Doesn't really fix PR44491, but it avoids treating it as an input.

This reverts commit 921f871a because it
causes libc++ code to trigger __warn_memset_zero_len.

See https://reviews.llvm.org/D71082.

Make it clear form the method names whether they are synchronous or
asynchronous.

This reverts commit 23058f9d. It breaks
builds of cuda code somehow in some cases.

Specifically, add a space between >> when closing templates.

This change optimizes the operator=() assignment for short strings by direcly
copying the raw data from the source into the current instance. This creates an
optimized / inlined mempcy up to over 2X faster for short string assignments.
With inlining enabled for operator=, performance is up to 6X faster.

Benchmarks 'as is':
name                                    old time/op   new time/op    delta
BM_StringAssignStr_Empty_Opaque         6.05ns ± 2%   3.59ns ± 0%  -40.67%
BM_StringAssignStr_Empty_Transparent    5.15ns ± 0%   3.08ns ± 0%  -40.12%
BM_StringAssignStr_Small_Opaque         7.71ns ± 0%   3.59ns ± 0%  -53.45%
BM_StringAssignStr_Small_Transparent    7.66ns ± 0%   3.09ns ± 0%  -59.66%
BM_StringAssignStr_Large_Opaque         24.1ns ± 0%   24.9ns ± 0%   +3.22%
BM_StringAssignStr_Large_Transparent    22.2ns ± 0%   22.8ns ± 0%   +2.77%
BM_StringAssignStr_Huge_Opaque           315ns ± 6%    320ns ± 5%     ~
BM_StringAssignStr_Huge_Transparent      318ns ± 5%    321ns ± 4%     ~

Benchmarks with partial inlining operator=():
name                                    old time/op   new time/op    delta
BM_StringAssignStr_Empty_Opaque         5.94ns ± 2%   1.95ns ± 0%  -67.21%
BM_StringAssignStr_Empty_Transparent    5.14ns ± 0%   1.04ns ± 1%  -79.73%
BM_StringAssignStr_Small_Opaque         7.69ns ± 0%   1.96ns ± 0%  -74.48%
BM_StringAssignStr_Small_Transparent    7.65ns ± 0%   1.04ns ± 0%  -86.40%
BM_StringAssignStr_Large_Opaque         24.1ns ± 0%   24.5ns ± 0%   +1.61%
BM_StringAssignStr_Large_Transparent    22.2ns ± 0%   21.1ns ± 0%   -4.70%
BM_StringAssignStr_Huge_Opaque           317ns ± 5%    323ns ± 4%     ~
BM_StringAssignStr_Huge_Transparent      318ns ± 5%    320ns ± 5%     ~

Patch by Martijn Vels (mvels@google.com)
Reviewed as https://reviews.llvm.org/D72704

This should fix a failure on the clang-cmake-armv7-quick bot.

This cleanup was suggested by Djordje in D72489.

It appears to be rather useful when analyzing Loops with multiple
deoptimizing exits, perhaps merged ones.
For now it is used in LoopPredication, will be adding more uses
in other loop passes.

Reviewers: asbirlea, fhahn, skatkov, spatel, reames
Reviewed By: reames

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D72754

  The GCC build failures have been addressed, and the LLDB failures were
  fixed by LLDB.

   I have also verified that the apple-clang 9.0 segfault no longer
   occurs.

Original Message:

 The external instantiation of std::string is a problem for libc++.
    Additions and removals of inline functions in string can cause ABI
    breakages, including introducing new symbols.

    This patch aims to:
      (1) Make clear which functions are explicitly instatiated.
      (2) Prevent new functions from being accidentally instantiated.
      (3) Allow a migration path for adding or removing functions from the
      explicit instantiation over time.

    Although this new formulation is uglier, it is preferable from a
    maintainability and readability standpoint because it explicitly
    enumerates the functions we've chosen to expose in our ABI. Changing
    this list is non-trivial and requires thought and planning.

    (3) is achieved by making it possible to control the extern template declaration
    separately from it's definition. Meaning we could add a new definition to
    the dylib, wait for it to roll out, then add the extern template
    declaration to the header. Similarly, we could remove existing extern
    template declarations while still keeping the definition to prevent ABI
    breakages.

visibility.

