OSDN Git Service
Alexey Bataev [Mon, 13 Feb 2017 19:08:19 +0000 (19:08 +0000)]
[SLP] Test for extractelement cost fix.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294980
91177308-0d34-0410-b5e6-
96231b3b80d8
Taewook Oh [Mon, 13 Feb 2017 18:15:31 +0000 (18:15 +0000)]
Make MachineBasicBlock::updateTerminator to update DebugLoc as well
Summary:
Currently MachineBasicBlock::updateTerminator simply drops DebugLoc for newly created branch instructions, which may cause incorrect stepping and/or imprecise sample profile data. Below is an example:
```
1 extern int bar(int x);
2
3 int foo(int *begin, int *end) {
4 int *i;
5 int ret = 0;
6 for (
7 i = begin ;
8 i != end ;
9 i++)
10 {
11 ret += bar(*i);
12 }
13 return ret;
14 }
```
Below is a bitcode of 'foo' at the end of LLVM-IR level optimizations with -O3:
```
define i32 @foo(i32* readonly %begin, i32* readnone %end) !dbg !4 {
entry:
%cmp6 = icmp eq i32* %begin, %end, !dbg !9
br i1 %cmp6, label %for.end, label %for.body.preheader, !dbg !12
for.body.preheader: ; preds = %entry
br label %for.body, !dbg !13
for.body: ; preds = %for.body.preheader, %for.body
%ret.08 = phi i32 [ %add, %for.body ], [ 0, %for.body.preheader ]
%i.07 = phi i32* [ %incdec.ptr, %for.body ], [ %begin, %for.body.preheader ]
%0 = load i32, i32* %i.07, align 4, !dbg !13, !tbaa !15
%call = tail call i32 @bar(i32 %0), !dbg !19
%add = add nsw i32 %call, %ret.08, !dbg !20
%incdec.ptr = getelementptr inbounds i32, i32* %i.07, i64 1, !dbg !21
%cmp = icmp eq i32* %incdec.ptr, %end, !dbg !9
br i1 %cmp, label %for.end.loopexit, label %for.body, !dbg !12, !llvm.loop !22
for.end.loopexit: ; preds = %for.body
br label %for.end, !dbg !24
for.end: ; preds = %for.end.loopexit, %entry
%ret.0.lcssa = phi i32 [ 0, %entry ], [ %add, %for.end.loopexit ]
ret i32 %ret.0.lcssa, !dbg !24
}
```
where
```
!12 = !DILocation(line: 6, column: 3, scope: !11)
```
. As you can see, the terminator of 'entry' block, which is a loop control branch, has a DebugLoc of line 6, column 3. Howerver, after the execution of 'MachineBlock::updateTerminator' function, which is triggered by MachineSinking pass, the DebugLoc info is dropped as below (see there's no debug-location for JNE_1):
```
bb.0.entry:
successors: %bb.4(0x30000000), %bb.1.for.body.preheader(0x50000000)
liveins: %rdi, %rsi
%6 = COPY %rsi
%5 = COPY %rdi
%8 = SUB64rr %5, %6, implicit-def %eflags, debug-location !9
JNE_1 %bb.1.for.body.preheader, implicit %eflags
```
This patch addresses this issue and make newly created branch instructions to keep debug-location info.
Reviewers: aprantl, MatzeB, craig.topper, qcolombet
Reviewed By: qcolombet
Subscribers: qcolombet, llvm-commits
Differential Revision: https://reviews.llvm.org/D29596
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294976
91177308-0d34-0410-b5e6-
96231b3b80d8
Matthew Simpson [Mon, 13 Feb 2017 18:02:35 +0000 (18:02 +0000)]
Revert "[LV] Extend trunc optimization to all IVs with constant integer steps"
This reverts commit r294967. This patch caused execution time slowdowns in a
few LLVM test-suite tests, as reported by the clang-cmake-aarch64-quick bot.
I'm reverting to investigate.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294973
91177308-0d34-0410-b5e6-
96231b3b80d8
Quentin Colombet [Mon, 13 Feb 2017 17:38:59 +0000 (17:38 +0000)]
[FastISel] Add a diagnostic to warm on fallback.
This is consistent with what we do for GlobalISel. That way, it is easy
to see whether or not FastISel is able to fully select a function.
At some point we may want to switch that to an optimization remark.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294970
91177308-0d34-0410-b5e6-
96231b3b80d8
James Molloy [Mon, 13 Feb 2017 17:18:00 +0000 (17:18 +0000)]
[ARM] Fix crash caused by r294945
I'd missed a creator of FCMP nodes - duplicateCmp().
