Lang Hames [Mon, 3 Sep 2012 06:31:45 +0000 (06:31 +0000)]
When updating live range endpoints, make sure to preserve the early clobber bit.
Fixs PR13719.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163107
91177308-0d34-0410-b5e6-
96231b3b80d8
Bob Wilson [Mon, 3 Sep 2012 05:15:15 +0000 (05:15 +0000)]
Fix more fallout from r158919, similar to PR13547.
This code used to only handle malloc-like calls, which do not read memory.
r158919 changed it to check isNoAliasFn(), which includes strdup-like and
realloc-like calls, but it was not checking for dependencies on the memory
read by those calls.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163106
91177308-0d34-0410-b5e6-
96231b3b80d8
Chad Rosier [Mon, 3 Sep 2012 03:16:09 +0000 (03:16 +0000)]
Removed unused argument.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163104
91177308-0d34-0410-b5e6-
96231b3b80d8
Chris Lattner [Mon, 3 Sep 2012 02:58:21 +0000 (02:58 +0000)]
some peepholes that should match horizontal add/sub operations.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163103
91177308-0d34-0410-b5e6-
96231b3b80d8
Chad Rosier [Mon, 3 Sep 2012 02:06:46 +0000 (02:06 +0000)]
[ms-inline asm] Expose the Kind and Opcode variables from the
MatchInstructionImpl() function.
These values are used by the ConvertToMCInst() function to index into the
ConversionTable. The values are also needed to call the GetMCInstOperandNum()
function.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163101
91177308-0d34-0410-b5e6-
96231b3b80d8
Chad Rosier [Mon, 3 Sep 2012 01:55:11 +0000 (01:55 +0000)]
Move ErrorLoc decl into the scope where it's actually used.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163100
91177308-0d34-0410-b5e6-
96231b3b80d8
Nuno Lopes [Sun, 2 Sep 2012 15:16:51 +0000 (15:16 +0000)]
escape special char when handling CXX_FOR_OCAMLOPT
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163098
91177308-0d34-0410-b5e6-
96231b3b80d8
Nuno Lopes [Sun, 2 Sep 2012 15:07:25 +0000 (15:07 +0000)]
fix test's RUN lines
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163097
91177308-0d34-0410-b5e6-
96231b3b80d8
Nuno Lopes [Sun, 2 Sep 2012 14:42:56 +0000 (14:42 +0000)]
add support for ocaml 3.12
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163096
91177308-0d34-0410-b5e6-
96231b3b80d8
Nuno Lopes [Sun, 2 Sep 2012 14:19:21 +0000 (14:19 +0000)]
replace a couple of single-line comments with /* */ to fix the build of stuff depending on the C headers
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163095
91177308-0d34-0410-b5e6-
96231b3b80d8
Nadav Rotem [Sun, 2 Sep 2012 12:21:50 +0000 (12:21 +0000)]
Fix a typo.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163094
91177308-0d34-0410-b5e6-
96231b3b80d8
Nadav Rotem [Sun, 2 Sep 2012 12:10:19 +0000 (12:10 +0000)]
Not all targets have efficient ISel code generation for select instructions.
For example, the ARM target does not have efficient ISel handling for vector
selects with scalar conditions. This patch adds a TLI hook which allows the
different targets to report which selects are supported well and which selects
should be converted to CF duting codegen prepare.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163093
91177308-0d34-0410-b5e6-
96231b3b80d8
Benjamin Kramer [Sun, 2 Sep 2012 11:57:22 +0000 (11:57 +0000)]
LoopRotation: Make the brute force DomTree update more brute force.
We update until we hit a fixpoint. This is probably slow but also
slightly simplifies the code. It should also fix the occasional
invalid domtrees observed when building with expensive checking.
I couldn't find a case where this had a measurable slowdown, but
if someone finds a pathological case where it does we may have
to find a cleverer way of updating dominators here.
Thanks to Duncan for the test case.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163091
91177308-0d34-0410-b5e6-
96231b3b80d8
Logan Chien [Sun, 2 Sep 2012 09:29:46 +0000 (09:29 +0000)]
Rename ANDROIDEABI to Android.
