[PATCH rdma-next v1 0/5] Cleanup of CONFIG_INFINIBAND_ON_DEMAND_PAGING usage

[PATCH rdma-next v1 0/5] Cleanup of CONFIG_INFINIBAND_ON_DEMAND_PAGING usage

[Leon Romanovsky]

From: Leon Romanovsky <leonro@mellanox.com>

Changelog v0->v1:
 * Don't set ODP caps always (patch #2)
 * Set invalidate_range callback based on device_cap only

-------------------------------------------------------------------------------

Hi,

As a followup to Jason's request to rethink CONFIG_INFINIBAND_ON_DEMAND_PAGING
usage, this series cleans mlx5_ib and RDMA/core code and it is based on accepted
patch https://patchwork.kernel.org/patch/10735547/

I think that we will see more patches in that area, I would like to
remove CONFIG_INFINIBAND_ON_DEMAND_PAGING completely.

Thanks

Leon Romanovsky (5):
  RDMA: Clean structures from CONFIG_INFINIBAND_ON_DEMAND_PAGING
  RDMA/core: Don't depend device ODP capabilities on kconfig option
  RDMA/mlx5: Introduce and reuse helper to identify ODP MR
  RDMA/mlx5: Embed into the code flow the ODP config option
  RDMA/mlx5: Delete declaration of already removed function

 drivers/infiniband/core/uverbs_cmd.c |  5 ---
 drivers/infiniband/hw/mlx5/main.c    | 51 ++++++++++++++--------------
 drivers/infiniband/hw/mlx5/mem.c     |  5 +--
 drivers/infiniband/hw/mlx5/mlx5_ib.h | 12 +++++--
 drivers/infiniband/hw/mlx5/mr.c      | 45 +++++++-----------------
 drivers/infiniband/hw/mlx5/odp.c     |  6 ++--
 include/linux/mlx5/driver.h          |  4 ---
 include/rdma/ib_umem_odp.h           | 26 +++++++-------
 include/rdma/ib_verbs.h              |  2 --
 9 files changed, 65 insertions(+), 91 deletions(-)

[PATCH rdma-next v1 1/5] RDMA: Clean structures from CONFIG_INFINIBAND_ON_DEMAND_PAGING

[Leon Romanovsky]

From: Leon Romanovsky <leonro@mellanox.com>

CONFIG_INFINIBAND_ON_DEMAND_PAGING is used in general structures
to micro-optimize the memory footprint. Remove it, so it will allow
us to simplify various ODP device flows.

Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
---
 drivers/infiniband/hw/mlx5/mlx5_ib.h | 2 --
 include/rdma/ib_verbs.h              | 2 --
 2 files changed, 4 deletions(-)

diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h
index d531f25a1105..58efcbd596fd 100644
--- a/drivers/infiniband/hw/mlx5/mlx5_ib.h
+++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h
@@ -911,7 +911,6 @@ struct mlx5_ib_dev {
 	/* Prevents soft lock on massive reg MRs */
 	struct mutex			slow_path_mutex;
 	int				fill_delay;
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
 	struct ib_odp_caps	odp_caps;
 	u64			odp_max_size;
 	struct mlx5_ib_pf_eq	odp_pf_eq;
@@ -923,7 +922,6 @@ struct mlx5_ib_dev {
 	struct srcu_struct      mr_srcu;
 	u32			null_mkey;
 	struct workqueue_struct *advise_mr_wq;
-#endif
 	struct mlx5_ib_flow_db	*flow_db;
 	/* protect resources needed as part of reset flow */
 	spinlock_t		reset_flow_resource_lock;
diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index a6e57f05410c..67339da6c72d 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -1504,12 +1504,10 @@ struct ib_ucontext {
 
 	bool cleanup_retryable;
 
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
 	void (*invalidate_range)(struct ib_umem_odp *umem_odp,
 				 unsigned long start, unsigned long end);
 	struct mutex per_mm_list_lock;
 	struct list_head per_mm_list;
-#endif
 
 	struct ib_rdmacg_object	cg_obj;
 	/*
-- 
2.19.1

[PATCH rdma-next v1 2/5] RDMA/core: Don't depend device ODP capabilities on kconfig option

[Leon Romanovsky]

From: Leon Romanovsky <leonro@mellanox.com>

Device capability bits are exposing what specific device supports from
HW perspective. Those bits are not dependent on kernel configurations
and RDMA/core should ensure that proper interfaces to users will be
disabled if CONFIG_INFINIBAND_ON_DEMAND_PAGING is not set.

