]> granicus.if.org Git - llvm/log
llvm
6 years agoRevert r351833 and r352250.
Hans Wennborg [Tue, 29 Jan 2019 13:43:22 +0000 (13:43 +0000)]
Revert r351833 and r352250.

They were breaking the Windows build when using MSBuild, see the
discussion on D56781.

r351833: "Use response file when generating LLVM-C.dll"

> Use response file when generating LLVM-C.dll
>
> As discovered in D56774 the command line gets to long, so use a response file to give the script the libs. This change has been tested and is confirmed working for me.
>
> Commited on behalf of Jakob Bornecrantz
>
> Differential Revision: https://reviews.llvm.org/D56781

r352250: "Build LLVM-C.dll by default on windows and enable in release package"

>  Build LLVM-C.dll by default on windows and enable in release package
>
>  With the fixes to the building of LLVM-C.dll in D56781 this should now
>  be safe to land. This will greatly simplify dealing with LLVM for people
>  that just want to use the C API on windows. This is a follow up from
>  D35077.
>
>  Patch by Jakob Bornecrantz!
>
>  Differential revision: https://reviews.llvm.org/D56774

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352492 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[CodeGen] Omit range checks from jump tables when lowering switches with unreachable...
Ayonam Ray [Tue, 29 Jan 2019 12:01:32 +0000 (12:01 +0000)]
[CodeGen] Omit range checks from jump tables when lowering switches with unreachable default

During the lowering of a switch that would result in the generation of a
jump table, a range check is performed before indexing into the jump
table, for the switch value being outside the jump table range and a
conditional branch is inserted to jump to the default block. In case the
default block is unreachable, this conditional jump can be omitted. This
patch implements omitting this conditional branch for unreachable
defaults.

Review ID: D52002
Reviewers: Hans Wennborg, Eli Freidman, Roman Lebedev

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352484 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[X86] Add PR40483 test case
Simon Pilgrim [Tue, 29 Jan 2019 10:58:42 +0000 (10:58 +0000)]
[X86] Add PR40483 test case

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352480 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[WebAssembly] Re-enable main-function signature rewriting
Dan Gohman [Tue, 29 Jan 2019 10:53:42 +0000 (10:53 +0000)]
[WebAssembly] Re-enable main-function signature rewriting

Re-enable the code to rewrite main-function signatures into
"int main(int argc, char *argv[])", but limited to only handling
the case of "int main(void)", so that it doesn't silently strip
an argument in the "int main(int argc, char *argv[], char *envp[])"
case.

This allows main to be called by C startup code, since WebAssembly
requires caller and callee signatures to match, so it can't rely
on passing main a different number of arguments than it expects.

Differential Revision: https://reviews.llvm.org/D57323

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352479 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[llvm-symbolizer][doc] Tweak wording of --adjust-vma switch description
James Henderson [Tue, 29 Jan 2019 10:43:48 +0000 (10:43 +0000)]
[llvm-symbolizer][doc] Tweak wording of --adjust-vma switch description

The address isn't dynamically relocated. The object is.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352477 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[X86] Fix linux32 pic tests to use correct relocation model (PR39684)
Simon Pilgrim [Tue, 29 Jan 2019 10:41:48 +0000 (10:41 +0000)]
[X86] Fix linux32 pic tests to use correct relocation model (PR39684)

Differential Revision: https://reviews.llvm.org/D57301

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352476 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[ARM] Use sub for negative offset load/store in thumb1
David Green [Tue, 29 Jan 2019 10:40:31 +0000 (10:40 +0000)]
[ARM] Use sub for negative offset load/store in thumb1

This attempts to optimise negative values used in load/store operands
a little. We currently try to selct them as rr, materialising the
negative constant using a MOV/MVN pair. This instead selects ri with
an immediate of 0, forcing the add node to become a simpler sub.

Differential Revision: https://reviews.llvm.org/D57121

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352475 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[X86] Regenerate abi-isel.ll test
Simon Pilgrim [Tue, 29 Jan 2019 10:39:02 +0000 (10:39 +0000)]
[X86] Regenerate abi-isel.ll test

Adds note requested in D57301 and fixes some missing GOTPCREL addressmath checks

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352474 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[ARM] Add extra testcases for D57121. NFC
David Green [Tue, 29 Jan 2019 10:25:56 +0000 (10:25 +0000)]
[ARM] Add extra testcases for D57121. NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352472 91177308-0d34-0410-b5e6-96231b3b80d8

6 years agoRemove 'XFAIL: powerpc64' from a debuginfo test
Jeremy Morse [Tue, 29 Jan 2019 10:23:43 +0000 (10:23 +0000)]
Remove 'XFAIL: powerpc64' from a debuginfo test

This test started XPASSing with r352467, and the change in behaviour
performed by that patch does appear to fix the cause of the original XFAIL
(missing FrameIndex DBG_VALUE), which I've replicated locally with
-mtriple=powerpc64--.

I'll write this up in PR21881 which documents the XFAIL, and seek
confirmation I haven't overlooked something here.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352471 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[IPCP] Don't crash due to arg count/type mismatch between caller/callee
Bjorn Pettersson [Tue, 29 Jan 2019 10:19:44 +0000 (10:19 +0000)]
[IPCP] Don't crash due to arg count/type mismatch between caller/callee

Summary:
This patch avoids an assert in IPConstantPropagation when
there is a argument count/type mismatch between the caller and
the callee.

While this is actually UB on C-level (clang emits a warning),
the IR verifier seems to accept it. I'm not sure what other
frontends/languages might think about this, so simply bailing out
to avoid hitting an assert (in CallSiteBase<>::getArgOperand or
Value::doRAUW) seems like a simple solution.

The problem is exposed by the fact that AbstractCallSites will look
through a bitcast at the callee position of a call/invoke.

