Subscribers: mgorny, llvm-commits

Differential Revision: https://reviews.llvm.org/D50528

llvm-svn: 339729

Add proper headers in CFGUpdate.h and add CFGDiff.h in the list of delayed headers for LLVM_intrinsic_gen.

Summary:
Fix module build after r339694.
Add headers needed in CFGUpdate.h.
Add CFGDiff.h in the list of delayed headers for LLVM_intrinsic_gen.
Up for post-commit review.

Subscribers: sanjoy, jlebar, llvm-commits

Differential Revision: https://reviews.llvm.org/D50734

llvm-svn: 339724

Clarifying the comment about PSCEV and external IV users by referencing
the bug in question.

llvm-svn: 339722

Flags in DIBasicType will be used to pass attributes used in
DW_TAG_base_type, such as DW_AT_endianity.

Patch by Chirag Patel!

Differential Revision: https://reviews.llvm.org/D49610

llvm-svn: 339714

See D50036.

llvm-svn: 339713

Modifies existing SIMD tests to also check that SIMD instructions are
lowered to the expected bytes. This CL depends on D50597.

Reviewers: aheejin

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

Differential Revision: https://reviews.llvm.org/D50660

Patch by Thomas Lively (tlively)

llvm-svn: 339712

D50036 should fix the missed optimizations.

llvm-svn: 339711

Previously SIMD_I was the same as a normal instruction except for the
addition of a HasSIM128 predicate. However, rL339186 changed the
encoding of SIMD_I instructions to automatically contain the SIMD
prefix byte. This broke the encoding of non-SIMD vector-typed
instructions, which had instantiated SIMD_I. This CL corrects this
error.

Reviewers: aheejin

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

Differential Revision: https://reviews.llvm.org/D50682

Patch by Thomas Lively (tlively)

llvm-svn: 339710

1) We print __restrict twice on member pointers.  This is fixed
   and relevant tests are re-enabled.

2) Several tests were disabled because of printing slightly
   different output than undname.  These were confirmed to be
   bugs in undname, so we just re-enable the tests.

3) The test for printing reference temporaries is re-enabled.  This
   is a clang mangling extension, so we have some flexibility with
   how we demangle it.  The output currently looks fine, so we just
   re-enable the test with no fixes.

llvm-svn: 339708

Implement instruction selection for all versions of the extract_lane
instruction. Use explicit sext/zext to differentiate between
extract_lane_s and extract_lane_u for applicable types, otherwise
default to extract_lane_u.

Reviewers: aheejin

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

Differential Revision: https://reviews.llvm.org/D50597

Patch by Thomas Lively (tlively)

llvm-svn: 339707

[Tablegen][MCInstPredicate] Removed redundant template argument from class TIIPredicate, and implemented verification rules for TIIPredicates.

This patch removes redundant template argument `TargetName` from TIIPredicate.
Tablegen can always infer the target name from the context. So we don't need to
force users of TIIPredicate to always specify it.

This allows us to better modularize the tablegen class hierarchy for the
so-called "function predicates". class FunctionPredicateBase has been added; it
is currently used as a building block for TIIPredicates. However, I plan to
reuse that class to model other function predicate classes too (i.e. not just
TIIPredicates). For example, this can be a first step towards implementing
proper support for dependency breaking instructions in tablegen.

This patch also adds a verification step on TIIPredicates in tablegen.
We cannot have multiple TIIPredicates with the same name. Otherwise, this will
cause build errors later on, when tablegen'd .inc files are included by cpp
files and then compiled.

Differential Revision: https://reviews.llvm.org/D50708

llvm-svn: 339706

Summary:
This patch teaches the loop vectorizer to vectorize loops with non
header phis that have have outside uses.  This is because the iteration
dependence distance for these phis can be widened upto VF (similar to
how we do for induction/reduction) if they do not have a cyclic
dependence with header phis. When identifying reduction/induction/first
order recurrence header phis, we already identify if there are any cyclic
dependencies that prevents vectorization.

The vectorizer is taught to extract the last element from the vectorized
phi and update the scalar loop exit block phi to contain this extracted
element from the vector loop.