Fixes: f4056bfd8ccf ("IB/core: Add on demand paging caps to ib_uverbs_ex_query_device")
Fixes: 8cdd312cfed7 ("IB/mlx5: Implement the ODP capability query verb")
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
---
 drivers/infiniband/core/uverbs_cmd.c |  2 --
 drivers/infiniband/hw/mlx5/main.c    | 10 +++++-----
 2 files changed, 5 insertions(+), 7 deletions(-)

diff --git a/drivers/infiniband/core/uverbs_cmd.c b/drivers/infiniband/core/uverbs_cmd.c
index 6b12cc5f97b2..549d9eedf62e 100644
--- a/drivers/infiniband/core/uverbs_cmd.c
+++ b/drivers/infiniband/core/uverbs_cmd.c
@@ -3609,7 +3609,6 @@ static int ib_uverbs_ex_query_device(struct uverbs_attr_bundle *attrs)
 
 	copy_query_dev_fields(ucontext, &resp.base, &attr);
 
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
 	resp.odp_caps.general_caps = attr.odp_caps.general_caps;
 	resp.odp_caps.per_transport_caps.rc_odp_caps =
 		attr.odp_caps.per_transport_caps.rc_odp_caps;
@@ -3617,7 +3616,6 @@ static int ib_uverbs_ex_query_device(struct uverbs_attr_bundle *attrs)
 		attr.odp_caps.per_transport_caps.uc_odp_caps;
 	resp.odp_caps.per_transport_caps.ud_odp_caps =
 		attr.odp_caps.per_transport_caps.ud_odp_caps;
-#endif
 
 	resp.timestamp_mask = attr.timestamp_mask;
 	resp.hca_core_clock = attr.hca_core_clock;
diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index d84b21c95ae5..c98343e50a00 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -923,11 +923,11 @@ static int mlx5_ib_query_device(struct ib_device *ibdev,
 	props->hca_core_clock = MLX5_CAP_GEN(mdev, device_frequency_khz);
 	props->timestamp_mask = 0x7FFFFFFFFFFFFFFFULL;
 
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
-	if (MLX5_CAP_GEN(mdev, pg))
-		props->device_cap_flags |= IB_DEVICE_ON_DEMAND_PAGING;
-	props->odp_caps = dev->odp_caps;
-#endif
+	if (IS_ENABLED(CONFIG_INFINIBAND_ON_DEMAND_PAGING)) {
+		if (MLX5_CAP_GEN(mdev, pg))
+			props->device_cap_flags |= IB_DEVICE_ON_DEMAND_PAGING;
+		props->odp_caps = dev->odp_caps;
+	}
 
 	if (MLX5_CAP_GEN(mdev, cd))
 		props->device_cap_flags |= IB_DEVICE_CROSS_CHANNEL;
-- 
2.19.1

[PATCH rdma-next v1 3/5] RDMA/mlx5: Introduce and reuse helper to identify ODP MR

[Leon Romanovsky]

From: Leon Romanovsky <leonro@mellanox.com>

Consolidate various checks if MR is ODP backed to one simple
helper and update call sites to use it.

Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
---
 drivers/infiniband/hw/mlx5/mlx5_ib.h | 10 ++++++++++
 drivers/infiniband/hw/mlx5/mr.c      | 29 ++++++----------------------
 drivers/infiniband/hw/mlx5/odp.c     |  6 +++---
 3 files changed, 19 insertions(+), 26 deletions(-)

diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h
index 58efcbd596fd..6563e531fdc2 100644
--- a/drivers/infiniband/hw/mlx5/mlx5_ib.h
+++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h
@@ -36,6 +36,7 @@
 #include <linux/kernel.h>
 #include <linux/sched.h>
 #include <rdma/ib_verbs.h>
+#include <rdma/ib_umem.h>
 #include <rdma/ib_smi.h>
 #include <linux/mlx5/driver.h>
 #include <linux/mlx5/cq.h>
@@ -589,6 +590,12 @@ struct mlx5_ib_mr {
 	wait_queue_head_t       q_leaf_free;
 };
 
+static inline bool is_odp_mr(struct mlx5_ib_mr *mr)
+{
+	return IS_ENABLED(CONFIG_INFINIBAND_ON_DEMAND_PAGING) && mr->umem &&
+	       mr->umem->is_odp;
+}
+
 struct mlx5_ib_mw {
 	struct ib_mw		ibmw;
 	struct mlx5_core_mkey	mmkey;
@@ -1212,6 +1219,9 @@ mlx5_ib_advise_mr_prefetch(struct ib_pd *pd,
 {
 	return -EOPNOTSUPP;
 }
+static inline void mlx5_ib_invalidate_range(struct ib_umem_odp *umem_odp,
+					    unsigned long start,
+					    unsigned long end){};
 #endif /* CONFIG_INFINIBAND_ON_DEMAND_PAGING */
 
 /* Needed for rep profile */
diff --git a/drivers/infiniband/hw/mlx5/mr.c b/drivers/infiniband/hw/mlx5/mr.c
index 1bd8c1b1dba1..b861b4a5b0e0 100644
--- a/drivers/infiniband/hw/mlx5/mr.c
+++ b/drivers/infiniband/hw/mlx5/mr.c
@@ -71,11 +71,9 @@ static int destroy_mkey(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr)
 {
 	int err = mlx5_core_destroy_mkey(dev->mdev, &mr->mmkey);
 
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
 	/* Wait until all page fault handlers using the mr complete. */
-	if (mr->umem && mr->umem->is_odp)
+	if (is_odp_mr(mr))
 		synchronize_srcu(&dev->mr_srcu);
-#endif
 
 	return err;
 }
@@ -96,10 +94,9 @@ static bool use_umr_mtt_update(struct mlx5_ib_mr *mr, u64 start, u64 length)
 		length + (start & (MLX5_ADAPTER_PAGE_SIZE - 1));
 }
 
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
 static void update_odp_mr(struct mlx5_ib_mr *mr)
 {
-	if (mr->umem->is_odp) {
+	if (is_odp_mr(mr)) {
 		/*
 		 * This barrier prevents the compiler from moving the
 		 * setting of umem->odp_data->private to point to our
@@ -122,7 +119,6 @@ static void update_odp_mr(struct mlx5_ib_mr *mr)
 		smp_wmb();
 	}
 }
-#endif
 
 static void reg_mr_callback(int status, void *context)
 {
@@ -238,9 +234,7 @@ static void remove_keys(struct mlx5_ib_dev *dev, int c, int num)
 {
 	struct mlx5_mr_cache *cache = &dev->cache;
 	struct mlx5_cache_ent *ent = &cache->ent[c];
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
 	bool odp_mkey_exist = false;
-#endif
 	struct mlx5_ib_mr *tmp_mr;
 	struct mlx5_ib_mr *mr;
 	LIST_HEAD(del_list);
@@ -253,10 +247,8 @@ static void remove_keys(struct mlx5_ib_dev *dev, int c, int num)
 			break;
 		}
 		mr = list_first_entry(&ent->head, struct mlx5_ib_mr, list);
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
-		if (mr->umem && mr->umem->is_odp)
+		if (is_odp_mr(mr))
 			odp_mkey_exist = true;
-#endif
 		list_move(&mr->list, &del_list);
 		ent->cur--;
 		ent->size--;
@@ -264,10 +256,8 @@ static void remove_keys(struct mlx5_ib_dev *dev, int c, int num)
 		mlx5_core_destroy_mkey(dev->mdev, &mr->mmkey);
 	}
 
