2
Tom Stellard [Mon, 24 Mar 2014 16:12:34 +0000 (16:12 +0000)]
R600/SI: Fix warning with gcc 4.8.2

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204618 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600/SI: Promote fp64 SELECT to i64
Tom Stellard [Mon, 24 Mar 2014 16:07:30 +0000 (16:07 +0000)]
R600/SI: Promote fp64 SELECT to i64

This type promotion is replacing a Tablegen pattern and it is already
covered by existing tests.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204617 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoSelectionDAG: Allow promotion of SELECT nodes from float to int types
Tom Stellard [Mon, 24 Mar 2014 16:07:28 +0000 (16:07 +0000)]
SelectionDAG: Allow promotion of SELECT nodes from float to int types

And vice-versa, as long as the types are the same width.

There are a few R600 tests that will cover this.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204616 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600: Reorganize tablegen instruction definitions
Tom Stellard [Mon, 24 Mar 2014 16:07:25 +0000 (16:07 +0000)]
R600: Reorganize tablegen instruction definitions

Each GPU family now has its own file.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204615 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[PPC64LE] ELFv2 ABI updates for the .opd section
Will Schmidt [Mon, 24 Mar 2014 16:04:15 +0000 (16:04 +0000)]
[PPC64LE] ELFv2 ABI updates for the .opd section

[PPC64LE] ELFv2 ABI updates for the .opd section
The PPC64 Little Endian (PPC64LE) target supports the ELFv2 ABI, and as
such, does not have a ".opd" section.  This is keyed off a _CALL_ELF=2
macro check.

The CALL_ELF check is not clearly documented at this time.  The basis
for usage in this patch is from the gcc thread here:
http://gcc.gnu.org/ml/gcc-patches/2013-11/msg01144.html

> Adding comment from Uli:
Looks good to me.  I think the old-style JIT doesn't really work
anyway for 64-bit, but at least with this patch LLVM will compile
and link again on a ppc64le host ...

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204614 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[mips] Add regression tests for parenthetic expressions in MIPS assembly.
Daniel Sanders [Mon, 24 Mar 2014 15:42:21 +0000 (15:42 +0000)]
[mips] Add regression tests for parenthetic expressions in MIPS assembly.

Summary:
These expressions already worked but weren't tested.

Patch by Robert N. M. Watson and David Chisnall (it was originally two patches)
Their work was sponsored by: DARPA, AFRL

Differential Revision: http://llvm-reviews.chandlerc.com/D3156

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204612 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[mips] Allow dsubu to take an immediate as an alias for dsubiu.
Daniel Sanders [Mon, 24 Mar 2014 15:38:00 +0000 (15:38 +0000)]
[mips] Allow dsubu to take an immediate as an alias for dsubiu.

Summary:
Patch by David Chisnall
His work was sponsored by: DARPA, AFRL

Differential Revision: http://llvm-reviews.chandlerc.com/D3155

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204611 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[PowerPC] Mark many instructions as commutative
Hal Finkel [Mon, 24 Mar 2014 15:07:28 +0000 (15:07 +0000)]
[PowerPC] Mark many instructions as commutative

I'm under the impression that we used to infer the isCommutable flag from the
instruction-associated pattern. Regardless, we don't seem to do this (at least
by default) any more. I've gone through all of our instruction definitions, and
marked as commutative all of those that should be trivial to commute (by
exchanging the first two operands). There has been special code for the RL*
instructions, and that's not changed.

Before this change, we had the following commutative instructions:

 RLDIMI
 RLDIMIo
 RLWIMI
 RLWIMI8
 RLWIMI8o
 RLWIMIo
 XSADDDP
 XSMULDP
 XVADDDP
 XVADDSP
 XVMULDP
 XVMULSP

After:

 ADD4
 ADD4o
 ADD8
 ADD8o
 ADDC
 ADDC8
 ADDC8o
 ADDCo
 ADDE
 ADDE8
 ADDE8o
 ADDEo
 AND
 AND8
 AND8o
 ANDo
 CRAND
 CREQV
 CRNAND
 CRNOR
 CROR
 CRXOR
 EQV
 EQV8
 EQV8o
 EQVo
 FADD
 FADDS
 FADDSo
 FADDo
 FMADD
 FMADDS
 FMADDSo
 FMADDo
 FMSUB
 FMSUBS
 FMSUBSo
 FMSUBo
 FMUL
 FMULS
 FMULSo
 FMULo
 FNMADD
 FNMADDS
 FNMADDSo
 FNMADDo
 FNMSUB
 FNMSUBS
 FNMSUBSo
 FNMSUBo
 MULHD
 MULHDU
 MULHDUo
 MULHDo
 MULHW
 MULHWU
 MULHWUo
 MULHWo
 MULLD
 MULLDo
 MULLW
 MULLWo
 NAND
 NAND8
 NAND8o
 NANDo
 NOR
 NOR8
 NOR8o
 NORo
 OR
 OR8
 OR8o
 ORo
 RLDIMI
 RLDIMIo
 RLWIMI
 RLWIMI8
 RLWIMI8o
 RLWIMIo
 VADDCUW
 VADDFP
 VADDSBS
 VADDSHS
 VADDSWS
 VADDUBM
 VADDUBS
 VADDUHM
 VADDUHS
 VADDUWM
 VADDUWS
 VAND
 VAVGSB
 VAVGSH
 VAVGSW
 VAVGUB
 VAVGUH
 VAVGUW
 VMADDFP
 VMAXFP
 VMAXSB
 VMAXSH
 VMAXSW
 VMAXUB
 VMAXUH
 VMAXUW
 VMHADDSHS
 VMHRADDSHS
 VMINFP
 VMINSB
 VMINSH
 VMINSW
 VMINUB
 VMINUH
 VMINUW
 VMLADDUHM
 VMULESB
 VMULESH
 VMULEUB
 VMULEUH
 VMULOSB
 VMULOSH
 VMULOUB
 VMULOUH
 VNMSUBFP
 VOR
 VXOR
 XOR
 XOR8
 XOR8o
 XORo
 XSADDDP
 XSMADDADP
 XSMAXDP
 XSMINDP
 XSMSUBADP
 XSMULDP
 XSNMADDADP
 XSNMSUBADP
 XVADDDP
 XVADDSP
 XVMADDADP
 XVMADDASP
 XVMAXDP
 XVMAXSP
 XVMINDP
 XVMINSP
 XVMSUBADP
 XVMSUBASP
 XVMULDP
 XVMULSP
 XVNMADDADP
 XVNMADDASP
 XVNMSUBADP
 XVNMSUBASP
 XXLAND
 XXLNOR
 XXLOR
 XXLXOR

This is a by-inspection change, and I'm not sure how to write a reliable test
case. I would like advice on this, however.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204609 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[mips] Implement shorthand add / sub forms for MIPS.
Daniel Sanders [Mon, 24 Mar 2014 14:05:39 +0000 (14:05 +0000)]
[mips] Implement shorthand add / sub forms for MIPS.

Summary:
- If only two registers are passed to a three-register operation, then the
  first argument is both source and destination register.

- If a non-register is passed as the last argument, generate the immediate
  version of the instruction.

Also mark DADD commutative and add scheduling information (to the generic
scheduler), and implement DSUB.

Patch by David Chisnall
His work was sponsored by: DARPA, AFRL

CC: theraven
Differential Revision: http://llvm-reviews.chandlerc.com/D3148

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204605 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[NVPTX] Add isel patterns for addrspacecast
Justin Holewinski [Mon, 24 Mar 2014 11:17:53 +0000 (11:17 +0000)]
[NVPTX] Add isel patterns for addrspacecast

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204600 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoUpdate release notes with EHABI current behaviour
Renato Golin [Mon, 24 Mar 2014 11:02:38 +0000 (11:02 +0000)]
Update release notes with EHABI current behaviour

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204598 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[PowerPC] Don't schedule VSX copy legalization unless VSX is enabled
Hal Finkel [Mon, 24 Mar 2014 09:51:41 +0000 (09:51 +0000)]
[PowerPC] Don't schedule VSX copy legalization unless VSX is enabled

There is no need to schedule this extra pass if it will have nothing to do.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204594 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[PowerPC] Update comment re: VSX copy-instruction selection
Hal Finkel [Mon, 24 Mar 2014 09:36:36 +0000 (09:36 +0000)]
[PowerPC] Update comment re: VSX copy-instruction selection

I've done some experimentation with this, and it looks like using the
lower-latency (but lower throughput) copy instruction is essentially always the
right thing to do.

My assumption is that, in order to be relatively sure that the higher-latency
copy will increase throughput, we'd want to have it unlikely to be in-flight
with its use. On the P7, the global completion table (GCT) can hold a maximum
of 120 instructions, shared among all active threads (up to 4), giving 30
instructions per thread.  So specifically, I'd require at least that many
instructions between the copy and the use before the high-latency variant is
used.

Trying this, however, over the entire test suite resulted in zero cases where
the high-latency form would be preferable. This may be a consequence of the
fact that the scheduler views copies as free, and so they tend to end up close
to their uses. For this experiment I created a function:

  unsigned chooseVSXCopy(MachineBasicBlock &MBB,
                         MachineBasicBlock::iterator I,
                         unsigned DestReg, unsigned SrcReg,
                         unsigned StartDist = 1,
                         unsigned Depth = 3) const;

with an implementation like:

  if (!Depth)
    return PPC::XXLOR;

  const unsigned MaxDist = 30;
  unsigned Dist = StartDist;
  for (auto J = I, JE = MBB.end(); J != JE && Dist <= MaxDist; ++J) {
    if (J->isTransient() && !J->isCopy())
      continue;

    if (J->isCall() || J->isReturn() || J->readsRegister(DestReg, TRI))
      return PPC::XXLOR;

