From 16b82367aa28bd31795e720548421b58824108e1 Mon Sep 17 00:00:00 2001
From: Easwar Hariharan <eahariha@linux.microsoft.com>
Date: Wed, 19 Feb 2025 21:36:40 +0000
Subject: [PATCH 01/57] RDMA/mlx5: convert timeouts to secs_to_jiffies()

Commit b35108a51cf7 ("jiffies: Define secs_to_jiffies()") introduced
secs_to_jiffies().  As the value here is a multiple of 1000, use
secs_to_jiffies() instead of msecs_to_jiffies to avoid the multiplication.

This is converted using scripts/coccinelle/misc/secs_to_jiffies.cocci with
the following Coccinelle rules:

@depends on patch@
expression E;
@@

-msecs_to_jiffies(E * 1000)
+secs_to_jiffies(E)

-msecs_to_jiffies(E * MSEC_PER_SEC)
+secs_to_jiffies(E)

Link: https://patch.msgid.link/r/20250219-rdma-secs-to-jiffies-v1-2-b506746561a9@linux.microsoft.com
Signed-off-by: Easwar Hariharan <eahariha@linux.microsoft.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
---
 drivers/infiniband/hw/mlx5/mr.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/infiniband/hw/mlx5/mr.c b/drivers/infiniband/hw/mlx5/mr.c
index 5fbebafc8774..6dd813bac5b2 100644
--- a/drivers/infiniband/hw/mlx5/mr.c
+++ b/drivers/infiniband/hw/mlx5/mr.c
@@ -525,7 +525,7 @@ static void queue_adjust_cache_locked(struct mlx5_cache_ent *ent)
 		ent->fill_to_high_water = false;
 		if (ent->pending)
 			queue_delayed_work(ent->dev->cache.wq, &ent->dwork,
-					   msecs_to_jiffies(1000));
+					   secs_to_jiffies(1));
 		else
 			mod_delayed_work(ent->dev->cache.wq, &ent->dwork, 0);
 	}
@@ -576,7 +576,7 @@ static void __cache_work_func(struct mlx5_cache_ent *ent)
 					"add keys command failed, err %d\n",
 					err);
 				queue_delayed_work(cache->wq, &ent->dwork,
-						   msecs_to_jiffies(1000));
+						   secs_to_jiffies(1));
 			}
 		}
 	} else if (ent->mkeys_queue.ci > 2 * ent->limit) {
@@ -2051,7 +2051,7 @@ static int mlx5_revoke_mr(struct mlx5_ib_mr *mr)
 			ent->in_use--;
 		if (ent->is_tmp && !ent->tmp_cleanup_scheduled) {
 			mod_delayed_work(ent->dev->cache.wq, &ent->dwork,
-					 msecs_to_jiffies(30 * 1000));
+					 secs_to_jiffies(30));
 			ent->tmp_cleanup_scheduled = true;
 		}
 		spin_unlock_irq(&ent->mkeys_queue.lock);

From 2633ddff9b0f6f2a41f9f05cf18afacc029f8ef7 Mon Sep 17 00:00:00 2001
From: Peng Jiang <jiang.peng9@zte.com.cn>
Date: Wed, 26 Mar 2025 22:19:55 +0800
Subject: [PATCH 02/57] =?UTF-8?q?RDMA:=20Replace=20msecs=5Fto=5Fjiffies=20?=
 =?UTF-8?q?with=20secs=5Fto=5Fjiffies=20for=C2=A0timeout?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

In drivers/infiniband/hw/mlx4/mcg.c and drivers/infiniband/hw/mlx5/mr.c,
`msecs_to_jiffies` is used to convert milliseconds to jiffies.
For constant milliseconds, using `msecs_to_jiffies` introduces additional
computational overhead. For example, it is unnecessary to check
if m > jiffies_to_msecs(MAX_JIFFY_OFFSET) or (int)m < 0 for constants,
while using `secs_to_jiffies` can avoid these extra calculations.

Link: https://patch.msgid.link/r/20250326221955611qu6Ix3Pt5WgKvhL6sTySX@zte.com.cn
Signed-off-by: Peng Jiang <jiang.peng9@zte.com.cn>
Signed-off-by: Ye Xingchen <ye.xingchen@zte.com.cn>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
---
 drivers/infiniband/hw/mlx4/mcg.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/drivers/infiniband/hw/mlx4/mcg.c b/drivers/infiniband/hw/mlx4/mcg.c
index 33f525b744f2..e279e69b9a51 100644
--- a/drivers/infiniband/hw/mlx4/mcg.c
+++ b/drivers/infiniband/hw/mlx4/mcg.c
@@ -43,7 +43,7 @@
 
 #define MAX_VFS		80
 #define MAX_PEND_REQS_PER_FUNC 4
-#define MAD_TIMEOUT_MS	2000
+#define MAD_TIMEOUT_SEC	2
 
 #define mcg_warn(fmt, arg...)	pr_warn("MCG WARNING: " fmt, ##arg)
 #define mcg_error(fmt, arg...)	pr_err(fmt, ##arg)
@@ -270,7 +270,7 @@ static int send_join_to_wire(struct mcast_group *group, struct ib_sa_mad *sa_mad
 	if (!ret) {
 		/* calls mlx4_ib_mcg_timeout_handler */
 		queue_delayed_work(group->demux->mcg_wq, &group->timeout_work,
-				msecs_to_jiffies(MAD_TIMEOUT_MS));
+				   secs_to_jiffies(MAD_TIMEOUT_SEC));
 	}
 
 	return ret;
@@ -309,7 +309,7 @@ static int send_leave_to_wire(struct mcast_group *group, u8 join_state)
 	if (!ret) {
 		/* calls mlx4_ib_mcg_timeout_handler */
 		queue_delayed_work(group->demux->mcg_wq, &group->timeout_work,
-				msecs_to_jiffies(MAD_TIMEOUT_MS));
+				   secs_to_jiffies(MAD_TIMEOUT_SEC));
 	}
 
 	return ret;
@@ -1091,7 +1091,7 @@ static void _mlx4_ib_mcg_port_cleanup(struct mlx4_ib_demux_ctx *ctx, int destroy
 	for (i = 0; i < MAX_VFS; ++i)
 		clean_vf_mcast(ctx, i);
 
-	end = jiffies + msecs_to_jiffies(MAD_TIMEOUT_MS + 3000);
+	end = jiffies + secs_to_jiffies(MAD_TIMEOUT_SEC + 3);
 	do {
 		count = 0;
 		mutex_lock(&ctx->mcg_table_lock);

From 41e2649c796e1f49c7b4d9176d2437504b78580f Mon Sep 17 00:00:00 2001
From: Li Haoran <li.haoran7@zte.com.cn>
Date: Tue, 1 Apr 2025 21:08:40 +0800
Subject: [PATCH 03/57] RDMA/core: Convert to use ERR_CAST()

As opposed to open-code, using the ERR_CAST macro clearly indicates that
this is a pointer to an error value and a type conversion was performed.

Link: https://patch.msgid.link/r/20250401210840146_IyrV3zlejzz3eAnDmMSB@zte.com.cn
Signed-off-by: Li Haoran <li.haoran7@zte.com.cn>
Signed-off-by: Shao Mingyin <shao.mingyin@zte.com.cn>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
---
 drivers/infiniband/core/mad_rmpp.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/infiniband/core/mad_rmpp.c b/drivers/infiniband/core/mad_rmpp.c
index 8af0619a39cd..b4b10e8a6495 100644
--- a/drivers/infiniband/core/mad_rmpp.c
+++ b/drivers/infiniband/core/mad_rmpp.c
@@ -158,7 +158,7 @@ static struct ib_mad_send_buf *alloc_response_msg(struct ib_mad_agent *agent,
 	ah = ib_create_ah_from_wc(agent->qp->pd, recv_wc->wc,
 				  recv_wc->recv_buf.grh, agent->port_num);
 	if (IS_ERR(ah))
-		return (void *) ah;
+		return ERR_CAST(ah);
 
 	hdr_len = ib_get_mad_data_offset(recv_wc->recv_buf.mad->mad_hdr.mgmt_class);
 	msg = ib_create_send_mad(agent, recv_wc->wc->src_qp,

From 7bc871af417a3011739ccddabb64db482ff7d752 Mon Sep 17 00:00:00 2001
From: Li Haoran <li.haoran7@zte.com.cn>
Date: Tue, 1 Apr 2025 21:09:23 +0800
Subject: [PATCH 04/57] RDMA/uverbs: Convert to use ERR_CAST()

As opposed to open-code, using the ERR_CAST macro clearly indicates that
this is a pointer to an error value and a type conversion was performed.

Link: https://patch.msgid.link/r/202504012109233981_YPVbd4wQzmAzP3tA5IG@zte.com.cn
Signed-off-by: Li Haoran <li.haoran7@zte.com.cn>
Signed-off-by: Shao Mingyin <shao.mingyin@zte.com.cn>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
---
 drivers/infiniband/core/uverbs_cmd.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/infiniband/core/uverbs_cmd.c b/drivers/infiniband/core/uverbs_cmd.c
index 3c3bb670c805..bc9fe3ceca4d 100644
--- a/drivers/infiniband/core/uverbs_cmd.c
+++ b/drivers/infiniband/core/uverbs_cmd.c
@@ -193,7 +193,7 @@ _ib_uverbs_lookup_comp_file(s32 fd, struct uverbs_attr_bundle *attrs)
 					       fd, attrs);
 
 	if (IS_ERR(uobj))
-		return (void *)uobj;
+		return ERR_CAST(uobj);
 
 	uverbs_uobject_get(uobj);
 	uobj_put_read(uobj);

From 8a94c42d831092ba65ce88bda020d93d2e267fe4 Mon Sep 17 00:00:00 2001
From: Li Haoran <li.haoran7@zte.com.cn>
Date: Tue, 1 Apr 2025 21:10:15 +0800
Subject: [PATCH 05/57] RDMA/core: Convert to use ERR_CAST()

As opposed to open-code, using the ERR_CAST macro clearly indicates that
this is a pointer to an error value and a type conversion was performed.

Link: https://patch.msgid.link/r/20250401211015750qxOfU9XZ8QgKizM1Lcyq2@zte.com.cn
Signed-off-by: Li Haoran <li.haoran7@zte.com.cn>
Signed-off-by: Shao Mingyin <shao.mingyin@zte.com.cn>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
---
 drivers/infiniband/core/verbs.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/infiniband/core/verbs.c b/drivers/infiniband/core/verbs.c
index c5e78bbefbd0..75fde0fe9989 100644
--- a/drivers/infiniband/core/verbs.c
+++ b/drivers/infiniband/core/verbs.c
@@ -572,7 +572,7 @@ struct ib_ah *rdma_create_ah(struct ib_pd *pd, struct rdma_ah_attr *ah_attr,
 					   GFP_KERNEL : GFP_ATOMIC);
 	if (IS_ERR(slave)) {
 		rdma_unfill_sgid_attr(ah_attr, old_sgid_attr);
-		return (void *)slave;
+		return ERR_CAST(slave);
 	}
 	ah = _rdma_create_ah(pd, ah_attr, flags, NULL, slave);
 	rdma_lag_put_ah_roce_slave(slave);

From aae85e007dffda49d1a611474f9653021ec763bd Mon Sep 17 00:00:00 2001
From: "Gustavo A. R. Silva" <gustavoars@kernel.org>
Date: Tue, 1 Apr 2025 11:29:06 -0600
Subject: [PATCH 06/57] IB/hfi1: Avoid -Wflex-array-member-not-at-end warning

-Wflex-array-member-not-at-end was introduced in GCC-14, and we are
getting ready to enable it, globally.

Remove unused flex-array member `class_data` from
`struct opa_mad_notice_attr`.

Fix the following warning:

drivers/infiniband/hw/hfi1/mad.c:23:36: warning: structure containing a flexible array member is not at the end of another structure [-Wflex-array-member-not-at-end]

Link: https://patch.msgid.link/r/Z-wiYkll8Vo3ME3P@kspp
Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
---
 drivers/infiniband/hw/hfi1/mad.h | 1 -
 1 file changed, 1 deletion(-)

diff --git a/drivers/infiniband/hw/hfi1/mad.h b/drivers/infiniband/hw/hfi1/mad.h
index b6e3141253c4..d6dde762921a 100644
--- a/drivers/infiniband/hw/hfi1/mad.h
+++ b/drivers/infiniband/hw/hfi1/mad.h
@@ -124,7 +124,6 @@ struct opa_mad_notice_attr {
 		} __packed ntc_2048;
 
 	};
-	u8	class_data[];
 };
 
 #define IB_VLARB_LOWPRI_0_31    1

From 3aadd652c2c816a908abe55079e2590e3259e8ba Mon Sep 17 00:00:00 2001
From: Chengchang Tang <tangchengchang@huawei.com>
Date: Thu, 27 Mar 2025 19:47:23 +0800
Subject: [PATCH 07/57] RDMA/hns: Remove unused parameters

Remove unused parameters.

Link: https://patch.msgid.link/r/20250327114724.3454268-2-huangjunxian6@hisilicon.com
Signed-off-by: Chengchang Tang <tangchengchang@huawei.com>
Signed-off-by: Junxian Huang <huangjunxian6@hisilicon.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
---
 drivers/infiniband/hw/hns/hns_roce_hw_v2.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
index 160e8927d364..0600e87936bf 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
@@ -4302,8 +4302,7 @@ static inline int get_pdn(struct ib_pd *ib_pd)
 }
 
 static void modify_qp_reset_to_init(struct ib_qp *ibqp,
-				    struct hns_roce_v2_qp_context *context,
-				    struct hns_roce_v2_qp_context *qpc_mask)
+				    struct hns_roce_v2_qp_context *context)
 {
 	struct hns_roce_dev *hr_dev = to_hr_dev(ibqp->device);
 	struct hns_roce_qp *hr_qp = to_hr_qp(ibqp);
@@ -5122,7 +5121,7 @@ static int hns_roce_v2_set_abs_fields(struct ib_qp *ibqp,
 
 	if (cur_state == IB_QPS_RESET && new_state == IB_QPS_INIT) {
 		memset(qpc_mask, 0, hr_dev->caps.qpc_sz);
-		modify_qp_reset_to_init(ibqp, context, qpc_mask);
+		modify_qp_reset_to_init(ibqp, context);
 	} else if (cur_state == IB_QPS_INIT && new_state == IB_QPS_INIT) {
 		modify_qp_init_to_init(ibqp, context, qpc_mask);
 	} else if (cur_state == IB_QPS_INIT && new_state == IB_QPS_RTR) {

From 4dab26bed543584577b64b36aadb8b5b165bf44f Mon Sep 17 00:00:00 2001
From: Jacob Moroni <jmoroni@google.com>
Date: Thu, 20 Feb 2025 17:56:12 +0000
Subject: [PATCH 08/57] IB/cm: use rwlock for MAD agent lock

In workloads where there are many processes establishing connections using
RDMA CM in parallel (large scale MPI), there can be heavy contention for
mad_agent_lock in cm_alloc_msg.

This contention can occur while inside of a spin_lock_irq region, leading
to interrupts being disabled for extended durations on many
cores. Furthermore, it leads to the serialization of rdma_create_ah calls,
which has negative performance impacts for NICs which are capable of
processing multiple address handle creations in parallel.

The end result is the machine becoming unresponsive, hung task warnings,
netdev TX timeouts, etc.

Since the lock appears to be only for protection from cm_remove_one, it
can be changed to a rwlock to resolve these issues.

Reproducer:

Server:
  for i in $(seq 1 512); do
    ucmatose -c 32 -p $((i + 5000)) &
  done

Client:
  for i in $(seq 1 512); do
    ucmatose -c 32 -p $((i + 5000)) -s 10.2.0.52 &
  done

Fixes: 76039ac9095f ("IB/cm: Protect cm_dev, cm_ports and mad_agent with kref and lock")
Link: https://patch.msgid.link/r/20250220175612.2763122-1-jmoroni@google.com
Signed-off-by: Jacob Moroni <jmoroni@google.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Zhu Yanjun <yanjun.zhu@linux.dev>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
---
 drivers/infiniband/core/cm.c | 16 ++++++++--------
 1 file changed, 8 insertions(+), 8 deletions(-)

diff --git a/drivers/infiniband/core/cm.c b/drivers/infiniband/core/cm.c
index 142170473e75..effa53dd6800 100644
--- a/drivers/infiniband/core/cm.c
+++ b/drivers/infiniband/core/cm.c
@@ -167,7 +167,7 @@ struct cm_port {
 struct cm_device {
 	struct kref kref;
 	struct list_head list;
-	spinlock_t mad_agent_lock;
+	rwlock_t mad_agent_lock;
 	struct ib_device *ib_device;
 	u8 ack_delay;
 	int going_down;
@@ -285,7 +285,7 @@ static struct ib_mad_send_buf *cm_alloc_msg(struct cm_id_private *cm_id_priv)
 	if (!cm_id_priv->av.port)
 		return ERR_PTR(-EINVAL);
 
-	spin_lock(&cm_id_priv->av.port->cm_dev->mad_agent_lock);
+	read_lock(&cm_id_priv->av.port->cm_dev->mad_agent_lock);
 	mad_agent = cm_id_priv->av.port->mad_agent;
 	if (!mad_agent) {
 		m = ERR_PTR(-EINVAL);
@@ -311,7 +311,7 @@ static struct ib_mad_send_buf *cm_alloc_msg(struct cm_id_private *cm_id_priv)
 	m->ah = ah;
 
 out:
-	spin_unlock(&cm_id_priv->av.port->cm_dev->mad_agent_lock);
+	read_unlock(&cm_id_priv->av.port->cm_dev->mad_agent_lock);
 	return m;
 }
 
@@ -1297,10 +1297,10 @@ static __be64 cm_form_tid(struct cm_id_private *cm_id_priv)
 	if (!cm_id_priv->av.port)
 		return cpu_to_be64(low_tid);
 
-	spin_lock(&cm_id_priv->av.port->cm_dev->mad_agent_lock);
+	read_lock(&cm_id_priv->av.port->cm_dev->mad_agent_lock);
 	if (cm_id_priv->av.port->mad_agent)
 		hi_tid = ((u64)cm_id_priv->av.port->mad_agent->hi_tid) << 32;
-	spin_unlock(&cm_id_priv->av.port->cm_dev->mad_agent_lock);
+	read_unlock(&cm_id_priv->av.port->cm_dev->mad_agent_lock);
 	return cpu_to_be64(hi_tid | low_tid);
 }
 
@@ -4378,7 +4378,7 @@ static int cm_add_one(struct ib_device *ib_device)
 		return -ENOMEM;
 
 	kref_init(&cm_dev->kref);
-	spin_lock_init(&cm_dev->mad_agent_lock);
+	rwlock_init(&cm_dev->mad_agent_lock);
 	cm_dev->ib_device = ib_device;
 	cm_dev->ack_delay = ib_device->attrs.local_ca_ack_delay;
 	cm_dev->going_down = 0;
@@ -4494,9 +4494,9 @@ static void cm_remove_one(struct ib_device *ib_device, void *client_data)
 		 * The above ensures no call paths from the work are running,
 		 * the remaining paths all take the mad_agent_lock.
 		 */
-		spin_lock(&cm_dev->mad_agent_lock);
+		write_lock(&cm_dev->mad_agent_lock);
 		port->mad_agent = NULL;
-		spin_unlock(&cm_dev->mad_agent_lock);
+		write_unlock(&cm_dev->mad_agent_lock);
 		ib_unregister_mad_agent(mad_agent);
 		ib_port_unregister_client_groups(ib_device, i,
 						 cm_counter_groups);

From 6703cb3dced01f32982b9f7069ef1336d0225077 Mon Sep 17 00:00:00 2001
From: Daisuke Matsuda <matsuda-daisuke@fujitsu.com>
Date: Mon, 24 Mar 2025 16:56:48 +0900
Subject: [PATCH 09/57] RDMA/rxe: Enable ODP in RDMA FLUSH operation

For persistent memories, add rxe_odp_flush_pmem_iova() so that ODP specific
steps are executed. Otherwise, no additional consideration is required.

Signed-off-by: Daisuke Matsuda <matsuda-daisuke@fujitsu.com>
Link: https://patch.msgid.link/20250324075649.3313968-2-matsuda-daisuke@fujitsu.com
Reviewed-by: Li Zhijian <lizhijian@fujitsu.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/sw/rxe/rxe.c      |  1 +
 drivers/infiniband/sw/rxe/rxe_loc.h  |  7 ++++
 drivers/infiniband/sw/rxe/rxe_mr.c   | 36 ++++++++++------
 drivers/infiniband/sw/rxe/rxe_odp.c  | 62 ++++++++++++++++++++++++++--
 drivers/infiniband/sw/rxe/rxe_resp.c |  4 --
 include/rdma/ib_verbs.h              |  1 +
 6 files changed, 91 insertions(+), 20 deletions(-)

diff --git a/drivers/infiniband/sw/rxe/rxe.c b/drivers/infiniband/sw/rxe/rxe.c
index b248c68bf9b1..bf711eb9440c 100644
--- a/drivers/infiniband/sw/rxe/rxe.c
+++ b/drivers/infiniband/sw/rxe/rxe.c
@@ -101,6 +101,7 @@ static void rxe_init_device_param(struct rxe_dev *rxe, struct net_device *ndev)
 		rxe->attr.odp_caps.per_transport_caps.rc_odp_caps |= IB_ODP_SUPPORT_READ;
 		rxe->attr.odp_caps.per_transport_caps.rc_odp_caps |= IB_ODP_SUPPORT_ATOMIC;
 		rxe->attr.odp_caps.per_transport_caps.rc_odp_caps |= IB_ODP_SUPPORT_SRQ_RECV;
+		rxe->attr.odp_caps.per_transport_caps.rc_odp_caps |= IB_ODP_SUPPORT_FLUSH;
 	}
 }
 
diff --git a/drivers/infiniband/sw/rxe/rxe_loc.h b/drivers/infiniband/sw/rxe/rxe_loc.h
index feb386d98d1d..0012bebe96ef 100644
--- a/drivers/infiniband/sw/rxe/rxe_loc.h
+++ b/drivers/infiniband/sw/rxe/rxe_loc.h
@@ -194,6 +194,8 @@ int rxe_odp_mr_copy(struct rxe_mr *mr, u64 iova, void *addr, int length,
 		    enum rxe_mr_copy_dir dir);
 int rxe_odp_atomic_op(struct rxe_mr *mr, u64 iova, int opcode,
 			 u64 compare, u64 swap_add, u64 *orig_val);
+int rxe_odp_flush_pmem_iova(struct rxe_mr *mr, u64 iova,
+			    unsigned int length);
 #else /* CONFIG_INFINIBAND_ON_DEMAND_PAGING */
 static inline int
 rxe_odp_mr_init_user(struct rxe_dev *rxe, u64 start, u64 length, u64 iova,
@@ -212,6 +214,11 @@ rxe_odp_atomic_op(struct rxe_mr *mr, u64 iova, int opcode,
 {
 	return RESPST_ERR_UNSUPPORTED_OPCODE;
 }
+static inline int rxe_odp_flush_pmem_iova(struct rxe_mr *mr, u64 iova,
+					  unsigned int length)
+{
+	return -EOPNOTSUPP;
+}
 #endif /* CONFIG_INFINIBAND_ON_DEMAND_PAGING */
 
 #endif /* RXE_LOC_H */
diff --git a/drivers/infiniband/sw/rxe/rxe_mr.c b/drivers/infiniband/sw/rxe/rxe_mr.c
index 868d2f0b74e9..93e4b5acd3ac 100644
--- a/drivers/infiniband/sw/rxe/rxe_mr.c
+++ b/drivers/infiniband/sw/rxe/rxe_mr.c
@@ -424,7 +424,7 @@ err1:
 	return err;
 }
 
-int rxe_flush_pmem_iova(struct rxe_mr *mr, u64 iova, unsigned int length)
+static int rxe_mr_flush_pmem_iova(struct rxe_mr *mr, u64 iova, unsigned int length)
 {
 	unsigned int page_offset;
 	unsigned long index;
@@ -433,16 +433,6 @@ int rxe_flush_pmem_iova(struct rxe_mr *mr, u64 iova, unsigned int length)
 	int err;
 	u8 *va;
 
-	/* mr must be valid even if length is zero */
-	if (WARN_ON(!mr))
-		return -EINVAL;
-
-	if (length == 0)
-		return 0;
-
-	if (mr->ibmr.type == IB_MR_TYPE_DMA)
-		return -EFAULT;
-
 	err = mr_check_range(mr, iova, length);
 	if (err)
 		return err;
@@ -454,7 +444,7 @@ int rxe_flush_pmem_iova(struct rxe_mr *mr, u64 iova, unsigned int length)
 		if (!page)
 			return -EFAULT;
 		bytes = min_t(unsigned int, length,
-				mr_page_size(mr) - page_offset);
+			      mr_page_size(mr) - page_offset);
 
 		va = kmap_local_page(page);
 		arch_wb_cache_pmem(va + page_offset, bytes);
@@ -468,6 +458,28 @@ int rxe_flush_pmem_iova(struct rxe_mr *mr, u64 iova, unsigned int length)
 	return 0;
 }
 
+int rxe_flush_pmem_iova(struct rxe_mr *mr, u64 start, unsigned int length)
+{
+	int err;
+
+	/* mr must be valid even if length is zero */
+	if (WARN_ON(!mr))
+		return -EINVAL;
+
+	if (length == 0)
+		return 0;
+
+	if (mr->ibmr.type == IB_MR_TYPE_DMA)
+		return -EFAULT;
+
+	if (mr->umem->is_odp)
+		err = rxe_odp_flush_pmem_iova(mr, start, length);
+	else
+		err = rxe_mr_flush_pmem_iova(mr, start, length);
+
+	return err;
+}
+
 /* Guarantee atomicity of atomic operations at the machine level. */
 DEFINE_SPINLOCK(atomic_ops_lock);
 
diff --git a/drivers/infiniband/sw/rxe/rxe_odp.c b/drivers/infiniband/sw/rxe/rxe_odp.c
index 9f6e2bb2a269..9a9aae967486 100644
--- a/drivers/infiniband/sw/rxe/rxe_odp.c
+++ b/drivers/infiniband/sw/rxe/rxe_odp.c
@@ -4,6 +4,7 @@
  */
 
 #include <linux/hmm.h>
+#include <linux/libnvdimm.h>
 
 #include <rdma/ib_umem_odp.h>
 
@@ -147,6 +148,16 @@ static inline bool rxe_check_pagefault(struct ib_umem_odp *umem_odp,
 	return need_fault;
 }
 
+static unsigned long rxe_odp_iova_to_index(struct ib_umem_odp *umem_odp, u64 iova)
+{
+	return (iova - ib_umem_start(umem_odp)) >> umem_odp->page_shift;
+}
+
+static unsigned long rxe_odp_iova_to_page_offset(struct ib_umem_odp *umem_odp, u64 iova)
+{
+	return iova & (BIT(umem_odp->page_shift) - 1);
+}
+
 static int rxe_odp_map_range_and_lock(struct rxe_mr *mr, u64 iova, int length, u32 flags)
 {
 	struct ib_umem_odp *umem_odp = to_ib_umem_odp(mr->umem);
@@ -190,8 +201,8 @@ static int __rxe_odp_mr_copy(struct rxe_mr *mr, u64 iova, void *addr,
 	size_t offset;
 	u8 *user_va;
 
-	idx = (iova - ib_umem_start(umem_odp)) >> umem_odp->page_shift;
-	offset = iova & (BIT(umem_odp->page_shift) - 1);
+	idx = rxe_odp_iova_to_index(umem_odp, iova);
+	offset = rxe_odp_iova_to_page_offset(umem_odp, iova);
 
 	while (length > 0) {
 		u8 *src, *dest;
@@ -277,8 +288,8 @@ static int rxe_odp_do_atomic_op(struct rxe_mr *mr, u64 iova, int opcode,
 		return RESPST_ERR_RKEY_VIOLATION;
 	}
 
-	idx = (iova - ib_umem_start(umem_odp)) >> umem_odp->page_shift;
-	page_offset = iova & (BIT(umem_odp->page_shift) - 1);
+	idx = rxe_odp_iova_to_index(umem_odp, iova);
+	page_offset = rxe_odp_iova_to_page_offset(umem_odp, iova);
 	page = hmm_pfn_to_page(umem_odp->pfn_list[idx]);
 	if (!page)
 		return RESPST_ERR_RKEY_VIOLATION;
@@ -324,3 +335,46 @@ int rxe_odp_atomic_op(struct rxe_mr *mr, u64 iova, int opcode,
 
 	return err;
 }
+
+int rxe_odp_flush_pmem_iova(struct rxe_mr *mr, u64 iova,
+			    unsigned int length)
+{
+	struct ib_umem_odp *umem_odp = to_ib_umem_odp(mr->umem);
+	unsigned int page_offset;
+	unsigned long index;
+	struct page *page;
+	unsigned int bytes;
+	int err;
+	u8 *va;
+
+	err = rxe_odp_map_range_and_lock(mr, iova, length,
+					 RXE_PAGEFAULT_DEFAULT);
+	if (err)
+		return err;
+
+	while (length > 0) {
+		index = rxe_odp_iova_to_index(umem_odp, iova);
+		page_offset = rxe_odp_iova_to_page_offset(umem_odp, iova);
+
+		page = hmm_pfn_to_page(umem_odp->pfn_list[index]);
+		if (!page) {
+			mutex_unlock(&umem_odp->umem_mutex);
+			return -EFAULT;
+		}
+
+		bytes = min_t(unsigned int, length,
+			      mr_page_size(mr) - page_offset);
+
+		va = kmap_local_page(page);
+		arch_wb_cache_pmem(va + page_offset, bytes);
+		kunmap_local(va);
+
+		length -= bytes;
+		iova += bytes;
+		page_offset = 0;
+	}
+
+	mutex_unlock(&umem_odp->umem_mutex);
+
+	return 0;
+}
diff --git a/drivers/infiniband/sw/rxe/rxe_resp.c b/drivers/infiniband/sw/rxe/rxe_resp.c
index 54ba9ee1acc5..304e3de740ad 100644
--- a/drivers/infiniband/sw/rxe/rxe_resp.c
+++ b/drivers/infiniband/sw/rxe/rxe_resp.c
@@ -649,10 +649,6 @@ static enum resp_states process_flush(struct rxe_qp *qp,
 	struct rxe_mr *mr = qp->resp.mr;
 	struct resp_res *res = qp->resp.res;
 
-	/* ODP is not supported right now. WIP. */
-	if (mr->umem->is_odp)
-		return RESPST_ERR_UNSUPPORTED_OPCODE;
-
 	/* oA19-14, oA19-15 */
 	if (res && res->replay)
 		return RESPST_ACKNOWLEDGE;
diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index d42eae69d9a8..41bf98ccb275 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -325,6 +325,7 @@ enum ib_odp_transport_cap_bits {
 	IB_ODP_SUPPORT_READ	= 1 << 3,
 	IB_ODP_SUPPORT_ATOMIC	= 1 << 4,
 	IB_ODP_SUPPORT_SRQ_RECV	= 1 << 5,
+	IB_ODP_SUPPORT_FLUSH	= 1 << 6,
 };
 
 struct ib_odp_caps {

From b84001ad0ceeb34bc3fd6c383f197326d4fe8353 Mon Sep 17 00:00:00 2001
From: Daisuke Matsuda <matsuda-daisuke@fujitsu.com>
Date: Mon, 24 Mar 2025 16:56:49 +0900
Subject: [PATCH 10/57] RDMA/rxe: Enable ODP in ATOMIC WRITE operation

Add rxe_odp_do_atomic_write() so that ODP specific steps are applied to
ATOMIC WRITE requests.