This patch can be extended to vectorize loops where instructions other
than phis have outside uses.

Reviewers: Ayal, mkuper, mssimpso, efriedma

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D50579

llvm-svn: 339703

This reverts commit cb8c5e417d55141f3f079a8a876e786f44308336 / r339676.

This causing a test to fail in http://green.lab.llvm.org/green/job/clang-stage1-configure-RA/48406/

    LLVM :: DebugInfo/Generic/debug-label.ll

llvm-svn: 339700

Summary:
Treating a graph in reverse is a property of the GraphDiff and should instead be a template argument, just like IsPostDom is one for DomTrees.
If it's just an argument to all methods, we could have mismatches between the constructor of the GraphDiff which may reverse the updates when filtering them, and the calls retrieving the filtered delete/insert updates.
Also, since this will be used in IDF, where we're using a DomTree, this creates a cleaner interface for the GraphTraits to use the existing template argument of DomTreeBase.

Separate patch from the one adding GraphDiff, so get a clear diff of what changed.

Reviewers: timshen, kuhar

Subscribers: sanjoy, llvm-commits, jlebar

Differential Revision: https://reviews.llvm.org/D50687

llvm-svn: 339699

rL339686 added the case where a faux shuffle might have repeated shuffle inputs coming from either side of the OR().

This patch improves the insertion of the inputs into the source ops lists to account for this, as well as making it trivial to add support for shuffles with more than 2 inputs in the future.

llvm-svn: 339696

Summary:
Clean-up following D50479.
Make Update and LegalizeUpdate refer to the utilities in Support/CFGUpdate.

Reviewers: kuhar

Subscribers: sanjoy, jlebar, mgrang, llvm-commits

Differential Revision: https://reviews.llvm.org/D50669

llvm-svn: 339694

llvm-svn: 339692

Summary:
Certain passes or analysis need to view a CFG snapshot rather than the actual CFG. This patch provides GraphTraits to offer such a view.

The patch defines GraphTraits for BasicBlock* and Inverse<BasicBlock*> to provide CFG successors and predecessors based on a list of CFG updates.

An Update is defined as a triple {InsertOrDeleteKind, BlockStartOfEdge, BlockEndOfEdge}.
A GraphDiff is defined as a list of Updates that has been preprocessed to treat the CFG as a graph rather than a multi-graph. As such, there can only exist a single Update given two nodes. All duplicates will be filtered and Insert/Delete edges that cancel out will be ignored.
The methods GraphDiff exposes are:
- Determine if an existing child needs to be ignored, i.e. an Update exists in the correct direction to assume the removal of that edge.
- Return a list of new children to be considered, i.e. an Update exists in the correct direction for each child in the list to assume the insertion of that edge.

Reviewers: timshen, kuhar, chandlerc

Subscribers: sanjoy, jlebar, llvm-commits

Differential Revision: https://reviews.llvm.org/D50479

llvm-svn: 339689

Patch by Henric Karlsson.

llvm-svn: 339688

If each element is zero from one (or both) inputs then we can combine these into a single shuffle mask.

llvm-svn: 339686

We generate these shuffle patterns but we fail to combine them.

llvm-svn: 339684

llvm-svn: 339683

Increment existing NumInlined and NumDeleted stats in InlinerPass::run.

llvm-svn: 339682

llvm-svn: 339681

We still only use this for v16i8, but this cleans up the code to support v32i8/v64i8 sometime in the future.

llvm-svn: 339679

There are two forms for label debug information in DWARF format.

1. Labels in a non-inlined function:

DW_TAG_label
  DW_AT_name
  DW_AT_decl_file
  DW_AT_decl_line
  DW_AT_low_pc

2. Labels in an inlined function:

DW_TAG_label
  DW_AT_abstract_origin
  DW_AT_low_pc

We will collect label information from DBG_LABEL. Before every DBG_LABEL,
we will generate a temporary symbol to denote the location of the label.
The symbol could be used to get DW_AT_low_pc afterwards. So, we create a
mapping between 'inlined label' and DBG_LABEL MachineInstr in DebugHandlerBase.
The DBG_LABEL in the mapping is used to query the symbol before it.

