oota-llvm.git
11 years agoR600/SI: remove shader type intrinsic
Christian Konig [Thu, 7 Mar 2013 09:03:46 +0000 (09:03 +0000)]
R600/SI: remove shader type intrinsic

Just encode the type as target specific attribute.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176622 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoR600/SI: switch types of SGPRs to v*i8
Christian Konig [Thu, 7 Mar 2013 09:03:38 +0000 (09:03 +0000)]
R600/SI: switch types of SGPRs to v*i8

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176621 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoR600/SI: fix unused variable warning
Christian Konig [Thu, 7 Mar 2013 09:03:30 +0000 (09:03 +0000)]
R600/SI: fix unused variable warning

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176620 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoSwitch from a version 4.2/4.4 switch to a four-byte version string to be put
Nick Lewycky [Thu, 7 Mar 2013 08:28:49 +0000 (08:28 +0000)]
Switch from a version 4.2/4.4 switch to a four-byte version string to be put
into the actual gcov file.

Instead of using the bottom 4 bytes as the function identifier, use a counter.
This makes the identifier numbers stable across multiple runs.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176616 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoNo need to go through int64 and APInt when generating a new constant.
Nadav Rotem [Thu, 7 Mar 2013 06:34:49 +0000 (06:34 +0000)]
No need to go through int64 and APInt when generating a new constant.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176615 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoSDAG: Handle scalarizing an extend of a <1 x iN> vector.
Jim Grosbach [Thu, 7 Mar 2013 05:47:54 +0000 (05:47 +0000)]
SDAG: Handle scalarizing an extend of a <1 x iN> vector.

Just scalarize the element and rebuild a vector of the result type
from that.

rdar://13281568

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176614 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoRemove GCDAProfiling.c. This copy is old, the copy in compiler-rt is newer and
Nick Lewycky [Thu, 7 Mar 2013 02:03:08 +0000 (02:03 +0000)]
Remove GCDAProfiling.c. This copy is old, the copy in compiler-rt is newer and
is the one that should be used.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176608 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoDebug Info: store the files and directories for each compile unit.
Manman Ren [Thu, 7 Mar 2013 01:42:00 +0000 (01:42 +0000)]
Debug Info: store the files and directories for each compile unit.

We now emit a line table for each compile unit. To reduce the prologue size
of each line table, the files and directories used by each compile unit are
stored in std::map<unsigned, std::vector< > > instead of std::vector< >.

The prologue for a lto'ed image can be as big as 93K. Duplicating 93K for each
compile unit causes a huge increase of debug info. With this patch, each
prologue will only emit the files required by the compile unit.

rdar://problem/13342023

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176605 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoArrayRef has a OneElt constructor. Beautify the code.
Nadav Rotem [Thu, 7 Mar 2013 01:38:04 +0000 (01:38 +0000)]
ArrayRef has a OneElt constructor. Beautify the code.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176604 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoSwitch from std::vector to ArrayRef. Speedup FoldBitCast by 5x.
Nadav Rotem [Thu, 7 Mar 2013 01:30:40 +0000 (01:30 +0000)]
Switch from std::vector to ArrayRef. Speedup FoldBitCast by 5x.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176602 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoSimplifyCFG fix for volatile load/store.
Andrew Trick [Thu, 7 Mar 2013 01:03:35 +0000 (01:03 +0000)]
SimplifyCFG fix for volatile load/store.

Fixes rdar:13349374.

Volatile loads and stores need to be preserved even if the language
standard says they are undefined. "volatile" in this context means "get
out of the way compiler, let my platform handle it".

Additionally, this is the only way I know of with llvm to write to the
first page (when hardware allows) without dropping to assembly.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176599 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoFix two remaining issue after fixing PR15355 when CMOV is not available
Michael Liao [Thu, 7 Mar 2013 01:01:29 +0000 (01:01 +0000)]
Fix two remaining issue after fixing PR15355 when CMOV is not available

- Phi nodes should be replaced/updated after lowering CMOV into branch
  because 'mainMBB' updating operand in Phi node is changed.
- Add EFLAGS in livein before lowering the 2nd CMOV. It's necessary as
  we will reuse the EFLAGS generated before the 1st lowered CMOV, which
  won't clobber EFLAGS. However, we need explicitly specify that.
- '-attr=-cmov' test case are added.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176598 91177308-0d34-0410-b5e6-96231b3b80d8

11 years ago[mips] Custom-legalize BR_JT.
Akira Hatanaka [Wed, 6 Mar 2013 21:32:03 +0000 (21:32 +0000)]
[mips] Custom-legalize BR_JT.

