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;
MIME-Version: 1.0
References: <20210730100158.3117319-1-ruansy.fnst@fujitsu.com>
 <20210730100158.3117319-7-ruansy.fnst@fujitsu.com> <ec5dd047-a420-8e17-d803-729e052b2377@oracle.com>
In-Reply-To: <ec5dd047-a420-8e17-d803-729e052b2377@oracle.com>
From:   Dan Williams <dan.j.williams@intel.com>
Date:   Fri, 20 Aug 2021 15:56:54 -0700
Message-ID: <CAPcyv4hitKKPByHkX-syRmc1rmF8B4sGRsGdUDsBAE5-yoBvXw@mail.gmail.com>
Subject: Re: [PATCH RESEND v6 6/9] xfs: Implement ->notify_failure() for XFS
To:     Jane Chu <jane.chu@oracle.com>
Cc:     Shiyang Ruan <ruansy.fnst@fujitsu.com>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        linux-xfs <linux-xfs@vger.kernel.org>,
        Linux NVDIMM <nvdimm@lists.linux.dev>,
        Linux MM <linux-mm@kvack.org>,
        linux-fsdevel <linux-fsdevel@vger.kernel.org>,
        device-mapper development <dm-devel@redhat.com>,
        "Darrick J. Wong" <djwong@kernel.org>, david <david@fromorbit.com>,
        Christoph Hellwig <hch@lst.de>,
        Alasdair Kergon <agk@redhat.com>,
        Mike Snitzer <snitzer@redhat.com>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Thu, Aug 5, 2021 at 5:50 PM Jane Chu <jane.chu@oracle.com> wrote:
>
>
> On 7/30/2021 3:01 AM, Shiyang Ruan wrote:
> > +     mapping = VFS_I(ip)->i_mapping;
> > +     if (IS_ENABLED(CONFIG_MEMORY_FAILURE)) {
> > +             for (i = 0; i < rec->rm_blockcount; i++) {
> > +                     error = mf_dax_kill_procs(mapping, rec->rm_offset + i,
> > +                                               *flags);
> > +                     if (error)
> > +                             break;
> > +             }
> > +     }
>
> If a poison is injected to a PMD dax page, after consuming the poison,
> how many SIGBUS signals are expected to be sent to the process?

I think it should only get one. I.e. just like the the generic code
does one shootdown per mapped page regardless of whether that page is
4K, 2M, or 1G. Once the application is notified it should be able to
query the filesystem to determine the full extent of the damage to
files.