Re: [PATCH 1/3] mm/mlock.c: convert put_page() to put_user_page*()

From: Michal Hocko <mhocko@kernel.org>
To: John Hubbard <jhubbard@nvidia.com>
Cc: Vlastimil Babka <vbabka@suse.cz>,
	Andrew Morton <akpm@linux-foundation.org>,
	Christoph Hellwig <hch@infradead.org>,
	Ira Weiny <ira.weiny@intel.com>, Jan Kara <jack@suse.cz>,
	Jason Gunthorpe <jgg@ziepe.ca>,
	Jerome Glisse <jglisse@redhat.com>,
	LKML <linux-kernel@vger.kernel.org>,
	linux-mm@kvack.org, linux-fsdevel@vger.kernel.org,
	Dan Williams <dan.j.williams@intel.com>,
	Daniel Black <daniel@linux.ibm.com>,
	Matthew Wilcox <willy@infradead.org>,
	Mike Kravetz <mike.kravetz@oracle.com>
Subject: Re: [PATCH 1/3] mm/mlock.c: convert put_page() to put_user_page*()
Date: Fri, 9 Aug 2019 11:16:14 +0200	[thread overview]
Message-ID: <20190809091614.GO18351@dhcp22.suse.cz> (raw)
In-Reply-To: <a83e4449-fc8d-7771-1b78-2fa645fa0772@nvidia.com>

On Fri 09-08-19 02:05:15, John Hubbard wrote:
> On 8/9/19 1:23 AM, Michal Hocko wrote:
> > On Fri 09-08-19 10:12:48, Vlastimil Babka wrote:
> > > On 8/9/19 12:59 AM, John Hubbard wrote:
> > > > > > That's true. However, I'm not sure munlocking is where the
> > > > > > put_user_page() machinery is intended to be used anyway? These are
> > > > > > short-term pins for struct page manipulation, not e.g. dirtying of page
> > > > > > contents. Reading commit fc1d8e7cca2d I don't think this case falls
> > > > > > within the reasoning there. Perhaps not all GUP users should be
> > > > > > converted to the planned separate GUP tracking, and instead we should
> > > > > > have a GUP/follow_page_mask() variant that keeps using get_page/put_page?
> > > > > 
> > > > > Interesting. So far, the approach has been to get all the gup callers to
> > > > > release via put_user_page(), but if we add in Jan's and Ira's vaddr_pin_pages()
> > > > > wrapper, then maybe we could leave some sites unconverted.
> > > > > 
> > > > > However, in order to do so, we would have to change things so that we have
> > > > > one set of APIs (gup) that do *not* increment a pin count, and another set
> > > > > (vaddr_pin_pages) that do.
> > > > > 
> > > > > Is that where we want to go...?
> > > > > 
> > > 
> > > We already have a FOLL_LONGTERM flag, isn't that somehow related? And if
> > > it's not exactly the same thing, perhaps a new gup flag to distinguish
> > > which kind of pinning to use?
> > 
> > Agreed. This is a shiny example how forcing all existing gup users into
> > the new scheme is subotimal at best. Not the mention the overal
> > fragility mention elsewhere. I dislike the conversion even more now.
> > 
> > Sorry if this was already discussed already but why the new pinning is
> > not bound to FOLL_LONGTERM (ideally hidden by an interface so that users
> > do not have to care about the flag) only?
> > 
> 
> Oh, it's been discussed alright, but given how some of the discussions have gone,
> I certainly am not surprised that there are still questions and criticisms!
> Especially since I may have misunderstood some of the points, along the way.
> It's been quite a merry go round. :)

Yeah, I've tried to follow them but just gave up at some point.

> Anyway, what I'm hearing now is: for gup(FOLL_LONGTERM), apply the pinned tracking.
> And therefore only do put_user_page() on pages that were pinned with
> FOLL_LONGTERM. For short term pins, let the locking do what it will:
> things can briefly block and all will be well.
> 
> Also, that may or may not come with a wrapper function, courtesy of Jan
> and Ira.
> 
> Is that about right? It's late here, but I don't immediately recall any
> problems with doing it that way...

Yes that makes more sense to me. Whoever needs that tracking should
opt-in for it. Otherwise you just risk problems like the one discussed
in the mlock path (because we do a strange stuff in the name of
performance) and a never ending whack a mole where new users do not
follow the new API usage and that results in all sorts of weird issues.

Thanks!
-- 
Michal Hocko
SUSE Labs