This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Unreviewed build, lto patch


On Wed, 16 Feb 2011, Rainer Orth wrote:

> Richard Guenther <rguenther@suse.de> writes:
> 
> > On Wed, 16 Feb 2011, Rainer Orth wrote:
> >
> >> Richard Guenther <rguenther@suse.de> writes:
> >> 
> >> > I think at this point it would be more useful to check for specific
> >> > known good linkers and versions, also for whether we use 
> >> > -fuse-linker-plugin by default or not.
> >> 
> >> Fully agreed, not only because it simplifies the configure part.  If
> >> anyone could tell me which gld and gold versions are considered good
> >> (and I suppose this means full support for all aspects of LTO and the
> >> lto-plugin, thus something newer than gld/gold 2.21?), I could update my
> >> patch accordingly.
> >
> > I think we should require at least GNU ld/gold 2.21 given the amount
> > of bugfixes over 2.20.  If 2.21.1 is released in-time I'd consider
> > even requiring that version (mind - for making -fuse-linker-plugin
> > the _default_, not for disabling support for it entirely).
> 
> And what about the testcases (at least one or two) that fail with 2.21?
> Live with them?

Yes, they should then only fail with -fuse-linker-plugin and show
people w/ older linker that they may have an issue.  It's only
bad if they fail with -flto only with no sign of the real cause IMHO.

Richard.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]