Kindly and promptly reported by Gabor Ballabas, due to his CSiBE test suite.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294968
91177308-0d34-0410-b5e6-
96231b3b80d8
Matthew Simpson [Mon, 13 Feb 2017 16:48:00 +0000 (16:48 +0000)]
[LV] Extend trunc optimization to all IVs with constant integer steps
This patch extends the optimization of truncations whose operand is an
induction variable with a constant integer step. Previously we were only
applying this optimization to the primary induction variable. However, the cost
model assumes the optimization is applied to the truncation of all integer
induction variables (even regardless of step type). The transformation is now
applied to the other induction variables, and I've updated the cost model to
ensure it is better in sync with the transformation we actually perform.
Differential Revision: https://reviews.llvm.org/D29847
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294967
91177308-0d34-0410-b5e6-
96231b3b80d8
Simon Dardis [Mon, 13 Feb 2017 16:42:35 +0000 (16:42 +0000)]
[mips] Fix failing test.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294966
91177308-0d34-0410-b5e6-
96231b3b80d8
Sanjay Patel [Mon, 13 Feb 2017 16:17:29 +0000 (16:17 +0000)]
fix documentation comments; NFC
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294964
91177308-0d34-0410-b5e6-
96231b3b80d8
Davide Italiano [Mon, 13 Feb 2017 16:08:36 +0000 (16:08 +0000)]
[llvm-lto2] Fix typo spotted by Teresa (r294885 post-commit review).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294962
91177308-0d34-0410-b5e6-
96231b3b80d8
Simon Dardis [Mon, 13 Feb 2017 16:06:48 +0000 (16:06 +0000)]
[mips] divide macro instruction cleanup.
Clean up the implementation of divide macro expansion by getting rid of a
FIXME regarding magic numbers and branch instructions. Match GAS' behaviour
for expansion of ddiv / div in the two and three operand cases. Add the two
operand alias for MIPSR6. Finally, optimize macro expansion cases where the
divisior is the $zero register.
Reviewers: slthakur
Differential Revision: https://reviews.llvm.org/D29887
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294960
91177308-0d34-0410-b5e6-
96231b3b80d8
Simon Pilgrim [Mon, 13 Feb 2017 15:31:08 +0000 (15:31 +0000)]
Fix indentation. NFCI.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294959
91177308-0d34-0410-b5e6-
96231b3b80d8
Davide Italiano [Mon, 13 Feb 2017 15:26:22 +0000 (15:26 +0000)]
[PM] Hook up the instrumented PGO machinery in the new PM.
Differential Revision: https://reviews.llvm.org/D29308
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294955
91177308-0d34-0410-b5e6-
96231b3b80d8
Davide Italiano [Mon, 13 Feb 2017 14:39:51 +0000 (14:39 +0000)]
[LTO] Make sure we flush buffers to work around linker shenanigans.
lld, at least, doesn't call global destructors by default (unless
--full-shutdown is passed) because it's, allegedly, expensive.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294953
91177308-0d34-0410-b5e6-
96231b3b80d8
Simon Pilgrim [Mon, 13 Feb 2017 14:20:13 +0000 (14:20 +0000)]
[X86][SSE] Add v4f32 and v2f64 extract to store tests
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294952
91177308-0d34-0410-b5e6-
96231b3b80d8
Sanne Wouda [Mon, 13 Feb 2017 14:07:45 +0000 (14:07 +0000)]
[CodeGen] fix alignment of JUMPTABLE_INSTS on v8M.base
Summary:
The attached test case fails with "fatal error: error in backend:
misaligned pc-relative fixup value" as the jump table is misaligned.
The EmitAlignment existed already for ARM and Thumb-1 code, but was
missing for Thumb-2.
The test checks that the fatal error disappears when generating an obj
file, as well as checking the align directive is there when producing an
asm file.
Reviewers: rengolin, grosbach, t.p.northover, jmolloy, SjoerdMeijer, samparker
Reviewed By: samparker
Subscribers: samparker, aemerson, llvm-commits
Differential Revision: https://reviews.llvm.org/D29650
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294950
91177308-0d34-0410-b5e6-
96231b3b80d8
James Molloy [Mon, 13 Feb 2017 14:07:39 +0000 (14:07 +0000)]
[Thumb-1] TBB generation: spot redefinitions of index register
We match a sequence of 3-4 instructions into a tTBB pseudo. One of our checks is that
a particular register in that sequence is killed (so it can be clobbered by the pseudo).
We weren't noticing if an errant MOV or other instruction had infiltrated the
sequence we were walking. If it had, and it defined the register we've already
identified as killed, it makes it live across the tBR_JT and thus unclobberable.
Notice this case and bail out.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294949
91177308-0d34-0410-b5e6-
96231b3b80d8
James Molloy [Mon, 13 Feb 2017 14:07:25 +0000 (14:07 +0000)]
[ARM] Register ConstantIslands with the pass manager
This allows us to use -stop-before/-stop-after/-run-pass - we can now write
.mir tests.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294948
91177308-0d34-0410-b5e6-
96231b3b80d8
Sanne Wouda [Mon, 13 Feb 2017 13:58:00 +0000 (13:58 +0000)]
[Assembler] Improve diagnostics for inline assembly.