Signed-off-by: Daisuke Matsuda <matsuda-daisuke@fujitsu.com>
Link: https://patch.msgid.link/20250324075649.3313968-3-matsuda-daisuke@fujitsu.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/sw/rxe/rxe.c      |  1 +
 drivers/infiniband/sw/rxe/rxe_loc.h  | 12 +++++++-
 drivers/infiniband/sw/rxe/rxe_mr.c   | 12 --------
 drivers/infiniband/sw/rxe/rxe_odp.c  | 46 ++++++++++++++++++++++++++++
 drivers/infiniband/sw/rxe/rxe_resp.c | 11 ++++++-
 include/rdma/ib_verbs.h              |  1 +
 6 files changed, 69 insertions(+), 14 deletions(-)

diff --git a/drivers/infiniband/sw/rxe/rxe.c b/drivers/infiniband/sw/rxe/rxe.c
index bf711eb9440c..3a77d6db1720 100644
--- a/drivers/infiniband/sw/rxe/rxe.c
+++ b/drivers/infiniband/sw/rxe/rxe.c
@@ -102,6 +102,7 @@ static void rxe_init_device_param(struct rxe_dev *rxe, struct net_device *ndev)
 		rxe->attr.odp_caps.per_transport_caps.rc_odp_caps |= IB_ODP_SUPPORT_ATOMIC;
 		rxe->attr.odp_caps.per_transport_caps.rc_odp_caps |= IB_ODP_SUPPORT_SRQ_RECV;
 		rxe->attr.odp_caps.per_transport_caps.rc_odp_caps |= IB_ODP_SUPPORT_FLUSH;
+		rxe->attr.odp_caps.per_transport_caps.rc_odp_caps |= IB_ODP_SUPPORT_ATOMIC_WRITE;
 	}
 }
 
diff --git a/drivers/infiniband/sw/rxe/rxe_loc.h b/drivers/infiniband/sw/rxe/rxe_loc.h
index 0012bebe96ef..6389d829b06a 100644
--- a/drivers/infiniband/sw/rxe/rxe_loc.h
+++ b/drivers/infiniband/sw/rxe/rxe_loc.h
@@ -187,7 +187,7 @@ static inline unsigned int wr_opcode_mask(int opcode, struct rxe_qp *qp)
 /* rxe_odp.c */
 extern const struct mmu_interval_notifier_ops rxe_mn_ops;
 
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
+#if defined CONFIG_INFINIBAND_ON_DEMAND_PAGING
 int rxe_odp_mr_init_user(struct rxe_dev *rxe, u64 start, u64 length,
 			 u64 iova, int access_flags, struct rxe_mr *mr);
 int rxe_odp_mr_copy(struct rxe_mr *mr, u64 iova, void *addr, int length,
@@ -221,4 +221,14 @@ static inline int rxe_odp_flush_pmem_iova(struct rxe_mr *mr, u64 iova,
 }
 #endif /* CONFIG_INFINIBAND_ON_DEMAND_PAGING */
 
+#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
+enum resp_states rxe_odp_do_atomic_write(struct rxe_mr *mr, u64 iova, u64 value);
+#else
+static inline enum resp_states rxe_odp_do_atomic_write(struct rxe_mr *mr,
+						       u64 iova, u64 value)
+{
+	return RESPST_ERR_UNSUPPORTED_OPCODE;
+}
+#endif /* CONFIG_INFINIBAND_ON_DEMAND_PAGING */
+
 #endif /* RXE_LOC_H */
diff --git a/drivers/infiniband/sw/rxe/rxe_mr.c b/drivers/infiniband/sw/rxe/rxe_mr.c
index 93e4b5acd3ac..d40fbe10633f 100644
--- a/drivers/infiniband/sw/rxe/rxe_mr.c
+++ b/drivers/infiniband/sw/rxe/rxe_mr.c
@@ -547,16 +547,6 @@ int rxe_mr_do_atomic_write(struct rxe_mr *mr, u64 iova, u64 value)
 	struct page *page;
 	u64 *va;
 
-	/* ODP is not supported right now. WIP. */
-	if (mr->umem->is_odp)
-		return RESPST_ERR_UNSUPPORTED_OPCODE;
-
-	/* See IBA oA19-28 */
-	if (unlikely(mr->state != RXE_MR_STATE_VALID)) {
-		rxe_dbg_mr(mr, "mr not in valid state\n");
-		return RESPST_ERR_RKEY_VIOLATION;
-	}
-
 	if (mr->ibmr.type == IB_MR_TYPE_DMA) {
 		page_offset = iova & (PAGE_SIZE - 1);
 		page = ib_virt_dma_to_page(iova);
@@ -584,10 +574,8 @@ int rxe_mr_do_atomic_write(struct rxe_mr *mr, u64 iova, u64 value)
 	}
 
 	va = kmap_local_page(page);
-
 	/* Do atomic write after all prior operations have completed */
 	smp_store_release(&va[page_offset >> 3], value);
-
 	kunmap_local(va);
 
 	return 0;
diff --git a/drivers/infiniband/sw/rxe/rxe_odp.c b/drivers/infiniband/sw/rxe/rxe_odp.c
index 9a9aae967486..02de05d759c6 100644
--- a/drivers/infiniband/sw/rxe/rxe_odp.c
+++ b/drivers/infiniband/sw/rxe/rxe_odp.c
@@ -378,3 +378,49 @@ int rxe_odp_flush_pmem_iova(struct rxe_mr *mr, u64 iova,
 
 	return 0;
 }
+
+/* CONFIG_64BIT=y */
+enum resp_states rxe_odp_do_atomic_write(struct rxe_mr *mr, u64 iova, u64 value)
+{
+	struct ib_umem_odp *umem_odp = to_ib_umem_odp(mr->umem);
+	unsigned int page_offset;
+	unsigned long index;
+	struct page *page;
+	int err;
+	u64 *va;
+
+	/* See IBA oA19-28 */
+	err = mr_check_range(mr, iova, sizeof(value));
+	if (unlikely(err)) {
+		rxe_dbg_mr(mr, "iova out of range\n");
+		return RESPST_ERR_RKEY_VIOLATION;
+	}
+
+	err = rxe_odp_map_range_and_lock(mr, iova, sizeof(value),
+					 RXE_PAGEFAULT_DEFAULT);
+	if (err)
+		return RESPST_ERR_RKEY_VIOLATION;
+
+	page_offset = rxe_odp_iova_to_page_offset(umem_odp, iova);
+	index = rxe_odp_iova_to_index(umem_odp, iova);
+	page = hmm_pfn_to_page(umem_odp->pfn_list[index]);
+	if (!page) {
+		mutex_unlock(&umem_odp->umem_mutex);
+		return RESPST_ERR_RKEY_VIOLATION;
+	}
+	/* See IBA A19.4.2 */
+	if (unlikely(page_offset & 0x7)) {
+		mutex_unlock(&umem_odp->umem_mutex);
+		rxe_dbg_mr(mr, "misaligned address\n");
+		return RESPST_ERR_MISALIGNED_ATOMIC;
+	}
+
+	va = kmap_local_page(page);
+	/* Do atomic write after all prior operations have completed */
+	smp_store_release(&va[page_offset >> 3], value);
+	kunmap_local(va);
+
+	mutex_unlock(&umem_odp->umem_mutex);
+
+	return RESPST_NONE;
+}
diff --git a/drivers/infiniband/sw/rxe/rxe_resp.c b/drivers/infiniband/sw/rxe/rxe_resp.c
index 304e3de740ad..fd7bac5bce18 100644
--- a/drivers/infiniband/sw/rxe/rxe_resp.c
+++ b/drivers/infiniband/sw/rxe/rxe_resp.c
@@ -749,7 +749,16 @@ static enum resp_states atomic_write_reply(struct rxe_qp *qp,
 	value = *(u64 *)payload_addr(pkt);
 	iova = qp->resp.va + qp->resp.offset;
 
-	err = rxe_mr_do_atomic_write(mr, iova, value);
+	/* See IBA oA19-28 */
+	if (unlikely(mr->state != RXE_MR_STATE_VALID)) {
+		rxe_dbg_mr(mr, "mr not in valid state\n");
+		return RESPST_ERR_RKEY_VIOLATION;
+	}
+
+	if (mr->umem->is_odp)
+		err = rxe_odp_do_atomic_write(mr, iova, value);
+	else
+		err = rxe_mr_do_atomic_write(mr, iova, value);
 	if (err)
 		return err;
 
diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index 41bf98ccb275..0a7ccd08b365 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -326,6 +326,7 @@ enum ib_odp_transport_cap_bits {
 	IB_ODP_SUPPORT_ATOMIC	= 1 << 4,
 	IB_ODP_SUPPORT_SRQ_RECV	= 1 << 5,
 	IB_ODP_SUPPORT_FLUSH	= 1 << 6,
+	IB_ODP_SUPPORT_ATOMIC_WRITE	= 1 << 7,
 };
 
 struct ib_odp_caps {

From 9334003d1335b1d086334f946eab158503487e58 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Thomas=20Wei=C3=9Fschuh?= <thomas.weissschuh@linutronix.de>
Date: Mon, 7 Apr 2025 10:25:09 +0200
Subject: [PATCH 11/57] RDMA: Don't use %pK through printk
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

In the past %pK was preferable to %p as it would not leak raw pointer
values into the kernel log.
Since commit ad67b74d2469 ("printk: hash addresses printed with %p")
the regular %p has been improved to avoid this issue.
Furthermore, restricted pointers ("%pK") were never meant to be used
through printk(). They can still unintentionally leak raw pointers or
acquire sleeping looks in atomic contexts.

Switch to the regular pointer formatting which is safer and
easier to reason about.

Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
Link: https://patch.msgid.link/20250407-restricted-pointers-infiniband-v1-1-22b20504b84d@linutronix.de
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/hw/irdma/ctrl.c       | 2 +-
 drivers/infiniband/hw/irdma/pble.c       | 2 +-
 drivers/infiniband/hw/usnic/usnic_uiom.c | 2 +-
 drivers/infiniband/sw/siw/siw.h          | 2 +-
 drivers/infiniband/sw/siw/siw_cq.c       | 2 +-
 drivers/infiniband/sw/siw/siw_mem.c      | 4 ++--
 drivers/infiniband/sw/siw/siw_qp_rx.c    | 8 ++++----
 drivers/infiniband/sw/siw/siw_verbs.c    | 6 +++---
 8 files changed, 14 insertions(+), 14 deletions(-)

diff --git a/drivers/infiniband/hw/irdma/ctrl.c b/drivers/infiniband/hw/irdma/ctrl.c
index 6aed6169c07d..99a7f1a6c0b5 100644
--- a/drivers/infiniband/hw/irdma/ctrl.c
+++ b/drivers/infiniband/hw/irdma/ctrl.c
@@ -3131,7 +3131,7 @@ int irdma_sc_cqp_init(struct irdma_sc_cqp *cqp,
 	writel(0, cqp->dev->hw_regs[IRDMA_CCQPSTATUS]);
 
 	ibdev_dbg(to_ibdev(cqp->dev),
-		  "WQE: sq_size[%04d] hw_sq_size[%04d] sq_base[%p] sq_pa[%pK] cqp[%p] polarity[x%04x]\n",
+		  "WQE: sq_size[%04d] hw_sq_size[%04d] sq_base[%p] sq_pa[%p] cqp[%p] polarity[x%04x]\n",
 		  cqp->sq_size, cqp->hw_sq_size, cqp->sq_base,
 		  (u64 *)(uintptr_t)cqp->sq_pa, cqp, cqp->polarity);
 	return 0;
diff --git a/drivers/infiniband/hw/irdma/pble.c b/drivers/infiniband/hw/irdma/pble.c
index e7ce6840755f..37ce35cb10e7 100644
--- a/drivers/infiniband/hw/irdma/pble.c
+++ b/drivers/infiniband/hw/irdma/pble.c
@@ -108,7 +108,7 @@ static int add_sd_direct(struct irdma_hmc_pble_rsrc *pble_rsrc,
 	chunk->vaddr = sd_entry->u.bp.addr.va + offset;
 	chunk->fpm_addr = pble_rsrc->next_fpm_addr;
 	ibdev_dbg(to_ibdev(dev),
-		  "PBLE: chunk_size[%lld] = 0x%llx vaddr=0x%pK fpm_addr = %llx\n",
+		  "PBLE: chunk_size[%lld] = 0x%llx vaddr=0x%p fpm_addr = %llx\n",
 		  chunk->size, chunk->size, chunk->vaddr, chunk->fpm_addr);
 
 	return 0;
diff --git a/drivers/infiniband/hw/usnic/usnic_uiom.c b/drivers/infiniband/hw/usnic/usnic_uiom.c
index f948b76f984d..3fbf99757b11 100644
--- a/drivers/infiniband/hw/usnic/usnic_uiom.c
+++ b/drivers/infiniband/hw/usnic/usnic_uiom.c
@@ -56,7 +56,7 @@ static int usnic_uiom_dma_fault(struct iommu_domain *domain,
 				unsigned long iova, int flags,
 				void *token)
 {
-	usnic_err("Device %s iommu fault domain 0x%pK va 0x%lx flags 0x%x\n",
+	usnic_err("Device %s iommu fault domain 0x%p va 0x%lx flags 0x%x\n",
 		dev_name(dev),
 		domain, iova, flags);
 	return -ENOSYS;
diff --git a/drivers/infiniband/sw/siw/siw.h b/drivers/infiniband/sw/siw/siw.h
index 385067e07faf..e49c8a76e22e 100644
--- a/drivers/infiniband/sw/siw/siw.h
+++ b/drivers/infiniband/sw/siw/siw.h
@@ -738,7 +738,7 @@ static inline void siw_crc_skb(struct siw_rx_stream *srx, unsigned int len)
 		  "MEM[0x%08x] %s: " fmt, mem->stag, __func__, ##__VA_ARGS__)
 
 #define siw_dbg_cep(cep, fmt, ...)                                             \
-	ibdev_dbg(&cep->sdev->base_dev, "CEP[0x%pK] %s: " fmt,                 \
+	ibdev_dbg(&cep->sdev->base_dev, "CEP[0x%p] %s: " fmt,                 \
 		  cep, __func__, ##__VA_ARGS__)
 
 void siw_cq_flush(struct siw_cq *cq);
diff --git a/drivers/infiniband/sw/siw/siw_cq.c b/drivers/infiniband/sw/siw/siw_cq.c
index f3c2226aff94..25b3c741b66b 100644
--- a/drivers/infiniband/sw/siw/siw_cq.c
+++ b/drivers/infiniband/sw/siw/siw_cq.c
@@ -72,7 +72,7 @@ int siw_reap_cqe(struct siw_cq *cq, struct ib_wc *wc)
 			wc->opcode = map_wc_opcode[cqe->opcode];
 			wc->status = map_cqe_status[cqe->status].ib;
 			siw_dbg_cq(cq,
-				   "idx %u, type %d, flags %2x, id 0x%pK\n",
+				   "idx %u, type %d, flags %2x, id 0x%p\n",
 				   cq->cq_get % cq->num_cqe, cqe->opcode,
 				   cqe->flags, (void *)(uintptr_t)cqe->id);
 		} else {
diff --git a/drivers/infiniband/sw/siw/siw_mem.c b/drivers/infiniband/sw/siw/siw_mem.c
index dcb963607c8b..b17156995595 100644
--- a/drivers/infiniband/sw/siw/siw_mem.c
+++ b/drivers/infiniband/sw/siw/siw_mem.c
@@ -181,10 +181,10 @@ int siw_check_mem(struct ib_pd *pd, struct siw_mem *mem, u64 addr,
 	 */
 	if (addr < mem->va || addr + len > mem->va + mem->len) {
 		siw_dbg_pd(pd, "MEM interval len %d\n", len);
-		siw_dbg_pd(pd, "[0x%pK, 0x%pK] out of bounds\n",
+		siw_dbg_pd(pd, "[0x%p, 0x%p] out of bounds\n",
 			   (void *)(uintptr_t)addr,
 			   (void *)(uintptr_t)(addr + len));
-		siw_dbg_pd(pd, "[0x%pK, 0x%pK] STag=0x%08x\n",
+		siw_dbg_pd(pd, "[0x%p, 0x%p] STag=0x%08x\n",
 			   (void *)(uintptr_t)mem->va,
 			   (void *)(uintptr_t)(mem->va + mem->len),
 			   mem->stag);
diff --git a/drivers/infiniband/sw/siw/siw_qp_rx.c b/drivers/infiniband/sw/siw/siw_qp_rx.c
index 32554eba1eac..a10820e33887 100644
--- a/drivers/infiniband/sw/siw/siw_qp_rx.c
+++ b/drivers/infiniband/sw/siw/siw_qp_rx.c
@@ -38,7 +38,7 @@ static int siw_rx_umem(struct siw_rx_stream *srx, struct siw_umem *umem,
 
 		p = siw_get_upage(umem, dest_addr);
 		if (unlikely(!p)) {
-			pr_warn("siw: %s: [QP %u]: bogus addr: %pK, %pK\n",
+			pr_warn("siw: %s: [QP %u]: bogus addr: %p, %p\n",
 				__func__, qp_id(rx_qp(srx)),
 				(void *)(uintptr_t)dest_addr,
 				(void *)(uintptr_t)umem->fp_addr);
@@ -51,7 +51,7 @@ static int siw_rx_umem(struct siw_rx_stream *srx, struct siw_umem *umem,
 		pg_off = dest_addr & ~PAGE_MASK;
 		bytes = min(len, (int)PAGE_SIZE - pg_off);
 
-		siw_dbg_qp(rx_qp(srx), "page %pK, bytes=%u\n", p, bytes);
+		siw_dbg_qp(rx_qp(srx), "page %p, bytes=%u\n", p, bytes);
 
 		dest = kmap_atomic(p);
 		rv = skb_copy_bits(srx->skb, srx->skb_offset, dest + pg_off,
@@ -105,11 +105,11 @@ static int siw_rx_kva(struct siw_rx_stream *srx, void *kva, int len)
 {
 	int rv;
 
-	siw_dbg_qp(rx_qp(srx), "kva: 0x%pK, len: %u\n", kva, len);
+	siw_dbg_qp(rx_qp(srx), "kva: 0x%p, len: %u\n", kva, len);
 
 	rv = skb_copy_bits(srx->skb, srx->skb_offset, kva, len);
 	if (unlikely(rv)) {
-		pr_warn("siw: [QP %u]: %s, len %d, kva 0x%pK, rv %d\n",
+		pr_warn("siw: [QP %u]: %s, len %d, kva 0x%p, rv %d\n",
 			qp_id(rx_qp(srx)), __func__, len, kva, rv);
 
 		return rv;
diff --git a/drivers/infiniband/sw/siw/siw_verbs.c b/drivers/infiniband/sw/siw/siw_verbs.c
index fd7b266a221b..7ce0035c54fa 100644
--- a/drivers/infiniband/sw/siw/siw_verbs.c
+++ b/drivers/infiniband/sw/siw/siw_verbs.c
@@ -936,7 +936,7 @@ int siw_post_send(struct ib_qp *base_qp, const struct ib_send_wr *wr,
 			rv = -EINVAL;
 			break;
 		}
-		siw_dbg_qp(qp, "opcode %d, flags 0x%x, wr_id 0x%pK\n",
+		siw_dbg_qp(qp, "opcode %d, flags 0x%x, wr_id 0x%p\n",
 			   sqe->opcode, sqe->flags,
 			   (void *)(uintptr_t)sqe->id);
 
@@ -1332,7 +1332,7 @@ struct ib_mr *siw_reg_user_mr(struct ib_pd *pd, u64 start, u64 len,
 	struct siw_device *sdev = to_siw_dev(pd->device);
 	int rv;
 
-	siw_dbg_pd(pd, "start: 0x%pK, va: 0x%pK, len: %llu\n",
+	siw_dbg_pd(pd, "start: 0x%p, va: 0x%p, len: %llu\n",
 		   (void *)(uintptr_t)start, (void *)(uintptr_t)rnic_va,
 		   (unsigned long long)len);
 
@@ -1525,7 +1525,7 @@ int siw_map_mr_sg(struct ib_mr *base_mr, struct scatterlist *sl, int num_sle,
 		mem->len = base_mr->length;
 		mem->va = base_mr->iova;
 		siw_dbg_mem(mem,
-			"%llu bytes, start 0x%pK, %u SLE to %u entries\n",
+			"%llu bytes, start 0x%p, %u SLE to %u entries\n",
 			mem->len, (void *)(uintptr_t)mem->va, num_sle,
 			pbl->num_buf);
 	}

From 29610226c33ffee687dedfdf1f19cbb2b117e9e7 Mon Sep 17 00:00:00 2001
From: Daisuke Matsuda <matsuda-daisuke@fujitsu.com>
Date: Wed, 9 Apr 2025 19:27:01 +0900
Subject: [PATCH 12/57] RDMA/rxe: Fix mismatched type declarations

Some functions return int values while they are defined as enum resp_states
variables. This patch resolves the mismatches in rxe.

Signed-off-by: Daisuke Matsuda <matsuda-daisuke@fujitsu.com>
Link: https://patch.msgid.link/20250409102701.1275265-1-matsuda-daisuke@fujitsu.com
Reviewed-by: Zhu Yanjun <yanjun.zhu@linux.dev>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/sw/rxe/rxe_loc.h | 14 +++++++-------
 drivers/infiniband/sw/rxe/rxe_mr.c  | 12 ++++++------
 drivers/infiniband/sw/rxe/rxe_odp.c | 11 ++++++-----
 3 files changed, 19 insertions(+), 18 deletions(-)

diff --git a/drivers/infiniband/sw/rxe/rxe_loc.h b/drivers/infiniband/sw/rxe/rxe_loc.h
index 6389d829b06a..30fa83c9c846 100644
--- a/drivers/infiniband/sw/rxe/rxe_loc.h
+++ b/drivers/infiniband/sw/rxe/rxe_loc.h
@@ -70,9 +70,9 @@ int copy_data(struct rxe_pd *pd, int access, struct rxe_dma_info *dma,
 	      void *addr, int length, enum rxe_mr_copy_dir dir);
 int rxe_map_mr_sg(struct ib_mr *ibmr, struct scatterlist *sg,
 		  int sg_nents, unsigned int *sg_offset);
-int rxe_mr_do_atomic_op(struct rxe_mr *mr, u64 iova, int opcode,
-			u64 compare, u64 swap_add, u64 *orig_val);
-int rxe_mr_do_atomic_write(struct rxe_mr *mr, u64 iova, u64 value);
+enum resp_states rxe_mr_do_atomic_op(struct rxe_mr *mr, u64 iova, int opcode,
+				     u64 compare, u64 swap_add, u64 *orig_val);
+enum resp_states rxe_mr_do_atomic_write(struct rxe_mr *mr, u64 iova, u64 value);
 struct rxe_mr *lookup_mr(struct rxe_pd *pd, int access, u32 key,
 			 enum rxe_mr_lookup_type type);
 int mr_check_range(struct rxe_mr *mr, u64 iova, size_t length);
@@ -192,8 +192,8 @@ int rxe_odp_mr_init_user(struct rxe_dev *rxe, u64 start, u64 length,
 			 u64 iova, int access_flags, struct rxe_mr *mr);
 int rxe_odp_mr_copy(struct rxe_mr *mr, u64 iova, void *addr, int length,
 		    enum rxe_mr_copy_dir dir);
-int rxe_odp_atomic_op(struct rxe_mr *mr, u64 iova, int opcode,
-			 u64 compare, u64 swap_add, u64 *orig_val);
+enum resp_states rxe_odp_atomic_op(struct rxe_mr *mr, u64 iova, int opcode,
+				   u64 compare, u64 swap_add, u64 *orig_val);
 int rxe_odp_flush_pmem_iova(struct rxe_mr *mr, u64 iova,
 			    unsigned int length);
 #else /* CONFIG_INFINIBAND_ON_DEMAND_PAGING */
@@ -208,9 +208,9 @@ static inline int rxe_odp_mr_copy(struct rxe_mr *mr, u64 iova, void *addr,
 {
 	return -EOPNOTSUPP;
 }
-static inline int
+static inline enum resp_states
 rxe_odp_atomic_op(struct rxe_mr *mr, u64 iova, int opcode,
-		     u64 compare, u64 swap_add, u64 *orig_val)
+		  u64 compare, u64 swap_add, u64 *orig_val)
 {
 	return RESPST_ERR_UNSUPPORTED_OPCODE;
 }
diff --git a/drivers/infiniband/sw/rxe/rxe_mr.c b/drivers/infiniband/sw/rxe/rxe_mr.c
index d40fbe10633f..1a74013a14ab 100644
--- a/drivers/infiniband/sw/rxe/rxe_mr.c
+++ b/drivers/infiniband/sw/rxe/rxe_mr.c
@@ -483,8 +483,8 @@ int rxe_flush_pmem_iova(struct rxe_mr *mr, u64 start, unsigned int length)
 /* Guarantee atomicity of atomic operations at the machine level. */
 DEFINE_SPINLOCK(atomic_ops_lock);
 
-int rxe_mr_do_atomic_op(struct rxe_mr *mr, u64 iova, int opcode,
-			u64 compare, u64 swap_add, u64 *orig_val)
+enum resp_states rxe_mr_do_atomic_op(struct rxe_mr *mr, u64 iova, int opcode,
+				     u64 compare, u64 swap_add, u64 *orig_val)
 {
 	unsigned int page_offset;
 	struct page *page;
@@ -536,12 +536,12 @@ int rxe_mr_do_atomic_op(struct rxe_mr *mr, u64 iova, int opcode,
 
 	kunmap_local(va);
 
-	return 0;
+	return RESPST_NONE;
 }
 
 #if defined CONFIG_64BIT
 /* only implemented or called for 64 bit architectures */
-int rxe_mr_do_atomic_write(struct rxe_mr *mr, u64 iova, u64 value)
+enum resp_states rxe_mr_do_atomic_write(struct rxe_mr *mr, u64 iova, u64 value)
 {
 	unsigned int page_offset;
 	struct page *page;
@@ -578,10 +578,10 @@ int rxe_mr_do_atomic_write(struct rxe_mr *mr, u64 iova, u64 value)
 	smp_store_release(&va[page_offset >> 3], value);
 	kunmap_local(va);
 
-	return 0;
+	return RESPST_NONE;
 }
 #else
-int rxe_mr_do_atomic_write(struct rxe_mr *mr, u64 iova, u64 value)
+enum resp_states rxe_mr_do_atomic_write(struct rxe_mr *mr, u64 iova, u64 value)
 {
 	return RESPST_ERR_UNSUPPORTED_OPCODE;
 }
diff --git a/drivers/infiniband/sw/rxe/rxe_odp.c b/drivers/infiniband/sw/rxe/rxe_odp.c
index 02de05d759c6..fa5e8f5017cc 100644
--- a/drivers/infiniband/sw/rxe/rxe_odp.c
+++ b/drivers/infiniband/sw/rxe/rxe_odp.c
@@ -266,8 +266,9 @@ int rxe_odp_mr_copy(struct rxe_mr *mr, u64 iova, void *addr, int length,
 	return err;
 }
 
-static int rxe_odp_do_atomic_op(struct rxe_mr *mr, u64 iova, int opcode,
-				u64 compare, u64 swap_add, u64 *orig_val)
+static enum resp_states rxe_odp_do_atomic_op(struct rxe_mr *mr, u64 iova,
+					     int opcode, u64 compare,
+					     u64 swap_add, u64 *orig_val)
 {
 	struct ib_umem_odp *umem_odp = to_ib_umem_odp(mr->umem);
 	unsigned int page_offset;
@@ -315,11 +316,11 @@ static int rxe_odp_do_atomic_op(struct rxe_mr *mr, u64 iova, int opcode,
 
 	kunmap_local(va);
 
-	return 0;
+	return RESPST_NONE;
 }
 
-int rxe_odp_atomic_op(struct rxe_mr *mr, u64 iova, int opcode,
-			 u64 compare, u64 swap_add, u64 *orig_val)
+enum resp_states rxe_odp_atomic_op(struct rxe_mr *mr, u64 iova, int opcode,
+				   u64 compare, u64 swap_add, u64 *orig_val)
 {
 	struct ib_umem_odp *umem_odp = to_ib_umem_odp(mr->umem);
 	int err;

From ffe1cee21f8b533ae27c3a31bfa56b8c1b27fa6e Mon Sep 17 00:00:00 2001
From: Chen Linxuan <chenlinxuan@uniontech.com>
Date: Fri, 11 Apr 2025 18:54:53 +0800
Subject: [PATCH 13/57] RDMA/hns: initialize db in update_srq_db()

On x86_64 with gcc version 13.3.0, I compile
drivers/infiniband/hw/hns/hns_roce_hw_v2.c with:

  make defconfig
  ./scripts/kconfig/merge_config.sh .config <(
    echo CONFIG_COMPILE_TEST=y
    echo CONFIG_HNS3=m
    echo CONFIG_INFINIBAND=m
    echo CONFIG_INFINIBAND_HNS_HIP08=m
  )
  make KCFLAGS="-fno-inline-small-functions -fno-inline-functions-called-once" \
    drivers/infiniband/hw/hns/hns_roce_hw_v2.o

Then I get a compile error:

    CALL    scripts/checksyscalls.sh
    DESCEND objtool
    INSTALL libsubcmd_headers
    CC [M]  drivers/infiniband/hw/hns/hns_roce_hw_v2.o
  In file included from drivers/infiniband/hw/hns/hns_roce_hw_v2.c:47:
  drivers/infiniband/hw/hns/hns_roce_hw_v2.c: In function 'update_srq_db':
  drivers/infiniband/hw/hns/hns_roce_common.h:74:17: error: 'db' is used uninitialized [-Werror=uninitialized]
     74 |                 *((__le32 *)_ptr + (field_h) / 32) &=                          \
        |                 ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  drivers/infiniband/hw/hns/hns_roce_common.h:90:17: note: in expansion of macro '_hr_reg_clear'
     90 |                 _hr_reg_clear(ptr, field_type, field_h, field_l);              \
        |                 ^~~~~~~~~~~~~
  drivers/infiniband/hw/hns/hns_roce_common.h:95:39: note: in expansion of macro '_hr_reg_write'
     95 | #define hr_reg_write(ptr, field, val) _hr_reg_write(ptr, field, val)
        |                                       ^~~~~~~~~~~~~
  drivers/infiniband/hw/hns/hns_roce_hw_v2.c:948:9: note: in expansion of macro 'hr_reg_write'
    948 |         hr_reg_write(&db, DB_TAG, srq->srqn);
        |         ^~~~~~~~~~~~
  drivers/infiniband/hw/hns/hns_roce_hw_v2.c:946:31: note: 'db' declared here
    946 |         struct hns_roce_v2_db db;
        |                               ^~
  cc1: all warnings being treated as errors

Signed-off-by: Chen Linxuan <chenlinxuan@uniontech.com>
Co-developed-by: Winston Wen <wentao@uniontech.com>
Signed-off-by: Winston Wen <wentao@uniontech.com>
Link: https://patch.msgid.link/FF922C77946229B6+20250411105459.90782-5-chenlinxuan@uniontech.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/hw/hns/hns_roce_hw_v2.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
index 0600e87936bf..c6399490e3a5 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
@@ -943,7 +943,7 @@ static void fill_wqe_idx(struct hns_roce_srq *srq, unsigned int wqe_idx)
 static void update_srq_db(struct hns_roce_srq *srq)
 {
 	struct hns_roce_dev *hr_dev = to_hr_dev(srq->ibsrq.device);
-	struct hns_roce_v2_db db;
+	struct hns_roce_v2_db db = {};
 
 	hr_reg_write(&db, DB_TAG, srq->srqn);
 	hr_reg_write(&db, DB_CMD, HNS_ROCE_V2_SRQ_DB);

From 7d40ccf018694ae894b37d7e849cde116eb37627 Mon Sep 17 00:00:00 2001
From: Konstantin Taranov <kotaranov@microsoft.com>
Date: Mon, 14 Apr 2025 02:00:32 -0700
Subject: [PATCH 14/57] RDMA/mana_ib: Access remote atomic for MRs

Add IB_ACCESS_REMOTE_ATOMIC to the valid flags for MRs and use
the corresponding flag bit during MR creation in the HW.

Signed-off-by: Konstantin Taranov <kotaranov@microsoft.com>
Link: https://patch.msgid.link/1744621234-26114-2-git-send-email-kotaranov@linux.microsoft.com
Reviewed-by: Long Li <longli@microsoft.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/hw/mana/mr.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/drivers/infiniband/hw/mana/mr.c b/drivers/infiniband/hw/mana/mr.c
index f99557ec7767..e4a9f537a6d0 100644
--- a/drivers/infiniband/hw/mana/mr.c
+++ b/drivers/infiniband/hw/mana/mr.c
@@ -5,8 +5,8 @@
 
 #include "mana_ib.h"
 
-#define VALID_MR_FLAGS                                                         \
-	(IB_ACCESS_LOCAL_WRITE | IB_ACCESS_REMOTE_WRITE | IB_ACCESS_REMOTE_READ)
+#define VALID_MR_FLAGS (IB_ACCESS_LOCAL_WRITE | IB_ACCESS_REMOTE_WRITE | IB_ACCESS_REMOTE_READ |\
+			IB_ACCESS_REMOTE_ATOMIC)
 
 #define VALID_DMA_MR_FLAGS (IB_ACCESS_LOCAL_WRITE)
 
@@ -24,6 +24,9 @@ mana_ib_verbs_to_gdma_access_flags(int access_flags)
 	if (access_flags & IB_ACCESS_REMOTE_READ)
 		flags |= GDMA_ACCESS_FLAG_REMOTE_READ;
 
+	if (access_flags & IB_ACCESS_REMOTE_ATOMIC)
+		flags |= GDMA_ACCESS_FLAG_REMOTE_ATOMIC;
+
 	return flags;
 }
 

From 8f49682d94f3a12a6a3e636a07bbe57c80329d1d Mon Sep 17 00:00:00 2001
From: Konstantin Taranov <kotaranov@microsoft.com>
Date: Mon, 14 Apr 2025 02:00:33 -0700
Subject: [PATCH 15/57] RDMA/mana_ib: support of the zero based MRs

Add IB_ZERO_BASED to the valid flags and use
the corresponding MR creation request for the zero
based memory.

Signed-off-by: Konstantin Taranov <kotaranov@microsoft.com>
Link: https://patch.msgid.link/1744621234-26114-3-git-send-email-kotaranov@linux.microsoft.com
Reviewed-by: Long Li <longli@microsoft.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/hw/mana/mr.c | 24 +++++++++++++++++-------
 include/net/mana/gdma.h         | 11 ++++++++++-
 2 files changed, 27 insertions(+), 8 deletions(-)

diff --git a/drivers/infiniband/hw/mana/mr.c b/drivers/infiniband/hw/mana/mr.c
index e4a9f537a6d0..6d974d0a8400 100644
--- a/drivers/infiniband/hw/mana/mr.c
+++ b/drivers/infiniband/hw/mana/mr.c
@@ -6,7 +6,7 @@
 #include "mana_ib.h"
 
 #define VALID_MR_FLAGS (IB_ACCESS_LOCAL_WRITE | IB_ACCESS_REMOTE_WRITE | IB_ACCESS_REMOTE_READ |\
-			IB_ACCESS_REMOTE_ATOMIC)
+			IB_ACCESS_REMOTE_ATOMIC | IB_ZERO_BASED)
 
 #define VALID_DMA_MR_FLAGS (IB_ACCESS_LOCAL_WRITE)
 
@@ -51,7 +51,10 @@ static int mana_ib_gd_create_mr(struct mana_ib_dev *dev, struct mana_ib_mr *mr,
 		req.gva.virtual_address = mr_params->gva.virtual_address;
 		req.gva.access_flags = mr_params->gva.access_flags;
 		break;
-
+	case GDMA_MR_TYPE_ZBVA:
+		req.zbva.dma_region_handle = mr_params->zbva.dma_region_handle;
+		req.zbva.access_flags = mr_params->zbva.access_flags;
+		break;
 	default:
 		ibdev_dbg(&dev->ib_dev,
 			  "invalid param (GDMA_MR_TYPE) passed, type %d\n",
@@ -147,11 +150,18 @@ struct ib_mr *mana_ib_reg_user_mr(struct ib_pd *ibpd, u64 start, u64 length,
 		  dma_region_handle);
 
 	mr_params.pd_handle = pd->pd_handle;
-	mr_params.mr_type = GDMA_MR_TYPE_GVA;
-	mr_params.gva.dma_region_handle = dma_region_handle;
-	mr_params.gva.virtual_address = iova;
-	mr_params.gva.access_flags =
-		mana_ib_verbs_to_gdma_access_flags(access_flags);
+	if (access_flags & IB_ZERO_BASED) {
+		mr_params.mr_type = GDMA_MR_TYPE_ZBVA;
+		mr_params.zbva.dma_region_handle = dma_region_handle;
+		mr_params.zbva.access_flags =
+			mana_ib_verbs_to_gdma_access_flags(access_flags);
+	} else {
+		mr_params.mr_type = GDMA_MR_TYPE_GVA;
+		mr_params.gva.dma_region_handle = dma_region_handle;
+		mr_params.gva.virtual_address = iova;
+		mr_params.gva.access_flags =
+			mana_ib_verbs_to_gdma_access_flags(access_flags);
+	}
 
 	err = mana_ib_gd_create_mr(dev, mr, &mr_params);
 	if (err)
diff --git a/include/net/mana/gdma.h b/include/net/mana/gdma.h
index 228603bf03f2..239a70032550 100644
--- a/include/net/mana/gdma.h
+++ b/include/net/mana/gdma.h
@@ -815,6 +815,8 @@ enum gdma_mr_type {
 	 * address that is set up in the MST
 	 */
 	GDMA_MR_TYPE_GVA = 2,
+	/* Guest zero-based address MRs */
+	GDMA_MR_TYPE_ZBVA = 4,
 };
 
 struct gdma_create_mr_params {
@@ -826,6 +828,10 @@ struct gdma_create_mr_params {
 			u64 virtual_address;
 			enum gdma_mr_access_flags access_flags;
 		} gva;
+		struct {
+			u64 dma_region_handle;
+			enum gdma_mr_access_flags access_flags;
+		} zbva;
 	};
 };
 
@@ -841,7 +847,10 @@ struct gdma_create_mr_request {
 			u64 virtual_address;
 			enum gdma_mr_access_flags access_flags;
 		} gva;
-
+		struct {
+			u64 dma_region_handle;
+			enum gdma_mr_access_flags access_flags;
+		} zbva;
 	};
 	u32 reserved_2;
 };/* HW DATA */

From f1652d76f4c51b5aefd14706eecbd70f05ca987a Mon Sep 17 00:00:00 2001
From: Konstantin Taranov <kotaranov@microsoft.com>
Date: Mon, 14 Apr 2025 02:00:34 -0700
Subject: [PATCH 16/57] RDMA/mana_ib: Add support of 4M, 1G, and 2G pages

Check PF capability flag whether the 4M, 1G, and 2G pages are
supported. Add these pages sizes to mana_ib, if supported.