Reviewers: jdoerfert, reames, efriedma

Reviewed By: jdoerfert, efriedma

Subscribers: eli.friedman, efriedma, llvm-commits

Differential Revision: https://reviews.llvm.org/D57052

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352469 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[DebugInfo][DAG] Process FrameIndex dbg.values unconditionally
Jeremy Morse [Tue, 29 Jan 2019 09:40:05 +0000 (09:40 +0000)]
[DebugInfo][DAG] Process FrameIndex dbg.values unconditionally

A FrameIndex should be valid throughout a block regardless of what instructions
get selected in that block -- therefore we shouldn't harness dbg.values that
refer to FrameIndexes to an SDNode. There are numerous codegen reasons why
an SDNode never appears or doesn't become a location that a DBG_VALUE can
refer to. None of them actually affect the variable location.

Therefore, before any other tests to encode dbg_values in a SelectionDAG,
identify FrameIndex operands and encode them unattached to any SDNode.

Differential Revision: https://reviews.llvm.org/D57328

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352467 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[NFC] Use ArrayRef instead of SmallVectorImpl where possible
Max Kazantsev [Tue, 29 Jan 2019 09:39:15 +0000 (09:39 +0000)]
[NFC] Use ArrayRef instead of SmallVectorImpl where possible

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352466 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[COFF, ARM64] Don't put jump table into a separate COFF section for EK_LabelDifference32
Martin Storsjo [Tue, 29 Jan 2019 09:36:48 +0000 (09:36 +0000)]
[COFF, ARM64] Don't put jump table into a separate COFF section for EK_LabelDifference32

Windows ARM64 has PIC relocation model and uses jump table kind
EK_LabelDifference32. This produces jump table entry as
".word LBB123 - LJTI1_2" which represents the distance between the block
and jump table.

A new relocation type (IMAGE_REL_ARM64_REL32) is needed to do the fixup
correctly if they are in different COFF section.

This change saves the jump table to the same COFF section as the
associated code. An ideal fix could be utilizing IMAGE_REL_ARM64_REL32
relocation type.

Patch by Tom Tan!

Differential Revision: https://reviews.llvm.org/D57277

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352465 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[CodeGenPrepare] Handle all debug calls in dupRetToEnableTailCallOpts()
Jonas Paulsson [Tue, 29 Jan 2019 09:03:35 +0000 (09:03 +0000)]
[CodeGenPrepare]  Handle all debug calls in dupRetToEnableTailCallOpts()

This patch makes sure that a debug value that is after the bitcast in
dupRetToEnableTailCallOpts() is also skipped.

The reduced test case is from SPEC-2006 on SystemZ.

Review: Vedant Kumar, Wolfgang Pieb
https://reviews.llvm.org/D57050

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352462 91177308-0d34-0410-b5e6-96231b3b80d8

6 years agoFix an incorrectly configured test.
Jeremy Morse [Tue, 29 Jan 2019 08:41:44 +0000 (08:41 +0000)]
Fix an incorrectly configured test.

This should have had a target triple in it, my mistake.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352460 91177308-0d34-0410-b5e6-96231b3b80d8

6 years agoFix compiler warning when using clang 3.6.0
Mikael Holmen [Tue, 29 Jan 2019 06:51:28 +0000 (06:51 +0000)]
Fix compiler warning when using clang 3.6.0

Without the fix we get the following (with -Werror):

../lib/Target/X86/X86ISelLowering.cpp:14181:58: error: suggest braces around initialization of subobject [-Werror,-Wmissing-braces]
  SmallVector<std::array<int, 2>, 2> LaneSrcs(NumLanes, {-1, -1});
                                                         ^~~~~~
                                                         {     }
1 error generated.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352455 91177308-0d34-0410-b5e6-96231b3b80d8

6 years agoCorrect contents for r352453
Philip Reames [Tue, 29 Jan 2019 06:40:02 +0000 (06:40 +0000)]
Correct contents for r352453

I had a local change I hadn't realized when submitting that auto-update.  As such, the auto-update was wrong.  This should fix it, and with that, it's clearly time to stop submitting changes and go to bed.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352454 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[Tests] Regen to remove future test diffs
Philip Reames [Tue, 29 Jan 2019 06:34:46 +0000 (06:34 +0000)]
[Tests] Regen to remove future test diffs

This file appears to have been manually editted at some point after being auto-updated. A future change adjusts this file slightly, and all of the updates makes the diff super confusing.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352453 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[Test] Add tests for gather/maked.load demanded elements, and convert the whole file...
Philip Reames [Tue, 29 Jan 2019 05:58:32 +0000 (05:58 +0000)]
[Test] Add tests for gather/maked.load demanded elements, and convert the whole file to auto generated checks.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352452 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[SCEV] Take correct loop in AddRec simplification. PR40420
Max Kazantsev [Tue, 29 Jan 2019 05:37:59 +0000 (05:37 +0000)]
[SCEV] Take correct loop in AddRec simplification. PR40420

The code of AddRec simplification is using wrong loop when it creates a new
AddRecExpr. It should be using AddRecLoop which we have saved and against which
all gate checks are made, and not calling AddRec->getLoop() over and over
again because AddRec may change and become an AddRecurrency from outer loop
during the transform iterations.

Considering this change trivial, commiting for postcommit review.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352451 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[NFC] Merge failing test from PR40420
Max Kazantsev [Tue, 29 Jan 2019 05:12:40 +0000 (05:12 +0000)]
[NFC] Merge failing test from PR40420

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352450 91177308-0d34-0410-b5e6-96231b3b80d8

6 years agoTry to make new test more resilient to different orderings
Teresa Johnson [Tue, 29 Jan 2019 02:04:01 +0000 (02:04 +0000)]
Try to make new test more resilient to different orderings

New test added in r352441 getting a bot failure which I believe is
due to different ordering in the dumping which isn't being handled
well. Try to make test more resilient to ordering differences.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352446 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[WebAssembly] Handle more types of uses in WebAssemblyAddMissingPrototypes
Sam Clegg [Tue, 29 Jan 2019 00:30:46 +0000 (00:30 +0000)]
[WebAssembly] Handle more types of uses in WebAssemblyAddMissingPrototypes

Previously we were only handling bitcast operations, however
prototypeless functions can also appear in other places such as
comparisons and as function params.

Switch to using replaceAllUsesWith() to replace the prototype-less
function uses.  This new approach results in some redundant bitcasting
but is much simpler and handles all cases.

