On 6/24/2019 3:04 PM, Jason Gunthorpe wrote:
On Tue, Jun 18, 2019 at 08:15:40PM +0300, Leon Romanovsky wrote:
From: Yishai Hadas <yishaih@xxxxxxxxxxxx>
Add DEVX support for CQ events by creating and destroying the CQ via
mlx5_core and set an handler to manage its completions.
Signed-off-by: Yishai Hadas <yishaih@xxxxxxxxxxxx>
Signed-off-by: Leon Romanovsky <leonro@xxxxxxxxxxxx>
drivers/infiniband/hw/mlx5/devx.c | 40 +++++++++++++++++++++++++++++++
1 file changed, 40 insertions(+)
diff --git a/drivers/infiniband/hw/mlx5/devx.c b/drivers/infiniband/hw/mlx5/devx.c
index 49fdce95d6d9..91ccd58ebc05 100644
+++ b/drivers/infiniband/hw/mlx5/devx.c
@@ -19,9 +19,12 @@
#define UVERBS_MODULE_NAME mlx5_ib
#include <rdma/uverbs_named_ioctl.h>
+static void dispatch_event_fd(struct list_head *fd_list, const void *data);
+
enum devx_obj_flags {
DEVX_OBJ_FLAGS_INDIRECT_MKEY = 1 << 0,
DEVX_OBJ_FLAGS_DCT = 1 << 1,
+ DEVX_OBJ_FLAGS_CQ = 1 << 2,
};
struct devx_async_data {
@@ -94,6 +97,7 @@ struct devx_async_event_file {
#define MLX5_MAX_DESTROY_INBOX_SIZE_DW MLX5_ST_SZ_DW(delete_fte_in)
struct devx_obj {
struct mlx5_core_dev *mdev;
+ struct mlx5_ib_dev *ib_dev;
This seems strange, why would we need to store the core_dev and the ib_dev
in a struct when ibdev->mdev == core_dev?
We need to add the ib_dev as we can't access it from the core_dev.
Most of this patch we can probably go and drop the mdev and access it
from ib_dev, I preferred to not handle that in this patch.