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
 	if (odp_mkey_exist)
 		synchronize_srcu(&dev->mr_srcu);
-#endif
 
 	list_for_each_entry_safe(mr, tmp_mr, &del_list, list) {
 		list_del(&mr->list);
@@ -594,7 +584,7 @@ static void clean_keys(struct mlx5_ib_dev *dev, int c)
 			break;
 		}
 		mr = list_first_entry(&ent->head, struct mlx5_ib_mr, list);
-		if (mr->umem && mr->umem->is_odp)
+		if (is_odp_mr(mr))
 			odp_mkey_exist = true;
 		list_move(&mr->list, &del_list);
 		ent->cur--;
@@ -603,10 +593,8 @@ static void clean_keys(struct mlx5_ib_dev *dev, int c)
 		mlx5_core_destroy_mkey(dev->mdev, &mr->mmkey);
 	}
 
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
 	if (odp_mkey_exist)
 		synchronize_srcu(&dev->mr_srcu);
-#endif
 
 	list_for_each_entry_safe(mr, tmp_mr, &del_list, list) {
 		list_del(&mr->list);
@@ -1399,9 +1387,7 @@ struct ib_mr *mlx5_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
 	mr->umem = umem;
 	set_mr_fields(dev, mr, npages, length, access_flags);
 
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
 	update_odp_mr(mr);
-#endif
 
 	if (!populate_mtts) {
 		int update_xlt_flags = MLX5_IB_UPD_XLT_ENABLE;
@@ -1566,9 +1552,7 @@ int mlx5_ib_rereg_user_mr(struct ib_mr *ib_mr, int flags, u64 start,
 
 	set_mr_fields(dev, mr, npages, len, access_flags);
 
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
 	update_odp_mr(mr);
-#endif
 	return 0;
 
 err:
@@ -1654,8 +1638,7 @@ static void dereg_mr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr)
 	int npages = mr->npages;
 	struct ib_umem *umem = mr->umem;
 
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
-	if (umem && umem->is_odp) {
+	if (is_odp_mr(mr)) {
 		struct ib_umem_odp *umem_odp = to_ib_umem_odp(umem);
 
 		/* Prevent new page faults from succeeding */
@@ -1679,7 +1662,7 @@ static void dereg_mr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr)
 		/* Avoid double-freeing the umem. */
 		umem = NULL;
 	}
-#endif
+
 	clean_mr(dev, mr);
 
 	/*
diff --git a/drivers/infiniband/hw/mlx5/odp.c b/drivers/infiniband/hw/mlx5/odp.c
index 80fa2438db8f..86c64c3468df 100644
--- a/drivers/infiniband/hw/mlx5/odp.c
+++ b/drivers/infiniband/hw/mlx5/odp.c
@@ -103,7 +103,7 @@ static int check_parent(struct ib_umem_odp *odp,
 
 struct ib_ucontext_per_mm *mr_to_per_mm(struct mlx5_ib_mr *mr)
 {
-	if (WARN_ON(!mr || !mr->umem || !mr->umem->is_odp))
+	if (WARN_ON(!mr || !is_odp_mr(mr)))
 		return NULL;
 
 	return to_ib_umem_odp(mr->umem)->per_mm;
@@ -740,12 +740,12 @@ static int pagefault_single_data_segment(struct mlx5_ib_dev *dev, u32 key,
 			goto srcu_unlock;
 		}
 
-		if (prefetch && !mr->umem->is_odp) {
+		if (prefetch && !is_odp_mr(mr)) {
 			ret = -EINVAL;
 			goto srcu_unlock;
 		}
 
-		if (!mr->umem->is_odp) {
+		if (!is_odp_mr(mr)) {
 			mlx5_ib_dbg(dev, "skipping non ODP MR (lkey=0x%06x) in page fault handler.\n",
 				    key);
 			if (bytes_mapped)
-- 
2.19.1

[PATCH rdma-next v1 4/5] RDMA/mlx5: Embed into the code flow the ODP config option

[Leon Romanovsky]

From: Leon Romanovsky <leonro@mellanox.com>

Convert various places to more readable code, which embeds
CONFIG_INFINIBAND_ON_DEMAND_PAGING into the code flow.

Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
---
 drivers/infiniband/core/uverbs_cmd.c |  3 --
 drivers/infiniband/hw/mlx5/main.c    | 41 ++++++++++++++--------------
 drivers/infiniband/hw/mlx5/mem.c     |  5 +---
 drivers/infiniband/hw/mlx5/mr.c      | 16 +++++------
 include/rdma/ib_umem_odp.h           | 26 +++++++++---------
 5 files changed, 41 insertions(+), 50 deletions(-)

diff --git a/drivers/infiniband/core/uverbs_cmd.c b/drivers/infiniband/core/uverbs_cmd.c
index 549d9eedf62e..d4f1a2ef5015 100644
--- a/drivers/infiniband/core/uverbs_cmd.c
+++ b/drivers/infiniband/core/uverbs_cmd.c
@@ -234,14 +234,11 @@ static int ib_uverbs_get_context(struct uverbs_attr_bundle *attrs)
 	ucontext->closing = false;
 	ucontext->cleanup_retryable = false;
 
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
 	mutex_init(&ucontext->per_mm_list_lock);
 	INIT_LIST_HEAD(&ucontext->per_mm_list);
 	if (!(ib_dev->attrs.device_cap_flags & IB_DEVICE_ON_DEMAND_PAGING))
 		ucontext->invalidate_range = NULL;
 
-#endif
-
 	resp.num_comp_vectors = file->device->num_comp_vectors;
 
 	ret = get_unused_fd_flags(O_CLOEXEC);
diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index c98343e50a00..9f5f0c638663 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -1763,9 +1763,9 @@ static struct ib_ucontext *mlx5_ib_alloc_ucontext(struct ib_device *ibdev,
 	if (err)
 		goto out_sys_pages;
 
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
-	context->ibucontext.invalidate_range = &mlx5_ib_invalidate_range;
-#endif
+	if (ibdev->attrs.device_cap_flags & IB_DEVICE_ON_DEMAND_PAGING)
+		context->ibucontext.invalidate_range =
+			&mlx5_ib_invalidate_range;
 
 	if (req.flags & MLX5_IB_ALLOC_UCTX_DEVX) {
 		err = mlx5_ib_devx_create(dev, true);
@@ -1897,12 +1897,10 @@ static int mlx5_ib_dealloc_ucontext(struct ib_ucontext *ibcontext)
 	struct mlx5_ib_dev *dev = to_mdev(ibcontext->device);
 	struct mlx5_bfreg_info *bfregi;
 
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
 	/* All umem's must be destroyed before destroying the ucontext. */
 	mutex_lock(&ibcontext->per_mm_list_lock);
 	WARN_ON(!list_empty(&ibcontext->per_mm_list));
 	mutex_unlock(&ibcontext->per_mm_list_lock);
-#endif
 
 	bfregi = &context->bfregi;
 	mlx5_ib_dealloc_transport_domain(dev, context->tdn, context->devx_uid);
@@ -5722,11 +5720,11 @@ static struct ib_counters *mlx5_ib_create_counters(struct ib_device *device,
 void mlx5_ib_stage_init_cleanup(struct mlx5_ib_dev *dev)
 {
 	mlx5_ib_cleanup_multiport_master(dev);
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
-	cleanup_srcu_struct(&dev->mr_srcu);
-	drain_workqueue(dev->advise_mr_wq);
-	destroy_workqueue(dev->advise_mr_wq);
-#endif
+	if (IS_ENABLED(CONFIG_INFINIBAND_ON_DEMAND_PAGING)) {
+		cleanup_srcu_struct(&dev->mr_srcu);
+		drain_workqueue(dev->advise_mr_wq);
+		destroy_workqueue(dev->advise_mr_wq);
+	}
 	kfree(dev->port);
 }
 
