OSDN Git Service

android-x86/external-llvm.git
5 years ago[InstCombine] Optimize `atomicrmw <op>, 0` into `load atomic` when possible
Quentin Colombet [Thu, 7 Feb 2019 21:27:23 +0000 (21:27 +0000)]
[InstCombine] Optimize `atomicrmw <op>, 0` into `load atomic` when possible

This commit teaches InstCombine how to replace an atomicrmw operation
into a simple load atomic.
For a given `atomicrmw <op>`, this is possible when:
1. The ordering of that operation is compatible with a load (i.e.,
   anything that doesn't have a release semantic).
2. <op> does not modify the value being stored

Differential Revision: https://reviews.llvm.org/D57854

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353471 91177308-0d34-0410-b5e6-96231b3b80d8

5 years agogn build: Make check-{clang,lld,llvm} pass on FreeBSD.
Peter Collingbourne [Thu, 7 Feb 2019 21:24:30 +0000 (21:24 +0000)]
gn build: Make check-{clang,lld,llvm} pass on FreeBSD.

Mostly achieved by assuming that anything that isn't Win or Mac is ELF,
which seems reasonable enough for now.

Differential Revision: https://reviews.llvm.org/D57870

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353470 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[LV] Remove unnecessary assignment to UserIC.
Florian Hahn [Thu, 7 Feb 2019 21:23:37 +0000 (21:23 +0000)]
[LV] Remove unnecessary assignment to UserIC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353469 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[InstCombine] Fix crashing from (icmp (bitcast ([su]itofp X)), Y)
Sanjay Patel [Thu, 7 Feb 2019 21:12:01 +0000 (21:12 +0000)]
[InstCombine] Fix crashing from (icmp (bitcast ([su]itofp X)), Y)

This fixes a class of bugs introduced by D44367,
which transforms various cases of icmp (bitcast ([su]itofp X)), Y to icmp X, Y.
If the bitcast is between vector types with a different number of elements,
the current code will produce bad IR along the lines of: icmp <N x i32> ..., <M x i32> <...>.

This patch suppresses the transform if the bitcast changes the number of vector elements.

Patch by: @AndrewScheidecker (Andrew Scheidecker)

Differential Revision: https://reviews.llvm.org/D57871

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353467 91177308-0d34-0410-b5e6-96231b3b80d8

5 years agoMove SMTSolver dump() methods out-of-line.
Adrian Prantl [Thu, 7 Feb 2019 21:03:18 +0000 (21:03 +0000)]
Move SMTSolver dump() methods out-of-line.

This broke modularized non-local-submodule-visibility builds because
the function bodies pulled in extra dependencies.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353465 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[CodeGen] Handle vector UADDO, SADDO, USUBO, SSUBO
Nikita Popov [Thu, 7 Feb 2019 21:02:22 +0000 (21:02 +0000)]
[CodeGen] Handle vector UADDO, SADDO, USUBO, SSUBO

This is part of https://bugs.llvm.org/show_bug.cgi?id=40442.

Vector legalization is implemented for the add/sub overflow opcodes.
UMULO/SMULO are also handled as far as legalization is concerned, but
they don't support vector expansion yet (so no tests for them).

The vector result widening implementation is suboptimal, because it
could result in a legalization loop.

Differential Revision: https://reviews.llvm.org/D57639

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353464 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[cmake] Pass LLVM_TEMPORARILY_ALLOW_OLD_TOOLCHAIN to NATIVE configure
Shoaib Meenai [Thu, 7 Feb 2019 20:58:04 +0000 (20:58 +0000)]
[cmake] Pass LLVM_TEMPORARILY_ALLOW_OLD_TOOLCHAIN to NATIVE configure

We should propagate this down to host builds so that e.g. people using
an optimized tablegen can do the sub-configure successfully.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353463 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[InstCombine] refactor folds for (icmp (bitcast X), Y); NFCI
Sanjay Patel [Thu, 7 Feb 2019 20:54:09 +0000 (20:54 +0000)]
[InstCombine] refactor folds for (icmp (bitcast X), Y); NFCI

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353462 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[LV] Prevent interleaving if computeMaxVF returned None.
Florian Hahn [Thu, 7 Feb 2019 20:49:10 +0000 (20:49 +0000)]
[LV] Prevent interleaving if computeMaxVF returned None.

As discussed in D57382, interleaving should be avoided if computeMaxVF
returns None, same as we currently do for vectorization.

Fixes https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=6477

Reviewers: Ayal, dcaballe, hsaito, mkuper, rengolin

Reviewed By: Ayal

Differential Revision: https://reviews.llvm.org/D57837

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353461 91177308-0d34-0410-b5e6-96231b3b80d8

5 years agoGlobalISel: Try to fix bot failures
Matt Arsenault [Thu, 7 Feb 2019 20:44:08 +0000 (20:44 +0000)]
GlobalISel: Try to fix bot failures

Don't rely on order of evaluation of function arguments.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353460 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[DAGCombiner] (add (umax X, C), -C) --> (usubsat X, C) (PR40111)
Simon Pilgrim [Thu, 7 Feb 2019 20:14:43 +0000 (20:14 +0000)]
[DAGCombiner] (add (umax X, C), -C) --> (usubsat X, C) (PR40111)

Move the (add (umax X, C), -C) --> (usubsat X, C) X86 combine into generic DAGCombiner

First of a number of saturated arithmetic folds that can be moved out of X86-specific code for PR40111.

Differential Revision: https://reviews.llvm.org/D57754

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353457 91177308-0d34-0410-b5e6-96231b3b80d8

5 years agoGlobalISel: Implement narrowScalar for shift main type
Matt Arsenault [Thu, 7 Feb 2019 19:37:44 +0000 (19:37 +0000)]
GlobalISel: Implement narrowScalar for shift main type

This is pretty much directly ported from SelectionDAG. Doesn't include
the shift by non-constant but known bits version, since there isn't a
globalisel version of computeKnownBits yet.

