[llvm-dev] LLVM Weekly - #100, Nov 30th 2015

Alex Bradbury via llvm-dev llvm-dev at lists.llvm.org
Mon Nov 30 04:36:42 PST 2015

LLVM Weekly - #100, Nov 30th 2015

If you prefer, you can read a HTML version of this email at

Welcome to the one hundredth 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.

Eagle-eyed readers will note we've now reached issue 100, marking 100 weeks of
uninterrupted service and of course meaning there's just 28 weeks to go until
an important numerical milestone.

## News and articles from around the web

There is going to be an LLVM Devroom at FOSDEM next year and the [call for
closes on December 1st. Get your submissions in!

Most slides from the recent LLVM in HPC workshop have [now been

Jeff Trull has posted a great [blog post on fuzzing C++ code with AFL and

## On the mailing lists

* The upcoming removal of the autoconf build system came up on the mailing
list again. Chris Bieneman explains [the policy on blocking vs non-blocking
bugs for
this](http://lists.llvm.org/pipermail/llvm-dev/2015-November/092841.html). If
you think you're likely to be affected, now is a very good time to kick the
tires on CMake.

* Geoffrey Romer is [interested in adding the ability to customise the
behaviour of
and is looking for feedback.

* Is it possible to use a static base register on ARM rather than PC-relative
addressing? Oliver Stannard [has a patch for
which should be upstreamed soon.

* Rail Shafigulin is looking for information on [how slots are assigned for
packets in
As usual, Krzystof Parzyszek provides some useful answers.

## LLVM commits

* A number of patches related to ARMv8.2-A have landed. Public documentation
doesn't seem to have been released for this architecture revision, but the
patches indicate some of the new features including: persistent memory
instruction and FP16 instructions. You can see the patches still in review

* A series of helper functions from SelectionDAGNodes have been exposed
(isNullConstant, isNullFPConstant, isAllOnesConstant, isOneConstant). These
helpers can help simplify code in your target's ISelLowering.

* The WebAssembly backend's block placement algorithm has been improved.

* Tests generated from `utils/update_llc_test_checks.py` are now marked as
autogenerated. [r253917](http://reviews.llvm.org/rL253917).

## Clang commits

* DataRecursiveASTVisitor has been removed, and RecursiveASTVisitor can be
used in its place. This resulted in the removal of 2912 lines of code.

* Sparc and SparcV9 default to using an external assembler again.

* Functions with the interrupt attribute are now supported for mips32r2+.

## Other project commits

* A single DataFlowSanitizer or ThreadSanitizer-instrumented binary can now
run on both 39-bit virtual address space and 42-bit virtual address space
AArch64 platforms. [r254151](http://reviews.llvm.org/rL254151),

* lldb gained a `swig_bot.py` for generating bindings.

More information about the llvm-dev mailing list