In N64-static, GOT address is needed to compute the branch address.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176580 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoGeneralize my previous fix for -print-options.
Andrew Trick [Wed, 6 Mar 2013 19:04:56 +0000 (19:04 +0000)]
Generalize my previous fix for -print-options.

Always print options that differ from their implicit default. At least
for simple option types.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176572 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoRemove tailing whitespaces
Michael Liao [Wed, 6 Mar 2013 18:24:34 +0000 (18:24 +0000)]
Remove tailing whitespaces

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176570 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoGive -loop-vectorize an explicit default.
Andrew Trick [Wed, 6 Mar 2013 18:22:22 +0000 (18:22 +0000)]
Give -loop-vectorize an explicit default.

This way, clang -mllvm -print-options shows that the driver is overriding it.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176569 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoMemory Dependence Analysis (not mem-dep test) take advantage of "invariant.load"...
Shuxin Yang [Wed, 6 Mar 2013 17:48:48 +0000 (17:48 +0000)]
Memory Dependence Analysis (not mem-dep test) take advantage of "invariant.load" metadata.

The "invariant.load" metadata indicates the memory unit being accessed is immutable.
A load annotated with this metadata can be moved across any store.

As I am not sure if it is legal to move such loads across barrier/fence, this
change dose not allow such transformation.

rdar://11311484

Thank Arnold for code review.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176562 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoInstCombine: Don't shrink allocas when combining with a bitcast.
Jim Grosbach [Wed, 6 Mar 2013 05:44:53 +0000 (05:44 +0000)]
InstCombine: Don't shrink allocas when combining with a bitcast.

When considering folding a bitcast of an alloca into the alloca itself,
make sure we don't shrink the amount of memory being allocated, or
things rapidly go sideways.

rdar://13324424

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176547 91177308-0d34-0410-b5e6-96231b3b80d8

11 years ago[mips] Add a line which checks function name. Rename file.
Akira Hatanaka [Wed, 6 Mar 2013 01:58:03 +0000 (01:58 +0000)]
[mips] Add a line which checks function name. Rename file.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176543 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoFix PR15355
Michael Liao [Wed, 6 Mar 2013 00:17:04 +0000 (00:17 +0000)]
Fix PR15355

- Clear 'mayStore' flag when loading from the atomic variable before the
  spin loop
- Clear kill flag from one use to multiple use in registers forming the
  address to that atomic variable
- don't use a physical register as live-in register in BB (neither entry
  nor landing pad.) by copying it into virtual register

(patch by Cameron Zwarich)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176538 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoUse dyn_cast instead of isa && cast. No functionality change.
Jakub Staszak [Wed, 6 Mar 2013 00:16:16 +0000 (00:16 +0000)]
Use dyn_cast instead of isa && cast. No functionality change.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176537 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoRemove duplicated forward declaration.
Jakub Staszak [Wed, 6 Mar 2013 00:04:32 +0000 (00:04 +0000)]
Remove duplicated forward declaration.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176536 91177308-0d34-0410-b5e6-96231b3b80d8

11 years ago[mips] Remove android calling convention.
Akira Hatanaka [Tue, 5 Mar 2013 23:22:30 +0000 (23:22 +0000)]
[mips] Remove android calling convention.

This calling convention was added just to handle functions which return vector
of floats. The fix committed in r165585 solves the problem.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176530 91177308-0d34-0410-b5e6-96231b3b80d8

11 years ago[mips] Fix MipsCC::analyzeReturn so that, in soft-float mode, fp128 gets
Akira Hatanaka [Tue, 5 Mar 2013 22:54:59 +0000 (22:54 +0000)]
[mips] Fix MipsCC::analyzeReturn so that, in soft-float mode, fp128 gets
returned in registers $2 and $4.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176527 91177308-0d34-0410-b5e6-96231b3b80d8

11 years ago[mips] Fix MipsTargetLowering::LowerCallResult and LowerReturn to correctly
Akira Hatanaka [Tue, 5 Mar 2013 22:41:55 +0000 (22:41 +0000)]
[mips] Fix MipsTargetLowering::LowerCallResult and LowerReturn to correctly
handle fp128 returns.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176523 91177308-0d34-0410-b5e6-96231b3b80d8