This shows a disadvantage of targets not specifically which type
should be used for the shift amount. If type 0 is legalized before
type 1, the operations on the shift amount type use the wider type
(which are also less likely to legalize). This can be avoided by
targets specifying legalization actions on type 1 earlier than for
type 0.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353455 91177308-0d34-0410-b5e6-96231b3b80d8

5 years agoAMDGPU/GlobalISel: Restrict g_implicit_def legality
Matt Arsenault [Thu, 7 Feb 2019 19:10:15 +0000 (19:10 +0000)]
AMDGPU/GlobalISel: Restrict g_implicit_def legality

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353452 91177308-0d34-0410-b5e6-96231b3b80d8

5 years agoGlobalISel: Fix artifact combiner constant legality checks for vectors
Matt Arsenault [Thu, 7 Feb 2019 18:58:28 +0000 (18:58 +0000)]
GlobalISel: Fix artifact combiner constant legality checks for vectors

Since G_CONSTANT is illegal for vectors, this needs to check
what buildConstant will produce for a splat vector.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353449 91177308-0d34-0410-b5e6-96231b3b80d8

5 years agoAMDGPU/GlobalISel: Don't use g_implicit_def in a few tests
Matt Arsenault [Thu, 7 Feb 2019 18:33:22 +0000 (18:33 +0000)]
AMDGPU/GlobalISel: Don't use g_implicit_def in a few tests

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353443 91177308-0d34-0410-b5e6-96231b3b80d8

5 years agoRevert "[DAG] Cleanup of unused node in SimplifySelectCC."
Nirav Dave [Thu, 7 Feb 2019 18:31:05 +0000 (18:31 +0000)]
Revert "[DAG] Cleanup of unused node in SimplifySelectCC."

Causes ASAN use-after-poison errors.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353442 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[InstrProf] Avoid reconstructing Triple, NFC
Reid Kleckner [Thu, 7 Feb 2019 18:16:22 +0000 (18:16 +0000)]
[InstrProf] Avoid reconstructing Triple, NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353439 91177308-0d34-0410-b5e6-96231b3b80d8

5 years agoAMDGPU/GlobalISel: Legalize fsqrt
Matt Arsenault [Thu, 7 Feb 2019 18:14:39 +0000 (18:14 +0000)]
AMDGPU/GlobalISel: Legalize fsqrt

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353438 91177308-0d34-0410-b5e6-96231b3b80d8

5 years agoAMDGPU/GlobalISel: Legalize some f16 operations
Matt Arsenault [Thu, 7 Feb 2019 18:03:11 +0000 (18:03 +0000)]
AMDGPU/GlobalISel: Legalize some f16 operations

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353436 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[HotColdSplit] With PGO add profile entry metadata to split cold function
Teresa Johnson [Thu, 7 Feb 2019 17:50:35 +0000 (17:50 +0000)]
[HotColdSplit] With PGO add profile entry metadata to split cold function

Summary:
When compiling with profile data, ensure the split cold function gets
cold function_entry_count metadata (just use 0 since it should be cold).
Otherwise with function sections it will not be placed in the unlikely
text section with other cold code.

Reviewers: vsk

Subscribers: sebpop, hiraditya, davidxl, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D57900

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353434 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[DAGCombiner] fold add/sub with bool operand based on target's boolean contents
Sanjay Patel [Thu, 7 Feb 2019 17:43:34 +0000 (17:43 +0000)]
[DAGCombiner] fold add/sub with bool operand based on target's boolean contents

I noticed that we are missing this canonicalization in IR:
rL352515
...and then realized that we don't get this right in SDAG either,
so this has to be fixed first regardless of what we choose to do in IR.

The existing fold was limited to scalars and using the wrong predicate
to guard the transform. We have a boolean contents TLI query that can
be used to decide which direction to fold.

This may eventually lead back to the problems/question in:
https://bugs.llvm.org/show_bug.cgi?id=40486
...but it makes no difference to that yet.

Differential Revision: https://reviews.llvm.org/D57401

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353433 91177308-0d34-0410-b5e6-96231b3b80d8

5 years agoGlobalISel: Implement fewerElementsVector for shifts
Matt Arsenault [Thu, 7 Feb 2019 17:38:00 +0000 (17:38 +0000)]
GlobalISel: Implement fewerElementsVector for shifts

Introduce a new function which handles instructions with multiple type
indices, but have the same number of vector elements.

Also legalize v2s16 shifts when applicable.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353432 91177308-0d34-0410-b5e6-96231b3b80d8

5 years agoGlobalISel: Try to make legalize rules more useful for vectors
Matt Arsenault [Thu, 7 Feb 2019 17:25:51 +0000 (17:25 +0000)]
GlobalISel: Try to make legalize rules more useful for vectors

Mostly keep the existing functions on scalars, but add versions which
also operate based on the vector element size.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353430 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[DAG] Cleanup of unused node in SimplifySelectCC.
Nirav Dave [Thu, 7 Feb 2019 17:13:55 +0000 (17:13 +0000)]
[DAG] Cleanup of unused node in SimplifySelectCC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353428 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[x86] split more 256/512-bit shuffles in lowering
Sanjay Patel [Thu, 7 Feb 2019 17:10:49 +0000 (17:10 +0000)]
[x86] split more 256/512-bit shuffles in lowering

This is intentionally a small step because it's hard to know exactly
where we might introduce a conflicting transform with the code that
tries to form wider shuffles. But I think this is safe - if we have
a wide shuffle with 2 operands, then we should do better with an
extract + narrow shuffle.

Differential Revision: https://reviews.llvm.org/D57867

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353427 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[DAG] Cleanup unused node on failed SELECT Combine.
Nirav Dave [Thu, 7 Feb 2019 16:57:50 +0000 (16:57 +0000)]
[DAG] Cleanup unused node on failed SELECT Combine.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353426 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[llvm-ar][libObject] Fix relative paths when nesting thin archives.
Jordan Rupprecht [Thu, 7 Feb 2019 16:41:06 +0000 (16:41 +0000)]
[llvm-ar][libObject] Fix relative paths when nesting thin archives.

Summary:
When adding one thin archive to another, we currently chop off the relative path to the flattened members. For instance, when adding `foo/child.a` (which contains `x.txt`) to `parent.a`, when flattening it we should add it as `foo/x.txt` (which exists) instead of `x.txt` (which does not exist).

As a note, this also undoes the `IsNew` parameter of handling relative paths in r288280. The unit test there still passes.

This was reported as part of testing the kernel build with llvm-ar: https://patchwork.kernel.org/patch/10767545/ (see the second point).

