Re: 'gegl' segfault breaks "gimp" build

[Joerg Sonnenberger <joerg%bec.de@localhost>]

Am Sun, Apr 10, 2022 at 11:13:50AM +0200 schrieb Tobias Nygren:
> On Sat, 9 Apr 2022 19:21:09 -0500 (CDT)
> "John D. Baker" <jdbaker%consolidated.net@localhost> wrote:
> 
> > On Wed, 23 Mar 2022, John D. Baker wrote:
> > 
> > > Same failure on netbsd-9/i386:
> > 
> > I see the same failure on amd64/9.99.96 w/pkgsrc-2022Q1:
> 
> There are no crashes on amd64 expressed in bulk builds though,
> so I'm relucant to apply the same fix everywhere. Apparently
> it works for some (most?) people.
> 
> Some x86 instructions can raise a general protection fault if SIMD
> operations are not performed on a 128-bit aligned boundary. Sometimes
> compiling code with -mstackrealign in CFLAGS can help with this. We
> have a few packages that do such.

Yeah, it would help tracking where the failing address is coming from.

Joerg