Most of the code guarded with ANDROIDEABI are not
ARM-specific, and having no relation with arm-eabi.
Thus, it will be more natural to call this
environment "Android" instead of "ANDROIDEABI".
Note: We are not using ANDROID because several projects
are using "-DANDROID" as the conditional compilation
flag.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163087
91177308-0d34-0410-b5e6-
96231b3b80d8
Nadav Rotem [Sun, 2 Sep 2012 08:20:07 +0000 (08:20 +0000)]
Generate better select code by allowing the target to use scalar select, and not sign-extend.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163086
91177308-0d34-0410-b5e6-
96231b3b80d8
Pete Cooper [Sat, 1 Sep 2012 22:27:48 +0000 (22:27 +0000)]
Only legalise a VSELECT in to bitwise operations if the vector mask bool is zeros or all ones. A vector bool with just ones isn't suitable for masking with.
No test case unfortunately as i couldn't find a target which fit all
the conditions needed to hit this code.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163075
91177308-0d34-0410-b5e6-
96231b3b80d8
Benjamin Kramer [Sat, 1 Sep 2012 20:50:18 +0000 (20:50 +0000)]
PR13689: Check for backtrace(3) in the cmake build.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163074
91177308-0d34-0410-b5e6-
96231b3b80d8
Tim Northover [Sat, 1 Sep 2012 18:07:29 +0000 (18:07 +0000)]
Limit domain conversion to cases where it won't break dep chains.
NEON domain conversion was too heavy-handed with its widened
registers, which could have stripped existing instructions of their
dependency, leaving them vulnerable to scheduling errors.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163070
91177308-0d34-0410-b5e6-
96231b3b80d8
Pete Cooper [Sat, 1 Sep 2012 17:37:55 +0000 (17:37 +0000)]
Revert "Take account of boolean vector contents when promoting a build vector from i1 to some other type. rdar://problem/
12210060"
This reverts commit
5dd9e214fb92847e947f9edab170f9b4e52b908f.
Thanks to Duncan for explaining how this should have been done.
Conflicts:
test/CodeGen/X86/vec_select.ll
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163064
91177308-0d34-0410-b5e6-
96231b3b80d8
Logan Chien [Sat, 1 Sep 2012 15:06:36 +0000 (15:06 +0000)]
Fix Thumb2 fixup kind in the integrated-as.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163063
91177308-0d34-0410-b5e6-
96231b3b80d8
Logan Chien [Sat, 1 Sep 2012 14:58:11 +0000 (14:58 +0000)]
Add ARM ELF support to llvm-objdump.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163062
91177308-0d34-0410-b5e6-
96231b3b80d8
Logan Chien [Sat, 1 Sep 2012 14:43:30 +0000 (14:43 +0000)]
Code cleanup: tools/opt/opt.cpp
Remove unused local variable.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163061
91177308-0d34-0410-b5e6-
96231b3b80d8
Logan Chien [Sat, 1 Sep 2012 12:11:41 +0000 (12:11 +0000)]
Fix typo.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163059
91177308-0d34-0410-b5e6-
96231b3b80d8
Benjamin Kramer [Sat, 1 Sep 2012 12:04:51 +0000 (12:04 +0000)]
LoopRotation: Check some invariants of the dominator updating code.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163058
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Sat, 1 Sep 2012 06:33:50 +0000 (06:33 +0000)]
Typos
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163053
91177308-0d34-0410-b5e6-
96231b3b80d8
Owen Anderson [Sat, 1 Sep 2012 06:04:27 +0000 (06:04 +0000)]
Teach DAG combine a number of tricks to simplify FMA expressions in fast-math mode.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163051
91177308-0d34-0410-b5e6-
96231b3b80d8
Michael Liao [Sat, 1 Sep 2012 04:09:16 +0000 (04:09 +0000)]
Fix typo
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163049
91177308-0d34-0410-b5e6-
96231b3b80d8
NAKAMURA Takumi [Sat, 1 Sep 2012 00:26:28 +0000 (00:26 +0000)]
llvm/test/CodeGen/X86/fp-fast.ll: Suppress FMA4 on AMD Bulldozer host, corresponding to r162999.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163041
91177308-0d34-0410-b5e6-
96231b3b80d8
Manman Ren [Sat, 1 Sep 2012 00:17:06 +0000 (00:17 +0000)]
Fix Atom bots for r163036.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163040
91177308-0d34-0410-b5e6-
96231b3b80d8
Manman Ren [Fri, 31 Aug 2012 23:16:57 +0000 (23:16 +0000)]
SelectionDAG: when constructing VZEXT_LOAD from other loads, make sure its
output chain is correctly setup.
