[lia64-kernel] Re: Re: Re: Re: clone and clone2

[Ulrich Drepper]

David Mosberger <davidm@hpl.hp.com> writes:

>   >> that is one solution. All the other CPUs who want thread_top need
>   >> to use thread_bottom + stack_size to get thread_top in their
>   >> clone.S.
> 
>   Uli> This only makes sense if the kernel is actually able to limit
>   Uli> the growth of the stack without the user creating an artificial
>   Uli> barrier (as we do in the moment).
> 
> Why do you think so?

If you have a stack which grows downwards and the you feed the low
stack address and the size, the kernel will have to compute the high
address for itself.  Then the user can assume that this address range
is taken.  But it's not.  The kernel can place arbitrary mmap()ed
blocks just in the middle and then?  This problem of course exists
with the current implementation as well but at least the interface
does not suggest something differently.

-- 
---------------.      drepper at gnu.org  ,-.   1325 Chesapeake Terrace
Ulrich Drepper  \    ,-------------------'   \  Sunnyvale, CA 94089 USA
Red Hat          `--' drepper at redhat.com   `------------------------