Summary:
Keep a vector of LocInfos around; one for each call to EmitInlineAsm.
Since each call to EmitInlineAsm creates a new buffer in the inline asm
SourceMgr, we can use the buffer number to map to the right LocInfo.
Reviewers: rengolin, grosbach, rnk, echristo
Reviewed By: rnk
Subscribers: mehdi_amini, llvm-commits
Differential Revision: https://reviews.llvm.org/D29769
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294947
91177308-0d34-0410-b5e6-
96231b3b80d8
Simon Pilgrim [Mon, 13 Feb 2017 13:40:12 +0000 (13:40 +0000)]
[X86][SSE] Add more thorough extract to store tests
Added v4i32 and v2i64 tests and test on i686 as well as x86_64.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294946
91177308-0d34-0410-b5e6-
96231b3b80d8
James Molloy [Mon, 13 Feb 2017 12:32:47 +0000 (12:32 +0000)]
[ARM] Use VCMP, not VCMPE, for floating point equality comparisons
When generating a floating point comparison we currently unconditionally
generate VCMPE. This has the sideeffect of setting the cumulative Invalid
bit in FPSCR if any of the operands are QNaN.
It is expected that use of a relational predicate on a QNaN value should
raise Invalid. Quoting from the C standard:
The relational and equality operators support the usual mathematical
relationships between numeric values. For any ordered pair of numeric
values exactly one of relationships the less, greater, equal and is true.
Relational operators may raise the floating-point exception when argument
values are NaNs.
The standard doesn't explicitly state the expectation for equality operators,
but the implication and obvious expectation is that equality operators
should not raise Invalid on a QNaN input, as those predicates are wholly
defined on unordered inputs (to return not equal).
Therefore, add a new operand to ARMISD::FPCMP and FPCMPZ indicating if
QNaN should raise Invalid, and pipe that through to TableGen.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294945
91177308-0d34-0410-b5e6-
96231b3b80d8
Simon Pilgrim [Mon, 13 Feb 2017 11:52:58 +0000 (11:52 +0000)]
[X86][SSE] Create matchVectorShuffleWithUNPCK helper function.
Currently only used by target shuffle combining - will use it for lowering as well in a future patch.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294943
91177308-0d34-0410-b5e6-
96231b3b80d8
Pierre Gousseau [Mon, 13 Feb 2017 09:57:17 +0000 (09:57 +0000)]
[X86] Improve readability of test/CodeGen/X86/lzcnt-zext-cmp.ll by adding a common check prefix ALL. NFC.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294938
91177308-0d34-0410-b5e6-
96231b3b80d8
Ayman Musa [Mon, 13 Feb 2017 09:55:48 +0000 (09:55 +0000)]
[X86][AVX512] Fix operand classes for some AVX512 instructions to keep consistency between VEX/EVEX versions of the same instruction.
Differential Revision: https://reviews.llvm.org/D29873
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294937
91177308-0d34-0410-b5e6-
96231b3b80d8
Andrew V. Tischenko [Mon, 13 Feb 2017 09:43:37 +0000 (09:43 +0000)]
Compile time decreasing in the case we're dealing with Machine Combiner.
Before this patch compile time was about 21s (see below). After this patch
we have less than 2s (see bellow).
Intel(R) Xeon(R) CPU E5-2676 v3 @ 2.40GHz
DAGCombiner - trunk
time ./llc spill_fdiv.ll -o /dev/null -enable-unsafe-fp-math
real 0m1.685s
DAGCombiner + Speed patch
time ./llc spill_fdiv.ll -o /dev/null -enable-unsafe-fp-math
real 0m1.655s
MachineCombiner w/o Speed patch
time ./llc spill_fdiv.ll -o /dev/null -enable-unsafe-fp-math
real 0m21.614s
MachineCombiner + Speed patch
time ./llc spill_fdiv.ll -o /dev/null -enable-unsafe-fp-math
real 0m1.593s
The test spill_fdiv.ll is attached to D29627
D29627 should be closed.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294936
91177308-0d34-0410-b5e6-
96231b3b80d8
Alexey Bataev [Mon, 13 Feb 2017 08:01:26 +0000 (08:01 +0000)]
[SLP] Fix for PR31690: Allow using of extra values in horizontal
reductions.
Currently, LLVM supports vectorization of horizontal reduction
instructions with initial value set to 0. Patch supports vectorization
of reduction with non-zero initial values. Also, it supports a
vectorization of instructions with some extra arguments, like:
```
float f(float x[], int a, int b) {
float p = a % b;
p += x[0] + 3;
for (int i = 1; i < 32; i++)
p += x[i];
return p;
}
```
Patch allows vectorization of this kind of horizontal reductions.