Define possible page sizes in enum gdma_page_type and
remove unused enum atb_page_size.

Signed-off-by: Konstantin Taranov <kotaranov@microsoft.com>
Link: https://patch.msgid.link/1744621234-26114-4-git-send-email-kotaranov@linux.microsoft.com
Reviewed-by: Long Li <longli@microsoft.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/hw/mana/main.c               | 10 +++++++---
 drivers/infiniband/hw/mana/mana_ib.h            |  1 +
 drivers/net/ethernet/microsoft/mana/gdma_main.c |  1 +
 include/net/mana/gdma.h                         | 17 +++--------------
 4 files changed, 12 insertions(+), 17 deletions(-)

diff --git a/drivers/infiniband/hw/mana/main.c b/drivers/infiniband/hw/mana/main.c
index eda9c5b971de..bb0f685babe6 100644
--- a/drivers/infiniband/hw/mana/main.c
+++ b/drivers/infiniband/hw/mana/main.c
@@ -479,7 +479,7 @@ int mana_ib_create_dma_region(struct mana_ib_dev *dev, struct ib_umem *umem,
 {
 	unsigned long page_sz;
 
-	page_sz = ib_umem_find_best_pgsz(umem, PAGE_SZ_BM, virt);
+	page_sz = ib_umem_find_best_pgsz(umem, dev->adapter_caps.page_size_cap, virt);
 	if (!page_sz) {
 		ibdev_dbg(&dev->ib_dev, "Failed to find page size.\n");
 		return -EINVAL;
@@ -494,7 +494,7 @@ int mana_ib_create_zero_offset_dma_region(struct mana_ib_dev *dev, struct ib_ume
 	unsigned long page_sz;
 
 	/* Hardware requires dma region to align to chosen page size */
-	page_sz = ib_umem_find_best_pgoff(umem, PAGE_SZ_BM, 0);
+	page_sz = ib_umem_find_best_pgoff(umem, dev->adapter_caps.page_size_cap, 0);
 	if (!page_sz) {
 		ibdev_dbg(&dev->ib_dev, "Failed to find page size.\n");
 		return -EINVAL;
@@ -577,7 +577,7 @@ int mana_ib_query_device(struct ib_device *ibdev, struct ib_device_attr *props,
 
 	memset(props, 0, sizeof(*props));
 	props->max_mr_size = MANA_IB_MAX_MR_SIZE;
-	props->page_size_cap = PAGE_SZ_BM;
+	props->page_size_cap = dev->adapter_caps.page_size_cap;
 	props->max_qp = dev->adapter_caps.max_qp_count;
 	props->max_qp_wr = dev->adapter_caps.max_qp_wr;
 	props->device_cap_flags = IB_DEVICE_RC_RNR_NAK_GEN;
@@ -696,6 +696,10 @@ int mana_ib_gd_query_adapter_caps(struct mana_ib_dev *dev)
 	caps->max_recv_sge_count = resp.max_recv_sge_count;
 	caps->feature_flags = resp.feature_flags;
 
+	caps->page_size_cap = PAGE_SZ_BM;
+	if (mdev_to_gc(dev)->pf_cap_flags1 & GDMA_DRV_CAP_FLAG_1_GDMA_PAGES_4MB_1GB_2GB)
+		caps->page_size_cap |= (SZ_4M | SZ_1G | SZ_2G);
+
 	return 0;
 }
 
diff --git a/drivers/infiniband/hw/mana/mana_ib.h b/drivers/infiniband/hw/mana/mana_ib.h
index 6903946677e5..f0dbd90b8600 100644
--- a/drivers/infiniband/hw/mana/mana_ib.h
+++ b/drivers/infiniband/hw/mana/mana_ib.h
@@ -60,6 +60,7 @@ struct mana_ib_adapter_caps {
 	u32 max_recv_sge_count;
 	u32 max_inline_data_size;
 	u64 feature_flags;
+	u64 page_size_cap;
 };
 
 struct mana_ib_queue {
diff --git a/drivers/net/ethernet/microsoft/mana/gdma_main.c b/drivers/net/ethernet/microsoft/mana/gdma_main.c
index 4ffaf7588885..8ee1aa3a7ec3 100644
--- a/drivers/net/ethernet/microsoft/mana/gdma_main.c
+++ b/drivers/net/ethernet/microsoft/mana/gdma_main.c
@@ -964,6 +964,7 @@ int mana_gd_verify_vf_version(struct pci_dev *pdev)
 			err, resp.hdr.status);
 		return err ? err : -EPROTO;
 	}
+	gc->pf_cap_flags1 = resp.pf_cap_flags1;
 	if (resp.pf_cap_flags1 & GDMA_DRV_CAP_FLAG_1_HWC_TIMEOUT_RECONFIG) {
 		err = mana_gd_query_hwc_timeout(pdev, &hwc->hwc_timeout);
 		if (err) {
diff --git a/include/net/mana/gdma.h b/include/net/mana/gdma.h
index 239a70032550..ffa9820f14ba 100644
--- a/include/net/mana/gdma.h
+++ b/include/net/mana/gdma.h
@@ -407,6 +407,8 @@ struct gdma_context {
 
 	/* Azure RDMA adapter */
 	struct gdma_dev		mana_ib;
+
+	u64 pf_cap_flags1;
 };
 
 static inline bool mana_gd_is_mana(struct gdma_dev *gd)
@@ -553,6 +555,7 @@ enum {
  */
 #define GDMA_DRV_CAP_FLAG_1_NAPI_WKDONE_FIX BIT(2)
 #define GDMA_DRV_CAP_FLAG_1_HWC_TIMEOUT_RECONFIG BIT(3)
+#define GDMA_DRV_CAP_FLAG_1_GDMA_PAGES_4MB_1GB_2GB BIT(4)
 #define GDMA_DRV_CAP_FLAG_1_VARIABLE_INDIRECTION_TABLE_SUPPORT BIT(5)
 
 /* Driver can handle holes (zeros) in the device list */
@@ -707,20 +710,6 @@ struct gdma_query_hwc_timeout_resp {
 	u32 reserved;
 };
 
-enum atb_page_size {
-	ATB_PAGE_SIZE_4K,
-	ATB_PAGE_SIZE_8K,
-	ATB_PAGE_SIZE_16K,
-	ATB_PAGE_SIZE_32K,
-	ATB_PAGE_SIZE_64K,
-	ATB_PAGE_SIZE_128K,
-	ATB_PAGE_SIZE_256K,
-	ATB_PAGE_SIZE_512K,
-	ATB_PAGE_SIZE_1M,
-	ATB_PAGE_SIZE_2M,
-	ATB_PAGE_SIZE_MAX,
-};
-
 enum gdma_mr_access_flags {
 	GDMA_ACCESS_FLAG_LOCAL_READ = BIT_ULL(0),
 	GDMA_ACCESS_FLAG_LOCAL_WRITE = BIT_ULL(1),

From 04039390cc3cb3d1e6ce6bb1680cbdfe117d6473 Mon Sep 17 00:00:00 2001
From: "Dr. David Alan Gilbert" <linux@treblig.org>
Date: Fri, 18 Apr 2025 17:58:48 +0100
Subject: [PATCH 17/57] RDMA/cma: Remove unused rdma_res_to_id

The last use of rdma_res_to_id() was removed in 2020 by
commi t211cd9459fda ("RDMA: Add dedicated CM_ID resource tracker function")

Remove it.

Signed-off-by: Dr. David Alan Gilbert <linux@treblig.org>
Link: https://patch.msgid.link/20250418165848.241305-1-linux@treblig.org
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/core/cma.c | 13 -------------
 include/rdma/rdma_cm.h        |  1 -
 2 files changed, 14 deletions(-)

diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c
index fedcdb56fb6b..cd7df623161d 100644
--- a/drivers/infiniband/core/cma.c
+++ b/drivers/infiniband/core/cma.c
@@ -144,19 +144,6 @@ struct iw_cm_id *rdma_iw_cm_id(struct rdma_cm_id *id)
 }
 EXPORT_SYMBOL(rdma_iw_cm_id);
 
-/**
- * rdma_res_to_id() - return the rdma_cm_id pointer for this restrack.
- * @res: rdma resource tracking entry pointer
- */
-struct rdma_cm_id *rdma_res_to_id(struct rdma_restrack_entry *res)
-{
-	struct rdma_id_private *id_priv =
-		container_of(res, struct rdma_id_private, res);
-
-	return &id_priv->id;
-}
-EXPORT_SYMBOL(rdma_res_to_id);
-
 static int cma_add_one(struct ib_device *device);
 static void cma_remove_one(struct ib_device *device, void *client_data);
 
diff --git a/include/rdma/rdma_cm.h b/include/rdma/rdma_cm.h
index 8a8ab2f793ab..d1593ad47e28 100644
--- a/include/rdma/rdma_cm.h
+++ b/include/rdma/rdma_cm.h
@@ -388,6 +388,5 @@ void rdma_read_gids(struct rdma_cm_id *cm_id, union ib_gid *sgid,
 		    union ib_gid *dgid);
 
 struct iw_cm_id *rdma_iw_cm_id(struct rdma_cm_id *cm_id);
-struct rdma_cm_id *rdma_res_to_id(struct rdma_restrack_entry *res);
 
 #endif /* RDMA_CM_H */

From 1c7eec4d5f3b39cdea2153abaebf1b7229a47072 Mon Sep 17 00:00:00 2001
From: Zhu Yanjun <yanjun.zhu@linux.dev>
Date: Sat, 19 Apr 2025 10:07:41 +0200
Subject: [PATCH 18/57] RDMA/rxe: Fix "trying to register non-static key in
 rxe_qp_do_cleanup" bug

Call Trace:
 <TASK>
 __dump_stack lib/dump_stack.c:94 [inline]
 dump_stack_lvl+0x116/0x1f0 lib/dump_stack.c:120
 assign_lock_key kernel/locking/lockdep.c:986 [inline]
 register_lock_class+0x4a3/0x4c0 kernel/locking/lockdep.c:1300
 __lock_acquire+0x99/0x1ba0 kernel/locking/lockdep.c:5110
 lock_acquire kernel/locking/lockdep.c:5866 [inline]
 lock_acquire+0x179/0x350 kernel/locking/lockdep.c:5823
 __timer_delete_sync+0x152/0x1b0 kernel/time/timer.c:1644
 rxe_qp_do_cleanup+0x5c3/0x7e0 drivers/infiniband/sw/rxe/rxe_qp.c:815
 execute_in_process_context+0x3a/0x160 kernel/workqueue.c:4596
 __rxe_cleanup+0x267/0x3c0 drivers/infiniband/sw/rxe/rxe_pool.c:232
 rxe_create_qp+0x3f7/0x5f0 drivers/infiniband/sw/rxe/rxe_verbs.c:604
 create_qp+0x62d/0xa80 drivers/infiniband/core/verbs.c:1250
 ib_create_qp_kernel+0x9f/0x310 drivers/infiniband/core/verbs.c:1361
 ib_create_qp include/rdma/ib_verbs.h:3803 [inline]
 rdma_create_qp+0x10c/0x340 drivers/infiniband/core/cma.c:1144
 rds_ib_setup_qp+0xc86/0x19a0 net/rds/ib_cm.c:600
 rds_ib_cm_initiate_connect+0x1e8/0x3d0 net/rds/ib_cm.c:944
 rds_rdma_cm_event_handler_cmn+0x61f/0x8c0 net/rds/rdma_transport.c:109
 cma_cm_event_handler+0x94/0x300 drivers/infiniband/core/cma.c:2184
 cma_work_handler+0x15b/0x230 drivers/infiniband/core/cma.c:3042
 process_one_work+0x9cc/0x1b70 kernel/workqueue.c:3238
 process_scheduled_works kernel/workqueue.c:3319 [inline]
 worker_thread+0x6c8/0xf10 kernel/workqueue.c:3400
 kthread+0x3c2/0x780 kernel/kthread.c:464
 ret_from_fork+0x45/0x80 arch/x86/kernel/process.c:153
 ret_from_fork_asm+0x1a/0x30 arch/x86/entry/entry_64.S:245
 </TASK>

The root cause is as below:

In the function rxe_create_qp, the function rxe_qp_from_init is called
to create qp, if this function rxe_qp_from_init fails, rxe_cleanup will
be called to handle all the allocated resources, including the timers:
retrans_timer and rnr_nak_timer.

The function rxe_qp_from_init calls the function rxe_qp_init_req to
initialize the timers: retrans_timer and rnr_nak_timer.

But these timers are initialized in the end of rxe_qp_init_req.
If some errors occur before the initialization of these timers, this
problem will occur.

The solution is to check whether these timers are initialized or not.
If these timers are not initialized, ignore these timers.

Fixes: 8700e3e7c485 ("Soft RoCE driver")
Reported-by: syzbot+4edb496c3cad6e953a31@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=4edb496c3cad6e953a31
Signed-off-by: Zhu Yanjun <yanjun.zhu@linux.dev>
Link: https://patch.msgid.link/20250419080741.1515231-1-yanjun.zhu@linux.dev
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/sw/rxe/rxe_qp.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/drivers/infiniband/sw/rxe/rxe_qp.c b/drivers/infiniband/sw/rxe/rxe_qp.c
index 7975fb0e2782..f2af3e0aef35 100644
--- a/drivers/infiniband/sw/rxe/rxe_qp.c
+++ b/drivers/infiniband/sw/rxe/rxe_qp.c
@@ -811,7 +811,12 @@ static void rxe_qp_do_cleanup(struct work_struct *work)
 	spin_unlock_irqrestore(&qp->state_lock, flags);
 	qp->qp_timeout_jiffies = 0;
 
-	if (qp_type(qp) == IB_QPT_RC) {
+	/* In the function timer_setup, .function is initialized. If .function
+	 * is NULL, it indicates the function timer_setup is not called, the
+	 * timer is not initialized. Or else, the timer is initialized.
+	 */
+	if (qp_type(qp) == IB_QPT_RC && qp->retrans_timer.function &&
+		qp->rnr_nak_timer.function) {
 		timer_delete_sync(&qp->retrans_timer);
 		timer_delete_sync(&qp->rnr_nak_timer);
 	}

From d85080df12f33ad42a452a9998b21813b29daf35 Mon Sep 17 00:00:00 2001
From: "Dr. David Alan Gilbert" <linux@treblig.org>
Date: Sat, 19 Apr 2025 14:27:25 +0100
Subject: [PATCH 19/57] RDMA/rxe: Remove unused rxe_run_task

rxe_run_task() has been unused since 2024's
commit 23bc06af547f ("RDMA/rxe: Don't call direct between tasks")

Remove it.

Signed-off-by: Dr. David Alan Gilbert <linux@treblig.org>
Link: https://patch.msgid.link/20250419132725.199785-1-linux@treblig.org
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/sw/rxe/rxe_task.c | 40 ++++++++--------------------
 drivers/infiniband/sw/rxe/rxe_task.h |  2 --
 2 files changed, 11 insertions(+), 31 deletions(-)

diff --git a/drivers/infiniband/sw/rxe/rxe_task.c b/drivers/infiniband/sw/rxe/rxe_task.c
index 80332638d9e3..6f8f353e9583 100644
--- a/drivers/infiniband/sw/rxe/rxe_task.c
+++ b/drivers/infiniband/sw/rxe/rxe_task.c
@@ -85,17 +85,17 @@ static bool is_done(struct rxe_task *task)
 
 /* do_task is a wrapper for the three tasks (requester,
  * completer, responder) and calls them in a loop until
- * they return a non-zero value. It is called either
- * directly by rxe_run_task or indirectly if rxe_sched_task
- * schedules the task. They must call __reserve_if_idle to
- * move the task to busy before calling or scheduling.
- * The task can also be moved to drained or invalid
- * by calls to rxe_cleanup_task or rxe_disable_task.
- * In that case tasks which get here are not executed but
- * just flushed. The tasks are designed to look to see if
- * there is work to do and then do part of it before returning
- * here with a return value of zero until all the work
- * has been consumed then it returns a non-zero value.
+ * they return a non-zero value. It is called indirectly
+ * when rxe_sched_task schedules the task. They must
+ * call __reserve_if_idle to move the task to busy before
+ * calling or scheduling. The task can also be moved to
+ * drained or invalid by calls to rxe_cleanup_task or
+ * rxe_disable_task. In that case tasks which get here
+ * are not executed but just flushed. The tasks are
+ * designed to look to see if there is work to do and
+ * then do part of it before returning here with a return
+ * value of zero until all the work has been consumed then
+ * it returns a non-zero value.
  * The number of times the task can be run is limited by
  * max iterations so one task cannot hold the cpu forever.
  * If the limit is hit and work remains the task is rescheduled.
@@ -234,24 +234,6 @@ void rxe_cleanup_task(struct rxe_task *task)
 	spin_unlock_irqrestore(&task->lock, flags);
 }
 
-/* run the task inline if it is currently idle
- * cannot call do_task holding the lock
- */
-void rxe_run_task(struct rxe_task *task)
-{
-	unsigned long flags;
-	bool run;
-
-	WARN_ON(rxe_read(task->qp) <= 0);
-
-	spin_lock_irqsave(&task->lock, flags);
-	run = __reserve_if_idle(task);
-	spin_unlock_irqrestore(&task->lock, flags);
-
-	if (run)
-		do_task(task);
-}
-
 /* schedule the task to run later as a work queue entry.
  * the queue_work call can be called holding
  * the lock.
diff --git a/drivers/infiniband/sw/rxe/rxe_task.h b/drivers/infiniband/sw/rxe/rxe_task.h
index a63e258b3d66..a8c9a77b6027 100644
--- a/drivers/infiniband/sw/rxe/rxe_task.h
+++ b/drivers/infiniband/sw/rxe/rxe_task.h
@@ -47,8 +47,6 @@ int rxe_init_task(struct rxe_task *task, struct rxe_qp *qp,
 /* cleanup task */
 void rxe_cleanup_task(struct rxe_task *task);
 
-void rxe_run_task(struct rxe_task *task);
-
 void rxe_sched_task(struct rxe_task *task);
 
 /* keep a task from scheduling */

From 23ea3c70ee426b12ce033cf3107bc0dd312ea1d9 Mon Sep 17 00:00:00 2001
From: Daisuke Matsuda <matsuda-daisuke@fujitsu.com>
Date: Mon, 21 Apr 2025 11:51:01 +0900
Subject: [PATCH 20/57] RDMA/rxe: Remove 32-bit architecture support

Major linux distibutions have phased out support for 32-bit machines. Since
rxe is primarily used for development and testing, the benefit of
maintaining 32-bit support is minimal. This change simplifies ATOMIC WRITE
implementations and improves maintainability of the driver.

Signed-off-by: Daisuke Matsuda <matsuda-daisuke@fujitsu.com>
Link: https://patch.msgid.link/20250421025101.3588139-1-matsuda-daisuke@fujitsu.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/sw/rxe/Kconfig     | 2 +-
 drivers/infiniband/sw/rxe/rxe_loc.h   | 6 +-----
 drivers/infiniband/sw/rxe/rxe_mr.c    | 8 --------
 drivers/infiniband/sw/rxe/rxe_odp.c   | 1 -
 drivers/infiniband/sw/rxe/rxe_param.h | 5 +----
 5 files changed, 3 insertions(+), 19 deletions(-)

diff --git a/drivers/infiniband/sw/rxe/Kconfig b/drivers/infiniband/sw/rxe/Kconfig
index c180e7ebcfc5..1ed5b63f8afc 100644
--- a/drivers/infiniband/sw/rxe/Kconfig
+++ b/drivers/infiniband/sw/rxe/Kconfig
@@ -1,7 +1,7 @@
 # SPDX-License-Identifier: GPL-2.0-only
 config RDMA_RXE
 	tristate "Software RDMA over Ethernet (RoCE) driver"
-	depends on INET && PCI && INFINIBAND
+	depends on INET && PCI && INFINIBAND && 64BIT
 	depends on INFINIBAND_VIRT_DMA
 	select NET_UDP_TUNNEL
 	select CRC32
diff --git a/drivers/infiniband/sw/rxe/rxe_loc.h b/drivers/infiniband/sw/rxe/rxe_loc.h
index 30fa83c9c846..f7dbb9cddd12 100644
--- a/drivers/infiniband/sw/rxe/rxe_loc.h
+++ b/drivers/infiniband/sw/rxe/rxe_loc.h
@@ -196,6 +196,7 @@ enum resp_states rxe_odp_atomic_op(struct rxe_mr *mr, u64 iova, int opcode,
 				   u64 compare, u64 swap_add, u64 *orig_val);
 int rxe_odp_flush_pmem_iova(struct rxe_mr *mr, u64 iova,
 			    unsigned int length);
+enum resp_states rxe_odp_do_atomic_write(struct rxe_mr *mr, u64 iova, u64 value);
 #else /* CONFIG_INFINIBAND_ON_DEMAND_PAGING */
 static inline int
 rxe_odp_mr_init_user(struct rxe_dev *rxe, u64 start, u64 length, u64 iova,
@@ -219,11 +220,6 @@ static inline int rxe_odp_flush_pmem_iova(struct rxe_mr *mr, u64 iova,
 {
 	return -EOPNOTSUPP;
 }
-#endif /* CONFIG_INFINIBAND_ON_DEMAND_PAGING */
-
-#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
-enum resp_states rxe_odp_do_atomic_write(struct rxe_mr *mr, u64 iova, u64 value);
-#else
 static inline enum resp_states rxe_odp_do_atomic_write(struct rxe_mr *mr,
 						       u64 iova, u64 value)
 {
diff --git a/drivers/infiniband/sw/rxe/rxe_mr.c b/drivers/infiniband/sw/rxe/rxe_mr.c
index 1a74013a14ab..c4936d63e26a 100644
--- a/drivers/infiniband/sw/rxe/rxe_mr.c
+++ b/drivers/infiniband/sw/rxe/rxe_mr.c
@@ -539,8 +539,6 @@ enum resp_states rxe_mr_do_atomic_op(struct rxe_mr *mr, u64 iova, int opcode,
 	return RESPST_NONE;
 }
 
-#if defined CONFIG_64BIT
-/* only implemented or called for 64 bit architectures */
 enum resp_states rxe_mr_do_atomic_write(struct rxe_mr *mr, u64 iova, u64 value)
 {
 	unsigned int page_offset;
@@ -580,12 +578,6 @@ enum resp_states rxe_mr_do_atomic_write(struct rxe_mr *mr, u64 iova, u64 value)
 
 	return RESPST_NONE;
 }
-#else
-enum resp_states rxe_mr_do_atomic_write(struct rxe_mr *mr, u64 iova, u64 value)
-{
-	return RESPST_ERR_UNSUPPORTED_OPCODE;
-}
-#endif
 
 int advance_dma_data(struct rxe_dma_info *dma, unsigned int length)
 {
diff --git a/drivers/infiniband/sw/rxe/rxe_odp.c b/drivers/infiniband/sw/rxe/rxe_odp.c
index fa5e8f5017cc..6149d9ffe7f7 100644
--- a/drivers/infiniband/sw/rxe/rxe_odp.c
+++ b/drivers/infiniband/sw/rxe/rxe_odp.c
@@ -380,7 +380,6 @@ int rxe_odp_flush_pmem_iova(struct rxe_mr *mr, u64 iova,
 	return 0;
 }
 
-/* CONFIG_64BIT=y */
 enum resp_states rxe_odp_do_atomic_write(struct rxe_mr *mr, u64 iova, u64 value)
 {
 	struct ib_umem_odp *umem_odp = to_ib_umem_odp(mr->umem);
diff --git a/drivers/infiniband/sw/rxe/rxe_param.h b/drivers/infiniband/sw/rxe/rxe_param.h
index 003f681e5dc0..767870568372 100644
--- a/drivers/infiniband/sw/rxe/rxe_param.h
+++ b/drivers/infiniband/sw/rxe/rxe_param.h
@@ -53,12 +53,9 @@ enum rxe_device_param {
 					| IB_DEVICE_MEM_WINDOW
 					| IB_DEVICE_FLUSH_GLOBAL
 					| IB_DEVICE_FLUSH_PERSISTENT
-#ifdef CONFIG_64BIT
 					| IB_DEVICE_MEM_WINDOW_TYPE_2B
 					| IB_DEVICE_ATOMIC_WRITE,
-#else
-					| IB_DEVICE_MEM_WINDOW_TYPE_2B,
-#endif /* CONFIG_64BIT */
+
 	RXE_MAX_SGE			= 32,
 	RXE_MAX_WQE_SIZE		= sizeof(struct rxe_send_wqe) +
 					  sizeof(struct ib_sge) * RXE_MAX_SGE,

From 685f9537a72877693a1ab116d155acc89562c29b Mon Sep 17 00:00:00 2001
From: Daisuke Matsuda <matsuda-daisuke@fujitsu.com>
Date: Fri, 18 Apr 2025 14:13:45 +0900
Subject: [PATCH 21/57] RDMA/core: Move ODP capability definitions to uapi

The bits are used from both kernel space and userland, so they should be
placed in UAPI.

Signed-off-by: Daisuke Matsuda <matsuda-daisuke@fujitsu.com>
Link: https://patch.msgid.link/20250418051345.1022339-2-matsuda-daisuke@fujitsu.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 include/rdma/ib_verbs.h           | 20 ++++++++++----------
 include/uapi/rdma/ib_user_verbs.h | 16 ++++++++++++++++
 2 files changed, 26 insertions(+), 10 deletions(-)

diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index 0a7ccd08b365..b06a0ed81bdd 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -314,19 +314,19 @@ enum ib_atomic_cap {
 };
 
 enum ib_odp_general_cap_bits {
-	IB_ODP_SUPPORT		= 1 << 0,
-	IB_ODP_SUPPORT_IMPLICIT = 1 << 1,
+	IB_ODP_SUPPORT		= IB_UVERBS_ODP_SUPPORT,
+	IB_ODP_SUPPORT_IMPLICIT = IB_UVERBS_ODP_SUPPORT_IMPLICIT,
 };
 
 enum ib_odp_transport_cap_bits {
-	IB_ODP_SUPPORT_SEND	= 1 << 0,
-	IB_ODP_SUPPORT_RECV	= 1 << 1,
-	IB_ODP_SUPPORT_WRITE	= 1 << 2,
-	IB_ODP_SUPPORT_READ	= 1 << 3,
-	IB_ODP_SUPPORT_ATOMIC	= 1 << 4,
-	IB_ODP_SUPPORT_SRQ_RECV	= 1 << 5,
-	IB_ODP_SUPPORT_FLUSH	= 1 << 6,
-	IB_ODP_SUPPORT_ATOMIC_WRITE	= 1 << 7,
+	IB_ODP_SUPPORT_SEND	= IB_UVERBS_ODP_SUPPORT_SEND,
+	IB_ODP_SUPPORT_RECV	= IB_UVERBS_ODP_SUPPORT_RECV,
+	IB_ODP_SUPPORT_WRITE	= IB_UVERBS_ODP_SUPPORT_WRITE,
+	IB_ODP_SUPPORT_READ	= IB_UVERBS_ODP_SUPPORT_READ,
+	IB_ODP_SUPPORT_ATOMIC	= IB_UVERBS_ODP_SUPPORT_ATOMIC,
+	IB_ODP_SUPPORT_SRQ_RECV	= IB_UVERBS_ODP_SUPPORT_SRQ_RECV,
+	IB_ODP_SUPPORT_FLUSH	= IB_UVERBS_ODP_SUPPORT_FLUSH,
+	IB_ODP_SUPPORT_ATOMIC_WRITE	= IB_UVERBS_ODP_SUPPORT_ATOMIC_WRITE,
 };
 
 struct ib_odp_caps {
diff --git a/include/uapi/rdma/ib_user_verbs.h b/include/uapi/rdma/ib_user_verbs.h
index e16650f0c85d..3b7bd99813e9 100644
--- a/include/uapi/rdma/ib_user_verbs.h
+++ b/include/uapi/rdma/ib_user_verbs.h
@@ -233,6 +233,22 @@ struct ib_uverbs_ex_query_device {
 	__u32 reserved;
 };
 
+enum ib_uverbs_odp_general_cap_bits {
+	IB_UVERBS_ODP_SUPPORT          = 1 << 0,
+	IB_UVERBS_ODP_SUPPORT_IMPLICIT = 1 << 1,
+};
+
+enum ib_uverbs_odp_transport_cap_bits {
+	IB_UVERBS_ODP_SUPPORT_SEND     = 1 << 0,
+	IB_UVERBS_ODP_SUPPORT_RECV     = 1 << 1,
+	IB_UVERBS_ODP_SUPPORT_WRITE    = 1 << 2,
+	IB_UVERBS_ODP_SUPPORT_READ     = 1 << 3,
+	IB_UVERBS_ODP_SUPPORT_ATOMIC   = 1 << 4,
+	IB_UVERBS_ODP_SUPPORT_SRQ_RECV = 1 << 5,
+	IB_UVERBS_ODP_SUPPORT_FLUSH    = 1 << 6,
+	IB_UVERBS_ODP_SUPPORT_ATOMIC_WRITE     = 1 << 7,
+};
+
 struct ib_uverbs_odp_caps {
 	__aligned_u64 general_caps;
 	struct {

From 02007e3ddc0790ed5a6a504892d7aa1a917076bc Mon Sep 17 00:00:00 2001
From: Junxian Huang <huangjunxian6@hisilicon.com>
Date: Mon, 21 Apr 2025 21:27:45 +0800
Subject: [PATCH 22/57] RDMA/hns: Add trace for flush CQE

Add trace to print the producer index of QP when triggering flush CQE.

Output example:
$ cat /sys/kernel/debug/tracing/trace
  tracer: nop

  entries-in-buffer/entries-written: 2/2   #P:128

                                 _-----=> irqs-off/BH-disabled
                                / _----=> need-resched
                               | / _---=> hardirq/softirq
                               || / _--=> preempt-depth
                               ||| / _-=> migrate-disable
                               |||| /     delay
            TASK-PID     CPU#  |||||  TIMESTAMP  FUNCTION
               | |         |   |||||     |         |
      ib_send_bw-11474   [075] d..1.  2393.434738: hns_sq_flush_cqe: SQ 0x2 flush head 0xb5c7.
      ib_send_bw-11474   [075] d..1.  2393.434739: hns_rq_flush_cqe: RQ 0x2 flush head 0.