This ensures that the version compiled into the library isn't
accidentally hidden.

This patch is needed in order to work around a GCC bug that fails to
explicitly instantiate a non-template function of a class template when
there is another overload that's a function template.
(See https://godbolt.org/z/4bUQ_b)

This patch SFINAE's away the function templates when the argument is
a basic_string.

See https://bugs.llvm.org/show_bug.cgi?id=44561, these tests are failing
on an aarch64/Linux bot:

http://lab.llvm.org:8011/builders/lldb-aarch64-ubuntu/builds/655

For some reason the backtrace the tests are expecting to find is
incomplete.

https://reviews.llvm.org/D72706 tried to reduce reordering due to mem op
clustering. This patch avoid doing the swap when we can't cluster.

Reviewed By: rampitec

Differential Revision: https://reviews.llvm.org/D72800

Summary:
InlineResult is used both in APIs assessing whether a call site is
inlinable (e.g. llvm::isInlineViable) as well as in the function
inlining utility (llvm::InlineFunction). It means slightly different
things (can/should inlining happen, vs did it happen), and the
implicit casting may introduce ambiguity (casting from 'false' in
InlineFunction will default a message about hight costs,
which is incorrect here).

The change renames the type to a more generic name, and disables
implicit constructors.

Reviewers: eraman, davidxl

Reviewed By: davidxl

Subscribers: kerbowa, arsenm, jvesely, nhaehnle, eraman, hiraditya, haicheng, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D72744

Summary:
* Add shaped container type interface which allows infering the shape, element
  type and attribute of shaped container type separately. Show usage by way of
  tensor type inference trait which combines the shape & element type in
  infering a tensor type;
  - All components need not be specified;
  - Attribute is added to allow for layout attribute that was previously
    discussed;
* Expand the test driver to make it easier to test new creation instances
  (adding new operands or ops with attributes or regions would trigger build
  functions/type inference methods);
  - The verification part will be moved out of the test and to verify method
    instead of ops implementing the type inference interface in a follow up;
* Add MLIRContext as arg to possible to create type for ops without arguments,
  region or location;
* Also move out the section in OpDefinitions doc to separate ShapeInference doc
  where the shape function requirements can be captured;
  - Part of this would move to the shape dialect and/or shape dialect ops be
    included as subsection of this doc;
* Update ODS's variable usage to match camelBack format for builder,
  state and arg variables;
  - I could have split this out, but I had to make some changes around
    these and the inconsistency bugged me :)

Differential Revision: https://reviews.llvm.org/D72432

Summary: Duplicate code in widenWithVariantLoadUseCodegen is removed and also use assert to check unknown extension type as it should be filtered out by the pre condition check before calling this function.

Reviewers: az, sanjoy, sebpop, efriedma, javed.absar, sanjoy.google

Reviewed By: efriedma

Subscribers: hiraditya, llvm-commits, amehsan

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D72652

PR17164: Change clang's default behavior from -flax-vector-conversions=all to -flax-vector-conversions=integer.

Summary:
See proposal on cfe-dev:
http://lists.llvm.org/pipermail/cfe-dev/2019-April/062030.html

Reviewers: SjoerdMeijer, eli.friedman

Subscribers: kristof.beyls, cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D67678

Work around PR43337: don't try to use the vec_sel overloads for vector long long, since clang's <altivec.h> doesn't provide it yet!

Factor out the logic needed to update debug locations contained within
MD_loop metadata.

This refactor is preparation for a future change that also needs to
rewrite MD_loop metadata.

rdar://45507940

This reverts D53469, which changed llvm's DWARF emission to emit
DW_AT_call_return_pc as a function-local offset. Such an encoding is not
compatible with post-link block re-ordering tools and isn't standards-
compliant.

In addition to reverting back to the original DW_AT_call_return_pc
encoding, teach lldb how to fix up DW_AT_call_return_pc when the address
comes from an object file pointed-to by a debug map. While doing this I
noticed that lldb's support for tail calls that cross a DSO/object file
boundary wasn't covered, so I added tests for that. This latter case
exercises the newly added return PC fixup.

The dsymutil changes in this patch were originally included in D49887:
the associated test should be sufficient to test DW_AT_call_return_pc
encoding purely on the llvm side.

