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;
From:   Chaitanya Kulkarni <Chaitanya.Kulkarni@wdc.com>
To:     Johannes Thumshirn <jthumshirn@suse.de>,
        Christoph Hellwig <hch@lst.de>
CC:     Linux Kernel Mailinglist <linux-kernel@vger.kernel.org>,
        Linux NVMe Mailinglist <linux-nvme@lists.infradead.org>
Subject: Re: [PATCH] nvmet: disable direct I/O when unavailable
Thread-Topic: [PATCH] nvmet: disable direct I/O when unavailable
Thread-Index: AQHUyhJ6d8gGSTBblkStjTf6aoWyPA==
Date:   Fri, 22 Feb 2019 00:41:10 +0000
Message-ID: <SN6PR04MB4527A810D247541AA7C4A512867F0@SN6PR04MB4527.namprd04.prod.outlook.com>
References: <20190221182234.22345-1-jthumshirn@suse.de>
Accept-Language: en-US
Content-Language: en-US
wdcipoutbound: EOP-TRUE
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
MIME-Version: 1.0
X-MS-Exchange-CrossTenant-Network-Message-Id: 544ba4f4-2edd-462b-33f0-08d6985e70d8
X-MS-Exchange-CrossTenant-originalarrivaltime: 22 Feb 2019 00:41:10.5397
 (UTC)
X-MS-Exchange-CrossTenant-fromentityheader: Hosted
X-MS-Exchange-CrossTenant-id: b61c8803-16f3-4c35-9b17-6f65f441df86
X-MS-Exchange-CrossTenant-mailboxtype: HOSTED
X-MS-Exchange-Transport-CrossTenantHeadersStamped: SN6PR04MB4717
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org

On 02/21/2019 10:23 AM, Johannes Thumshirn wrote:=0A=
> Some file-systems, like tmpfs, do not support direct IO, but file-backed=
=0A=
> namespaces default to using direct IO. If direct IO is unavailable fall=
=0A=
> back to using buffered IO for the file-backed namespace.=0A=
>=0A=
> This might not ultimately be a solution for production environments but=
=0A=
> for test environments it sometimes is feasible to use tmpfs.=0A=
>=0A=
> Signed-off-by: Johannes Thumshirn <jthumshirn@suse.de>=0A=
> ---=0A=
>   drivers/nvme/target/io-cmd-file.c | 10 ++++++++++=0A=
>   1 file changed, 10 insertions(+)=0A=
>=0A=
> diff --git a/drivers/nvme/target/io-cmd-file.c b/drivers/nvme/target/io-c=
md-file.c=0A=
> index 517522305e5c..8a861cc0160e 100644=0A=
> --- a/drivers/nvme/target/io-cmd-file.c=0A=
> +++ b/drivers/nvme/target/io-cmd-file.c=0A=
> @@ -38,11 +38,21 @@ int nvmet_file_ns_enable(struct nvmet_ns *ns)=0A=
>=0A=
>   	ns->file =3D filp_open(ns->device_path, flags, 0);=0A=
>   	if (IS_ERR(ns->file)) {=0A=
> +		if (ns->file =3D=3D ERR_PTR(-EINVAL) && (flags & O_DIRECT)) {=0A=
> +			flags &=3D ~O_DIRECT;=0A=
> +			ns->buffered_io =3D 0;=0A=
This needs to be set to 1 when we enable buffered I/O.=0A=
> +			ns->file =3D filp_open(ns->device_path, flags, 0);=0A=
> +			if (!IS_ERR(ns->file)) {=0A=
> +				pr_info("direct I/O unavailable, falling back to buffered I/O\n");=
=0A=
> +				goto getattr;=0A=
> +			}=0A=
> +		}=0A=
>   		pr_err("failed to open file %s: (%ld)\n",=0A=
>   				ns->device_path, PTR_ERR(ns->file));=0A=
>   		return PTR_ERR(ns->file);=0A=
>   	}=0A=
>=0A=
> +getattr:=0A=
>   	ret =3D vfs_getattr(&ns->file->f_path,=0A=
>   			&stat, STATX_SIZE, AT_STATX_FORCE_SYNC);=0A=
>   	if (ret)=0A=
> kk=0A=
=0A=
As per specified in the patch, this is only useful for testing, then we =0A=
should modify the test scripts so that on creation of the ctrl we switch =
=0A=
to the buffered I/O before running fio.=0A=
=0A=
OR=0A=
=0A=
Similar result can be achieved by setting buffered I/O flag=0A=
buffered_io=3D1 before enabling the name-space in the test script.=0A=
=0A=
=0A=