    ++Dist;
  }

  // We've exceeded the required distance for the high-latency form, use it.
  if (Dist > MaxDist)
    return PPC::XVCPSGNDP;

  // If this is only an exit block, use the low-latency form.
  if (MBB.succ_empty())
    return PPC::XXLOR;

  // We've reached the end of the block, check the successor blocks (up to some
  // depth), and use the high-latency form if that is okay with all successors.
  for (auto J = MBB.succ_begin(), JE = MBB.succ_end(); J != JE; ++J) {
    if (chooseVSXCopy(**J, (*J)->begin(), DestReg, SrcReg,
                      Dist, --Depth) == PPC::XXLOR)
      return PPC::XXLOR;
  }

  // All of our successor blocks seem okay with the high-latency variant, so
  // we'll use it.
  return PPC::XVCPSGNDP;

and then changed the copy opcode selection from:
    Opc = PPC::XXLOR;
to:
    Opc = chooseVSXCopy(MBB, std::next(I), DestReg, SrcReg);

In conclusion, I'm removing the FIXME from the comment, because I believe that
there is, at least absent other examples, nothing to fix.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204591 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoTeach llvm-readobj to print human friendly description of reserved sections.
Rafael Espindola [Mon, 24 Mar 2014 05:00:34 +0000 (05:00 +0000)]
Teach llvm-readobj to print human friendly description of reserved sections.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204584 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAllow constant folding of ceil function whenever feasible
Karthik Bhat [Mon, 24 Mar 2014 04:36:06 +0000 (04:36 +0000)]
Allow constant folding of ceil function whenever feasible

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204583 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdd back tests that were reverted in r204203.
Rafael Espindola [Mon, 24 Mar 2014 03:48:15 +0000 (03:48 +0000)]
Add back tests that were reverted in r204203.

They pass again with the fix in r204581.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204582 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoPropagate section from base to derived symbol.
Rafael Espindola [Mon, 24 Mar 2014 03:43:21 +0000 (03:43 +0000)]
Propagate section from base to derived symbol.

We were already propagating the section in

a = b

With this patch we also propagate it for

a = b + 1

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204581 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoInstrProf: Silence spurious warnings in GCC 4.8
Duncan P. N. Exon Smith [Mon, 24 Mar 2014 00:47:18 +0000 (00:47 +0000)]
InstrProf: Silence spurious warnings in GCC 4.8

No functionality change.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204580 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoSupportTests.LockFileManagerTest: Add assertions for Win32.
NAKAMURA Takumi [Sun, 23 Mar 2014 23:55:57 +0000 (23:55 +0000)]
SupportTests.LockFileManagerTest: Add assertions for Win32.

  - create_link doesn't work for nonexistent file.
  - remove cannot remove working directory.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204579 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoARM: no need to update SplatBits as it is not used
Arnaud A. de Grandmaison [Sun, 23 Mar 2014 21:14:32 +0000 (21:14 +0000)]
ARM: no need to update SplatBits as it is not used

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204575 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agollvm-profdata: Check for bad data in the show command
Justin Bogner [Sun, 23 Mar 2014 20:55:53 +0000 (20:55 +0000)]
llvm-profdata: Check for bad data in the show command

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204573 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agollvm-profdata: Use Format.h instead of handrolling a formatter
Justin Bogner [Sun, 23 Mar 2014 20:43:50 +0000 (20:43 +0000)]
llvm-profdata: Use Format.h instead of handrolling a formatter

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204571 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoWinCOFF: Add support for -ffunction-sections
David Majnemer [Sun, 23 Mar 2014 17:47:39 +0000 (17:47 +0000)]
WinCOFF: Add support for -ffunction-sections

This is a pretty straight forward translation for COFF, we just need to
stick the function in a COMDAT section marked as
IMAGE_COMDAT_SELECT_NODUPLICATES.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204565 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoremove a bunch of unused private methods
Nuno Lopes [Sun, 23 Mar 2014 17:09:26 +0000 (17:09 +0000)]
remove a bunch of unused private methods
found with a smarter version of -Wunused-member-function that I'm playwing with.
Appologies in advance if I removed someone's WIP code.