11 years ago[mips] Fix MipsTargetLowering::LowerCall to pass fp128 arguments in floating
Akira Hatanaka [Tue, 5 Mar 2013 22:20:28 +0000 (22:20 +0000)]
[mips] Fix MipsTargetLowering::LowerCall to pass fp128 arguments in floating
point registers.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176521 91177308-0d34-0410-b5e6-96231b3b80d8

11 years ago[mips] Correct handling of fp128 (long double) formals and read long double
Akira Hatanaka [Tue, 5 Mar 2013 22:13:04 +0000 (22:13 +0000)]
[mips] Correct handling of fp128 (long double) formals and read long double
parameters from floating point registers if target is mips64 hard float.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176520 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoFix a few typos in comments.
Jakub Staszak [Tue, 5 Mar 2013 22:05:16 +0000 (22:05 +0000)]
Fix a few typos in comments.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176519 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoAdd some constantness.
Jakub Staszak [Tue, 5 Mar 2013 22:01:15 +0000 (22:01 +0000)]
Add some constantness.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176518 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agostd::distance() == 0 means that iterators are equal. No functionality change.
Jakub Staszak [Tue, 5 Mar 2013 21:53:57 +0000 (21:53 +0000)]
std::distance() == 0 means that iterators are equal. No functionality change.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176516 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoAdd more functions to the TLI.
Meador Inge [Tue, 5 Mar 2013 21:47:40 +0000 (21:47 +0000)]
Add more functions to the TLI.

This patch adds many more functions to the target library information.
All of the functions being added were discovered while doing the migration
of the simplify-libcalls attribute annotation functionality to the
functionattrs pass.  As a part of that work the attribute annotation logic
will query TLI to determine if a function should be annotated or not.

Signed-off-by: Meador Inge <meadori@codesourcery.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176514 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoreverting patch 176508.
Jyotsna Verma [Tue, 5 Mar 2013 20:29:23 +0000 (20:29 +0000)]
reverting patch 176508.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176513 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoHexagon: Add support for lowering block address.
Jyotsna Verma [Tue, 5 Mar 2013 19:37:46 +0000 (19:37 +0000)]
Hexagon: Add support for lowering block address.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176508 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoR600: Do not predicate vector op
Vincent Lejeune [Tue, 5 Mar 2013 19:12:06 +0000 (19:12 +0000)]
R600: Do not predicate vector op

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176507 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoHexagon: Expand addc, adde, subc and sube.
Jyotsna Verma [Tue, 5 Mar 2013 19:04:47 +0000 (19:04 +0000)]
Hexagon: Expand addc, adde, subc and sube.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176505 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoUse the right number of slashes in comment string
Arnold Schwaighofer [Tue, 5 Mar 2013 19:04:12 +0000 (19:04 +0000)]
Use the right number of slashes in comment string

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176504 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoFixes a test by replacing .align by .p2align and setting triples explicitly.
Eli Bendersky [Tue, 5 Mar 2013 18:56:14 +0000 (18:56 +0000)]
Fixes a test by replacing .align by .p2align and setting triples explicitly.

Patch by David Sehr

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176502 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoUpdate cmake build.
Benjamin Kramer [Tue, 5 Mar 2013 18:54:05 +0000 (18:54 +0000)]
Update cmake build.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176501 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoHexagon: Use MO operand flags to mark constant extended instructions.
Jyotsna Verma [Tue, 5 Mar 2013 18:51:42 +0000 (18:51 +0000)]
Hexagon: Use MO operand flags to mark constant extended instructions.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176500 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoHexagon: Add encoding bits to the TFR64 instructions.
Jyotsna Verma [Tue, 5 Mar 2013 18:42:28 +0000 (18:42 +0000)]
Hexagon: Add encoding bits to the TFR64 instructions.
Set imMoveImm, isAsCheapAsAMove flags for TFRI instructions.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176499 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoR600: initial scheduler code
Vincent Lejeune [Tue, 5 Mar 2013 18:41:32 +0000 (18:41 +0000)]
R600: initial scheduler code

This is a skeleton for a pre-RA MachineInstr scheduler strategy. Currently
it only tries to expose more parallelism for ALU instructions (this also
makes the distribution of GPR channels more uniform and increases the
chances of ALU instructions to be packed together in a single VLIW group).
Also it tries to reduce clause switching by grouping instruction of the
same kind (ALU/FETCH/CF) together.