@@ -5779,19 +5777,20 @@ int mlx5_ib_stage_init_init(struct mlx5_ib_dev *dev)
 	spin_lock_init(&dev->memic.memic_lock);
 	dev->memic.dev = mdev;
 
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
-	dev->advise_mr_wq = alloc_ordered_workqueue("mlx5_ib_advise_mr_wq", 0);
-	if (!dev->advise_mr_wq) {
-		err = -ENOMEM;
-		goto err_mp;
-	}
+	if (IS_ENABLED(CONFIG_INFINIBAND_ON_DEMAND_PAGING)) {
+		dev->advise_mr_wq =
+			alloc_ordered_workqueue("mlx5_ib_advise_mr_wq", 0);
+		if (!dev->advise_mr_wq) {
+			err = -ENOMEM;
+			goto err_mp;
+		}
 
-	err = init_srcu_struct(&dev->mr_srcu);
-	if (err) {
-		destroy_workqueue(dev->advise_mr_wq);
-		goto err_mp;
+		err = init_srcu_struct(&dev->mr_srcu);
+		if (err) {
+			destroy_workqueue(dev->advise_mr_wq);
+			goto err_mp;
+		}
 	}
-#endif
 
 	return 0;
 err_mp:
diff --git a/drivers/infiniband/hw/mlx5/mem.c b/drivers/infiniband/hw/mlx5/mem.c
index 549234988bb4..9f90be296ee0 100644
--- a/drivers/infiniband/hw/mlx5/mem.c
+++ b/drivers/infiniband/hw/mlx5/mem.c
@@ -111,7 +111,6 @@ void mlx5_ib_cont_pages(struct ib_umem *umem, u64 addr,
 	*count = i;
 }
 
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
 static u64 umem_dma_to_mtt(dma_addr_t umem_dma)
 {
 	u64 mtt_entry = umem_dma & ODP_DMA_ADDR_MASK;
@@ -123,7 +122,6 @@ static u64 umem_dma_to_mtt(dma_addr_t umem_dma)
 
 	return mtt_entry;
 }
-#endif
 
 /*
  * Populate the given array with bus addresses from the umem.
@@ -151,7 +149,7 @@ void __mlx5_ib_populate_pas(struct mlx5_ib_dev *dev, struct ib_umem *umem,
 	int len;
 	struct scatterlist *sg;
 	int entry;
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
+
 	if (umem->is_odp) {
 		WARN_ON(shift != 0);
 		WARN_ON(access_flags != (MLX5_IB_MTT_READ | MLX5_IB_MTT_WRITE));
@@ -164,7 +162,6 @@ void __mlx5_ib_populate_pas(struct mlx5_ib_dev *dev, struct ib_umem *umem,
 		}
 		return;
 	}
-#endif
 
 	i = 0;
 	for_each_sg(umem->sg_head.sgl, sg, umem->nmap, entry) {
diff --git a/drivers/infiniband/hw/mlx5/mr.c b/drivers/infiniband/hw/mlx5/mr.c
index b861b4a5b0e0..65d07c111d42 100644
--- a/drivers/infiniband/hw/mlx5/mr.c
+++ b/drivers/infiniband/hw/mlx5/mr.c
@@ -1332,8 +1332,8 @@ struct ib_mr *mlx5_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
 	mlx5_ib_dbg(dev, "start 0x%llx, virt_addr 0x%llx, length 0x%llx, access_flags 0x%x\n",
 		    start, virt_addr, length, access_flags);
 
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
-	if (!start && length == U64_MAX) {
+	if (IS_ENABLED(CONFIG_INFINIBAND_ON_DEMAND_PAGING) && !start &&
+	    length == U64_MAX) {
 		if (!(access_flags & IB_ACCESS_ON_DEMAND) ||
 		    !(dev->odp_caps.general_caps & IB_ODP_SUPPORT_IMPLICIT))
 			return ERR_PTR(-EINVAL);
@@ -1343,7 +1343,6 @@ struct ib_mr *mlx5_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
 			return ERR_CAST(mr);
 		return &mr->ibmr;
 	}
-#endif
 
 	err = mr_umem_get(pd, start, length, access_flags, &umem, &npages,
 			   &page_shift, &ncont, &order);
@@ -1404,9 +1403,9 @@ struct ib_mr *mlx5_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
 		}
 	}
 
