diff options
| author | Yishai Hadas <yishaih@nvidia.com> | 2021-02-02 10:13:09 +0300 |
|---|---|---|
| committer | Jason Gunthorpe <jgg@nvidia.com> | 2021-02-09 03:31:11 +0300 |
| commit | db72438c9319cfd37e3c237a7754ca862ae12d63 (patch) | |
| tree | c5fe461d002f4fb853f9b944da24d4423cc66bec /drivers/infiniband/hw/mlx5/devx.c | |
| parent | a5887d62072e09a9d3b01549ac5d555bbeb9be3d (diff) | |
| download | linux-db72438c9319cfd37e3c237a7754ca862ae12d63.tar.xz | |
RDMA/mlx5: Cleanup the synchronize_srcu() from the ODP flow
Cleanup the synchronize_srcu() from the ODP flow as it was found to be a
very heavy time consumer as part of dereg_mr.
For example de-registration of 10000 ODP MRs each with size of 2M hugepage
took 19.6 sec comparing de-registration of same number of non ODP MRs that
took 172 ms.
The new locking scheme uses the wait_event() mechanism which follows the
use count of the MR instead of using synchronize_srcu().
By that change, the time required for the above test took 95 ms which is
even better than the non ODP flow.
Once fully dropped the srcu usage, had to come with a lock to protect the
XA access.
As part of using the above mechanism we could also clean the
num_deferred_work stuff and follow the use count instead.
Link: https://lore.kernel.org/r/20210202071309.2057998-1-leon@kernel.org
Signed-off-by: Yishai Hadas <yishaih@nvidia.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Diffstat (limited to 'drivers/infiniband/hw/mlx5/devx.c')
| -rw-r--r-- | drivers/infiniband/hw/mlx5/devx.c | 13 |
1 files changed, 6 insertions, 7 deletions
diff --git a/drivers/infiniband/hw/mlx5/devx.c b/drivers/infiniband/hw/mlx5/devx.c index e6d2709eb09e..e39661db0d2f 100644 --- a/drivers/infiniband/hw/mlx5/devx.c +++ b/drivers/infiniband/hw/mlx5/devx.c @@ -1310,9 +1310,9 @@ static int devx_handle_mkey_indirect(struct devx_obj *obj, mkey->size = MLX5_GET64(mkc, mkc, len); mkey->pd = MLX5_GET(mkc, mkc, pd); devx_mr->ndescs = MLX5_GET(mkc, mkc, translations_octword_size); + init_waitqueue_head(&mkey->wait); - return xa_err(xa_store(&dev->odp_mkeys, mlx5_base_mkey(mkey->key), mkey, - GFP_KERNEL)); + return mlx5r_store_odp_mkey(dev, mkey); } static int devx_handle_mkey_create(struct mlx5_ib_dev *dev, @@ -1385,16 +1385,15 @@ static int devx_obj_cleanup(struct ib_uobject *uobject, int ret; dev = mlx5_udata_to_mdev(&attrs->driver_udata); - if (obj->flags & DEVX_OBJ_FLAGS_INDIRECT_MKEY) { + if (obj->flags & DEVX_OBJ_FLAGS_INDIRECT_MKEY && + xa_erase(&obj->ib_dev->odp_mkeys, + mlx5_base_mkey(obj->devx_mr.mmkey.key))) /* * The pagefault_single_data_segment() does commands against * the mmkey, we must wait for that to stop before freeing the * mkey, as another allocation could get the same mkey #. */ - xa_erase(&obj->ib_dev->odp_mkeys, - mlx5_base_mkey(obj->devx_mr.mmkey.key)); - synchronize_srcu(&dev->odp_srcu); - } + mlx5r_deref_wait_odp_mkey(&obj->devx_mr.mmkey); if (obj->flags & DEVX_OBJ_FLAGS_DCT) ret = mlx5_core_destroy_dct(obj->ib_dev, &obj->core_dct); |