[<prev] [next>] [day] [month] [year] [list]
Message-ID: <20160929134248.GI408@dhcp22.suse.cz>
Date: Thu, 29 Sep 2016 15:42:48 +0200
From: Michal Hocko <mhocko@...nel.org>
To: robert.hu@...el.com
Cc: Oleg Nesterov <oleg@...hat.com>, pbonzini@...hat.com,
akpm@...ux-foundation.org, dan.j.williams@...el.com,
dave.hansen@...el.com, guangrong.xiao@...ux.intel.com,
gleb@...nel.org, mtosatti@...hat.com, kvm@...r.kernel.org,
linux-kernel@...r.kernel.org, stefanha@...hat.com,
yuhuang@...hat.com, linux-mm@...ck.org,
ross.zwisler@...ux.intel.com
Subject: Re: [PATCH v3 1/2] mm, proc: Fix region lost in /proc/self/smaps
On Thu 29-09-16 21:14:40, Robert Hu wrote:
> On Mon, 2016-09-26 at 10:46 +0200, Michal Hocko wrote:
> > On Fri 23-09-16 17:53:51, Oleg Nesterov wrote:
> > > On 09/23, Michal Hocko wrote:
> > > >
> > > > On Fri 23-09-16 15:56:36, Oleg Nesterov wrote:
> > > > >
> > > > > I think we can simplify this patch. And imo make it better. How about
> > > >
> > > > it is certainly less subtle because it doesn't report "sub-vmas".
> > > >
> > > > > if (last_addr) {
> > > > > vma = find_vma(mm, last_addr - 1);
> > > > > if (vma && vma->vm_start <= last_addr)
> > > > > vma = m_next_vma(priv, vma);
> > > > > if (vma)
> > > > > return vma;
> > > > > }
> > > >
> > > > we would still miss a VMA if the last one got shrunk/split
> > >
> > > Not sure I understand what you mean... If the last one was split
> > > we probably should not report the new vma.
> >
> > Right, VMA split is less of a problem. I meant to say that if the
> > last_vma->vm_end got lower for whatever reason then we could miss a VMA
> > right after. We actually might want to display such a VMA because it
> > could be a completely new one. We just do not know whether it is a
> > former split with enlarged VMA or a completely new one
> >
> > [ old VMA ] Hole [ VMA ]
> > [ old VMA ][ New VMa ] [ VMA ]
>
> This is indeed possible. But I see this is like the last_vma enlargement
> case. I suggest we accept such missing, as we accept the enlargement
> part of last_vma is not printed.
>
> How about we set such target:
0) consistent output can be achieved only in the single read call
> 1) no duplicate print; 2) no old vma
> missing (unless it's unmapped); 3) monotonic printing.
> We accept those newly added/changed parts between 2 partial reads is not
> printed.
OK
--
Michal Hocko
SUSE Labs
Powered by blists - more mailing lists