Re: [PATCH rdma-next 4/5] IB/mlx5: Introduce VAR object and its alloc/destroy methods

Yishai Hadas <yishaih@xxxxxxxxxxxxxxxxxx> · Wed, 8 Jan 2020 10:12:20 +0200

On 1/7/2020 9:36 PM, Jason Gunthorpe wrote:
On Thu, Dec 12, 2019 at 01:09:27PM +0200, Leon Romanovsky wrote:
From: Yishai Hadas <yishaih@xxxxxxxxxxxx>

Introduce VAR object and its alloc/destroy KABI methods. The internal
implementation uses the IB core API to manage mmap/munamp calls.

Signed-off-by: Yishai Hadas <yishaih@xxxxxxxxxxxx>
Signed-off-by: Leon Romanovsky <leonro@xxxxxxxxxxxx>
  drivers/infiniband/hw/mlx5/main.c        | 157 +++++++++++++++++++++++
  drivers/infiniband/hw/mlx5/mlx5_ib.h     |   7 +
  include/uapi/rdma/mlx5_user_ioctl_cmds.h |  17 +++
  3 files changed, 181 insertions(+)

diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 79a5b8824b9d..873480b07686 100644
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -2078,6 +2078,7 @@ static void mlx5_ib_mmap_free(struct rdma_user_mmap_entry *entry)
  {
  	struct mlx5_user_mmap_entry *mentry = to_mmmap(entry);
  	struct mlx5_ib_dev *dev = to_mdev(entry->ucontext->device);
+	struct mlx5_var_table *var_table = &dev->var_table;
  	struct mlx5_ib_dm *mdm;
  
  	switch (mentry->mmap_flag) {
@@ -2087,6 +2088,12 @@ static void mlx5_ib_mmap_free(struct rdma_user_mmap_entry *entry)
  				       mdm->size);
  		kfree(mdm);
  		break;
+	case MLX5_IB_MMAP_TYPE_VAR:
+		mutex_lock(&var_table->bitmap_lock);
+		clear_bit(mentry->page_idx, var_table->bitmap);
+		mutex_unlock(&var_table->bitmap_lock);
+		kfree(mentry);
+		break;
  	default:
  		WARN_ON(true);
  	}
@@ -2255,6 +2262,15 @@ static int mlx5_ib_mmap_offset(struct mlx5_ib_dev *dev,
  	return ret;
  }
  
+static u64 mlx5_entry_to_mmap_offset(struct mlx5_user_mmap_entry *entry)
+{
+	u16 cmd = entry->rdma_entry.start_pgoff >> 16;
+	u16 index = entry->rdma_entry.start_pgoff & 0xFFFF;
+
+	return (((index >> 8) << 16) | (cmd << MLX5_IB_MMAP_CMD_SHIFT) |
+		(index & 0xFF)) << PAGE_SHIFT;
+}
+
  static int mlx5_ib_mmap(struct ib_ucontext *ibcontext, struct vm_area_struct *vma)
  {
  	struct mlx5_ib_ucontext *context = to_mucontext(ibcontext);
@@ -6034,6 +6050,145 @@ static void mlx5_ib_cleanup_multiport_master(struct mlx5_ib_dev *dev)
  	mlx5_nic_vport_disable_roce(dev->mdev);
  }
  