Differential Revision: https://reviews.llvm.org/D29727
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294934
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Mon, 13 Feb 2017 04:53:33 +0000 (04:53 +0000)]
[DAGCombiner] Teach DAG combine that inserting an extract_subvector result into the same location of a an undef vector can just use the original input to the extract.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294932
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Mon, 13 Feb 2017 04:53:29 +0000 (04:53 +0000)]
[X86] Genericize the handling of INSERT_SUBVECTOR from an EXTRACT_SUBVECTOR to support 512-bit vectors with 128-bit or 256-bit subvectors.
We now detect that both the extract and insert indices are non-zero and convert to a shuffle. This will be lowered as a blend for 256-bit vectors or as a vshuf operations for 512-bit vectors.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294931
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Sun, 12 Feb 2017 23:49:49 +0000 (23:49 +0000)]
[DAGCombiner] Remove the half vector width check for the combine of EXTRACT_SUBVECTOR from an INSERT_SUBVECTOR.
This gives more parallelism opportunities for AVX-512 when dealing with 128-bit extracts from 512-bit vectors.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294930
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Sun, 12 Feb 2017 23:49:46 +0000 (23:49 +0000)]
[X86] Don't let LowerEXTRACT_SUBVECTOR call getNode for EXTRACT_SUBVECTOR.
This results in the simplifications inside of getNode running while we're legalizing nodes popped off the worklist during the final DAG combine. This basically makes a DAG combine like operation occur during this legalize step, but we don't handle something quite the same way. I think we don't recursively added the removed nodes to the DAG combiner worklist.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294929
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Berlin [Sun, 12 Feb 2017 23:28:06 +0000 (23:28 +0000)]
NewGVN: Update a number of xfailed tests to either be correct or note
why they fail.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294928
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Berlin [Sun, 12 Feb 2017 23:24:47 +0000 (23:24 +0000)]
NewGVN: We really pass TBAA if we enable DCE and fix the test. Note that GVN eliminates no-use readonly/readnone calls, even if they are not marked nounwind. NewGVN only eliminates them if they are marked nounwind, and thus, trivially dead.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294927
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Berlin [Sun, 12 Feb 2017 23:24:45 +0000 (23:24 +0000)]
NewGVN: Reverse order of congruence class elimination to maximize trivial deadness
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294926
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Berlin [Sun, 12 Feb 2017 23:24:42 +0000 (23:24 +0000)]
NewGVN: Use shouldSwapOperands in one more place
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294925
91177308-0d34-0410-b5e6-
96231b3b80d8
Sanjay Patel [Sun, 12 Feb 2017 23:07:52 +0000 (23:07 +0000)]
[TargetLowering] fix SETCC SETLT folding with FP types
The bug was introduced with:
https://reviews.llvm.org/rL294863
...and manifests as a selection failure in x86, but that's actually
another bug. This fix prevents wrong codegen with -0.0, but in the
more common case when we have NSZ and NNAN (-ffast-math), we should
still be able to fold this setcc/compare.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294924
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Berlin [Sun, 12 Feb 2017 22:40:10 +0000 (22:40 +0000)]
Revert accidental commit titled "testing"
This reverts commit r294919
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294923
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Berlin [Sun, 12 Feb 2017 22:25:20 +0000 (22:25 +0000)]
NewGVN: Apply the fast math flags fix in r267113 to NewGVN as well.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294922
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Berlin [Sun, 12 Feb 2017 22:12:20 +0000 (22:12 +0000)]
PredicateInfo: Handle critical edges
Summary:
This adds support for placing predicateinfo such that it affects critical edges.
This fixes the issues mentioned by Nuno on the mailing list.
Depends on D29519
Reviewers: davide, nlopes
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D29606
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294921
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Berlin [Sun, 12 Feb 2017 22:02:47 +0000 (22:02 +0000)]
NewGVN: Fix missed call that should be to shouldSwapOperands
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294920
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Berlin [Sun, 12 Feb 2017 22:02:20 +0000 (22:02 +0000)]
testing
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294919
91177308-0d34-0410-b5e6-
96231b3b80d8
Simon Pilgrim [Sun, 12 Feb 2017 20:53:44 +0000 (20:53 +0000)]
[X86] Fix typo in function name. NFCI.
convertBitVectorToUnsiged - convertBitVectorToUnsigned
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294914
91177308-0d34-0410-b5e6-
96231b3b80d8
Saleem Abdulrasool [Sun, 12 Feb 2017 18:55:33 +0000 (18:55 +0000)]
llvm-readobj: process FreeBSD core notes
core files on FreeBSD have additional notes to capture state. Process
those notes when dumping the notes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294909
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Sun, 12 Feb 2017 18:47:46 +0000 (18:47 +0000)]
[AVX-512] Add various EVEX move instructions to load folding tables using the VEX equivalents as a guide.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294908
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Sun, 12 Feb 2017 18:47:44 +0000 (18:47 +0000)]
[AVX-512] Add VMOV64toSDZrm CodeGenOnly instruction based on the same instruction from AVX/SSE.