Reviewers: mstorsjo, pcc, ruiu, davide, david2050

Subscribers: hiraditya, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D57842

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353424 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[X86] Simplify casing. NFC.
Nirav Dave [Thu, 7 Feb 2019 15:43:40 +0000 (15:43 +0000)]
[X86] Simplify casing. NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353417 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[DAG] Cleanup unused nodes on failed store-to-load forward combine.
Nirav Dave [Thu, 7 Feb 2019 15:38:14 +0000 (15:38 +0000)]
[DAG] Cleanup unused nodes on failed store-to-load forward combine.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353416 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[CodeView] Fix cycles in debug info when merging Types with global hashes
Alexandre Ganea [Thu, 7 Feb 2019 15:24:18 +0000 (15:24 +0000)]
[CodeView] Fix cycles in debug info when merging Types with global hashes

When type streams with forward references were merged using GHashes, cycles
were introduced in the debug info. This was caused by
GlobalTypeTableBuilder::insertRecordAs() not inserting the record on the second
pass, thus yielding an empty ArrayRef at that record slot. Later on, upon PDB
emission, TpiStreamBuilder::commit() would skip that empty record, thus
offseting all indices that came after in the stream.

This solution comes in two steps:

1. Fix the hash calculation, by doing a multiple-step resolution, iff there are
forward references in the input stream.
2. Fix merge by resolving with multiple passes, therefore moving records with
forward references at the end of the stream.

This patch also adds support for llvm-readoj --codeview-ghash.
Finally, fix dumpCodeViewMergedTypes() which previously could reference deleted
memory.

Fixes PR40221

Differential Revision: https://reviews.llvm.org/D57790

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353412 91177308-0d34-0410-b5e6-96231b3b80d8

5 years agoFix misspelled filenames in file headers
Fangrui Song [Thu, 7 Feb 2019 14:38:25 +0000 (14:38 +0000)]
Fix misspelled filenames in file headers

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353408 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[LSR] Generate cross iteration indexes
Sam Parker [Thu, 7 Feb 2019 13:32:54 +0000 (13:32 +0000)]
[LSR] Generate cross iteration indexes

Modify GenerateConstantOffsetsImpl to create offsets that can be used
by indexed addressing modes. If formulae can be generated which
result in the constant offset being the same size as the recurrence,
we can generate a pre-indexed access. This allows the pointer to be
updated via the single pre-indexed access so that (hopefully) no
add/subs are required to update it for the next iteration. For small
cores, this can significantly improve performance DSP-like loops.

Differential Revision: https://reviews.llvm.org/D55373

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353403 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[ARM GlobalISel] Support G_ICMP for Thumb2
Diana Picus [Thu, 7 Feb 2019 11:05:33 +0000 (11:05 +0000)]
[ARM GlobalISel] Support G_ICMP for Thumb2

Mark as legal and use the t2* equivalents of the arm mode instructions,
e.g. t2CMPrr instead of plain CMPrr.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353392 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[ARM] Reformat isRedundantFlagInstr for D57833. NFC
David Green [Thu, 7 Feb 2019 10:51:04 +0000 (10:51 +0000)]
[ARM] Reformat isRedundantFlagInstr for D57833. NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353386 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[BPF] add code-gen support for JMP32 instructions
Jiong Wang [Thu, 7 Feb 2019 10:43:09 +0000 (10:43 +0000)]
[BPF] add code-gen support for JMP32 instructions

JMP32 instructions has been added to eBPF ISA. They are 32-bit variants of
existing BPF conditional jump instructions, but the comparison happens on
low 32-bit sub-register only, therefore some unnecessary extensions could
be saved.

JMP32 instructions will only be available for -mcpu=v3. Host probe hook has
been updated accordingly.

JMP32 instructions will only be enabled in code-gen when -mattr=+alu32
enabled, meaning compiling the program using sub-register mode.

For JMP32 encoding, it is a new instruction class, and is using the
reserved eBPF class number 0x6.

This patch has been tested by compiling and running kernel bpf selftests
with JMP32 enabled.

Acked-by: Yonghong Song <yhs@fb.com>
Signed-off-by: Jiong Wang <jiong.wang@netronome.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353384 91177308-0d34-0410-b5e6-96231b3b80d8

5 years agoAArch64: implement copy for paired GPR registers.
Tim Northover [Thu, 7 Feb 2019 10:35:34 +0000 (10:35 +0000)]
AArch64: implement copy for paired GPR registers.

When doing 128-bit atomics using CASP we might need to copy a GPRPair to a
different register, but that was unimplemented up to now.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353383 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[BranchFolding] Remove dead code for handling EHPad blocks
Craig Topper [Thu, 7 Feb 2019 06:21:28 +0000 (06:21 +0000)]
[BranchFolding] Remove dead code for handling EHPad blocks

Summary: This code tries to handle the case where IBB is an EHPad, but there's an earlier check that uses PBB->hasEHPadSuccessor(). Where PBB is a predecessor of IBB. The hasEHPadSuccessor function would have visited IBB and seen that it was an EHPad and returned false. This would prevent us from reaching this code with IBB as an EHPad.

Looks like this code was originally added in rL37427 (ancient) and made dead in rL143001.

Reviewers: rnk, void, efriedma

Reviewed By: rnk

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D57358

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353375 91177308-0d34-0410-b5e6-96231b3b80d8

5 years agoBump minimum toolchain version
JF Bastien [Thu, 7 Feb 2019 05:20:00 +0000 (05:20 +0000)]
Bump minimum toolchain version

Summary:
The RFC on moving past C++11 got good traction:
  http://lists.llvm.org/pipermail/llvm-dev/2019-January/129452.html

This patch therefore bumps the toolchain versions according to our policy:
  llvm.org/docs/DeveloperPolicy.html#toolchain

Subscribers: mgorny, jkorous, dexonsmith, llvm-commits, mehdi_amini, jyknight, rsmith, chandlerc, smeenai, hans, reames, lattner, lhames, erichkeane

Differential Revision: https://reviews.llvm.org/D57264

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353374 91177308-0d34-0410-b5e6-96231b3b80d8

5 years agoMove the SMT API to LLVM
Mikhail R. Gadelha [Thu, 7 Feb 2019 03:19:45 +0000 (03:19 +0000)]
Move the SMT API to LLVM

Moved everything SMT-related to LLVM and updated the cmake scripts.