As an example, if the original load must happen before later stores, we need
to make sure the constructed VZEXT_LOAD is constrained to be before the stores.
rdar://
11457792
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163036
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Fri, 31 Aug 2012 23:10:34 +0000 (23:10 +0000)]
Mark FMA4 instructions as commutable and add them to the folding tables.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163035
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Fri, 31 Aug 2012 22:39:14 +0000 (22:39 +0000)]
Make sure to build libpthread to check for HAVE_PTHREAD_MUTEX_LOCK.
Patch by Brad Smith!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163033
91177308-0d34-0410-b5e6-
96231b3b80d8
Chad Rosier [Fri, 31 Aug 2012 22:12:31 +0000 (22:12 +0000)]
Remove an unused argument. The MCInst opcode is set in the ConvertToMCInst()
function nowadays.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163030
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Fri, 31 Aug 2012 22:12:16 +0000 (22:12 +0000)]
Add selection of RegOp2MemOpTable3 to canFoldMemoryOperand
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163029
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Fri, 31 Aug 2012 20:50:53 +0000 (20:50 +0000)]
Add MachineInstr::tieOperands, remove setIsTied().
Manage tied operands entirely internally to MachineInstr. This makes it
possible to change the representation of tied operands, as I will do
shortly.
The constraint that tied uses and defs must be in the same order was too
restrictive.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163021
91177308-0d34-0410-b5e6-
96231b3b80d8
Michael Liao [Fri, 31 Aug 2012 20:12:31 +0000 (20:12 +0000)]
Fix PR12359
- In addition to undefined, if V2 is zero vector, skip 2nd PSHUFB and POR as
well as PSHUFB will zero elements with negative indices.
Patch by Sriram Murali <sriram.murali@intel.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163018
91177308-0d34-0410-b5e6-
96231b3b80d8
Jack Carter [Fri, 31 Aug 2012 18:06:48 +0000 (18:06 +0000)]
The instruction DINS may be transformed into DINSU or DEXTM depending
on the size of the extraction and its position in the 64 bit word.
This patch allows support of the dext transformations with mips64 direct
object output.
0 <= msb < 32 0 <= lsb < 32 0 <= pos < 32 1 <= size <= 32
DINS
The field is entirely contained in the right-most word of the doubleword
32 <= msb < 64 0 <= lsb < 32 0 <= pos < 32 2 <= size <= 64
DINSM
The field straddles the words of the doubleword
32 <= msb < 64 32 <= lsb < 64 32 <= pos < 64 1 <= size <= 32
DINSU
The field is entirely contained in the left-most word of the doubleword
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163010
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Fri, 31 Aug 2012 17:31:28 +0000 (17:31 +0000)]
Move the GCOVFormat enums into their own namespace per the LLVM coding standard.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163008
91177308-0d34-0410-b5e6-
96231b3b80d8
Chad Rosier [Fri, 31 Aug 2012 17:24:10 +0000 (17:24 +0000)]
Add a comment to explain what's really going on.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163005
91177308-0d34-0410-b5e6-
96231b3b80d8
Chad Rosier [Fri, 31 Aug 2012 16:41:07 +0000 (16:41 +0000)]
The ConvertToMCInst() function can't fail, so remove the now dead Match_ConversionFail enum.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163002
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Fri, 31 Aug 2012 16:31:13 +0000 (16:31 +0000)]
Mark FMA3 instructions as commutable so that the operands to the multiply part can be commuted.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163001
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Fri, 31 Aug 2012 16:30:05 +0000 (16:30 +0000)]
Use CloneMachineInstr to make a new MI in commuteInstruction to make the code tolerant of instructions with more than two input operands.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163000
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Fri, 31 Aug 2012 15:40:30 +0000 (15:40 +0000)]
Add support for converting llvm.fma to fma4 instructions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162999
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Fri, 31 Aug 2012 15:34:59 +0000 (15:34 +0000)]
Don't enforce ordered inline asm operands.
