Re: [PATCH] tools/testing/selftests/vm/mlock2-tests: fix mlock2 false-negative errors

From: Rafael Aquini <aquini@redhat.com>
To: Michal Hocko <mhocko@kernel.org>
Cc: Shakeel Butt <shakeelb@google.com>,
	LKML <linux-kernel@vger.kernel.org>,
	linux-kselftest@vger.kernel.org, shuah@kernel.org,
	Andrew Morton <akpm@linux-foundation.org>
Subject: Re: [PATCH] tools/testing/selftests/vm/mlock2-tests: fix mlock2 false-negative errors
Date: Mon, 23 Mar 2020 10:55:42 -0400	[thread overview]
Message-ID: <20200323145542.GC23364@optiplex-lnx> (raw)
In-Reply-To: <20200323142941.GK7524@dhcp22.suse.cz>

On Mon, Mar 23, 2020 at 03:29:41PM +0100, Michal Hocko wrote:
> On Mon 23-03-20 10:16:59, Rafael Aquini wrote:
> > On Sun, Mar 22, 2020 at 09:31:04AM -0700, Shakeel Butt wrote:
> > > On Sat, Mar 21, 2020 at 6:35 PM Rafael Aquini <aquini@redhat.com> wrote:
> > > >
> > > > Changes for commit 9c4e6b1a7027f ("mm, mlock, vmscan: no more skipping pagevecs")
> > > > break this test expectations on the behavior of mlock syscall family immediately
> > > > inserting the recently faulted pages into the UNEVICTABLE_LRU, when MCL_ONFAULT is
> > > > passed to the syscall as part of its flag-set.
> > > 
> > > mlock* syscalls do not provide any guarantee that the pages will be in
> > > unevictable LRU, only that the pages will not be paged-out. The test
> > > is checking something very internal to the kernel and this is expected
> > > to break.
> > 
> > It was a check expected to be satisfied before the commit, though. 
> > Getting the mlocked pages inserted directly into the unevictable LRU,
> > skipping the pagevec, was established behavior before the aforementioned
> > commit, and even though one could argue userspace should not be aware,
> > or care, about such inner kernel circles the program in question is not an 
> > ordinary userspace app, but a kernel selftest that is supposed to check
> > for the functionality correctness.
> 
> But mlock (in neither mode) is reall forced to put pages to the
> UNEVICTABLE_LRU for correctness. If the test is really depending on it
> then the selftest is bogus. A real MCL_ONFAULT test should focus on the
> user observable contract of this api. And that is that a new mapping
> doesn't fault in the page during the mlock call but the memory is locked
> after the memory is faulted in. You can use different methods to observe
> locked memory - e.g. try to reclaim it and check or check /proc/<pid>/smaps
>

Again, I don't think the test is bogus, although it's (now) expecting
something that is not guaranteed after the referred commit.
The check for PG_unevictable set on the page backing up the mapping
seems reasonable, as the flag is supposed to be there, if everything 
went on fine after the mlock call. 

-- Rafael