Differential Revision: https://reviews.llvm.org/D54978

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353373 91177308-0d34-0410-b5e6-96231b3b80d8

5 years agogn build: Merge the test part of r353237.
Peter Collingbourne [Thu, 7 Feb 2019 02:40:49 +0000 (02:40 +0000)]
gn build: Merge the test part of r353237.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353369 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[WebAssembly] Update test output after rL353357. NFC.
Sam Clegg [Thu, 7 Feb 2019 02:35:22 +0000 (02:35 +0000)]
[WebAssembly] Update test output after rL353357. NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353368 91177308-0d34-0410-b5e6-96231b3b80d8

5 years agoAdd OpenBSD support to be able to get the thread name
Brad Smith [Thu, 7 Feb 2019 02:06:58 +0000 (02:06 +0000)]
Add OpenBSD support to be able to get the thread name

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353367 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[WebAssembly] Add symbol flag to the binary format llvm.used
Sam Clegg [Thu, 7 Feb 2019 01:24:44 +0000 (01:24 +0000)]
[WebAssembly] Add symbol flag to the binary format llvm.used

Summary:
Rather than add a new attribute
See https://github.com/WebAssembly/tool-conventions/issues/64

Subscribers: dschuff, jgravelle-google, aheejin, sunfish, rupprecht, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D57864

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353360 91177308-0d34-0410-b5e6-96231b3b80d8

5 years agoFix a minor grammar thinko.
Eric Christopher [Thu, 7 Feb 2019 01:22:07 +0000 (01:22 +0000)]
Fix a minor grammar thinko.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353359 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[WebAssembly] Expand symbol flags shown by llvm-objdump --symbols
Sam Clegg [Thu, 7 Feb 2019 01:17:34 +0000 (01:17 +0000)]
[WebAssembly] Expand symbol flags shown by llvm-objdump --symbols

Differential Revision: https://reviews.llvm.org/D57861

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353357 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[cmake] Drop clang-tools-extra from LLVM_ALL_PROJECTS
Shoaib Meenai [Thu, 7 Feb 2019 01:12:56 +0000 (01:12 +0000)]
[cmake] Drop clang-tools-extra from LLVM_ALL_PROJECTS

We iterate over the list and only enable projects from that list that
are present in LLVM_ENABLE_PROJECTS and disable all other projects. Most
users will only specify clang in LLVM_ENABLE_PROJECTS and expect
clang-tools-extra to be implicitly enabled, so remove clang-tools-extra
from LLVM_ALL_PROJECTS so that it doesn't get disabled instead.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353354 91177308-0d34-0410-b5e6-96231b3b80d8

5 years agoRemove reference to non-existent function. NFC.
Sam Clegg [Thu, 7 Feb 2019 00:11:43 +0000 (00:11 +0000)]
Remove reference to non-existent function. NFC.

This comment is old. The code in question was removed in rL203174

Differential Revision: https://reviews.llvm.org/D57856

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353352 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[llvm-ar] Remove leading slash when printing thin archive members
Jordan Rupprecht [Wed, 6 Feb 2019 21:50:45 +0000 (21:50 +0000)]
[llvm-ar] Remove leading slash when printing thin archive members

Reviewers: ruiu

Subscribers: llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D57845

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353347 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[cmake] Add all subprojects to LLVM_ALL_PROJECTS
Shoaib Meenai [Wed, 6 Feb 2019 21:49:47 +0000 (21:49 +0000)]
[cmake] Add all subprojects to LLVM_ALL_PROJECTS

Make LLVM_ALL_PROJECTS reflect all top-level directories in the monorepo
rather than an arbitrary subset. clang-tools-extra is technically
unnecessary since it gets enabled by clang, but having it there for
consistency shouldn't hurt either.

Differential Revision: https://reviews.llvm.org/D57843

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353346 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[PowerPC] Add vector truncate test to prep for D56507 NFC
Roland Froese [Wed, 6 Feb 2019 21:34:44 +0000 (21:34 +0000)]
[PowerPC] Add vector truncate test to prep for D56507 NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353344 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[cmake] Add openmp to LLVM_ALL_PROJECTS
Shoaib Meenai [Wed, 6 Feb 2019 21:08:17 +0000 (21:08 +0000)]
[cmake] Add openmp to LLVM_ALL_PROJECTS

It'll get ignored in LLVM_ENABLE_PROJECTS after r353148 otherwise.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353343 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[libObject][NFC] Include filename in error message
Jordan Rupprecht [Wed, 6 Feb 2019 20:51:04 +0000 (20:51 +0000)]
[libObject][NFC] Include filename in error message

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353341 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[LICM/MSSA] Add promotion to scalars by building an AliasSetTracker with MemorySSA.
Alina Sbirlea [Wed, 6 Feb 2019 20:25:17 +0000 (20:25 +0000)]
[LICM/MSSA] Add promotion to scalars by building an AliasSetTracker with MemorySSA.

Summary:
Experimentally we found that promotion to scalars carries less benefits
than sinking and hoisting in LICM. When using MemorySSA, we build an
AliasSetTracker on demand in order to reuse the current infrastructure.
We only build it if less than AccessCapForMSSAPromotion exist in the
loop, a cap that is by default set to 250. This value ensures there are
no runtime regressions, and there are small compile time gains for
pathological cases. A much lower value (20) was found to yield a single
regression in the llvm-test-suite and much higher benefits for compile
times. Conservatively we set the current cap to a high value, but we will
explore lowering it when MemorySSA is enabled by default.

Reviewers: sanjoy, chandlerc

Subscribers: nemanjai, jlebar, Prazek, george.burgess.iv, jfb, jsji, llvm-commits

Differential Revision: https://reviews.llvm.org/D56625

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353339 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[DAG] Immediately cleanup unused nodes from extend-based combines.
Nirav Dave [Wed, 6 Feb 2019 20:12:03 +0000 (20:12 +0000)]
[DAG] Immediately cleanup unused nodes from extend-based combines.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353338 91177308-0d34-0410-b5e6-96231b3b80d8

5 years agoMove IR flag handling directly into builder calls for cases translated from Instructi...
Michael Berg [Wed, 6 Feb 2019 19:57:06 +0000 (19:57 +0000)]
Move IR flag handling directly into builder calls for cases translated from Instructions in GlobalIsel

Reviewers: aditya_nandakumar, volkan

Reviewed By: aditya_nandakumar

Subscribers: rovka, kristof.beyls, volkan, Petar.Avramovic

Differential Revision: https://reviews.llvm.org/D57630

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353336 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[AliasSetTracker] Pass MustAlias to addPointer more often.
Alina Sbirlea [Wed, 6 Feb 2019 19:55:12 +0000 (19:55 +0000)]
[AliasSetTracker] Pass MustAlias to addPointer more often.