Differential Revision: https://reviews.llvm.org/D56938

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352445 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[PPC] Include tablegenerated PPCGenCallingConv.inc once
Reid Kleckner [Tue, 29 Jan 2019 00:30:35 +0000 (00:30 +0000)]
[PPC] Include tablegenerated PPCGenCallingConv.inc once

Move the CC analysis implementation to its own .cpp file instead of
duplicating it and artificually using functions in PPCISelLowering.cpp
and PPCFastISel.cpp. Follow-up to the same change done for X86, ARM, and
AArch64.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352444 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[WebAssembly] Expand BUILD_PAIR nodes
Thomas Lively [Mon, 28 Jan 2019 23:44:31 +0000 (23:44 +0000)]
[WebAssembly] Expand BUILD_PAIR nodes

Reviewers: aheejin

Subscribers: dschuff, sbc100, jgravelle-google, hiraditya, sunfish

Differential Revision: https://reviews.llvm.org/D57276

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352442 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[ThinLTO] Add option to dump per-module summary dot graph
Teresa Johnson [Mon, 28 Jan 2019 23:43:26 +0000 (23:43 +0000)]
[ThinLTO] Add option to dump per-module summary dot graph

Summary:
I found that there currently isn't a way to invoke exportToDot from
the command line for a per-module summary index, and therefore no
testing of that case. Add an internal option and use it to test dumping
of per module summary indexes.

In particular, I am looking at fixing the limitation that causes the
aliasee GUID in the per-module summary to be 0, and want to be able to
test that change.

Reviewers: evgeny777

Subscribers: mehdi_amini, inglorion, eraman, steven_wu, dexonsmith, llvm-commits

Differential Revision: https://reviews.llvm.org/D57206

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352441 91177308-0d34-0410-b5e6-96231b3b80d8

6 years agoDemanded elements support for vector GEPs
Philip Reames [Mon, 28 Jan 2019 23:24:49 +0000 (23:24 +0000)]
Demanded elements support for vector GEPs

GEPs can produce either scalar or vector results. If we're extracting only a subset of the vector lanes, simplifying the operands is helpful in eliminating redundant computation, and (eventually) allowing further optimizations

Differential Revision: https://reviews.llvm.org/D57177

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352440 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[docs] Fix a couple spelling errors.
Eli Friedman [Mon, 28 Jan 2019 23:03:41 +0000 (23:03 +0000)]
[docs] Fix a couple spelling errors.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352439 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[ThinLTO] Refine reachability check to fix compile time increase
Teresa Johnson [Mon, 28 Jan 2019 22:27:05 +0000 (22:27 +0000)]
[ThinLTO] Refine reachability check to fix compile time increase

Summary:
A recent fix to the ThinLTO whole program dead code elimination (D56117)
increased the thin link time on a large MSAN'ed binary by 2x.
It's likely that the time increased elsewhere, but was more noticeable
here since it was already large and ended up timing out.

That change made it so we would repeatedly scan all copies of linkonce
symbols for liveness every time they were encountered during the graph
traversal. This was needed since we only mark one copy of an aliasee as
live when we encounter a live alias. This patch fixes the issue in a
more efficient manner by simply proactively visiting the aliasee (thus
marking all copies live) when we encounter a live alias.

Two notes: One, this requires a hash table lookup (finding the aliasee
summary in the index based on aliasee GUID). However, the impact of this
seems to be small compared to the original pre-D56117 thin link time. It
could be addressed if we keep the aliasee ValueInfo in the alias summary
instead of the aliasee GUID, which I am exploring in a separate patch.

Second, we only populate the aliasee GUID field when reading summaries
from bitcode (whether we are reading individual summaries and merging on
the fly to form the compiled index, or reading in a serialized combined
index). Thankfully, that's currently the only way we can get to this
code as we don't yet support reading summaries from LLVM assembly
directly into a tool that performs the thin link (they must be converted
to bitcode first). I added a FIXME, however I have the fix under test
already. The easiest fix is to simply populate this field always, which
isn't hard, but more likely the change I am exploring to store the
ValueInfo instead as described above will subsume this. I don't want to
hold up the regression fix for this though.

Reviewers: trentxintong

Subscribers: mehdi_amini, inglorion, dexonsmith, llvm-commits

Differential Revision: https://reviews.llvm.org/D57203

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352438 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[CGP] auto-generate complete checks for add overflow tests; NFC
Sanjay Patel [Mon, 28 Jan 2019 22:07:37 +0000 (22:07 +0000)]
[CGP] auto-generate complete checks for add overflow tests; NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352437 91177308-0d34-0410-b5e6-96231b3b80d8

6 years agoRecommit r352255 "[SelectionDAG][X86] Don't use SEXTLOAD for promoting masked loads...
Craig Topper [Mon, 28 Jan 2019 21:38:47 +0000 (21:38 +0000)]
Recommit r352255 "[SelectionDAG][X86] Don't use SEXTLOAD for promoting masked loads in the type legalizer"

This did not cause the buildbot failure it was previously reverted for.

Original commit message:

I'm not sure why we were using SEXTLOAD. EXTLOAD seems more appropriate since we don't care about the upper bits.

This patch changes this and then modifies the X86 post legalization combine to emit a extending shuffle instead of a sign_extend_vector_inreg. Could maybe use an any_extend_vector_inre

On AVX512 targets I think we might be able to use a masked vpmovzx and not have to expand this at all.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352433 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[RuntimeDyld] load all sections with ProcessAllSections
Yonghong Song [Mon, 28 Jan 2019 21:35:23 +0000 (21:35 +0000)]
[RuntimeDyld] load all sections with ProcessAllSections

This patch tried to address the following use case.
  . bcc (https://github.com/iovisor/bcc) utilizes llvm JIT to
    compile for BTF target.
  . with -g, .BTF and .BTF.ext sections (BPF debug info)
    will be generated by LLVM.
  . .BTF does not have relocations and .BTF.ext has some
    relocations.
  . With ProcessAllSections, .BTF.ext is loaded by JIT dynamic linker
    and is available to application. But .BTF is not loaded.

The bcc application needs both .BTF.ext and .BTF for debugging
purpose, and .BTF is not loaded. This patch addressed this issue
by iterating over all sections and loading any missing
sections, after symbol/relocation processing in loadObjectImpl().