Vincent Lejeune:
 - Support for VLIW4 Slot assignement
 - Recomputation of ScheduleDAG to get more parallelism opportunities

Tom Stellard:
 - Fix assertion failure when trying to determine an instruction's slot
   based on its destination register's class
 - Fix some compiler warnings

Vincent Lejeune: [v2]
 - Remove recomputation of ScheduleDAG (will be provided in a later patch)
 - Improve estimation of an ALU clause size so that heuristic does not emit cf
 instructions at the wrong position.
 - Make schedule heuristic smarter using SUnit Depth
 - Take constant read limitations into account

Vincent Lejeune: [v3]
 - Fix some uninitialized values in ConstPair
 - Add asserts to ensure an ALU slot is always populated

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176498 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoClarify comment for function getObjectSize
Arnold Schwaighofer [Tue, 5 Mar 2013 16:53:24 +0000 (16:53 +0000)]
Clarify comment for function getObjectSize

Clarify that we mean the object starting at the pointer to the end of the
underlying object and not the size of the whole allocated object.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176491 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoAdd a test that .align directives on capable processors use long NOPs.
David Sehr [Tue, 5 Mar 2013 16:46:54 +0000 (16:46 +0000)]
Add a test that .align directives on capable processors use long NOPs.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176490 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoR600: Remove LowerConstCopyPass and lower CONST_COPY right after ISel.
Vincent Lejeune [Tue, 5 Mar 2013 15:04:55 +0000 (15:04 +0000)]
R600: Remove LowerConstCopyPass and lower CONST_COPY right after ISel.

Maintaining CONST_COPY Instructions until Pre Emit may prevent some ifcvt case
and taking them in account for scheduling is difficult for no real benefit.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176488 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoR600: Turn BUILD_VECTOR into Reg_Sequence
Vincent Lejeune [Tue, 5 Mar 2013 15:04:49 +0000 (15:04 +0000)]
R600: Turn BUILD_VECTOR into Reg_Sequence

Reviewed-by: Tom Stellard <thomas.stellard at amd.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176487 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoR600: CONST_ADDRESS node is not marked as mayLoad anymore
Vincent Lejeune [Tue, 5 Mar 2013 15:04:42 +0000 (15:04 +0000)]
R600: CONST_ADDRESS node is not marked as mayLoad anymore

Reviewed-by: Tom Stellard <thomas.stellard at amd.com>
mayLoad complexify scheduling and does not bring any usefull info
as the location is not writeable at all.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176486 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoR600: Use MUL_IEEE for trig/fdiv intrinsic
Vincent Lejeune [Tue, 5 Mar 2013 15:04:37 +0000 (15:04 +0000)]
R600: Use MUL_IEEE for trig/fdiv intrinsic

Reviewed-by: Tom Stellard <thomas.stellard at amd.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176485 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoR600: Add support for indirect addressing of non default const buffer
Vincent Lejeune [Tue, 5 Mar 2013 15:04:29 +0000 (15:04 +0000)]
R600: Add support for indirect addressing of non default const buffer

NOTE: This is a candidate for the Mesa stable branch.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176484 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoPrint a warning message if compiler-rt can't be built because of old CMake version...
Alexey Samsonov [Tue, 5 Mar 2013 14:43:07 +0000 (14:43 +0000)]
Print a warning message if compiler-rt can't be built because of old CMake version to make this requirement more visible to users

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176481 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agollvm/test/CodeGen/Mips/mips64-f128.ll: Add explicit -mtriple=mips64el-unknown-unknown...
NAKAMURA Takumi [Tue, 5 Mar 2013 02:18:59 +0000 (02:18 +0000)]
llvm/test/CodeGen/Mips/mips64-f128.ll: Add explicit -mtriple=mips64el-unknown-unknown to appease win32.

FIXME: Is it expected for win32 to affect mips targets?

