oota-llvm.git
10 years agoRename DIBuilder::createNullPtrType to createUnspecifiedType and introduce
Peter Collingbourne [Thu, 27 Jun 2013 22:50:59 +0000 (22:50 +0000)]
Rename DIBuilder::createNullPtrType to createUnspecifiedType and introduce
a zero-argument createNullPtrType function for creating the canonical
nullptr type.

Differential Revision: http://llvm-reviews.chandlerc.com/D1050

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185114 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAt the request of Richard Smith, swapped the order of cold/builtin so it is in alphab...
Michael Gottesman [Thu, 27 Jun 2013 22:48:08 +0000 (22:48 +0000)]
At the request of Richard Smith, swapped the order of cold/builtin so it is in alphabetical order.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185113 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRevert "Revert "[APFloat] Removed APFloat constructor which initialized to either...
Michael Gottesman [Thu, 27 Jun 2013 21:58:19 +0000 (21:58 +0000)]
Revert "Revert "[APFloat] Removed APFloat constructor which initialized to either zero/NaN but allowed you to arbitrarily set the category of the float.""

This reverts commit r185099.

Looks like both the ppc-64 and mips bots are still failing after I reverted this
change.

Since:

1. The mips bot always performs a clean build,
2. The ppc64-bot failed again after a clean build (I asked the ppc-64
maintainers to clean the bot which they did... Thanks Will!),

I think it is safe to assume that this change was not the cause of the failures
that said builders were seeing. Thus I am recomitting.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185111 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRevert "[APFloat] Removed APFloat constructor which initialized to either zero/NaN...
Michael Gottesman [Thu, 27 Jun 2013 20:40:11 +0000 (20:40 +0000)]
Revert "[APFloat] Removed APFloat constructor which initialized to either zero/NaN but allowed you to arbitrarily set the category of the float."

This reverts commit r185095. This is causing a FileCheck failure on
the 3dnow intrinsics on at least the mips/ppc bots but not on the x86
bots.

Reverting while I figure out what is going on.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185099 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoLoopVectorize: Cache edge masks created during if-conversion
Arnold Schwaighofer [Thu, 27 Jun 2013 20:31:06 +0000 (20:31 +0000)]
LoopVectorize: Cache edge masks created during if-conversion

Otherwise, we end up with an exponential IR blowup.
Fixes PR16472.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185097 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRemove unnecessary conditional checks.
Chad Rosier [Thu, 27 Jun 2013 20:19:13 +0000 (20:19 +0000)]
Remove unnecessary conditional checks.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185096 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[APFloat] Removed APFloat constructor which initialized to either zero/NaN but allowe...
Michael Gottesman [Thu, 27 Jun 2013 19:50:52 +0000 (19:50 +0000)]
[APFloat] Removed APFloat constructor which initialized to either zero/NaN but allowed you to arbitrarily set the category of the float.

The category which an APFloat belongs to should be dependent on the
actual value that the APFloat has, not be arbitrarily passed in by the
user. This will prevent inconsistency bugs where the category and the
actual value in APFloat differ.

I also fixed up all of the references to this constructor (which were
only in LLVM).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185095 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoImprove the compression of the tablegen DiffLists by introducing a new sort
Chad Rosier [Thu, 27 Jun 2013 19:38:13 +0000 (19:38 +0000)]
Improve the compression of the tablegen DiffLists by introducing a new sort
algorithm when assigning EnumValues to the synthesized registers.

The current algorithm, LessRecord, uses the StringRef compare_numeric
function.  This function compares strings, while handling embedded numbers.
For example, the R600 backend registers are sorted as follows:

  T1
  T1_W
  T1_X
  T1_XYZW
  T1_Y
  T1_Z
  T2
  T2_W
  T2_X
  T2_XYZW
  T2_Y
  T2_Z

In this example, the 'scaling factor' is dEnum/dN = 6 because T0, T1, T2
have an EnumValue offset of 6 from one another.  However, in other parts
of the register bank, the scaling factors are different:

dEnum/dN = 5:
  KC0_128_W
  KC0_128_X
  KC0_128_XYZW
  KC0_128_Y
  KC0_128_Z
  KC0_129_W
  KC0_129_X
  KC0_129_XYZW
  KC0_129_Y
  KC0_129_Z

The diff lists do not work correctly because different kinds of registers have
different 'scaling factors'.  This new algorithm, LessRecordRegister, tries to
enforce a scaling factor of 1.  For example, the registers are now sorted as
follows:

  T1
  T2
  T3
  ...
  T0_W
  T1_W
  T2_W
  ...
  T0_X
  T1_X
  T2_X
  ...
  KC0_128_W
  KC0_129_W
  KC0_130_W
  ...

