Re: [PATCH] Do not allow to inline ifunc resolvers (PR ipa/81128).

2017-06-28 Thread Martin Liška
gt;>> b/gcc/testsuite/gcc.target/i386/pr81128.c >>>> new file mode 100644 >>>> index 000..90a567ad690 >>>> --- /dev/null >>>> +++ b/gcc/testsuite/gcc.target/i386/pr81128.c >>>> @@ -0,0 +1,65 @@ >>>> +/* PR ipa/811

Re: [PATCH] Do not allow to inline ifunc resolvers (PR ipa/81128).

2017-06-27 Thread Jan Hubicka
> diff --git a/gcc/ipa-visibility.c b/gcc/ipa-visibility.c > index d5a3ae56c46..69e6e295d55 100644 > --- a/gcc/ipa-visibility.c > +++ b/gcc/ipa-visibility.c > @@ -97,7 +97,8 @@ non_local_p (struct cgraph_node *node, void *data > ATTRIBUTE_UNUSED) >&& !DECL_EXTERNAL (node->decl) >

Re: [PATCH] Do not allow to inline ifunc resolvers (PR ipa/81128).

2017-06-27 Thread Martin Liška
On 06/27/2017 04:57 PM, Jan Hubicka wrote: >> Hello. >> >> Currently ifunc is interpreted as normal alias by IPA optimizations. That's >> problematic >> as should not consider ifunc alias as candidate for inlining, or redirection. >> >> Patch can bootstrap on ppc64le-redhat-linux and survives regr

Re: [PATCH] Do not allow to inline ifunc resolvers (PR ipa/81128).

2017-06-27 Thread Jan Hubicka
> Hello. > > Currently ifunc is interpreted as normal alias by IPA optimizations. That's > problematic > as should not consider ifunc alias as candidate for inlining, or redirection. > > Patch can bootstrap on ppc64le-redhat-linux and survives regression tests. > And survives MVC tests on x86_64

[PATCH] Do not allow to inline ifunc resolvers (PR ipa/81128).

2017-06-27 Thread Martin Liška
Hello. Currently ifunc is interpreted as normal alias by IPA optimizations. That's problematic as should not consider ifunc alias as candidate for inlining, or redirection. Patch can bootstrap on ppc64le-redhat-linux and survives regression tests. And survives MVC tests on x86_64-linux-gnu. Rea