[PATCH] MC: Don't emit min version directives when -fno-integrated-as is on

[David Blaikie]

On Sat, Nov 22, 2014 at 11:28 AM, David Majnemer <david.majnemer at gmail.com>
wrote:

> >>! In D6371#7, @dblaikie wrote:
> >>>! In D6371#6, @majnemer wrote:
> >>>>! In D6371#4, @dblaikie wrote:
> >>> I believe we've been trying to avoid special casing the integrated
> >>> assembler as much as possible (I think Rafael removed many instances of
> >>> this) - any other signal that would be appropriate to use?
> >>
> >> This patch is working around a directive which the cctools assembler
> does not have.  Asking it to trigger on another condition would essentially
> be a proxy for "MC is targeting the cctools assembler" anyway.
> >
> > Could we do this down in the MCStreamer instead?
> >
> > It looks like MCAsmStreamer::EmitVersionMin should just be a no-op,
> maybe? (or a no-op on Darwin?
>
> How would we be able to emit the version information for Darwin if it were
> a no-op? I might be missing something but it sounds like .ios_version_min
> and .macosx_version_min would not properly round-trip.
>

The MCAsmStreamer is for printing textual assembly, as I understand it(?) -
and it's this case where want to not use these features, yes?

The MCObjectStreamer is the integrated assembler, that can continue to
support this feature - that said, I can't find any support for emitting
these directives in there... so I'm not sure where they're handled for the
integrated assembler case.


>
> > Is the version directive valid anywhere else other than LLVM's
> integrated assembler?)
>
> AFAIK, only the LLVM integrated assembler understands .ios_version_min and
> .macosx_version_min.  I've checked the cctools sources and the latest
> binutils GAS sources as well.
>
> http://reviews.llvm.org/D6371
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20141122/b3c933c1/attachment.html>