r206081 - CodeGen: Fix handling of C++11 lambdas in profiling

[Keith Walker]

Fix to optionally expect zeroext on the matching line committed as r206467.

Keith

> -----Original Message-----
> From: Justin Bogner [mailto:justin at justinbogner.com] On Behalf Of
> Justin Bogner
> Sent: 15 April 2014 21:21
> To: Keith Walker
> Cc: cfe-commits at cs.uiuc.edu
> Subject: Re: r206081 - CodeGen: Fix handling of C++11 lambdas in
> profiling
> 
> "Keith Walker" <kwalker at arm.com> writes:
> > Justin,
> >
> > If the default clang target is changed to aarch64-arm-none-eabi then
> the new
> > test cxx-lambda.cpp breaks.
> >
> > This is because the "expected" line:
> >    %call = call zeroext i1 @"_ZZ7lambdasvENK3$_0clEi"(%class.anon*
> %f, i32
> > %sub)
> > Instead comes out as:
> >    %call = call i1 @"_ZZ7lambdasvENK3$_0clEi"(%class.anon* %f, i32
> %sub)
> >
> > As you can see there is no "zeroext" in the line.
> >
> > Does the following change to the expected results from:
> >
> >   // LMBGEN-LABEL: define internal zeroext i1
> @"_ZZ7lambdasvENK3$_0clEi"(
> >   // LMBUSE-LABEL: define internal zeroext i1
> @"_ZZ7lambdasvENK3$_0clEi"(
> >
> > to:
> >
> >   // LMBGEN-LABEL: define internal{{.*}} i1
> @"_ZZ7lambdasvENK3$_0clEi"(
> >   // LMBUSE-LABEL: define internal{{.*}} i1
> @"_ZZ7lambdasvENK3$_0clEi"(
> >
> > look sensible to you?
> 
> Yes, that should be fine. Sorry for the trouble.