I can't prove that we can select this instruction or the AVX/SSE version, but I'm adding it for consistency for now so I can continue matching the load folding tables.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294907
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Sun, 12 Feb 2017 18:47:40 +0000 (18:47 +0000)]
[X86] Fix a couple instruction names to use 'mr' instead of 'rm' to indicate they are stores. AVX-512 version was already named with 'mr'.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294906
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Sun, 12 Feb 2017 18:47:37 +0000 (18:47 +0000)]
[AVX-512] Add VPEXTRD/Q to load folding tables.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294905
91177308-0d34-0410-b5e6-
96231b3b80d8
Simon Pilgrim [Sun, 12 Feb 2017 16:46:41 +0000 (16:46 +0000)]
[X86][SSE] Update argument names to match function name. NFCI.
The target shuffle match function arguments were using the term 'Ops' but the function names referred to them as 'Inputs' - use 'Inputs' consistently.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294900
91177308-0d34-0410-b5e6-
96231b3b80d8
Sanjay Patel [Sun, 12 Feb 2017 16:40:30 +0000 (16:40 +0000)]
[InstCombine] fold icmp sgt/slt (add nsw X, C2), C --> icmp sgt/slt X, (C - C2)
I found one special case of this transform for 'slt 0', so I removed that and added the general transform.
Alive code to check correctness:
Name: slt_no_overflow
Pre: WillNotOverflowSignedSub(C1, C2)
%a = add nsw i8 %x, C2
%b = icmp slt %a, C1
=>
%b = icmp slt %x, C1 - C2
Name: sgt_no_overflow
Pre: WillNotOverflowSignedSub(C1, C2)
%a = add nsw i8 %x, C2
%b = icmp sgt %a, C1
=>
%b = icmp sgt %x, C1 - C2
http://rise4fun.com/Alive/MH
Differential Revision: https://reviews.llvm.org/D29774
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294898
91177308-0d34-0410-b5e6-
96231b3b80d8
Sanjay Patel [Sun, 12 Feb 2017 15:35:34 +0000 (15:35 +0000)]
[ValueTracking] use nonnull argument attribute to eliminate null checks
Enhancing value tracking's analysis of null-ness was suggested in D27855, so here's a first attempt at that.
This is part of solving:
https://llvm.org/bugs/show_bug.cgi?id=28430
Differential Revision: https://reviews.llvm.org/D28204
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294897
91177308-0d34-0410-b5e6-
96231b3b80d8
Simon Pilgrim [Sun, 12 Feb 2017 14:31:23 +0000 (14:31 +0000)]
[X86][AVX2] Add support for combining target shuffles to VPMOVZX
Initial 256-bit vector support - 512-bit support requires extra checks for AVX512BW support (PMOVZXBW) that will be handled in a future patch.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294896
91177308-0d34-0410-b5e6-
96231b3b80d8
NAKAMURA Takumi [Sun, 12 Feb 2017 13:15:31 +0000 (13:15 +0000)]
AMDGPU::expandMemIntrinsicUses(): Fix an uninitialized variable. This function returned true or undef.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294895
91177308-0d34-0410-b5e6-
96231b3b80d8
Dorit Nuzman [Sun, 12 Feb 2017 09:32:53 +0000 (09:32 +0000)]
[LV/LoopAccess] Check statically if an unknown dependence distance can be
proven larger than the loop-count
This fixes PR31098: Try to resolve statically data-dependences whose
compile-time-unknown distance can be proven larger than the loop-count,
instead of resorting to runtime dependence checking (which are not always
possible).
For vectorization it is sufficient to prove that the dependence distance
is >= VF; But in some cases we can prune unknown dependence distances early,
and even before selecting the VF, and without a runtime test, by comparing
the distance against the loop iteration count. Since the vectorized code
will be executed only if LoopCount >= VF, proving distance >= LoopCount
also guarantees that distance >= VF. This check is also equivalent to the
Strong SIV Test.
Reviewers: mkuper, anemet, sanjoy
Differential Revision: https://reviews.llvm.org/D28044
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294892
91177308-0d34-0410-b5e6-
96231b3b80d8
Elena Demikhovsky [Sun, 12 Feb 2017 07:56:50 +0000 (07:56 +0000)]
AVX-512: Fixed DWARF register numbers for XMM16-31
The reference is here:
https://software.intel.com/sites/default/files/article/402129/mpx-linux64-abi.pdf
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294890
91177308-0d34-0410-b5e6-
96231b3b80d8
Davide Italiano [Sun, 12 Feb 2017 05:43:25 +0000 (05:43 +0000)]
[LTO] Remove useless redirection from test. NFCI.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294889
91177308-0d34-0410-b5e6-
96231b3b80d8
Chandler Carruth [Sun, 12 Feb 2017 05:38:04 +0000 (05:38 +0000)]
[PM] Add devirtualization-based iteration utility into the new PM's
default pipeline.
A clang with this patch built with ASan and asserts can build all of the
test-suite as well, so it seems to not uncover any latent problems.