For the Mips and R600 I see a 19% and 6% reduction in size, respectively.  I
did see a few small regressions, but the differences were on the order of a
few bytes (e.g., AArch64 was 16 bytes).  I suspect there will be even
greater wins for targets with larger register files.

Patch reviewed by Jakob.
rdar://14006013

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185094 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[Support/Registry.h] Include llvm/Support/Compiler.h.
Argyrios Kyrtzidis [Thu, 27 Jun 2013 17:57:40 +0000 (17:57 +0000)]
[Support/Registry.h] Include llvm/Support/Compiler.h.

Because Registry.h is using the LLVM_DELETED_FUNCTION macro.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185087 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoGet rid of the unused class member.
Nadav Rotem [Thu, 27 Jun 2013 17:54:10 +0000 (17:54 +0000)]
Get rid of the unused class member.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185086 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoCostModel: improve the cost model for load/store of non power-of-two types such as...
Nadav Rotem [Thu, 27 Jun 2013 17:52:04 +0000 (17:52 +0000)]
CostModel: improve the cost model for load/store of non power-of-two types such as <3 x float>, which are popular in graphics.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185085 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600: Remove alu-split.ll test
Tom Stellard [Thu, 27 Jun 2013 17:00:38 +0000 (17:00 +0000)]
R600: Remove alu-split.ll test

The purpose of this test was to check boundary conditions for the size
of an ALU clause.  This test is very sensitive to changes to the
optimizer or scheduler, because it requires an exact number of ALU
instructions in order to remain valid.  It's not good to have a test
this sensitive, because it is confusing to developers who implement
optimizations and then 'break' the test.

I'm not sure if there is a good way to test these limits using lit, but
if I can come up with replacement test that isn't as sensitive I'll add
it back to the tree.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185084 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoLoopVectorize: Use vectorized loop invariant gep index anchored in loop
Arnold Schwaighofer [Thu, 27 Jun 2013 15:11:55 +0000 (15:11 +0000)]
LoopVectorize: Use vectorized loop invariant gep index anchored in loop

Use vectorized instruction instead of original instruction anchored in the
original loop.

Fixes PR16452 and t2075.c of PR16455.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185081 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoUse MCFillFragment for zero-initialized data.
Serge Pavlov [Thu, 27 Jun 2013 14:35:03 +0000 (14:35 +0000)]
Use MCFillFragment for zero-initialized data.
It fixes PR16338 (ICE when compiling very large two-dimensional array).

Differential Revision: http://llvm-reviews.chandlerc.com/D1043

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185080 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdd a Subtarget feature 'v8fp' to the ARM backend.
Joey Gouly [Thu, 27 Jun 2013 11:49:26 +0000 (11:49 +0000)]
Add a Subtarget feature 'v8fp' to the ARM backend.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185073 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRemove unused variable.
Benjamin Kramer [Thu, 27 Jun 2013 11:26:41 +0000 (11:26 +0000)]
Remove unused variable.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185072 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoDon't cast away constness.
Benjamin Kramer [Thu, 27 Jun 2013 11:07:42 +0000 (11:07 +0000)]
Don't cast away constness.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185071 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[SystemZ] Fix some embarrassing test typos
Richard Sandiford [Thu, 27 Jun 2013 09:49:34 +0000 (09:49 +0000)]
[SystemZ] Fix some embarrassing test typos

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185070 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[SystemZ] Allow LA and LARL to be rematerialized
Richard Sandiford [Thu, 27 Jun 2013 09:42:10 +0000 (09:42 +0000)]
[SystemZ] Allow LA and LARL to be rematerialized

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185069 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[SystemZ] Allow immediate moves to be rematerialized
Richard Sandiford [Thu, 27 Jun 2013 09:38:48 +0000 (09:38 +0000)]
[SystemZ] Allow immediate moves to be rematerialized

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185068 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[SystemZ] Add conditional store patterns
Richard Sandiford [Thu, 27 Jun 2013 09:27:40 +0000 (09:27 +0000)]
[SystemZ] Add conditional store patterns