Signed-off-by: Yonghong Song <yhs@fb.com>
Differential Revision: https://reviews.llvm.org/D55943

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352432 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[ARM] Deduplicate table generated CC analysis code
Reid Kleckner [Mon, 28 Jan 2019 21:28:43 +0000 (21:28 +0000)]
[ARM] Deduplicate table generated CC analysis code

Create ARMCallingConv.cpp and emit code for calling convention analysis
from there.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352431 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[AArch64] Include AArch64GenCallingConv.inc once
Reid Kleckner [Mon, 28 Jan 2019 21:28:40 +0000 (21:28 +0000)]
[AArch64] Include AArch64GenCallingConv.inc once

Summary:
Avoids duplicating generated static helpers for calling convention
analysis.

This also means you can modify AArch64CallingConv.td without recompiling
the AArch64ISelLowering.cpp monolith, so it provides faster incremental
rebuilds.

Saves 12K in llc.exe, but adds a new object file, which is large.

Reviewers: efriedma, t.p.northover

Subscribers: mgorny, javed.absar, kristof.beyls, hiraditya, llvm-commits

Differential Revision: https://reviews.llvm.org/D56948

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352430 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[GlobalISel][AArch64] Add legalization for G_FLOG
Jessica Paquette [Mon, 28 Jan 2019 21:27:23 +0000 (21:27 +0000)]
[GlobalISel][AArch64] Add legalization for G_FLOG

This adds support for legalizing G_FLOG into a RTLib call.

It adds a legalizer test, and updates the existing floating point tests.

https://reviews.llvm.org/D57347

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352429 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[InstCombine] add another saturating uadd test (no undefs); NFC
Sanjay Patel [Mon, 28 Jan 2019 20:37:18 +0000 (20:37 +0000)]
[InstCombine] add another saturating uadd test (no undefs); NFC

I forgot that our undef matching hasn't been completed in the previous commit.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352424 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[InstCombine] add tests for saturating uadd with constant; NFC
Sanjay Patel [Mon, 28 Jan 2019 20:32:48 +0000 (20:32 +0000)]
[InstCombine] add tests for saturating uadd with constant; NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352423 91177308-0d34-0410-b5e6-96231b3b80d8

6 years agoAMDGPU: Add DS append/consume intrinsics
Matt Arsenault [Mon, 28 Jan 2019 20:14:49 +0000 (20:14 +0000)]
AMDGPU: Add DS append/consume intrinsics

Since these pass the pointer in m0 unlike other DS instructions, these
need to worry about whether the address is uniform or not. This
assumes the address is dynamically uniform, and just uses
readfirstlane to get a copy into an SGPR.

I don't know if these have the same 16-bit add for the addressing mode
offset problem on SI or not, but I've just assumed they do.

Also includes some misc. changes to avoid test differences between the
LDS and GDS versions.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352422 91177308-0d34-0410-b5e6-96231b3b80d8

6 years agogn build: Add get.py script to download prebuilt gn, make gn.py run downloaded gn...
Nico Weber [Mon, 28 Jan 2019 19:54:41 +0000 (19:54 +0000)]
gn build: Add get.py script to download prebuilt gn, make gn.py run downloaded gn if gn is not on PATH

Prebuilts are available for x86_64 Linux, macOS, Windows. The script always
pulls the latest GN version.

Differential Revision: https://reviews.llvm.org/D57256

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352420 91177308-0d34-0410-b5e6-96231b3b80d8

6 years agogn build: Make cmake sync script work on Windows if git is a bat file
Nico Weber [Mon, 28 Jan 2019 19:53:52 +0000 (19:53 +0000)]
gn build: Make cmake sync script work on Windows if git is a bat file

Differential Revision: https://reviews.llvm.org/D57338

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352419 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[GlobalISel][AArch64] Add instruction selection support for @llvm.log10
Jessica Paquette [Mon, 28 Jan 2019 19:53:14 +0000 (19:53 +0000)]
[GlobalISel][AArch64] Add instruction selection support for @llvm.log10

This adds instruction selection support for @llvm.log10 in AArch64. It teaches
GISel to lower it to a library call, updates the relevant tests, and adds a
legalizer test for log10.

https://reviews.llvm.org/D57341

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352418 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[AliasSetTracker] Cleanup more comments. [NFCI]
Alina Sbirlea [Mon, 28 Jan 2019 19:38:03 +0000 (19:38 +0000)]
[AliasSetTracker] Cleanup more comments. [NFCI]

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352416 91177308-0d34-0410-b5e6-96231b3b80d8

6 years agogn build: Fix `lld-link: unknown flag: -fuse-ld=lld` warnings on Windows
Nico Weber [Mon, 28 Jan 2019 19:32:52 +0000 (19:32 +0000)]
gn build: Fix `lld-link: unknown flag: -fuse-ld=lld` warnings on Windows

Fixes a minor regression from r351248.

While here, also make it possible to opt out of lld by saying
use_lld=false when clang_base_path is set. (use_lld still defaults to
true if clang_base_path is set.)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352415 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[MC] Do not consider .ifdef/.ifndef as a use
Scott Linder [Mon, 28 Jan 2019 19:32:08 +0000 (19:32 +0000)]
[MC] Do not consider .ifdef/.ifndef as a use

This is allowed by GAS and seems correct.

Differential Revision: https://reviews.llvm.org/D55439

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352414 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[AArch64] Add 'apple-latest' CPU alias
Francis Visoiu Mistrih [Mon, 28 Jan 2019 19:27:33 +0000 (19:27 +0000)]
[AArch64] Add 'apple-latest' CPU alias

The 'apple-latest' alias is supposed to provide a CPU that contains the
latest Apple processor model supported by LLVM.

This is supposed to be used by tools like lldb to provide a target that
supports most of the CPU features.

For now, this is mapped to Cyclone.