Differential Revision: https://reviews.llvm.org/D29853
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294888
91177308-0d34-0410-b5e6-
96231b3b80d8
Chandler Carruth [Sun, 12 Feb 2017 05:34:04 +0000 (05:34 +0000)]
[PM] Enable GlobalsAA in the new PM's pipeline by default.
All the invalidation issues and bugs in this seem to be fixed, it has
survived a full build of the test suite plus SPEC with asserts and ASan
enabled on the Clang binary used.
Differential Revision: https://reviews.llvm.org/D29815
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294887
91177308-0d34-0410-b5e6-
96231b3b80d8
Davide Italiano [Sun, 12 Feb 2017 05:05:35 +0000 (05:05 +0000)]
[lib/LTO] Add support for hotness optremarks in the new API.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294885
91177308-0d34-0410-b5e6-
96231b3b80d8
Davide Italiano [Sun, 12 Feb 2017 03:47:54 +0000 (03:47 +0000)]
[LTO] Simplify this test quite a bit, @func2 is unused/unneeded.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294884
91177308-0d34-0410-b5e6-
96231b3b80d8
Davide Italiano [Sun, 12 Feb 2017 03:42:09 +0000 (03:42 +0000)]
[llvm-lto2] Fix typo in error message.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294883
91177308-0d34-0410-b5e6-
96231b3b80d8
Davide Italiano [Sun, 12 Feb 2017 03:31:30 +0000 (03:31 +0000)]
[lib/LTO] Initial support for optimization remarks in the new API.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294882
91177308-0d34-0410-b5e6-
96231b3b80d8
NAKAMURA Takumi [Sun, 12 Feb 2017 01:18:32 +0000 (01:18 +0000)]
Kaleidoscope-Ch7: Add TranformUtils for llvm::createPromoteMemoryToRegisterPass() added in r294870.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294881
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Sat, 11 Feb 2017 23:23:11 +0000 (23:23 +0000)]
[X86] Update test case I missed in r294876.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294878
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Sat, 11 Feb 2017 22:57:12 +0000 (22:57 +0000)]
[X86] Move code for using blendi for insert_subvector out to an isel pattern. This gives the DAG combiner more opportunity to optimize without needing to dig through the blend.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294876
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Sat, 11 Feb 2017 22:57:09 +0000 (22:57 +0000)]
[DAGCombiner] Make the combine of INSERT_SUBVECTOR into a CONCAT_VECTOR more generic to support larger concats.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294875
91177308-0d34-0410-b5e6-
96231b3b80d8
Simon Pilgrim [Sat, 11 Feb 2017 22:47:06 +0000 (22:47 +0000)]
[X86][SSE] Use VSEXT/VZEXT constant folding for SIGN_EXTEND_VECTOR_INREG/ZERO_EXTEND_VECTOR_INREG
Preparatory step for PR31712
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294874
91177308-0d34-0410-b5e6-
96231b3b80d8
Simon Pilgrim [Sat, 11 Feb 2017 21:55:24 +0000 (21:55 +0000)]
[X86][SSE] Improve VSEXT/VZEXT constant folding.
Generalize VSEXT/VZEXT constant folding to work with any target constant bits source not just BUILD_VECTOR .
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294873
91177308-0d34-0410-b5e6-
96231b3b80d8
Mehdi Amini [Sat, 11 Feb 2017 21:26:52 +0000 (21:26 +0000)]
Update Kaleidoscope tutorial and improve Windows support
Many quoted code blocks were not in sync with the actual toy.cpp
files. Improve tutorial text slightly in several places.
Added some step descriptions crucial to avoid crashes (like
InitializeNativeTarget* calls).
Solve/workaround problems with Windows (JIT'ed method not found, using
custom and standard library functions from host process).
Patch by: Moritz Kroll <moritz.kroll@gmx.de>
Differential Revision: https://reviews.llvm.org/D29864
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294870
91177308-0d34-0410-b5e6-
96231b3b80d8
Amaury Sechet [Sat, 11 Feb 2017 19:34:11 +0000 (19:34 +0000)]
Fix atomic-minmax-i6432.ll .
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294867
91177308-0d34-0410-b5e6-
96231b3b80d8
Amaury Sechet [Sat, 11 Feb 2017 19:27:15 +0000 (19:27 +0000)]
Regen expected tests result. NFC
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294866
91177308-0d34-0410-b5e6-
96231b3b80d8
Aaron Ballman [Sat, 11 Feb 2017 18:45:24 +0000 (18:45 +0000)]
Correcting several sphinx errors; should fix the LLVM documentation build.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294865
91177308-0d34-0410-b5e6-
96231b3b80d8
Simon Pilgrim [Sat, 11 Feb 2017 18:06:24 +0000 (18:06 +0000)]
[X86][SSE] Add early-out when trying to match blend shuffle. NFCI.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294864
91177308-0d34-0410-b5e6-
96231b3b80d8
Sanjay Patel [Sat, 11 Feb 2017 18:01:55 +0000 (18:01 +0000)]
[TargetLowering] check for sign-bit comparisons in SimplifyDemandedBits
I don't know if anything other than x86 vectors is affected by this change, but this may allow
us to remove target-specific intrinsics for blendv* (vector selects). The simplification arises
from the fact that blendv* instructions only use the sign-bit when deciding which vector element
to choose for the destination vector. The mechanism to fold VSELECT into SHRUNKBLEND nodes already
exists in x86 lowering; this demanded bits change just enables the transform to fire more often.