Signed-off-by: Junxian Huang <huangjunxian6@hisilicon.com>
Link: https://patch.msgid.link/20250421132750.1363348-2-huangjunxian6@hisilicon.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/hw/hns/hns_roce_device.h | 17 +++++++
 drivers/infiniband/hw/hns/hns_roce_hw_v2.c  |  5 +++
 drivers/infiniband/hw/hns/hns_roce_trace.h  | 50 +++++++++++++++++++++
 3 files changed, 72 insertions(+)
 create mode 100644 drivers/infiniband/hw/hns/hns_roce_trace.h

diff --git a/drivers/infiniband/hw/hns/hns_roce_device.h b/drivers/infiniband/hw/hns/hns_roce_device.h
index 560a1d9de408..080bd049b0f8 100644
--- a/drivers/infiniband/hw/hns/hns_roce_device.h
+++ b/drivers/infiniband/hw/hns/hns_roce_device.h
@@ -1027,6 +1027,23 @@ struct hns_roce_dev {
 	atomic64_t *dfx_cnt;
 };
 
+enum hns_roce_trace_type {
+	TRACE_SQ,
+	TRACE_RQ,
+};
+
+static inline const char *trace_type_to_str(enum hns_roce_trace_type type)
+{
+	switch (type) {
+	case TRACE_SQ:
+		return "SQ";
+	case TRACE_RQ:
+		return "RQ";
+	default:
+		return "UNKNOWN";
+	}
+}
+
 static inline struct hns_roce_dev *to_hr_dev(struct ib_device *ib_dev)
 {
 	return container_of(ib_dev, struct hns_roce_dev, ib_dev);
diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
index c6399490e3a5..e1c27c9556fd 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
@@ -50,6 +50,9 @@
 #include "hns_roce_hem.h"
 #include "hns_roce_hw_v2.h"
 
+#define CREATE_TRACE_POINTS
+#include "hns_roce_trace.h"
+
 enum {
 	CMD_RST_PRC_OTHERS,
 	CMD_RST_PRC_SUCCESS,
@@ -5312,6 +5315,7 @@ static void v2_set_flushed_fields(struct ib_qp *ibqp,
 		return;
 
 	spin_lock_irqsave(&hr_qp->sq.lock, sq_flag);
+	trace_hns_sq_flush_cqe(hr_qp->qpn, hr_qp->sq.head, TRACE_SQ);
 	hr_reg_write(context, QPC_SQ_PRODUCER_IDX, hr_qp->sq.head);
 	hr_reg_clear(qpc_mask, QPC_SQ_PRODUCER_IDX);
 	hr_qp->state = IB_QPS_ERR;
@@ -5321,6 +5325,7 @@ static void v2_set_flushed_fields(struct ib_qp *ibqp,
 		return;
 
 	spin_lock_irqsave(&hr_qp->rq.lock, rq_flag);
+	trace_hns_rq_flush_cqe(hr_qp->qpn, hr_qp->rq.head, TRACE_RQ);
 	hr_reg_write(context, QPC_RQ_PRODUCER_IDX, hr_qp->rq.head);
 	hr_reg_clear(qpc_mask, QPC_RQ_PRODUCER_IDX);
 	spin_unlock_irqrestore(&hr_qp->rq.lock, rq_flag);
diff --git a/drivers/infiniband/hw/hns/hns_roce_trace.h b/drivers/infiniband/hw/hns/hns_roce_trace.h
new file mode 100644
index 000000000000..3deace357fb7
--- /dev/null
+++ b/drivers/infiniband/hw/hns/hns_roce_trace.h
@@ -0,0 +1,50 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * Copyright (c) 2025 Hisilicon Limited.
+ */
+
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM hns_roce
+
+#if !defined(__HNS_ROCE_TRACE_H) || defined(TRACE_HEADER_MULTI_READ)
+#define __HNS_ROCE_TRACE_H
+
+#include <linux/tracepoint.h>
+#include "hns_roce_device.h"
+
+DECLARE_EVENT_CLASS(flush_head_template,
+		    TP_PROTO(unsigned long qpn, u32 pi,
+			     enum hns_roce_trace_type type),
+		    TP_ARGS(qpn, pi, type),
+
+		    TP_STRUCT__entry(__field(unsigned long, qpn)
+				     __field(u32, pi)
+				     __field(enum hns_roce_trace_type, type)
+		    ),
+
+		    TP_fast_assign(__entry->qpn = qpn;
+				   __entry->pi = pi;
+				   __entry->type = type;
+		    ),
+
+		    TP_printk("%s 0x%lx flush head 0x%x.",
+			      trace_type_to_str(__entry->type),
+			      __entry->qpn, __entry->pi)
+);
+
+DEFINE_EVENT(flush_head_template, hns_sq_flush_cqe,
+	     TP_PROTO(unsigned long qpn, u32 pi,
+		      enum hns_roce_trace_type type),
+	     TP_ARGS(qpn, pi, type));
+DEFINE_EVENT(flush_head_template, hns_rq_flush_cqe,
+	     TP_PROTO(unsigned long qpn, u32 pi,
+		      enum hns_roce_trace_type type),
+	     TP_ARGS(qpn, pi, type));
+
+#endif /* __HNS_ROCE_TRACE_H */
+
+#undef TRACE_INCLUDE_FILE
+#define TRACE_INCLUDE_FILE hns_roce_trace
+#undef TRACE_INCLUDE_PATH
+#define TRACE_INCLUDE_PATH .
+#include <trace/define_trace.h>

From 6c98c86708063af509b82f35bfac9577b61b4a94 Mon Sep 17 00:00:00 2001
From: Junxian Huang <huangjunxian6@hisilicon.com>
Date: Mon, 21 Apr 2025 21:27:46 +0800
Subject: [PATCH 23/57] RDMA/hns: Add trace for WQE dumping

Add trace for WQE dumping, including SQ, RQ and SRQ.

Output example:
$ cat /sys/kernel/debug/tracing/trace
  tracer: nop

  entries-in-buffer/entries-written: 2/2   #P:128

                                 _-----=> irqs-off/BH-disabled
                               / _----=> need-resched
                               | / _---=> hardirq/softirq
			       || / _--=> preempt-depth
                               ||| / _-=> migrate-disable
                               |||| /     delay
            TASK-PID     CPU#  |||||  TIMESTAMP  FUNCTION
               | |         |   |||||     |         |
  roce_test_main-22730   [074] d..1. 16133.898282: hns_sq_wqe: SQ 0xc wqe
(0x0/0xffff0820a6076060): {0x180,0x639c,0x0,0x1000000,0x0,0x0,0x0,0x0,
0x639c,0x300,0xf7e38000,0x0,0x0,0x0,0x0,0x0}

Signed-off-by: Junxian Huang <huangjunxian6@hisilicon.com>
Link: https://patch.msgid.link/20250421132750.1363348-3-huangjunxian6@hisilicon.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/hw/hns/hns_roce_device.h |  3 ++
 drivers/infiniband/hw/hns/hns_roce_hw_v2.c  |  8 ++++
 drivers/infiniband/hw/hns/hns_roce_trace.h  | 44 +++++++++++++++++++++
 3 files changed, 55 insertions(+)

diff --git a/drivers/infiniband/hw/hns/hns_roce_device.h b/drivers/infiniband/hw/hns/hns_roce_device.h
index 080bd049b0f8..1dcc9cbb4678 100644
--- a/drivers/infiniband/hw/hns/hns_roce_device.h
+++ b/drivers/infiniband/hw/hns/hns_roce_device.h
@@ -1030,6 +1030,7 @@ struct hns_roce_dev {
 enum hns_roce_trace_type {
 	TRACE_SQ,
 	TRACE_RQ,
+	TRACE_SRQ,
 };
 
 static inline const char *trace_type_to_str(enum hns_roce_trace_type type)
@@ -1039,6 +1040,8 @@ static inline const char *trace_type_to_str(enum hns_roce_trace_type type)
 		return "SQ";
 	case TRACE_RQ:
 		return "RQ";
+	case TRACE_SRQ:
+		return "SRQ";
 	default:
 		return "UNKNOWN";
 	}
diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
index e1c27c9556fd..ecc1e0b1df4e 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
@@ -741,6 +741,8 @@ static int hns_roce_v2_post_send(struct ib_qp *ibqp,
 		else
 			ret = set_ud_wqe(qp, wr, wqe, &sge_idx, owner_bit);
 
+		trace_hns_sq_wqe(qp->qpn, wqe_idx, wqe, 1 << qp->sq.wqe_shift,
+				 wr->wr_id, TRACE_SQ);
 		if (unlikely(ret)) {
 			*bad_wr = wr;
 			goto out;
@@ -810,6 +812,9 @@ static void fill_rq_wqe(struct hns_roce_qp *hr_qp, const struct ib_recv_wr *wr,
 
 	wqe = hns_roce_get_recv_wqe(hr_qp, wqe_idx);
 	fill_recv_sge_to_wqe(wr, wqe, max_sge, hr_qp->rq.rsv_sge);
+
+	trace_hns_rq_wqe(hr_qp->qpn, wqe_idx, wqe, 1 << hr_qp->rq.wqe_shift,
+			 wr->wr_id, TRACE_RQ);
 }
 
 static int hns_roce_v2_post_recv(struct ib_qp *ibqp,
@@ -987,6 +992,9 @@ static int hns_roce_v2_post_srq_recv(struct ib_srq *ibsrq,
 		fill_recv_sge_to_wqe(wr, wqe, max_sge, srq->rsv_sge);
 		fill_wqe_idx(srq, wqe_idx);
 		srq->wrid[wqe_idx] = wr->wr_id;
+
+		trace_hns_srq_wqe(srq->srqn, wqe_idx, wqe, 1 << srq->wqe_shift,
+				  wr->wr_id, TRACE_SRQ);
 	}
 
 	if (likely(nreq)) {
diff --git a/drivers/infiniband/hw/hns/hns_roce_trace.h b/drivers/infiniband/hw/hns/hns_roce_trace.h
index 3deace357fb7..56e152387db4 100644
--- a/drivers/infiniband/hw/hns/hns_roce_trace.h
+++ b/drivers/infiniband/hw/hns/hns_roce_trace.h
@@ -41,6 +41,50 @@ DEFINE_EVENT(flush_head_template, hns_rq_flush_cqe,
 		      enum hns_roce_trace_type type),
 	     TP_ARGS(qpn, pi, type));
 
+#define MAX_SGE_PER_WQE 64
+#define MAX_WQE_SIZE (MAX_SGE_PER_WQE * HNS_ROCE_SGE_SIZE)
+DECLARE_EVENT_CLASS(wqe_template,
+		    TP_PROTO(unsigned long qpn, u32 idx, void *wqe, u32 len,
+			     u64 id, enum hns_roce_trace_type type),
+		    TP_ARGS(qpn, idx, wqe, len, id, type),
+
+		    TP_STRUCT__entry(__field(unsigned long, qpn)
+				     __field(u32, idx)
+				     __array(__le32, wqe,
+					     MAX_WQE_SIZE / sizeof(__le32))
+				     __field(u32, len)
+				     __field(u64, id)
+				     __field(enum hns_roce_trace_type, type)
+				     ),
+
+		    TP_fast_assign(__entry->qpn = qpn;
+				   __entry->idx = idx;
+				   __entry->id = id;
+				   memcpy(__entry->wqe, wqe, len);
+				   __entry->len = len / sizeof(__le32);
+				   __entry->type = type;
+				   ),
+
+		    TP_printk("%s 0x%lx wqe(0x%x/0x%llx): %s",
+			      trace_type_to_str(__entry->type),
+			      __entry->qpn, __entry->idx, __entry->id,
+			      __print_array(__entry->wqe, __entry->len,
+					    sizeof(__le32)))
+);
+
+DEFINE_EVENT(wqe_template, hns_sq_wqe,
+	     TP_PROTO(unsigned long qpn, u32 idx, void *wqe, u32 len, u64 id,
+		      enum hns_roce_trace_type type),
+	     TP_ARGS(qpn, idx, wqe, len, id, type));
+DEFINE_EVENT(wqe_template, hns_rq_wqe,
+	     TP_PROTO(unsigned long qpn, u32 idx, void *wqe, u32 len, u64 id,
+		      enum hns_roce_trace_type type),
+	     TP_ARGS(qpn, idx, wqe, len, id, type));
+DEFINE_EVENT(wqe_template, hns_srq_wqe,
+	     TP_PROTO(unsigned long qpn, u32 idx, void *wqe, u32 len, u64 id,
+		      enum hns_roce_trace_type type),
+	     TP_ARGS(qpn, idx, wqe, len, id, type));
+
 #endif /* __HNS_ROCE_TRACE_H */
 
 #undef TRACE_INCLUDE_FILE

From 1e63e2f96613bc6471e7497e47d1767e3846608e Mon Sep 17 00:00:00 2001
From: Junxian Huang <huangjunxian6@hisilicon.com>
Date: Mon, 21 Apr 2025 21:27:47 +0800
Subject: [PATCH 24/57] RDMA/hns: Add trace for AEQE dumping

Add trace for AEQE dumping.

Output example:
$ cat /sys/kernel/debug/tracing/trace
  tracer: nop

  entries-in-buffer/entries-written: 2/2   #P:128

                                 _-----=> irqs-off/BH-disabled
                               / _----=> need-resched
                               | / _---=> hardirq/softirq
                               || / _--=> preempt-depth
                               ||| / _-=> migrate-disable
                               |||| /     delay
            TASK-PID     CPU#  |||||  TIMESTAMP  FUNCTION
               | |         |   |||||     |         |
	  <idle>-0       [120] d.h1.  7995.835587: hns_ae_info: event 19 aeqe:
{0x80006013,0x0,0x0,0x10d2c,0x0,0x0,0x0,0x0,0x0,0x0,0x0,0x0,0x0,0x0,0x0,0x0}

Signed-off-by: Junxian Huang <huangjunxian6@hisilicon.com>
Link: https://patch.msgid.link/20250421132750.1363348-4-huangjunxian6@hisilicon.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/hw/hns/hns_roce_hw_v2.c |  1 +
 drivers/infiniband/hw/hns/hns_roce_trace.h | 19 +++++++++++++++++++
 2 files changed, 20 insertions(+)

diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
index ecc1e0b1df4e..5d3bf84c1aae 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
@@ -6260,6 +6260,7 @@ static irqreturn_t hns_roce_v2_aeq_int(struct hns_roce_dev *hr_dev,
 		eq->sub_type = sub_type;
 		++eq->cons_index;
 		aeqe_found = IRQ_HANDLED;
+		trace_hns_ae_info(event_type, aeqe, eq->eqe_size);
 
 		atomic64_inc(&hr_dev->dfx_cnt[HNS_ROCE_DFX_AEQE_CNT]);
 
diff --git a/drivers/infiniband/hw/hns/hns_roce_trace.h b/drivers/infiniband/hw/hns/hns_roce_trace.h
index 56e152387db4..75c89aff68c3 100644
--- a/drivers/infiniband/hw/hns/hns_roce_trace.h
+++ b/drivers/infiniband/hw/hns/hns_roce_trace.h
@@ -85,6 +85,25 @@ DEFINE_EVENT(wqe_template, hns_srq_wqe,
 		      enum hns_roce_trace_type type),
 	     TP_ARGS(qpn, idx, wqe, len, id, type));
 
+TRACE_EVENT(hns_ae_info,
+	    TP_PROTO(int event_type, void *aeqe, unsigned int len),
+	    TP_ARGS(event_type, aeqe, len),
+
+	    TP_STRUCT__entry(__field(int, event_type)
+			     __array(__le32, aeqe,
+				     HNS_ROCE_V3_EQE_SIZE / sizeof(__le32))
+			     __field(u32, len)
+	    ),
+
+	    TP_fast_assign(__entry->event_type = event_type;
+			   __entry->len = len / sizeof(__le32);
+			   memcpy(__entry->aeqe, aeqe, len);
+	    ),
+
+	    TP_printk("event %2d aeqe: %s", __entry->event_type,
+		      __print_array(__entry->aeqe, __entry->len, sizeof(__le32)))
+);
+
 #endif /* __HNS_ROCE_TRACE_H */
 
 #undef TRACE_INCLUDE_FILE

From 48ffc152576d633d05eda4251aaef5dd53970cb8 Mon Sep 17 00:00:00 2001
From: Junxian Huang <huangjunxian6@hisilicon.com>
Date: Mon, 21 Apr 2025 21:27:48 +0800
Subject: [PATCH 25/57] RDMA/hns: Add trace for MR/MTR attribute dumping

Add trace for MR/MTR attribute dumping.

Output example:
$ cat /sys/kernel/debug/tracing/trace
  tracer: nop

  entries-in-buffer/entries-written: 2/2   #P:128

                                 _-----=> irqs-off/BH-disabled
                               / _----=> need-resched
                               | / _---=> hardirq/softirq
                               || / _--=> preempt-depth
                               ||| / _-=> migrate-disable
                               |||| /     delay
            TASK-PID     CPU#  |||||  TIMESTAMP  FUNCTION
               | |         |   |||||     |         |
      ib_send_bw-14751   [111] .....  8763.823038: hns_buf_attr: rg cnt:1,
pg_sft:0xc, mtt_only:no, rg 0 (sz:131072, hop:2), rg 1 (sz:0, hop:0),
rg 2 (sz:0, hop:0)

      ib_send_bw-14751   [111] .....  8763.823118: hns_mr:
iova:0xffffb2968000, size:131072, key:512, pd:1, pbl_hop:1, npages:4,
type:0, status:0

Signed-off-by: Junxian Huang <huangjunxian6@hisilicon.com>
Link: https://patch.msgid.link/20250421132750.1363348-5-huangjunxian6@hisilicon.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/hw/hns/hns_roce_mr.c    |  3 +
 drivers/infiniband/hw/hns/hns_roce_trace.h | 65 ++++++++++++++++++++++
 2 files changed, 68 insertions(+)

diff --git a/drivers/infiniband/hw/hns/hns_roce_mr.c b/drivers/infiniband/hw/hns/hns_roce_mr.c
index 09da3496843b..93a48b41955b 100644
--- a/drivers/infiniband/hw/hns/hns_roce_mr.c
+++ b/drivers/infiniband/hw/hns/hns_roce_mr.c
@@ -38,6 +38,7 @@
 #include "hns_roce_device.h"
 #include "hns_roce_cmd.h"
 #include "hns_roce_hem.h"
+#include "hns_roce_trace.h"
 
 static u32 hw_index_to_key(int ind)
 {
@@ -159,6 +160,7 @@ static int hns_roce_mr_enable(struct hns_roce_dev *hr_dev,
 	if (IS_ERR(mailbox))
 		return PTR_ERR(mailbox);
 
+	trace_hns_mr(mr);
 	if (mr->type != MR_TYPE_FRMR)
 		ret = hr_dev->hw->write_mtpt(hr_dev, mailbox->buf, mr);
 	else
@@ -1146,6 +1148,7 @@ int hns_roce_mtr_create(struct hns_roce_dev *hr_dev, struct hns_roce_mtr *mtr,
 	struct ib_device *ibdev = &hr_dev->ib_dev;
 	int ret;
 
+	trace_hns_buf_attr(buf_attr);
 	/* The caller has its own buffer list and invokes the hns_roce_mtr_map()
 	 * to finish the MTT configuration.
 	 */
diff --git a/drivers/infiniband/hw/hns/hns_roce_trace.h b/drivers/infiniband/hw/hns/hns_roce_trace.h
index 75c89aff68c3..b0bc22b9cc28 100644
--- a/drivers/infiniband/hw/hns/hns_roce_trace.h
+++ b/drivers/infiniband/hw/hns/hns_roce_trace.h
@@ -10,6 +10,7 @@
 #define __HNS_ROCE_TRACE_H
 
 #include <linux/tracepoint.h>
+#include <linux/string_choices.h>
 #include "hns_roce_device.h"
 
 DECLARE_EVENT_CLASS(flush_head_template,
@@ -104,6 +105,70 @@ TRACE_EVENT(hns_ae_info,
 		      __print_array(__entry->aeqe, __entry->len, sizeof(__le32)))
 );
 
+TRACE_EVENT(hns_mr,
+	    TP_PROTO(struct hns_roce_mr *mr),
+	    TP_ARGS(mr),
+
+	    TP_STRUCT__entry(__field(u64, iova)
+			     __field(u64, size)
+			     __field(u32, key)
+			     __field(u32, pd)
+			     __field(u32, pbl_hop_num)
+			     __field(u32, npages)
+			     __field(int, type)
+			     __field(int, enabled)
+	    ),
+
+	    TP_fast_assign(__entry->iova = mr->iova;
+			   __entry->size = mr->size;
+			   __entry->key = mr->key;
+			   __entry->pd = mr->pd;
+			   __entry->pbl_hop_num = mr->pbl_hop_num;
+			   __entry->npages = mr->npages;
+			   __entry->type = mr->type;
+			   __entry->enabled = mr->enabled;
+	    ),
+
+	    TP_printk("iova:0x%llx, size:%llu, key:%u, pd:%u, pbl_hop:%u, npages:%u, type:%d, status:%d",
+		      __entry->iova, __entry->size, __entry->key,
+		      __entry->pd, __entry->pbl_hop_num, __entry->npages,
+		      __entry->type, __entry->enabled)
+);
+
+TRACE_EVENT(hns_buf_attr,
+	    TP_PROTO(struct hns_roce_buf_attr *attr),
+	    TP_ARGS(attr),
+
+	    TP_STRUCT__entry(__field(unsigned int, region_count)
+			     __field(unsigned int, region0_size)
+			     __field(int, region0_hopnum)
+			     __field(unsigned int, region1_size)
+			     __field(int, region1_hopnum)
+			     __field(unsigned int, region2_size)
+			     __field(int, region2_hopnum)
+			     __field(unsigned int, page_shift)
+			     __field(bool, mtt_only)
+	    ),
+
+	    TP_fast_assign(__entry->region_count = attr->region_count;
+			   __entry->region0_size = attr->region[0].size;
+			   __entry->region0_hopnum = attr->region[0].hopnum;
+			   __entry->region1_size = attr->region[1].size;
+			   __entry->region1_hopnum = attr->region[1].hopnum;
+			   __entry->region2_size = attr->region[2].size;
+			   __entry->region2_hopnum = attr->region[2].hopnum;
+			   __entry->page_shift = attr->page_shift;
+			   __entry->mtt_only = attr->mtt_only;
+	    ),
+
+	    TP_printk("rg cnt:%u, pg_sft:0x%x, mtt_only:%s, rg 0 (sz:%u, hop:%u), rg 1 (sz:%u, hop:%u), rg 2 (sz:%u, hop:%u)\n",
+		      __entry->region_count, __entry->page_shift,
+		      str_yes_no(__entry->mtt_only),
+		      __entry->region0_size, __entry->region0_hopnum,
+		      __entry->region1_size, __entry->region1_hopnum,
+		      __entry->region2_size, __entry->region2_hopnum)
+);
+
 #endif /* __HNS_ROCE_TRACE_H */
 
 #undef TRACE_INCLUDE_FILE

From 2b11d33de23262cb20d1dcb24b586dbb8f54d463 Mon Sep 17 00:00:00 2001
From: Junxian Huang <huangjunxian6@hisilicon.com>
Date: Mon, 21 Apr 2025 21:27:49 +0800
Subject: [PATCH 26/57] RDMA/hns: Include hnae3.h in hns_roce_hw_v2.h

hns_roce_hw_v2.h has a direct dependency on hnae3.h due to the
inline function hns_roce_write64(), but it doesn't include this
header currently. This leads to that files including
hns_roce_hw_v2.h must also include hnae3.h to avoid compilation
errors, even if they themselves don't really rely on hnae3.h.
This doesn't make sense, hns_roce_hw_v2.h should include hnae3.h
directly.

Fixes: d3743fa94ccd ("RDMA/hns: Fix the chip hanging caused by sending doorbell during reset")
Signed-off-by: Junxian Huang <huangjunxian6@hisilicon.com>
Link: https://patch.msgid.link/20250421132750.1363348-6-huangjunxian6@hisilicon.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/hw/hns/hns_roce_ah.c       | 1 -
 drivers/infiniband/hw/hns/hns_roce_hw_v2.c    | 1 -
 drivers/infiniband/hw/hns/hns_roce_hw_v2.h    | 1 +
 drivers/infiniband/hw/hns/hns_roce_main.c     | 1 -
 drivers/infiniband/hw/hns/hns_roce_restrack.c | 1 -
 5 files changed, 1 insertion(+), 4 deletions(-)

diff --git a/drivers/infiniband/hw/hns/hns_roce_ah.c b/drivers/infiniband/hw/hns/hns_roce_ah.c
index 4fc5b9d5fea8..307c35888b30 100644
--- a/drivers/infiniband/hw/hns/hns_roce_ah.c
+++ b/drivers/infiniband/hw/hns/hns_roce_ah.c
@@ -33,7 +33,6 @@
 #include <linux/pci.h>
 #include <rdma/ib_addr.h>
 #include <rdma/ib_cache.h>
-#include "hnae3.h"
 #include "hns_roce_device.h"
 #include "hns_roce_hw_v2.h"
 
diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
index 5d3bf84c1aae..9cb9fe6a0a98 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
@@ -43,7 +43,6 @@
 #include <rdma/ib_umem.h>
 #include <rdma/uverbs_ioctl.h>
 
-#include "hnae3.h"
 #include "hns_roce_common.h"
 #include "hns_roce_device.h"
 #include "hns_roce_cmd.h"
diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.h b/drivers/infiniband/hw/hns/hns_roce_hw_v2.h
index 91a5665465ff..bc7466830eaf 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.h
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.h
@@ -34,6 +34,7 @@
 #define _HNS_ROCE_HW_V2_H
 
 #include <linux/bitops.h>
+#include "hnae3.h"
 
 #define HNS_ROCE_V2_MAX_RC_INL_INN_SZ		32
 #define HNS_ROCE_V2_MTT_ENTRY_SZ		64
diff --git a/drivers/infiniband/hw/hns/hns_roce_main.c b/drivers/infiniband/hw/hns/hns_roce_main.c
index cf89a8db4f64..02f7acb30c5d 100644
--- a/drivers/infiniband/hw/hns/hns_roce_main.c
+++ b/drivers/infiniband/hw/hns/hns_roce_main.c
@@ -37,7 +37,6 @@
 #include <rdma/ib_smi.h>
 #include <rdma/ib_user_verbs.h>
 #include <rdma/ib_cache.h>
-#include "hnae3.h"
 #include "hns_roce_common.h"
 #include "hns_roce_device.h"
 #include "hns_roce_hem.h"
diff --git a/drivers/infiniband/hw/hns/hns_roce_restrack.c b/drivers/infiniband/hw/hns/hns_roce_restrack.c
index 356d98816949..f637b73b946e 100644
--- a/drivers/infiniband/hw/hns/hns_roce_restrack.c
+++ b/drivers/infiniband/hw/hns/hns_roce_restrack.c
@@ -4,7 +4,6 @@
 #include <rdma/rdma_cm.h>
 #include <rdma/restrack.h>
 #include <uapi/rdma/rdma_netlink.h>
-#include "hnae3.h"
 #include "hns_roce_common.h"
 #include "hns_roce_device.h"
 #include "hns_roce_hw_v2.h"

From 6bd18dabf1c94653b9a96402544bb980a7ca7ce1 Mon Sep 17 00:00:00 2001
From: Junxian Huang <huangjunxian6@hisilicon.com>
Date: Mon, 21 Apr 2025 21:27:50 +0800
Subject: [PATCH 27/57] RDMA/hns: Add trace for CMDQ dumping

Add trace for CMDQ dumping.

Output example:
$ cat /sys/kernel/debug/tracing/trace
  tracer: nop

  entries-in-buffer/entries-written: 2/2   #P:128

                                 _-----=> irqs-off/BH-disabled
                               / _----=> need-resched
                               | / _---=> hardirq/softirq
                               || / _--=> preempt-depth
                               ||| / _-=> migrate-disable
                               |||| /     delay
            TASK-PID     CPU#  |||||  TIMESTAMP  FUNCTION
               | |         |   |||||     |         |
  kworker/u512:1-14003   [089] b..1. 50737.238304: hns_cmdq_req: 0000:bd:00.0 cmdq opcode:0x8500, flag:0x1, retval:0x0, data:{0x2,0x0,0x0,0xffff0000,0x32323232,0x0}

  kworker/u512:1-14003   [089] b..1. 50737.238316: hns_cmdq_resp: 0000:bd:00.0 cmdq opcode:0x8500, flag:0x2, retval:0x0, data:{0x2,0x0,0x0,0xffff0000,0x32323232,0x0}

Signed-off-by: Junxian Huang <huangjunxian6@hisilicon.com>
Link: https://patch.msgid.link/20250421132750.1363348-7-huangjunxian6@hisilicon.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/hw/hns/hns_roce_hw_v2.c |  4 +++
 drivers/infiniband/hw/hns/hns_roce_trace.h | 35 ++++++++++++++++++++++
 2 files changed, 39 insertions(+)

diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
index 9cb9fe6a0a98..fa8747656f25 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
@@ -1321,6 +1321,8 @@ static int __hns_roce_cmq_send_one(struct hns_roce_dev *hr_dev,
 	tail = csq->head;
 
 	for (i = 0; i < num; i++) {
+		trace_hns_cmdq_req(hr_dev, &desc[i]);
+
 		csq->desc[csq->head++] = desc[i];
 		if (csq->head == csq->desc_num)
 			csq->head = 0;
@@ -1335,6 +1337,8 @@ static int __hns_roce_cmq_send_one(struct hns_roce_dev *hr_dev,
 	if (hns_roce_cmq_csq_done(hr_dev)) {
 		ret = 0;
 		for (i = 0; i < num; i++) {
+			trace_hns_cmdq_resp(hr_dev, &csq->desc[tail]);
+
 			/* check the result of hardware write back */
 			desc_ret = le16_to_cpu(csq->desc[tail++].retval);
 			if (tail == csq->desc_num)
diff --git a/drivers/infiniband/hw/hns/hns_roce_trace.h b/drivers/infiniband/hw/hns/hns_roce_trace.h
index b0bc22b9cc28..23cbdbaeffaa 100644
--- a/drivers/infiniband/hw/hns/hns_roce_trace.h
+++ b/drivers/infiniband/hw/hns/hns_roce_trace.h
@@ -12,6 +12,7 @@
 #include <linux/tracepoint.h>
 #include <linux/string_choices.h>
 #include "hns_roce_device.h"
+#include "hns_roce_hw_v2.h"
 
 DECLARE_EVENT_CLASS(flush_head_template,
 		    TP_PROTO(unsigned long qpn, u32 pi,
@@ -169,6 +170,40 @@ TRACE_EVENT(hns_buf_attr,
 		      __entry->region2_size, __entry->region2_hopnum)
 );
 
+DECLARE_EVENT_CLASS(cmdq,
+		    TP_PROTO(struct hns_roce_dev *hr_dev,
+			     struct hns_roce_cmq_desc *desc),
+		    TP_ARGS(hr_dev, desc),
+
+		    TP_STRUCT__entry(__string(dev_name, dev_name(hr_dev->dev))
+				     __field(u16, opcode)
+				     __field(u16, flag)
+				     __field(u16, retval)
+				     __array(__le32, data, 6)
+		    ),
+
+		    TP_fast_assign(__assign_str(dev_name);
+				   __entry->opcode = le16_to_cpu(desc->opcode);
+				   __entry->flag = le16_to_cpu(desc->flag);
+				   __entry->retval = le16_to_cpu(desc->retval);
+				   memcpy(__entry->data, desc->data, 6 * sizeof(__le32));
+		    ),
+
+		    TP_printk("%s cmdq opcode:0x%x, flag:0x%x, retval:0x%x, data:%s\n",
+			      __get_str(dev_name), __entry->opcode,
+			      __entry->flag, __entry->retval,
+			      __print_array(__entry->data, 6, sizeof(__le32)))
+);
+
+DEFINE_EVENT(cmdq, hns_cmdq_req,
+	     TP_PROTO(struct hns_roce_dev *hr_dev,
+		      struct hns_roce_cmq_desc *desc),
+	     TP_ARGS(hr_dev, desc));
+DEFINE_EVENT(cmdq, hns_cmdq_resp,
+	     TP_PROTO(struct hns_roce_dev *hr_dev,
+		      struct hns_roce_cmq_desc *desc),
+	     TP_ARGS(hr_dev, desc));
+
 #endif /* __HNS_ROCE_TRACE_H */
 
 #undef TRACE_INCLUDE_FILE

From 7c04bc97171780e15798433d9e83a241d9956014 Mon Sep 17 00:00:00 2001
From: Kees Cook <kees@kernel.org>
Date: Fri, 25 Apr 2025 23:12:09 -0700
Subject: [PATCH 28/57] IB/mthca: Adjust buddy->bits allocation type

In preparation for making the kmalloc family of allocators type aware,
we need to make sure that the returned type from the allocation matches
the type of the variable being assigned. (Before, the allocator would
always return "void *", which can be implicitly cast to any pointer type.)

The assigned type is "unsigned long **", but the returned type will be
"long **". These are the same allocation size (pointer size), but the
types do not match. Adjust the allocation type to match the assignment.

Signed-off-by: Kees Cook <kees@kernel.org>
Link: https://patch.msgid.link/20250426061208.work.000-kees@kernel.org
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/hw/mthca/mthca_mr.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/infiniband/hw/mthca/mthca_mr.c b/drivers/infiniband/hw/mthca/mthca_mr.c
index 192f83fd7c8a..dacb8ceeebe0 100644
--- a/drivers/infiniband/hw/mthca/mthca_mr.c
+++ b/drivers/infiniband/hw/mthca/mthca_mr.c
@@ -144,7 +144,7 @@ static int mthca_buddy_init(struct mthca_buddy *buddy, int max_order)
 	buddy->max_order = max_order;
 	spin_lock_init(&buddy->lock);
 
-	buddy->bits = kcalloc(buddy->max_order + 1, sizeof(long *),
+	buddy->bits = kcalloc(buddy->max_order + 1, sizeof(*buddy->bits),
 			      GFP_KERNEL);
 	buddy->num_free = kcalloc((buddy->max_order + 1), sizeof *buddy->num_free,
 				  GFP_KERNEL);

From 3db60cf9b7da4ae35646c00d9241c44d2dd6caf8 Mon Sep 17 00:00:00 2001
From: Kees Cook <kees@kernel.org>
Date: Fri, 25 Apr 2025 23:12:48 -0700
Subject: [PATCH 29/57] IB/hfi1: Adjust fd->entry_to_rb allocation type

In preparation for making the kmalloc family of allocators type aware,
we need to make sure that the returned type from the allocation matches
the type of the variable being assigned. (Before, the allocator would
always return "void *", which can be implicitly cast to any pointer type.)

The assigned type is "struct tid_rb_node **", but the return type will be
"struct rb_node **". These are the same allocation size (pointer size),
but the types do not match. Adjust the allocation type to match the
assignment.

Signed-off-by: Kees Cook <kees@kernel.org>
Link: https://patch.msgid.link/20250426061247.work.261-kees@kernel.org
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/hw/hfi1/user_exp_rcv.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/infiniband/hw/hfi1/user_exp_rcv.c b/drivers/infiniband/hw/hfi1/user_exp_rcv.c
index cf2d29098406..62b4f16dab27 100644
--- a/drivers/infiniband/hw/hfi1/user_exp_rcv.c
+++ b/drivers/infiniband/hw/hfi1/user_exp_rcv.c
@@ -53,7 +53,7 @@ int hfi1_user_exp_rcv_init(struct hfi1_filedata *fd,
 	int ret = 0;
 
 	fd->entry_to_rb = kcalloc(uctxt->expected_count,
-				  sizeof(struct rb_node *),
+				  sizeof(*fd->entry_to_rb),
 				  GFP_KERNEL);
 	if (!fd->entry_to_rb)
 		return -ENOMEM;

From 7590649ee7af381a9d1153143026dec124c5798e Mon Sep 17 00:00:00 2001
From: Vlad Dumitrescu <vdumitrescu@nvidia.com>
Date: Mon, 28 Apr 2025 14:30:18 +0300
Subject: [PATCH 30/57] IB/cm: Drop lockdep assert and WARN when freeing old
 msg

The send completion handler can run after cm_id has advanced to another
message.  The cm_id lock is not needed in this case, but a recent change
re-used cm_free_priv_msg(), which asserts that the lock is held and
WARNs if the cm_id's currently outstanding msg is different than the one
being freed.

