Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Tue, 18 Apr 2023 16:32:27 +0100
From:   Matthew Wilcox <willy@infradead.org>
To:     Peter Xu <peterx@redhat.com>
Cc:     Suren Baghdasaryan <surenb@google.com>, akpm@linux-foundation.org,
        hannes@cmpxchg.org, mhocko@suse.com, josef@toxicpanda.com,
        jack@suse.cz, ldufour@linux.ibm.com, laurent.dufour@fr.ibm.com,
        michel@lespinasse.org, liam.howlett@oracle.com, jglisse@google.com,
        vbabka@suse.cz, minchan@google.com, dave@stgolabs.net,
        punit.agrawal@bytedance.com, lstoakes@gmail.com,
        linux-mm@kvack.org, linux-kernel@vger.kernel.org,
        kernel-team@android.com
Subject: Re: [PATCH v2 1/1] mm: do not increment pgfault stats when page
 fault handler retries
Message-ID: <ZD64C3R6BzqpSfYX@casper.infradead.org>
References: <20230415000818.1955007-1-surenb@google.com>
 <ZD25bBPbZYSb7grA@x1n>
 <CAJuCfpHf06cr2d277DXQUtBto_0bVgK3ykMHLYRgZXgnot=e4w@mail.gmail.com>
 <ZD61DLJNilUeDCnC@x1n>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <ZD61DLJNilUeDCnC@x1n>
Precedence: bulk

On Tue, Apr 18, 2023 at 11:19:40AM -0400, Peter Xu wrote:
> On Mon, Apr 17, 2023 at 03:47:57PM -0700, Suren Baghdasaryan wrote:
> > On Mon, Apr 17, 2023 at 2:26 PM Peter Xu <peterx@redhat.com> wrote:
> > >
> > > On Fri, Apr 14, 2023 at 05:08:18PM -0700, Suren Baghdasaryan wrote:
> > > > @@ -5223,8 +5230,8 @@ vm_fault_t handle_mm_fault(struct vm_area_struct *vma, unsigned long address,
> > > >               if (task_in_memcg_oom(current) && !(ret & VM_FAULT_OOM))
> > > >                       mem_cgroup_oom_synchronize(false);
> > > >       }
> > > > -
> > > > -     mm_account_fault(regs, address, flags, ret);
> > > > +out:
> > > > +     mm_account_fault(mm, regs, address, flags, ret);
> > >
> > > Ah, one more question.. can this cached mm race with a destroying mm (just
> > > like the vma race we wanted to avoid)?  Still a question only applies to
> > > COMPLETE case when mmap read lock can be released.  Thanks,
> > 
> > I believe that is impossible because whoever is calling the page fault
> > handler has stabilized the mm by getting a refcount.
> 
> Do you have a hint on where that refcount is taken?

... when we called clone()?  A thread by definition has a reference to
its own mm.

> Btw, it's definitely not a question sololy for this patch but a more common
> question to the page fault path.  It's just that when I wanted to look for
> any refcount boost (which I also expect to have somewhere) I didn't really
> see that in current path (e.g. do_user_addr_fault() for x86_64).
> 
> I also had a quick look on do_exit() but I also didn't see where do we
> e.g. wait for all the threads to stop before recycles a mm.
> 
> I had a feeling that I must have missed something, but just want to make
> sure it's the case.
> 
> -- 
> Peter Xu
>