Re: Commonize handling of attr-fnspec

2020-10-01 Thread Richard Biener
On Thu, 1 Oct 2020, Jan Hubicka wrote: > Hi > > > + if (!fnspec.arg_specified_p (arg)) > > > +; > > > + else if (!fnspec.arg_used_p (arg)) > > > +flags = EAF_UNUSED; > > > + else > > > +{ > > > + if (!fnspec.arg_direct_p (arg)) > > > > negated test > > > > > + flags |= EAF_DI

Re: Commonize handling of attr-fnspec

2020-10-01 Thread Jan Hubicka
Hi > > + if (!fnspec.arg_specified_p (arg)) > > +; > > + else if (!fnspec.arg_used_p (arg)) > > +flags = EAF_UNUSED; > > + else > > +{ > > + if (!fnspec.arg_direct_p (arg)) > > negated test > > > + flags |= EAF_DIRECT; > > + if (!fnspec.arg_noescape_p (arg)) > > + fla

Re: Commonize handling of attr-fnspec

2020-10-01 Thread Richard Biener
On Thu, 1 Oct 2020, Jan Hubicka wrote: > Hi, > this patch adds the simple class for parsing fnspec attribute. I plan > to add support for generating and modifying it too (it is used by > fortran and I plan to make modref to detect noclobbers and stuff). > Verification is disabled until we fix rem

Commonize handling of attr-fnspec

2020-10-01 Thread Jan Hubicka
Hi, this patch adds the simple class for parsing fnspec attribute. I plan to add support for generating and modifying it too (it is used by fortran and I plan to make modref to detect noclobbers and stuff). Verification is disabled until we fix remaining fortran specifier (I got a promised help at