Differential Revision: https://reviews.llvm.org/D72489

This updates the discussion of lazy reexports, fixes a TBD for a usage example,
and adds a reference to the fully worked lazy reexports example that was added
in e9e26c01.

I believe the generated code here can suffer from double rounding.
So I wanted to capture the existing codegen so we can make
decisions about how to fix it.

Based on Don Hinton's patch in https://reviews.llvm.org/D72406. This feature
was accidentally left out of e9e26c01, and
would have pessimized concurrent compilation in the default case.

Thanks for spotting this Don!

Revert "Revert rG6078f2fe - "[AArch64][GlobalISel]: Support @llvm.{return,frame}address selection.""

The original change wasn't constraining the operand regclasses which broke EXPENSIVE_CHECKS.

Ran `git ls-files '*.gn' '*.gni' | xargs llvm/utils/gn/gn.py format`.
The motivation is to reformat them with trunk gn again right after.
Trunk gn changed formatting of some single-element lists.

Flags are clang's default UI is flags.

We can have an env var in addition to that, but in D69825 nobody has yet
mentioned why this needs an env var, so omit it for now.  If someone
needs to set the flag via env var, the existing CCC_OVERRIDE_OPTIONS
mechanism works for it (set CCC_OVERRIDE_OPTIONS=+-fno-integrated-cc1
for example).

Also mention the cc1-in-process change in the release notes.

Also spruce up the test a bit so it actually tests something :)

Differential Revision: https://reviews.llvm.org/D72769

Summary: Add VMINAQ, VMINNMAQ, VMAXAQ, VMAXNMAQ intrinsics and unit tests.

Reviewers: simon_tatham, miyuki, dmgreen

Subscribers: kristof.beyls, hiraditya, cfe-commits, llvm-commits

Tags: #clang, #llvm

Differential Revision: https://reviews.llvm.org/D72761

This reverts commit 4624a1e8. Causing
problems downstream.

I noticed one bot failure due to
24a00ef2 because the wildcard matching
was not working as intended, fixed it to act similar to other checks of
CGSCCToFunctionPassAdaptor.

This patch imports constant variables even when they can't be internalized
(which results in promotion). This offers some extra constant folding
opportunities.

Differential revision: https://reviews.llvm.org/D70404

Summary:
Current peeling implementation bails out in case of loop nests.
The patch introduces a field in TargetTransformInfo structure that
certain targets can use to relax the constraints if it's
profitable (disabled by default).
Also additional option is added to enable peeling manually for
experimenting and testing purposes.

Reviewers: fhahn, lebedev.ri, xbolva00

Reviewed By: xbolva00

Subscribers: xbolva00, hiraditya, zzheng, llvm-commits

Differential Revision: https://reviews.llvm.org/D70304

As discussed in the motivating PR44509:
https://bugs.llvm.org/show_bug.cgi?id=44509

...we can end up with worse code using fast-math than without.
This is because the reassociate pass greedily transforms fsub
into fneg/fadd and apparently (based on the regression tests
seen here) expects instcombine to clean that up if it wasn't
profitable. But we were missing this fold:

(X - Y) - Z --> X - (Y + Z)

There's another, more specific case that I think we should
handle as shown in the "fake" fneg test (but missed with a real
fneg), but that's another patch. That may be tricky to get
right without conflicting with existing transforms for fneg.

Differential Revision: https://reviews.llvm.org/D72521

Summary:
This diff moves the conversion pass declaration closer to its definition
and makes the namespacing of passes consistent with the rest of the
infrastructure (i.e. `mlir::linalg::createXXXPass` -> `mlir::createXXXPass`).

Reviewers: ftynse, jpienaar, mehdi_amini

Subscribers: rriddle, burmako, shauheen, antiagainst, arpith-jacob, mgester, lucyrfox, aartbik, liufengdb, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D72766

This patch makes the target triple available via the LLJIT interface, and moves
the IRTransformLayer from LLLazyJIT down into LLJIT. Together these changes make
it easier to use the lazyReexports utility with LLJIT, and to apply IR
transforms to code as it is compiled in LLJIT (rather than requiring transforms
to be applied manually before code is added). An code example is added in
llvm/examples/LLJITExamples/LLJITWithLazyReexports