Differential Revision: https://reviews.llvm.org/D56384

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352412 91177308-0d34-0410-b5e6-96231b3b80d8

6 years agoRevert "[CMake] Use __libc_start_main rather than fopen when checking for C library"
Petr Hosek [Mon, 28 Jan 2019 19:26:41 +0000 (19:26 +0000)]
Revert "[CMake] Use __libc_start_main rather than fopen when checking for C library"

This reverts commit r352341: it broke the build on macOS which doesn't
seem to provide __libc_start_main in its C library.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352411 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[GlobalISel] Add ISel support for @llvm.lifetime.start and @llvm.lifetime.end
Jessica Paquette [Mon, 28 Jan 2019 19:22:29 +0000 (19:22 +0000)]
[GlobalISel] Add ISel support for @llvm.lifetime.start and @llvm.lifetime.end

This adds ISel support for lifetime markers in opt levels above O0.

It also updates the arm64-irtranslator test, and updates some AArch64 tests that
use them for added coverage.

It also adds a testcase taken from the X86 codegen tests which verified a bug
caused by lifetime markers + stack colouring in the past. This is intended to
make sure that GISel doesn't re-introduce the bug.

(This is basically a straight copy from what SelectionDAG does in
SelectionDAGBuilder.cpp)

https://reviews.llvm.org/D57187

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352410 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[CodeGen][X86] Expand UADDSAT to NOT+UMIN+ADD
Nikita Popov [Mon, 28 Jan 2019 19:19:09 +0000 (19:19 +0000)]
[CodeGen][X86] Expand UADDSAT to NOT+UMIN+ADD

Followup to D56636, this time handling the UADDSAT case by expanding
uadd.sat(a, b) to umin(a, ~b) + b.

Differential Revision: https://reviews.llvm.org/D56869

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352409 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[CodeExtractor] Add support for the `swifterror` attribute
Vedant Kumar [Mon, 28 Jan 2019 19:13:37 +0000 (19:13 +0000)]
[CodeExtractor] Add support for the `swifterror` attribute

When passing a `swifterror` argument or alloca as an input to an
extraction region, mark the input parameter `swifterror`.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352408 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[AliasSetTracker] Cleanup comments. [NFCI]
Alina Sbirlea [Mon, 28 Jan 2019 19:01:32 +0000 (19:01 +0000)]
[AliasSetTracker] Cleanup comments. [NFCI]

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352406 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[GlobalISel][AArch64] Add instruction selection support for G_FCOS and G_FSIN
Jessica Paquette [Mon, 28 Jan 2019 18:34:18 +0000 (18:34 +0000)]
[GlobalISel][AArch64] Add instruction selection support for G_FCOS and G_FSIN

This contains all of the legalizer changes from D57197 necessary to select
G_FCOS and G_FSIN. It also updates several existing IR tests in
test/CodeGen/AArch64 that verify that we correctly lower the G_FCOS and G_FSIN
instructions.

https://reviews.llvm.org/D57197
3/3

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352402 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[GlobalISel][AArch64] Add IRTranslator support for G_FCOS and G_FSIN
Jessica Paquette [Mon, 28 Jan 2019 18:34:17 +0000 (18:34 +0000)]
[GlobalISel][AArch64] Add IRTranslator support for G_FCOS and G_FSIN

This adds IRTranslator support for the G_FCOS and G_FSIN generic instructions.

https://reviews.llvm.org/D57197
2/3

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352401 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[GlobalISel] Add G_FSIN and G_FCOS generic instructions
Jessica Paquette [Mon, 28 Jan 2019 18:34:16 +0000 (18:34 +0000)]
[GlobalISel] Add G_FSIN and G_FCOS generic instructions

This introduces generic instrutions for floating point sin and cos, G_FCOS and
G_FSIN. It updates the tests, etc.

https://reviews.llvm.org/D57197
1/3

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352400 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[AliasSetTracker] Update signature to aliasesPointer [NFCI].
Alina Sbirlea [Mon, 28 Jan 2019 18:30:05 +0000 (18:30 +0000)]
[AliasSetTracker] Update signature to aliasesPointer [NFCI].

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352399 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[NFC] TLI query with default(on) behavior wrt DAG combines for fmin/fmax target control
Michael Berg [Mon, 28 Jan 2019 18:03:08 +0000 (18:03 +0000)]
[NFC] TLI query with default(on) behavior wrt DAG combines for fmin/fmax target control

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352396 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[SimpleLoopUnswitch] Early check exit for trivial unswitch with MemorySSA.
Alina Sbirlea [Mon, 28 Jan 2019 17:48:45 +0000 (17:48 +0000)]
[SimpleLoopUnswitch] Early check exit for trivial unswitch with MemorySSA.

Summary:
If MemorySSA is avaiable, we can skip checking all instructions if block has any Defs.
(volatile loads are also Defs).
We still need to check all instructions for "canThrow", even if no Defs are found.

Reviewers: chandlerc

Subscribers: sanjoy, jlebar, Prazek, george.burgess.iv, llvm-commits

Differential Revision: https://reviews.llvm.org/D57129

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352393 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[X86][AVX] Remove lowerShuffleByMerging128BitLanes 2-lane restriction
Simon Pilgrim [Mon, 28 Jan 2019 17:02:35 +0000 (17:02 +0000)]
[X86][AVX] Remove lowerShuffleByMerging128BitLanes 2-lane restriction

First step towards adding support for 64-bit unary "sublane" handling (a bit like lowerShuffleAsRepeatedMaskAndLanePermute).

This allows us to add lowerV64I8Shuffle handling.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352389 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[LangRef] Mention vector support for bitreverse/bswap intrinsics (PR38012)
Simon Pilgrim [Mon, 28 Jan 2019 16:56:38 +0000 (16:56 +0000)]
[LangRef] Mention vector support for bitreverse/bswap intrinsics (PR38012)

Differential Revision: https://reviews.llvm.org/D57309

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352386 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[llvm-objdump] - Restore a piece of code removed by mistake in r352366.
George Rimar [Mon, 28 Jan 2019 16:36:12 +0000 (16:36 +0000)]
[llvm-objdump] - Restore a piece of code removed by mistake in r352366.

Seems when committed the r352366
("[llvm-objdump] - Print LMAs when dumping section headers.")
I resolved merge conflict incorrectly and removed this piece by mistake.