Add pseudo conditional store instructions, so that we use:

    branch foo:
    store
foo:

instead of:

    load
    branch foo:
    move
foo:
    store

z196 has real 32-bit and 64-bit conditional stores, but we don't use
any z196 instructions yet.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185065 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdd more owners to CODE_OWNERS.TXT (Kostya Serebryany: AddressSanitizer and ThreadSan...
Kostya Serebryany [Thu, 27 Jun 2013 08:47:12 +0000 (08:47 +0000)]
Add more owners to CODE_OWNERS.TXT (Kostya Serebryany: AddressSanitizer and ThreadSanitizer; Evgeniy Stepanov: MemorySanitizer)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185064 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoUpdate testing case to make DI nodes have the correct format.
Manman Ren [Thu, 27 Jun 2013 06:40:18 +0000 (06:40 +0000)]
Update testing case to make DI nodes have the correct format.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185061 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdd CXXFLAGS back to the Link command.
Bob Wilson [Thu, 27 Jun 2013 06:09:14 +0000 (06:09 +0000)]
Add CXXFLAGS back to the Link command.

This is essentially reverting one piece of 184793 to try to fix one of Apple's
buildbots.  I will check with Eric to see if this is OK or if we need to find
some other solution.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185060 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdd a convenience createUniqueDirectory function.
Rafael Espindola [Thu, 27 Jun 2013 03:45:31 +0000 (03:45 +0000)]
Add a convenience createUniqueDirectory function.

There are a few valid situation where we care about the structure inside a
directory, but not about the directory itself. A simple example is for unit
testing directory traversal.

PathV1 had a function like this, add one to V2 and port existing users of the
created temp file and delete it hack to using it.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185059 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoFix spelling.
Arnold Schwaighofer [Thu, 27 Jun 2013 01:01:11 +0000 (01:01 +0000)]
Fix spelling.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185052 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoLoopVectorize: Don't store a reversed value in the vectorized value map
Arnold Schwaighofer [Thu, 27 Jun 2013 00:45:41 +0000 (00:45 +0000)]
LoopVectorize: Don't store a reversed value in the vectorized value map

When we store values for reversed induction stores we must not store the
reversed value in the vectorized value map. Another instruction might use this
value.

This fixes 3 test cases of PR16455.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185051 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[APFloat] Added unittest for APFloat.divide that checks special cases, result categor...
Michael Gottesman [Thu, 27 Jun 2013 00:42:00 +0000 (00:42 +0000)]
[APFloat] Added unittest for APFloat.divide that checks special cases, result categories, and result statuses.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185050 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdded support for the Builtin attribute.
Michael Gottesman [Thu, 27 Jun 2013 00:25:01 +0000 (00:25 +0000)]
Added support for the Builtin attribute.

The Builtin attribute is an attribute that can be placed on function call site that signal that even though a function is declared as being a builtin,

rdar://problem/13727199

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185049 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoNo need to use a Set when a vector would do.
Nadav Rotem [Thu, 27 Jun 2013 00:14:13 +0000 (00:14 +0000)]
No need to use a Set when a vector would do.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185047 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRemoved two logging printf statements from the APFloat tests... = /.
Michael Gottesman [Thu, 27 Jun 2013 00:00:45 +0000 (00:00 +0000)]
Removed two logging printf statements from the APFloat tests... = /.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185045 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[APFloat] Added unittest for APFloat.multiply that checks special cases, result categ...
Michael Gottesman [Wed, 26 Jun 2013 23:55:26 +0000 (23:55 +0000)]
[APFloat] Added unittest for APFloat.multiply that checks special cases, result categories, and result status.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185044 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[APFloat] Added unittest for APFloat.subtract that checks special cases, result categ...
Michael Gottesman [Wed, 26 Jun 2013 23:55:23 +0000 (23:55 +0000)]
[APFloat] Added unittest for APFloat.subtract that checks special cases, result categories, and result status.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185043 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoSLP: When searching for vectorization opportunities scan the blocks in post-order...
Nadav Rotem [Wed, 26 Jun 2013 23:44:45 +0000 (23:44 +0000)]
SLP: When searching for vectorization opportunities scan the blocks in post-order because we grow chains upwards.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185041 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoSLP: Dont erase instructions during vectorization because it prevents the outerloops...
Nadav Rotem [Wed, 26 Jun 2013 23:43:23 +0000 (23:43 +0000)]
SLP: Dont erase instructions during vectorization because it prevents the outerloops from iterating over the instructions.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185040 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoIn InstCombine{AddSub,MulDivRem} convert APFloat.isFiniteNonZero() && !APFloat.isDeno...
Michael Gottesman [Wed, 26 Jun 2013 23:17:31 +0000 (23:17 +0000)]
In InstCombine{AddSub,MulDivRem} convert APFloat.isFiniteNonZero() && !APFloat.isDenormal => APFloat.isNormal.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185037 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[APFloat] Convert all references to fcNormal to references to isFiniteNonZero().
Michael Gottesman [Wed, 26 Jun 2013 23:17:28 +0000 (23:17 +0000)]
[APFloat] Convert all references to fcNormal to references to isFiniteNonZero().

