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;
IronPort-SDR: xVGfzDaeXy9LMyXE6Mf78oJ+eeELgzQsN2Gr8HZk0zC8Za45bWJSbbVFFg3VvV5Oa67qdlvOGL
 mtLKPL7PcxeC+H8k/Wyae159j2BlzKKP39V7IWzpmw1umJ2TCSKGl2WrECXxGe1v0jHEJD+l92
 dIoGOrRW3s+/SwBz9V7+qhXJcWA0d8i+9wFVl+A25UNmo25b0PMOKKyuEsmHJlsK0gwNNeS/Li
 xg6cFSgSWmZDArZ15LNk3eBX2CIhVVuiRTgtzL3LOKeWJPVcVoWnOQBrxBrY3dbJTyiX+WPi8U
 kUs=
From:   Damien Le Moal <Damien.LeMoal@wdc.com>
To:     "javier.gonz@samsung.com" <javier@javigon.com>
CC:     Kanchan Joshi <joshi.k@samsung.com>,
        "axboe@kernel.dk" <axboe@kernel.dk>,
        "viro@zeniv.linux.org.uk" <viro@zeniv.linux.org.uk>,
        "bcrl@kvack.org" <bcrl@kvack.org>,
        "linux-fsdevel@vger.kernel.org" <linux-fsdevel@vger.kernel.org>,
        "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
        "linux-aio@kvack.org" <linux-aio@kvack.org>,
        "io-uring@vger.kernel.org" <io-uring@vger.kernel.org>,
        "linux-block@vger.kernel.org" <linux-block@vger.kernel.org>,
        "selvakuma.s1@samsung.com" <selvakuma.s1@samsung.com>,
        "nj.shetty@samsung.com" <nj.shetty@samsung.com>
Subject: Re: [PATCH 3/3] io_uring: add support for zone-append
Thread-Topic: [PATCH 3/3] io_uring: add support for zone-append
Thread-Index: AQHWRMyUe8chR0VkUU2K4ON9AWdJKQ==
Date:   Thu, 18 Jun 2020 08:47:29 +0000
Message-ID: <CY4PR04MB3751D5D6AFB0DA7B8A2DFF61E79B0@CY4PR04MB3751.namprd04.prod.outlook.com>
References: <1592414619-5646-1-git-send-email-joshi.k@samsung.com>
 <CGME20200617172713epcas5p352f2907a12bd4ee3c97be1c7d8e1569e@epcas5p3.samsung.com>
 <1592414619-5646-4-git-send-email-joshi.k@samsung.com>
 <CY4PR04MB37510E916B6F243D189B4EB0E79B0@CY4PR04MB3751.namprd04.prod.outlook.com>
 <20200618083529.ciifu4chr4vrv2j5@mpHalley.local>
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: 27803948-8030-42c8-b0c2-08d813643bb1
X-MS-Exchange-CrossTenant-originalarrivaltime: 18 Jun 2020 08:47:29.0624
 (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-CrossTenant-userprincipalname: C2BY3/DQpKiWKaE9kPDe/OL0+rAKDh74nHPV9nkJZVPjKjXF632yF9W15/ihOecILckJ77q7LroeZjgafb07jQ==
X-MS-Exchange-Transport-CrossTenantHeadersStamped: CY4PR0401MB3586
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 2020/06/18 17:35, javier.gonz@samsung.com wrote:=0A=
> On 18.06.2020 07:39, Damien Le Moal wrote:=0A=
>> On 2020/06/18 2:27, Kanchan Joshi wrote:=0A=
>>> From: Selvakumar S <selvakuma.s1@samsung.com>=0A=
>>>=0A=
>>> Introduce three new opcodes for zone-append -=0A=
>>>=0A=
>>>    IORING_OP_ZONE_APPEND     : non-vectord, similiar to IORING_OP_WRITE=
=0A=
>>>    IORING_OP_ZONE_APPENDV    : vectored, similar to IORING_OP_WRITEV=0A=
>>>    IORING_OP_ZONE_APPEND_FIXED : append using fixed-buffers=0A=
>>>=0A=
>>> Repurpose cqe->flags to return zone-relative offset.=0A=
>>>=0A=
>>> Signed-off-by: SelvaKumar S <selvakuma.s1@samsung.com>=0A=
>>> Signed-off-by: Kanchan Joshi <joshi.k@samsung.com>=0A=
>>> Signed-off-by: Nitesh Shetty <nj.shetty@samsung.com>=0A=
>>> Signed-off-by: Javier Gonzalez <javier.gonz@samsung.com>=0A=
>>> ---=0A=
>>>  fs/io_uring.c                 | 72 +++++++++++++++++++++++++++++++++++=
++++++--=0A=
>>>  include/uapi/linux/io_uring.h |  8 ++++-=0A=
>>>  2 files changed, 77 insertions(+), 3 deletions(-)=0A=
>>>=0A=
>>> diff --git a/fs/io_uring.c b/fs/io_uring.c=0A=
>>> index 155f3d8..c14c873 100644=0A=
>>> --- a/fs/io_uring.c=0A=
>>> +++ b/fs/io_uring.c=0A=
>>> @@ -649,6 +649,10 @@ struct io_kiocb {=0A=
>>>  	unsigned long		fsize;=0A=
>>>  	u64			user_data;=0A=
>>>  	u32			result;=0A=
>>> +#ifdef CONFIG_BLK_DEV_ZONED=0A=
>>> +	/* zone-relative offset for append, in bytes */=0A=
>>> +	u32			append_offset;=0A=
>>=0A=
>> this can overflow. u64 is needed.=0A=
> =0A=
> We chose to do it this way to start with because struct io_uring_cqe=0A=
> only has space for u32 when we reuse the flags.=0A=
> =0A=
> We can of course create a new cqe structure, but that will come with=0A=
> larger changes to io_uring for supporting append.=0A=
> =0A=
> Do you believe this is a better approach?=0A=
=0A=
The problem is that zone size are 32 bits in the kernel, as a number of sec=
tors.=0A=
So any device that has a zone size smaller or equal to 2^31 512B sectors ca=
n be=0A=
accepted. Using a zone relative offset in bytes for returning zone append r=
esult=0A=
is OK-ish, but to match the kernel supported range of possible zone size, y=
ou=0A=
need 31+9 bits... 32 does not cut it.=0A=
=0A=
Since you need a 64-bit sized result, I would also prefer that you drop the=
 zone=0A=
relative offset as a result and return the absolute offset instead. That ma=
kes=0A=
life easier for the applications since the zone append requests also must u=
se=0A=
absolute offsets for zone start. An absolute offset as a result becomes=0A=
consistent with that and all other read/write system calls that all use abs=
olute=0A=
offsets (seek() is the only one that I know of that can use a relative offs=
et,=0A=
but that is not an IO system call).=0A=
=0A=
=0A=
> =0A=
> Javier=0A=
> =0A=
=0A=
=0A=
-- =0A=
Damien Le Moal=0A=
Western Digital Research=0A=