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176471 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agollvm/test/CodeGen/Thumb/iabs.ll: Add explicit -mtriple=thumb-unknown-unknown to appea...
NAKAMURA Takumi [Tue, 5 Mar 2013 02:18:52 +0000 (02:18 +0000)]
llvm/test/CodeGen/Thumb/iabs.ll: Add explicit -mtriple=thumb-unknown-unknown to appease win32 hosts.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176470 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoRemove unused #includes.
Bill Wendling [Tue, 5 Mar 2013 01:00:45 +0000 (01:00 +0000)]
Remove unused #includes.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176467 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoThe current X86 NOP padding uses one long NOP followed by the remainder in
David Sehr [Tue, 5 Mar 2013 00:02:23 +0000 (00:02 +0000)]
The current X86 NOP padding uses one long NOP followed by the remainder in
one-byte NOPs.  If the processor actually executes those NOPs, as it sometimes
does with aligned bundling, this can have a performance impact.  From my
micro-benchmarks run on my one machine, a 15-byte NOP followed by twelve
one-byte NOPs is about 20% worse than a 15 followed by a 12.  This patch
changes NOP emission to emit as many 15-byte (the maximum) as possible followed
by at most one shorter NOP.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176464 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoCheck isDiscardableIfUnused, rather than hasLocalLinkage, when bumping
Lang Hames [Mon, 4 Mar 2013 22:40:44 +0000 (22:40 +0000)]
Check isDiscardableIfUnused, rather than hasLocalLinkage, when bumping
GlobalValue linkage up to ExternalLinkage in the ExtractGV pass. This
prevents linkonce and linkonce_odr symbols from being DCE'd.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176459 91177308-0d34-0410-b5e6-96231b3b80d8

11 years ago[mips] Print move instructions.
Akira Hatanaka [Mon, 4 Mar 2013 22:25:01 +0000 (22:25 +0000)]
[mips] Print move instructions.

"move $4, $5" is printed instead of "or $4, $5, $zero".

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176455 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoMips specific inline assembler constraint 'R'
Jack Carter [Mon, 4 Mar 2013 21:33:15 +0000 (21:33 +0000)]
Mips specific inline assembler constraint 'R'

'R' An address that can be sued in a non-macro load or store.
This patch includes a positive test case.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176452 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoReapply r176381, writing the CHECKs in a more forgiving manner to account for
Eli Bendersky [Mon, 4 Mar 2013 18:20:31 +0000 (18:20 +0000)]
Reapply r176381, writing the CHECKs in a more forgiving manner to account for
running llvm-objdump on Darwin.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176443 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoBypass Slow Divides
Preston Gurd [Mon, 4 Mar 2013 18:13:57 +0000 (18:13 +0000)]
Bypass Slow Divides

* Only apply divide bypass optimization when not optimizing for size.
* Fixed bug caused by constant for 0 value of type Int32,
  used dividend type to generate the constant instead.
* For atom x86-64 apply the divide bypass to use 16-bit divides instead of
  64-bit divides when operand values are small enough.
* Added lit tests for 64-bit divide bypass.

Patch by Tyler Nowicki!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176442 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoR600: Clean up datalayout strings so they better match hardware capabilities
Tom Stellard [Mon, 4 Mar 2013 17:40:28 +0000 (17:40 +0000)]
R600: Clean up datalayout strings so they better match hardware capabilities

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176439 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoMips ISD typo
Jia Liu [Mon, 4 Mar 2013 01:06:54 +0000 (01:06 +0000)]
Mips ISD typo

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176426 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoARM: Creating a vector from a lane of another.
Jim Grosbach [Sat, 2 Mar 2013 20:16:24 +0000 (20:16 +0000)]
ARM: Creating a vector from a lane of another.

The VDUP instruction source register doesn't allow a non-constant lane
index, so make sure we don't construct a ARM::VDUPLANE node asking it to
do so.

rdar://13328063
http://llvm.org/bugs/show_bug.cgi?id=13963

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176413 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoClean up code format a bit.
Jim Grosbach [Sat, 2 Mar 2013 20:16:19 +0000 (20:16 +0000)]
Clean up code format a bit.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176412 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoTidy up. Trailing whitespace.
Jim Grosbach [Sat, 2 Mar 2013 20:16:15 +0000 (20:16 +0000)]
Tidy up. Trailing whitespace.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176411 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoARM NEON: Fix v2f32 float intrinsics
Arnold Schwaighofer [Sat, 2 Mar 2013 19:38:33 +0000 (19:38 +0000)]
ARM NEON: Fix v2f32 float intrinsics

Mark them as expand, they are not legal as our backend does not match them.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176410 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agorecommit r172363 & r171325 (reverted in r172756)
Nuno Lopes [Sat, 2 Mar 2013 11:36:24 +0000 (11:36 +0000)]
recommit r172363 & r171325 (reverted in r172756)
This adds minimalistic support for PHI nodes to llvm.objectsize() evaluation

fingers crossed so that it does break clang boostrap again..