 include/llvm/CodeGen/MachineSSAUpdater.h            |    1
 include/llvm/IR/DebugInfo.h                         |    3
 lib/CodeGen/MachineSSAUpdater.cpp                   |   10 --
 lib/CodeGen/PostRASchedulerList.cpp                 |    1
 lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp    |   10 --
 lib/IR/DebugInfo.cpp                                |   12 --
 lib/MC/MCAsmStreamer.cpp                            |    2
 lib/Support/YAMLParser.cpp                          |   39 ---------
 lib/TableGen/TGParser.cpp                           |   16 ---
 lib/TableGen/TGParser.h                             |    1
 lib/Target/AArch64/AArch64TargetTransformInfo.cpp   |    9 --
 lib/Target/ARM/ARMCodeEmitter.cpp                   |   12 --
 lib/Target/ARM/ARMFastISel.cpp                      |   84 --------------------
 lib/Target/Mips/MipsCodeEmitter.cpp                 |   11 --
 lib/Target/Mips/MipsConstantIslandPass.cpp          |   12 --
 lib/Target/NVPTX/NVPTXISelDAGToDAG.cpp              |   21 -----
 lib/Target/NVPTX/NVPTXISelDAGToDAG.h                |    2
 lib/Target/PowerPC/PPCFastISel.cpp                  |    1
 lib/Transforms/Instrumentation/AddressSanitizer.cpp |    2
 lib/Transforms/Instrumentation/BoundsChecking.cpp   |    2
 lib/Transforms/Instrumentation/MemorySanitizer.cpp  |    1
 lib/Transforms/Scalar/LoopIdiomRecognize.cpp        |    8 -
 lib/Transforms/Scalar/SCCP.cpp                      |    1
 utils/TableGen/CodeEmitterGen.cpp                   |    2
 24 files changed, 2 insertions(+), 261 deletions(-)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204560 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[PowerPC] Make use of VSX f64 <-> i64 conversion instructions
Hal Finkel [Sun, 23 Mar 2014 05:35:00 +0000 (05:35 +0000)]
[PowerPC] Make use of VSX f64 <-> i64 conversion instructions

When VSX is available, these instructions should be used in preference to the
older variants that only have access to the scalar floating-point registers.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204559 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRevert r204076 for now - it caused significant regressions in a number of
Lang Hames [Sun, 23 Mar 2014 04:22:31 +0000 (04:22 +0000)]
Revert r204076 for now - it caused significant regressions in a number of
benchmarks.

<rdar://problem/16368461>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204558 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoInstrProf: Check pointer size in raw profile
Duncan P. N. Exon Smith [Sun, 23 Mar 2014 03:38:12 +0000 (03:38 +0000)]
InstrProf: Check pointer size in raw profile

Since the profile can come from 32-bit machines, we need to check the
pointer size.  Change the magic number to facilitate this.

Adds tests for reading 32-bit and 64-bit binaries (both big- and
little-endian).  The tests write a binary using printf in RUN lines
(like raw-magic-but-no-header.test).  Assuming the bots don't complain,
this seems like a better way forward for testing RawInstrProfReader than
committing binary files.

<rdar://problem/16400648>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204557 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoPropagate types from symbol to aliases.
Rafael Espindola [Sun, 23 Mar 2014 03:33:20 +0000 (03:33 +0000)]
Propagate types from symbol to aliases.

This is similar, but not identical to what gas does. The logic in MC is to just
compute the symbol table after parsing the entire file. GAS is mixed, given

.type b, @object
a = b
b:
.type b, @function

It will propagate the change and make 'a' a function. Given

.type b, @object
b:
a = b
.type b, @function

the type of 'a' is still object.

Since we do the computation in the end, we produce a function in both cases.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204555 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[CMake] LLVMProfileData: No need to add LINK_LIBS here. LLVMBuild should do.
NAKAMURA Takumi [Sun, 23 Mar 2014 01:23:36 +0000 (01:23 +0000)]
[CMake] LLVMProfileData: No need to add LINK_LIBS here. LLVMBuild should do.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204553 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agollvm-profdata doesn't require LLVMCore.
NAKAMURA Takumi [Sun, 23 Mar 2014 01:23:26 +0000 (01:23 +0000)]
llvm-profdata doesn't require LLVMCore.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204552 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agollvm-profdata: Don't pipe stderr into show for the tests
Justin Bogner [Sat, 22 Mar 2014 23:53:43 +0000 (23:53 +0000)]
llvm-profdata: Don't pipe stderr into show for the tests

Some text shows up on stderr when using guard malloc, and this test
was trying to treat that as input to llvm-profdata show. There's no
reason to pipe stderr into show at all here.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204549 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoPrune includes in ARM target.
Craig Topper [Sat, 22 Mar 2014 23:51:00 +0000 (23:51 +0000)]
Prune includes in ARM target.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204548 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoARM IAS: properly handle function entries in .thumb
Saleem Abdulrasool [Sat, 22 Mar 2014 19:26:18 +0000 (19:26 +0000)]
ARM IAS: properly handle function entries in .thumb

When a label is parsed, check if there is type information available for the
label.  If so, check if the symbol is a function.  If the symbol is a function
and we are in thumb mode and no explicit thumb_func has been emitted, adjust the
symbol data to indicate that the function definition is a thumb function.

The application of this inferencing is improved value handling in the object
file (the required thumb bit is set on symbols which are thumb functions).  It
also helps improve compatibility with binutils.

