[PATCH] Add HLE target feature

[Jeffrey Yasskin]

On Wed, Feb 20, 2013 at 5:34 PM, Michael Liao <michael.liao at intel.com> wrote:
> Hi Jeffery
>
> Thanks for reminder. I will add documentation soon. So far, at IR level,
> the only change is adding a metadata (!targetflags <N>) through the
> patch. The flag word will be defined by target and not a generic one.
>
> Target-specific flag should be helpful for other architectures, such as
> Power. The similar HLE hint could be used by target backend to replace
> regular atomic builtin with a HLE version if they have efficient TM
> support.

The patch to the langref is pretty much the most important part of
this patch series, so I hope you send it out soon. I don't think it's
possible to review the rest of the patches without it. Please cc at
least me and Chandler. Eli Friedman did the initial implementation of
the atomic instructions, so he might be interested too.

I don't think !targetflags <N> is appropriate for this feature since
we don't have a large number of examples of other target-specific
boolean modifications to instructions. I think you should propose a
particular metadata node and define its meaning in the context of the
IR. Even if you used !targetflags <N>, you'd *still* need to create a
section of the langref where you defined the meanings of each <N> for
each target, so you're not saving any work by preferring the more
obscure syntax.

Jeffrey