Fixes: 1e5159219076 ("IB/cm: Do not hold reference on cm_id unless needed")
Signed-off-by: Vlad Dumitrescu <vdumitrescu@nvidia.com>
Reviewed-by: Sean Hefty <shefty@nvidia.com>
Link: https://patch.msgid.link/0c364c29142f72b7875fdeba51f3c9bd6ca863ee.1745839788.git.leon@kernel.org
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/core/cm.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/infiniband/core/cm.c b/drivers/infiniband/core/cm.c
index effa53dd6800..e64cbd034a2a 100644
--- a/drivers/infiniband/core/cm.c
+++ b/drivers/infiniband/core/cm.c
@@ -3786,7 +3786,8 @@ static void cm_process_send_error(struct cm_id_private *cm_id_priv,
 	spin_lock_irq(&cm_id_priv->lock);
 	if (msg != cm_id_priv->msg) {
 		spin_unlock_irq(&cm_id_priv->lock);
-		cm_free_priv_msg(msg);
+		cm_free_msg(msg);
+		cm_deref_id(cm_id_priv);
 		return;
 	}
 	cm_free_priv_msg(msg);

From 5d2ea5aebbb2f3ebde4403f9c55b2b057e5dd2d6 Mon Sep 17 00:00:00 2001
From: Patrisious Haddad <phaddad@nvidia.com>
Date: Mon, 28 Apr 2025 14:34:07 +0300
Subject: [PATCH 31/57] RDMA/mlx5: Fix error flow upon firmware failure for RQ
 destruction

Upon RQ destruction if the firmware command fails which is the
last resource to be destroyed some SW resources were already cleaned
regardless of the failure.

Now properly rollback the object to its original state upon such failure.

In order to avoid a use-after free in case someone tries to destroy the
object again, which results in the following kernel trace:
refcount_t: underflow; use-after-free.
WARNING: CPU: 0 PID: 37589 at lib/refcount.c:28 refcount_warn_saturate+0xf4/0x148
Modules linked in: rdma_ucm(OE) rdma_cm(OE) iw_cm(OE) ib_ipoib(OE) ib_cm(OE) ib_umad(OE) mlx5_ib(OE) rfkill mlx5_core(OE) mlxdevm(OE) ib_uverbs(OE) ib_core(OE) psample mlxfw(OE) mlx_compat(OE) macsec tls pci_hyperv_intf sunrpc vfat fat virtio_net net_failover failover fuse loop nfnetlink vsock_loopback vmw_vsock_virtio_transport_common vmw_vsock_vmci_transport vmw_vmci vsock xfs crct10dif_ce ghash_ce sha2_ce sha256_arm64 sha1_ce virtio_console virtio_gpu virtio_blk virtio_dma_buf virtio_mmio dm_mirror dm_region_hash dm_log dm_mod xpmem(OE)
CPU: 0 UID: 0 PID: 37589 Comm: python3 Kdump: loaded Tainted: G           OE     -------  ---  6.12.0-54.el10.aarch64 #1
Tainted: [O]=OOT_MODULE, [E]=UNSIGNED_MODULE
Hardware name: QEMU KVM Virtual Machine, BIOS 0.0.0 02/06/2015
pstate: 60400005 (nZCv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--)
pc : refcount_warn_saturate+0xf4/0x148
lr : refcount_warn_saturate+0xf4/0x148
sp : ffff80008b81b7e0
x29: ffff80008b81b7e0 x28: ffff000133d51600 x27: 0000000000000001
x26: 0000000000000000 x25: 00000000ffffffea x24: ffff00010ae80f00
x23: ffff00010ae80f80 x22: ffff0000c66e5d08 x21: 0000000000000000
x20: ffff0000c66e0000 x19: ffff00010ae80340 x18: 0000000000000006
x17: 0000000000000000 x16: 0000000000000020 x15: ffff80008b81b37f
x14: 0000000000000000 x13: 2e656572662d7265 x12: ffff80008283ef78
x11: ffff80008257efd0 x10: ffff80008283efd0 x9 : ffff80008021ed90
x8 : 0000000000000001 x7 : 00000000000bffe8 x6 : c0000000ffff7fff
x5 : ffff0001fb8e3408 x4 : 0000000000000000 x3 : ffff800179993000
x2 : 0000000000000000 x1 : 0000000000000000 x0 : ffff000133d51600
Call trace:
 refcount_warn_saturate+0xf4/0x148
 mlx5_core_put_rsc+0x88/0xa0 [mlx5_ib]
 mlx5_core_destroy_rq_tracked+0x64/0x98 [mlx5_ib]
 mlx5_ib_destroy_wq+0x34/0x80 [mlx5_ib]
 ib_destroy_wq_user+0x30/0xc0 [ib_core]
 uverbs_free_wq+0x28/0x58 [ib_uverbs]
 destroy_hw_idr_uobject+0x34/0x78 [ib_uverbs]
 uverbs_destroy_uobject+0x48/0x240 [ib_uverbs]
 __uverbs_cleanup_ufile+0xd4/0x1a8 [ib_uverbs]
 uverbs_destroy_ufile_hw+0x48/0x120 [ib_uverbs]
 ib_uverbs_close+0x2c/0x100 [ib_uverbs]
 __fput+0xd8/0x2f0
 __fput_sync+0x50/0x70
 __arm64_sys_close+0x40/0x90
 invoke_syscall.constprop.0+0x74/0xd0
 do_el0_svc+0x48/0xe8
 el0_svc+0x44/0x1d0
 el0t_64_sync_handler+0x120/0x130
 el0t_64_sync+0x1a4/0x1a8

Fixes: e2013b212f9f ("net/mlx5_core: Add RQ and SQ event handling")
Signed-off-by: Patrisious Haddad <phaddad@nvidia.com>
Link: https://patch.msgid.link/3181433ccdd695c63560eeeb3f0c990961732101.1745839855.git.leon@kernel.org
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/hw/mlx5/qpc.c | 30 ++++++++++++++++++++++++++++--
 include/linux/mlx5/driver.h      |  1 +
 2 files changed, 29 insertions(+), 2 deletions(-)

diff --git a/drivers/infiniband/hw/mlx5/qpc.c b/drivers/infiniband/hw/mlx5/qpc.c
index d3dcc272200a..146d03ae40bd 100644
--- a/drivers/infiniband/hw/mlx5/qpc.c
+++ b/drivers/infiniband/hw/mlx5/qpc.c
@@ -21,8 +21,10 @@ mlx5_get_rsc(struct mlx5_qp_table *table, u32 rsn)
 	spin_lock_irqsave(&table->lock, flags);
 
 	common = radix_tree_lookup(&table->tree, rsn);
-	if (common)
+	if (common && !common->invalid)
 		refcount_inc(&common->refcount);
+	else
+		common = NULL;
 
 	spin_unlock_irqrestore(&table->lock, flags);
 
@@ -178,6 +180,18 @@ static int create_resource_common(struct mlx5_ib_dev *dev,
 	return 0;
 }
 
+static void modify_resource_common_state(struct mlx5_ib_dev *dev,
+					 struct mlx5_core_qp *qp,
+					 bool invalid)
+{
+	struct mlx5_qp_table *table = &dev->qp_table;
+	unsigned long flags;
+
+	spin_lock_irqsave(&table->lock, flags);
+	qp->common.invalid = invalid;
+	spin_unlock_irqrestore(&table->lock, flags);
+}
+
 static void destroy_resource_common(struct mlx5_ib_dev *dev,
 				    struct mlx5_core_qp *qp)
 {
@@ -609,8 +623,20 @@ err_destroy_rq:
 int mlx5_core_destroy_rq_tracked(struct mlx5_ib_dev *dev,
 				 struct mlx5_core_qp *rq)
 {
+	int ret;
+
+	/* The rq destruction can be called again in case it fails, hence we
+	 * mark the common resource as invalid and only once FW destruction
+	 * is completed successfully we actually destroy the resources.
+	 */
+	modify_resource_common_state(dev, rq, true);
+	ret = destroy_rq_tracked(dev, rq->qpn, rq->uid);
+	if (ret) {
+		modify_resource_common_state(dev, rq, false);
+		return ret;
+	}
 	destroy_resource_common(dev, rq);
-	return destroy_rq_tracked(dev, rq->qpn, rq->uid);
+	return 0;
 }
 
 static void destroy_sq_tracked(struct mlx5_ib_dev *dev, u32 sqn, u16 uid)
diff --git a/include/linux/mlx5/driver.h b/include/linux/mlx5/driver.h
index d1dfbad9a447..e6ba8f4f4bd1 100644
--- a/include/linux/mlx5/driver.h
+++ b/include/linux/mlx5/driver.h
@@ -398,6 +398,7 @@ struct mlx5_core_rsc_common {
 	enum mlx5_res_type	res;
 	refcount_t		refcount;
 	struct completion	free;
+	bool			invalid;
 };
 
 struct mlx5_uars_page {

From 7ccc2a0646ecbf2f8676b7ee0dc4aaccaa2e5ce1 Mon Sep 17 00:00:00 2001
From: "Dr. David Alan Gilbert" <linux@treblig.org>
Date: Mon, 5 May 2025 21:54:19 +0100
Subject: [PATCH 32/57] IB/hfi1: Remove unused sc_drop and sdma_all_idle

sc_drop() and sdma_all_idle() were both added in 2015's commit
7724105686e7 ("IB/hfi1: add driver files") but have remained unused.

Remove them.

Link: https://patch.msgid.link/r/20250505205419.88131-1-linux@treblig.org
Signed-off-by: Dr. David Alan Gilbert <linux@treblig.org>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
---
 drivers/infiniband/hw/hfi1/pio.c  | 10 ----------
 drivers/infiniband/hw/hfi1/pio.h  |  1 -
 drivers/infiniband/hw/hfi1/sdma.c | 18 ------------------
 drivers/infiniband/hw/hfi1/sdma.h |  1 -
 4 files changed, 30 deletions(-)

diff --git a/drivers/infiniband/hw/hfi1/pio.c b/drivers/infiniband/hw/hfi1/pio.c
index 5a91cbda4aee..764286da2ce8 100644
--- a/drivers/infiniband/hw/hfi1/pio.c
+++ b/drivers/infiniband/hw/hfi1/pio.c
@@ -1361,16 +1361,6 @@ void sc_flush(struct send_context *sc)
 	sc_wait_for_packet_egress(sc, 1);
 }
 
-/* drop all packets on the context, no waiting until they are sent */
-void sc_drop(struct send_context *sc)
-{
-	if (!sc)
-		return;
-
-	dd_dev_info(sc->dd, "%s: context %u(%u) - not implemented\n",
-		    __func__, sc->sw_index, sc->hw_context);
-}
-
 /*
  * Start the software reaction to a context halt or SPC freeze:
  *	- mark the context as halted or frozen
diff --git a/drivers/infiniband/hw/hfi1/pio.h b/drivers/infiniband/hw/hfi1/pio.h
index d07cc6ea7c63..ab0f9a3a8d12 100644
--- a/drivers/infiniband/hw/hfi1/pio.h
+++ b/drivers/infiniband/hw/hfi1/pio.h
@@ -246,7 +246,6 @@ void sc_disable(struct send_context *sc);
 int sc_restart(struct send_context *sc);
 void sc_return_credits(struct send_context *sc);
 void sc_flush(struct send_context *sc);
-void sc_drop(struct send_context *sc);
 void sc_stop(struct send_context *sc, int bit);
 struct pio_buf *sc_buffer_alloc(struct send_context *sc, u32 dw_len,
 				pio_release_cb cb, void *arg);
diff --git a/drivers/infiniband/hw/hfi1/sdma.c b/drivers/infiniband/hw/hfi1/sdma.c
index 0d2b39b7c8b5..16a749d16ee9 100644
--- a/drivers/infiniband/hw/hfi1/sdma.c
+++ b/drivers/infiniband/hw/hfi1/sdma.c
@@ -1520,24 +1520,6 @@ void sdma_all_running(struct hfi1_devdata *dd)
 	}
 }
 
-/**
- * sdma_all_idle() - called when the link goes down
- * @dd: hfi1_devdata
- *
- * This routine moves all engines to the idle state.
- */
-void sdma_all_idle(struct hfi1_devdata *dd)
-{
-	struct sdma_engine *sde;
-	unsigned int i;
-
-	/* idle all engines */
-	for (i = 0; i < dd->num_sdma; ++i) {
-		sde = &dd->per_sdma[i];
-		sdma_process_event(sde, sdma_event_e70_go_idle);
-	}
-}
-
 /**
  * sdma_start() - called to kick off state processing for all engines
  * @dd: hfi1_devdata
diff --git a/drivers/infiniband/hw/hfi1/sdma.h b/drivers/infiniband/hw/hfi1/sdma.h
index d77246b48434..91dfd5d0c419 100644
--- a/drivers/infiniband/hw/hfi1/sdma.h
+++ b/drivers/infiniband/hw/hfi1/sdma.h
@@ -373,7 +373,6 @@ void sdma_start(struct hfi1_devdata *dd);
 void sdma_exit(struct hfi1_devdata *dd);
 void sdma_clean(struct hfi1_devdata *dd, size_t num_engines);
 void sdma_all_running(struct hfi1_devdata *dd);
-void sdma_all_idle(struct hfi1_devdata *dd);
 void sdma_freeze_notify(struct hfi1_devdata *dd, int go_idle);
 void sdma_freeze(struct hfi1_devdata *dd);
 void sdma_unfreeze(struct hfi1_devdata *dd);

From e56b4eab9cde7db525bf8dc57b489d157d5a201e Mon Sep 17 00:00:00 2001
From: "Dr. David Alan Gilbert" <linux@treblig.org>
Date: Mon, 5 May 2025 22:02:26 +0100
Subject: [PATCH 33/57] RDMA/siw: Remove unused siw_mem_add

siw_mem_add() was added in 2019 by commit 2251334dcac9 ("rdma/siw:
application buffer management") but has remained unused.

Remove it.

Link: https://patch.msgid.link/r/20250505210226.88994-1-linux@treblig.org
Signed-off-by: Dr. David Alan Gilbert <linux@treblig.org>
Acked-by: Bernard Metzler <bmt@zurich.ibm.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
---
 drivers/infiniband/sw/siw/siw_mem.c | 24 ------------------------
 drivers/infiniband/sw/siw/siw_mem.h |  1 -
 2 files changed, 25 deletions(-)

diff --git a/drivers/infiniband/sw/siw/siw_mem.c b/drivers/infiniband/sw/siw/siw_mem.c
index b17156995595..d5ddeb17bd22 100644
--- a/drivers/infiniband/sw/siw/siw_mem.c
+++ b/drivers/infiniband/sw/siw/siw_mem.c
@@ -17,30 +17,6 @@
 /* Stag lookup is based on its index part only (24 bits). */
 #define SIW_STAG_MAX_INDEX	0x00ffffff
 
-/*
- * The code avoids special Stag of zero and tries to randomize
- * STag values between 1 and SIW_STAG_MAX_INDEX.
- */
-int siw_mem_add(struct siw_device *sdev, struct siw_mem *m)
-{
-	struct xa_limit limit = XA_LIMIT(1, SIW_STAG_MAX_INDEX);
-	u32 id, next;
-
-	get_random_bytes(&next, 4);
-	next &= SIW_STAG_MAX_INDEX;
-
-	if (xa_alloc_cyclic(&sdev->mem_xa, &id, m, limit, &next,
-	    GFP_KERNEL) < 0)
-		return -ENOMEM;
-
-	/* Set the STag index part */
-	m->stag = id << 8;
-
-	siw_dbg_mem(m, "new MEM object\n");
-
-	return 0;
-}
-
 /*
  * siw_mem_id2obj()
  *
diff --git a/drivers/infiniband/sw/siw/siw_mem.h b/drivers/infiniband/sw/siw/siw_mem.h
index e74cfcd6dbc1..8e769d30e2ac 100644
--- a/drivers/infiniband/sw/siw/siw_mem.h
+++ b/drivers/infiniband/sw/siw/siw_mem.h
@@ -12,7 +12,6 @@ void siw_umem_release(struct siw_umem *umem);
 struct siw_pbl *siw_pbl_alloc(u32 num_buf);
 dma_addr_t siw_pbl_get_buffer(struct siw_pbl *pbl, u64 off, int *len, int *idx);
 struct siw_mem *siw_mem_id2obj(struct siw_device *sdev, int stag_index);
-int siw_mem_add(struct siw_device *sdev, struct siw_mem *m);
 int siw_invalidate_stag(struct ib_pd *pd, u32 stag);
 int siw_check_mem(struct ib_pd *pd, struct siw_mem *mem, u64 addr,
 		  enum ib_access_flags perms, int len);

From 4ffb62fa8925c1c1591145ff0f5ccfd5918b40df Mon Sep 17 00:00:00 2001
From: Junxian Huang <huangjunxian6@hisilicon.com>
Date: Wed, 7 May 2025 11:39:03 +0800
Subject: [PATCH 34/57] RDMA/hns: Fix build error of hns_roce_trace

Add include path to find hns_roce_trace.h to fix the following
build error:

In file included from drivers/infiniband/hw/hns/hns_roce_trace.h:213,
                 from drivers/infiniband/hw/hns/hns_roce_hw_v2.c:53:
./include/trace/define_trace.h:110:42: fatal error: ./hns_roce_trace.h: No such file or directory
  110 | #include TRACE_INCLUDE(TRACE_INCLUDE_FILE)
      |                                          ^
compilation terminated.

Fixes: 02007e3ddc07 ("RDMA/hns: Add trace for flush CQE")
Reported-by: Paul E. McKenney <paulmck@kernel.org>
Closes: https://lore.kernel.org/linux-next/b7dd4dda-37d8-47e4-8d78-b6585be21cfd@paulmck-laptop/T/#t
Tested-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Junxian Huang <huangjunxian6@hisilicon.com>
Link: https://patch.msgid.link/20250507033903.2879433-1-huangjunxian6@hisilicon.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/hw/hns/Makefile | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/infiniband/hw/hns/Makefile b/drivers/infiniband/hw/hns/Makefile
index 7917af8e6380..baf592e6f21b 100644
--- a/drivers/infiniband/hw/hns/Makefile
+++ b/drivers/infiniband/hw/hns/Makefile
@@ -4,6 +4,7 @@
 #
 
 ccflags-y :=  -I $(srctree)/drivers/net/ethernet/hisilicon/hns3
+ccflags-y +=  -I $(src)
 
 hns-roce-hw-v2-objs := hns_roce_main.o hns_roce_cmd.o hns_roce_pd.o \
 	hns_roce_ah.o hns_roce_hem.o hns_roce_mr.o hns_roce_qp.o \

From 285e871884ff3dc31c0c2c1a87f0018481bc8471 Mon Sep 17 00:00:00 2001
From: Leon Romanovsky <leonro@nvidia.com>
Date: Mon, 28 Apr 2025 12:22:16 +0300
Subject: [PATCH 35/57] mm/hmm: let users to tag specific PFN with DMA mapped
 bit

Introduce new sticky flag (HMM_PFN_DMA_MAPPED), which isn't overwritten
by HMM range fault. Such flag allows users to tag specific PFNs with
information if this specific PFN was already DMA mapped.

Tested-by: Jens Axboe <axboe@kernel.dk>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
---
 include/linux/hmm.h | 20 ++++++++++++++++--
 mm/hmm.c            | 51 ++++++++++++++++++++++++++++-----------------
 2 files changed, 50 insertions(+), 21 deletions(-)

diff --git a/include/linux/hmm.h b/include/linux/hmm.h
index 126a36571667..a43e56f273a1 100644
--- a/include/linux/hmm.h
+++ b/include/linux/hmm.h
@@ -23,6 +23,8 @@ struct mmu_interval_notifier;
  * HMM_PFN_WRITE - if the page memory can be written to (requires HMM_PFN_VALID)
  * HMM_PFN_ERROR - accessing the pfn is impossible and the device should
  *                 fail. ie poisoned memory, special pages, no vma, etc
+ * HMM_PFN_DMA_MAPPED - Flag preserved on input-to-output transformation
+ *                      to mark that page is already DMA mapped
  *
  * On input:
  * 0                 - Return the current state of the page, do not fault it.
@@ -36,13 +38,19 @@ enum hmm_pfn_flags {
 	HMM_PFN_VALID = 1UL << (BITS_PER_LONG - 1),
 	HMM_PFN_WRITE = 1UL << (BITS_PER_LONG - 2),
 	HMM_PFN_ERROR = 1UL << (BITS_PER_LONG - 3),
-	HMM_PFN_ORDER_SHIFT = (BITS_PER_LONG - 8),
+	/*
+	 * Sticky flags, carried from input to output,
+	 * don't forget to update HMM_PFN_INOUT_FLAGS
+	 */
+	HMM_PFN_DMA_MAPPED = 1UL << (BITS_PER_LONG - 4),
+
+	HMM_PFN_ORDER_SHIFT = (BITS_PER_LONG - 9),
 
 	/* Input flags */
 	HMM_PFN_REQ_FAULT = HMM_PFN_VALID,
 	HMM_PFN_REQ_WRITE = HMM_PFN_WRITE,
 
-	HMM_PFN_FLAGS = 0xFFUL << HMM_PFN_ORDER_SHIFT,
+	HMM_PFN_FLAGS = ~((1UL << HMM_PFN_ORDER_SHIFT) - 1),
 };
 
 /*
@@ -57,6 +65,14 @@ static inline struct page *hmm_pfn_to_page(unsigned long hmm_pfn)
 	return pfn_to_page(hmm_pfn & ~HMM_PFN_FLAGS);
 }
 
+/*
+ * hmm_pfn_to_phys() - return physical address pointed to by a device entry
+ */
+static inline phys_addr_t hmm_pfn_to_phys(unsigned long hmm_pfn)
+{
+	return __pfn_to_phys(hmm_pfn & ~HMM_PFN_FLAGS);
+}
+
 /*
  * hmm_pfn_to_map_order() - return the CPU mapping size order
  *
diff --git a/mm/hmm.c b/mm/hmm.c
index 082f7b7c0b9e..51fe8b011cc7 100644
--- a/mm/hmm.c
+++ b/mm/hmm.c
@@ -39,13 +39,20 @@ enum {
 	HMM_NEED_ALL_BITS = HMM_NEED_FAULT | HMM_NEED_WRITE_FAULT,
 };
 
+enum {
+	/* These flags are carried from input-to-output */
+	HMM_PFN_INOUT_FLAGS = HMM_PFN_DMA_MAPPED,
+};
+
 static int hmm_pfns_fill(unsigned long addr, unsigned long end,
 			 struct hmm_range *range, unsigned long cpu_flags)
 {
 	unsigned long i = (addr - range->start) >> PAGE_SHIFT;
 
-	for (; addr < end; addr += PAGE_SIZE, i++)
-		range->hmm_pfns[i] = cpu_flags;
+	for (; addr < end; addr += PAGE_SIZE, i++) {
+		range->hmm_pfns[i] &= HMM_PFN_INOUT_FLAGS;
+		range->hmm_pfns[i] |= cpu_flags;
+	}
 	return 0;
 }
 
@@ -202,8 +209,10 @@ static int hmm_vma_handle_pmd(struct mm_walk *walk, unsigned long addr,
 		return hmm_vma_fault(addr, end, required_fault, walk);
 
 	pfn = pmd_pfn(pmd) + ((addr & ~PMD_MASK) >> PAGE_SHIFT);
-	for (i = 0; addr < end; addr += PAGE_SIZE, i++, pfn++)
-		hmm_pfns[i] = pfn | cpu_flags;
+	for (i = 0; addr < end; addr += PAGE_SIZE, i++, pfn++) {
+		hmm_pfns[i] &= HMM_PFN_INOUT_FLAGS;
+		hmm_pfns[i] |= pfn | cpu_flags;
+	}
 	return 0;
 }
 #else /* CONFIG_TRANSPARENT_HUGEPAGE */
@@ -230,14 +239,14 @@ static int hmm_vma_handle_pte(struct mm_walk *walk, unsigned long addr,
 	unsigned long cpu_flags;
 	pte_t pte = ptep_get(ptep);
 	uint64_t pfn_req_flags = *hmm_pfn;
+	uint64_t new_pfn_flags = 0;
 
 	if (pte_none_mostly(pte)) {
 		required_fault =
 			hmm_pte_need_fault(hmm_vma_walk, pfn_req_flags, 0);
 		if (required_fault)
 			goto fault;
-		*hmm_pfn = 0;
-		return 0;
+		goto out;
 	}
 
 	if (!pte_present(pte)) {
@@ -253,16 +262,14 @@ static int hmm_vma_handle_pte(struct mm_walk *walk, unsigned long addr,
 			cpu_flags = HMM_PFN_VALID;
 			if (is_writable_device_private_entry(entry))
 				cpu_flags |= HMM_PFN_WRITE;
-			*hmm_pfn = swp_offset_pfn(entry) | cpu_flags;
-			return 0;
+			new_pfn_flags = swp_offset_pfn(entry) | cpu_flags;
+			goto out;
 		}
 
 		required_fault =
 			hmm_pte_need_fault(hmm_vma_walk, pfn_req_flags, 0);
-		if (!required_fault) {
-			*hmm_pfn = 0;
-			return 0;
-		}
+		if (!required_fault)
+			goto out;
 
 		if (!non_swap_entry(entry))
 			goto fault;
@@ -304,11 +311,13 @@ static int hmm_vma_handle_pte(struct mm_walk *walk, unsigned long addr,
 			pte_unmap(ptep);
 			return -EFAULT;
 		}
-		*hmm_pfn = HMM_PFN_ERROR;
-		return 0;
+		new_pfn_flags = HMM_PFN_ERROR;
+		goto out;
 	}
 
-	*hmm_pfn = pte_pfn(pte) | cpu_flags;
+	new_pfn_flags = pte_pfn(pte) | cpu_flags;
+out:
+	*hmm_pfn = (*hmm_pfn & HMM_PFN_INOUT_FLAGS) | new_pfn_flags;
 	return 0;
 
 fault:
@@ -448,8 +457,10 @@ static int hmm_vma_walk_pud(pud_t *pudp, unsigned long start, unsigned long end,
 		}
 
 		pfn = pud_pfn(pud) + ((addr & ~PUD_MASK) >> PAGE_SHIFT);
-		for (i = 0; i < npages; ++i, ++pfn)
-			hmm_pfns[i] = pfn | cpu_flags;
+		for (i = 0; i < npages; ++i, ++pfn) {
+			hmm_pfns[i] &= HMM_PFN_INOUT_FLAGS;
+			hmm_pfns[i] |= pfn | cpu_flags;
+		}
 		goto out_unlock;
 	}
 
@@ -507,8 +518,10 @@ static int hmm_vma_walk_hugetlb_entry(pte_t *pte, unsigned long hmask,
 	}
 
 	pfn = pte_pfn(entry) + ((start & ~hmask) >> PAGE_SHIFT);
-	for (; addr < end; addr += PAGE_SIZE, i++, pfn++)
-		range->hmm_pfns[i] = pfn | cpu_flags;
+	for (; addr < end; addr += PAGE_SIZE, i++, pfn++) {
+		range->hmm_pfns[i] &= HMM_PFN_INOUT_FLAGS;
+		range->hmm_pfns[i] |= pfn | cpu_flags;
+	}
 
 	spin_unlock(ptl);
 	return 0;

From 8cad47130566123b2c70ef2aa53be02ef1aee5e5 Mon Sep 17 00:00:00 2001
From: Leon Romanovsky <leonro@nvidia.com>
Date: Mon, 28 Apr 2025 12:22:17 +0300
Subject: [PATCH 36/57] mm/hmm: provide generic DMA managing logic

HMM callers use PFN list to populate range while calling
to hmm_range_fault(), the conversion from PFN to DMA address
is done by the callers with help of another DMA list. However,
it is wasteful on any modern platform and by doing the right
logic, that DMA list can be avoided.

Provide generic logic to manage these lists and gave an interface
to map/unmap PFNs to DMA addresses, without requiring from the callers
to be an experts in DMA core API.

Tested-by: Jens Axboe <axboe@kernel.dk>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
---
 include/linux/hmm-dma.h |  33 +++++++
 include/linux/hmm.h     |   6 +-
 mm/hmm.c                | 214 +++++++++++++++++++++++++++++++++++++++-
 3 files changed, 251 insertions(+), 2 deletions(-)
 create mode 100644 include/linux/hmm-dma.h

diff --git a/include/linux/hmm-dma.h b/include/linux/hmm-dma.h
new file mode 100644
index 000000000000..f58b9fc71999
--- /dev/null
+++ b/include/linux/hmm-dma.h
@@ -0,0 +1,33 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/* Copyright (c) 2024 NVIDIA Corporation & Affiliates */
+#ifndef LINUX_HMM_DMA_H
+#define LINUX_HMM_DMA_H
+
+#include <linux/dma-mapping.h>
+
+struct dma_iova_state;
+struct pci_p2pdma_map_state;
+
+/*
+ * struct hmm_dma_map - array of PFNs and DMA addresses
+ *
+ * @state: DMA IOVA state
+ * @pfns: array of PFNs
+ * @dma_list: array of DMA addresses
+ * @dma_entry_size: size of each DMA entry in the array
+ */
+struct hmm_dma_map {
+	struct dma_iova_state state;
+	unsigned long *pfn_list;
+	dma_addr_t *dma_list;
+	size_t dma_entry_size;
+};
+
+int hmm_dma_map_alloc(struct device *dev, struct hmm_dma_map *map,
+		      size_t nr_entries, size_t dma_entry_size);
+void hmm_dma_map_free(struct device *dev, struct hmm_dma_map *map);
+dma_addr_t hmm_dma_map_pfn(struct device *dev, struct hmm_dma_map *map,
+			   size_t idx,
+			   struct pci_p2pdma_map_state *p2pdma_state);
+bool hmm_dma_unmap_pfn(struct device *dev, struct hmm_dma_map *map, size_t idx);
+#endif /* LINUX_HMM_DMA_H */
diff --git a/include/linux/hmm.h b/include/linux/hmm.h
index a43e56f273a1..db75ffc949a7 100644
--- a/include/linux/hmm.h
+++ b/include/linux/hmm.h
@@ -23,6 +23,8 @@ struct mmu_interval_notifier;
  * HMM_PFN_WRITE - if the page memory can be written to (requires HMM_PFN_VALID)
  * HMM_PFN_ERROR - accessing the pfn is impossible and the device should
  *                 fail. ie poisoned memory, special pages, no vma, etc
+ * HMM_PFN_P2PDMA - P2P page
+ * HMM_PFN_P2PDMA_BUS - Bus mapped P2P transfer
  * HMM_PFN_DMA_MAPPED - Flag preserved on input-to-output transformation
  *                      to mark that page is already DMA mapped
  *
@@ -43,8 +45,10 @@ enum hmm_pfn_flags {
 	 * don't forget to update HMM_PFN_INOUT_FLAGS
 	 */
 	HMM_PFN_DMA_MAPPED = 1UL << (BITS_PER_LONG - 4),
+	HMM_PFN_P2PDMA     = 1UL << (BITS_PER_LONG - 5),
+	HMM_PFN_P2PDMA_BUS = 1UL << (BITS_PER_LONG - 6),
 
-	HMM_PFN_ORDER_SHIFT = (BITS_PER_LONG - 9),
+	HMM_PFN_ORDER_SHIFT = (BITS_PER_LONG - 11),
 
 	/* Input flags */
 	HMM_PFN_REQ_FAULT = HMM_PFN_VALID,
diff --git a/mm/hmm.c b/mm/hmm.c
index 51fe8b011cc7..a8bf097677f3 100644
--- a/mm/hmm.c
+++ b/mm/hmm.c
@@ -10,6 +10,7 @@
  */
 #include <linux/pagewalk.h>
 #include <linux/hmm.h>
+#include <linux/hmm-dma.h>
 #include <linux/init.h>
 #include <linux/rmap.h>
 #include <linux/swap.h>
@@ -23,6 +24,7 @@
 #include <linux/sched/mm.h>
 #include <linux/jump_label.h>
 #include <linux/dma-mapping.h>
+#include <linux/pci-p2pdma.h>
 #include <linux/mmu_notifier.h>
 #include <linux/memory_hotplug.h>
 