The AbstractLabels in DwarfCompileUnit is used to process labels in inlined
functions.

We also keep a mapping between scope and labels in DwarfFile to help to
generate correct tree structure of DIEs.

It also generates label debug information under global isel.

Differential Revision: https://reviews.llvm.org/D45556

llvm-svn: 339676

Differential Revision: https://reviews.llvm.org/D49442

llvm-svn: 339674

llvm-svn: 339670

MSVC only accepts if-else chains up to 127 blocks long. I've had to merge a number of intrinsic cases together to get back below this limit, resulting in some duplication of string matches; this shouldn't cause any notable increase in runtime (and even then only for old IR, nothing that clang currently emits).

llvm-svn: 339666

Summary: This adds constant folding of signed add/sub with saturation intrinsics.

Reviewers: craig.topper, spatel, RKSimon, chandlerc, efriedma

Reviewed By: craig.topper

Subscribers: rnk, llvm-commits

Differential Revision: https://reviews.llvm.org/D50499

llvm-svn: 339659

This is a fix for r339314.

MCInstBuilder uses the named parameter idiom and an 'operator MCInst&' to ease
the creation of MCInsts. As the object of MCInstBuilder owns the MCInst is
manipulating, the lifetime of the MCInst is bound to that of MCInstBuilder.

In r339314 I bound a reference to the MCInst in an initializer. The
temporary of MCInstBuilder (and also its MCInst) is destroyed at the end of
the declaration leading to a dangling reference.

Fix this by using MCInstBuilder inside an argument of a function call.
Temporaries in function calls are destroyed in the enclosing full expression,
so the the reference to MCInst is still valid when emitToStreamer executes.

llvm-svn: 339654

llvm-svn: 339652

Summary: This revision improves previous version (rL330322) which has been reverted due to crashes.

This is the patch that lowers x86 intrinsics to native IR
in order to enable optimizations. The patch also includes folding
of previously missing saturation patterns so that IR emits the same
machine instructions as the intrinsics.

Reviewers: craig.topper, spatel, RKSimon

Reviewed By: craig.topper

Subscribers: mike.dvoretsky, DavidKreitzer, sroland, llvm-commits

Differential Revision: https://reviews.llvm.org/D46179

llvm-svn: 339650

Differential Revision: https://reviews.llvm.org/D50511

llvm-svn: 339645

llvm-svn: 339644

llvm-svn: 339643

Summary:
When WPD is performed in a ThinLTO backend, the function may be created
if it isn't already in that module. Module::getOrInsertFunction may
add a bitcast, in which case the returned Constant is not a Function and
doesn't have a name. Invoke stripPointerCasts() on the returned value
where we access its name.

Reviewers: pcc

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

Differential Revision: https://reviews.llvm.org/D49959

llvm-svn: 339640

llvm-strip doesn't handle -gx spelling, so we need to split these
as two separate flags.

Differential Revision: https://reviews.llvm.org/D50684

llvm-svn: 339639

Summary:
The AsmWriter was only writing the Args for a ConstVCall if it was
non-empty, however, the LLParser was always expecting it. To aid
in making it optional, surround the ConstVCall VFuncId and Args in
parentheses when writing, then make the Args optional when reading.

Reviewers: pcc

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

Differential Revision: https://reviews.llvm.org/D49960

llvm-svn: 339637

Summary:
Calls marked 'tail' cannot read or write allocas from the current frame
because the current frame might be destroyed by the time they run.
However, a tail call may use an alloca with byval. Calling with byval
copies the contents of the alloca into argument registers or stack
slots, so there is no lifetime issue. Tail calls never modify allocas,
so we can return just ModRefInfo::Ref.

Fixes PR38466, a longstanding bug.

Reviewers: hfinkel, nlewycky, gbiv, george.burgess.iv

Subscribers: hiraditya, llvm-commits

Differential Revision: https://reviews.llvm.org/D50679

llvm-svn: 339636