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176408 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoadd getUnderlyingObjectSize()
Nuno Lopes [Sat, 2 Mar 2013 11:23:34 +0000 (11:23 +0000)]
add getUnderlyingObjectSize()
this is similar to getObjectSize(), but doesnt subtract the offset
tweak the BasicAA code accordingly (per PR14988)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176407 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoX86 cost model: Adjust cost for custom lowered vector multiplies
Arnold Schwaighofer [Sat, 2 Mar 2013 04:02:52 +0000 (04:02 +0000)]
X86 cost model: Adjust cost for custom lowered vector multiplies

This matters for example in following matrix multiply:

int **mmult(int rows, int cols, int **m1, int **m2, int **m3) {
  int i, j, k, val;
  for (i=0; i<rows; i++) {
    for (j=0; j<cols; j++) {
      val = 0;
      for (k=0; k<cols; k++) {
        val += m1[i][k] * m2[k][j];
      }
      m3[i][j] = val;
    }
  }
  return(m3);
}

Taken from the test-suite benchmark Shootout.

We estimate the cost of the multiply to be 2 while we generate 9 instructions
for it and end up being quite a bit slower than the scalar version (48% on my
machine).

Also, properly differentiate between avx1 and avx2. On avx-1 we still split the
vector into 2 128bits and handle the subvector muls like above with 9
instructions.
Only on avx-2 will we have a cost of 9 for v4i64.

I changed the test case in test/Transforms/LoopVectorize/X86/avx1.ll to use an
add instead of a mul because with a mul we now no longer vectorize. I did
verify that the mul would be indeed more expensive when vectorized with 3
kernels:

for (i ...)
   r += a[i] * 3;
for (i ...)
  m1[i] = m1[i] * 3; // This matches the test case in avx1.ll
and a matrix multiply.

In each case the vectorized version was considerably slower.

radar://13304919

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176403 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoAdded FIXME for future Hexagon cleanup.
Andrew Trick [Sat, 2 Mar 2013 01:43:08 +0000 (01:43 +0000)]
Added FIXME for future Hexagon cleanup.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176400 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoPR14448 - prevent the loop vectorizer from vectorizing the same loop twice.
Nadav Rotem [Sat, 2 Mar 2013 01:33:49 +0000 (01:33 +0000)]
PR14448 - prevent the loop vectorizer from vectorizing the same loop twice.
The LoopVectorizer often runs multiple times on the same function due to inlining.
When this happens the loop vectorizer often vectorizes the same loops multiple times, increasing code size and adding unneeded branches.
With this patch, the vectorizer during vectorization puts metadata on scalar loops and marks them as 'already vectorized' so that it knows to ignore them when it sees them a second time.

PR14448.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176399 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoModify {Call,Invoke}Inst::addAttribute to take an AttrKind.
Peter Collingbourne [Sat, 2 Mar 2013 01:20:18 +0000 (01:20 +0000)]
Modify {Call,Invoke}Inst::addAttribute to take an AttrKind.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176397 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoCMake: Always include the CheckCXXCompilerFlag in HandleLLVMOptions.cmake.
Jordan Rose [Sat, 2 Mar 2013 01:00:40 +0000 (01:00 +0000)]
CMake: Always include the CheckCXXCompilerFlag in HandleLLVMOptions.cmake.

Previously we relied on it being included by config-ix.cmake.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176396 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoRevert "Rewrite a test to count emitted instructions without using -stats"
Michael Gottesman [Sat, 2 Mar 2013 00:53:20 +0000 (00:53 +0000)]
Revert "Rewrite a test to count emitted instructions without using -stats"

This reverts commit aac7922b8fe7ae733d3fe6697e6789fd730315dc. I am reverting the
commit since it broke the phase 1 public buildbot for a few hours.

http://lab.llvm.org:8013/builders/clang-x86_64-darwin11-nobootstrap-RA/builds/2137

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176394 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoRemove duplicate line and move another closer to its actual use
Eli Bendersky [Fri, 1 Mar 2013 23:32:40 +0000 (23:32 +0000)]
Remove duplicate line and move another closer to its actual use

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176391 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoMIsched machine model: tablegen subtarget emitter improvement.
Andrew Trick [Fri, 1 Mar 2013 23:31:26 +0000 (23:31 +0000)]
MIsched machine model: tablegen subtarget emitter improvement.

