Re: [PATCH 14/38] Remove struct mm_struct::exe_file et al

[Alexey Dobriyan <adobriyan@gmail.com>]

On Tue, May 26, 2009 at 04:24:15PM -0700, Andrew Morton wrote:
> On Tue, 26 May 2009 04:36:18 -0700
> Matt Helsley <matthltc@us.ibm.com> wrote:
> 
> > I don't see any mention in the changelog of the point brought up by Ingo:
> > 
> > http://lkml.org/lkml/2009/4/10/105
> 
> Nor of Eric's comments.
> 
> Alexey, pleeeze don't do this.  We (read: I) heavily depend upon patch
> submitters to keep track of outstanding issues and review comments,
> etc.
> 
> If the patch submitter simply blows these things off then it devolves
> to me having to keep track of each patch's issue list as well as the
> patch itself.  My workload goes up by a factor of N and the error rate
> goes up by N^2 :(

grmbh..

"Security" and "holding ->mmap_sem" were answered and dismissed.

You can't do readlink(2) on /proc/*/exe if you can't ptrace task.
So no new possible holes are created.

->mmap_sem was held since /proc/*/exe was added and nobody cared.
And, again, you can't readlink _any_ /proc/*/exe.

Patch simply restores code to year-back state.

I'll send removal and readddition of "struct path" as separate things
next time.


And BTW, there is something unnatural when executable path is attached
to mm_struct(!) not task_struct, so yet another argument to ->exe_file
removal.