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
-	mr->live = 1;
-#endif
+	if (IS_ENABLED(CONFIG_INFINIBAND_ON_DEMAND_PAGING))
+		mr->live = 1;
+
 	return &mr->ibmr;
 error:
 	ib_umem_release(umem);
@@ -1521,9 +1520,8 @@ int mlx5_ib_rereg_user_mr(struct ib_mr *ib_mr, int flags, u64 start,
 		}
 
 		mr->allocated_from_cache = 0;
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
-		mr->live = 1;
-#endif
+		if (IS_ENABLED(CONFIG_INFINIBAND_ON_DEMAND_PAGING))
+			mr->live = 1;
 	} else {
 		/*
 		 * Send a UMR WQE
diff --git a/include/rdma/ib_umem_odp.h b/include/rdma/ib_umem_odp.h
index 0b1446fe2fab..d3725cf13ecd 100644
--- a/include/rdma/ib_umem_odp.h
+++ b/include/rdma/ib_umem_odp.h
@@ -83,6 +83,19 @@ static inline struct ib_umem_odp *to_ib_umem_odp(struct ib_umem *umem)
 	return container_of(umem, struct ib_umem_odp, umem);
 }
 
+/*
+ * The lower 2 bits of the DMA address signal the R/W permissions for
+ * the entry. To upgrade the permissions, provide the appropriate
+ * bitmask to the map_dma_pages function.
+ *
+ * Be aware that upgrading a mapped address might result in change of
+ * the DMA address for the page.
+ */
+#define ODP_READ_ALLOWED_BIT  (1<<0ULL)
+#define ODP_WRITE_ALLOWED_BIT (1<<1ULL)
+
+#define ODP_DMA_ADDR_MASK (~(ODP_READ_ALLOWED_BIT | ODP_WRITE_ALLOWED_BIT))
+
 #ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
 
 struct ib_ucontext_per_mm {
@@ -107,19 +120,6 @@ struct ib_umem_odp *ib_alloc_odp_umem(struct ib_ucontext_per_mm *per_mm,
 				      unsigned long addr, size_t size);
 void ib_umem_odp_release(struct ib_umem_odp *umem_odp);
 
-/*
- * The lower 2 bits of the DMA address signal the R/W permissions for
- * the entry. To upgrade the permissions, provide the appropriate
- * bitmask to the map_dma_pages function.
- *
- * Be aware that upgrading a mapped address might result in change of
- * the DMA address for the page.
- */
-#define ODP_READ_ALLOWED_BIT  (1<<0ULL)
-#define ODP_WRITE_ALLOWED_BIT (1<<1ULL)
-
-#define ODP_DMA_ADDR_MASK (~(ODP_READ_ALLOWED_BIT | ODP_WRITE_ALLOWED_BIT))
-
 int ib_umem_odp_map_dma_pages(struct ib_umem_odp *umem_odp, u64 start_offset,
 			      u64 bcnt, u64 access_mask,
 			      unsigned long current_seq);
-- 
2.19.1

[PATCH mlx5-next v1 5/5] RDMA/mlx5: Delete declaration of already removed function

[Leon Romanovsky]

From: Leon Romanovsky <leonro@mellanox.com>

The implementation of mlx5_core_page_fault_resume() was removed
in commit d5d284b829a6 ("{net,IB}/mlx5: Move Page fault EQ and
ODP logic to RDMA"). This patch removes declaration too.