@@ -41,7 +43,8 @@ enum {
 
 enum {
 	/* These flags are carried from input-to-output */
-	HMM_PFN_INOUT_FLAGS = HMM_PFN_DMA_MAPPED,
+	HMM_PFN_INOUT_FLAGS = HMM_PFN_DMA_MAPPED | HMM_PFN_P2PDMA |
+			      HMM_PFN_P2PDMA_BUS,
 };
 
 static int hmm_pfns_fill(unsigned long addr, unsigned long end,
@@ -620,3 +623,212 @@ int hmm_range_fault(struct hmm_range *range)
 	return ret;
 }
 EXPORT_SYMBOL(hmm_range_fault);
+
+/**
+ * hmm_dma_map_alloc - Allocate HMM map structure
+ * @dev: device to allocate structure for
+ * @map: HMM map to allocate
+ * @nr_entries: number of entries in the map
+ * @dma_entry_size: size of the DMA entry in the map
+ *
+ * Allocate the HMM map structure and all the lists it contains.
+ * Return 0 on success, -ENOMEM on failure.
+ */
+int hmm_dma_map_alloc(struct device *dev, struct hmm_dma_map *map,
+		      size_t nr_entries, size_t dma_entry_size)
+{
+	bool dma_need_sync = false;
+	bool use_iova;
+
+	if (!(nr_entries * PAGE_SIZE / dma_entry_size))
+		return -EINVAL;
+
+	/*
+	 * The HMM API violates our normal DMA buffer ownership rules and can't
+	 * transfer buffer ownership.  The dma_addressing_limited() check is a
+	 * best approximation to ensure no swiotlb buffering happens.
+	 */
+#ifdef CONFIG_DMA_NEED_SYNC
+	dma_need_sync = !dev->dma_skip_sync;
+#endif /* CONFIG_DMA_NEED_SYNC */
+	if (dma_need_sync || dma_addressing_limited(dev))
+		return -EOPNOTSUPP;
+
+	map->dma_entry_size = dma_entry_size;
+	map->pfn_list = kvcalloc(nr_entries, sizeof(*map->pfn_list),
+				 GFP_KERNEL | __GFP_NOWARN);
+	if (!map->pfn_list)
+		return -ENOMEM;
+
+	use_iova = dma_iova_try_alloc(dev, &map->state, 0,
+			nr_entries * PAGE_SIZE);
+	if (!use_iova && dma_need_unmap(dev)) {
+		map->dma_list = kvcalloc(nr_entries, sizeof(*map->dma_list),
+					 GFP_KERNEL | __GFP_NOWARN);
+		if (!map->dma_list)
+			goto err_dma;
+	}
+	return 0;
+
+err_dma:
+	kvfree(map->pfn_list);
+	return -ENOMEM;
+}
+EXPORT_SYMBOL_GPL(hmm_dma_map_alloc);
+
+/**
+ * hmm_dma_map_free - iFree HMM map structure
+ * @dev: device to free structure from
+ * @map: HMM map containing the various lists and state
+ *
+ * Free the HMM map structure and all the lists it contains.
+ */
+void hmm_dma_map_free(struct device *dev, struct hmm_dma_map *map)
+{
+	if (dma_use_iova(&map->state))
+		dma_iova_free(dev, &map->state);
+	kvfree(map->pfn_list);
+	kvfree(map->dma_list);
+}
+EXPORT_SYMBOL_GPL(hmm_dma_map_free);
+
+/**
+ * hmm_dma_map_pfn - Map a physical HMM page to DMA address
+ * @dev: Device to map the page for
+ * @map: HMM map
+ * @idx: Index into the PFN and dma address arrays
+ * @p2pdma_state: PCI P2P state.
+ *
+ * dma_alloc_iova() allocates IOVA based on the size specified by their use in
+ * iova->size. Call this function after IOVA allocation to link whole @page
+ * to get the DMA address. Note that very first call to this function
+ * will have @offset set to 0 in the IOVA space allocated from
+ * dma_alloc_iova(). For subsequent calls to this function on same @iova,
+ * @offset needs to be advanced by the caller with the size of previous
+ * page that was linked + DMA address returned for the previous page that was
+ * linked by this function.
+ */
+dma_addr_t hmm_dma_map_pfn(struct device *dev, struct hmm_dma_map *map,
+			   size_t idx,
+			   struct pci_p2pdma_map_state *p2pdma_state)
+{
+	struct dma_iova_state *state = &map->state;
+	dma_addr_t *dma_addrs = map->dma_list;
+	unsigned long *pfns = map->pfn_list;
+	struct page *page = hmm_pfn_to_page(pfns[idx]);
+	phys_addr_t paddr = hmm_pfn_to_phys(pfns[idx]);
+	size_t offset = idx * map->dma_entry_size;
+	unsigned long attrs = 0;
+	dma_addr_t dma_addr;
+	int ret;
+
+	if ((pfns[idx] & HMM_PFN_DMA_MAPPED) &&
+	    !(pfns[idx] & HMM_PFN_P2PDMA_BUS)) {
+		/*
+		 * We are in this flow when there is a need to resync flags,
+		 * for example when page was already linked in prefetch call
+		 * with READ flag and now we need to add WRITE flag
+		 *
+		 * This page was already programmed to HW and we don't want/need
+		 * to unlink and link it again just to resync flags.
+		 */
+		if (dma_use_iova(state))
+			return state->addr + offset;
+
+		/*
+		 * Without dma_need_unmap, the dma_addrs array is NULL, thus we
+		 * need to regenerate the address below even if there already
+		 * was a mapping. But !dma_need_unmap implies that the
+		 * mapping stateless, so this is fine.
+		 */
+		if (dma_need_unmap(dev))
+			return dma_addrs[idx];
+
+		/* Continue to remapping */
+	}
+
+	switch (pci_p2pdma_state(p2pdma_state, dev, page)) {
+	case PCI_P2PDMA_MAP_NONE:
+		break;
+	case PCI_P2PDMA_MAP_THRU_HOST_BRIDGE:
+		attrs |= DMA_ATTR_SKIP_CPU_SYNC;
+		pfns[idx] |= HMM_PFN_P2PDMA;
+		break;
+	case PCI_P2PDMA_MAP_BUS_ADDR:
+		pfns[idx] |= HMM_PFN_P2PDMA_BUS | HMM_PFN_DMA_MAPPED;
+		return pci_p2pdma_bus_addr_map(p2pdma_state, paddr);
+	default:
+		return DMA_MAPPING_ERROR;
+	}
+
+	if (dma_use_iova(state)) {
+		ret = dma_iova_link(dev, state, paddr, offset,
+				    map->dma_entry_size, DMA_BIDIRECTIONAL,
+				    attrs);
+		if (ret)
+			goto error;
+
+		ret = dma_iova_sync(dev, state, offset, map->dma_entry_size);
+		if (ret) {
+			dma_iova_unlink(dev, state, offset, map->dma_entry_size,
+					DMA_BIDIRECTIONAL, attrs);
+			goto error;
+		}
+
+		dma_addr = state->addr + offset;
+	} else {
+		if (WARN_ON_ONCE(dma_need_unmap(dev) && !dma_addrs))
+			goto error;
+
+		dma_addr = dma_map_page(dev, page, 0, map->dma_entry_size,
+					DMA_BIDIRECTIONAL);
+		if (dma_mapping_error(dev, dma_addr))
+			goto error;
+
+		if (dma_need_unmap(dev))
+			dma_addrs[idx] = dma_addr;
+	}
+	pfns[idx] |= HMM_PFN_DMA_MAPPED;
+	return dma_addr;
+error:
+	pfns[idx] &= ~HMM_PFN_P2PDMA;
+	return DMA_MAPPING_ERROR;
+
+}
+EXPORT_SYMBOL_GPL(hmm_dma_map_pfn);
+
+/**
+ * hmm_dma_unmap_pfn - Unmap a physical HMM page from DMA address
+ * @dev: Device to unmap the page from
+ * @map: HMM map
+ * @idx: Index of the PFN to unmap
+ *
+ * Returns true if the PFN was mapped and has been unmapped, false otherwise.
+ */
+bool hmm_dma_unmap_pfn(struct device *dev, struct hmm_dma_map *map, size_t idx)
+{
+	const unsigned long valid_dma = HMM_PFN_VALID | HMM_PFN_DMA_MAPPED;
+	struct dma_iova_state *state = &map->state;
+	dma_addr_t *dma_addrs = map->dma_list;
+	unsigned long *pfns = map->pfn_list;
+	unsigned long attrs = 0;
+
+	if ((pfns[idx] & valid_dma) != valid_dma)
+		return false;
+
+	if (pfns[idx] & HMM_PFN_P2PDMA_BUS)
+		; /* no need to unmap bus address P2P mappings */
+	else if (dma_use_iova(state)) {
+		if (pfns[idx] & HMM_PFN_P2PDMA)
+			attrs |= DMA_ATTR_SKIP_CPU_SYNC;
+		dma_iova_unlink(dev, state, idx * map->dma_entry_size,
+				map->dma_entry_size, DMA_BIDIRECTIONAL, attrs);
+	} else if (dma_need_unmap(dev))
+		dma_unmap_page(dev, dma_addrs[idx], map->dma_entry_size,
+			       DMA_BIDIRECTIONAL);
+
+	pfns[idx] &=
+		~(HMM_PFN_DMA_MAPPED | HMM_PFN_P2PDMA | HMM_PFN_P2PDMA_BUS);
+	return true;
+}
+EXPORT_SYMBOL_GPL(hmm_dma_unmap_pfn);

From eedd5b1276e76d6b260a7a77a149ef5155aa76f0 Mon Sep 17 00:00:00 2001
From: Leon Romanovsky <leonro@nvidia.com>
Date: Mon, 28 Apr 2025 12:22:18 +0300
Subject: [PATCH 37/57] RDMA/umem: Store ODP access mask information in PFN

As a preparation to remove dma_list, store access mask in PFN pointer
and not in dma_addr_t.

Tested-by: Jens Axboe <axboe@kernel.dk>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
---
 drivers/infiniband/core/umem_odp.c   | 101 +++++++++++----------------
 drivers/infiniband/hw/mlx5/mlx5_ib.h |   1 +
 drivers/infiniband/hw/mlx5/odp.c     |  37 +++++-----
 drivers/infiniband/sw/rxe/rxe_odp.c  |  14 ++--
 include/rdma/ib_umem_odp.h           |  14 +---
 5 files changed, 69 insertions(+), 98 deletions(-)

diff --git a/drivers/infiniband/core/umem_odp.c b/drivers/infiniband/core/umem_odp.c
index e9fa22d31c23..e1a5a567efb3 100644
--- a/drivers/infiniband/core/umem_odp.c
+++ b/drivers/infiniband/core/umem_odp.c
@@ -296,22 +296,11 @@ EXPORT_SYMBOL(ib_umem_odp_release);
 static int ib_umem_odp_map_dma_single_page(
 		struct ib_umem_odp *umem_odp,
 		unsigned int dma_index,
-		struct page *page,
-		u64 access_mask)
+		struct page *page)
 {
 	struct ib_device *dev = umem_odp->umem.ibdev;
 	dma_addr_t *dma_addr = &umem_odp->dma_list[dma_index];
 
-	if (*dma_addr) {
-		/*
-		 * If the page is already dma mapped it means it went through
-		 * a non-invalidating trasition, like read-only to writable.
-		 * Resync the flags.
-		 */
-		*dma_addr = (*dma_addr & ODP_DMA_ADDR_MASK) | access_mask;
-		return 0;
-	}
-
 	*dma_addr = ib_dma_map_page(dev, page, 0, 1 << umem_odp->page_shift,
 				    DMA_BIDIRECTIONAL);
 	if (ib_dma_mapping_error(dev, *dma_addr)) {
@@ -319,7 +308,6 @@ static int ib_umem_odp_map_dma_single_page(
 		return -EFAULT;
 	}
 	umem_odp->npages++;
-	*dma_addr |= access_mask;
 	return 0;
 }
 
@@ -355,9 +343,6 @@ int ib_umem_odp_map_dma_and_lock(struct ib_umem_odp *umem_odp, u64 user_virt,
 	struct hmm_range range = {};
 	unsigned long timeout;
 
-	if (access_mask == 0)
-		return -EINVAL;
-
 	if (user_virt < ib_umem_start(umem_odp) ||
 	    user_virt + bcnt > ib_umem_end(umem_odp))
 		return -EFAULT;
@@ -383,7 +368,7 @@ int ib_umem_odp_map_dma_and_lock(struct ib_umem_odp *umem_odp, u64 user_virt,
 	if (fault) {
 		range.default_flags = HMM_PFN_REQ_FAULT;
 
-		if (access_mask & ODP_WRITE_ALLOWED_BIT)
+		if (access_mask & HMM_PFN_WRITE)
 			range.default_flags |= HMM_PFN_REQ_WRITE;
 	}
 
@@ -415,22 +400,17 @@ retry:
 	for (pfn_index = 0; pfn_index < num_pfns;
 		pfn_index += 1 << (page_shift - PAGE_SHIFT), dma_index++) {
 
-		if (fault) {
-			/*
-			 * Since we asked for hmm_range_fault() to populate
-			 * pages it shouldn't return an error entry on success.
-			 */
-			WARN_ON(range.hmm_pfns[pfn_index] & HMM_PFN_ERROR);
-			WARN_ON(!(range.hmm_pfns[pfn_index] & HMM_PFN_VALID));
-		} else {
-			if (!(range.hmm_pfns[pfn_index] & HMM_PFN_VALID)) {
-				WARN_ON(umem_odp->dma_list[dma_index]);
-				continue;
-			}
-			access_mask = ODP_READ_ALLOWED_BIT;
-			if (range.hmm_pfns[pfn_index] & HMM_PFN_WRITE)
-				access_mask |= ODP_WRITE_ALLOWED_BIT;
-		}
+		/*
+		 * Since we asked for hmm_range_fault() to populate
+		 * pages it shouldn't return an error entry on success.
+		 */
+		WARN_ON(fault && range.hmm_pfns[pfn_index] & HMM_PFN_ERROR);
+		WARN_ON(fault && !(range.hmm_pfns[pfn_index] & HMM_PFN_VALID));
+		if (!(range.hmm_pfns[pfn_index] & HMM_PFN_VALID))
+			continue;
+
+		if (range.hmm_pfns[pfn_index] & HMM_PFN_DMA_MAPPED)
+			continue;
 
 		hmm_order = hmm_pfn_to_map_order(range.hmm_pfns[pfn_index]);
 		/* If a hugepage was detected and ODP wasn't set for, the umem
@@ -445,13 +425,14 @@ retry:
 		}
 
 		ret = ib_umem_odp_map_dma_single_page(
-				umem_odp, dma_index, hmm_pfn_to_page(range.hmm_pfns[pfn_index]),
-				access_mask);
+			umem_odp, dma_index,
+			hmm_pfn_to_page(range.hmm_pfns[pfn_index]));
 		if (ret < 0) {
 			ibdev_dbg(umem_odp->umem.ibdev,
 				  "ib_umem_odp_map_dma_single_page failed with error %d\n", ret);
 			break;
 		}
+		range.hmm_pfns[pfn_index] |= HMM_PFN_DMA_MAPPED;
 	}
 	/* upon success lock should stay on hold for the callee */
 	if (!ret)
@@ -471,7 +452,6 @@ EXPORT_SYMBOL(ib_umem_odp_map_dma_and_lock);
 void ib_umem_odp_unmap_dma_pages(struct ib_umem_odp *umem_odp, u64 virt,
 				 u64 bound)
 {
-	dma_addr_t dma_addr;
 	dma_addr_t dma;
 	int idx;
 	u64 addr;
@@ -482,34 +462,37 @@ void ib_umem_odp_unmap_dma_pages(struct ib_umem_odp *umem_odp, u64 virt,
 	virt = max_t(u64, virt, ib_umem_start(umem_odp));
 	bound = min_t(u64, bound, ib_umem_end(umem_odp));
 	for (addr = virt; addr < bound; addr += BIT(umem_odp->page_shift)) {
+		unsigned long pfn_idx = (addr - ib_umem_start(umem_odp)) >>
+					PAGE_SHIFT;
+		struct page *page =
+			hmm_pfn_to_page(umem_odp->pfn_list[pfn_idx]);
+
 		idx = (addr - ib_umem_start(umem_odp)) >> umem_odp->page_shift;
 		dma = umem_odp->dma_list[idx];
 
-		/* The access flags guaranteed a valid DMA address in case was NULL */
-		if (dma) {
-			unsigned long pfn_idx = (addr - ib_umem_start(umem_odp)) >> PAGE_SHIFT;
-			struct page *page = hmm_pfn_to_page(umem_odp->pfn_list[pfn_idx]);
+		if (!(umem_odp->pfn_list[pfn_idx] & HMM_PFN_VALID))
+			goto clear;
+		if (!(umem_odp->pfn_list[pfn_idx] & HMM_PFN_DMA_MAPPED))
+			goto clear;
 
-			dma_addr = dma & ODP_DMA_ADDR_MASK;
-			ib_dma_unmap_page(dev, dma_addr,
-					  BIT(umem_odp->page_shift),
-					  DMA_BIDIRECTIONAL);
-			if (dma & ODP_WRITE_ALLOWED_BIT) {
-				struct page *head_page = compound_head(page);
-				/*
-				 * set_page_dirty prefers being called with
-				 * the page lock. However, MMU notifiers are
-				 * called sometimes with and sometimes without
-				 * the lock. We rely on the umem_mutex instead
-				 * to prevent other mmu notifiers from
-				 * continuing and allowing the page mapping to
-				 * be removed.
-				 */
-				set_page_dirty(head_page);
-			}
-			umem_odp->dma_list[idx] = 0;
-			umem_odp->npages--;
+		ib_dma_unmap_page(dev, dma, BIT(umem_odp->page_shift),
+				  DMA_BIDIRECTIONAL);
+		if (umem_odp->pfn_list[pfn_idx] & HMM_PFN_WRITE) {
+			struct page *head_page = compound_head(page);
+			/*
+			 * set_page_dirty prefers being called with
+			 * the page lock. However, MMU notifiers are
+			 * called sometimes with and sometimes without
+			 * the lock. We rely on the umem_mutex instead
+			 * to prevent other mmu notifiers from
+			 * continuing and allowing the page mapping to
+			 * be removed.
+			 */
+			set_page_dirty(head_page);
 		}
+		umem_odp->npages--;
+clear:
+		umem_odp->pfn_list[pfn_idx] &= ~HMM_PFN_FLAGS;
 	}
 }
 EXPORT_SYMBOL(ib_umem_odp_unmap_dma_pages);
diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h
index ace2df3e1d9f..7424b23bb0d9 100644
--- a/drivers/infiniband/hw/mlx5/mlx5_ib.h
+++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h
@@ -351,6 +351,7 @@ struct mlx5_ib_flow_db {
 #define MLX5_IB_UPD_XLT_PD	      BIT(4)
 #define MLX5_IB_UPD_XLT_ACCESS	      BIT(5)
 #define MLX5_IB_UPD_XLT_INDIRECT      BIT(6)
+#define MLX5_IB_UPD_XLT_DOWNGRADE     BIT(7)
 
 /* Private QP creation flags to be passed in ib_qp_init_attr.create_flags.
  *
diff --git a/drivers/infiniband/hw/mlx5/odp.c b/drivers/infiniband/hw/mlx5/odp.c
index 86d8fa63bf69..6074c541885c 100644
--- a/drivers/infiniband/hw/mlx5/odp.c
+++ b/drivers/infiniband/hw/mlx5/odp.c
@@ -34,6 +34,7 @@
 #include <linux/kernel.h>
 #include <linux/dma-buf.h>
 #include <linux/dma-resv.h>
+#include <linux/hmm.h>
 
 #include "mlx5_ib.h"
 #include "cmd.h"
@@ -158,22 +159,12 @@ static void populate_klm(struct mlx5_klm *pklm, size_t idx, size_t nentries,
 	}
 }
 
-static u64 umem_dma_to_mtt(dma_addr_t umem_dma)
-{
-	u64 mtt_entry = umem_dma & ODP_DMA_ADDR_MASK;
-
-	if (umem_dma & ODP_READ_ALLOWED_BIT)
-		mtt_entry |= MLX5_IB_MTT_READ;
-	if (umem_dma & ODP_WRITE_ALLOWED_BIT)
-		mtt_entry |= MLX5_IB_MTT_WRITE;
-
-	return mtt_entry;
-}
-
 static void populate_mtt(__be64 *pas, size_t idx, size_t nentries,
 			 struct mlx5_ib_mr *mr, int flags)
 {
 	struct ib_umem_odp *odp = to_ib_umem_odp(mr->umem);
+	bool downgrade = flags & MLX5_IB_UPD_XLT_DOWNGRADE;
+	unsigned long pfn;
 	dma_addr_t pa;
 	size_t i;
 
@@ -181,8 +172,17 @@ static void populate_mtt(__be64 *pas, size_t idx, size_t nentries,
 		return;
 
 	for (i = 0; i < nentries; i++) {
+		pfn = odp->pfn_list[idx + i];
+		if (!(pfn & HMM_PFN_VALID))
+			/* ODP initialization */
+			continue;
+
 		pa = odp->dma_list[idx + i];
-		pas[i] = cpu_to_be64(umem_dma_to_mtt(pa));
+		pa |= MLX5_IB_MTT_READ;
+		if ((pfn & HMM_PFN_WRITE) && !downgrade)
+			pa |= MLX5_IB_MTT_WRITE;
+
+		pas[i] = cpu_to_be64(pa);
 	}
 }
 
@@ -303,8 +303,7 @@ static bool mlx5_ib_invalidate_range(struct mmu_interval_notifier *mni,
 		 * estimate the cost of another UMR vs. the cost of bigger
 		 * UMR.
 		 */
-		if (umem_odp->dma_list[idx] &
-		    (ODP_READ_ALLOWED_BIT | ODP_WRITE_ALLOWED_BIT)) {
+		if (umem_odp->pfn_list[idx] & HMM_PFN_VALID) {
 			if (!in_block) {
 				blk_start_idx = idx;
 				in_block = 1;
@@ -687,7 +686,7 @@ static int pagefault_real_mr(struct mlx5_ib_mr *mr, struct ib_umem_odp *odp,
 {
 	int page_shift, ret, np;
 	bool downgrade = flags & MLX5_PF_FLAGS_DOWNGRADE;
-	u64 access_mask;
+	u64 access_mask = 0;
 	u64 start_idx;
 	bool fault = !(flags & MLX5_PF_FLAGS_SNAPSHOT);
 	u32 xlt_flags = MLX5_IB_UPD_XLT_ATOMIC;
@@ -695,12 +694,14 @@ static int pagefault_real_mr(struct mlx5_ib_mr *mr, struct ib_umem_odp *odp,
 	if (flags & MLX5_PF_FLAGS_ENABLE)
 		xlt_flags |= MLX5_IB_UPD_XLT_ENABLE;
 
+	if (flags & MLX5_PF_FLAGS_DOWNGRADE)
+		xlt_flags |= MLX5_IB_UPD_XLT_DOWNGRADE;
+
 	page_shift = odp->page_shift;
 	start_idx = (user_va - ib_umem_start(odp)) >> page_shift;
-	access_mask = ODP_READ_ALLOWED_BIT;
 
 	if (odp->umem.writable && !downgrade)
-		access_mask |= ODP_WRITE_ALLOWED_BIT;
+		access_mask |= HMM_PFN_WRITE;
 
 	np = ib_umem_odp_map_dma_and_lock(odp, user_va, bcnt, access_mask, fault);
 	if (np < 0)
diff --git a/drivers/infiniband/sw/rxe/rxe_odp.c b/drivers/infiniband/sw/rxe/rxe_odp.c
index 6149d9ffe7f7..aae63e336e53 100644
--- a/drivers/infiniband/sw/rxe/rxe_odp.c
+++ b/drivers/infiniband/sw/rxe/rxe_odp.c
@@ -27,7 +27,7 @@ static bool rxe_ib_invalidate_range(struct mmu_interval_notifier *mni,
 	start = max_t(u64, ib_umem_start(umem_odp), range->start);
 	end = min_t(u64, ib_umem_end(umem_odp), range->end);
 
-	/* update umem_odp->dma_list */
+	/* update umem_odp->map.pfn_list */
 	ib_umem_odp_unmap_dma_pages(umem_odp, start, end);
 
 	mutex_unlock(&umem_odp->umem_mutex);
@@ -45,12 +45,11 @@ static int rxe_odp_do_pagefault_and_lock(struct rxe_mr *mr, u64 user_va, int bcn
 {
 	struct ib_umem_odp *umem_odp = to_ib_umem_odp(mr->umem);
 	bool fault = !(flags & RXE_PAGEFAULT_SNAPSHOT);
-	u64 access_mask;
+	u64 access_mask = 0;
 	int np;
 
-	access_mask = ODP_READ_ALLOWED_BIT;
 	if (umem_odp->umem.writable && !(flags & RXE_PAGEFAULT_RDONLY))
-		access_mask |= ODP_WRITE_ALLOWED_BIT;
+		access_mask |= HMM_PFN_WRITE;
 
 	/*
 	 * ib_umem_odp_map_dma_and_lock() locks umem_mutex on success.
@@ -138,7 +137,7 @@ static inline bool rxe_check_pagefault(struct ib_umem_odp *umem_odp,
 	while (addr < iova + length) {
 		idx = (addr - ib_umem_start(umem_odp)) >> umem_odp->page_shift;
 
-		if (!(umem_odp->dma_list[idx] & perm)) {
+		if (!(umem_odp->map.pfn_list[idx] & perm)) {
 			need_fault = true;
 			break;
 		}
@@ -162,15 +161,14 @@ static int rxe_odp_map_range_and_lock(struct rxe_mr *mr, u64 iova, int length, u
 {
 	struct ib_umem_odp *umem_odp = to_ib_umem_odp(mr->umem);
 	bool need_fault;
-	u64 perm;
+	u64 perm = 0;
 	int err;
 
 	if (unlikely(length < 1))
 		return -EINVAL;
 
-	perm = ODP_READ_ALLOWED_BIT;
 	if (!(flags & RXE_PAGEFAULT_RDONLY))
-		perm |= ODP_WRITE_ALLOWED_BIT;
+		perm |= HMM_PFN_WRITE;
 
 	mutex_lock(&umem_odp->umem_mutex);
 
diff --git a/include/rdma/ib_umem_odp.h b/include/rdma/ib_umem_odp.h
index 0844c1d05ac6..a345c26a745d 100644
--- a/include/rdma/ib_umem_odp.h
+++ b/include/rdma/ib_umem_odp.h
@@ -8,6 +8,7 @@
 
 #include <rdma/ib_umem.h>
 #include <rdma/ib_verbs.h>
+#include <linux/hmm.h>
 
 struct ib_umem_odp {
 	struct ib_umem umem;
@@ -67,19 +68,6 @@ static inline size_t ib_umem_odp_num_pages(struct ib_umem_odp *umem_odp)
 	       umem_odp->page_shift;
 }
 
-/*
- * 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_umem_odp *

From 1efe8c0670d6a6883faa09c9abc746c741f5664a Mon Sep 17 00:00:00 2001
From: Leon Romanovsky <leonro@nvidia.com>
Date: Mon, 28 Apr 2025 12:22:19 +0300
Subject: [PATCH 38/57] RDMA/core: Convert UMEM ODP DMA mapping to caching IOVA
 and page linkage

Reuse newly added DMA API to cache IOVA and only link/unlink pages
in fast path for UMEM ODP flow.

Tested-by: Jens Axboe <axboe@kernel.dk>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
---
 drivers/infiniband/core/umem_odp.c   | 104 ++++++---------------------
 drivers/infiniband/hw/mlx5/mlx5_ib.h |  11 +--
 drivers/infiniband/hw/mlx5/odp.c     |  42 +++++++----
 drivers/infiniband/hw/mlx5/umr.c     |  12 +++-
 drivers/infiniband/sw/rxe/rxe_odp.c  |   8 +--
 include/rdma/ib_umem_odp.h           |  13 +---
 6 files changed, 74 insertions(+), 116 deletions(-)

diff --git a/drivers/infiniband/core/umem_odp.c b/drivers/infiniband/core/umem_odp.c
index e1a5a567efb3..30cd8f353476 100644
--- a/drivers/infiniband/core/umem_odp.c
+++ b/drivers/infiniband/core/umem_odp.c
@@ -41,6 +41,7 @@
 #include <linux/hugetlb.h>
 #include <linux/interval_tree.h>
 #include <linux/hmm.h>
+#include <linux/hmm-dma.h>
 #include <linux/pagemap.h>
 
 #include <rdma/ib_umem_odp.h>
@@ -50,6 +51,7 @@
 static inline int ib_init_umem_odp(struct ib_umem_odp *umem_odp,
 				   const struct mmu_interval_notifier_ops *ops)
 {
+	struct ib_device *dev = umem_odp->umem.ibdev;
 	int ret;
 
 	umem_odp->umem.is_odp = 1;
@@ -59,7 +61,6 @@ static inline int ib_init_umem_odp(struct ib_umem_odp *umem_odp,
 		size_t page_size = 1UL << umem_odp->page_shift;
 		unsigned long start;
 		unsigned long end;
-		size_t ndmas, npfns;
 
 		start = ALIGN_DOWN(umem_odp->umem.address, page_size);
 		if (check_add_overflow(umem_odp->umem.address,
@@ -70,36 +71,23 @@ static inline int ib_init_umem_odp(struct ib_umem_odp *umem_odp,
 		if (unlikely(end < page_size))
 			return -EOVERFLOW;
 
-		ndmas = (end - start) >> umem_odp->page_shift;
-		if (!ndmas)
-			return -EINVAL;
-
-		npfns = (end - start) >> PAGE_SHIFT;
-		umem_odp->pfn_list = kvcalloc(
-			npfns, sizeof(*umem_odp->pfn_list), GFP_KERNEL);
-		if (!umem_odp->pfn_list)
-			return -ENOMEM;
-
-		umem_odp->dma_list = kvcalloc(
-			ndmas, sizeof(*umem_odp->dma_list), GFP_KERNEL);
-		if (!umem_odp->dma_list) {
-			ret = -ENOMEM;
-			goto out_pfn_list;
-		}
+		ret = hmm_dma_map_alloc(dev->dma_device, &umem_odp->map,
+					(end - start) >> PAGE_SHIFT,
+					1 << umem_odp->page_shift);
+		if (ret)
+			return ret;
 
 		ret = mmu_interval_notifier_insert(&umem_odp->notifier,
 						   umem_odp->umem.owning_mm,
 						   start, end - start, ops);
 		if (ret)
-			goto out_dma_list;
+			goto out_free_map;
 	}
 
 	return 0;
 
-out_dma_list:
-	kvfree(umem_odp->dma_list);
-out_pfn_list:
-	kvfree(umem_odp->pfn_list);
+out_free_map:
+	hmm_dma_map_free(dev->dma_device, &umem_odp->map);
 	return ret;
 }
 
@@ -262,6 +250,8 @@ EXPORT_SYMBOL(ib_umem_odp_get);
 
 void ib_umem_odp_release(struct ib_umem_odp *umem_odp)
 {
+	struct ib_device *dev = umem_odp->umem.ibdev;
+
 	/*
 	 * Ensure that no more pages are mapped in the umem.
 	 *
@@ -274,48 +264,17 @@ void ib_umem_odp_release(struct ib_umem_odp *umem_odp)
 					    ib_umem_end(umem_odp));
 		mutex_unlock(&umem_odp->umem_mutex);
 		mmu_interval_notifier_remove(&umem_odp->notifier);
-		kvfree(umem_odp->dma_list);
-		kvfree(umem_odp->pfn_list);
+		hmm_dma_map_free(dev->dma_device, &umem_odp->map);
 	}
 	put_pid(umem_odp->tgid);
 	kfree(umem_odp);
 }
 EXPORT_SYMBOL(ib_umem_odp_release);
 
-/*
- * Map for DMA and insert a single page into the on-demand paging page tables.
- *
- * @umem: the umem to insert the page to.
- * @dma_index: index in the umem to add the dma to.
- * @page: the page struct to map and add.
- * @access_mask: access permissions needed for this page.
- *
- * The function returns -EFAULT if the DMA mapping operation fails.
- *
- */
-static int ib_umem_odp_map_dma_single_page(
-		struct ib_umem_odp *umem_odp,
-		unsigned int dma_index,
-		struct page *page)
-{
-	struct ib_device *dev = umem_odp->umem.ibdev;
-	dma_addr_t *dma_addr = &umem_odp->dma_list[dma_index];
-
-	*dma_addr = ib_dma_map_page(dev, page, 0, 1 << umem_odp->page_shift,
-				    DMA_BIDIRECTIONAL);
-	if (ib_dma_mapping_error(dev, *dma_addr)) {
-		*dma_addr = 0;
-		return -EFAULT;
-	}
-	umem_odp->npages++;
-	return 0;
-}
-
 /**
  * ib_umem_odp_map_dma_and_lock - DMA map userspace memory in an ODP MR and lock it.
  *
  * Maps the range passed in the argument to DMA addresses.
- * The DMA addresses of the mapped pages is updated in umem_odp->dma_list.
  * Upon success the ODP MR will be locked to let caller complete its device
  * page table update.
  *
@@ -372,7 +331,7 @@ int ib_umem_odp_map_dma_and_lock(struct ib_umem_odp *umem_odp, u64 user_virt,
 			range.default_flags |= HMM_PFN_REQ_WRITE;
 	}
 
-	range.hmm_pfns = &(umem_odp->pfn_list[pfn_start_idx]);
+	range.hmm_pfns = &(umem_odp->map.pfn_list[pfn_start_idx]);
 	timeout = jiffies + msecs_to_jiffies(HMM_RANGE_DEFAULT_TIMEOUT);
 
 retry:
@@ -423,16 +382,6 @@ retry:
 				  __func__, hmm_order, page_shift);
 			break;
 		}
-
-		ret = ib_umem_odp_map_dma_single_page(
-			umem_odp, dma_index,
-			hmm_pfn_to_page(range.hmm_pfns[pfn_index]));
-		if (ret < 0) {
-			ibdev_dbg(umem_odp->umem.ibdev,
-				  "ib_umem_odp_map_dma_single_page failed with error %d\n", ret);
-			break;
-		}
-		range.hmm_pfns[pfn_index] |= HMM_PFN_DMA_MAPPED;
 	}
 	/* upon success lock should stay on hold for the callee */
 	if (!ret)
@@ -452,32 +401,23 @@ EXPORT_SYMBOL(ib_umem_odp_map_dma_and_lock);
 void ib_umem_odp_unmap_dma_pages(struct ib_umem_odp *umem_odp, u64 virt,
 				 u64 bound)
 {
-	dma_addr_t dma;
-	int idx;
-	u64 addr;
 	struct ib_device *dev = umem_odp->umem.ibdev;
+	u64 addr;
 
 	lockdep_assert_held(&umem_odp->umem_mutex);
 
 	virt = max_t(u64, virt, ib_umem_start(umem_odp));
 	bound = min_t(u64, bound, ib_umem_end(umem_odp));
 	for (addr = virt; addr < bound; addr += BIT(umem_odp->page_shift)) {
-		unsigned long pfn_idx = (addr - ib_umem_start(umem_odp)) >>
-					PAGE_SHIFT;
-		struct page *page =
-			hmm_pfn_to_page(umem_odp->pfn_list[pfn_idx]);
+		u64 offset = addr - ib_umem_start(umem_odp);
+		size_t idx = offset >> umem_odp->page_shift;
+		unsigned long pfn = umem_odp->map.pfn_list[idx];
 
