Re: sys_kcmp

[Cyrill Gorcunov <gorcunov@openvz.org>]

On Mon, Sep 24, 2012 at 01:44:47PM -0700, Eric W. Biederman wrote:
> Cyrill Gorcunov <gorcunov@openvz.org> writes:
> 
> > On Mon, Sep 24, 2012 at 10:51:19PM +0400, Cyrill Gorcunov wrote:
> >> > I expect what you want is a call to access_ok, rather than hard coding
> >> > details about task layout here.  This test certainly looks wrong
> >> > for a 32bit process on a 64bit kernel. If I read your test right it
> >> > appears I can set values of say 0x100000000 on a 32bit process...
> >> > 
> >> > As for mmap_min_addr I would expect your find_vma check would make that
> >> > test unnecessary, simply by not finding a vma...
> >> 
> >> Good point, Eric, thanks! I'm cooking a new patch now.
> >
> > Btw, Eric, I somehow miss one bit -- how would you set this 0x100000000
> > if TASK_SIZE is a macro which does check for TIF_ADDR32 and sets limit
> > acordingly? What i'm missing?
> 
> How odd.  Last time I had looked TASK_SIZE was a simple constant.

Ah, I see.

> Still I wonder a little if all architectures currently run from 0 to
> TASK_SIZE, for address space available.  I seem to remember there have
> been some exceptions to that rule.  But I can't recall what they were.

Actually I;ve tuned up the code to use access_ok instead but now I'm trying
to fugure out situation if it can somehow affect c/r process (well, i've
ran all test cases we use for c/r and all are passed well, but still...).

Mark, after some more thinking, I agree that your proposal with min-address
should work better than mine explicit CONFIG_MMU. Could you please send
your patch for that? As to access_ok -- gimme some more time, i need to double
check everything and I'll patch the code on top of your patch a bit later, ok?