Currently inside APFloat fcNormal still implies the old definition of Normal
(i.e. isFiniteNonZero) instead of the proper IEEE-754R definition that the
external method isNormal() uses.

This patch prepares for the internal switch inside APFloat by converting all
references that check if a category is fcNormal directly with an indirect call
via isFiniteNonZero().

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185036 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRevert "Debug Info: clean up usage of Verify." as it's breaking bots.
Eric Christopher [Wed, 26 Jun 2013 22:44:57 +0000 (22:44 +0000)]
Revert "Debug Info: clean up usage of Verify." as it's breaking bots.

This reverts commit r185020

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185032 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoFix a crash bug in dumping options with groups
Reid Kleckner [Wed, 26 Jun 2013 22:43:37 +0000 (22:43 +0000)]
Fix a crash bug in dumping options with groups

Option groups don't have prefixes.  Option dumping is basically dead
code unless there is something wrong with the option table, so this
isn't an important crasher.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185031 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoClarify and doxygen-ify comments
Stephen Lin [Wed, 26 Jun 2013 22:27:50 +0000 (22:27 +0000)]
Clarify and doxygen-ify comments

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185030 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[Mips Disassembler] Have the DecodeCCRRegisterClass function use the getReg
Chad Rosier [Wed, 26 Jun 2013 22:23:32 +0000 (22:23 +0000)]
[Mips Disassembler] Have the DecodeCCRRegisterClass function use the getReg
function to lookup the proper tablegen'ed register enumeration.  Previously,
it was using the encoded value directly.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185026 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoARM: Proactively ensure that the LowerCallResult hack for 'this'-returns is not used...
Stephen Lin [Wed, 26 Jun 2013 21:42:14 +0000 (21:42 +0000)]
ARM: Proactively ensure that the LowerCallResult hack for 'this'-returns is not used for incompatible calling conventions.

(Currently, ARM 'this'-returns are handled in the standard calling convention case by treating R0 as preserved and doing some extra magic in LowerCallResult; this may not apply to calling conventions added in the future so this patch provides and documents an interface for indicating such)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185024 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoDebug Info: clean up usage of Verify.
Manman Ren [Wed, 26 Jun 2013 21:26:10 +0000 (21:26 +0000)]
Debug Info: clean up usage of Verify.

No functionality change.
It should suffice to check the type of a debug info metadata, instead of
calling Verify.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185020 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoMinor formatting fix to ARMBaseRegisterInfo::getCalleeSavedRegs
Stephen Lin [Wed, 26 Jun 2013 20:19:06 +0000 (20:19 +0000)]
Minor formatting fix to ARMBaseRegisterInfo::getCalleeSavedRegs

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185016 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRename PathV2 to just Path now that it is the only one.
Rafael Espindola [Wed, 26 Jun 2013 19:33:03 +0000 (19:33 +0000)]
Rename PathV2 to just Path now that it is the only one.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185015 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[mips] Do not emit ".option pic0" if target is mips64.
Akira Hatanaka [Wed, 26 Jun 2013 19:08:49 +0000 (19:08 +0000)]
[mips] Do not emit ".option pic0" if target is mips64.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185012 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[mips] Improve code generation for constant multiplication using shifts, adds and
Akira Hatanaka [Wed, 26 Jun 2013 18:48:17 +0000 (18:48 +0000)]
[mips] Improve code generation for constant multiplication using shifts, adds and
subs.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185011 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[APFloat] Removed debugging cruft that snuck in.
Michael Gottesman [Wed, 26 Jun 2013 17:59:36 +0000 (17:59 +0000)]
[APFloat] Removed debugging cruft that snuck in.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184974 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoThe SLP Vectorizer works across basic blocks. Update the docs.
Nadav Rotem [Wed, 26 Jun 2013 17:59:35 +0000 (17:59 +0000)]
The SLP Vectorizer works across basic blocks. Update the docs.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184973 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoUse enums instead of raw octal values.
Rafael Espindola [Wed, 26 Jun 2013 17:28:04 +0000 (17:28 +0000)]
Use enums instead of raw octal values.