The one complication that arises from this handling is the MCAsmStreamer.  The
default implementation of getOrCreateSymbolData in MCStreamer does not support
tracking the symbol data.  In order to support the semantics of thumb functions,
track symbol data in assembly streamer.  Although O(n) in number of labels in
the TU, this is already done in various other streamers and as such the memory
overhead is not a practical concern in this scenario.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204544 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[PowerPC] Fix the VSX v2f64 return register
Hal Finkel [Sat, 22 Mar 2014 18:24:43 +0000 (18:24 +0000)]
[PowerPC] Fix the VSX v2f64 return register

v2f64 values, like other 128-bit values, are returned under VSX in register
vs34 (Altivec register v2).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204543 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[TableGen] Don't assert, produce an error, when an instruction has too few operands
Hal Finkel [Sat, 22 Mar 2014 11:33:32 +0000 (11:33 +0000)]
[TableGen] Don't assert, produce an error, when an instruction has too few operands

When an instruction's operand list does not have a sufficient number of
operands to match with all of the variables that contribute to its
encoding, instead of asserting inside a call to getSubOperandNumber, produce an
informative error.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204542 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agollvm-profdata: Avoid F_Text in "merge" for now, since "llvm-profdata show" is confuse...
NAKAMURA Takumi [Sat, 22 Mar 2014 05:38:22 +0000 (05:38 +0000)]
llvm-profdata: Avoid F_Text in "merge" for now, since "llvm-profdata show" is confused with CRLF.

FIXME: line_iterator should be tolerant of CR.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204540 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[Constant Hoisting] Erase dead cast instructions.
Juergen Ributzka [Sat, 22 Mar 2014 01:49:30 +0000 (01:49 +0000)]
[Constant Hoisting] Erase dead cast instructions.

The cleanup code that removes dead cast instructions only removed them from the
basic block, but didn't delete them. This fix erases them now too.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204538 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[Constant Hoisting] Fix multiple entries for the same basic block in PHI nodes.
Juergen Ributzka [Sat, 22 Mar 2014 01:49:27 +0000 (01:49 +0000)]
[Constant Hoisting] Fix multiple entries for the same basic block in PHI nodes.

A PHI node usually has only one value/basic block pair per incoming basic block.
In the case of a switch statement it is possible that a following PHI node may
have more than one such pair per incoming basic block. E.g.:
%0 = phi i64 [ 123456, %case2 ], [ 654321, %Entry ], [ 654321, %Entry ]
This is valid and the verfier doesn't complain, because both values are the
same.

Constant hoisting materializes the constant for each operand separately and the
value is still the same, but the variable names have changed. As a result the
verfier can't recognize anymore that they are the same value and complains.

This fix adds special update code for PHI node in constant hoisting to prevent
this corner case.

This fixes <rdar://problem/16394449>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204537 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[DAG] Fix an assertion failure caused by an invalid cast in method 'BuildVectorSDNode...
Andrea Di Biagio [Sat, 22 Mar 2014 01:47:22 +0000 (01:47 +0000)]
[DAG] Fix an assertion failure caused by an invalid cast in method 'BuildVectorSDNode::isConstantSplat'

This patch renames method 'isConstantSplat' as 'getConstantSplatValue'
(mainly for consistency reasons), and rewrites its logic to ensure
that we always perform a legal 'cast<ConstantSDNode>'.

Added test shift-combine-crash.ll to verify that DAGCombiner no longer crashes with an assertion failure in the attempt to simplify a vector shift by a vector of all undef counts.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204536 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoSuppress SupportTests.LockFileManagerTest on win32 for investigating.
NAKAMURA Takumi [Sat, 22 Mar 2014 00:27:17 +0000 (00:27 +0000)]
Suppress SupportTests.LockFileManagerTest on win32 for investigating.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204533 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoDelete stale comment. Thanks, Eric!
Adrian Prantl [Fri, 21 Mar 2014 22:58:25 +0000 (22:58 +0000)]
Delete stale comment. Thanks, Eric!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204530 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoDwarf Debug: Remove some cargo-cult type uniquing. Scopes do not have
Adrian Prantl [Fri, 21 Mar 2014 22:16:32 +0000 (22:16 +0000)]
Dwarf Debug: Remove some cargo-cult type uniquing. Scopes do not have
an ID, so this is a noop.
Thanks Manman for catching this!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204528 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoFix the value computation in
Rafael Espindola [Fri, 21 Mar 2014 22:00:29 +0000 (22:00 +0000)]
Fix the value computation in

sym_a:
sym_d = sym_a + 1

This is the smallest fix I was able to extract from what got reverted in
r204203.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204527 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRemove some dead assignements found by scan-build
Arnaud A. de Grandmaison [Fri, 21 Mar 2014 21:54:46 +0000 (21:54 +0000)]
Remove some dead assignements found by scan-build

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204526 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRegister allocator: add condition to hoist a spill to outer loop.
Manman Ren [Fri, 21 Mar 2014 21:46:24 +0000 (21:46 +0000)]
Register allocator: add condition to hoist a spill to outer loop.

