[llvm-dev] LLVM Weekly - #189, Aug 14th 2017

[Alex Bradbury via llvm-dev]

LLVM Weekly - #189, Aug 14th 2017
=================================

If you prefer, you can read a HTML version of this email at
<http://llvmweekly.org/issue/189>.

Welcome to the one hundred and eighty-ninth issue of LLVM Weekly, a weekly
newsletter (published every Monday) covering developments in LLVM, Clang, and
related projects. LLVM Weekly is brought to you by [Alex
Bradbury](http://asbradbury.org). Subscribe to future issues at
<http://llvmweekly.org> and pass it on to anyone else you think may be
interested. Please send any tips or feedback to <asb at asbradbury.org>, or
@llvmweekly or @asbradbury on Twitter.


## News and articles from around the web

If you haven't submitted your talk/tutorial/BoF/panel proposal for the
upcoming LLVM Dev Meeting, you [have a little more
time](http://lists.llvm.org/pipermail/llvm-dev/2017-August/116437.html). Get
your submission in by today, Monday 14th at 11:59PM PDT.

LLVM 5.0.0-rc2 has [been
tagged](http://lists.llvm.org/pipermail/llvm-dev/2017-August/116405.html).

The call for papers for the LLVM in HPC worksohp at SC17 is [still
open](http://lists.llvm.org/pipermail/llvm-dev/2017-August/116342.html).


## On the mailing lists

* Marc-André Laperle has given an
[update](http://lists.llvm.org/pipermail/cfe-dev/2017-August/054955.html) on
his prototype to implement indexing support in clangd.

* Brian Gesiak has posted an [RFC on expanding
-ftime-report](http://lists.llvm.org/pipermail/cfe-dev/2017-August/054964.html)
to give more information about how time was spent in tasks such as
preprocessing and template instantiation.

* Alex Bradbury has posted an RFC on [the future of the `va_arg`
instruction](http://lists.llvm.org/pipermail/llvm-dev/2017-August/116337.html)
in LLVM IR. This RFC hopes to explore where `va_arg` can be used for more
targets and be expanded to handle aggregates, as well as whether that's even a
desirable direction.

* Kuba Mracek has proposed [dropping the %T expansion in
lit](http://lists.llvm.org/pipermail/llvm-dev/2017-August/116320.html).


## LLVM commits

* Assembler support has been added for ARMv8.3-A pointer authentication
instructions. [r310709](http://reviews.llvm.org/rL310709).

* Assembler support has been added for the new release consistent and
processor consistent loads in ARMv8.3-A.
[r310575](http://reviews.llvm.org/rL310575).

* A basic RISCVAsmParser has landed.
[r310363](http://reviews.llvm.org/rL310363).

* The TargetTransformInfo::isFoldableMemAccess hook has been removed, as its
functionality is subsumed by isLegalAddressingMode.
[r310463](http://reviews.llvm.org/rL310463).

* The AArch64 and ARM backends gained assembler support for the ARMv8.2a dot
product instructions. (Question to my readers: Given Arm's recent rebranding,
should that be Armv8.2a or armv8.2a instead?).
[r310480](http://reviews.llvm.org/rL310480),
[r310701](http://reviews.llvm.org/rL310701).

* A new "Restored" flag has been added to CalleeSavedInfo.
[r310619](http://reviews.llvm.org/rL310619).

* Most AVX512 broadcast intrinsics have been removed and will be auto-upgraded
when loaded old IR. [r310725](http://reviews.llvm.org/rL310725).


## Clang commits

* The clang-proto-fuzzer tool has been created. This models a subset of C++ as
a protobuf and uses libprotobuf-mutator to generate mutations.
[r310408](http://reviews.llvm.org/rL310408).

* A couple of minor ABI differences between Clang and GCC related to move
constructors and deleted copy constructors have been fixed.
[r310401](http://reviews.llvm.org/rL310401).

* clang-tidy gained a new integer division check.
[r310589](http://reviews.llvm.org/rL310589).

* clang-import-test gained a `-dump-ir` argument, which can be used to help
test completions. [r310318](http://reviews.llvm.org/rL310318).


## Other project commits

* A new tool has been added to LLDB to support Intel Processor Trace.
[r310261](http://reviews.llvm.org/rL310261).

* AddressSanitizer now supports Fuchsia.
[r310431](http://reviews.llvm.org/rL310431).

* Polly is now run "before-vectorizer" rather than at the beginning of the
pass pipeline. The advantages of this move are explained in the commit
message. [r310319](http://reviews.llvm.org/rL310319).