Patch by ç½—勇刚(Yonggang Luo).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184971 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoErase all of the instructions that we RAUWed
Nadav Rotem [Wed, 26 Jun 2013 17:16:09 +0000 (17:16 +0000)]
Erase all of the instructions that we RAUWed

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184969 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdd a subtarget feature 'v8' to the ARM backend.
Joey Gouly [Wed, 26 Jun 2013 16:58:26 +0000 (16:58 +0000)]
Add a subtarget feature 'v8' to the ARM backend.
This allows for targeting the ARMv8 AArch32 variant.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184967 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoDo not add cse-ed instructions into the visited map because we dont want to consider...
Nadav Rotem [Wed, 26 Jun 2013 16:54:53 +0000 (16:54 +0000)]
Do not add cse-ed instructions into the visited map because we dont want to consider them as a candidate for replacement of instructions to be visited.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184966 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoARM: fix more cases where predication may or may not be allowed
Tim Northover [Wed, 26 Jun 2013 16:52:40 +0000 (16:52 +0000)]
ARM: fix more cases where predication may or may not be allowed

Unfortunately this addresses two issues (by the time I'd disentangled the logic
it wasn't worth putting it back to half-broken):

+ Coprocessor instructions should all be predicable in Thumb mode.
+ BKPT should never be predicable.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184965 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoARM: allow predicated barriers in Thumb mode
Tim Northover [Wed, 26 Jun 2013 16:52:32 +0000 (16:52 +0000)]
ARM: allow predicated barriers in Thumb mode

The barrier instructions are only "always-execute" in ARM mode, they can quite
happily sit inside an IT block in Thumb.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184964 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRemove the 'generic' CPU from the ARM eabi attributes printer.
Joey Gouly [Wed, 26 Jun 2013 16:39:06 +0000 (16:39 +0000)]
Remove the 'generic' CPU from the ARM eabi attributes printer.
Make v4 the default ARM architecture attribute, to match CodeGen.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184962 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoPathV1 is deprecated since the 18th of Dec 2010. Remove it.
Rafael Espindola [Wed, 26 Jun 2013 16:24:35 +0000 (16:24 +0000)]
PathV1 is deprecated since the 18th of Dec 2010. Remove it.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184960 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdd a convenience functions that don't return if the directory existed.
Rafael Espindola [Wed, 26 Jun 2013 15:21:13 +0000 (15:21 +0000)]
Add a convenience functions that don't return if the directory existed.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184955 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRemove unused includes.
Rafael Espindola [Wed, 26 Jun 2013 13:54:34 +0000 (13:54 +0000)]
Remove unused includes.

llvm itself is now PathV1 clean.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184947 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[PowerPC] Accept 17-bit signed immediates for addis
Ulrich Weigand [Wed, 26 Jun 2013 13:49:53 +0000 (13:49 +0000)]
[PowerPC] Accept 17-bit signed immediates for addis

The assembler currently strictly verifies that immediates for
s16imm operands are in range (-32768 ... 32767).  This matches
the behaviour of the GNU assembler, with one exception: gas
allows, as a special case, operands in an extended range
(-65536 .. 65535) for the addis instruction only (and its
extended mnemonic lis).

The main reason for this seems to be to allow using unsigned
16-bit operands for lis, e.g. like lis %r1, 0xfedc.

Since this has been supported by gas for a long time, and
assembler source code seen "in the wild" actually exploits
this feature, this patch adds equivalent support to LLVM
for compatibility reasons.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184946 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[PowerPC] Support symbolic u16imm operands
Ulrich Weigand [Wed, 26 Jun 2013 13:49:15 +0000 (13:49 +0000)]
[PowerPC] Support symbolic u16imm operands

Currently, all instructions taking s16imm operands support symbolic
operands.  However, for u16imm operands, we only support actual
immediate integers.  This causes the assembler to reject code like

  ori %r5, %r5, symbol@l

This patch changes the u16imm operand definition to likewise
accept symbolic operands.  In fact, s16imm and u16imm can
share the same encoding routine, now renamed to getImm16Encoding.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184944 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoARM: operands should be explicit when disassembled
Amaury de la Vieuville [Wed, 26 Jun 2013 13:39:07 +0000 (13:39 +0000)]
ARM: operands should be explicit when disassembled