-		idx = (addr - ib_umem_start(umem_odp)) >> umem_odp->page_shift;
-		dma = umem_odp->dma_list[idx];
-
-		if (!(umem_odp->pfn_list[pfn_idx] & HMM_PFN_VALID))
-			goto clear;
-		if (!(umem_odp->pfn_list[pfn_idx] & HMM_PFN_DMA_MAPPED))
+		if (!hmm_dma_unmap_pfn(dev->dma_device, &umem_odp->map, idx))
 			goto clear;
 
-		ib_dma_unmap_page(dev, dma, BIT(umem_odp->page_shift),
-				  DMA_BIDIRECTIONAL);
-		if (umem_odp->pfn_list[pfn_idx] & HMM_PFN_WRITE) {
+		if (pfn & HMM_PFN_WRITE) {
+			struct page *page = hmm_pfn_to_page(pfn);
 			struct page *head_page = compound_head(page);
 			/*
 			 * set_page_dirty prefers being called with
@@ -492,7 +432,7 @@ void ib_umem_odp_unmap_dma_pages(struct ib_umem_odp *umem_odp, u64 virt,
 		}
 		umem_odp->npages--;
 clear:
-		umem_odp->pfn_list[pfn_idx] &= ~HMM_PFN_FLAGS;
+		umem_odp->map.pfn_list[idx] &= ~HMM_PFN_FLAGS;
 	}
 }
 EXPORT_SYMBOL(ib_umem_odp_unmap_dma_pages);
diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h
index 7424b23bb0d9..27fbf5f3a2d6 100644
--- a/drivers/infiniband/hw/mlx5/mlx5_ib.h
+++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h
@@ -1474,8 +1474,8 @@ void mlx5_ib_odp_cleanup_one(struct mlx5_ib_dev *ibdev);
 int __init mlx5_ib_odp_init(void);
 void mlx5_ib_odp_cleanup(void);
 int mlx5_odp_init_mkey_cache(struct mlx5_ib_dev *dev);
-void mlx5_odp_populate_xlt(void *xlt, size_t idx, size_t nentries,
-			   struct mlx5_ib_mr *mr, int flags);
+int mlx5_odp_populate_xlt(void *xlt, size_t idx, size_t nentries,
+			  struct mlx5_ib_mr *mr, int flags);
 
 int mlx5_ib_advise_mr_prefetch(struct ib_pd *pd,
 			       enum ib_uverbs_advise_mr_advice advice,
@@ -1496,8 +1496,11 @@ static inline int mlx5_odp_init_mkey_cache(struct mlx5_ib_dev *dev)
 {
 	return 0;
 }
-static inline void mlx5_odp_populate_xlt(void *xlt, size_t idx, size_t nentries,
-					 struct mlx5_ib_mr *mr, int flags) {}
+static inline int mlx5_odp_populate_xlt(void *xlt, size_t idx, size_t nentries,
+					struct mlx5_ib_mr *mr, int flags)
+{
+	return -EOPNOTSUPP;
+}
 
 static inline int
 mlx5_ib_advise_mr_prefetch(struct ib_pd *pd,
diff --git a/drivers/infiniband/hw/mlx5/odp.c b/drivers/infiniband/hw/mlx5/odp.c
index 6074c541885c..eaa2f9f5f3a9 100644
--- a/drivers/infiniband/hw/mlx5/odp.c
+++ b/drivers/infiniband/hw/mlx5/odp.c
@@ -35,6 +35,8 @@
 #include <linux/dma-buf.h>
 #include <linux/dma-resv.h>
 #include <linux/hmm.h>
+#include <linux/hmm-dma.h>
+#include <linux/pci-p2pdma.h>
 
 #include "mlx5_ib.h"
 #include "cmd.h"
@@ -159,40 +161,50 @@ static void populate_klm(struct mlx5_klm *pklm, size_t idx, size_t nentries,
 	}
 }
 
-static void populate_mtt(__be64 *pas, size_t idx, size_t nentries,
-			 struct mlx5_ib_mr *mr, int flags)
+static int populate_mtt(__be64 *pas, size_t start, size_t nentries,
+			struct mlx5_ib_mr *mr, int flags)
 {
 	struct ib_umem_odp *odp = to_ib_umem_odp(mr->umem);
 	bool downgrade = flags & MLX5_IB_UPD_XLT_DOWNGRADE;
-	unsigned long pfn;
-	dma_addr_t pa;
+	struct pci_p2pdma_map_state p2pdma_state = {};
+	struct ib_device *dev = odp->umem.ibdev;
 	size_t i;
 
 	if (flags & MLX5_IB_UPD_XLT_ZAP)
-		return;
+		return 0;
 
 	for (i = 0; i < nentries; i++) {
-		pfn = odp->pfn_list[idx + i];
+		unsigned long pfn = odp->map.pfn_list[start + i];
+		dma_addr_t dma_addr;
+
+		pfn = odp->map.pfn_list[start + i];
 		if (!(pfn & HMM_PFN_VALID))
 			/* ODP initialization */
 			continue;
 
-		pa = odp->dma_list[idx + i];
-		pa |= MLX5_IB_MTT_READ;
-		if ((pfn & HMM_PFN_WRITE) && !downgrade)
-			pa |= MLX5_IB_MTT_WRITE;
+		dma_addr = hmm_dma_map_pfn(dev->dma_device, &odp->map,
+					   start + i, &p2pdma_state);
+		if (ib_dma_mapping_error(dev, dma_addr))
+			return -EFAULT;
 
-		pas[i] = cpu_to_be64(pa);
+		dma_addr |= MLX5_IB_MTT_READ;
+		if ((pfn & HMM_PFN_WRITE) && !downgrade)
+			dma_addr |= MLX5_IB_MTT_WRITE;
+
+		pas[i] = cpu_to_be64(dma_addr);
+		odp->npages++;
 	}
+	return 0;
 }
 
-void mlx5_odp_populate_xlt(void *xlt, size_t idx, size_t nentries,
-			   struct mlx5_ib_mr *mr, int flags)
+int mlx5_odp_populate_xlt(void *xlt, size_t idx, size_t nentries,
+			  struct mlx5_ib_mr *mr, int flags)
 {
 	if (flags & MLX5_IB_UPD_XLT_INDIRECT) {
 		populate_klm(xlt, idx, nentries, mr, flags);
+		return 0;
 	} else {
-		populate_mtt(xlt, idx, nentries, mr, flags);
+		return populate_mtt(xlt, idx, nentries, mr, flags);
 	}
 }
 
@@ -303,7 +315,7 @@ static bool mlx5_ib_invalidate_range(struct mmu_interval_notifier *mni,
 		 * estimate the cost of another UMR vs. the cost of bigger
 		 * UMR.
 		 */
-		if (umem_odp->pfn_list[idx] & HMM_PFN_VALID) {
+		if (umem_odp->map.pfn_list[idx] & HMM_PFN_VALID) {
 			if (!in_block) {
 				blk_start_idx = idx;
 				in_block = 1;
diff --git a/drivers/infiniband/hw/mlx5/umr.c b/drivers/infiniband/hw/mlx5/umr.c
index 793f3c5c4d01..5be4426a2884 100644
--- a/drivers/infiniband/hw/mlx5/umr.c
+++ b/drivers/infiniband/hw/mlx5/umr.c
@@ -840,7 +840,17 @@ int mlx5r_umr_update_xlt(struct mlx5_ib_mr *mr, u64 idx, int npages,
 		size_to_map = npages * desc_size;
 		dma_sync_single_for_cpu(ddev, sg.addr, sg.length,
 					DMA_TO_DEVICE);
-		mlx5_odp_populate_xlt(xlt, idx, npages, mr, flags);
+		/*
+		 * npages is the maximum number of pages to map, but we
+		 * can't guarantee that all pages are actually mapped.
+		 *
+		 * For example, if page is p2p of type which is not supported
+		 * for mapping, the number of pages mapped will be less than
+		 * requested.
+		 */
+		err = mlx5_odp_populate_xlt(xlt, idx, npages, mr, flags);
+		if (err)
+			return err;
 		dma_sync_single_for_device(ddev, sg.addr, sg.length,
 					   DMA_TO_DEVICE);
 		sg.length = ALIGN(size_to_map, MLX5_UMR_FLEX_ALIGNMENT);
diff --git a/drivers/infiniband/sw/rxe/rxe_odp.c b/drivers/infiniband/sw/rxe/rxe_odp.c
index aae63e336e53..a1416626f61a 100644
--- a/drivers/infiniband/sw/rxe/rxe_odp.c
+++ b/drivers/infiniband/sw/rxe/rxe_odp.c
@@ -205,7 +205,7 @@ static int __rxe_odp_mr_copy(struct rxe_mr *mr, u64 iova, void *addr,
 	while (length > 0) {
 		u8 *src, *dest;
 
-		page = hmm_pfn_to_page(umem_odp->pfn_list[idx]);
+		page = hmm_pfn_to_page(umem_odp->map.pfn_list[idx]);
 		user_va = kmap_local_page(page);
 		if (!user_va)
 			return -EFAULT;
@@ -289,7 +289,7 @@ static enum resp_states rxe_odp_do_atomic_op(struct rxe_mr *mr, u64 iova,
 
 	idx = rxe_odp_iova_to_index(umem_odp, iova);
 	page_offset = rxe_odp_iova_to_page_offset(umem_odp, iova);
-	page = hmm_pfn_to_page(umem_odp->pfn_list[idx]);
+	page = hmm_pfn_to_page(umem_odp->map.pfn_list[idx]);
 	if (!page)
 		return RESPST_ERR_RKEY_VIOLATION;
 
@@ -355,7 +355,7 @@ int rxe_odp_flush_pmem_iova(struct rxe_mr *mr, u64 iova,
 		index = rxe_odp_iova_to_index(umem_odp, iova);
 		page_offset = rxe_odp_iova_to_page_offset(umem_odp, iova);
 
-		page = hmm_pfn_to_page(umem_odp->pfn_list[index]);
+		page = hmm_pfn_to_page(umem_odp->map.pfn_list[index]);
 		if (!page) {
 			mutex_unlock(&umem_odp->umem_mutex);
 			return -EFAULT;
@@ -401,7 +401,7 @@ enum resp_states rxe_odp_do_atomic_write(struct rxe_mr *mr, u64 iova, u64 value)
 
 	page_offset = rxe_odp_iova_to_page_offset(umem_odp, iova);
 	index = rxe_odp_iova_to_index(umem_odp, iova);
-	page = hmm_pfn_to_page(umem_odp->pfn_list[index]);
+	page = hmm_pfn_to_page(umem_odp->map.pfn_list[index]);
 	if (!page) {
 		mutex_unlock(&umem_odp->umem_mutex);
 		return RESPST_ERR_RKEY_VIOLATION;
diff --git a/include/rdma/ib_umem_odp.h b/include/rdma/ib_umem_odp.h
index a345c26a745d..2a24bf791c10 100644
--- a/include/rdma/ib_umem_odp.h
+++ b/include/rdma/ib_umem_odp.h
@@ -8,24 +8,17 @@
 
 #include <rdma/ib_umem.h>
 #include <rdma/ib_verbs.h>
-#include <linux/hmm.h>
+#include <linux/hmm-dma.h>
 
 struct ib_umem_odp {
 	struct ib_umem umem;
 	struct mmu_interval_notifier notifier;
 	struct pid *tgid;
 
-	/* An array of the pfns included in the on-demand paging umem. */
-	unsigned long *pfn_list;
+	struct hmm_dma_map map;
 
 	/*
-	 * An array with DMA addresses mapped for pfns in pfn_list.
-	 * The lower two bits designate access permissions.
-	 * See ODP_READ_ALLOWED_BIT and ODP_WRITE_ALLOWED_BIT.
-	 */
-	dma_addr_t		*dma_list;
-	/*
-	 * The umem_mutex protects the page_list and dma_list fields of an ODP
+	 * The umem_mutex protects the page_list field of an ODP
 	 * umem, allowing only a single thread to map/unmap pages. The mutex
 	 * also protects access to the mmu notifier counters.
 	 */

From 15a9f67e286b37e87dce0d9fa6ab702631c22d15 Mon Sep 17 00:00:00 2001
From: Leon Romanovsky <leonro@nvidia.com>
Date: Mon, 28 Apr 2025 12:22:20 +0300
Subject: [PATCH 39/57] RDMA/umem: Separate implicit ODP initialization from
 explicit ODP

Create separate functions for the implicit ODP initialization
which is different from the explicit ODP initialization.

Tested-by: Jens Axboe <axboe@kernel.dk>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
---
 drivers/infiniband/core/umem_odp.c | 87 +++++++++++++++---------------
 1 file changed, 44 insertions(+), 43 deletions(-)

diff --git a/drivers/infiniband/core/umem_odp.c b/drivers/infiniband/core/umem_odp.c
index 30cd8f353476..51d518989914 100644
--- a/drivers/infiniband/core/umem_odp.c
+++ b/drivers/infiniband/core/umem_odp.c
@@ -48,41 +48,44 @@
 
 #include "uverbs.h"
 
-static inline int ib_init_umem_odp(struct ib_umem_odp *umem_odp,
-				   const struct mmu_interval_notifier_ops *ops)
+static void ib_init_umem_implicit_odp(struct ib_umem_odp *umem_odp)
+{
+	umem_odp->is_implicit_odp = 1;
+	umem_odp->umem.is_odp = 1;
+	mutex_init(&umem_odp->umem_mutex);
+}
+
+static int ib_init_umem_odp(struct ib_umem_odp *umem_odp,
+			    const struct mmu_interval_notifier_ops *ops)
 {
 	struct ib_device *dev = umem_odp->umem.ibdev;
+	size_t page_size = 1UL << umem_odp->page_shift;
+	unsigned long start;
+	unsigned long end;
 	int ret;
 
 	umem_odp->umem.is_odp = 1;
 	mutex_init(&umem_odp->umem_mutex);
 
-	if (!umem_odp->is_implicit_odp) {
-		size_t page_size = 1UL << umem_odp->page_shift;
-		unsigned long start;
-		unsigned long end;
+	start = ALIGN_DOWN(umem_odp->umem.address, page_size);
+	if (check_add_overflow(umem_odp->umem.address,
+			       (unsigned long)umem_odp->umem.length, &end))
+		return -EOVERFLOW;
+	end = ALIGN(end, page_size);
+	if (unlikely(end < page_size))
+		return -EOVERFLOW;
 
-		start = ALIGN_DOWN(umem_odp->umem.address, page_size);
-		if (check_add_overflow(umem_odp->umem.address,
-				       (unsigned long)umem_odp->umem.length,
-				       &end))
-			return -EOVERFLOW;
-		end = ALIGN(end, page_size);
-		if (unlikely(end < page_size))
-			return -EOVERFLOW;
+	ret = hmm_dma_map_alloc(dev->dma_device, &umem_odp->map,
+				(end - start) >> PAGE_SHIFT,
+				1 << umem_odp->page_shift);
+	if (ret)
+		return ret;
 
-		ret = hmm_dma_map_alloc(dev->dma_device, &umem_odp->map,
-					(end - start) >> PAGE_SHIFT,
-					1 << umem_odp->page_shift);
-		if (ret)
-			return ret;
-
-		ret = mmu_interval_notifier_insert(&umem_odp->notifier,
-						   umem_odp->umem.owning_mm,
-						   start, end - start, ops);
-		if (ret)
-			goto out_free_map;
-	}
+	ret = mmu_interval_notifier_insert(&umem_odp->notifier,
+					   umem_odp->umem.owning_mm, start,
+					   end - start, ops);
+	if (ret)
+		goto out_free_map;
 
 	return 0;
 
@@ -106,7 +109,6 @@ struct ib_umem_odp *ib_umem_odp_alloc_implicit(struct ib_device *device,
 {
 	struct ib_umem *umem;
 	struct ib_umem_odp *umem_odp;
-	int ret;
 
 	if (access & IB_ACCESS_HUGETLB)
 		return ERR_PTR(-EINVAL);
@@ -118,16 +120,10 @@ struct ib_umem_odp *ib_umem_odp_alloc_implicit(struct ib_device *device,
 	umem->ibdev = device;
 	umem->writable = ib_access_writable(access);
 	umem->owning_mm = current->mm;
-	umem_odp->is_implicit_odp = 1;
 	umem_odp->page_shift = PAGE_SHIFT;
 
 	umem_odp->tgid = get_task_pid(current->group_leader, PIDTYPE_PID);
-	ret = ib_init_umem_odp(umem_odp, NULL);
-	if (ret) {
-		put_pid(umem_odp->tgid);
-		kfree(umem_odp);
-		return ERR_PTR(ret);
-	}
+	ib_init_umem_implicit_odp(umem_odp);
 	return umem_odp;
 }
 EXPORT_SYMBOL(ib_umem_odp_alloc_implicit);
@@ -248,7 +244,7 @@ err_put_pid:
 }
 EXPORT_SYMBOL(ib_umem_odp_get);
 
-void ib_umem_odp_release(struct ib_umem_odp *umem_odp)
+static void ib_umem_odp_free(struct ib_umem_odp *umem_odp)
 {
 	struct ib_device *dev = umem_odp->umem.ibdev;
 
@@ -258,14 +254,19 @@ void ib_umem_odp_release(struct ib_umem_odp *umem_odp)
 	 * It is the driver's responsibility to ensure, before calling us,
 	 * that the hardware will not attempt to access the MR any more.
 	 */
-	if (!umem_odp->is_implicit_odp) {
-		mutex_lock(&umem_odp->umem_mutex);
-		ib_umem_odp_unmap_dma_pages(umem_odp, ib_umem_start(umem_odp),
-					    ib_umem_end(umem_odp));
-		mutex_unlock(&umem_odp->umem_mutex);
-		mmu_interval_notifier_remove(&umem_odp->notifier);
-		hmm_dma_map_free(dev->dma_device, &umem_odp->map);
-	}
+	mutex_lock(&umem_odp->umem_mutex);
+	ib_umem_odp_unmap_dma_pages(umem_odp, ib_umem_start(umem_odp),
+				    ib_umem_end(umem_odp));
+	mutex_unlock(&umem_odp->umem_mutex);
+	mmu_interval_notifier_remove(&umem_odp->notifier);
+	hmm_dma_map_free(dev->dma_device, &umem_odp->map);
+}
+
+void ib_umem_odp_release(struct ib_umem_odp *umem_odp)
+{
+	if (!umem_odp->is_implicit_odp)
+		ib_umem_odp_free(umem_odp);
+
 	put_pid(umem_odp->tgid);
 	kfree(umem_odp);
 }

From 8536666a52833da326ab17d43ee2cd6c66751716 Mon Sep 17 00:00:00 2001
From: Colin Ian King <colin.i.king@gmail.com>
Date: Wed, 7 May 2025 14:18:34 +0100
Subject: [PATCH 40/57] RDMA/siw: replace redundant ternary operator with just
 rv

The use of the ternary operator on rv is redundant, rv is
either the initialized value of 0 or a negative error return
code, so it can never be greater than zero, and hence the
zero assignment in ternary operator is redundant. Just return
rv instead.

Signed-off-by: Colin Ian King <colin.i.king@gmail.com>
Link: https://patch.msgid.link/20250507131834.253823-1-colin.i.king@gmail.com
Acked-by: Bernard Metzler <bmt@zurich.ibm.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/sw/siw/siw_verbs.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/infiniband/sw/siw/siw_verbs.c b/drivers/infiniband/sw/siw/siw_verbs.c
index 7ce0035c54fa..2b2a7b8e93b0 100644
--- a/drivers/infiniband/sw/siw/siw_verbs.c
+++ b/drivers/infiniband/sw/siw/siw_verbs.c
@@ -1102,7 +1102,7 @@ int siw_post_receive(struct ib_qp *base_qp, const struct ib_recv_wr *wr,
 		siw_dbg_qp(qp, "error %d\n", rv);
 		*bad_wr = wr;
 	}
-	return rv > 0 ? 0 : rv;
+	return rv;
 }
 
 int siw_destroy_cq(struct ib_cq *base_cq, struct ib_udata *udata)

From ced82fce77e93315239f54caebbc88e263078e31 Mon Sep 17 00:00:00 2001
From: Konstantin Taranov <kotaranov@microsoft.com>
Date: Wed, 7 May 2025 08:59:02 -0700
Subject: [PATCH 41/57] net: mana: Probe rdma device in mana driver

Initialize gdma device for rdma inside mana module.
For each gdma device, initialize an auxiliary ib device.

Signed-off-by: Konstantin Taranov <kotaranov@microsoft.com>
Link: https://patch.msgid.link/1746633545-17653-2-git-send-email-kotaranov@linux.microsoft.com
Reviewed-by: Long Li <longli@microsoft.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 .../net/ethernet/microsoft/mana/gdma_main.c   | 15 ++++++-
 drivers/net/ethernet/microsoft/mana/mana_en.c | 39 +++++++++++++++++--
 include/net/mana/mana.h                       |  3 ++
 3 files changed, 52 insertions(+), 5 deletions(-)

diff --git a/drivers/net/ethernet/microsoft/mana/gdma_main.c b/drivers/net/ethernet/microsoft/mana/gdma_main.c
index 8ee1aa3a7ec3..59e7814f0b70 100644
--- a/drivers/net/ethernet/microsoft/mana/gdma_main.c
+++ b/drivers/net/ethernet/microsoft/mana/gdma_main.c
@@ -1005,7 +1005,6 @@ int mana_gd_register_device(struct gdma_dev *gd)
 
 	return 0;
 }
-EXPORT_SYMBOL_NS(mana_gd_register_device, "NET_MANA");
 
 int mana_gd_deregister_device(struct gdma_dev *gd)
 {
@@ -1036,7 +1035,6 @@ int mana_gd_deregister_device(struct gdma_dev *gd)
 
 	return err;
 }
-EXPORT_SYMBOL_NS(mana_gd_deregister_device, "NET_MANA");
 
 u32 mana_gd_wq_avail_space(struct gdma_queue *wq)
 {
@@ -1579,8 +1577,14 @@ static int mana_gd_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
 	if (err)
 		goto cleanup_gd;
 
+	err = mana_rdma_probe(&gc->mana_ib);
+	if (err)
+		goto cleanup_mana;
+
 	return 0;
 
+cleanup_mana:
+	mana_remove(&gc->mana, false);
 cleanup_gd:
 	mana_gd_cleanup(pdev);
 unmap_bar:
@@ -1608,6 +1612,7 @@ static void mana_gd_remove(struct pci_dev *pdev)
 {
 	struct gdma_context *gc = pci_get_drvdata(pdev);
 
+	mana_rdma_remove(&gc->mana_ib);
 	mana_remove(&gc->mana, false);
 
 	mana_gd_cleanup(pdev);
@@ -1631,6 +1636,7 @@ static int mana_gd_suspend(struct pci_dev *pdev, pm_message_t state)
 {
 	struct gdma_context *gc = pci_get_drvdata(pdev);
 
+	mana_rdma_remove(&gc->mana_ib);
 	mana_remove(&gc->mana, true);
 
 	mana_gd_cleanup(pdev);
@@ -1655,6 +1661,10 @@ static int mana_gd_resume(struct pci_dev *pdev)
 	if (err)
 		return err;
 
+	err = mana_rdma_probe(&gc->mana_ib);
+	if (err)
+		return err;
+
 	return 0;
 }
 
@@ -1665,6 +1675,7 @@ static void mana_gd_shutdown(struct pci_dev *pdev)
 
 	dev_info(&pdev->dev, "Shutdown was called\n");
 
+	mana_rdma_remove(&gc->mana_ib);
 	mana_remove(&gc->mana, true);
 
 	mana_gd_cleanup(pdev);
diff --git a/drivers/net/ethernet/microsoft/mana/mana_en.c b/drivers/net/ethernet/microsoft/mana/mana_en.c
index 2bac6be8f6a0..fa83f935ffe1 100644
--- a/drivers/net/ethernet/microsoft/mana/mana_en.c
+++ b/drivers/net/ethernet/microsoft/mana/mana_en.c
@@ -2945,7 +2945,7 @@ static void remove_adev(struct gdma_dev *gd)
 	gd->adev = NULL;
 }
 
-static int add_adev(struct gdma_dev *gd)
+static int add_adev(struct gdma_dev *gd, const char *name)
 {
 	struct auxiliary_device *adev;
 	struct mana_adev *madev;
@@ -2961,7 +2961,7 @@ static int add_adev(struct gdma_dev *gd)
 		goto idx_fail;
 	adev->id = ret;
 
-	adev->name = "rdma";
+	adev->name = name;
 	adev->dev.parent = gd->gdma_context->dev;
 	adev->dev.release = adev_release;
 	madev->mdev = gd;
@@ -3077,7 +3077,7 @@ int mana_probe(struct gdma_dev *gd, bool resuming)
 		}
 	}
 
-	err = add_adev(gd);
+	err = add_adev(gd, "eth");
 out:
 	if (err) {
 		mana_remove(gd, false);
@@ -3151,6 +3151,39 @@ out:
 	dev_dbg(dev, "%s succeeded\n", __func__);
 }
 
+int mana_rdma_probe(struct gdma_dev *gd)
+{
+	int err = 0;
+
+	if (gd->dev_id.type != GDMA_DEVICE_MANA_IB) {
+		/* RDMA device is not detected on pci */
+		return err;
+	}
+
+	err = mana_gd_register_device(gd);
+	if (err)
+		return err;
+
+	err = add_adev(gd, "rdma");
+	if (err)
+		mana_gd_deregister_device(gd);
+
+	return err;
+}
+
+void mana_rdma_remove(struct gdma_dev *gd)
+{
+	if (gd->dev_id.type != GDMA_DEVICE_MANA_IB) {
+		/* RDMA device is not detected on pci */
+		return;
+	}
+
+	if (gd->adev)
+		remove_adev(gd);
+
+	mana_gd_deregister_device(gd);
+}
+
 struct net_device *mana_get_primary_netdev(struct mana_context *ac,
 					   u32 port_index,
 					   netdevice_tracker *tracker)
diff --git a/include/net/mana/mana.h b/include/net/mana/mana.h
index 0f78065de8fe..5857efc885a6 100644
--- a/include/net/mana/mana.h
+++ b/include/net/mana/mana.h
@@ -488,6 +488,9 @@ int mana_detach(struct net_device *ndev, bool from_close);
 int mana_probe(struct gdma_dev *gd, bool resuming);
 void mana_remove(struct gdma_dev *gd, bool suspending);
 
+int mana_rdma_probe(struct gdma_dev *gd);
+void mana_rdma_remove(struct gdma_dev *gd);
+
 void mana_xdp_tx(struct sk_buff *skb, struct net_device *ndev);
 int mana_xdp_xmit(struct net_device *ndev, int n, struct xdp_frame **frames,
 		  u32 flags);

From c390828d4d7b453c21c6fc0787c714db82731093 Mon Sep 17 00:00:00 2001
From: Konstantin Taranov <kotaranov@microsoft.com>
Date: Wed, 7 May 2025 08:59:03 -0700
Subject: [PATCH 42/57] RDMA/mana_ib: Add support of mana_ib for RNIC and ETH
 nic

Allow mana_ib to be created over ethernet gdma device and
over rnic gdma device. The HW has two devices with different
capabilities and different use-cases. Initialize required
resources depending on the used gdma device.

Signed-off-by: Konstantin Taranov <kotaranov@microsoft.com>
Link: https://patch.msgid.link/1746633545-17653-3-git-send-email-kotaranov@linux.microsoft.com
Reviewed-by: Long Li <longli@microsoft.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/hw/mana/device.c  | 174 +++++++++++++--------------
 drivers/infiniband/hw/mana/main.c    |  55 ++++++++-
 drivers/infiniband/hw/mana/mana_ib.h |   6 +
 3 files changed, 138 insertions(+), 97 deletions(-)

diff --git a/drivers/infiniband/hw/mana/device.c b/drivers/infiniband/hw/mana/device.c
index b31089320aa5..165c0a1e67d1 100644
--- a/drivers/infiniband/hw/mana/device.c
+++ b/drivers/infiniband/hw/mana/device.c
@@ -101,103 +101,95 @@ static int mana_ib_probe(struct auxiliary_device *adev,
 			 const struct auxiliary_device_id *id)
 {
 	struct mana_adev *madev = container_of(adev, struct mana_adev, adev);
+	struct gdma_context *gc = madev->mdev->gdma_context;
+	struct mana_context *mc = gc->mana.driver_data;
 	struct gdma_dev *mdev = madev->mdev;
 	struct net_device *ndev;
-	struct mana_context *mc;
 	struct mana_ib_dev *dev;
 	u8 mac_addr[ETH_ALEN];
 	int ret;
 
-	mc = mdev->driver_data;
-
 	dev = ib_alloc_device(mana_ib_dev, ib_dev);
 	if (!dev)
 		return -ENOMEM;
 
 	ib_set_device_ops(&dev->ib_dev, &mana_ib_dev_ops);
-
-	dev->ib_dev.phys_port_cnt = mc->num_ports;
-
-	ibdev_dbg(&dev->ib_dev, "mdev=%p id=%d num_ports=%d\n", mdev,
-		  mdev->dev_id.as_uint32, dev->ib_dev.phys_port_cnt);
-
 	dev->ib_dev.node_type = RDMA_NODE_IB_CA;
-
-	/*
-	 * num_comp_vectors needs to set to the max MSIX index
-	 * when interrupts and event queues are implemented
-	 */
-	dev->ib_dev.num_comp_vectors = mdev->gdma_context->max_num_queues;
-	dev->ib_dev.dev.parent = mdev->gdma_context->dev;
-
-	ndev = mana_get_primary_netdev(mc, 0, &dev->dev_tracker);
-	if (!ndev) {
-		ret = -ENODEV;
-		ibdev_err(&dev->ib_dev, "Failed to get netdev for IB port 1");
-		goto free_ib_device;
-	}
-	ether_addr_copy(mac_addr, ndev->dev_addr);
-	addrconf_addr_eui48((u8 *)&dev->ib_dev.node_guid, ndev->dev_addr);
-	ret = ib_device_set_netdev(&dev->ib_dev, ndev, 1);
-	/* mana_get_primary_netdev() returns ndev with refcount held */
-	netdev_put(ndev, &dev->dev_tracker);
-	if (ret) {
-		ibdev_err(&dev->ib_dev, "Failed to set ib netdev, ret %d", ret);
-		goto free_ib_device;
-	}
-
-	ret = mana_gd_register_device(&mdev->gdma_context->mana_ib);
-	if (ret) {
-		ibdev_err(&dev->ib_dev, "Failed to register device, ret %d",
-			  ret);
-		goto free_ib_device;
-	}
-	dev->gdma_dev = &mdev->gdma_context->mana_ib;
-
-	dev->nb.notifier_call = mana_ib_netdev_event;
-	ret = register_netdevice_notifier(&dev->nb);
-	if (ret) {
-		ibdev_err(&dev->ib_dev, "Failed to register net notifier, %d",
-			  ret);
-		goto deregister_device;
-	}
-
-	ret = mana_ib_gd_query_adapter_caps(dev);
-	if (ret) {
-		ibdev_err(&dev->ib_dev, "Failed to query device caps, ret %d",
-			  ret);
-		goto deregister_net_notifier;
-	}
-
-	ib_set_device_ops(&dev->ib_dev, &mana_ib_stats_ops);
-
-	ret = mana_ib_create_eqs(dev);
-	if (ret) {
-		ibdev_err(&dev->ib_dev, "Failed to create EQs, ret %d", ret);
-		goto deregister_net_notifier;
-	}
-
-	ret = mana_ib_gd_create_rnic_adapter(dev);
-	if (ret)
-		goto destroy_eqs;
-
+	dev->ib_dev.num_comp_vectors = gc->max_num_queues;
+	dev->ib_dev.dev.parent = gc->dev;
+	dev->gdma_dev = mdev;
 	xa_init_flags(&dev->qp_table_wq, XA_FLAGS_LOCK_IRQ);
-	ret = mana_ib_gd_config_mac(dev, ADDR_OP_ADD, mac_addr);
-	if (ret) {
-		ibdev_err(&dev->ib_dev, "Failed to add Mac address, ret %d",
-			  ret);
-		goto destroy_rnic;
+
+	if (mana_ib_is_rnic(dev)) {
+		dev->ib_dev.phys_port_cnt = 1;
+		ndev = mana_get_primary_netdev(mc, 0, &dev->dev_tracker);
+		if (!ndev) {
+			ret = -ENODEV;
+			ibdev_err(&dev->ib_dev, "Failed to get netdev for IB port 1");
+			goto free_ib_device;
+		}
+		ether_addr_copy(mac_addr, ndev->dev_addr);
+		addrconf_addr_eui48((u8 *)&dev->ib_dev.node_guid, ndev->dev_addr);
+		ret = ib_device_set_netdev(&dev->ib_dev, ndev, 1);
+		/* mana_get_primary_netdev() returns ndev with refcount held */
+		netdev_put(ndev, &dev->dev_tracker);
+		if (ret) {
+			ibdev_err(&dev->ib_dev, "Failed to set ib netdev, ret %d", ret);
+			goto free_ib_device;
+		}
+
+		dev->nb.notifier_call = mana_ib_netdev_event;
+		ret = register_netdevice_notifier(&dev->nb);
+		if (ret) {
+			ibdev_err(&dev->ib_dev, "Failed to register net notifier, %d",
+				  ret);
+			goto free_ib_device;
+		}
+
+		ret = mana_ib_gd_query_adapter_caps(dev);
+		if (ret) {
+			ibdev_err(&dev->ib_dev, "Failed to query device caps, ret %d", ret);
+			goto deregister_net_notifier;
+		}
+
+		ib_set_device_ops(&dev->ib_dev, &mana_ib_stats_ops);
+
+		ret = mana_ib_create_eqs(dev);
+		if (ret) {
+			ibdev_err(&dev->ib_dev, "Failed to create EQs, ret %d", ret);
+			goto deregister_net_notifier;
+		}
+
+		ret = mana_ib_gd_create_rnic_adapter(dev);
+		if (ret)
+			goto destroy_eqs;
+
+		ret = mana_ib_gd_config_mac(dev, ADDR_OP_ADD, mac_addr);
+		if (ret) {
+			ibdev_err(&dev->ib_dev, "Failed to add Mac address, ret %d", ret);
+			goto destroy_rnic;
+		}
+	} else {
+		dev->ib_dev.phys_port_cnt = mc->num_ports;
+		ret = mana_eth_query_adapter_caps(dev);
+		if (ret) {
+			ibdev_err(&dev->ib_dev, "Failed to query ETH device caps, ret %d", ret);
+			goto free_ib_device;
+		}
 	}
 
