[uClibc] Re: uclibc and heap randomisation (ET_EXEC)

Joakim Tjernlund joakim.tjernlund at lumentis.se
Tue Aug 31 10:50:40 UTC 2004


> On Tue Aug 31, 2004 at 11:27:50AM +0200, Joakim Tjernlund wrote:
> > Don't know, but since Eric just removed the __curbrk stuff from ldso we
> > will soon find out :)
> > 
> > I still not sure that SPARC and ARM's brk() will work as expected, can you
> > test?
> 
> Arm works.  I don't have any sparc boxes to test with so I
> can only speak for arm,

hmm, after some more digging it seems like the __curbrk == 0 tests
in some arches is rendundant and could probably be removed. It depends
on the kernel brk impl. Current 2.4 kernels seems to do the right thing.
Some OS:es always returns a page aligned address for brk(0) and that could
break stuff, but linux doesn't do that.

 Jocke



More information about the uClibc mailing list