Bots did not catch this yet, seems they are slow today,
but the `X86/adjust-vma.test` test case fails locally for me without that.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352383 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[x86] allow more shuffle splitting to avoid vpermps (PR40434)
Sanjay Patel [Mon, 28 Jan 2019 15:51:34 +0000 (15:51 +0000)]
[x86] allow more shuffle splitting to avoid vpermps (PR40434)

This is tricky to make optimal: sometimes we're better off using
a single wider op, but other times it makes more sense to combine
a narrow ops to achieve the same result.

This solves the case from:
https://bugs.llvm.org/show_bug.cgi?id=40434

There's potentially a similar change for vectors with 64-bit elements,
but it needs adjustments similar to rL352333 to avoid creating infinite
loops.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352380 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[llvm-objdump] - Update test after r352366. NFC.
George Rimar [Mon, 28 Jan 2019 15:49:41 +0000 (15:49 +0000)]
[llvm-objdump] - Update test after r352366. NFC.

Change the column name.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352379 91177308-0d34-0410-b5e6-96231b3b80d8

6 years agoVERSION_GREATER_EQUAL not supported in llvm cmake.
Ranjeet Singh [Mon, 28 Jan 2019 15:48:07 +0000 (15:48 +0000)]
VERSION_GREATER_EQUAL not supported in llvm cmake.

Patch https://reviews.llvm.org/D56329 caused build failures for me when
building on Windows because of the use of cmake operator
'VERSION_GREATER_EQUAL' which isn't supported in older versions of cmake. The
llvm website states that minimum required version of cmake for building llvm is
3.4.3 https://llvm.org/docs/CMake.html

Differential Revision: https://reviews.llvm.org/D57326

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352378 91177308-0d34-0410-b5e6-96231b3b80d8

6 years agoRemove no longer needed Arm specific LICENSE.TXT file.
Arnaud A. de Grandmaison [Mon, 28 Jan 2019 15:38:01 +0000 (15:38 +0000)]
Remove no longer needed Arm specific LICENSE.TXT file.

As the codebase is now under the Apache 2.0 license with LLVM
Exceptions, and all Arm's contributions, past or future, are under that
new license, this Arm specific LICENSE.TXT is no longer needed, thus
removing it.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352376 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[cmake] Fix get_llvm_lit_path() to respect LLVM_EXTERNAL_LIT always
Michal Gorny [Mon, 28 Jan 2019 15:16:03 +0000 (15:16 +0000)]
[cmake] Fix get_llvm_lit_path() to respect LLVM_EXTERNAL_LIT always

Refactor the get_llvm_lit_path() logic to respect LLVM_EXTERNAL_LIT,
and require the fallback to be defined explicitly
as LLVM_DEFAULT_EXTERNAL_LIT. This fixes building libcxx standalone
after r346888.

The old logic was using LLVM_EXTERNAL_LIT both as user-defined cache
variable and an optional pre-definition of default value from caller
(e.g. libcxx). It included a hack to make this work by assigning
the value back and forth but it was fragile and stopped working
in libcxx.

The new logic is simpler and more transparent. Default value is
provided in a separate variable, and used only when user-specified
variable is empty (i.e. not overriden).

Differential Revision: https://reviews.llvm.org/D57282

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352374 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[obj2yaml] - Dump the sh_entsize section field.
George Rimar [Mon, 28 Jan 2019 15:05:10 +0000 (15:05 +0000)]
[obj2yaml] - Dump the sh_entsize section field.

I faced with the fact that obj2yaml does not dump the sh_entsize field.
A problem arose when I tried to dump ELF versioning sections.

This is close to what D50235 did, but D50235 did the change for yaml2obj, and now
I had to do the same for obj2yaml.

Differential revision: https://reviews.llvm.org/D57229

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352373 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[llvm-objcopy] Fix crash when writing empty binary output
Jordan Rupprecht [Mon, 28 Jan 2019 15:02:40 +0000 (15:02 +0000)]
[llvm-objcopy] Fix crash when writing empty binary output

Summary: When using llvm-objcopy -O binary and the resulting file will be empty (e.g. removing the only section that would be written, or using --only-keep with a section that doesn't exist/isn't SHF_ALLOC), we crash because FileOutputBuffer expects Size > 0. Add a regression test, and change Buffer to open/truncate the output file in this case.

Reviewers: alexshap, jhenderson, jakehehrlich, espindola

Reviewed By: alexshap, jhenderson

Subscribers: jfb, llvm-commits, emaste, arichardson

Differential Revision: https://reviews.llvm.org/D56806

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352371 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[mips] Support for +abs2008 attribute
Aleksandar Beserminji [Mon, 28 Jan 2019 14:59:30 +0000 (14:59 +0000)]
[mips] Support for +abs2008 attribute

Instruction abs.[ds] is not generating correct result when working
with NaNs for revisions prior mips32r6 and mips64r6.

To generate a sequence which always produce a correct result, but also
to allow user more control on how his code is compiled, attribute
+abs2008 is added, so user can choose legacy or 2008.

By default legacy mode is used on revisions prior R6. Mips32r6 and
mips64r6 use abs2008 mode by default.

Differential Revision: https://reviews.llvm.org/D35983

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352370 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[llvm-objdump] - Print LMAs when dumping section headers.
George Rimar [Mon, 28 Jan 2019 14:11:35 +0000 (14:11 +0000)]
[llvm-objdump] - Print LMAs when dumping section headers.

When --section-headers is used, GNU objdump prints both LMA and VMA for sections.
llvm-objdump does not do that what makes it's output be slightly inconsistent.

Patch teaches llvm-objdump to print LMA/VMA for ELF file formats.
The behavior for other formats remains unchanged.

Differential revision: https://reviews.llvm.org/D57146

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352366 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[AMDGPU] Add intrinsics for 16 bit interpolation
Tim Corringham [Mon, 28 Jan 2019 13:48:59 +0000 (13:48 +0000)]
[AMDGPU] Add intrinsics for 16 bit interpolation

Summary:
Added the intrinsics llvm.amdgcn.interp.p1.f16() and
llvm.amdgcn.interp.p2.f16() and related LIT test.

The p1 intrinsic generates code appropriate for both 16 and 32
bank LDS.