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184943 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[Sparc]: Add memory operands for the frame references in the storeRegToStackSlot
Venkatraman Govindaraju [Wed, 26 Jun 2013 12:40:16 +0000 (12:40 +0000)]
[Sparc]: Add memory operands for the frame references in the storeRegToStackSlot
and loadRegFromStackSlot.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184935 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoFixed a comment.
Elena Demikhovsky [Wed, 26 Jun 2013 12:15:53 +0000 (12:15 +0000)]
Fixed a comment.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184933 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoSuppress llvm/test/Other/can-execute.txt on msys bash.
NAKAMURA Takumi [Wed, 26 Jun 2013 10:56:44 +0000 (10:56 +0000)]
Suppress llvm/test/Other/can-execute.txt on msys bash.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184932 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoOptimized integer vector multiplication operation by replacing it with shift/xor...
Elena Demikhovsky [Wed, 26 Jun 2013 10:55:03 +0000 (10:55 +0000)]
Optimized integer vector multiplication operation by replacing it with shift/xor/sub when it is possible. Fixed a bug in SDIV, where the const operand is not a splat constant vector.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184931 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[asan] workaround for PR16277: don't instrument AllocaInstr with alignment more than...
Kostya Serebryany [Wed, 26 Jun 2013 09:49:52 +0000 (09:49 +0000)]
[asan] workaround for PR16277: don't instrument AllocaInstr with alignment more than the redzone size

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184928 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[asan] add option -asan-keep-uninstrumented-functions
Kostya Serebryany [Wed, 26 Jun 2013 09:18:17 +0000 (09:18 +0000)]
[asan] add option -asan-keep-uninstrumented-functions

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184927 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoProvide bootstrap support to build only llvm+clang when using cmake.
Andy Gibbs [Wed, 26 Jun 2013 08:05:08 +0000 (08:05 +0000)]
Provide bootstrap support to build only llvm+clang when using cmake.

Where a source tree is complete with lld, lldb and polly, it may not be possible to use cmake to configure build scripts if the host compiler it not capable of compiling these sub-projects.  This change makes it possible to first build a bootstrap clang compiler when can then be used to build a complete llvm toolchain.  An example bootstrap build sequence could be as follows:

$ mkdir bootstrap
$ cd bootstrap
$ cmake -G 'Unix Makefiles'
        -DCMAKE_BUILD_TYPE:STRING=Release
        -DCMAKE_PREFIX_PATH:STRING=$(pwd)
        -DLLVM_TARGETS_TO_BUILD:STRING=host
        -DLLVM_INCLUDE_TOOLS:STRING=bootstrap-only
        ../source
$ make clang  # build clang only for host

$ cd ..
$ export CC=$(realpath bootstrap/bin)/clang
$ export CXX=$(realpath bootstrap/bin)/clang++

$ mkdir final
$ cd final
$ cmake -G 'Unix Makefiles' ../source
$ make all check-all

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184924 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoSupport using "host" as a target in LLVM_TARGETS_TO_BUILD when using cmake.
Andy Gibbs [Wed, 26 Jun 2013 07:57:53 +0000 (07:57 +0000)]
Support using "host" as a target in LLVM_TARGETS_TO_BUILD when using cmake.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184923 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRemove calls to Path in #ifdefs that don't seem to be used in any of the bots :-(
Rafael Espindola [Wed, 26 Jun 2013 06:10:32 +0000 (06:10 +0000)]
Remove calls to Path in #ifdefs that don't seem to be used in any of the bots :-(

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184920 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdd a simpler version of remove_all.
Rafael Espindola [Wed, 26 Jun 2013 06:06:54 +0000 (06:06 +0000)]
Add a simpler version of remove_all.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184919 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoFix the build when __APPLE__ is defined.
Rafael Espindola [Wed, 26 Jun 2013 05:25:44 +0000 (05:25 +0000)]
Fix the build when __APPLE__ is defined.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184917 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRemove sys::GetMainExecutable.
Rafael Espindola [Wed, 26 Jun 2013 05:05:37 +0000 (05:05 +0000)]
Remove sys::GetMainExecutable.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184916 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoPort GetMainExecutable over to PathV2.
Rafael Espindola [Wed, 26 Jun 2013 05:01:35 +0000 (05:01 +0000)]
Port GetMainExecutable over to PathV2.