I was too optimistic, inline asm can have tied operands that don't
follow the def order.
Fixes PR13742.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162998
91177308-0d34-0410-b5e6-
96231b3b80d8
Benjamin Kramer [Fri, 31 Aug 2012 12:43:07 +0000 (12:43 +0000)]
Clean up ProfileDataLoader a bit.
- Overloading operator<< for raw_ostream and pointers is dangerous, it alters
the behavior of code that includes the header.
- Remove unused ID.
- Use LLVM's byte swapping helpers instead of a hand-coded.
- Make ReadProfilingData work directly on a pointer.
No functionality change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162992
91177308-0d34-0410-b5e6-
96231b3b80d8
NAKAMURA Takumi [Fri, 31 Aug 2012 10:02:22 +0000 (10:02 +0000)]
llvm/test/CodeGen/X86/vec_select.ll: Fix failure on xmm-less hosts, to add -mattr=+sse2.
FIXME: Should this be tested with both +avx and -avx,+sse2?
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162983
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Fri, 31 Aug 2012 05:18:31 +0000 (05:18 +0000)]
Cleanups due to feedback. No functionality change. Patch by Alistair.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162979
91177308-0d34-0410-b5e6-
96231b3b80d8
Michael Liao [Fri, 31 Aug 2012 03:01:35 +0000 (03:01 +0000)]
Clean up AddedComplexity further after adding UseSSEx
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162973
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Fri, 31 Aug 2012 02:08:34 +0000 (02:08 +0000)]
Fix a couple of typos in EmitAtomic.
Thumb2 instructions are mostly constrained to rGPR, not tGPR which is
for Thumb1.
rdar://problem/
12203728
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162968
91177308-0d34-0410-b5e6-
96231b3b80d8
Jim Grosbach [Fri, 31 Aug 2012 00:30:30 +0000 (00:30 +0000)]
X86: Fix encoding of 'movd %xmm0, %rax'
The assembly string for the VMOVPQIto64rr instruction incorrectly lacked the 'v'
prefix, resulting in mis-assembly of the vanilla movd instruction.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162963
91177308-0d34-0410-b5e6-
96231b3b80d8
Chad Rosier [Fri, 31 Aug 2012 00:03:31 +0000 (00:03 +0000)]
With the fix in r162954/162955 every cvt function returns true. Thus, have
the ConvertToMCInst() return void, rather then a bool. Update all the cvt
functions as well.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162961
91177308-0d34-0410-b5e6-
96231b3b80d8
Pete Cooper [Thu, 30 Aug 2012 23:58:52 +0000 (23:58 +0000)]
Take account of boolean vector contents when promoting a build vector from i1 to some other type. rdar://problem/
12210060
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162960
91177308-0d34-0410-b5e6-
96231b3b80d8
Owen Anderson [Thu, 30 Aug 2012 23:51:20 +0000 (23:51 +0000)]
Try to make this test more generic to unbreak buildbots.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162958
91177308-0d34-0410-b5e6-
96231b3b80d8
Owen Anderson [Thu, 30 Aug 2012 23:35:16 +0000 (23:35 +0000)]
Teach the DAG combiner to turn chains of FADDs (x+x+x+x+...) into FMULs by constants. This is only enabled in unsafe FP math mode, since it does not preserve rounding effects for all such constants.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162956
91177308-0d34-0410-b5e6-
96231b3b80d8
Chad Rosier [Thu, 30 Aug 2012 23:22:05 +0000 (23:22 +0000)]
Fix for r162954. Return the Error.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162955
91177308-0d34-0410-b5e6-
96231b3b80d8
Chad Rosier [Thu, 30 Aug 2012 23:20:38 +0000 (23:20 +0000)]
Move a check to the validateInstruction() function where it more properly belongs.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162954
91177308-0d34-0410-b5e6-
96231b3b80d8
Michael Gottesman [Thu, 30 Aug 2012 23:11:49 +0000 (23:11 +0000)]
