Re: [PATCH v2 1/7] fs/proc/task_mmu.c: don't use task->mm in m_start() and show_*map()

[Oleg Nesterov <oleg@redhat.com>]

On 08/06, Oleg Nesterov wrote:
>
> On 08/06, Kirill A. Shutemov wrote:
> >
> > On Tue, Aug 05, 2014 at 09:46:44PM +0200, Oleg Nesterov wrote:
> > >
> > >  	if (m->count < m->size)  /* vma is copied successfully */
> > > -		m->version = (vma != get_gate_vma(task->mm))
> > > +		m->version = (vma != priv->tail_vma)
> > >  			? vma->vm_start : 0;
> >
> > Drop excessive parenthesis while you're there? And in the next hunk.
>
> I agree. But lets do this in another patch? I forgot to mention this in
> the yesterday's TODO, but we should factor out this code. Perhaps move
> this into vma_stop(). Or at least add a trivial helper, this pattern
> repeats 3 times, plus another one in m_start(). This problem is that I
> do not really understand this logic right now, but I'll do something
> today in any case.

Damn, this needs even more cleanups than I thought. Even "version == -1"
logic is absolutely wrong/dead. And "last_addr rather than next_addr" is
wrong too afaics, this just means we need the extra ->vm_next step plus
vma-or-tail_vma recheck.

Hopefully I'll finish this tomorrow.

Oleg.