Reviewers: #amdgpu, dstuttard, arsenm, tpr

Reviewed By: #amdgpu, arsenm

Subscribers: jvesely, mgorny, arsenm, kzhuravl, wdng, nhaehnle, yaxunl, dstuttard, tpr, t-tye, llvm-commits

Differential Revision: https://reviews.llvm.org/D46754

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352357 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[opaque pointer types] Remove GraphTraits specialization for Type.
James Y Knight [Mon, 28 Jan 2019 13:25:57 +0000 (13:25 +0000)]
[opaque pointer types] Remove GraphTraits specialization for Type.

The only caller has been deleted in r352076, and I'd like to minimize
the amount of code walking Type hierarchies generically, to make it
easier to identify code depending on pointee types.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352353 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[MIPS GlobalISel] Select sub
Petar Avramovic [Mon, 28 Jan 2019 12:10:17 +0000 (12:10 +0000)]
[MIPS GlobalISel] Select sub

Lower G_USUBO and G_USUBE. Add narrowScalar for G_SUB.
Legalize and select G_SUB for MIPS 32.

Differential Revision: https://reviews.llvm.org/D53416

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352351 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[DebugInfo][DAG] Avoid re-ordering of DBG_VALUEs
Jeremy Morse [Mon, 28 Jan 2019 12:08:31 +0000 (12:08 +0000)]
[DebugInfo][DAG] Avoid re-ordering of DBG_VALUEs

This patch improves the placement of DBG_VALUEs when by SelectionDAG, which
as documented in PR40427 can go very wrong. At the core of this is
ProcessSourceNode, which assumes the last instruction in a BB is the start
of the last processed IR instruction, which isn't always true.

Instead, use a helper function to call InstrEmitter::EmitNode, that records
before-and-after iterators and determines the first of any new instruction
created during emission. This is passed to ProcessSourceNode, which can
then make more elightened decisions about ordering for DBG_VALUE placement.

Differential revision: https://reviews.llvm.org/D57163

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352350 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[llvm-objdump] - Fix comment. NFC.
George Rimar [Mon, 28 Jan 2019 10:48:54 +0000 (10:48 +0000)]
[llvm-objdump] - Fix comment. NFC.

This was mentioned by James Henderson
in review for https://reviews.llvm.org/D57051.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352348 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[llvm-objdump] - Implement the --adjust-vma option.
George Rimar [Mon, 28 Jan 2019 10:44:01 +0000 (10:44 +0000)]
[llvm-objdump] - Implement the --adjust-vma option.

GNU objdump's help says: "--adjust-vma: Add OFFSET to all displayed section addresses"
In real life what it does is a bit more complicated
(and IMO not always reasonable. For example, GNU objdump prints not only VMA, but also LMA
for sections. And with --adjust-vma it adjusts LMA, but only when a section has relocations.
llvm-objsump does not seem to support printing LMAs yet, but GNU's logic anyways does not
make sense for me here).

This patch tries to adjust VMA. I tried to implement a reasonable approach.
I am not adjusting sections that are not allocatable. As, for example, adjusting debug sections
VA's and rel[a] sections VA's should not make sense. This behavior seems to be GNU compatible.

Differential revision: https://reviews.llvm.org/D57051

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352347 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[ARM GlobalISel] Support integer division for Thumb2
Diana Picus [Mon, 28 Jan 2019 10:37:30 +0000 (10:37 +0000)]
[ARM GlobalISel] Support integer division for Thumb2

Support G_SDIV, G_UDIV, G_SREM and G_UREM.

The only significant difference between arm and thumb mode is that we
need to check a different subtarget feature.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352346 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[X86] Add new variadic avx512 compress/expand intrinsics that use vXi1 types for...
Craig Topper [Mon, 28 Jan 2019 07:03:03 +0000 (07:03 +0000)]
[X86] Add new variadic avx512 compress/expand intrinsics that use vXi1 types for the mask argument.

Remove and autoupgrade the old intrinsics

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352343 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[X86] Add vbmi2 compressstore and expandload tests that aren't fast-isel tests.
Craig Topper [Mon, 28 Jan 2019 05:42:39 +0000 (05:42 +0000)]
[X86] Add vbmi2 compressstore and expandload tests that aren't fast-isel tests.

These got removed when we autoupgraded to target independent intrinsics, but we didn't have coverage anywhere else. The avx512f/avx512vl versions do have coverage.

Also move some tests back from the upgrade file that aren't really upgraded.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352342 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[CMake] Use __libc_start_main rather than fopen when checking for C library
Petr Hosek [Mon, 28 Jan 2019 04:12:54 +0000 (04:12 +0000)]
[CMake] Use __libc_start_main rather than fopen when checking for C library

The check_library_exists CMake uses a custom symbol definition. This
is a problem when checking for C library symbols because Clang
recognizes many of them as builtins, and returns the
-Wbuiltin-requires-header (or -Wincompatible-library-redeclaration)
error. When building with -Werror which is the default, this causes
the check_library_exists check fail making the build think that C
library isn't available.

To avoid this issue, we should use a symbol that isn't recognized by
Clang and wouldn't cause the same issue. __libc_start_main seems like
reasonable choice that fits the bill.

Differential Revision: https://reviews.llvm.org/D57142

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352341 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[AArch64][GlobalISel] Teach RBS about G_FNEG default mapping.
Amara Emerson [Mon, 28 Jan 2019 03:21:14 +0000 (03:21 +0000)]
[AArch64][GlobalISel] Teach RBS about G_FNEG default mapping.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352340 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[AArch64][GlobalISel] Add some missing vector support for FP arithmetic ops.
Amara Emerson [Mon, 28 Jan 2019 02:28:22 +0000 (02:28 +0000)]
[AArch64][GlobalISel] Add some missing vector support for FP arithmetic ops.

Moved the fneg lowering legalization test from AArch64 to X86, as we want to
specify that it's already legal.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352338 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[AArch64][GlobalISel] Add some vector support for fp <-> int conversions.
Amara Emerson [Mon, 28 Jan 2019 02:27:59 +0000 (02:27 +0000)]
[AArch64][GlobalISel] Add some vector support for fp <-> int conversions.