Summary:
Pass the alias info to addPointer when available. Will save an alias()
call for must sets when adding a known Must or May alias.
[Part of a series of cleanup patches]

Reviewers: reames, mkazantsev

Subscribers: sanjoy, jlebar, llvm-commits

Differential Revision: https://reviews.llvm.org/D56613

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353335 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[X86] Change the CPU on the test case for pr40529.ll to really show the bug. NFC
Craig Topper [Wed, 6 Feb 2019 19:50:59 +0000 (19:50 +0000)]
[X86] Change the CPU on the test case for pr40529.ll to really show the bug. NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353334 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[X86][DAG] Avoid creating dangling bitcast.
Nirav Dave [Wed, 6 Feb 2019 19:45:47 +0000 (19:45 +0000)]
[X86][DAG] Avoid creating dangling bitcast.

combineExtractWithShuffle may leave a dangling bitcast which may
prevent further optimization in later passes. Avoid constructing it
unless it is used.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353333 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[x86] add tests for horizontal ops (PR38971, PR33758); NFC
Sanjay Patel [Wed, 6 Feb 2019 19:40:11 +0000 (19:40 +0000)]
[x86] add tests for horizontal ops (PR38971, PR33758); NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353332 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[SystemZ] Improved handling of the @llvm.ctlz intrinsic.
Jonas Paulsson [Wed, 6 Feb 2019 19:23:31 +0000 (19:23 +0000)]
[SystemZ]  Improved handling of the @llvm.ctlz intrinsic.

Since SystemZ supports counting of leading zeros with the FLOGR instruction,
isCheapToSpeculateCtlz() should return true, which it now does.

ISD::CTLZ_ZERO_UNDEF i32 is now handled the same way as ISD::CTLZ is, which
is needed since promotion to i64 is required and CTLZ_ZERO_UNDEF is only
expanded to CTLZ if it is Legal or Custom.

Review: Ulrich Weigand
https://reviews.llvm.org/D57710

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353330 91177308-0d34-0410-b5e6-96231b3b80d8

5 years agobuild: Remove the cmake check for malloc.h.
Peter Collingbourne [Wed, 6 Feb 2019 19:20:47 +0000 (19:20 +0000)]
build: Remove the cmake check for malloc.h.

As far as I can tell, malloc.h is only being used here to provide
a definition of mallinfo (malloc itself is declared in stdlib.h via
cstdlib). We already have a macro for whether mallinfo is available,
so switch to using that instead.

Differential Revision: https://reviews.llvm.org/D57807

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353329 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[SystemZ] Wait with VGBM selection until after DAGCombine2.
Jonas Paulsson [Wed, 6 Feb 2019 18:59:19 +0000 (18:59 +0000)]
[SystemZ]  Wait with VGBM selection until after DAGCombine2.

Don't lower BUILD_VECTORs to BYTE_MASK, but instead expose the BUILD_VECTORs
to the DAGCombiner and select them to VGBM in Select(). This allows the
DAGCombiner to understand the constant vector values.

For floating point, only all-zeros vectors are now generated with VGBM, as it
turned out to be somewhat complicated to handle any arbitrary constants,
while in practice this is very rare and hardly needed.

The SystemZ ISD opcodes z_byte_mask, z_vzero and z_vones have been removed.

Review: Ulrich Weigand
https://reviews.llvm.org/D57152

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353325 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[opt-viewer] Add --filter option to select remarks for displaying.
Florian Hahn [Wed, 6 Feb 2019 18:43:37 +0000 (18:43 +0000)]
[opt-viewer] Add --filter option to select remarks for displaying.

This allows limiting the displayed remarks to the ones with names
matching the filter (regular) expression.

Generating html pages for a larger project with optimization remarks can
result in a huge HTML documents and using --filter allows to focus on a
set of interesting remarks.

Reviewers: hfinkel, anemet, thegameg, serge-sans-paille

Reviewed By: anemet

Differential Revision: https://reviews.llvm.org/D57827

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353322 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[SelectionDAG] Cleanup some code comments. NFC
Bjorn Pettersson [Wed, 6 Feb 2019 17:36:18 +0000 (17:36 +0000)]
[SelectionDAG] Cleanup some code comments. NFC

Don't repeat the function name in some doxygen
comments.

(Just a minor cleanup, while testing to push
from the git monorepo setup.)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353317 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[GlobalISel][NFC] Gardening: Factor out code for simple unary intrinsics
Jessica Paquette [Wed, 6 Feb 2019 17:25:54 +0000 (17:25 +0000)]
[GlobalISel][NFC] Gardening: Factor out code for simple unary intrinsics

There was a lot of repeated code wrt unary math intrinsics in
translateKnownIntrinsic. This factors out the repeated MIRBuilder code into
two functions: translateSimpleUnaryIntrinsic and getSimpleUnaryIntrinsicOpcode.

This simplifies adding simple unary intrinsics, since after this, all you have
to do is add the mapping to SimpleUnaryIntrinsicOpcodes.

Differential Revision: https://reviews.llvm.org/D57774

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353316 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[yaml2obj]Allow number for ELF symbol type
James Henderson [Wed, 6 Feb 2019 17:16:33 +0000 (17:16 +0000)]
[yaml2obj]Allow number for ELF symbol type

yaml2obj previously only recognised standard STT_* names, and didn't
allow arbitrary numbers. This change allows the user to specify a number
for the type instead. It also adds a test to verify the existing
behaviour for obj2yaml for unkown symbol types.

Reviewed by: grimar

Differential Revision: https://reviews.llvm.org/D57822

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353315 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[InstCombine] X | C == C --> (X & ~C) == 0
Sanjay Patel [Wed, 6 Feb 2019 16:43:54 +0000 (16:43 +0000)]
[InstCombine] X | C == C --> (X & ~C) == 0

We should canonicalize to one of these forms,
and compare-with-zero could be more conducive
to follow-on transforms. This also leads to
generally better codegen as shown in PR40611:
https://bugs.llvm.org/show_bug.cgi?id=40611

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353313 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[InstCombine] add tests for PR40611 and regenerate checks; NFC
Sanjay Patel [Wed, 6 Feb 2019 16:17:05 +0000 (16:17 +0000)]
[InstCombine] add tests for PR40611 and regenerate checks; NFC

Lots of unrelated diffs here from the newer version of the script.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353312 91177308-0d34-0410-b5e6-96231b3b80d8

5 years agoAArch64: enforce even/odd register pairs for CASP instructions.
Tim Northover [Wed, 6 Feb 2019 15:26:35 +0000 (15:26 +0000)]
AArch64: enforce even/odd register pairs for CASP instructions.