Fix the way resources are counted. I'm taking some time to cleanup the
way MachineScheduler handles in-order machine resources. Eventually
we'll need more PPC/Atom test cases in tree.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176390 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoIn llvm::MemoryBuffer::getFile() remove an unnecessary stat call check.
Argyrios Kyrtzidis [Fri, 1 Mar 2013 22:48:51 +0000 (22:48 +0000)]
In llvm::MemoryBuffer::getFile() remove an unnecessary stat call check.

The sys::fs::is_directory() check is unnecessary because, if the filename is
a directory, the function will fail anyway with the same error code returned.
Remove the check to avoid an unnecessary stat call.

Someone needs to review on windows and see if the check is necessary there or not.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176386 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoFix my email address in CREDITS.TXT.
Stefanus Du Toit [Fri, 1 Mar 2013 22:20:03 +0000 (22:20 +0000)]
Fix my email address in CREDITS.TXT.

Checking to see if svn notifications also use correct address now.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176385 91177308-0d34-0410-b5e6-96231b3b80d8

11 years ago[mips] Fix inefficient code generation.
Akira Hatanaka [Fri, 1 Mar 2013 21:52:08 +0000 (21:52 +0000)]
[mips] Fix inefficient code generation.

This patch eliminates the need to emit a constant move instruction when this
pattern is matched:

(select (setgt a, Constant), T, F)

The pattern above effectively turns into this:

(conditional-move (setlt a, Constant + 1), F, T)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176384 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoRemoved extraneous #include "LLVMContextImpl.h" from lib/IR/Module.cpp
Jean-Luc Duprat [Fri, 1 Mar 2013 21:37:24 +0000 (21:37 +0000)]
Removed extraneous #include "LLVMContextImpl.h" from lib/IR/Module.cpp

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176382 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoRewrite a test to count emitted instructions without using -stats
Eli Bendersky [Fri, 1 Mar 2013 21:34:37 +0000 (21:34 +0000)]
Rewrite a test to count emitted instructions without using -stats

Also removed the comments of "should produce..." because they completely
don't match the actually produced output.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176381 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoFix indentation.
Akira Hatanaka [Fri, 1 Mar 2013 21:22:21 +0000 (21:22 +0000)]
Fix indentation.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176380 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoSet properties for f128 type.
Akira Hatanaka [Fri, 1 Mar 2013 21:11:44 +0000 (21:11 +0000)]
Set properties for f128 type.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176378 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoRewrite a test to check actual output rather than intermediate implementation
Eli Bendersky [Fri, 1 Mar 2013 20:54:00 +0000 (20:54 +0000)]
Rewrite a test to check actual output rather than intermediate implementation
detail.

The was this test was written, it was relying on an implementation detail
(fixups) and hence was very brittle (relying, among other things, on the
exact ordering of statistics printed by MC).

The test was rewritten to check a more observable output difference. While it
doesn't cover 100% of the things the original test covered, it's a good
practice to write regression tests this way. If we want to check that
internal details and invariants hold, such tests should be expressed as unit
tests.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176377 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoNo need to force-create clang-tools-extra lit.site.cfg
Edwin Vane [Fri, 1 Mar 2013 19:58:58 +0000 (19:58 +0000)]
No need to force-create clang-tools-extra lit.site.cfg

The make (all) target takes care of creating lit configs and auto-generating
tests. The problem with the original 'lit.site.cfg' target is it's not
recursive and doesn't fully create everything necessary for testing
clang-tools-extra.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176374 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoAdd regression tests (WORKSFORME)
Michael Liao [Fri, 1 Mar 2013 19:23:37 +0000 (19:23 +0000)]
Add regression tests (WORKSFORME)

- These tests wont't crash on trunk but would be better to add them so that
  they don't break again in the future.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176369 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoGenerate an error message instead of asserting or segfaulting when we can't
Chad Rosier [Fri, 1 Mar 2013 19:12:05 +0000 (19:12 +0000)]
Generate an error message instead of asserting or segfaulting when we can't
handle indirect register inputs.
rdar://13322011

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176367 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoLoopVectorize: Don't hang forever if a PHI only has skipped PHI uses.
Benjamin Kramer [Fri, 1 Mar 2013 19:07:31 +0000 (19:07 +0000)]
LoopVectorize: Don't hang forever if a PHI only has skipped PHI uses.

Fixes PR15384.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176366 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoCache the result of Function::getIntrinsicID() in a DenseMap attached to the LLVMContext.
Michael Ilseman [Fri, 1 Mar 2013 18:48:54 +0000 (18:48 +0000)]
Cache the result of Function::getIntrinsicID() in a DenseMap attached to the LLVMContext.

