Re: firefox52 core dump on RPI2 NetBSD9.1

[tlaronde%polynum.com@localhost]

On Fri, Nov 27, 2020 at 01:57:52PM +0100, Martin Husemann wrote:
> On Fri, Nov 27, 2020 at 01:52:02PM +0100, tlaronde%polynum.com@localhost wrote:
> > BTW, are these packages "cross-compiled" on a system with an arm
> > processor that is not the target one (a more "muscled" machine than is
> > generally available with a earmv7)? Because the fix was for
> > elf32-arm.h and I wonder if there could be an elfxx-aarch64.c coming in
> > the way...
> 
> They are build with a earmv7hf userland (don't know about processor details).
> 
> Maybe the fix was not complete, or some strange other reason leads
> to similar symptoms and it is actually a different bug?

The fix solved the problem for 32bits, where it appeared. And I as
others have built an entire 8.99 userland and packages without seeing
the problem anymore. For the system, it was cross-compiled from amd64
(in my case) and for the packages they were mostly compiled directly on
an earmv7hf target. But (for me), this was just after the fix, and we
just the fix and not updating the rest of binutils.

I hope there is some binary stuff that is linked without being built, or
that there are some "old" packages lying around, or that some packages
require an old version of the binutils, so that we can find the reason.

We need to know for sure that the binutils are the culprit and that
there is not something "old" lying around, having the incorrect String
Table processing.
-- 
        Thierry Laronde <tlaronde +AT+ polynum +dot+ com>
                     http://www.kergis.com/
                    http://kertex.kergis.com/
                       http://www.sbfa.fr/
Key fingerprint = 0FF7 E906 FBAF FE95 FD89  250D 52B1 AE95 6006 F40C