llvm-svn: 353477

llvm-svn: 353476

Summary:
From runtime side looks it's OK to RoundUpTo to needed alignment as buffer is
going to be RoundUpTo to page size anyway.

Reviewers: eugenis, pcc

Subscribers: #sanitizers

Tags: #sanitizers

Differential Revision: https://reviews.llvm.org/D57866

llvm-svn: 353475

[WebAssembly] Fix imported function symbol names that differ from their import names in the .o format

Add a flag to allow symbols to have a wasm import name which differs from the
linker symbol name, allowing the linker to link code using the import_module
attribute.

This is the MC/Object portion of the patch.

Differential Revision: https://reviews.llvm.org/D57632

llvm-svn: 353474

[WebAssembly] Fix imported function symbol names that differ from their import names in the .o format

Add a flag to allow symbols to have a wasm import name which differs from the
linker symbol name, allowing the linker to link code using the import_module
attribute.

Differential Revision: https://reviews.llvm.org/D57632

llvm-svn: 353473

We save two levels of indentation by returning early if the given file
doesn't exists or cannot be opened.

llvm-svn: 353472

This commit teaches InstCombine how to replace an atomicrmw operation
into a simple load atomic.
For a given `atomicrmw <op>`, this is possible when:
1. The ordering of that operation is compatible with a load (i.e.,
   anything that doesn't have a release semantic).
2. <op> does not modify the value being stored

Differential Revision: https://reviews.llvm.org/D57854

llvm-svn: 353471

Mostly achieved by assuming that anything that isn't Win or Mac is ELF,
which seems reasonable enough for now.

Differential Revision: https://reviews.llvm.org/D57870

llvm-svn: 353470

llvm-svn: 353469

llvm-svn: 353468

This fixes a class of bugs introduced by D44367, 
which transforms various cases of icmp (bitcast ([su]itofp X)), Y to icmp X, Y. 
If the bitcast is between vector types with a different number of elements, 
the current code will produce bad IR along the lines of: icmp <N x i32> ..., <M x i32> <...>.

This patch suppresses the transform if the bitcast changes the number of vector elements.

Patch by: @AndrewScheidecker (Andrew Scheidecker)

Differential Revision: https://reviews.llvm.org/D57871

llvm-svn: 353467

This broke modularized non-local-submodule-visibility builds because
the function bodies pulled in extra dependencies.

llvm-svn: 353465

This is part of https://bugs.llvm.org/show_bug.cgi?id=40442.

Vector legalization is implemented for the add/sub overflow opcodes.
UMULO/SMULO are also handled as far as legalization is concerned, but
they don't support vector expansion yet (so no tests for them).

The vector result widening implementation is suboptimal, because it
could result in a legalization loop.

Differential Revision: https://reviews.llvm.org/D57639

llvm-svn: 353464

We should propagate this down to host builds so that e.g. people using
an optimized tablegen can do the sub-configure successfully.

llvm-svn: 353463

llvm-svn: 353462

As discussed in D57382, interleaving should be avoided if computeMaxVF
returns None, same as we currently do for vectorization.

Fixes https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=6477

Reviewers: Ayal, dcaballe, hsaito, mkuper, rengolin

Reviewed By: Ayal

Differential Revision: https://reviews.llvm.org/D57837

llvm-svn: 353461

Don't rely on order of evaluation of function arguments.

llvm-svn: 353460

This patch fixes a bug where clang doesn’t reject union fields of
non-trivial C struct types. For example:

```
// This struct is non-trivial under ARC.
struct S0 {
  id x;
};

union U0 {
  struct S0 s0; // clang should reject this.
  struct S0 s1; // clang should reject this.
};

void test(union U0 a) {
  // Previously, both 'a.s0.x' and 'a.s1.x' were released in this
  // function.
}
```

rdar://problem/46677858

Differential Revision: https://reviews.llvm.org/D55659

llvm-svn: 353459

Move the (add (umax X, C), -C) --> (usubsat X, C) X86 combine into generic DAGCombiner

First of a number of saturated arithmetic folds that can be moved out of X86-specific code for PR40111.

Differential Revision: https://reviews.llvm.org/D57754

llvm-svn: 353457

Summary:
Deferred diagnostic interface is going to be used for OpenMP device
compilation. Generalized previously existed deferred diagnostic
interface for CUDA to be used with OpenMP and, possibly, other models.

Reviewers: rjmccall, tra

Subscribers: caomhin, cfe-commits, kkwli0

Tags: #clang

Differential Revision: https://reviews.llvm.org/D57908

llvm-svn: 353456

This is pretty much directly ported from SelectionDAG. Doesn't include
the shift by non-constant but known bits version, since there isn't a
globalisel version of computeKnownBits yet.