We make sure a spill is not hoisted to a hotter outer loop by adding
a condition. Hoist a spill to outer loop if there are multiple dependents
(it can be beneficial if more than one dependents are hoisted) or
if DepSV (the hoisting source) is hotter than SV (the hoisting destination).

rdar://16268194

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204522 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[Support] Follow up to r204426, for LockFileManager, make the given path absolute...
Argyrios Kyrtzidis [Fri, 21 Mar 2014 21:45:07 +0000 (21:45 +0000)]
[Support] Follow up to r204426, for LockFileManager, make the given path absolute so relative paths are properly handled in both Windows and Unix.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204520 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoInstrProf: Cleanup binary profdata testcase
Duncan P. N. Exon Smith [Fri, 21 Mar 2014 21:20:35 +0000 (21:20 +0000)]
InstrProf: Cleanup binary profdata testcase

Cleanup the current binary testcase for profile data.

  - Rename it to something more specific.
  - Remove the text comparison.
  - Check the output of llvm-profdata show.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204518 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoInstrProf: Move constructor to the header
Duncan P. N. Exon Smith [Fri, 21 Mar 2014 20:59:19 +0000 (20:59 +0000)]
InstrProf: Move constructor to the header

Fixes 80-column violation at the same time.

<rdar://problem/15950346>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204516 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoInstrProf: Change magic number to have non-text characters
Duncan P. N. Exon Smith [Fri, 21 Mar 2014 20:42:37 +0000 (20:42 +0000)]
InstrProf: Change magic number to have non-text characters

Include non-text characters in the magic number so that text files can't
match.

<rdar://problem/15950346>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204513 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoInstrProf: Use move semantics with unique_ptr
Duncan P. N. Exon Smith [Fri, 21 Mar 2014 20:42:34 +0000 (20:42 +0000)]
InstrProf: Use move semantics with unique_ptr

<rdar://problem/15950346>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204512 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoInstrProf: Detect magic numbers in a more scalable way
Duncan P. N. Exon Smith [Fri, 21 Mar 2014 20:42:31 +0000 (20:42 +0000)]
InstrProf: Detect magic numbers in a more scalable way

No functionality change.

<rdar://problem/15950346>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204511 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoInstrProf: Actually detect bad headers
Duncan P. N. Exon Smith [Fri, 21 Mar 2014 20:42:28 +0000 (20:42 +0000)]
InstrProf: Actually detect bad headers

<rdar://problem/15950346>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204510 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[RuntimeDyld] Fix comment for previous commit (r204439)
Juergen Ributzka [Fri, 21 Mar 2014 20:38:46 +0000 (20:38 +0000)]
[RuntimeDyld] Fix comment for previous commit (r204439)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204508 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[RuntimeDyld] clang-format files.
Juergen Ributzka [Fri, 21 Mar 2014 20:28:42 +0000 (20:28 +0000)]
[RuntimeDyld] clang-format files.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204507 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoDebugInfo: Omit DW_AT_addr_base from skeletal type units.
David Blaikie [Fri, 21 Mar 2014 20:27:21 +0000 (20:27 +0000)]
DebugInfo: Omit DW_AT_addr_base from skeletal type units.

Type units have no addresses, so there's no need for DW_AT_addr_base.
This removes another relocation from every skeletal type unit and brings
LLVM's skeletal type units in line with GCC's (containing only
GNU_dwo_name (strp), comp_dir (strp), and GNU_pubnames (flag_present)).

Cary's got some ideas about using str_index in the .o file to reduce
those last two relocations (well, replace two relocations with one
relocation (pointing to the string index) and two indicies)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204506 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[AArch64] Add SchedRW lists to NEON instructions.
Chad Rosier [Fri, 21 Mar 2014 19:34:41 +0000 (19:34 +0000)]
[AArch64] Add SchedRW lists to NEON instructions.

Previously, only regular AArch64 instructions were annotated with SchedRW lists.
This patch does the same for NEON enabling these instructions to be scheduled by
the MIScheduler. Additionally, store operations are now modeled and a few
SchedRW lists were updated for bug fixes (e.g. multiple def operands).

Reviewers: apazos, mcrosier, atrick
Patch by Dave Estes <cestes@codeaurora.org>!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204505 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoProfileData: Avoid double underscores in header guards
Justin Bogner [Fri, 21 Mar 2014 18:46:29 +0000 (18:46 +0000)]
ProfileData: Avoid double underscores in header guards

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204501 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoInstrProf: Read raw binary profile in llvm-profdata
Duncan P. N. Exon Smith [Fri, 21 Mar 2014 18:26:05 +0000 (18:26 +0000)]
InstrProf: Read raw binary profile in llvm-profdata

Read a raw binary profile that corresponds to a memory dump from the
runtime profile.

The test is a binary file generated from
cfe/trunk/test/Profile/c-general.c with the new compiler-rt runtime and
the matching text version of the input.  It includes instructions on how
to regenerate.

<rdar://problem/15950346>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204496 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoProfileData: Avoid brace initialization, windows doesn't like it
Justin Bogner [Fri, 21 Mar 2014 18:22:16 +0000 (18:22 +0000)]
ProfileData: Avoid brace initialization, windows doesn't like it

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204494 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600/SI: Move instruction patterns to scalar versions.
Matt Arsenault [Fri, 21 Mar 2014 18:01:18 +0000 (18:01 +0000)]
R600/SI: Move instruction patterns to scalar versions.

