Re: gcc/jemalloc failure

[Manuel Bouyer <bouyer%antioche.eu.org@localhost>]

On Wed, Apr 24, 2019 at 07:07:46PM -0000, Christos Zoulas wrote:
> In article <20190424102534.GA20995%mail.soc.lip6.fr@localhost>,
> Manuel Bouyer  <bouyer%antioche.eu.org@localhost> wrote:
> >Hello
> >while building wxGTK30 (with gtk2) on earmv7hf, I got:
> ><jemalloc>:
> >/usr/src/external/bsd/jemalloc/lib/../include/jemalloc/internal/rtree.h:397:
> >Failed assertion: "rtree_leaf_elm_extent_read(tsdn, rtree, elm, false)
> >== NULL"
> ><jemalloc>: Should own 0 locks of rank >= 1: extent_mutex_pool(16)
> >g++: internal compiler error: Abort trap (program cc1plus received signal 6)
> ><jemalloc>:
> >/usr/src/external/bsd/jemalloc/lib/../include/jemalloc/internal/rtree.h:397:
> >Failed assertion: "rtree_leaf_elm_extent_read(tsdn, rtree, elm, false)
> >== NULL"
> ><jemalloc>: Should own 0 locks of rank >= 1: extent_mutex_pool(16)
> >g++: internal compiler error: Abort trap (program cc1plus received signal 6)
> >
> >any idea ?
> 
> Stack trace please?

Unfortunably it's not easy to reproduce. From several runs, gcc will
either die with a segfault, or hit an assert in jemalloc. There is
no core dump (I've not investigated why).

wxGTK30 disables precompiled headers for previous gcc versions, for
what looks like the same problem. I'll probably add gcc7 to the list.

-- 
Manuel Bouyer <bouyer%antioche.eu.org@localhost>
     NetBSD: 26 ans d'experience feront toujours la difference
--