[llvm] Updated the test fold-vector-select so that we test the vector selects exhaustively.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162953
91177308-0d34-0410-b5e6-
96231b3b80d8
Chad Rosier [Thu, 30 Aug 2012 23:00:00 +0000 (23:00 +0000)]
Typo.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162952
91177308-0d34-0410-b5e6-
96231b3b80d8
Chad Rosier [Thu, 30 Aug 2012 21:47:00 +0000 (21:47 +0000)]
Whitespace.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162946
91177308-0d34-0410-b5e6-
96231b3b80d8
Chad Rosier [Thu, 30 Aug 2012 21:46:00 +0000 (21:46 +0000)]
Whitespace.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162945
91177308-0d34-0410-b5e6-
96231b3b80d8
Chad Rosier [Thu, 30 Aug 2012 21:43:05 +0000 (21:43 +0000)]
Hoist a check to eliminate obvious mismatches as early as possible. Also, fix
an 80-column violation in the generated code. No functional change intended.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162944
91177308-0d34-0410-b5e6-
96231b3b80d8
Nadav Rotem [Thu, 30 Aug 2012 19:17:29 +0000 (19:17 +0000)]
Currently targets that do not support selects with scalar conditions and vector operands - scalarize the code. ARM is such a target
because it does not support CMOV of vectors. To implement this efficientlyi, we broadcast the condition bit and use a sequence of NAND-OR
to select between the two operands. This is the same sequence we use for targets that don't have vector BLENDs (like SSE2).
rdar://
12201387
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162926
91177308-0d34-0410-b5e6-
96231b3b80d8
Chad Rosier [Thu, 30 Aug 2012 17:59:25 +0000 (17:59 +0000)]
[ms-inline asm] Add a new function, GetMCInstOperandNum, to the
AsmMatcherEmitter. This function maps inline assembly operands to MCInst
operands.
For example, '__asm mov j, eax' is represented by the follow MCInst:
<MCInst 1460 <MCOperand Reg:0> <MCOperand Imm:1> <MCOperand Reg:0>
<MCOperand Expr:(j)> <MCOperand Reg:0> <MCOperand Reg:43>>
The first 5 MCInst operands are a result of j matching as a memory operand
consisting of a BaseReg (Reg:0), MemScale (Imm:1), MemIndexReg(Reg:0),
Expr (Expr:(j), and a MemSegReg (Reg:0). The 6th MCInst operand represents
the eax register (Reg:43).
This translation is necessary to determine the Input and Output Exprs. If a
single asm operand maps to multiple MCInst operands, the index of the first
MCInst operand is returned. Ideally, it would return the operand we really
care out (i.e., the Expr:(j) in this case), but I haven't found an easy way
of doing this yet.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162920
91177308-0d34-0410-b5e6-
96231b3b80d8
Michael Liao [Thu, 30 Aug 2012 16:54:46 +0000 (16:54 +0000)]
Introduce 'UseSSEx' to force SSE legacy encoding
- Add 'UseSSEx' to force SSE legacy insn not being selected when AVX is
enabled.
As the penalty of inter-mixing SSE and AVX instructions, we need
prevent SSE legacy insn from being generated except explicitly
specified through some intrinsics. For patterns supported by both
SSE and AVX, so far, we force AVX insn will be tried first relying on
AddedComplexity or position in td file. It's error-prone and
introduces bugs accidentally.
'UseSSEx' is disabled when AVX is turned on. For SSE insns inherited
by AVX, we need this predicate to force VEX encoding or SSE legacy
encoding only.
For insns not inherited by AVX, we still use the previous predicates,
i.e. 'HasSSEx'. So far, these insns fall into the following
categories:
* SSE insns with MMX operands
* SSE insns with GPR/MEM operands only (xFENCE, PREFETCH, CLFLUSH,
CRC, and etc.)
* SSE4A insns.
* MMX insns.
* x87 insns added by SSE.
2 test cases are modified:
- test/CodeGen/X86/fast-isel-x86-64.ll
AVX code generation is different from SSE one. 'vcvtsi2sdq' cannot be
selected by fast-isel due to complicated pattern and fast-isel
fallback to materialize it from constant pool.