+static int var_obj_cleanup(struct ib_uobject *uobject,
+			   enum rdma_remove_reason why,
+			   struct uverbs_attr_bundle *attrs)
+{
+	struct mlx5_user_mmap_entry *obj = uobject->object;
+
+	rdma_user_mmap_entry_remove(&obj->rdma_entry);
+	return 0;
+}
+
+static struct mlx5_user_mmap_entry *
+alloc_var_entry(struct mlx5_ib_ucontext *c)
+{
+	struct mlx5_user_mmap_entry *entry;
+	struct mlx5_var_table *var_table;
+	u32 page_idx;
+	int err;
+
+	var_table = &to_mdev(c->ibucontext.device)->var_table;
+	entry = kzalloc(sizeof(*entry), GFP_KERNEL);
+	if (!entry)
+		return ERR_PTR(-ENOMEM);
+
+	mutex_lock(&var_table->bitmap_lock);
+	page_idx = find_first_zero_bit(var_table->bitmap,
+				       var_table->num_var_hw_entries);
+	if (page_idx >= var_table->num_var_hw_entries) {
+		err = -ENOSPC;
+		mutex_unlock(&var_table->bitmap_lock);
+		goto end;
+	}
+
+	set_bit(page_idx, var_table->bitmap);
+	mutex_unlock(&var_table->bitmap_lock);
+
+	entry->address = var_table->hw_start_addr +
+				(page_idx * var_table->stride_size);
+	entry->page_idx = page_idx;
+	entry->mmap_flag = MLX5_IB_MMAP_TYPE_VAR;
+
+	err = rdma_user_mmap_entry_insert_range(
+		&c->ibucontext, &entry->rdma_entry, var_table->stride_size,
+		MLX5_IB_MMAP_OFFSET_START << 16,
+		(MLX5_IB_MMAP_OFFSET_END << 16) + (1UL << 16) - 1);
+	if (err)
+		goto err_insert;
+
+	return entry;
+
+err_insert:
+	mutex_lock(&var_table->bitmap_lock);
+	clear_bit(page_idx, var_table->bitmap);
+	mutex_unlock(&var_table->bitmap_lock);
+end:
+	kfree(entry);
+	return ERR_PTR(err);
+}
+
+static int UVERBS_HANDLER(MLX5_IB_METHOD_VAR_OBJ_ALLOC)(
+	struct uverbs_attr_bundle *attrs)
+{
+	struct ib_uobject *uobj = uverbs_attr_get_uobject(
+		attrs, MLX5_IB_ATTR_VAR_OBJ_ALLOC_HANDLE);
+	struct mlx5_ib_ucontext *c;
+	struct mlx5_user_mmap_entry *entry;
+	u64 mmap_offset;
+	u32 length;
+	int err;
+
+	c = to_mucontext(ib_uverbs_get_ucontext(attrs));
+	if (IS_ERR(c))
+		return PTR_ERR(c);
+
+	entry = alloc_var_entry(c);
+	if (IS_ERR(entry))
+		return PTR_ERR(entry);
+
+	mmap_offset = mlx5_entry_to_mmap_offset(entry);
+	length = entry->rdma_entry.npages * PAGE_SIZE;
+	uobj->object = entry;
+
+	err = uverbs_copy_to(attrs, MLX5_IB_ATTR_VAR_OBJ_ALLOC_MMAP_OFFSET,
+			     &mmap_offset, sizeof(mmap_offset));
+	if (err)
+		goto err;
+
+	err = uverbs_copy_to(attrs, MLX5_IB_ATTR_VAR_OBJ_ALLOC_PAGE_ID,
+			     &entry->page_idx, sizeof(entry->page_idx));
+	if (err)
+		goto err;
+
+	err = uverbs_copy_to(attrs, MLX5_IB_ATTR_VAR_OBJ_ALLOC_MMAP_LENGTH,
+			     &length, sizeof(length));
+	if (err)
+		goto err;
+
+	return 0;
+
+err:
+	rdma_user_mmap_entry_remove(&entry->rdma_entry);
+	return err;
+}
+
+DECLARE_UVERBS_NAMED_METHOD(
+	MLX5_IB_METHOD_VAR_OBJ_ALLOC,
+	UVERBS_ATTR_IDR(MLX5_IB_ATTR_VAR_OBJ_ALLOC_HANDLE,
+			MLX5_IB_OBJECT_VAR,
+			UVERBS_ACCESS_NEW,
+			UA_MANDATORY),
+	UVERBS_ATTR_PTR_OUT(MLX5_IB_ATTR_VAR_OBJ_ALLOC_PAGE_ID,
+			   UVERBS_ATTR_TYPE(u32),
+			   UA_MANDATORY),
+	UVERBS_ATTR_PTR_OUT(MLX5_IB_ATTR_VAR_OBJ_ALLOC_MMAP_LENGTH,
+			   UVERBS_ATTR_TYPE(u32),
+			   UA_MANDATORY),
+	UVERBS_ATTR_PTR_OUT(MLX5_IB_ATTR_VAR_OBJ_ALLOC_MMAP_OFFSET,
+			    UVERBS_ATTR_TYPE(u64),
+			    UA_MANDATORY));
+
+DECLARE_UVERBS_NAMED_METHOD_DESTROY(
+	MLX5_IB_METHOD_VAR_OBJ_DESTROY,
+	UVERBS_ATTR_IDR(MLX5_IB_ATTR_VAR_OBJ_DESTROY_HANDLE,
+			MLX5_IB_OBJECT_VAR,
+			UVERBS_ACCESS_DESTROY,
+			UA_MANDATORY));
+
+DECLARE_UVERBS_NAMED_OBJECT(MLX5_IB_OBJECT_VAR,
+			    UVERBS_TYPE_ALLOC_IDR(var_obj_cleanup),
+			    &UVERBS_METHOD(MLX5_IB_METHOD_VAR_OBJ_ALLOC),
+			    &UVERBS_METHOD(MLX5_IB_METHOD_VAR_OBJ_DESTROY));
+
+static bool var_is_supported(struct ib_device *device)
+{
+	struct mlx5_ib_dev *dev = to_mdev(device);
+
+	return (MLX5_CAP_GEN_64(dev->mdev, general_obj_types) &
+			MLX5_GENERAL_OBJ_TYPES_CAP_VIRTIO_NET_Q);
+}
+
  ADD_UVERBS_ATTRIBUTES_SIMPLE(
  	mlx5_ib_dm,
  	UVERBS_OBJECT_DM,
@@ -6064,6 +6219,8 @@ static const struct uapi_definition mlx5_ib_defs[] = {
  	UAPI_DEF_CHAIN_OBJ_TREE(UVERBS_OBJECT_FLOW_ACTION,
  				&mlx5_ib_flow_action),
  	UAPI_DEF_CHAIN_OBJ_TREE(UVERBS_OBJECT_DM, &mlx5_ib_dm),
+	UAPI_DEF_CHAIN_OBJ_TREE_NAMED(MLX5_IB_OBJECT_VAR,
+				UAPI_DEF_IS_OBJ_SUPPORTED(var_is_supported)),
  	{}
  };
  
diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h
index 23ad949e247f..489128fe8603 100644
+++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h
@@ -71,6 +71,11 @@
  
  #define MLX5_MKEY_PAGE_SHIFT_MASK __mlx5_mask(mkc, log_page_size)
  
+enum {
+	MLX5_IB_MMAP_OFFSET_START = 9,
+	MLX5_IB_MMAP_OFFSET_END = 255,
+};
+
  enum {
  	MLX5_IB_MMAP_CMD_SHIFT	= 8,
  	MLX5_IB_MMAP_CMD_MASK	= 0xff,
@@ -120,6 +125,7 @@ enum {
  
  enum mlx5_ib_mmap_type {
  	MLX5_IB_MMAP_TYPE_MEMIC = 1,
+	MLX5_IB_MMAP_TYPE_VAR = 2,
  };
  
  #define MLX5_LOG_SW_ICM_BLOCK_SIZE(dev)                                        \
@@ -563,6 +569,7 @@ struct mlx5_user_mmap_entry {
  	struct rdma_user_mmap_entry rdma_entry;
  	u8 mmap_flag;
  	u64 address;
+	u32 page_idx;

Why are we storing this in the global struct when it is never read
except by the caller of alloc_var_entry()? Return it from
alloc_var_entry?


It's required as part of mlx5_ib_mmap_free() to claer the matching bit 
map entry of the device var table, see above in this patch.

Also the final patch in the series should be here as at this point
mmap will succeed but return the wrong cachability flags.


Right, let's squash it to this patch.

Since Leon is away I can fix this two things if you agree.

Yes, thanks.

Yishai