ARMv8.1a CASP instructions need the first of the pair to be an even register
(otherwise the encoding is unallocated). We enforced this during assembly, but
not CodeGen before.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353308 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[InlineAsm][X86] Add backend support for X86 flag output parameters.
Nirav Dave [Wed, 6 Feb 2019 15:26:29 +0000 (15:26 +0000)]
[InlineAsm][X86] Add backend support for X86 flag output parameters.

Allow custom handling of inline assembly output parameters and add X86
flag parameter support.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353307 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[SelectionDAGBuilder] Refactor Inline Asm output check. NFCI.
Nirav Dave [Wed, 6 Feb 2019 15:12:46 +0000 (15:12 +0000)]
[SelectionDAGBuilder] Refactor Inline Asm output check. NFCI.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353305 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[SystemZ] Do not return INT_MIN from strcmp/memcmp
Ulrich Weigand [Wed, 6 Feb 2019 15:10:13 +0000 (15:10 +0000)]
[SystemZ] Do not return INT_MIN from strcmp/memcmp

The IPM sequence currently generated to compute the strcmp/memcmp
result will return INT_MIN for the "less than zero" case.  While
this is in compliance with the standard, strictly speaking, it
turns out that common applications cannot handle this, e.g. because
they negate a comparison result in order to implement reverse
compares.

This patch changes code to use a different sequence that will result
in -2 for the "less than zero" case (same as GCC).  However, this
requires that the two source operands of the compare instructions
are inverted, which breaks the optimization in removeIPMBasedCompare.
Therefore, I've removed this (and all of optimizeCompareInstr), and
replaced it with a mostly equivalent optimization in combineCCMask
at the DAGcombine level.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353304 91177308-0d34-0410-b5e6-96231b3b80d8

5 years agoAArch64: annotate atomics with dropped acquire semantics when printing.
Tim Northover [Wed, 6 Feb 2019 15:07:59 +0000 (15:07 +0000)]
AArch64: annotate atomics with dropped acquire semantics when printing.

A quirk of the v8.1a spec is that when the writeback regiser for an atomic
read-modify-write instruction is wzr/xzr, the instruction no longer enforces
acquire ordering. However, it's still written with the misleading 'a' mnemonic.

So this adds an annotation when disassembling such instructions, mentioning the
change.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353303 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[x86] vectorize cast ops in lowering to avoid register file transfers
Sanjay Patel [Wed, 6 Feb 2019 14:59:39 +0000 (14:59 +0000)]
[x86] vectorize cast ops in lowering to avoid register file transfers

The proposal in D56796 may cross the line because we're trying to avoid vectorization
transforms in generic DAG combining. So this is an alternate, later, x86-specific
translation of that patch.

There are several potential follow-ups to enhance this:
1. Allow extraction from non-zero element index.
2. Peek through extends of smaller width integers.
3. Support x86-specific conversion opcodes like X86ISD::CVTSI2P

Differential Revision: https://reviews.llvm.org/D56864

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353302 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[MCA] Speedup ResourceManager queries. NFCI
Andrea Di Biagio [Wed, 6 Feb 2019 14:57:28 +0000 (14:57 +0000)]
[MCA] Speedup ResourceManager queries. NFCI

When a resource unit R is released, the ResourceManager notifies groups that
contain R. Before this patch, the logic in method ResourceManager::release()
implemented a potentially slow iterative search of dependent groups on the
entire set of processor resources.
This patch replaces that logic with a simpler (and often faster) lookup on array
`Resource2Groups`.  This patch gives an average speedup of ~3-4% (observed on a
release build when testing for target btver2).
No functional change intended.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353301 91177308-0d34-0410-b5e6-96231b3b80d8

5 years agogn build: Merge r353265, r353237
Nico Weber [Wed, 6 Feb 2019 13:53:47 +0000 (13:53 +0000)]
gn build: Merge r353265, r353237

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353298 91177308-0d34-0410-b5e6-96231b3b80d8

5 years agoAttempt to fix buildbot after r353289
Eugene Leviant [Wed, 6 Feb 2019 13:45:22 +0000 (13:45 +0000)]
Attempt to fix buildbot after r353289

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353294 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[DAGCombine][NFC] GatherAllAliases should take a LSBaseSDNode.
Clement Courbet [Wed, 6 Feb 2019 12:36:17 +0000 (12:36 +0000)]
[DAGCombine][NFC] GatherAllAliases should take a LSBaseSDNode.

GatherAllAliases only makes sense for LSBaseSDNode. Enforce it with
static typing instead of runtime cast.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353291 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[NFC] Simplify check in guard widening
Max Kazantsev [Wed, 6 Feb 2019 11:27:00 +0000 (11:27 +0000)]
[NFC] Simplify check in guard widening

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353290 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[llvm-objcopy] Allow regular expressions in name comparison
Eugene Leviant [Wed, 6 Feb 2019 11:00:07 +0000 (11:00 +0000)]
[llvm-objcopy] Allow regular expressions in name comparison

Differential revision: https://reviews.llvm.org/D57517

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353289 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[DebugInfo]Print correct value for special opcode address increment
James Henderson [Wed, 6 Feb 2019 10:31:50 +0000 (10:31 +0000)]
[DebugInfo]Print correct value for special opcode address increment

The wrong variable was being used when printing the address increment in
verbose output of .debug_line. This patch fixes this.

Reviewed by: JDevlieghere

Differential Revision: https://reviews.llvm.org/D57693

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353288 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[DebugInfo][llvm-symbolizer]Add some tests for edge cases when symbolizing
James Henderson [Wed, 6 Feb 2019 10:13:14 +0000 (10:13 +0000)]
[DebugInfo][llvm-symbolizer]Add some tests for edge cases when symbolizing

This patch adds half a dozen new tests that test various edge cases in
the behaviour of the symbolizer and DWARF data parsing. All of them test
the current behaviour.

Reviewed by: JDevlieghere, aprantl

Differential Revision: https://reviews.llvm.org/D57741

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353286 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[yaml::BinaryRef] Slight perf tuning (for llvm-exegesis analysis mode)
Roman Lebedev [Wed, 6 Feb 2019 08:57:01 +0000 (08:57 +0000)]
[yaml::BinaryRef] Slight perf tuning (for llvm-exegesis analysis mode)

Summary:
llvm-exegesis uses this functionality to read it's benchmark dumps.
This reading of `.yaml`s takes ~60% of runtime for 14656 benchmark points (i.e. one sweep over all x86 instructions),
but only 30% of time for 3x as much benchmark points.

