Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Date:   Tue, 21 Apr 2020 16:46:03 +0200
From:   Michal Hocko <mhocko@kernel.org>
To:     Peter Xu <peterx@redhat.com>
Cc:     Andrew Morton <akpm@linux-foundation.org>,
        Linus Torvalds <torvalds@linux-foundation.org>,
        linux-mm@kvack.org, LKML <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH] mm, mempolicy: fix up gup usage in lookup_node
Message-ID: <20200421144603.GI27314@dhcp22.suse.cz>
References: <20200421071026.18394-1-mhocko@kernel.org>
 <20200421132916.GE420399@xz-x1>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20200421132916.GE420399@xz-x1>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Tue 21-04-20 09:29:16, Peter Xu wrote:
> On Tue, Apr 21, 2020 at 09:10:26AM +0200, Michal Hocko wrote:
> > From: Michal Hocko <mhocko@suse.com>
> > 
> > ba841078cd05 ("mm/mempolicy: Allow lookup_node() to handle fatal signal") has
> > added a special casing for 0 return value because that was a possible
> > gup return value when interrupted by fatal signal. This has been fixed
> > by ae46d2aa6a7f ("mm/gup: Let __get_user_pages_locked() return -EINTR
> > for fatal signal") in the mean time so ba841078cd05 can be reverted.
> > 
> > This patch however doesn't go all the way to revert it because the check
> > for 0 is wrong and confusing here. Firstly it is inherently unsafe to
> > access the page when get_user_pages_locked returns 0 (aka no page
> > returned).
> > Fortunatelly this will not happen because get_user_pages_locked will not
> > return 0 when nr_pages > 0 unless FOLL_NOWAIT is specified which is not
> > the case here. Document this potential error code in gup code while we
> > are at it.
> > 
> > Signed-off-by: Michal Hocko <mhocko@suse.com>
> > ---
> >  mm/gup.c       | 5 +++++
> >  mm/mempolicy.c | 5 +----
> >  2 files changed, 6 insertions(+), 4 deletions(-)
> > 
> > diff --git a/mm/gup.c b/mm/gup.c
> > index 50681f0286de..a8575b880baf 100644
> > --- a/mm/gup.c
> > +++ b/mm/gup.c
> > @@ -980,6 +980,7 @@ static int check_vma_flags(struct vm_area_struct *vma, unsigned long gup_flags)
> >   * -- If nr_pages is >0, but no pages were pinned, returns -errno.
> >   * -- If nr_pages is >0, and some pages were pinned, returns the number of
> >   *    pages pinned. Again, this may be less than nr_pages.
> > + * -- 0 return value is possible when the fault would need to be retried.
> >   *
> >   * The caller is responsible for releasing returned @pages, via put_page().
> >   *
> > @@ -1247,6 +1248,10 @@ int fixup_user_fault(struct task_struct *tsk, struct mm_struct *mm,
> >  }
> >  EXPORT_SYMBOL_GPL(fixup_user_fault);
> >  
> > +/*
> > + * Please note that this function, unlike __get_user_pages will not
> > + * return 0 for nr_pages > 0 without FOLL_NOWAIT
> 
> It's a bit unclear to me on whether "will not return 0" applies to "this
> function" or "__get_user_pages"...  Might be easier just to avoid mentioning
> __get_user_pages?

I really wanted to call out __get_user_pages because the semantic of
0 return value is different. If you have a suggestion how to reformulate
this to be more clear then I will incorporate that.

> > + */
> >  static __always_inline long __get_user_pages_locked(struct task_struct *tsk,
> >  						struct mm_struct *mm,
> >  						unsigned long start,
> > diff --git a/mm/mempolicy.c b/mm/mempolicy.c
> > index 48ba9729062e..1965e2681877 100644
> > --- a/mm/mempolicy.c
> > +++ b/mm/mempolicy.c
> > @@ -927,10 +927,7 @@ static int lookup_node(struct mm_struct *mm, unsigned long addr)
> >  
> >  	int locked = 1;
> >  	err = get_user_pages_locked(addr & PAGE_MASK, 1, 0, &p, &locked);
> > -	if (err == 0) {
> > -		/* E.g. GUP interrupted by fatal signal */
> > -		err = -EFAULT;
> > -	} else if (err > 0) {
> > +	if (err > 0) {
> >  		err = page_to_nid(p);
> >  		put_page(p);
> >  	}
> 
> Again, this is my totally humble opinion: I'm fine with removing the comment,
> however I still don't think it's helpful at all to explicitly remove a check
> against invalid return value (err==0), especially if that's the only functional
> change in this patch.

I thought I have explained that when we have discussed last time and the
changelog is explaining that as well. Checking for impossible error code
is simply confusing and provokes for copy&pasting this pattern. I
wouldn't really bother if I haven't seen this cargo cult pattern in the
so many times.
-- 
Michal Hocko
SUSE Labs