- test/CodeGen/X86/widen_load-1.ll
AVX code generation is different from SSE one after fixing SSE/AVX
inter-mixing. Exec-domain fixing prefers 'vmovapd' instead of
'vmovaps'.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162919
91177308-0d34-0410-b5e6-
96231b3b80d8
NAKAMURA Takumi [Thu, 30 Aug 2012 16:22:32 +0000 (16:22 +0000)]
ADTTests: [CMake] Exclude DenseMapTest.cpp and SmallVectorTest.cpp on MSVC9 due to its bug.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162918
91177308-0d34-0410-b5e6-
96231b3b80d8
NAKAMURA Takumi [Thu, 30 Aug 2012 16:22:26 +0000 (16:22 +0000)]
Apply "/Og-" also to MSC15(aka VS9) on VMCore/Function.cpp.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162917
91177308-0d34-0410-b5e6-
96231b3b80d8
NAKAMURA Takumi [Thu, 30 Aug 2012 15:52:29 +0000 (15:52 +0000)]
PPCISelLowering.cpp: Fix r162725.
[Tobias von Koch] What's happening here is that the CR6SET/CR6UNSET is breaking the chain of register copies glued to the function call (BL_SVR4 node). The scheduler then moves other instructions in between those and the function call, which isn't good!
Right. That's the case where there is no chain of register copies before the call, so InFlag == 0... Attached is a new revision of the patch which should fix this for good.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162916
91177308-0d34-0410-b5e6-
96231b3b80d8
NAKAMURA Takumi [Thu, 30 Aug 2012 15:52:23 +0000 (15:52 +0000)]
PPCISelLowering.cpp: Whitespace.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162915
91177308-0d34-0410-b5e6-
96231b3b80d8
Michael Ilseman [Thu, 30 Aug 2012 15:45:16 +0000 (15:45 +0000)]
test
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162914
91177308-0d34-0410-b5e6-
96231b3b80d8
Benjamin Kramer [Thu, 30 Aug 2012 15:42:45 +0000 (15:42 +0000)]
Fix test case.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162913
91177308-0d34-0410-b5e6-
96231b3b80d8
Benjamin Kramer [Thu, 30 Aug 2012 15:39:42 +0000 (15:39 +0000)]
LoopRotate: Also rotate loops with multiple exits.
The old PHI updating code in loop-rotate was replaced with SSAUpdater a while
ago, it has no problems with comples PHIs. What had to be fixed is detecting
whether a loop was already rotated and updating dominators when multiple exits
were present.
This change increases overall code size a bit, mostly due to additional loop
unrolling opportunities. Passes test-suite and selfhost with -verify-dom-info.
Fixes PR7447.
Thanks to Andy for the input on the domtree updating code.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162912
91177308-0d34-0410-b5e6-
96231b3b80d8
Benjamin Kramer [Thu, 30 Aug 2012 15:07:40 +0000 (15:07 +0000)]
InstCombine: Fix comment to reflect the code.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162911
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Thu, 30 Aug 2012 14:39:06 +0000 (14:39 +0000)]
Don't use MCInstrDesc flags for implicit operands.
When a MachineInstr is constructed, its implicit operands are added
first, then the explicit operands are inserted before the implicits.
MCInstrDesc has oprand flags like early clobber and operand ties that
apply to the explicit operands.
Don't look at those flags when the implicit operands are first added in
the explicit operands's positions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162910
91177308-0d34-0410-b5e6-
96231b3b80d8
Alexey Samsonov [Thu, 30 Aug 2012 13:47:13 +0000 (13:47 +0000)]
Whitespace
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162907
91177308-0d34-0410-b5e6-
96231b3b80d8
Nadav Rotem [Thu, 30 Aug 2012 11:23:20 +0000 (11:23 +0000)]
It is illegal to transform (sdiv (ashr X c1) c2) -> (sdiv x (2^c1 * c2)),
because C always rounds towards zero.
