llvm-svn: 324033

Every instruction that has the word TEST in its name seems to have been buried into EmitTest. But that code is largely concerned with trying to reuse the flags from instructions that update flags in a pretty normal way.

PTEST/TESTP/KTEST do not update flags in a normal way. They only update Z and C and the C flag update is non-standard. Rather than try to bend EmitTest's already complex logic to accomodate this, just move the call up to LowerSETCC and replicate the few pre-checks that are needed.

While there add a FIXME for using the C flag for checking for all 1s which we definitely couldn't do from EmitTEST.

llvm-svn: 324029

Legalizing vectors may keep the element type the same but change the number of
elements, the assert didn't take this into account.

llvm-svn: 324028

Indeed, "CHANGE" is not a thing yet, and we should probably not carry
around dead code which does not do anything apart from confusing the
reader.

Differential Revision: https://reviews.llvm.org/D42819

llvm-svn: 324027

llvm-svn: 324024

llvm-svn: 324022

[analyzer] [tests] Add an option to show the histogram of path differences between the analyzer runs

Differential Revision: https://reviews.llvm.org/D42778

llvm-svn: 324021

Patch by Chris Kennelly!

Differential Revision: https://reviews.llvm.org/D41746

llvm-svn: 324020

LLDB queries Spotlight to locate .dSYM bundles based on the UUID
embedded in a binary, and because the UUID is a hash of filename and
.text section, there *will* be conflicts inside $BUILD_DIR.

This should fix the broken green dragon bots.

llvm-svn: 324019

We use CXXTempObjectRegion exclusively as a bailout value for construction
targets when we are unable to find the correct construction region.
Sometimes it works correctly, but rather accidentally than intentionally.

Now that we want to increase the amount of situations where it works correctly,
the first step is to introduce a different way of communicating our failure
to find the correct construction region. EvalCallOptions are introduced
for this purpose.

For now EvalCallOptions are communicating two kinds of problems:
- We have been completely unable to find the correct construction site.
- We have found the construction site correctly, and there's more than one of
  them (i.e. array construction which we currently don't support).

Accidentally find and fix a test in which the new approach to communicating
failures produces better results.

Differential Revision: https://reviews.llvm.org/D42457

llvm-svn: 324018

llvm-svn: 324017

llvm-svn: 324016

llvm-svn: 324015

This is the enhancement suggested in D42536 to fix a shortcoming in 
regular InstCombine's canEvaluate* functionality.
When we have multiple uses of a value, but they're all in one instruction, we can 
allow that expression to be narrowed or widened for the same cost as a single-use 
value.

AFAICT, this can only matter for multiply: sub/and/or/xor/select would be simplified 
away if the operands are the same value; add becomes shl; shifts with a variable shift 
amount aren't handled.

Differential Revision: https://reviews.llvm.org/D42739

llvm-svn: 324014

Remove obsolete measurements.

This check in requires at least 10.11

Reviewed: Jason Molenda, Jim Ingham

<rdar://problem/37047106> Xcode Memory gauge should show the jetsam ledger footprint rather than anonymous

llvm-svn: 324013

Summary:
Implement `MonotonicNanoTime` using `QueryPerformanceCounter`.

This function is used by Scudo & the 64-bit Primary allocator. Implementing it
now means that the release-to-OS mechanism of the Primary will kick in (it
never did since the function returned 0 always), but `ReleaseMemoryPagesToOS` is
still not currently implemented for Windows.

Performance wise, this adds a syscall & a 64-bit division per call to
`MonotonicNanoTime` so the impact might not be negligible, but I don't think
there is a way around it.

Reviewers: rnk, alekseyshl, amccarth

Reviewed By: alekseyshl, amccarth

Subscribers: amccarth, flowerhack, kubamracek, delcypher, llvm-commits, #sanitizers

Differential Revision: https://reviews.llvm.org/D42579

llvm-svn: 324011

Using the "run_to_{source,name}_breakpoint will allow us to remove
a lot of boiler-plate from the testsuite.  We mostly use source
breakpoints, but some tests use by name ones so this was needed.

llvm-svn: 324010

llvm-svn: 324008

This is a rather non-controversial change. We were missing these instructions
from the list of instructions that are lane-sensitive. These two put the result
into lane 0 (BE) or 3 (LE) regardless of the input. This patch fixes PR36068.

llvm-svn: 324005

(I suppose these two pieces could be separated - but seemed related
enough)

As discussed on llvm-dev, this documents the general expectation of how
library layering should be handled. There are a few existing cases where
these constraints are not met, but as with most style guide things -
this is forward looking and provides guidance when cleaning up existing
code, it doesn't immediately require that all previous code be cleaned
up to match. (see: naming conventions, etc)

Differential Revision: https://reviews.llvm.org/D42771

llvm-svn: 324004

llvm-svn: 324003

[DAGCombiner] When folding (insert_subvector undef, (bitcast (extract_subvector N1, Idx)), Idx) -> (bitcast N1) make sure that N1 has the same total size as the original output

We were only checking the element count, but not the total width. This could cause illegal bitcasts to be created if for example the output was 512-bits, but N1 is 256 bits, and the extraction size was 128-bits.

