[llvm-dev] LLVM Weekly - #104, Dec 28th 2015

Alex Bradbury via llvm-dev llvm-dev at lists.llvm.org
Mon Dec 28 07:44:48 PST 2015


LLVM Weekly - #104, Dec 28th 2015
=================================

If you prefer, you can read a HTML version of this email at
<http://llvmweekly.org/issue/104>.

Welcome to the one hundred and fourth 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

The [schedule for the LLVM devroom at FOSDEM](http://llvm.org/devmtg/2016-01/)
has been published. This will be on January 30th 2016 in Brussels at
[FOSDEM](https://fosdem.org/2016/).

Andy Finnell spent some time over the Christmas vacation porting the LLVM
Kaleidoscope tutorial to Erlang and has kindly [shared the fruits of his
labours](http://losingfight.com/blog/2015/12/24/christmas-and-compilers/).

Richard Pennington has written another blog post about ELLCC, this time about
[using it to cross-compile the Linux kernel for the Raspberry
Pi](http://ellcc.org/blog/?p=26435).

Tim Jones (lecturer at the University of Cambridge Computer Laboratory) has
written about the [alias analysis used in the HELIX
compiler](http://www-dyn.cl.cam.ac.uk/~tmj32/wordpress/alias-analysis-in-helix/).
There's nothing LLVM-specific here, indeed it was implemented using ILDJIT but
should be of general interest to compiler developers.


## On the mailing lists

* Keno Fischer has posted a [proposal for multi-location debug info support in
LLVM IR](http://lists.llvm.org/pipermail/llvm-dev/2015-December/093535.html).
This would allow, for instance, modelling when a variable is available either
on the stack or in a register.

* Adam Nemet is proposing to [extend the PowerPC software prefetching pass to
work on other
targets](http://lists.llvm.org/pipermail/llvm-dev/2015-December/093514.html),
specifically AArch64.

* Russel Wallace asks for [advice on finding all pointers to
functions](http://lists.llvm.org/pipermail/llvm-dev/2015-December/093486.html),
and had a number of suggestions in response.

* For a while now, Galina Kistanova has been posting [statistics from the LLVM
buildbots](http://lists.llvm.org/pipermail/llvm-dev/2015-December/093503.html).
This includes the number of commits for each project, the number of failed
builds, and average build time. I haven't linked to it before, so felt I
should rectify that.

## LLVM commits

* An initial implementation of an LLVMCodeView library has landed. This
implements support for emitting debug info in the CodeView format.
[r256385](http://reviews.llvm.org/rL256385).

* lit has gained support for a per-test timeout which can be set using
`--timeout=`. [r256471](http://reviews.llvm.org/rL256471).

* All uses of edge wights in BranchProbabilityInfo have been replaced with
probabilities. [r256263](http://reviews.llvm.org/rL256263).

* The LLVM project documentation on patch reviews via Phabricator now has
advice on choosing reviewers. [r256265](http://reviews.llvm.org/rL256265).

* The gc.statepoint intrinsic's return type is now a token type rather than
i32. [r256443](http://reviews.llvm.org/rL256443).


## Clang commits

* ASTtemplateKWAndArgsInfo and ASTTemplateArgumentListInfo have been converted
to use the TrailingObjects header. This abstracts away `reinterpret_cast`,
pointer arithmetic, and size calculations needed for the case where a class
has some other objects appended to the end of it.
[r256359](http://reviews.llvm.org/rL256359).


## Other project commits

* Development of LLD's new ELF linker is continuing, with support for new
relocations on x86, x86-64, and MIPS.
[r256143](http://reviews.llvm.org/rL256143),
[r256144](http://reviews.llvm.org/rL256144),
[r256172](http://reviews.llvm.org/rL256172),
[r256416](http://reviews.llvm.org/rL256416).


More information about the llvm-dev mailing list