The memory patterns were near the MOVDDUP definition, but the non-memory patterns were near the broadcast instructions.

llvm-svn: 313494

llvm-svn: 313493

[X86] Remove isel patterns for X86Movhlps and X86Movlhps with integer types. Lowering doesn't emit these.

llvm-svn: 313492

llvm-svn: 313491

Differential Revision: https://reviews.llvm.org/D37962

llvm-svn: 313490

This reverts commit r313488, because it breaks compilation on Android and breaks llvm-lto. 

llvm-svn: 313489

Summary:
ld64 on OSX uses the old ThinLTOCodegenerator API. When two modules have the same name in an archive (valid archive), a name collision happens for the modules' buffer identifiers.
This PR resolves this, by suffixing the module name with an increasing number such that the identifiers are guaranteed to be unique.

For a similar fix in LLD, see https://reviews.llvm.org/D25495

Reviewers: mehdi_amini, tejohnson

Reviewed By: mehdi_amini

Subscribers: inglorion, eraman, llvm-commits

Differential Revision: https://reviews.llvm.org/D37961

llvm-svn: 313488

Fix the second half of PR34266:  Don't implicitly capture '*this' if the members are found in a class unrelated to the enclosing class.

https://bugs.llvm.org/show_bug.cgi?id=34266

For e.g.
  struct A {
     void f(int);
     static void f(char);
  };
  struct B {
    auto foo() {
      return [&] (auto a) {
         A::f(a); // this should not cause a capture of '*this'
      };
    }
  };

The patch does the following:
1) It moves the check to attempt an implicit capture of '*this' by reference into the more logical location of when the call is actually built within ActOnCallExpr (as opposed to when the unresolved-member-lookup node is created).
  - Reminder: A capture of '*this' by value has to always be an explicit capture.

2) It additionally checks whether the naming class of the UnresolvedMemberExpr ('A' in the example above) is related to the enclosing class ('B' above).

P.S. If you have access to ISO-C++'s CWG reflector, see this thread for some potentially related discussion: http://lists.isocpp.org/core/2017/08/2851.php

llvm-svn: 313487

This Disassembly support allows for 'round-trip' testing, and rv32i-valid.s
has been updated appropriately.

Differential Revision: https://reviews.llvm.org/D23567

llvm-svn: 313486

This patch supports all RV32I instructions as described in the RISC-V manual.
A future patch will add support for pseudoinstructions and other instruction
expansions (e.g. 0-arg fence -> fence iorw, iorw).

Differential Revision: https://reviews.llvm.org/D23566

llvm-svn: 313485

llvm-svn: 313484

llvm-svn: 313483

Lambdas are slightly more verbose, but also more readable. No
functionality change intended.

llvm-svn: 313482

It's removed in C++17. No functionality change intended.

llvm-svn: 313481

Commit r313277 moved IsStackOverflow to inside the SignalContext
class, but didn't update a code block in #ifdef s390x accordingly.

llvm-svn: 313480

llvm-svn: 313479

Summary: G_FCONSTANT support, port the implementation from X86FastIsel.

Reviewers: zvi, delena, guyblank

Reviewed By: delena

Subscribers: rovka, llvm-commits, kristof.beyls

Differential Revision: https://reviews.llvm.org/D37734

llvm-svn: 313478

llvm-svn: 313477

llvm-svn: 313476

llvm-svn: 313475

This can be used to add a relocatable object to the JIT session.

llvm-svn: 313474

Summary: Fix a few typos and update names to match current source.

Differential Revision: https://reviews.llvm.org/D37948

llvm-svn: 313473

This adds new patterns and SDNodeXForm to enable the immediate to commuted.

llvm-svn: 313472

Previously we were just relying on pattern order to define precedence. Which works, but isn't the best way.

llvm-svn: 313471

llvm-svn: 313470

It doesn't make sense to me why these bots are failing as the
traceback does not agree with the source code.  It's possible
something is stale or there is some other mysterious error,
but in any case hopefully this fixes it.

llvm-svn: 313469

This was a bug in the test that was only exposed as a result of
refactoring some code in lit configuration files.  Previously,
llvm's lit configuration would only set the target-windows feature
if the system was also windows.  Since cross-compilation is
a thing, this isn't correct.  target-windows should be set
independently of system-windows.

Adding to that bug, this particular test then checked for
target-windows when it really meant "can I call a certain API on
the host machine", which is what system-windows is for.

Ultimately, this test only works if *both* the target and host
are Windows, so I've updated the test to reflect that.

llvm-svn: 313468

There were some issues surrounding Py2 / Py3 compatibility, but
I've now tested with both Py2 and Py3 and everything seems to
work.

llvm-svn: 313467

llvm-svn: 313466

The same code appears earlier in the function. This represents an earlier version of what became r313373 that I still had sitting in my local repo.

llvm-svn: 313465

This will print all children of a DIE when selectively printing only
one DIE at a given offset.

llvm-svn: 313464

llvm-svn: 313463

Summary:
For a lot of older CPUs we have a 1:1 mapping between CPU name and enum name. But many of them are effectively aliases of each other and as a result are always repeated together at every usage

This patch removes most of the duplication. It also uses StringSwitch::Cases to make the many to one mapping in the StringSwitch more obvious.

Reviewers: RKSimon, spatel, zvi, igorb

Reviewed By: RKSimon

Subscribers: cfe-commits

Differential Revision: https://reviews.llvm.org/D37938

llvm-svn: 313462

A few tests were manually constructing a LitConfig object, since
I added a new argument to it this was triggering some failures
I didn't detect.  `ninja check-lit` passes now.

llvm-svn: 313461

This is helpful for debugging test failures since it removes
the multiprocessing pool from the picture.  This will obviously
slow down the test suite by a few orders of magnitude, so it
should only be used for debugging specific failures.

llvm-svn: 313460

readelf tool reports an error when output contains the same section
in multiple COMDAT groups. That can be useful.
Path teaches llvm-readobj to do the same.

Differential revision: https://reviews.llvm.org/D37567

llvm-svn: 313459

This allows vector-sized store merging of constants in DAGCombiner using the existing code in MergeConsecutiveStores(). 
All of the twisted logic that decides exactly what vector operations are legal and fast for each particular CPU are 
handled separately in there using the appropriate hooks.

For the motivating tests in merge-store-constants.ll, we already produce the same vector code in IR via the SLP vectorizer. 
So this is just providing a backend backstop for code that doesn't go through that pass (-O1). More details in PR24449:
https://bugs.llvm.org/show_bug.cgi?id=24449 (this change should be the last step to resolve that bug)

Differential Revision: https://reviews.llvm.org/D37451

llvm-svn: 313458

llvm-svn: 313457

llvm-svn: 313456

[X86] Add isel patterns to be able to fold loads into VPERM2F128 even when the load is on the first input to the SDNode.

We just need to toggle bits 1 and 5 of the immediate and swap the sources. The peephole pass could trigger commuting/folding for this later, but its easy enough to fix in isel.

Disable the peephole pass on the main vperm2x128 test so we know we're doing this through isel.

llvm-svn: 313455