Some of them also had the pattern on both, so this removes the
duplication.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204492 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRemove redundant test.
Rafael Espindola [Fri, 21 Mar 2014 18:00:51 +0000 (18:00 +0000)]
Remove redundant test.

This is tested from MC already.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204491 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoMove codegen test over to MC.
Rafael Espindola [Fri, 21 Mar 2014 17:55:34 +0000 (17:55 +0000)]
Move codegen test over to MC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204490 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoProfileData: Introduce InstrProfWriter using the naive text format
Justin Bogner [Fri, 21 Mar 2014 17:46:22 +0000 (17:46 +0000)]
ProfileData: Introduce InstrProfWriter using the naive text format

This isn't a format we'll want to write out in practice, but moving it
to the writer library simplifies llvm-profdata and isolates it from
further changes to the format.

This also allows us to update the tests to not rely on the text output
format.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204489 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoConvert test to using cfi.
Rafael Espindola [Fri, 21 Mar 2014 17:38:01 +0000 (17:38 +0000)]
Convert test to using cfi.

An unnamed global in llvm still produces a regular symbol.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204488 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRefactor llvm/test/lit.cfg to use lit.util.which.
Paul Robinson [Fri, 21 Mar 2014 17:31:35 +0000 (17:31 +0000)]
Refactor llvm/test/lit.cfg to use lit.util.which.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204486 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agollvm-profdata: Implement show command
Justin Bogner [Fri, 21 Mar 2014 17:29:44 +0000 (17:29 +0000)]
llvm-profdata: Implement show command

The `llvm-profdata show` command summarizes a profdata file's contents
in a human readable format.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204485 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRemove redundant test.
Rafael Espindola [Fri, 21 Mar 2014 17:26:35 +0000 (17:26 +0000)]
Remove redundant test.

The production of the .eh symbols is done from MC now and we already have tests
for it.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204483 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoProfileData: Introduce the InstrProfReader interface and a text reader
Justin Bogner [Fri, 21 Mar 2014 17:24:48 +0000 (17:24 +0000)]
ProfileData: Introduce the InstrProfReader interface and a text reader

This introduces the ProfileData library and updates llvm-profdata to
use this library for reading profiles. InstrProfReader is an abstract
base class that will be subclassed for both the raw instrprof data
from compiler-rt and the efficient instrprof format that will be used
for PGO.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204482 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoSplit out the MC part of this test.
Rafael Espindola [Fri, 21 Mar 2014 17:16:11 +0000 (17:16 +0000)]
Split out the MC part of this test.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204481 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[mips] Correct lowering of VECTOR_SHUFFLE to VSHF.
Daniel Sanders [Fri, 21 Mar 2014 16:56:51 +0000 (16:56 +0000)]
[mips] Correct lowering of VECTOR_SHUFFLE to VSHF.

Summary:
VECTOR_SHUFFLE concatenates the vectors in an vectorwise fashion.
  <0b00, 0b01> + <0b10, 0b11> -> <0b00, 0b01, 0b10, 0b11>
VSHF concatenates the vectors in a bitwise fashion:
  <0b00, 0b01> + <0b10, 0b11> ->
  0b0100       + 0b1110       -> 0b01001110
                                 <0b10, 0b11, 0b00, 0b01>
We must therefore swap the operands to get the correct result.

The test case that discovered the issue was MultiSource/Benchmarks/nbench.

Reviewers: matheusalmeida

Reviewed By: matheusalmeida

Differential Revision: http://llvm-reviews.chandlerc.com/D3142

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204480 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdd overall description, file comments, some structure
Renato Golin [Fri, 21 Mar 2014 16:49:43 +0000 (16:49 +0000)]
Add overall description, file comments, some structure

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204479 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600/SI: Handle MUBUF instructions in SIInstrInfo::moveToVALU()
Tom Stellard [Fri, 21 Mar 2014 15:51:57 +0000 (15:51 +0000)]
R600/SI: Handle MUBUF instructions in SIInstrInfo::moveToVALU()

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204476 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600/SI: Handle S_MOV_B64 in SIInstrInfo::moveToVALU()
Tom Stellard [Fri, 21 Mar 2014 15:51:54 +0000 (15:51 +0000)]
R600/SI: Handle S_MOV_B64 in SIInstrInfo::moveToVALU()

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204475 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600/SI: Use SGPR_(32|64) reg clases when lowering SI_ADDR64_RSRC
Tom Stellard [Fri, 21 Mar 2014 15:51:53 +0000 (15:51 +0000)]
R600/SI: Use SGPR_(32|64) reg clases when lowering SI_ADDR64_RSRC

The SReg_(32|64) register classes contain special registers in addition
to the numbered SGPRs.  This can lead to machine verifier errors when
these register classes are used as sub-registers for SReg_128, since
SReg_128 only uses the numbered SGPRs.

Replacing SReg_(32|64) with SGPR_(32|64) fixes this problem, since
the SGPR_(32|64) register classes contain only numbered SGPRs.

