llvm-svn: 251367

instructions that aren't relevant for instruction selection of vector
min and max.

llvm-svn: 251366

http://reviews.llvm.org/D14095

llvm-svn: 251365

llvm-svn: 251364

LLVMSymbolizer::Options is mostly used in LLVMSymbolizer class anyway.
Let's keep their usage restricted to that class, especially given that
it's worth to move ModuleInfo to a different header, independent from
the symbolizer class.

llvm-svn: 251363

llvm-svn: 251362

The test is sensitive to stack layout changes. Tweak it a bit to
expose the bug on aarch64 as well as on arm.

llvm-svn: 251361

We've switched to Gold earlier because of a minor misconfiguration
of the BFD linker in Android NDK. It turns out, Gold has much bigger
problems:

https://sourceware.org/bugzilla/show_bug.cgi?id=19163
(a bug is actually in the android runtime loader, but it means that
gold does not work with android L and even M).

Switching back to BFD and adding a workaround by explicitly linking
libm to all tests.

llvm-svn: 251360

Asanwrapper is required on older android versions to work around undesired
linker behavior. It is not required on L and newer, and does not fully
support multiarch devices.

llvm-svn: 251359

Summary:
This check flags all array to pointer decays.

Pointers should not be used as arrays. array_view is a bounds-checked,
safe alternative to using pointers to access arrays.

This rule is part of the "Bounds safety" profile of the C++ Core
Guidelines, see
https://github.com/isocpp/CppCoreGuidelines/blob/master/CppCoreGuidelines.md#-bounds3-no-array-to-pointer-decay

Reviewers: alexfh, sbenza, bkramer, aaron.ballman

Subscribers: cfe-commits

Differential Revision: http://reviews.llvm.org/D13640

llvm-svn: 251358

This is a preliminary step before adding another optimization
to PerformBITCASTCombine().

..and I really hope it's NFC this time!

llvm-svn: 251357

llvm-svn: 251356

Summary: Adapt clang-tidy/add_new_check.py according to commit r251010 "Add %check_clang_tidy and %clang_tidy_diff"

Reviewers: alexfh, sbenza, bkramer, aaron.ballman

Subscribers: cfe-commits

Differential Revision: http://reviews.llvm.org/D14049

llvm-svn: 251355

This reverts commit r251340.

Breaks the Windows build because Windows doesn't have getuid.  The fix is not obvious.

llvm-svn: 251354

Summary:
In particular, this CL speeds up the official Chrome linking with LTO by
1.8x.

See more details in https://crbug.com/542426

Reviewers: dblaikie

Subscribers: jevinskie

Differential Revision: http://reviews.llvm.org/D13918

llvm-svn: 251353

Both VLDRS and VLDRD fault if the memory is not 4 byte aligned, which wasn't
really being checked before, leading to faults at runtime.

llvm-svn: 251352

Summary:
This does not have any functionnal difference but I found myself
searching for 'struct RegisterInfo$' to find a definition and couldn't
find it. The version without typedef seems more common so let's try and
be consistent.

Reviewers: clayborg, zturner

Subscribers: lldb-commits

Differential Revision: http://reviews.llvm.org/D14091

llvm-svn: 251351

llvm-svn: 251350

This is a preliminary step before adding another optimization
to PerformBITCASTCombine().

llvm-svn: 251349

Summary: This idiom is used elsewhere in LLVM, but was overlooked here.

Reviewers: chandlerc

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D13628

llvm-svn: 251348

llvm-svn: 251347

llvm-svn: 251346

The check for the glibc version was not working as expected (dlsym
was finding the interceptor instead of the libc implementation).

llvm-svn: 251345

llvm-svn: 251344

llvm-svn: 251343

Unbreaks linking with gold, which cannot resolve direct relocations referring
to global symbols.

llvm-svn: 251342

These maps are only needed during the construction of a single region statement.
Clearing them is important, as we otherwise get an assert in case some of the
referenced values are erased before the RegionGenerator is deleted.

llvm-svn: 251341

On UNIX (but not Darwin) the username needs to be respected when creating a 
temporary module directory, so that different users don't pollute each others'
module caches.

llvm-svn: 251340

Now it's enough to just specify -functions=short without additionally
providing -use-symbol-table=false.

llvm-svn: 251339

llvm-svn: 251338

This is a patch to improve StringTableBuilder's performance. That class'
finalize function is very hot particularly in LLD because the function
does tail-merge strings in string tables or SHF_MERGE sections.

Generic std::sort-style sorter is not efficient for sorting strings.
The function implemented in this patch seems to be more efficient.

Here's a benchmark of LLD to link Clang with or without this patch.
The numbers are medians of 50 runs.

-O0
real 0m0.455s
real 0m0.430s (5.5% faster)

-O3
real 0m0.487s
real 0m0.452s (7.2% faster)

Since that is a benchmark of the whole linker, the speedup of
StringTableBuilder itself is much more than that.

http://reviews.llvm.org/D14053

llvm-svn: 251337

llvm-svn: 251336

Summary:
In `MismatchingNewDeleteDetector::analyzeInClassInitializer`, if
`Field`'s initializer expression is null, lookup the field in
implicit instantiation, and use found field's the initializer.

Reviewers: rsmith, rtrieu

Subscribers: cfe-commits

Differential Revision: http://reviews.llvm.org/D9898

llvm-svn: 251335

llvm-svn: 251334

We should remove noalias along with dereference and dereference_or_null attributes 
because statepoint could potentially touch the entire heap including noalias objects.

Differential Revision: http://reviews.llvm.org/D14032

llvm-svn: 251333

They are aliases to --start-group and --end-group, respectively.

llvm-svn: 251332

This patch fixes the ptrace interceptor for aarch64. The PTRACE_GETREGSET
ptrace syscall with with invalid memory might zero the iovec::iov_base
field and then masking the subsequent check after the syscall (since it
will be 0 and it will not trigger an invalid access). The fix is to copy
the value on a local variable and use its value on the checks.

The patch also adds more coverage on the Linux/ptrace.cc testcase by addding
check for PTRACE_GETREGSET for both general and floating registers (aarch64
definitions added only).

llvm-svn: 251331

This adds a couple of optimization remarks to the SamplePGO
transformation. When it decides to inline a hot function (to mimic the
inline stack and repeat useful inline decisions in the original build).

It will also report branch destinations. For instance, given the code
fragment:

     6      if (i < 1000)
     7        sum -= i;
     8      else
     9        sum += -i * rand();

If the 'else' branch is taken most of the time, building this code with
-Rpass=sample-profile will produce:

a.cc:9:14: remark: most popular destination for conditional branches at small.cc:6:9 [-Rpass=sample-profile]
      sum += -i * rand();
             ^

llvm-svn: 251330

llvm-svn: 251329

Python 3 has a different syntax for octal literals than Python 2
and they are incompatible with each other.  Six doesn't provide
a transparent wrapper around this, so the most sane thing to do
is to not use octal literals.  If you need an octal literal,
use a decimal literal and if it's not obvious what the value is,
provide the value in octal as a comment.

llvm-svn: 251328