In particular, this `BinaryRef` appears to be an obvious pain point.
Without patch:
```
$ perf stat -r 25 ./bin/llvm-exegesis -mode=analysis -analysis-epsilon=1.0 -benchmarks-file=/tmp/benchmarks-inverse_throughput-onefull.yaml -analysis-clusters-output-file="" -analysis-inconsistencies-output-file=/tmp/clusters-orig.html
no exegesis target for x86_64-unknown-linux-gnu, using default
Parsed 14656 benchmark points
Printing sched class consistency analysis results to file '/tmp/clusters-orig.html'
...
no exegesis target for x86_64-unknown-linux-gnu, using default
Parsed 14656 benchmark points
Printing sched class consistency analysis results to file '/tmp/clusters-orig.html'

 Performance counter stats for './bin/llvm-exegesis -mode=analysis -analysis-epsilon=1.0 -benchmarks-file=/tmp/benchmarks-inverse_throughput-onefull.yaml -analysis-clusters-output-file= -analysis-inconsistencies-output-file=/tmp/clusters-orig.html' (25 runs):

            972.86 msec task-clock                #    0.994 CPUs utilized            ( +-  0.25% )
                30      context-switches          #   30.774 M/sec                    ( +- 21.74% )
                 0      cpu-migrations            #    0.370 M/sec                    ( +- 67.81% )
             11873      page-faults               # 12211.512 M/sec                   ( +-  0.00% )
        3898373408      cycles                    # 4009682.186 GHz                   ( +-  0.25% )  (83.12%)
         360399748      stalled-cycles-frontend   #    9.24% frontend cycles idle     ( +-  0.54% )  (83.24%)
        1099450483      stalled-cycles-backend    #   28.20% backend cycles idle      ( +-  0.59% )  (33.63%)
        4910528820      instructions              #    1.26  insn per cycle
                                                  #    0.22  stalled cycles per insn  ( +-  0.13% )  (50.21%)
        1111976775      branches                  # 1143726625.854 M/sec              ( +-  0.10% )  (66.77%)
          23248474      branch-misses             #    2.09% of all branches          ( +-  0.19% )  (83.29%)

           0.97850 +- 0.00647 seconds time elapsed  ( +-  0.66% )
```
With the patch:
```
$ perf stat -r 25 ./bin/llvm-exegesis -mode=analysis -analysis-epsilon=1.0 -benchmarks-file=/tmp/benchmarks-inverse_throughput-onefull.yaml -analysis-clusters-output-file="" -analysis-inconsistencies-output-file=/tmp/clusters-new.html
no exegesis target for x86_64-unknown-linux-gnu, using default
Parsed 14656 benchmark points
Printing sched class consistency analysis results to file '/tmp/clusters-new.html'
...
no exegesis target for x86_64-unknown-linux-gnu, using default
Parsed 14656 benchmark points
Printing sched class consistency analysis results to file '/tmp/clusters-new.html'

 Performance counter stats for './bin/llvm-exegesis -mode=analysis -analysis-epsilon=1.0 -benchmarks-file=/tmp/benchmarks-inverse_throughput-onefull.yaml -analysis-clusters-output-file= -analysis-inconsistencies-output-file=/tmp/clusters-new.html' (25 runs):

            905.29 msec task-clock                #    0.999 CPUs utilized            ( +-  0.11% )
                15      context-switches          #   16.533 M/sec                    ( +- 32.27% )
                 0      cpu-migrations            #    0.000 K/sec
             11873      page-faults               # 13121.789 M/sec                   ( +-  0.00% )
        3627759720      cycles                    # 4009283.100 GHz                   ( +-  0.11% )  (83.19%)
         370401480      stalled-cycles-frontend   #   10.21% frontend cycles idle     ( +-  0.22% )  (83.19%)
        1007114438      stalled-cycles-backend    #   27.76% backend cycles idle      ( +-  0.34% )  (33.62%)
        4414014304      instructions              #    1.22  insn per cycle
                                                  #    0.23  stalled cycles per insn  ( +-  0.08% )  (50.36%)
        1003751700      branches                  # 1109314021.971 M/sec              ( +-  0.07% )  (66.97%)
          24611010      branch-misses             #    2.45% of all branches          ( +-  0.10% )  (83.41%)

           0.90593 +- 0.00105 seconds time elapsed  ( +-  0.12% )
```
So this decreases the overall run time of llvm-exegesis analysis mode (on one sweep) by roughly -7%.

To be noted, `BinaryRef::writeAsBinary()` change is the reason for the perf changes,
usage of `llvm::isHexDigit()` instead of `isxdigit()` does not appear to have any perf impact,
i have only changed it "for symmetry".

`writeAsBinary()` change is correct, it produces identical de-hex-ified buffer, and the final output is thus identical:
```
$ sha512sum /tmp/clusters-*
db4bbd904fe8840853b589b032c5041bc060b91bcd9c27b914b56581fbc473550eea74b852238c79963b5adf2419f379e9f5db76784048b48e3937f9f3e732bf  /tmp/clusters-new.html
db4bbd904fe8840853b589b032c5041bc060b91bcd9c27b914b56581fbc473550eea74b852238c79963b5adf2419f379e9f5db76784048b48e3937f9f3e732bf  /tmp/clusters-orig.html
```

Reviewers: silvas, espindola, sbc100, zturner, courbet, gchatelet

Reviewed By: gchatelet

Subscribers: tschuett, RKSimon, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D57699

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353282 91177308-0d34-0410-b5e6-96231b3b80d8

5 years agoFix misspelled filenames in file headers of llvm/{MC,Object,CodeGen}/*.h
Fangrui Song [Wed, 6 Feb 2019 08:02:46 +0000 (08:02 +0000)]
Fix misspelled filenames in file headers of llvm/{MC,Object,CodeGen}/*.h

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353278 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[NFC] Factor out detatchment of dead blocks from their erasing
Max Kazantsev [Wed, 6 Feb 2019 07:56:36 +0000 (07:56 +0000)]
[NFC] Factor out detatchment of dead blocks from their erasing

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353277 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[LoopSimplifyCFG] Do not count dead exit blocks twice, make CFG simpler
Max Kazantsev [Wed, 6 Feb 2019 07:49:17 +0000 (07:49 +0000)]
[LoopSimplifyCFG] Do not count dead exit blocks twice, make CFG simpler

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353276 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[NFC] Revert rL353274
Max Kazantsev [Wed, 6 Feb 2019 06:33:02 +0000 (06:33 +0000)]
[NFC] Revert rL353274

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353275 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[NFC] Extend API of DeleteDeadBlock(s) to collect updates without DTU
Max Kazantsev [Wed, 6 Feb 2019 06:00:02 +0000 (06:00 +0000)]
[NFC] Extend API of DeleteDeadBlock(s) to collect updates without DTU

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353274 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[NFC] Replace readonly SmallVectorImpl with ArrayRef
Max Kazantsev [Wed, 6 Feb 2019 05:40:31 +0000 (05:40 +0000)]
[NFC] Replace readonly SmallVectorImpl with ArrayRef

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353273 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[HotColdSplit] Move splitting after instrumented PGO use
Teresa Johnson [Wed, 6 Feb 2019 04:29:39 +0000 (04:29 +0000)]
[HotColdSplit] Move splitting after instrumented PGO use

Summary:
Follow up to D57082 which moved splitting earlier in the pipeline, in
order to perform it before inlining. However, it was moved too early,
before the IR is annotated with instrumented PGO data. This caused the
splitting to incorrectly determine cold functions.

