pthread wastes memory with mlockall(MCL_FUTURE)

[Rich Felker]

On Mon, Sep 21, 2015 at 06:41:23PM +0100, Balazs Kezes wrote:
> On 2015-09-20 19:32 -0400, Rich Felker wrote:
> > You're still adding a new mprotect rather than just inverting the one
> > that's already there (somewhere else) adding the PROT_NONE for the
> > guard page. Is there a reason this is hard to do right on glibc?
> 
> Even if I manage to fix it properly (hah, I wish!) I won't be able to
> test it as I lack the various hardware this function supports. I've
> opened a bug instead so that more experienced people pick this up:
> 
> https://sourceware.org/bugzilla/show_bug.cgi?id=18988
> 
> Thanks!

Thanks! This looks like the best approach to get some more attention
on it from people who might have better ideas for what a fix would
look like.

Rich