Some unrelated, but benign, test changes as well due to the test update script.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352337 91177308-0d34-0410-b5e6-96231b3b80d8

6 years agoGlobalISel: Don't reduce elements for atomic load/store
Matt Arsenault [Sun, 27 Jan 2019 22:36:24 +0000 (22:36 +0000)]
GlobalISel: Don't reduce elements for atomic load/store

This is invalid for the same reason as in the narrowScalar handling
for load.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352334 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[x86] add restriction for lowering to vpermps
Sanjay Patel [Sun, 27 Jan 2019 21:53:33 +0000 (21:53 +0000)]
[x86] add restriction for lowering to vpermps

This transform was added with rL351346, and we had
an escape for shufps, but we also want one for
unpckps vs. vpermps because vpermps doesn't take
an immediate shuffle index operand.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352333 91177308-0d34-0410-b5e6-96231b3b80d8

6 years agoGlobalISel: Factor fewerElementVectors into separate functions
Matt Arsenault [Sun, 27 Jan 2019 21:53:09 +0000 (21:53 +0000)]
GlobalISel: Factor fewerElementVectors into separate functions

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352332 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[x86] add tests for extract/extract/unpack; NFC
Sanjay Patel [Sun, 27 Jan 2019 21:34:51 +0000 (21:34 +0000)]
[x86] add tests for extract/extract/unpack; NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352331 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[X86][SSE] Add UNDEF handling to combineSelect ISD::USUBSAT matching (PR40083)
Simon Pilgrim [Sun, 27 Jan 2019 21:01:23 +0000 (21:01 +0000)]
[X86][SSE] Add UNDEF handling to combineSelect ISD::USUBSAT matching (PR40083)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352330 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[X86][SSE] Add UNDEF test case for combineSelect ISD::USUBSAT matching (PR40083)
Simon Pilgrim [Sun, 27 Jan 2019 20:52:34 +0000 (20:52 +0000)]
[X86][SSE] Add UNDEF test case for combineSelect ISD::USUBSAT matching (PR40083)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352329 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[X86][SSE] Permit UNDEFs in combineAddToSUBUS matching (PR40083)
Simon Pilgrim [Sun, 27 Jan 2019 20:36:37 +0000 (20:36 +0000)]
[X86][SSE] Permit UNDEFs in combineAddToSUBUS matching (PR40083)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352328 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[x86] add more tests for lowerShuffleWithUndefHalf; NFC
Sanjay Patel [Sun, 27 Jan 2019 20:17:02 +0000 (20:17 +0000)]
[x86] add more tests for lowerShuffleWithUndefHalf; NFC

Some other transform is creating the opposite form and causing
an infinite loop if we try to split some of these.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352327 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[X86][SSE] Add PSUBUS undef element test case (PR40083)
Simon Pilgrim [Sun, 27 Jan 2019 20:09:30 +0000 (20:09 +0000)]
[X86][SSE] Add PSUBUS undef element test case (PR40083)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352326 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[COFF] Add new relocation types.
Martin Storsjo [Sun, 27 Jan 2019 19:53:36 +0000 (19:53 +0000)]
[COFF] Add new relocation types.

Differential Revision: https://reviews.llvm.org/D57291

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352324 91177308-0d34-0410-b5e6-96231b3b80d8

6 years agoFix some warnings on MSVC
Alexandre Ganea [Sun, 27 Jan 2019 18:41:40 +0000 (18:41 +0000)]
Fix some warnings on MSVC

Differential Revision: https://reviews.llvm.org/D56329

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352322 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[X86] Add test cases for PR36721 (unnecessary andl for %cl when shifting)
Simon Pilgrim [Sun, 27 Jan 2019 18:31:33 +0000 (18:31 +0000)]
[X86] Add test cases for PR36721 (unnecessary andl for %cl when shifting)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352321 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[x86] refactor logic in lowerShuffleWithUndefHalf
Sanjay Patel [Sun, 27 Jan 2019 18:12:03 +0000 (18:12 +0000)]
[x86] refactor logic in lowerShuffleWithUndefHalf

Although this is longer code, this is no-functional-change-intended.
The goal is to untangle the conditions under which we bail out, so
that's easier to adjust.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352320 91177308-0d34-0410-b5e6-96231b3b80d8

6 years agoGlobalISel: Verify load/store has a pointer input
Matt Arsenault [Sun, 27 Jan 2019 15:57:23 +0000 (15:57 +0000)]
GlobalISel: Verify load/store has a pointer input

I expected this to be automatically verified, but it seems
nothing uses that the type index was declared as a "ptype"

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352319 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[X86][NFC] Replace "<%s" with "< %s" in run-lines.
Roman Lebedev [Sun, 27 Jan 2019 15:36:35 +0000 (15:36 +0000)]
[X86][NFC] Replace "<%s" with "< %s" in run-lines.

While i have no intention of actually commiting regeneration
of the check lines in these test files with update_llc_test_checks,
lack of that whitespace breaks that util, which is mildly inconvenient.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352318 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[NFC][MCA][X86][BdVer2] Cherry-pick int-to-ivec forwarding tests from BtVer2
Roman Lebedev [Sun, 27 Jan 2019 14:35:54 +0000 (14:35 +0000)]
[NFC][MCA][X86][BdVer2] Cherry-pick int-to-ivec forwarding tests from BtVer2

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352317 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[X86] Add CGP tests for PR40486
Simon Pilgrim [Sun, 27 Jan 2019 14:04:45 +0000 (14:04 +0000)]
[X86] Add CGP tests for PR40486

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352316 91177308-0d34-0410-b5e6-96231b3b80d8

6 years ago[TTI] Add generic SADDSAT/SSUBSAT costs
Simon Pilgrim [Sun, 27 Jan 2019 13:51:59 +0000 (13:51 +0000)]
[TTI] Add generic SADDSAT/SSUBSAT costs

Add generic costs calculation for SADDSAT/SSUBSAT intrinsics, this uses generic costs for sadd_with_overflow/ssub_with_overflow, an extra sign comparison + a selects based on the sign/overflow.

This completes PR40316

Differential Revision: https://reviews.llvm.org/D57239

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@352315 91177308-0d34-0410-b5e6-96231b3b80d8