[llvm-dev] LLVM Weekly - #326, Mar 30th 2020

[Alex Bradbury via llvm-dev]

LLVM Weekly - #326, Mar 30th 2020
=================================

If you prefer, you can read a HTML version of this email at
<http://llvmweekly.org/issue/326>.

Welcome to the three hundred and twenty-sixth 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](https://www.linkedin.com/in/alex-bradbury/). 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

LLVM 10.0.0 [has been
released](http://lists.llvm.org/pipermail/llvm-dev/2020-March/140233.html). As
always, thank you to everyone who has contributed in any way!

## On the mailing lists

* Kyungwoo Lee wrote an RFC on [homogeneous prolog and epilog for size
optimisation in
AArch64](http://lists.llvm.org/pipermail/llvm-dev/2020-March/140206.html).
This is a code size optimisation that attempts to emit homogeneous stores and
loads with the same offset for prolog and epilog in order to increase the
benefit of machine outlining.

* Philip Reames started a discussion on [the status of Intel JCC
mitigations](http://lists.llvm.org/pipermail/llvm-dev/2020-March/140237.html)
and the desired next steps. As Philip explains, there is currently a
disagreement about how to move forwards.

* Anton Bikineev proposes [allowing custom sections to be under
`GNU_RELRO`](http://lists.llvm.org/pipermail/llvm-dev/2020-March/140348.html).
The motivation is to help security hardening applications that define custom
sections.

* Louis Dionne started a discussion on [bumping the CMake version requirement
for libc++ and
libc++abi](http://lists.llvm.org/pipermail/llvm-dev/2020-March/140239.html).
This triggered a lot of discussion.

* Bardia Mahjour shared [meeting minutes from the March 25th loop optimisation
working group
meeting](http://lists.llvm.org/pipermail/llvm-dev/2020-March/140283.html).

* "Whisperity" posted a proposal for [adding 'experimental' checks to
clang-tidy](http://lists.llvm.org/pipermail/cfe-dev/2020-March/064980.html).


## LLVM commits

* A new algorithm was added to the LLVM Support library to perform the
'optimal' layout of a struct, and it was put to use for coroutine frames.
[49e5a97](https://reviews.llvm.org/rG49e5a97ec36),
[9514c048](https://reviews.llvm.org/rG9514c048d89).

* MC layer (assembler/disassembler) support was added for Armv8.6-a BFloat16.
[71ae267](https://reviews.llvm.org/rG71ae267d1f4).

* GlobalISel gained new generic opcodes for saturating add/subtract.
[0444d16](https://reviews.llvm.org/rG0444d16a16c).

* The FREEZE node was added to the SelectionDAG.
[7802be4](https://reviews.llvm.org/rG7802be4a3d8).

* The use of `llvm_unreachable` was further clarified in the LLVM coding
standards. [4778e40](https://reviews.llvm.org/rG4778e409de1).

* Documentation was added on how to deal with issues of slow build time or
memory exhaustion when compiling LLVM.
[37943e5](https://reviews.llvm.org/rG37943e518c5).


## Clang commits

* An attribute plugin example was added.
[3f03c12](https://reviews.llvm.org/rG3f03c12a51b).

* Support was added for the `cmse_nonsecure_call` and `cmse_nonsecure_entry`
Arm CMSE (Cortex-M Security Extensions) attributes.
[080d046](https://reviews.llvm.org/rG080d046c91d).

* All Arm MVE intrinsics are now implemented.
[f282b6a](https://reviews.llvm.org/rGf282b6ab23a).


## Other project commits

* LLVM's libc gained a simple x86-64 Linux loader.
[f6ccb4f](https://reviews.llvm.org/rGf6ccb4fef24).

* `SignalContext::GetWriteFlag`, part of the sanitizer infrastructure, was
implemented for RISC-V. [ad1466f](https://reviews.llvm.org/rGad1466f8cbc).

* ThinLTO can now use all hardware threads on the system, rather than being
limited to one thread per core.
[0915825](https://reviews.llvm.org/rG09158252f77).

* Parallel loop collapsing was implemented in MLIR.
[27c201a](https://reviews.llvm.org/rG27c201aa1d9).