Fixes PR36199

Differential Revision: https://reviews.llvm.org/D42809

llvm-svn: 324002

Until we support extending loads properly we're going to fall back for these.
We already handle stores in the same way, so this is just being consistent.

llvm-svn: 324001

Increment the field list member count for base classes and virtual base
classes.

Differential Revision: https://reviews.llvm.org/D41874

llvm-svn: 324000

This is a bit faster in theory, in practice it's cold code that's only
active in !NDEBUG, so it probably doesn't make a difference. This is one
of the last users of our homegrown Atomic.h.

llvm-svn: 323999

DeclContext rather than injecting it wherever we happen to be.

This avoids creating functions whose DeclContext is a struct or similar.

llvm-svn: 323998

Summary:
Few changes to the secondary:
- mark `const` variables as such;
- change some `CHECK` to `DCHECK`: I don't feel we need to be as conservative as
  we were with out checks, as they are the results of our own computation.
- mark a condition as `UNLIKELY`.

Reviewers: alekseyshl

Reviewed By: alekseyshl

Subscribers: delcypher, #sanitizers, llvm-commits

Differential Revision: https://reviews.llvm.org/D42696

llvm-svn: 323997

I added this comment with D42323, but as discussed in D42806, the architecture
does the right thing for denorms. We don't even need the select on 0.0 here?

llvm-svn: 323996

Summary:
With the change, one can choose not to report comparison (or subtraction)
of a pointer with nullptr pointer.

Reviewers: kcc, jakubjelinek, alekseyshl

Reviewed By: alekseyshl

Subscribers: kubamracek

Differential Revision: https://reviews.llvm.org/D41479

llvm-svn: 323995

Summary:
D42698 adds child_edge_{begin|end} and children_edges to GraphTraits
which are used here. The reason for this change is to make it easy to
use count propagation on ModulesummaryIndex. As it stands,
CallGraphTraits is in Analysis while ModuleSummaryIndex is in IR.

Reviewers: davidxl, dberlin

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D42703

llvm-svn: 323994

This fixes pr36190.

Thanks to James Henderson for the testcase and for pointing out how to
fix this.

llvm-svn: 323993

Summary:
clangd drops diagnostics coming outside the main file, but it is still
useful to see that something went wrong in the logs.

Reviewers: hokein, ioeric, sammccall

Reviewed By: sammccall

Subscribers: klimek, jkorous-apple, cfe-commits

Differential Revision: https://reviews.llvm.org/D42803

llvm-svn: 323992

Summary:
This change extends MachineCopyPropagation to do COPY source forwarding
and adds an additional run of the pass to the default pass pipeline just
after register allocation.

This version of this patch uses the newly added
MachineOperand::isRenamable bit to avoid forwarding registers is such a
way as to violate constraints that aren't captured in the
Machine IR (e.g. ABI or ISA constraints).

This change is a continuation of the work started in D30751.

Reviewers: qcolombet, javed.absar, MatzeB, jonpa, tstellar

Subscribers: tpr, mgorny, mcrosier, nhaehnle, nemanjai, jyknight, hfinkel, arsenm, inouehrs, eraman, sdardis, guyblank, fedor.sergeev, aheejin, dschuff, jfb, myatsina, llvm-commits

Differential Revision: https://reviews.llvm.org/D41835

llvm-svn: 323991

Summary:
This change is mostly adding comments to GraphTraits describing
interfaces to iterate over children edges of a node. These will
have to be implemented by specializations of GraphTraits. The
non-comment change is the addition of children_edges template
function that returns an iterator range.

The motivation for this is to use it in synthetic count propagation
algorithm and remove the CallGraphTraits class that provide similar
interfaces.

Reviewers: dberlin, davidxl

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D42698

llvm-svn: 323990

Put the exception classes for experimental::optional and experimental::any back in the dylib for binary compatibility

llvm-svn: 323989

AMDGPU/SI: Adjust the encoding family for D16 buffer instructions when the target has UnpackedD16VMem feature.

Reviewers:
  Matt and Brian

Differential Revision:
  https://reviews.llvm.org/D42548

llvm-svn: 323988

This allows us to use PSHUFB for v8i16/v4i32 and VPERMD/PERMPS for v4i64/v4f64 variable shuffles.

Differential Revision: https://reviews.llvm.org/D42487

llvm-svn: 323987

llvm-svn: 323986

As noted in D42323, we're not checking for denorms as we should.

llvm-svn: 323985

llvm-svn: 323984