Fwd: glibc and unexec

[Paul Eggert]

Florian Weimer wrote:
> In my test, Emacs simply interposed
> its own bundled malloc implementation, which is quite the opposite of
> “getting out of the malloc business”.

Yes, this approach is merely a stopgap and we need something better; it's just 
that whatever we come up with probably won't be ready in time for Emacs 25.

> a multi-threaded Emacs will
> likely exercise gmalloc.c in new ways not originally expected.

That's one worry, yes. That being said, gmalloc does work on non-glibc platforms 
that have multithreaded Emacs.