The original motivation starts with a bug for DSE of masked stores that seems completely unrelated,
but I've explained the likely steps in this series here:
https://llvm.org/bugs/show_bug.cgi?id=11210
Differential Revision: https://reviews.llvm.org/D29687
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294863
91177308-0d34-0410-b5e6-
96231b3b80d8
Amaury Sechet [Sat, 11 Feb 2017 17:48:49 +0000 (17:48 +0000)]
Fix typo in test filename. NFC
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294860
91177308-0d34-0410-b5e6-
96231b3b80d8
Amaury Sechet [Sat, 11 Feb 2017 17:48:48 +0000 (17:48 +0000)]
Fix indentation in X86ISelLowering. NFC
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294859
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Sat, 11 Feb 2017 17:35:28 +0000 (17:35 +0000)]
[AVX-512] Add VPMINS/MINU/MAXS/MAXU instructions to load folding tables.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294858
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Sat, 11 Feb 2017 17:35:25 +0000 (17:35 +0000)]
[X86] Improve alphabetizing of load folding tables. NFC
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294857
91177308-0d34-0410-b5e6-
96231b3b80d8
Simon Pilgrim [Sat, 11 Feb 2017 17:27:21 +0000 (17:27 +0000)]
[X86][SSE] Convert getTargetShuffleMaskIndices to use getTargetConstantBitsFromNode.
Removes duplicate constant extraction code in getTargetShuffleMaskIndices.
getTargetConstantBitsFromNode - adds support for VZEXT_MOVL(SCALAR_TO_VECTOR) and fail if the caller doesn't support undef bits.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294856
91177308-0d34-0410-b5e6-
96231b3b80d8
Simon Pilgrim [Sat, 11 Feb 2017 16:42:07 +0000 (16:42 +0000)]
[X86] Merge repeated getScalarValueSizeInBits calls. NFCI.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294852
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Berlin [Sat, 11 Feb 2017 15:20:15 +0000 (15:20 +0000)]
NewGVN: Reverse sense of this test to make it clearer
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294851
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Berlin [Sat, 11 Feb 2017 15:13:49 +0000 (15:13 +0000)]
NewGVN: Add missing initialization of NumFuncArgs lost due to bad merge.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294850
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Berlin [Sat, 11 Feb 2017 15:07:01 +0000 (15:07 +0000)]
NewGVN: Rank and order commutative operands consistently.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294849
91177308-0d34-0410-b5e6-
96231b3b80d8
Simon Pilgrim [Sat, 11 Feb 2017 14:01:37 +0000 (14:01 +0000)]
[X86][3DNow!] Add tests to ensure PFMAX/PFMIN are not commuted.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294848
91177308-0d34-0410-b5e6-
96231b3b80d8
Simon Pilgrim [Sat, 11 Feb 2017 13:51:14 +0000 (13:51 +0000)]
[X86][3DNow!] Enable PFSUB<->PFSUBR commutation
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294847
91177308-0d34-0410-b5e6-
96231b3b80d8
Simon Pilgrim [Sat, 11 Feb 2017 13:32:55 +0000 (13:32 +0000)]
[X86][3DNow!] Enable commutation for PFADD/PFMUL/PFCMPEQ/PAVGUSB/PMULHRW
All commutations confirmed to give identical results - note PFMAX/PFMIN do not
PFSUB<->PFSUBR should be commutable as well
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294846
91177308-0d34-0410-b5e6-
96231b3b80d8
Simon Pilgrim [Sat, 11 Feb 2017 13:00:32 +0000 (13:00 +0000)]
[X86][3DNow!] Add tests showing missed commutation opportunities.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294845
91177308-0d34-0410-b5e6-
96231b3b80d8
Daniel Berlin [Sat, 11 Feb 2017 12:48:50 +0000 (12:48 +0000)]
NewGVN: Clean up how we handle the INITIAL class so that everything in
it is dead or unreachable, as it should be.
This also makes the leader of INITIAL undef, enabling us to handle
irreducibility properly.
Summary:
This lets us verify, more than we do now, that we didn't screw up
value numbering.
Reviewers: davide
Subscribers: Prazek, llvm-commits
Differential Revision: https://reviews.llvm.org/D29842
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294844
91177308-0d34-0410-b5e6-
96231b3b80d8
Vitaly Buka [Sat, 11 Feb 2017 12:44:03 +0000 (12:44 +0000)]
Fix "left shift of negative value -1" introduced by r294805
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294843
91177308-0d34-0410-b5e6-
96231b3b80d8
Simon Pilgrim [Sat, 11 Feb 2017 12:30:59 +0000 (12:30 +0000)]
[X86][XOP] Regenerate XOP commutation tests.
