[LLVMdev] Proposal: Adding aligned instruction bundle support to MC

Greg Fitzgerald garious at gmail.com
Thu Dec 6 11:59:01 PST 2012


Maybe I'm jumping the gun, but some initial thoughts after browsing the MC
layer changes:

* .bundle_align_start  -> .bundle_align_mode N  ?
* .bundle_align_end -> .bundle_align_mode 0  ?
* Add unit test showing .bundle_lock/unlock can be nested.

I like the way you modified the existing ARM tests to show how the existing
instructions are predicated.  Overall, I think the concept of Native Client
is well-documented and easy to follow.  Has anyone done work to show that
assembly in this form can be disassembled to a memory-safe assembly
language like TAL?


> groups of instructions between .bundle_lock and
> .bundle_unlock directives cannot cross a bundle boundary

Can this be relaxed to:  A data instruction cannot fall on a bundle
boundary.  ?


Thanks,
Greg


On Wed, Dec 5, 2012 at 7:22 PM, Eli Bendersky <eliben at google.com> wrote:

> On Wed, Dec 5, 2012 at 6:01 PM, Greg Fitzgerald <garious at gmail.com> wrote:
> >> But if you just want to see the code and the diff against upstream...
> >
> > If others are interested as well, here's all your changes via Github's
> > "compare" feature:
> >
> > https://github.com/garious/llvm/compare/master...chromium
> >
> > Fascinating stuff guys.  What a great contribution.
> >
>
> Thanks for the feedback, and any help reviewing future patches will be
> most appreciated :-)
> It's also important to state that we plan to significantly clean-up
> and refactor our code prior to upstreaming. Since in some areas the
> changes are significant, this also means some refactoring of existing
> LLVM code. My recent patches and commits in MC are the beginning of
> this effort.
>
> Eli
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20121206/b99ad357/attachment.html>


More information about the llvm-dev mailing list