Move it to just after PGO annotation (still before inlining), in both
pass managers.

Reviewers: vsk, hiraditya, sebpop

Subscribers: mehdi_amini, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D57805

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353270 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[CMake] Unify scripts for generating VCS headers
Petr Hosek [Wed, 6 Feb 2019 03:51:00 +0000 (03:51 +0000)]
[CMake] Unify scripts for generating VCS headers

Previously, there were two different scripts for generating VCS headers:
one used by LLVM and one used by Clang and lldb. They were both similar,
but different. They were both broken in their own ways, for example the
one used by Clang didn't properly handle monorepo resulting in an
incorrect version information reported by Clang.

This change unifies two the scripts by introducing a new script that's
used from both LLVM, Clang and lldb, ensures that the new script
supports both monorepo and standalone SVN and Git setups, and removes
the old scripts.

Differential Revision: https://reviews.llvm.org/D57063

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353268 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[AliasSetTracker] Minor style tweak to avoid a variable w/two distinct live ranges...
Philip Reames [Wed, 6 Feb 2019 03:46:40 +0000 (03:46 +0000)]
[AliasSetTracker] Minor style tweak to avoid a variable w/two distinct live ranges [NFC]

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353267 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[Test] Add codegen tests for unordered and monotonic integer operations
Philip Reames [Wed, 6 Feb 2019 03:19:04 +0000 (03:19 +0000)]
[Test] Add codegen tests for unordered and monotonic integer operations

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353266 91177308-0d34-0410-b5e6-96231b3b80d8

5 years agoMove DomTreeUpdater from IR to Analysis
Richard Trieu [Wed, 6 Feb 2019 02:52:52 +0000 (02:52 +0000)]
Move DomTreeUpdater from IR to Analysis

DomTreeUpdater depends on headers from Analysis, but is in IR.  This is a
layering violation since Analysis depends on IR.  Relocate this code from IR
to Analysis to fix the layering violation.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353265 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[x86] add tests for extract+sitofp; NFC
Sanjay Patel [Wed, 6 Feb 2019 00:19:56 +0000 (00:19 +0000)]
[x86] add tests for extract+sitofp; NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353249 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[WebAssembly] Tidy up `let` statements in .td files (NFC)
Heejin Ahn [Wed, 6 Feb 2019 00:17:03 +0000 (00:17 +0000)]
[WebAssembly] Tidy up `let` statements in .td files (NFC)

Summary:
- Delete {} for one-line `let` statements
- Don't indent within `let` blocks
- Add comments after `let` block's closing braces

Reviewers: tlively

Subscribers: dschuff, sbc100, jgravelle-google, sunfish, jfb, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D57730

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353248 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[BasicAA] Cache nonEscapingLocalObjects for alias() calls.
Alina Sbirlea [Tue, 5 Feb 2019 23:52:08 +0000 (23:52 +0000)]
[BasicAA] Cache nonEscapingLocalObjects for alias() calls.

Summary:
Use a small cache for Values tested by nonEscapingLocalObject().
Since the calls to PointerMayBeCaptured are fairly expensive, this saves
a good amount of compile time for anything relying heavily on
BasicAA.alias() calls.

This uses the same approach as the AliasCache, i.e. the cache is reset
after each alias() call. The cache is not used or updated by modRefInfo
calls since it's harder to know when to reset the cache.

Testcases that show improvements with this patch are too large to
include. Example compile time improvement: 7s to 6s.

Reviewers: chandlerc, sunfish

Subscribers: sanjoy, jlebar, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D57627

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353245 91177308-0d34-0410-b5e6-96231b3b80d8

5 years agogn build: Fix clang-tidy build
Nico Weber [Tue, 5 Feb 2019 23:48:13 +0000 (23:48 +0000)]
gn build: Fix clang-tidy build

Not depending on //clang/lib/StaticAnalyzer/Core and
//clang/lib/StaticAnalyzer/Frontend causes a linker error even if
ClangSACheckers are not supported.

Undefined symbols for architecture x86_64:
  "clang::ento::CreateAnalysisConsumer(clang::CompilerInstance&)", referenced from:
      clang::tidy::ClangTidyASTConsumerFactory::CreateASTConsumer(
              clang::CompilerInstance&, llvm::StringRef)
          in libclangTidy.a(libclangTidy.ClangTidy.o)

Patch from Mirko Bonadei <mbonadei@webrtc.org>!

Differential Revision: https://reviews.llvm.org/D57777

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353244 91177308-0d34-0410-b5e6-96231b3b80d8

5 years ago[HotColdSplit] Do not split out `resume` instructions
Vedant Kumar [Tue, 5 Feb 2019 23:39:02 +0000 (23:39 +0000)]
[HotColdSplit] Do not split out `resume` instructions

Resumes that are not reachable from a cleanup landing pad are considered
to be unreachable. It’s not safe to split them out.

rdar://47808235

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353242 91177308-0d34-0410-b5e6-96231b3b80d8

5 years agoOrc: Simplify RPC naming system by using function-local statics
David Blaikie [Tue, 5 Feb 2019 23:38:55 +0000 (23:38 +0000)]
Orc: Simplify RPC naming system by using function-local statics

The existing scheme of class template static members for Name and
NameMutex is a bit verbose, involves global ctors (even if they're cheap
for string and mutex, still not entirely free), and (importantly/my
immediate motivation here) trips over a bug in LLVM's modules
implementation that's a bit involved (hmm, sounds like Mr. Smith has a
fix for the modules thing - but I'm still inclined to commit this patch
as general goodness).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353241 91177308-0d34-0410-b5e6-96231b3b80d8