Added 32-bit tests as well.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294841
91177308-0d34-0410-b5e6-
96231b3b80d8
Simon Pilgrim [Sat, 11 Feb 2017 12:29:56 +0000 (12:29 +0000)]
[X86][SSE] Regenerate float comparison commutation tests.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294840
91177308-0d34-0410-b5e6-
96231b3b80d8
Simon Pilgrim [Sat, 11 Feb 2017 12:23:22 +0000 (12:23 +0000)]
[X86] Regenerate CLMUL commutation tests.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294839
91177308-0d34-0410-b5e6-
96231b3b80d8
Benjamin Kramer [Sat, 11 Feb 2017 11:06:55 +0000 (11:06 +0000)]
Move symbols from the global namespace into (anonymous) namespaces. NFC.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294837
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Sat, 11 Feb 2017 07:01:40 +0000 (07:01 +0000)]
[AVX-512] Add VPINSRB/W/D/Q instructions to load folding tables.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294830
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Sat, 11 Feb 2017 07:01:38 +0000 (07:01 +0000)]
[AVX-512] Fix apparent typo in instruction name VMOVSSDrr_REV->VMOVSDZrr_REV.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294829
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Sat, 11 Feb 2017 06:24:03 +0000 (06:24 +0000)]
[AVX-512] Add VPSADBW instructions to load folding tables.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294827
91177308-0d34-0410-b5e6-
96231b3b80d8
Evgeny Stupachenko [Sat, 11 Feb 2017 05:39:00 +0000 (05:39 +0000)]
The patch fixes r294821
Summary:
Update register match for windows testing
From: Evgeny Stupachenko <evstupac@gmail.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294825
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Sat, 11 Feb 2017 05:32:57 +0000 (05:32 +0000)]
[X86] Don't base domain decisions on VEXTRACTF128/VINSERTF128 if only AVX1 is available.
Seems the execution dependency pass likes to use FP instructions when most of the consuming code is integer if a vextractf128 instruction produced the register. Without AVX2 we don't have the corresponding integer instruction available.
This patch suppresses the domain on these instructions to GenericDomain if AVX2 is not supported so that they are ignored by domain fixing. If AVX2 is supported we'll report the correct domain and allow them to switch between integer and fp.
Overall I think this produces better results in the modified test cases.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294824
91177308-0d34-0410-b5e6-
96231b3b80d8
Peter Collingbourne [Sat, 11 Feb 2017 03:19:22 +0000 (03:19 +0000)]
Address Mehdi's post-commit review comments on r294795.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294822
91177308-0d34-0410-b5e6-
96231b3b80d8
Evgeny Stupachenko [Sat, 11 Feb 2017 02:57:43 +0000 (02:57 +0000)]
Fix PR23384 (under "-lsr-insns-cost" option)
Summary:
The patch adds instructions number generated by a solution
to LSR cost under "-lsr-insns-cost" option.
Reviewers: qcolombet, hfinkel
Differential Revision: http://reviews.llvm.org/D28307
From: Evgeny Stupachenko <evstupac@gmail.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294821
91177308-0d34-0410-b5e6-
96231b3b80d8
Ahmed Bougacha [Sat, 11 Feb 2017 01:53:04 +0000 (01:53 +0000)]
[ARM] Make f16 interleaved accesses expensive.
There are no vldN/vstN f16 variants, even with +fullfp16.
We could use the i16 variants, but, in practice, even with +fullfp16,
the f16 sequence leading to the i16 shuffle usually gets scalarized.
We'd need to improve our support for f16 codegen before getting there.
Teach the cost model to consider f16 interleaved operations as
expensive. Otherwise, we are all but guaranteed to end up with
a large block of scalarized vector code.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294819
91177308-0d34-0410-b5e6-
96231b3b80d8
Ahmed Bougacha [Sat, 11 Feb 2017 01:53:00 +0000 (01:53 +0000)]
[ARM] Don't lower f16 interleaved accesses.
There are no vldN/vstN f16 variants, even with +fullfp16.
We could use the i16 variants, but, in practice, even with +fullfp16,
the f16 sequence leading to the i16 shuffle usually gets scalarized.
We'd need to improve our support for f16 codegen before getting there.
Reject f16 interleaved accesses. If we try to emit the f16 intrinsics,
we'll just end up with a selection failure.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294818
91177308-0d34-0410-b5e6-
96231b3b80d8
Ahmed Bougacha [Sat, 11 Feb 2017 01:52:57 +0000 (01:52 +0000)]
[ARM] Unique some redundant CHECK lines. NFC.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294817
91177308-0d34-0410-b5e6-
96231b3b80d8