This shows a disadvantage of targets not specifically which type
should be used for the shift amount. If type 0 is legalized before
type 1, the operations on the shift amount type use the wider type
(which are also less likely to legalize). This can be avoided by
targets specifying legalization actions on type 1 earlier than for
type 0.

llvm-svn: 353455

llvm-svn: 353452

Differential Revision: https://reviews.llvm.org/D57913

llvm-svn: 353451

Add static_asserts to tuple's comparison operators to enforce the requirement that the tuples be the same size. See PR39183 for an example where we give unexpected results for this bad input case. With this change, we will reject it at compile-time

llvm-svn: 353450

Since G_CONSTANT is illegal for vectors, this needs to check
what buildConstant will produce for a splat vector.

llvm-svn: 353449

Add UBSAN annotation to __hash_table::rehash; we don't do anything wrong, but UBSAN's checker flags it as suspicious. See PR38606. NFC

llvm-svn: 353448

llvm-svn: 353447

Summary:
This commit contains the following changes:

  - Rewrite vfile close/read/write packet handlers with portable routines from lldb.
    This removes #if(s) and allows the handlers to work on Windows.

  - Fix a bug in File::Write. This is intended to write data at an offset to a file
    but actually writes at the current position of the file.

  - Add a default boolean argument 'should_close_fd' to FileSystem::Open to
    let the user decide whether to close the fd or not.

Reviewers: zturner, llvm-commits, labath

Reviewed By: zturner

Subscribers: Hui, labath, abidh, lldb-commits

Differential Revision: https://reviews.llvm.org/D56231

llvm-svn: 353446

llvm-svn: 353444

llvm-svn: 353443

Causes ASAN use-after-poison errors.

llvm-svn: 353442

Reviewers: zturner, llvm-commits, labath, serge-sans-paille

Reviewed By: labath

Subscribers: Hui, labath, lldb-commits

Differential Revision: https://reviews.llvm.org/D56230

llvm-svn: 353440

llvm-svn: 353439

llvm-svn: 353438

R_X86_64_PC{8,16} relocations are sign-extended, so when we check
for relocation overflow, we had to use checkInt instead of checkUInt.
I confirmed that GNU linkers create the same output for the test case.

llvm-svn: 353437

llvm-svn: 353436

Summary:
Before this change, check-profile would run, but all tests would be
marked unsupported on Windows. This is the new status of 'check-profile'
after this change:

Testing Time: 6.66s
  Expected Passes    : 29
  Expected Failures  : 5
  Unsupported Tests  : 39

I moved many tests that exercise posix-y features like dlopen and DSOs
into the Posix subdirectory, and ran the tests on Linux to validate my
changes.

These are the remaining tests that I handled on a case by case basis:
- instrprof-path.c
  Passes, Fixed some path portability issues
- instrprof-gcov-exceptions.test
  Passes, the FileCheck actually succeeds on Windows, so I RUNX'd it
- instrprof-icall-promo.test
  XFAILed, probably due to C++ ABI differences in vtables
- instrprof-merge-match.test
- instrprof-merge.c
- instrprof-merging.cpp
  XFAILed, These seem like real bugs that need fixing
- instrprof-version-mismatch.c
  XFAILed, Overriding the weak version symbol doesn't work
- instrprof-without-libc.c
  UNSUPPORTED, test needs an executable symbol table, Windows has none

Reviewers: davidxl, wmi, void

Subscribers: fedor.sergeev, #sanitizers, llvm-commits

Tags: #sanitizers

Differential Revision: https://reviews.llvm.org/D57853

llvm-svn: 353435

Summary:
When compiling with profile data, ensure the split cold function gets
cold function_entry_count metadata (just use 0 since it should be cold).
Otherwise with function sections it will not be placed in the unlikely
text section with other cold code.

Reviewers: vsk

Subscribers: sebpop, hiraditya, davidxl, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D57900

llvm-svn: 353434

  
I noticed that we are missing this canonicalization in IR:
rL352515
...and then realized that we don't get this right in SDAG either,
so this has to be fixed first regardless of what we choose to do in IR.

The existing fold was limited to scalars and using the wrong predicate
to guard the transform. We have a boolean contents TLI query that can
be used to decide which direction to fold.

This may eventually lead back to the problems/question in:
https://bugs.llvm.org/show_bug.cgi?id=40486
...but it makes no difference to that yet.

Differential Revision: https://reviews.llvm.org/D57401

llvm-svn: 353433

Introduce a new function which handles instructions with multiple type
indices, but have the same number of vector elements.

Also legalize v2s16 shifts when applicable.

llvm-svn: 353432