Fixes: d5d284b829a6 ("{net,IB}/mlx5: Move Page fault EQ and ODP logic to RDMA")
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
---
 include/linux/mlx5/driver.h | 4 ----
 1 file changed, 4 deletions(-)

diff --git a/include/linux/mlx5/driver.h b/include/linux/mlx5/driver.h
index cb1c5b10279c..709446bea72c 100644
--- a/include/linux/mlx5/driver.h
+++ b/include/linux/mlx5/driver.h
@@ -937,10 +937,6 @@ int mlx5_query_odp_caps(struct mlx5_core_dev *dev,
 			struct mlx5_odp_caps *odp_caps);
 int mlx5_core_query_ib_ppcnt(struct mlx5_core_dev *dev,
 			     u8 port_num, void *out, size_t sz);
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
-int mlx5_core_page_fault_resume(struct mlx5_core_dev *dev, u32 token,
-				u32 wq_num, u8 type, int error);
-#endif
 
 int mlx5_init_rl_table(struct mlx5_core_dev *dev);
 void mlx5_cleanup_rl_table(struct mlx5_core_dev *dev);
-- 
2.19.1

Re: [PATCH rdma-next v1 0/5] Cleanup of CONFIG_INFINIBAND_ON_DEMAND_PAGING usage

[Jason Gunthorpe]

On Sun, Dec 23, 2018 at 11:14:13AM +0200, Leon Romanovsky wrote:
> From: Leon Romanovsky <leonro@mellanox.com>
> 
> Changelog v0->v1:
>  * Don't set ODP caps always (patch #2)
>  * Set invalidate_range callback based on device_cap only
> 
> -------------------------------------------------------------------------------
> 
> Hi,
> 
> As a followup to Jason's request to rethink CONFIG_INFINIBAND_ON_DEMAND_PAGING
> usage, this series cleans mlx5_ib and RDMA/core code and it is based on accepted
> patch https://patchwork.kernel.org/patch/10735547/
> 
> I think that we will see more patches in that area, I would like to
> remove CONFIG_INFINIBAND_ON_DEMAND_PAGING completely.
> 
> Thanks
> 
> Leon Romanovsky (5):
>   RDMA: Clean structures from CONFIG_INFINIBAND_ON_DEMAND_PAGING
>   RDMA/core: Don't depend device ODP capabilities on kconfig option
>   RDMA/mlx5: Introduce and reuse helper to identify ODP MR
>   RDMA/mlx5: Embed into the code flow the ODP config option
>   RDMA/mlx5: Delete declaration of already removed function

This doesn't apply anymore, can you resend it please?

Jason

Re: [PATCH rdma-next v1 0/5] Cleanup of CONFIG_INFINIBAND_ON_DEMAND_PAGING usage

[Leon Romanovsky]

[-- Attachment #1: Type: text/plain, Size: 1232 bytes --]

On Mon, Jan 07, 2019 at 11:31:28AM -0700, Jason Gunthorpe wrote:
> On Sun, Dec 23, 2018 at 11:14:13AM +0200, Leon Romanovsky wrote:
> > From: Leon Romanovsky <leonro@mellanox.com>
> >
> > Changelog v0->v1:
> >  * Don't set ODP caps always (patch #2)
> >  * Set invalidate_range callback based on device_cap only
> >
> > -------------------------------------------------------------------------------
> >
> > Hi,
> >
> > As a followup to Jason's request to rethink CONFIG_INFINIBAND_ON_DEMAND_PAGING
> > usage, this series cleans mlx5_ib and RDMA/core code and it is based on accepted
> > patch https://patchwork.kernel.org/patch/10735547/
> >
> > I think that we will see more patches in that area, I would like to
> > remove CONFIG_INFINIBAND_ON_DEMAND_PAGING completely.
> >
> > Thanks
> >
> > Leon Romanovsky (5):
> >   RDMA: Clean structures from CONFIG_INFINIBAND_ON_DEMAND_PAGING
> >   RDMA/core: Don't depend device ODP capabilities on kconfig option
> >   RDMA/mlx5: Introduce and reuse helper to identify ODP MR
> >   RDMA/mlx5: Embed into the code flow the ODP config option
> >   RDMA/mlx5: Delete declaration of already removed function
>
> This doesn't apply anymore, can you resend it please?

Done, thanks

>
> Jason

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 801 bytes --]