Linux-RDMA Archive on lore.kernel.org
 help / color / Atom feed
From: Yishai Hadas <yishaih@mellanox.com>
To: linux-rdma@vger.kernel.org, jgg@mellanox.com, dledford@redhat.com
Cc: yishaih@mellanox.com, maorg@mellanox.com, michaelgur@mellanox.com
Subject: [PATCH rdma-next 03/14] RDMA/mlx5: Simplify devx async commands
Date: Wed,  8 Jan 2020 19:21:55 +0200
Message-ID: <1578504126-9400-4-git-send-email-yishaih@mellanox.com> (raw)
In-Reply-To: <1578504126-9400-1-git-send-email-yishaih@mellanox.com>

From: Jason Gunthorpe <jgg@mellanox.com>

With the new FD structure the async commands do not need to hold any
references while running. The existing mlx5_cmd_exec_cb() and
mlx5_cmd_cleanup_async_ctx() provide enough synchronization to ensure
that all outstanding commands are completed before the uobject can be
destructed.

Remove the now confusing get_file() and the type erasure of the
devx_async_cmd_event_file.

Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/mlx5/devx.c | 24 ++++++++++--------------
 1 file changed, 10 insertions(+), 14 deletions(-)

diff --git a/drivers/infiniband/hw/mlx5/devx.c b/drivers/infiniband/hw/mlx5/devx.c
index 02125d8..7bb91d2 100644
--- a/drivers/infiniband/hw/mlx5/devx.c
+++ b/drivers/infiniband/hw/mlx5/devx.c
@@ -30,7 +30,7 @@ enum devx_obj_flags {
 struct devx_async_data {
 	struct mlx5_ib_dev *mdev;
 	struct list_head list;
-	struct ib_uobject *fd_uobj;
+	struct devx_async_cmd_event_file *ev_file;
 	struct mlx5_async_work cb_work;
 	u16 cmd_out_len;
 	/* must be last field in this structure */
@@ -1673,21 +1673,20 @@ static void devx_query_callback(int status, struct mlx5_async_work *context)
 {
 	struct devx_async_data *async_data =
 		container_of(context, struct devx_async_data, cb_work);
-	struct ib_uobject *fd_uobj = async_data->fd_uobj;
-	struct devx_async_cmd_event_file *ev_file;
-	struct devx_async_event_queue *ev_queue;
+	struct devx_async_cmd_event_file *ev_file = async_data->ev_file;
+	struct devx_async_event_queue *ev_queue = &ev_file->ev_queue;
 	unsigned long flags;
 
-	ev_file = container_of(fd_uobj, struct devx_async_cmd_event_file,
-			       uobj);
-	ev_queue = &ev_file->ev_queue;
-
+	/*
+	 * Note that if the struct devx_async_cmd_event_file uobj begins to be
+	 * destroyed it will block at mlx5_cmd_cleanup_async_ctx() until this
+	 * routine returns, ensuring that it always remains valid here.
+	 */
 	spin_lock_irqsave(&ev_queue->lock, flags);
 	list_add_tail(&async_data->list, &ev_queue->event_list);
 	spin_unlock_irqrestore(&ev_queue->lock, flags);
 
 	wake_up_interruptible(&ev_queue->poll_wait);
-	fput(fd_uobj->object);
 }
 
 #define MAX_ASYNC_BYTES_IN_USE (1024 * 1024) /* 1MB */
@@ -1756,9 +1755,8 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_OBJ_ASYNC_QUERY)(
 
 	async_data->cmd_out_len = cmd_out_len;
 	async_data->mdev = mdev;
-	async_data->fd_uobj = fd_uobj;
+	async_data->ev_file = ev_file;
 
-	get_file(fd_uobj->object);
 	MLX5_SET(general_obj_in_cmd_hdr, cmd_in, uid, uid);
 	err = mlx5_cmd_exec_cb(&ev_file->async_ctx, cmd_in,
 		    uverbs_attr_get_len(attrs,
@@ -1768,12 +1766,10 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_OBJ_ASYNC_QUERY)(
 		    devx_query_callback, &async_data->cb_work);
 
 	if (err)
-		goto cb_err;
+		goto free_async;
 
 	return 0;
 
-cb_err:
-	fput(fd_uobj->object);
 free_async:
 	kvfree(async_data);
 sub_bytes:
-- 
1.8.3.1


  parent reply index

Thread overview: 17+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-01-08 17:21 [PATCH rdma-next 00/14] Refactoring FD usage Yishai Hadas
2020-01-08 17:21 ` [PATCH rdma-next 01/14] RDMA/mlx5: Use RCU and direct refcounts to keep memory alive Yishai Hadas
2020-01-08 17:21 ` [PATCH rdma-next 02/14] RDMA/core: Simplify destruction of FD uobjects Yishai Hadas
2020-01-08 17:21 ` Yishai Hadas [this message]
2020-01-08 17:21 ` [PATCH rdma-next 04/14] RDMA/core: Do not allow alloc_commit to fail Yishai Hadas
2020-01-08 17:21 ` [PATCH rdma-next 05/14] RDMA/core: Make ib_ucq_object use ib_uevent_object Yishai Hadas
2020-01-08 17:21 ` [PATCH rdma-next 06/14] RDMA/core: Do not erase the type of ib_cq.uobject Yishai Hadas
2020-01-08 17:21 ` [PATCH rdma-next 07/14] RDMA/core: Do not erase the type of ib_qp.uobject Yishai Hadas
2020-01-08 17:22 ` [PATCH rdma-next 08/14] RDMA/core: Do not erase the type of ib_srq.uobject Yishai Hadas
2020-01-08 17:22 ` [PATCH rdma-next 09/14] RDMA/core: Do not erase the type of ib_wq.uobject Yishai Hadas
2020-01-08 17:22 ` [PATCH rdma-next 10/14] RDMA/core: Simplify type usage for ib_uverbs_async_handler() Yishai Hadas
2020-01-08 17:22 ` [PATCH rdma-next 11/14] RDMA/core: Fix locking in ib_uverbs_event_read Yishai Hadas
2020-01-10 14:54   ` Jason Gunthorpe
2020-01-08 17:22 ` [PATCH rdma-next 12/14] RDMA/core: Remove the ufile arg from rdma_alloc_begin_uobject Yishai Hadas
2020-01-08 17:22 ` [PATCH rdma-next 13/14] RDMA/core: Make ib_uverbs_async_event_file into a uobject Yishai Hadas
2020-01-08 17:22 ` [PATCH rdma-next 14/14] RDMA/core: Use READ_ONCE for ib_ufile.async_file Yishai Hadas
2020-01-13 20:37 ` [PATCH rdma-next 00/14] Refactoring FD usage Jason Gunthorpe

Reply instructions:

You may reply publically 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=1578504126-9400-4-git-send-email-yishaih@mellanox.com \
    --to=yishaih@mellanox.com \
    --cc=dledford@redhat.com \
    --cc=jgg@mellanox.com \
    --cc=linux-rdma@vger.kernel.org \
    --cc=maorg@mellanox.com \
    --cc=michaelgur@mellanox.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

Linux-RDMA Archive on lore.kernel.org

Archives are clonable:
	git clone --mirror https://lore.kernel.org/linux-rdma/0 linux-rdma/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 linux-rdma linux-rdma/ https://lore.kernel.org/linux-rdma \
		linux-rdma@vger.kernel.org
	public-inbox-index linux-rdma

Example config snippet for mirrors

Newsgroup available over NNTP:
	nntp://nntp.lore.kernel.org/org.kernel.vger.linux-rdma


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git