From: John Garry <john.garry@huawei.com>
To: Kashyap Desai <kashyap.desai@broadcom.com>, <linux-scsi@vger.kernel.org>
Cc: <sumit.saxena@broadcom.com>, <chandrakanth.patil@broadcom.com>,
<linux-block@vger.kernel.org>
Subject: Re: [PATCH v1 1/3] add io_uring with IOPOLL support in scsi layer
Date: Fri, 13 Nov 2020 11:51:05 +0000 [thread overview]
Message-ID: <0531d781-38ed-0098-d5b8-727a3e143dde@huawei.com> (raw)
In-Reply-To: <20201015133633.61836-1-kashyap.desai@broadcom.com>
On 15/10/2020 14:36, Kashyap Desai wrote:
> io_uring with IOPOLL is not currently supported in scsi mid layer.
> Outside of that everything else should work and no extra support in the driver is needed.
> Currently io_uring with IOPOLL support is only available in block layer.
> This patch is to extend support of mq_poll in scsi layer. >
> Signed-off-by: Kashyap Desai <kashyap.desai@broadcom.com>
> Cc: sumit.saxena@broadcom.com
> Cc: chandrakanth.patil@broadcom.com
> Cc: linux-block@vger.kernel.org
>
> ---
> drivers/scsi/scsi_lib.c | 16 ++++++++++++++++
> include/scsi/scsi_cmnd.h | 1 +
> include/scsi/scsi_host.h | 11 +++++++++++
> 3 files changed, 28 insertions(+)
>
> diff --git a/drivers/scsi/scsi_lib.c b/drivers/scsi/scsi_lib.c
> index 72b12102f777..5a3c383a2bb3 100644
> --- a/drivers/scsi/scsi_lib.c
> +++ b/drivers/scsi/scsi_lib.c
> @@ -1766,6 +1766,19 @@ static void scsi_mq_exit_request(struct blk_mq_tag_set *set, struct request *rq,
> cmd->sense_buffer);
> }
>
> +
> +static int scsi_mq_poll(struct blk_mq_hw_ctx *hctx)
> +{
> + struct request_queue *q = hctx->queue;
> + struct scsi_device *sdev = q->queuedata;
> + struct Scsi_Host *shost = sdev->host;
could we separately set hctx->driver_data = shost or similar for a
quicker lookup? I don't see hctx->driver_data set for SCSI currently.
Going through the scsi_device looks strange - I know that it is done in
scsi_commit_rqs.
> +
> + if (shost->hostt->mq_poll)
to avoid this check, could we reject if .mq_poll is not set and
HCTX_TYPE_POLL is?
> + return shost->hostt->mq_poll(shost, hctx->queue_num);
> +
> + return 0;
> +}
> +
> static int scsi_map_queues(struct blk_mq_tag_set *set)
> {
> struct Scsi_Host *shost = container_of(set, struct Scsi_Host, tag_set);
> @@ -1833,6 +1846,7 @@ static const struct blk_mq_ops scsi_mq_ops_no_commit = {
> .cleanup_rq = scsi_cleanup_rq,
> .busy = scsi_mq_lld_busy,
> .map_queues = scsi_map_queues,
> + .poll = scsi_mq_poll,
> };
>
>
> @@ -1861,6 +1875,7 @@ static const struct blk_mq_ops scsi_mq_ops = {
> .cleanup_rq = scsi_cleanup_rq,
> .busy = scsi_mq_lld_busy,
> .map_queues = scsi_map_queues,
> + .poll = scsi_mq_poll,
> };
>
> struct request_queue *scsi_mq_alloc_queue(struct scsi_device *sdev)
> @@ -1893,6 +1908,7 @@ int scsi_mq_setup_tags(struct Scsi_Host *shost)
> else
> tag_set->ops = &scsi_mq_ops_no_commit;
> tag_set->nr_hw_queues = shost->nr_hw_queues ? : 1;
> + tag_set->nr_maps = shost->nr_maps ? : 1;
> tag_set->queue_depth = shost->can_queue;
> tag_set->cmd_size = cmd_size;
> tag_set->numa_node = NUMA_NO_NODE;
> diff --git a/include/scsi/scsi_cmnd.h b/include/scsi/scsi_cmnd.h
> index e76bac4d14c5..5844374a85b1 100644
> --- a/include/scsi/scsi_cmnd.h
> +++ b/include/scsi/scsi_cmnd.h
> @@ -9,6 +9,7 @@
> #include <linux/types.h>
> #include <linux/timer.h>
> #include <linux/scatterlist.h>
> +#include <scsi/scsi_host.h>
can we maintain alphabetic ordering?
> #include <scsi/scsi_device.h>
> #include <scsi/scsi_request.h>
>
> diff --git a/include/scsi/scsi_host.h b/include/scsi/scsi_host.h
> index 701f178b20ae..905ee6b00c55 100644
> --- a/include/scsi/scsi_host.h
> +++ b/include/scsi/scsi_host.h
> @@ -270,6 +270,16 @@ struct scsi_host_template {
> */
> int (* map_queues)(struct Scsi_Host *shost);
>
> + /*
> + * SCSI interface of blk_poll - poll for IO completions.
> + * Possible interface only if scsi LLD expose multiple h/w queues.
> + *
> + * Return values: Number of completed entries found.
/s/values/value/
> + *
> + * Status: OPTIONAL
> + */
> + int (* mq_poll)(struct Scsi_Host *shost, unsigned int queue_num);
> +
> /*
> * Check if scatterlists need to be padded for DMA draining.
> *
> @@ -610,6 +620,7 @@ struct Scsi_Host {
> * the total queue depth is can_queue.
> */
> unsigned nr_hw_queues;
> + unsigned nr_maps; > unsigned active_mode:2;
> unsigned unchecked_isa_dma:1;
>
>
next prev parent reply other threads:[~2020-11-13 12:12 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-10-15 13:36 [PATCH v1 1/3] add io_uring with IOPOLL support in scsi layer Kashyap Desai
2020-11-13 11:51 ` John Garry [this message]
2020-11-30 7:41 ` Kashyap Desai
2020-12-01 9:49 ` John Garry
2020-11-20 7:16 ` Hannes Reinecke
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=0531d781-38ed-0098-d5b8-727a3e143dde@huawei.com \
--to=john.garry@huawei.com \
--cc=chandrakanth.patil@broadcom.com \
--cc=kashyap.desai@broadcom.com \
--cc=linux-block@vger.kernel.org \
--cc=linux-scsi@vger.kernel.org \
--cc=sumit.saxena@broadcom.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).