I will remove the V1 version as soon as I change clang in the next commit.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184914 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRemove PathWithStatus.
Rafael Espindola [Wed, 26 Jun 2013 04:15:55 +0000 (04:15 +0000)]
Remove PathWithStatus.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184910 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agodbgs() << Instruction doesn't print a newline on the end any more. Update these
Nick Lewycky [Wed, 26 Jun 2013 00:30:18 +0000 (00:30 +0000)]
dbgs() << Instruction doesn't print a newline on the end any more. Update these
debug statements to add a missing newline. Also canonicalize to '\n' instead of
"\n"; the latter calls a function with a loop the former does not.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184897 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agos/C++0x/C++11/
Adrian Prantl [Tue, 25 Jun 2013 23:42:37 +0000 (23:42 +0000)]
s/C++0x/C++11/

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184892 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoMerge isReachable into isBackedge.
Jakob Stoklund Olesen [Tue, 25 Jun 2013 23:32:10 +0000 (23:32 +0000)]
Merge isReachable into isBackedge.

Prefer using RPO.lookup() instead of RPO[] which can mutate the map.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184891 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoSLPVectorizer: support slp-vectorization of PHINodes between basic blocks
Nadav Rotem [Tue, 25 Jun 2013 23:04:09 +0000 (23:04 +0000)]
SLPVectorizer: support slp-vectorization of PHINodes between basic blocks

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184888 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoPrint block frequencies in decimal form.
Jakob Stoklund Olesen [Tue, 25 Jun 2013 21:57:38 +0000 (21:57 +0000)]
Print block frequencies in decimal form.

This is easier to read than the internal fixed-point representation.

If anybody knows the correct algorithm for converting fixed-point
numbers to base 10, feel free to fix it.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184881 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600: Use new getNamedOperandIdx function generated by TableGen
Tom Stellard [Tue, 25 Jun 2013 21:22:18 +0000 (21:22 +0000)]
R600: Use new getNamedOperandIdx function generated by TableGen

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184880 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoTableGen: Generate a function for getting operand indices based on their defined...
Tom Stellard [Tue, 25 Jun 2013 21:22:09 +0000 (21:22 +0000)]
TableGen: Generate a function for getting operand indices based on their defined names

This patch modifies TableGen to generate a function in
${TARGET}GenInstrInfo.inc called getNamedOperandIdx(), which can be used
to look up indices for operands based on their names.

In order to activate this feature for an instruction, you must set the
UseNamedOperandTable bit.

For example, if you have an instruction like:

def ADD : TargetInstr <(outs GPR:$dst), (ins GPR:$src0, GPR:$src1)>;

You can look up the operand indices using the new function, like this:

Target::getNamedOperandIdx(Target::ADD, Target::OpName::dst)  => 0
Target::getNamedOperandIdx(Target::ADD, Target::OpName::src0) => 1
Target::getNamedOperandIdx(Target::ADD, Target::OpName::src1) => 2

The operand names are case sensitive, so $dst and $DST are considered
different operands.

This change is useful for R600 which has instructions with a large number
of operands, many of which model single bit instruction configuration
values.  These configuration bits are common across most instructions,
but may have a different operand index depending on the instruction type.
It is useful to have a convenient way to look up the operand indices,
so these bits can be generically set on any instruction.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184879 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoX86 cost model: Vectorizing integer division is a bad idea
Arnold Schwaighofer [Tue, 25 Jun 2013 19:14:09 +0000 (19:14 +0000)]
X86 cost model: Vectorizing integer division is a bad idea

radar://14057959

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184872 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoFix SROA to avoid unnecessary scalar conversions for 1-element vectors.
Bob Wilson [Tue, 25 Jun 2013 19:09:50 +0000 (19:09 +0000)]
Fix SROA to avoid unnecessary scalar conversions for 1-element vectors.

When a 1-element vector alloca is promoted, a store instruction can often be
rewritten without converting the value to a scalar and using an insertelement
instruction to stuff it into the new alloca.  This patch just adds a check
to skip that conversion when it is unnecessary.  This turns out to be really
important for some ARM Neon operations where <1 x i64> is used to get around
the fact that i64 is not a legal type.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184870 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRemove unused code. No functionality change.
Manman Ren [Tue, 25 Jun 2013 18:49:55 +0000 (18:49 +0000)]
Remove unused code. No functionality change.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184866 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoThe GCDA 402 format won't have a second checksum either.
Bill Wendling [Tue, 25 Jun 2013 18:13:52 +0000 (18:13 +0000)]
The GCDA 402 format won't have a second checksum either.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184864 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[PowerPC] Support @got modifier
Ulrich Weigand [Tue, 25 Jun 2013 16:49:50 +0000 (16:49 +0000)]
[PowerPC] Support @got modifier