This reduces the time actually spent doing string to ID conversion and shows a 10% improvement in compile time for a particularly bad case that involves ARM Neon intrinsics (these have many overloads).

Patch by Jean-Luc Duprat!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176365 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoFix PR10475
Michael Liao [Fri, 1 Mar 2013 18:40:30 +0000 (18:40 +0000)]
Fix PR10475

- ISD::SHL/SRL/SRA must have either both scalar or both vector operands
  but TLI.getShiftAmountTy() so far only return scalar type. As a
  result, backend logic assuming that breaks.
- Rename the original TLI.getShiftAmountTy() to
  TLI.getScalarShiftAmountTy() and re-define TLI.getShiftAmountTy() to
  return target-specificed scalar type or the same vector type as the
  1st operand.
- Fix most TICG logic assuming TLI.getShiftAmountTy() a simple scalar
  type.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176364 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoAdd support for using non-pic code for arm and thumb1 when emitting the sjlj
Chad Rosier [Fri, 1 Mar 2013 18:30:38 +0000 (18:30 +0000)]
Add support for using non-pic code for arm and thumb1 when emitting the sjlj
dispatch code.  As far as I can tell the thumb2 code is behaving as expected.
I was able to compile and run the associated test case for both arm and thumb1.
rdar://13066352

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176363 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoR600/SI: fix sampler tests after fixing wait insertions
Christian Konig [Fri, 1 Mar 2013 17:39:05 +0000 (17:39 +0000)]
R600/SI: fix sampler tests after fixing wait insertions

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176359 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoHexagon: Add constant extender support framework.
Jyotsna Verma [Fri, 1 Mar 2013 17:37:13 +0000 (17:37 +0000)]
Hexagon: Add constant extender support framework.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176358 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agotest commit to use consistent comment notation.
Peng Cheng [Fri, 1 Mar 2013 16:49:35 +0000 (16:49 +0000)]
test commit to use consistent comment notation.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176353 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoGCInfoDeleter code cleanup after r175528
Yiannis Tsiouris [Fri, 1 Mar 2013 11:40:32 +0000 (11:40 +0000)]
GCInfoDeleter code cleanup after r175528

Remove GCInfoDeleter from passes and comments.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176347 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoR600/SI: handle all registers in copyPhysReg v2
Christian Konig [Fri, 1 Mar 2013 09:46:27 +0000 (09:46 +0000)]
R600/SI: handle all registers in copyPhysReg v2

v2: based on Michels patch, but now allows copying of all registers sizes.

Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Christian König <christian.koenig@amd.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176346 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoR600/SI: remove S_MOV immediate patterns
Christian Konig [Fri, 1 Mar 2013 09:46:22 +0000 (09:46 +0000)]
R600/SI: remove S_MOV immediate patterns

They won't match anyway.

Signed-off-by: Christian König <christian.koenig@amd.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176345 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoR600/SI: remove GPR*AlignEncode
Christian Konig [Fri, 1 Mar 2013 09:46:17 +0000 (09:46 +0000)]
R600/SI: remove GPR*AlignEncode

It's much easier to specify the encoding with tablegen directly.

Signed-off-by: Christian König <christian.koenig@amd.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176344 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoR600/SI: fix warning about overloaded virtual
Christian Konig [Fri, 1 Mar 2013 09:46:11 +0000 (09:46 +0000)]
R600/SI: fix warning about overloaded virtual

Signed-off-by: Christian König <christian.koenig@amd.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176343 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoR600/SI: fix inserting waits for unordered defines
Christian Konig [Fri, 1 Mar 2013 09:46:04 +0000 (09:46 +0000)]
R600/SI: fix inserting waits for unordered defines

Signed-off-by: Christian König <christian.koenig@amd.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176342 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoGCC thinks that this variable might be used uninitialized (it isn't).
Duncan Sands [Fri, 1 Mar 2013 09:46:03 +0000 (09:46 +0000)]
GCC thinks that this variable might be used uninitialized (it isn't).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176341 91177308-0d34-0410-b5e6-96231b3b80d8

11 years agoDocs for llvm-symbolizer command-line tool
Alexey Samsonov [Fri, 1 Mar 2013 07:58:27 +0000 (07:58 +0000)]
Docs for llvm-symbolizer command-line tool

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176337 91177308-0d34-0410-b5e6-96231b3b80d8