Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Tue, 5 Feb 2019 02:14:34 -0500 (EST)
From:   Jan Stancek <jstancek@redhat.com>
To:     Lars Persson <lists@bofh.nu>
Cc:     linux-mm@kvack.org, lersek@redhat.com,
        alex williamson <alex.williamson@redhat.com>,
        aarcange@redhat.com, rientjes@google.com, kirill@shutemov.name,
        mgorman@techsingularity.net, mhocko@suse.com,
        linux-kernel@vger.kernel.org
Message-ID: <997509746.100933786.1549350874925.JavaMail.zimbra@redhat.com>
In-Reply-To: <CADnJP=vsum7_YYWBpknpahTQFAzm7G40_E2dLMB_poFEhPKEfw@mail.gmail.com>
References: <eabca57aa14f4df723173b24891f4a2d9c501f21.1543526537.git.jstancek@redhat.com> <c440d69879e34209feba21e12d236d06bc0a25db.1543577156.git.jstancek@redhat.com> <CADnJP=vsum7_YYWBpknpahTQFAzm7G40_E2dLMB_poFEhPKEfw@mail.gmail.com>
Subject: Re: [PATCH v2] mm: page_mapped: don't assume compound page is huge
 or THP
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit
Thread-Topic: page_mapped: don't assume compound page is huge or THP
Thread-Index: UZRA0ONgTI/KE7JRWOwpjzv1lbRMeg==
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk


----- Original Message -----
> On Fri, Nov 30, 2018 at 1:07 PM Jan Stancek <jstancek@redhat.com> wrote:
> >
> > LTP proc01 testcase has been observed to rarely trigger crashes
> > on arm64:
> >     page_mapped+0x78/0xb4
> >     stable_page_flags+0x27c/0x338
> >     kpageflags_read+0xfc/0x164
> >     proc_reg_read+0x7c/0xb8
> >     __vfs_read+0x58/0x178
> >     vfs_read+0x90/0x14c
> >     SyS_read+0x60/0xc0
> >
> > Issue is that page_mapped() assumes that if compound page is not
> > huge, then it must be THP. But if this is 'normal' compound page
> > (COMPOUND_PAGE_DTOR), then following loop can keep running
> > (for HPAGE_PMD_NR iterations) until it tries to read from memory
> > that isn't mapped and triggers a panic:
> >         for (i = 0; i < hpage_nr_pages(page); i++) {
> >                 if (atomic_read(&page[i]._mapcount) >= 0)
> >                         return true;
> >         }
> >
> > I could replicate this on x86 (v4.20-rc4-98-g60b548237fed) only
> > with a custom kernel module [1] which:
> > - allocates compound page (PAGEC) of order 1
> > - allocates 2 normal pages (COPY), which are initialized to 0xff
> >   (to satisfy _mapcount >= 0)
> > - 2 PAGEC page structs are copied to address of first COPY page
> > - second page of COPY is marked as not present
> > - call to page_mapped(COPY) now triggers fault on access to 2nd
> >   COPY page at offset 0x30 (_mapcount)
> >
> > [1]
> > https://github.com/jstancek/reproducers/blob/master/kernel/page_mapped_crash/repro.c
> >
> > Fix the loop to iterate for "1 << compound_order" pages.
> >
> > Debugged-by: Laszlo Ersek <lersek@redhat.com>
> > Suggested-by: "Kirill A. Shutemov" <kirill@shutemov.name>
> > Signed-off-by: Jan Stancek <jstancek@redhat.com>
> > ---
> >  mm/util.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > Changes in v2:
> > - change the loop instead so we check also mapcount of subpages
> >
> > diff --git a/mm/util.c b/mm/util.c
> > index 8bf08b5b5760..5c9c7359ee8a 100644
> > --- a/mm/util.c
> > +++ b/mm/util.c
> > @@ -478,7 +478,7 @@ bool page_mapped(struct page *page)
> >                 return true;
> >         if (PageHuge(page))
> >                 return false;
> > -       for (i = 0; i < hpage_nr_pages(page); i++) {
> > +       for (i = 0; i < (1 << compound_order(page)); i++) {
> >                 if (atomic_read(&page[i]._mapcount) >= 0)
> >                         return true;
> >         }
> > --
> > 1.8.3.1
> 
> Hi all
> 
> This patch landed in the 4.9-stable tree starting from 4.9.151 and it
> broke our MIPS1004kc system with CONFIG_HIGHMEM=y.

Hi,

are you using THP (CONFIG_TRANSPARENT_HUGEPAGE)?

The changed line should affect only THP and normal compound pages,
so a test with THP disabled might be interesting. 

> 
> The breakage consists of random processes dying with SIGILL or SIGSEGV
> when we stress test the system with high memory pressure and explicit
> memory compaction requested through /proc/sys/vm/compact_memory.
> Reverting this patch fixes the crashes.
> 
> We can put some effort on debugging if there are no obvious
> explanations for this. Keep in mind that this is 32-bit system with
> HIGHMEM.

Nothing obvious that I can see. I've been trying to reproduce on
32-bit x86 Fedora with no luck so far.

Regards,
Jan