Thanks Dirk and Ben.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162899
91177308-0d34-0410-b5e6-
96231b3b80d8
Tim Northover [Thu, 30 Aug 2012 10:17:45 +0000 (10:17 +0000)]
Add support for moving pure S-register to NEON pipeline if desired
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162898
91177308-0d34-0410-b5e6-
96231b3b80d8
Nadav Rotem [Thu, 30 Aug 2012 09:23:13 +0000 (09:23 +0000)]
remove old docs
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162896
91177308-0d34-0410-b5e6-
96231b3b80d8
Alexey Samsonov [Thu, 30 Aug 2012 07:49:50 +0000 (07:49 +0000)]
Refactor fetching file/line info from DWARFContext to simplify the
code and allow better code reuse. Make the code a bit more conforming
to LLVM code style.
No functionality change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162895
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Thu, 30 Aug 2012 07:34:22 +0000 (07:34 +0000)]
Add FMA to switch statement in VectorLegalizer::LegalizeOp so that it can be expanded when it isn't legal.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162894
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Thu, 30 Aug 2012 07:13:41 +0000 (07:13 +0000)]
Add support for FMA to WidenVectorResult.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162893
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Thu, 30 Aug 2012 06:56:15 +0000 (06:56 +0000)]
Only perform DAG combine on FMAs of legal types.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162892
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Thu, 30 Aug 2012 01:32:31 +0000 (01:32 +0000)]
Pass by pointer and not std::string.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162888
91177308-0d34-0410-b5e6-
96231b3b80d8
Michael Liao [Thu, 30 Aug 2012 00:43:57 +0000 (00:43 +0000)]
Should put test case under test/ExecutionEngine/MCJIT/
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162885
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Thu, 30 Aug 2012 00:34:21 +0000 (00:34 +0000)]
Revert r162855 in favor of changing clang to emit the absolute coverage file path.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162883
91177308-0d34-0410-b5e6-
96231b3b80d8
Will Dietz [Thu, 30 Aug 2012 00:30:21 +0000 (00:30 +0000)]
Fix HashString's Bernstein hash to use unsigned chars, as is usually done.
Changes the hash result for strings containing characters
with values >= 128, such as UTF8 strings (not normal ASCII).
Changed mostly so we match other implementations.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162882
91177308-0d34-0410-b5e6-
96231b3b80d8
Michael Liao [Thu, 30 Aug 2012 00:30:16 +0000 (00:30 +0000)]
Fix PR13727
- The root cause is that target constant materialization in X86 fast-isel
creates a PC-rel addressing which may overflow 32-bit range in non-Small code
model if .rodata section is allocated too far away from code segment in
MCJIT, which uses Large code model so far.
- Follow the similar logic to fix non-Small code model in fast-isel by skipping
non-Small code model.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162881
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Wed, 29 Aug 2012 23:52:55 +0000 (23:52 +0000)]
Add a bit of documentation to copyPhysReg.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162879
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Wed, 29 Aug 2012 23:52:52 +0000 (23:52 +0000)]
Verify the order of tied operands in inline asm.
When there are multiple tied use-def pairs on an inline asm instruction,
the tied uses must appear in the same order as the defs.
It is possible to write an LLVM IR inline asm instruction that breaks
this constraint, but there is no reason for a front end to emit the
operands out of order.
The gnu inline asm syntax specifies tied operands as a single read/write
constraint "+r", so ouf of order operands are not possible.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162878
91177308-0d34-0410-b5e6-
96231b3b80d8
Benjamin Kramer [Wed, 29 Aug 2012 23:28:45 +0000 (23:28 +0000)]
Explicitly cast an expression to bool before handing it off to __builtin_expect.
Avoids surprises when someone uses LLVM_(UN)LIKELY with an integer.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162877
91177308-0d34-0410-b5e6-
96231b3b80d8
Benjamin Kramer [Wed, 29 Aug 2012 22:57:04 +0000 (22:57 +0000)]
Add some __builtin_expect magic to StringMap.