Add VK_... values and relocation types necessary to support
the @got family of modifiers.  Used by the asm parser only.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184860 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoMove GetEXESuffix to the one place it is used.
Rafael Espindola [Tue, 25 Jun 2013 14:42:30 +0000 (14:42 +0000)]
Move GetEXESuffix to the one place it is used.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184853 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRemove sys::PathSeparator.
Rafael Espindola [Tue, 25 Jun 2013 14:32:45 +0000 (14:32 +0000)]
Remove sys::PathSeparator.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184852 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600: Consolidate expansion of v2i32/v4i32 ops for EG/SI
Aaron Watry [Tue, 25 Jun 2013 13:55:57 +0000 (13:55 +0000)]
R600: Consolidate expansion of v2i32/v4i32 ops for EG/SI

By default, we expand these operations for both EG and SI. Move the
duplicated code into a common space for now. If the targets ever actually
implement these operations as instructions, we can override that in the relevant
target.

Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184848 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600: Add v2i32 test for vselect
Aaron Watry [Tue, 25 Jun 2013 13:55:54 +0000 (13:55 +0000)]
R600: Add v2i32 test for vselect

Note: Only adding test for evergreen, not SI yet.

When I attempted to expand vselect for SI, I got the following:
llc: /home/awatry/src/llvm/lib/CodeGen/SelectionDAG/LegalizeIntegerTypes.cpp:522:
llvm::SDValue llvm::DAGTypeLegalizer::PromoteIntRes_SETCC(llvm::SDNode*):
Assertion `SVT.isVector() == N->getOperand(0).getValueType().isVector() &&
"Vector compare must return a vector result!"' failed.

Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184847 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600/SI: Expand xor v2i32/v4i32
Aaron Watry [Tue, 25 Jun 2013 13:55:52 +0000 (13:55 +0000)]
R600/SI: Expand xor v2i32/v4i32

Add test cases for both vector sizes on SI and also add v2i32 test for EG.

Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184846 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600: Add v2i32 test for setcc on evergreen
Aaron Watry [Tue, 25 Jun 2013 13:55:49 +0000 (13:55 +0000)]
R600: Add v2i32 test for setcc on evergreen

No test/expansion for SI has been added yet. Attempts to expand this
operation for SI resulted in a stacktrace in (IIRC) LegalizeIntegerTypes
which was complaining about vector comparisons being required to return
a vector type.

Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184845 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600/SI: Expand urem of v2i32/v4i32 for SI
Aaron Watry [Tue, 25 Jun 2013 13:55:46 +0000 (13:55 +0000)]
R600/SI: Expand urem of v2i32/v4i32 for SI

Also add lit test for both cases on SI, and v2i32 for evergreen.

Note: I followed the guidance of the v4i32 EG check... UREM produces really
complex code, so let's just check that the instruction was lowered
successfully.

Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184844 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600/SI: Expand udiv v[24]i32 for SI and v2i32 for EG
Aaron Watry [Tue, 25 Jun 2013 13:55:43 +0000 (13:55 +0000)]
R600/SI: Expand udiv v[24]i32 for SI and v2i32 for EG

Also add lit test for both cases on SI, and v2i32 for evergreen.

Note: I followed the guidance of the v4i32 EG check... UDIV produces really
complex code, so let's just check that the instruction was lowered
successfully.

Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184843 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600/SI: Expand ashr of v2i32/v4i32 for SI
Aaron Watry [Tue, 25 Jun 2013 13:55:40 +0000 (13:55 +0000)]
R600/SI: Expand ashr of v2i32/v4i32 for SI

Also add lit test for both cases on SI, and v2i32 for evergreen.

Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184842 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600/SI: Expand srl of v2i32/v4i32 for SI
Aaron Watry [Tue, 25 Jun 2013 13:55:37 +0000 (13:55 +0000)]
R600/SI: Expand srl of v2i32/v4i32 for SI

Also add lit test for both cases on SI, and v2i32 for evergreen.

Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184841 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600/SI: Expand shl of v2i32/v4i32 for SI
Aaron Watry [Tue, 25 Jun 2013 13:55:32 +0000 (13:55 +0000)]
R600/SI: Expand shl of v2i32/v4i32 for SI

Also add lit test for both cases on SI, and v2i32 for evergreen.

Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184840 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600/SI: Expand or of v2i32/v4i32 for SI
Aaron Watry [Tue, 25 Jun 2013 13:55:29 +0000 (13:55 +0000)]
R600/SI: Expand or of v2i32/v4i32 for SI

Also add lit test for both cases on SI, and v2i32 for evergreen.

Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@184839 91177308-0d34-0410-b5e6-96231b3b80d8