Tests cases for this are comming in a later commit.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204474 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoSink: Don't sink static allocas from the entry block
Tom Stellard [Fri, 21 Mar 2014 15:51:51 +0000 (15:51 +0000)]
Sink: Don't sink static allocas from the entry block

CodeGen treats allocas outside the entry block as dynamically sized
stack objects.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204473 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[SystemZ] Use "let Predicates =" for blocks of new instructions
Richard Sandiford [Fri, 21 Mar 2014 11:04:54 +0000 (11:04 +0000)]
[SystemZ] Use "let Predicates =" for blocks of new instructions

...instead of a separate Requires for each one.  This style was already
used in some places and seems more compact.

No behavioral change intended.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204452 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[SystemZ] Add support for z196 float<->unsigned conversions
Richard Sandiford [Fri, 21 Mar 2014 10:56:30 +0000 (10:56 +0000)]
[SystemZ] Add support for z196 float<->unsigned conversions

These complement the older float<->signed instructions.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204451 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[mips] Update namespace.
Matheus Almeida [Fri, 21 Mar 2014 10:35:14 +0000 (10:35 +0000)]
[mips] Update namespace.

We should be using the llvm namespace and not an anonymous namespace
in a header file.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204450 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[RuntimeDyld] Allow processRelocationRef to process more than one relocation entry...
Juergen Ributzka [Fri, 21 Mar 2014 07:26:41 +0000 (07:26 +0000)]
[RuntimeDyld] Allow processRelocationRef to process more than one relocation entry at a time.

Some targets require more than one relocation entry to perform a relocation.
This change allows processRelocationRef to process more than one relocation
entry at a time by passing the relocation iterator itself instead of just
the relocation entry.

Related to <rdar://problem/16199095>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204439 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoFix test command line to avoid generating output file.
Kevin Qin [Fri, 21 Mar 2014 07:20:29 +0000 (07:20 +0000)]
Fix test command line to avoid generating output file.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204437 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[Constant Hoisting] Make the constant materialization cost operand dependent
Juergen Ributzka [Fri, 21 Mar 2014 06:04:45 +0000 (06:04 +0000)]
[Constant Hoisting] Make the constant materialization cost operand dependent

Extend the target hook to take also the operand index into account when
calculating the cost of the constant materialization.

Related to <rdar://problem/16381500>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204435 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[Constant Hoisting] Lazily compute the idom and cache the result.
Juergen Ributzka [Fri, 21 Mar 2014 06:04:39 +0000 (06:04 +0000)]
[Constant Hoisting] Lazily compute the idom and cache the result.

Related to <rdar://problem/16381500>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204434 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[Constant Hoisting] Change the algorithm to only track constants for instructions.
Juergen Ributzka [Fri, 21 Mar 2014 06:04:36 +0000 (06:04 +0000)]
[Constant Hoisting] Change the algorithm to only track constants for instructions.

Originally the algorithm would search for expensive constants and track their
users, which could be instructions and constant expressions. This change only
tracks the constants for instructions, but constant expressions are indirectly
covered too. If an operand is an constant expression, then we look through the
expression to find anny expensive constants.

The algorithm keep now track of the instruction and the operand index where the
constant is used. This allows more precise hoisting of constant materialization
code for PHI instructions, because we only hoist to the basic block of the
incoming operand. Before we had to find the idom of all PHI operands and hoist
the materialization code there.

This also makes updating of instructions easier. Before we had to keep track of
the original constant, find it in the instructions, and then replace it. Now we
can just simply update the operand.

Related to <rdar://problem/16381500>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204433 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[Constant Hoisting] Fix capitalization of function names.
Juergen Ributzka [Fri, 21 Mar 2014 06:04:33 +0000 (06:04 +0000)]
[Constant Hoisting] Fix capitalization of function names.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204432 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[Constant Hoisting] Replace the MapVector with a separate Map and Vector to keep...
Juergen Ributzka [Fri, 21 Mar 2014 06:04:30 +0000 (06:04 +0000)]
[Constant Hoisting] Replace the MapVector with a separate Map and Vector to keep track of constant candidates.

This simplifies working with the constant candidates and removes the tight
coupling between the map and the vector.

Related to <rdar://problem/16381500>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204431 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoMCParser: add an assertion
Saleem Abdulrasool [Fri, 21 Mar 2014 05:13:23 +0000 (05:13 +0000)]
MCParser: add an assertion

Add an assertion that the section is not NULL.  Potential NULL pointer
dereference identified by clang static analyzer.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204429 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoThis reverts commit r203762, "ARM: support emission of complex SO expressions".
Jiangning Liu [Fri, 21 Mar 2014 02:51:01 +0000 (02:51 +0000)]
This reverts commit r203762, "ARM: support emission of complex SO expressions".

The commit r203762 introduced silent failure for complext SO expression, and it's even worse than compiler crash.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@204427 91177308-0d34-0410-b5e6-96231b3b80d8