Tombstones and full hash collisions are rare, mark the "empty"
and "no collision" paths as likely. The bug in simplifycfg
that prevented the hints from being picked during selfhost
up was fixed recently :)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162874
91177308-0d34-0410-b5e6-
96231b3b80d8
Benjamin Kramer [Wed, 29 Aug 2012 22:57:00 +0000 (22:57 +0000)]
Replace the BUILTIN_EXPECT macro with a less horrible LLVM_LIKELY/LLVM_UNLIKELY interface.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162873
91177308-0d34-0410-b5e6-
96231b3b80d8
Owen Anderson [Wed, 29 Aug 2012 22:18:56 +0000 (22:18 +0000)]
Allow targets to specify a minimum supported NOP size when performing NOP padding. If the desired padding is smaller than the supported NOP size,
we will enlarge the padding to make it work.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162870
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Wed, 29 Aug 2012 22:02:00 +0000 (22:02 +0000)]
Set the isTied flags when building INLINEASM MachineInstrs.
For normal instructions, isTied() is set automatically by addOperand(),
based on MCInstrDesc, but inline asm has tied operands outside the
descriptor.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162869
91177308-0d34-0410-b5e6-
96231b3b80d8
Andrew Trick [Wed, 29 Aug 2012 21:46:38 +0000 (21:46 +0000)]
Preserve branch profile metadata during switch formation.
Patch by Michael Ilseman!
This fixes SimplifyCFGOpt::FoldValueComparisonIntoPredecessors to preserve metata when folding conditional branches into switches.
void foo(int x) {
if (x == 0)
bar(1);
else if (__builtin_expect(x == 10, 1))
bar(2);
else if (x == 20)
bar(3);
}
CFG:
B0
| \
| X0
B10
| \
| X10
B20
| \
E X20
Merge B0-B10:
w(B0-X0) = w(B0-X0)*sum-weights(B10) = w(B0-X0) * (w(B10-X10) + w(B10-B20))
w(B0-X10) = w(B0-B10) * w(B10-X10)
w(B0-B20) = w(B0-B10) * w(B10-B20)
B0 __
| \ \
| X10 X0
B20
| \
E X20
Merge B0-B20:
w(B0-X0) = w(B0-X0) * sum-weights(B20) = w(B0-X0) * (w(B20-E) + w(B20-X20))
w(B0-X10) = w(B0-X10) * sum-weights(B20) = ...
w(B0-X20) = w(B0-B20) * w(B20-X20)
w(B0-E) = w(B0-B20) * w(B20-E)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162868
91177308-0d34-0410-b5e6-
96231b3b80d8
Andrew Trick [Wed, 29 Aug 2012 21:46:36 +0000 (21:46 +0000)]
whitespace
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162867
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Wed, 29 Aug 2012 21:19:21 +0000 (21:19 +0000)]
Rename hasVolatileMemoryRef() to hasOrderedMemoryRef().
Ordered memory operations are more constrained than volatile loads and
stores because they must be ordered with respect to all other memory
operations.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162861
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Wed, 29 Aug 2012 21:08:52 +0000 (21:08 +0000)]
Add MachineMemOperand::isUnordered().
This means the same as LoadInst/StoreInst::isUnordered(), and implies
!isVolatile().
Atomic loads and stored are also ordered, and this is the right method
to check if it is safe to reorder memory operations. Ordered atomics
can't be reordered wrt normal loads and stores, which is a stronger
constraint than volatile.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162859
91177308-0d34-0410-b5e6-
96231b3b80d8
Jakob Stoklund Olesen [Wed, 29 Aug 2012 20:48:45 +0000 (20:48 +0000)]
Don't move normal loads across volatile/atomic loads.
It is technically allowed to move a normal load across a volatile load,
but probably not a good idea.
It is not allowed to move a load across an atomic load with
Ordering > Monotonic, and we model those with MOVolatile as well.
I recently removed the mayStore flag from atomic load instructions, so
they don't need a pseudo-opcode. This patch makes up for the difference.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162857
91177308-0d34-0410-b5e6-
96231b3b80d8
Michael Liao [Wed, 29 Aug 2012 20:32:13 +0000 (20:32 +0000)]
fix C++ comment in C header
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162856
91177308-0d34-0410-b5e6-
96231b3b80d8
Bill Wendling [Wed, 29 Aug 2012 20:30:44 +0000 (20:30 +0000)]
Use the full path to output the .gcda file.
This lets the user run the program from a different directory and still have the
.gcda files show up in the correct place.
<rdar://problem/
12179524>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162855
91177308-0d34-0410-b5e6-
96231b3b80d8