Thursday, January 31, 2019

Re: sparc64 and C++ exceptions

On Friday, February 1, 2019 1:24 AM, Otto Moerbeek <otto@drijf.net> wrote:

> On Thu, Jan 31, 2019 at 05:17:33PM +0000, Stuart Henderson wrote:
>
> > On 2019/01/31 01:12, Tinker wrote:
> >
> > > Reading your log file
> > > http://build-failures.rhaalovely.net//sparc64/2019-01-10/net/amule,-daemon.log
> > > , there is mentioning of libestdc++ in there.
> >
> > That doesn't use g++ from base though, only the ports version.
> > I went through all C++ ports not on the direct path to building gcc
> > and added a COMPILER line so they are built with ports gcc.
>
> Meanwhile, kettenis@ mentioned the probably cause. I'm currently
> building ports gcc on sparc64 with a diff that could fix things.
> Specifically, the stack frames on OpenBSD are different compared to
> other systems since we have stackghost.
>
> See http://projects.cerias.purdue.edu/stackghost/stackghost/stackghost.html
>
> -Otto

Wait, the system g++ uses one stack frame format and the ports eg++
uses another one, this is how it is and has been for a long time?

If so no wonder that g++ and eg++ produced incompatible binaries.

So right now you are doing the historic maneuver of making g++ and
eg++-produced binaries compatible?

Also how does LLVM play in here.

Can you please confirm how it is?

No comments:

Post a Comment