-	dev->av_pool = dma_pool_create("mana_ib_av", mdev->gdma_context->dev,
-				       MANA_AV_BUFFER_SIZE, MANA_AV_BUFFER_SIZE, 0);
+	dev->av_pool = dma_pool_create("mana_ib_av", gc->dev, MANA_AV_BUFFER_SIZE,
+				       MANA_AV_BUFFER_SIZE, 0);
 	if (!dev->av_pool) {
 		ret = -ENOMEM;
 		goto destroy_rnic;
 	}
 
-	ret = ib_register_device(&dev->ib_dev, "mana_%d",
-				 mdev->gdma_context->dev);
+	ibdev_dbg(&dev->ib_dev, "mdev=%p id=%d num_ports=%d\n", mdev,
+		  mdev->dev_id.as_uint32, dev->ib_dev.phys_port_cnt);
+
+	ret = ib_register_device(&dev->ib_dev, mana_ib_is_rnic(dev) ? "mana_%d" : "manae_%d",
+				 gc->dev);
 	if (ret)
 		goto deallocate_pool;
 
@@ -208,15 +200,16 @@ static int mana_ib_probe(struct auxiliary_device *adev,
 deallocate_pool:
 	dma_pool_destroy(dev->av_pool);
 destroy_rnic:
-	xa_destroy(&dev->qp_table_wq);
-	mana_ib_gd_destroy_rnic_adapter(dev);
+	if (mana_ib_is_rnic(dev))
+		mana_ib_gd_destroy_rnic_adapter(dev);
 destroy_eqs:
-	mana_ib_destroy_eqs(dev);
+	if (mana_ib_is_rnic(dev))
+		mana_ib_destroy_eqs(dev);
 deregister_net_notifier:
-	unregister_netdevice_notifier(&dev->nb);
-deregister_device:
-	mana_gd_deregister_device(dev->gdma_dev);
+	if (mana_ib_is_rnic(dev))
+		unregister_netdevice_notifier(&dev->nb);
 free_ib_device:
+	xa_destroy(&dev->qp_table_wq);
 	ib_dealloc_device(&dev->ib_dev);
 	return ret;
 }
@@ -227,25 +220,24 @@ static void mana_ib_remove(struct auxiliary_device *adev)
 
 	ib_unregister_device(&dev->ib_dev);
 	dma_pool_destroy(dev->av_pool);
+	if (mana_ib_is_rnic(dev)) {
+		mana_ib_gd_destroy_rnic_adapter(dev);
+		mana_ib_destroy_eqs(dev);
+		unregister_netdevice_notifier(&dev->nb);
+	}
 	xa_destroy(&dev->qp_table_wq);
-	mana_ib_gd_destroy_rnic_adapter(dev);
-	mana_ib_destroy_eqs(dev);
-	unregister_netdevice_notifier(&dev->nb);
-	mana_gd_deregister_device(dev->gdma_dev);
 	ib_dealloc_device(&dev->ib_dev);
 }
 
 static const struct auxiliary_device_id mana_id_table[] = {
-	{
-		.name = "mana.rdma",
-	},
+	{ .name = "mana.rdma", },
+	{ .name = "mana.eth", },
 	{},
 };
 
 MODULE_DEVICE_TABLE(auxiliary, mana_id_table);
 
 static struct auxiliary_driver mana_driver = {
-	.name = "rdma",
 	.probe = mana_ib_probe,
 	.remove = mana_ib_remove,
 	.id_table = mana_id_table,
diff --git a/drivers/infiniband/hw/mana/main.c b/drivers/infiniband/hw/mana/main.c
index bb0f685babe6..3837e30ed819 100644
--- a/drivers/infiniband/hw/mana/main.c
+++ b/drivers/infiniband/hw/mana/main.c
@@ -4,6 +4,7 @@
  */
 
 #include "mana_ib.h"
+#include "linux/pci.h"
 
 void mana_ib_uncfg_vport(struct mana_ib_dev *dev, struct mana_ib_pd *pd,
 			 u32 port)
@@ -551,6 +552,7 @@ int mana_ib_mmap(struct ib_ucontext *ibcontext, struct vm_area_struct *vma)
 int mana_ib_get_port_immutable(struct ib_device *ibdev, u32 port_num,
 			       struct ib_port_immutable *immutable)
 {
+	struct mana_ib_dev *dev = container_of(ibdev, struct mana_ib_dev, ib_dev);
 	struct ib_port_attr attr;
 	int err;
 
@@ -560,10 +562,12 @@ int mana_ib_get_port_immutable(struct ib_device *ibdev, u32 port_num,
 
 	immutable->pkey_tbl_len = attr.pkey_tbl_len;
 	immutable->gid_tbl_len = attr.gid_tbl_len;
-	immutable->core_cap_flags = RDMA_CORE_PORT_RAW_PACKET;
-	if (port_num == 1) {
-		immutable->core_cap_flags |= RDMA_CORE_PORT_IBA_ROCE_UDP_ENCAP;
+
+	if (mana_ib_is_rnic(dev)) {
+		immutable->core_cap_flags = RDMA_CORE_PORT_IBA_ROCE_UDP_ENCAP;
 		immutable->max_mad_size = IB_MGMT_MAD_SIZE;
+	} else {
+		immutable->core_cap_flags = RDMA_CORE_PORT_RAW_PACKET;
 	}
 
 	return 0;
@@ -572,10 +576,12 @@ int mana_ib_get_port_immutable(struct ib_device *ibdev, u32 port_num,
 int mana_ib_query_device(struct ib_device *ibdev, struct ib_device_attr *props,
 			 struct ib_udata *uhw)
 {
-	struct mana_ib_dev *dev = container_of(ibdev,
-			struct mana_ib_dev, ib_dev);
+	struct mana_ib_dev *dev = container_of(ibdev, struct mana_ib_dev, ib_dev);
+	struct pci_dev *pdev = to_pci_dev(mdev_to_gc(dev)->dev);
 
 	memset(props, 0, sizeof(*props));
+	props->vendor_id = pdev->vendor;
+	props->vendor_part_id = dev->gdma_dev->dev_id.type;
 	props->max_mr_size = MANA_IB_MAX_MR_SIZE;
 	props->page_size_cap = dev->adapter_caps.page_size_cap;
 	props->max_qp = dev->adapter_caps.max_qp_count;
@@ -596,6 +602,8 @@ int mana_ib_query_device(struct ib_device *ibdev, struct ib_device_attr *props,
 	props->max_ah = INT_MAX;
 	props->max_pkeys = 1;
 	props->local_ca_ack_delay = MANA_CA_ACK_DELAY;
+	if (!mana_ib_is_rnic(dev))
+		props->raw_packet_caps = IB_RAW_PACKET_CAP_IP_CSUM;
 
 	return 0;
 }
@@ -603,6 +611,7 @@ int mana_ib_query_device(struct ib_device *ibdev, struct ib_device_attr *props,
 int mana_ib_query_port(struct ib_device *ibdev, u32 port,
 		       struct ib_port_attr *props)
 {
+	struct mana_ib_dev *dev = container_of(ibdev, struct mana_ib_dev, ib_dev);
 	struct net_device *ndev = mana_ib_get_netdev(ibdev, port);
 
 	if (!ndev)
@@ -623,7 +632,7 @@ int mana_ib_query_port(struct ib_device *ibdev, u32 port,
 	props->active_width = IB_WIDTH_4X;
 	props->active_speed = IB_SPEED_EDR;
 	props->pkey_tbl_len = 1;
-	if (port == 1) {
+	if (mana_ib_is_rnic(dev)) {
 		props->gid_tbl_len = 16;
 		props->port_cap_flags = IB_PORT_CM_SUP;
 		props->ip_gids = true;
@@ -703,6 +712,37 @@ int mana_ib_gd_query_adapter_caps(struct mana_ib_dev *dev)
 	return 0;
 }
 
+int mana_eth_query_adapter_caps(struct mana_ib_dev *dev)
+{
+	struct mana_ib_adapter_caps *caps = &dev->adapter_caps;
+	struct gdma_query_max_resources_resp resp = {};
+	struct gdma_general_req req = {};
+	int err;
+
+	mana_gd_init_req_hdr(&req.hdr, GDMA_QUERY_MAX_RESOURCES,
+			     sizeof(req), sizeof(resp));
+
+	err = mana_gd_send_request(mdev_to_gc(dev), sizeof(req), &req, sizeof(resp), &resp);
+	if (err) {
+		ibdev_err(&dev->ib_dev,
+			  "Failed to query adapter caps err %d", err);
+		return err;
+	}
+
+	caps->max_qp_count = min_t(u32, resp.max_sq, resp.max_rq);
+	caps->max_cq_count = resp.max_cq;
+	caps->max_mr_count = resp.max_mst;
+	caps->max_pd_count = 0x6000;
+	caps->max_qp_wr = min_t(u32,
+				0x100000 / GDMA_MAX_SQE_SIZE,
+				0x100000 / GDMA_MAX_RQE_SIZE);
+	caps->max_send_sge_count = 30;
+	caps->max_recv_sge_count = 15;
+	caps->page_size_cap = PAGE_SZ_BM;
+
+	return 0;
+}
+
 static void
 mana_ib_event_handler(void *ctx, struct gdma_queue *q, struct gdma_event *event)
 {
@@ -921,6 +961,9 @@ int mana_ib_gd_create_cq(struct mana_ib_dev *mdev, struct mana_ib_cq *cq, u32 do
 	struct mana_rnic_create_cq_req req = {};
 	int err;
 
+	if (!mdev->eqs)
+		return -EINVAL;
+
 	mana_gd_init_req_hdr(&req.hdr, MANA_IB_CREATE_CQ, sizeof(req), sizeof(resp));
 	req.hdr.dev_id = gc->mana_ib.dev_id;
 	req.adapter = mdev->adapter_handle;
diff --git a/drivers/infiniband/hw/mana/mana_ib.h b/drivers/infiniband/hw/mana/mana_ib.h
index f0dbd90b8600..42bebd6cd4f7 100644
--- a/drivers/infiniband/hw/mana/mana_ib.h
+++ b/drivers/infiniband/hw/mana/mana_ib.h
@@ -544,6 +544,11 @@ static inline void mana_put_qp_ref(struct mana_ib_qp *qp)
 		complete(&qp->free);
 }
 
+static inline bool mana_ib_is_rnic(struct mana_ib_dev *mdev)
+{
+	return mdev->gdma_dev->dev_id.type == GDMA_DEVICE_MANA_IB;
+}
+
 static inline struct net_device *mana_ib_get_netdev(struct ib_device *ibdev, u32 port)
 {
 	struct mana_ib_dev *mdev = container_of(ibdev, struct mana_ib_dev, ib_dev);
@@ -643,6 +648,7 @@ int mana_ib_query_gid(struct ib_device *ibdev, u32 port, int index,
 void mana_ib_disassociate_ucontext(struct ib_ucontext *ibcontext);
 
 int mana_ib_gd_query_adapter_caps(struct mana_ib_dev *mdev);
+int mana_eth_query_adapter_caps(struct mana_ib_dev *mdev);
 
 int mana_ib_create_eqs(struct mana_ib_dev *mdev);
 

From d4293f96ce0b6d27d9ca13fdcd7caa135f3d2732 Mon Sep 17 00:00:00 2001
From: Konstantin Taranov <kotaranov@microsoft.com>
Date: Wed, 7 May 2025 08:59:04 -0700
Subject: [PATCH 43/57] RDMA/mana_ib: unify mana_ib functions to support any
 gdma device

Use the installed gdma_device instead of hard-coded device
in requests to the HW.

Signed-off-by: Konstantin Taranov <kotaranov@microsoft.com>
Link: https://patch.msgid.link/1746633545-17653-4-git-send-email-kotaranov@linux.microsoft.com
Reviewed-by: Long Li <longli@microsoft.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/hw/mana/cq.c   |  4 +---
 drivers/infiniband/hw/mana/main.c | 27 +++++++++++++--------------
 drivers/infiniband/hw/mana/qp.c   |  5 ++---
 3 files changed, 16 insertions(+), 20 deletions(-)

diff --git a/drivers/infiniband/hw/mana/cq.c b/drivers/infiniband/hw/mana/cq.c
index 0fc4e2679218..28e154bbb50f 100644
--- a/drivers/infiniband/hw/mana/cq.c
+++ b/drivers/infiniband/hw/mana/cq.c
@@ -15,14 +15,12 @@ int mana_ib_create_cq(struct ib_cq *ibcq, const struct ib_cq_init_attr *attr,
 	struct ib_device *ibdev = ibcq->device;
 	struct mana_ib_create_cq ucmd = {};
 	struct mana_ib_dev *mdev;
-	struct gdma_context *gc;
 	bool is_rnic_cq;
 	u32 doorbell;
 	u32 buf_size;
 	int err;
 
 	mdev = container_of(ibdev, struct mana_ib_dev, ib_dev);
-	gc = mdev_to_gc(mdev);
 
 	cq->comp_vector = attr->comp_vector % ibdev->num_comp_vectors;
 	cq->cq_handle = INVALID_MANA_HANDLE;
@@ -65,7 +63,7 @@ int mana_ib_create_cq(struct ib_cq *ibcq, const struct ib_cq_init_attr *attr,
 			ibdev_dbg(ibdev, "Failed to create kernel queue for create cq, %d\n", err);
 			return err;
 		}
-		doorbell = gc->mana_ib.doorbell;
+		doorbell = mdev->gdma_dev->doorbell;
 	}
 
 	if (is_rnic_cq) {
diff --git a/drivers/infiniband/hw/mana/main.c b/drivers/infiniband/hw/mana/main.c
index 3837e30ed819..41a24a186f9d 100644
--- a/drivers/infiniband/hw/mana/main.c
+++ b/drivers/infiniband/hw/mana/main.c
@@ -244,7 +244,6 @@ void mana_ib_dealloc_ucontext(struct ib_ucontext *ibcontext)
 int mana_ib_create_kernel_queue(struct mana_ib_dev *mdev, u32 size, enum gdma_queue_type type,
 				struct mana_ib_queue *queue)
 {
-	struct gdma_context *gc = mdev_to_gc(mdev);
 	struct gdma_queue_spec spec = {};
 	int err;
 
@@ -253,7 +252,7 @@ int mana_ib_create_kernel_queue(struct mana_ib_dev *mdev, u32 size, enum gdma_qu
 	spec.type = type;
 	spec.monitor_avl_buf = false;
 	spec.queue_size = size;
-	err = mana_gd_create_mana_wq_cq(&gc->mana_ib, &spec, &queue->kmem);
+	err = mana_gd_create_mana_wq_cq(mdev->gdma_dev, &spec, &queue->kmem);
 	if (err)
 		return err;
 	/* take ownership into mana_ib from mana */
@@ -784,7 +783,7 @@ int mana_ib_create_eqs(struct mana_ib_dev *mdev)
 	spec.eq.log2_throttle_limit = LOG2_EQ_THROTTLE;
 	spec.eq.msix_index = 0;
 
-	err = mana_gd_create_mana_eq(&gc->mana_ib, &spec, &mdev->fatal_err_eq);
+	err = mana_gd_create_mana_eq(mdev->gdma_dev, &spec, &mdev->fatal_err_eq);
 	if (err)
 		return err;
 
@@ -835,7 +834,7 @@ int mana_ib_gd_create_rnic_adapter(struct mana_ib_dev *mdev)
 
 	mana_gd_init_req_hdr(&req.hdr, MANA_IB_CREATE_ADAPTER, sizeof(req), sizeof(resp));
 	req.hdr.req.msg_version = GDMA_MESSAGE_V2;
-	req.hdr.dev_id = gc->mana_ib.dev_id;
+	req.hdr.dev_id = mdev->gdma_dev->dev_id;
 	req.notify_eq_id = mdev->fatal_err_eq->id;
 
 	if (mdev->adapter_caps.feature_flags & MANA_IB_FEATURE_CLIENT_ERROR_CQE_SUPPORT)
@@ -860,7 +859,7 @@ int mana_ib_gd_destroy_rnic_adapter(struct mana_ib_dev *mdev)
 
 	gc = mdev_to_gc(mdev);
 	mana_gd_init_req_hdr(&req.hdr, MANA_IB_DESTROY_ADAPTER, sizeof(req), sizeof(resp));
-	req.hdr.dev_id = gc->mana_ib.dev_id;
+	req.hdr.dev_id = mdev->gdma_dev->dev_id;
 	req.adapter = mdev->adapter_handle;
 
 	err = mana_gd_send_request(gc, sizeof(req), &req, sizeof(resp), &resp);
@@ -887,7 +886,7 @@ int mana_ib_gd_add_gid(const struct ib_gid_attr *attr, void **context)
 	}
 
 	mana_gd_init_req_hdr(&req.hdr, MANA_IB_CONFIG_IP_ADDR, sizeof(req), sizeof(resp));
-	req.hdr.dev_id = gc->mana_ib.dev_id;
+	req.hdr.dev_id = mdev->gdma_dev->dev_id;
 	req.adapter = mdev->adapter_handle;
 	req.op = ADDR_OP_ADD;
 	req.sgid_type = (ntype == RDMA_NETWORK_IPV6) ? SGID_TYPE_IPV6 : SGID_TYPE_IPV4;
@@ -917,7 +916,7 @@ int mana_ib_gd_del_gid(const struct ib_gid_attr *attr, void **context)
 	}
 
 	mana_gd_init_req_hdr(&req.hdr, MANA_IB_CONFIG_IP_ADDR, sizeof(req), sizeof(resp));
-	req.hdr.dev_id = gc->mana_ib.dev_id;
+	req.hdr.dev_id = mdev->gdma_dev->dev_id;
 	req.adapter = mdev->adapter_handle;
 	req.op = ADDR_OP_REMOVE;
 	req.sgid_type = (ntype == RDMA_NETWORK_IPV6) ? SGID_TYPE_IPV6 : SGID_TYPE_IPV4;
@@ -940,7 +939,7 @@ int mana_ib_gd_config_mac(struct mana_ib_dev *mdev, enum mana_ib_addr_op op, u8
 	int err;
 
 	mana_gd_init_req_hdr(&req.hdr, MANA_IB_CONFIG_MAC_ADDR, sizeof(req), sizeof(resp));
-	req.hdr.dev_id = gc->mana_ib.dev_id;
+	req.hdr.dev_id = mdev->gdma_dev->dev_id;
 	req.adapter = mdev->adapter_handle;
 	req.op = op;
 	copy_in_reverse(req.mac_addr, mac, ETH_ALEN);
@@ -965,7 +964,7 @@ int mana_ib_gd_create_cq(struct mana_ib_dev *mdev, struct mana_ib_cq *cq, u32 do
 		return -EINVAL;
 
 	mana_gd_init_req_hdr(&req.hdr, MANA_IB_CREATE_CQ, sizeof(req), sizeof(resp));
-	req.hdr.dev_id = gc->mana_ib.dev_id;
+	req.hdr.dev_id = mdev->gdma_dev->dev_id;
 	req.adapter = mdev->adapter_handle;
 	req.gdma_region = cq->queue.gdma_region;
 	req.eq_id = mdev->eqs[cq->comp_vector]->id;
@@ -997,7 +996,7 @@ int mana_ib_gd_destroy_cq(struct mana_ib_dev *mdev, struct mana_ib_cq *cq)
 		return 0;
 
 	mana_gd_init_req_hdr(&req.hdr, MANA_IB_DESTROY_CQ, sizeof(req), sizeof(resp));
-	req.hdr.dev_id = gc->mana_ib.dev_id;
+	req.hdr.dev_id = mdev->gdma_dev->dev_id;
 	req.adapter = mdev->adapter_handle;
 	req.cq_handle = cq->cq_handle;
 
@@ -1023,7 +1022,7 @@ int mana_ib_gd_create_rc_qp(struct mana_ib_dev *mdev, struct mana_ib_qp *qp,
 	int err, i;
 
 	mana_gd_init_req_hdr(&req.hdr, MANA_IB_CREATE_RC_QP, sizeof(req), sizeof(resp));
-	req.hdr.dev_id = gc->mana_ib.dev_id;
+	req.hdr.dev_id = mdev->gdma_dev->dev_id;
 	req.adapter = mdev->adapter_handle;
 	req.pd_handle = pd->pd_handle;
 	req.send_cq_handle = send_cq->cq_handle;
@@ -1059,7 +1058,7 @@ int mana_ib_gd_destroy_rc_qp(struct mana_ib_dev *mdev, struct mana_ib_qp *qp)
 	int err;
 
 	mana_gd_init_req_hdr(&req.hdr, MANA_IB_DESTROY_RC_QP, sizeof(req), sizeof(resp));
-	req.hdr.dev_id = gc->mana_ib.dev_id;
+	req.hdr.dev_id = mdev->gdma_dev->dev_id;
 	req.adapter = mdev->adapter_handle;
 	req.rc_qp_handle = qp->qp_handle;
 	err = mana_gd_send_request(gc, sizeof(req), &req, sizeof(resp), &resp);
@@ -1082,7 +1081,7 @@ int mana_ib_gd_create_ud_qp(struct mana_ib_dev *mdev, struct mana_ib_qp *qp,
 	int err, i;
 
 	mana_gd_init_req_hdr(&req.hdr, MANA_IB_CREATE_UD_QP, sizeof(req), sizeof(resp));
-	req.hdr.dev_id = gc->mana_ib.dev_id;
+	req.hdr.dev_id = mdev->gdma_dev->dev_id;
 	req.adapter = mdev->adapter_handle;
 	req.pd_handle = pd->pd_handle;
 	req.send_cq_handle = send_cq->cq_handle;
@@ -1117,7 +1116,7 @@ int mana_ib_gd_destroy_ud_qp(struct mana_ib_dev *mdev, struct mana_ib_qp *qp)
 	int err;
 
 	mana_gd_init_req_hdr(&req.hdr, MANA_IB_DESTROY_UD_QP, sizeof(req), sizeof(resp));
-	req.hdr.dev_id = gc->mana_ib.dev_id;
+	req.hdr.dev_id = mdev->gdma_dev->dev_id;
 	req.adapter = mdev->adapter_handle;
 	req.qp_handle = qp->qp_handle;
 	err = mana_gd_send_request(gc, sizeof(req), &req, sizeof(resp), &resp);
diff --git a/drivers/infiniband/hw/mana/qp.c b/drivers/infiniband/hw/mana/qp.c
index c928af58f38b..14fd7d6c54a2 100644
--- a/drivers/infiniband/hw/mana/qp.c
+++ b/drivers/infiniband/hw/mana/qp.c
@@ -635,7 +635,6 @@ static int mana_ib_create_ud_qp(struct ib_qp *ibqp, struct ib_pd *ibpd,
 {
 	struct mana_ib_dev *mdev = container_of(ibpd->device, struct mana_ib_dev, ib_dev);
 	struct mana_ib_qp *qp = container_of(ibqp, struct mana_ib_qp, ibqp);
-	struct gdma_context *gc = mdev_to_gc(mdev);
 	u32 doorbell, queue_size;
 	int i, err;
 
@@ -654,7 +653,7 @@ static int mana_ib_create_ud_qp(struct ib_qp *ibqp, struct ib_pd *ibpd,
 			goto destroy_queues;
 		}
 	}
-	doorbell = gc->mana_ib.doorbell;
+	doorbell = mdev->gdma_dev->doorbell;
 
 	err = create_shadow_queue(&qp->shadow_rq, attr->cap.max_recv_wr,
 				  sizeof(struct ud_rq_shadow_wqe));
@@ -736,7 +735,7 @@ static int mana_ib_gd_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr,
 	int err;
 
 	mana_gd_init_req_hdr(&req.hdr, MANA_IB_SET_QP_STATE, sizeof(req), sizeof(resp));
-	req.hdr.dev_id = gc->mana_ib.dev_id;
+	req.hdr.dev_id = mdev->gdma_dev->dev_id;
 	req.adapter = mdev->adapter_handle;
 	req.qp_handle = qp->qp_handle;
 	req.qp_state = attr->qp_state;

From 505cc26bcae00699bacaee66cd50ede7a9cc89cb Mon Sep 17 00:00:00 2001
From: Shiraz Saleem <shirazsaleem@microsoft.com>
Date: Wed, 7 May 2025 08:59:05 -0700
Subject: [PATCH 44/57] net: mana: Add support for auxiliary device servicing
 events

Handle soc servicing events which require the rdma auxiliary device resources to
be cleaned up during a suspend, and re-initialized during a resume.

Signed-off-by: Shiraz Saleem <shirazsaleem@microsoft.com>
Signed-off-by: Konstantin Taranov <kotaranov@microsoft.com>
Link: https://patch.msgid.link/1746633545-17653-5-git-send-email-kotaranov@linux.microsoft.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 .../net/ethernet/microsoft/mana/gdma_main.c   | 11 ++-
 .../net/ethernet/microsoft/mana/hw_channel.c  | 19 +++++
 drivers/net/ethernet/microsoft/mana/mana_en.c | 69 +++++++++++++++++++
 include/net/mana/gdma.h                       | 19 +++++
 include/net/mana/hw_channel.h                 |  9 +++
 5 files changed, 126 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/microsoft/mana/gdma_main.c b/drivers/net/ethernet/microsoft/mana/gdma_main.c
index 59e7814f0b70..3504507477c6 100644
--- a/drivers/net/ethernet/microsoft/mana/gdma_main.c
+++ b/drivers/net/ethernet/microsoft/mana/gdma_main.c
@@ -391,6 +391,7 @@ static void mana_gd_process_eqe(struct gdma_queue *eq)
 	case GDMA_EQE_HWC_INIT_EQ_ID_DB:
 	case GDMA_EQE_HWC_INIT_DATA:
 	case GDMA_EQE_HWC_INIT_DONE:
+	case GDMA_EQE_HWC_SOC_SERVICE:
 	case GDMA_EQE_RNIC_QP_FATAL:
 		if (!eq->eq.callback)
 			break;
@@ -1468,10 +1469,14 @@ static int mana_gd_setup(struct pci_dev *pdev)
 	mana_gd_init_registers(pdev);
 	mana_smc_init(&gc->shm_channel, gc->dev, gc->shm_base);
 
+	gc->service_wq = alloc_ordered_workqueue("gdma_service_wq", 0);
+	if (!gc->service_wq)
+		return -ENOMEM;
+
 	err = mana_gd_setup_irqs(pdev);
 	if (err) {
 		dev_err(gc->dev, "Failed to setup IRQs: %d\n", err);
-		return err;
+		goto free_workqueue;
 	}
 
 	err = mana_hwc_create_channel(gc);
@@ -1497,6 +1502,8 @@ destroy_hwc:
 	mana_hwc_destroy_channel(gc);
 remove_irq:
 	mana_gd_remove_irqs(pdev);
+free_workqueue:
+	destroy_workqueue(gc->service_wq);
 	dev_err(&pdev->dev, "%s failed (error %d)\n", __func__, err);
 	return err;
 }
@@ -1508,6 +1515,8 @@ static void mana_gd_cleanup(struct pci_dev *pdev)
 	mana_hwc_destroy_channel(gc);
 
 	mana_gd_remove_irqs(pdev);
+
+	destroy_workqueue(gc->service_wq);
 	dev_dbg(&pdev->dev, "mana gdma cleanup successful\n");
 }
 
diff --git a/drivers/net/ethernet/microsoft/mana/hw_channel.c b/drivers/net/ethernet/microsoft/mana/hw_channel.c
index 1ba49602089b..a8c4d8db75a5 100644
--- a/drivers/net/ethernet/microsoft/mana/hw_channel.c
+++ b/drivers/net/ethernet/microsoft/mana/hw_channel.c
@@ -112,11 +112,13 @@ out:
 static void mana_hwc_init_event_handler(void *ctx, struct gdma_queue *q_self,
 					struct gdma_event *event)
 {
+	union hwc_init_soc_service_type service_data;
 	struct hw_channel_context *hwc = ctx;
 	struct gdma_dev *gd = hwc->gdma_dev;
 	union hwc_init_type_data type_data;
 	union hwc_init_eq_id_db eq_db;
 	u32 type, val;
+	int ret;
 
 	switch (event->type) {
 	case GDMA_EQE_HWC_INIT_EQ_ID_DB:
@@ -199,7 +201,24 @@ static void mana_hwc_init_event_handler(void *ctx, struct gdma_queue *q_self,
 		}
 
 		break;
+	case GDMA_EQE_HWC_SOC_SERVICE:
+		service_data.as_uint32 = event->details[0];
+		type = service_data.type;
 
+		switch (type) {
+		case GDMA_SERVICE_TYPE_RDMA_SUSPEND:
+		case GDMA_SERVICE_TYPE_RDMA_RESUME:
+			ret = mana_rdma_service_event(gd->gdma_context, type);
+			if (ret)
+				dev_err(hwc->dev, "Failed to schedule adev service event: %d\n",
+					ret);
+			break;
+		default:
+			dev_warn(hwc->dev, "Received unknown SOC service type %u\n", type);
+			break;
+		}
+
+		break;
 	default:
 		dev_warn(hwc->dev, "Received unknown gdma event %u\n", event->type);
 		/* Ignore unknown events, which should never happen. */
diff --git a/drivers/net/ethernet/microsoft/mana/mana_en.c b/drivers/net/ethernet/microsoft/mana/mana_en.c
index fa83f935ffe1..1758281b2a51 100644
--- a/drivers/net/ethernet/microsoft/mana/mana_en.c
+++ b/drivers/net/ethernet/microsoft/mana/mana_en.c
@@ -2993,6 +2993,70 @@ idx_fail:
 	return ret;
 }
 
+static void mana_rdma_service_handle(struct work_struct *work)
+{
+	struct mana_service_work *serv_work =
+		container_of(work, struct mana_service_work, work);
+	struct gdma_dev *gd = serv_work->gdma_dev;
+	struct device *dev = gd->gdma_context->dev;
+	int ret;
+
+	if (READ_ONCE(gd->rdma_teardown))
+		goto out;
+
+	switch (serv_work->event) {
+	case GDMA_SERVICE_TYPE_RDMA_SUSPEND:
+		if (!gd->adev || gd->is_suspended)
+			break;
+
+		remove_adev(gd);
+		gd->is_suspended = true;
+		break;
+
+	case GDMA_SERVICE_TYPE_RDMA_RESUME:
+		if (!gd->is_suspended)
+			break;
+
+		ret = add_adev(gd, "rdma");
+		if (ret)
+			dev_err(dev, "Failed to add adev on resume: %d\n", ret);
+		else
+			gd->is_suspended = false;
+		break;
+
+	default:
+		dev_warn(dev, "unknown adev service event %u\n",
+			 serv_work->event);
+		break;
+	}
+
+out:
+	kfree(serv_work);
+}
+
+int mana_rdma_service_event(struct gdma_context *gc, enum gdma_service_type event)
+{
+	struct gdma_dev *gd = &gc->mana_ib;
+	struct mana_service_work *serv_work;
+
+	if (gd->dev_id.type != GDMA_DEVICE_MANA_IB) {
+		/* RDMA device is not detected on pci */
+		return 0;
+	}
+
+	serv_work = kzalloc(sizeof(*serv_work), GFP_ATOMIC);
+	if (!serv_work)
+		return -ENOMEM;
+
+	serv_work->event = event;
+	serv_work->gdma_dev = gd;
+
+	INIT_WORK(&serv_work->work, mana_rdma_service_handle);
+	queue_work(gc->service_wq, &serv_work->work);
+
+	return 0;
+}
+
 int mana_probe(struct gdma_dev *gd, bool resuming)
 {
 	struct gdma_context *gc = gd->gdma_context;
@@ -3173,11 +3237,16 @@ int mana_rdma_probe(struct gdma_dev *gd)
 
 void mana_rdma_remove(struct gdma_dev *gd)
 {
+	struct gdma_context *gc = gd->gdma_context;
+
 	if (gd->dev_id.type != GDMA_DEVICE_MANA_IB) {
 		/* RDMA device is not detected on pci */
 		return;
 	}
 
+	WRITE_ONCE(gd->rdma_teardown, true);
+	flush_workqueue(gc->service_wq);
+
 	if (gd->adev)
 		remove_adev(gd);
 
diff --git a/include/net/mana/gdma.h b/include/net/mana/gdma.h
index ffa9820f14ba..3ce56a816425 100644
--- a/include/net/mana/gdma.h
+++ b/include/net/mana/gdma.h
@@ -60,6 +60,7 @@ enum gdma_eqe_type {
 	GDMA_EQE_HWC_INIT_DONE		= 131,
 	GDMA_EQE_HWC_SOC_RECONFIG	= 132,
 	GDMA_EQE_HWC_SOC_RECONFIG_DATA	= 133,
+	GDMA_EQE_HWC_SOC_SERVICE	= 134,
 	GDMA_EQE_RNIC_QP_FATAL		= 176,
 };
 
@@ -70,6 +71,18 @@ enum {
 	GDMA_DEVICE_MANA_IB	= 3,
 };
 
+enum gdma_service_type {
+	GDMA_SERVICE_TYPE_NONE		= 0,
+	GDMA_SERVICE_TYPE_RDMA_SUSPEND	= 1,
+	GDMA_SERVICE_TYPE_RDMA_RESUME	= 2,
+};
+
+struct mana_service_work {
+	struct work_struct work;
+	struct gdma_dev *gdma_dev;
+	enum gdma_service_type event;
+};
+
 struct gdma_resource {
 	/* Protect the bitmap */
 	spinlock_t lock;
@@ -224,6 +237,8 @@ struct gdma_dev {
 	void *driver_data;
 
 	struct auxiliary_device *adev;
+	bool is_suspended;
+	bool rdma_teardown;
 };
 
 /* MANA_PAGE_SIZE is the DMA unit */
@@ -409,6 +424,8 @@ struct gdma_context {
 	struct gdma_dev		mana_ib;
 
 	u64 pf_cap_flags1;
+
+	struct workqueue_struct *service_wq;
 };
 
 static inline bool mana_gd_is_mana(struct gdma_dev *gd)
@@ -891,4 +908,6 @@ int mana_gd_destroy_dma_region(struct gdma_context *gc, u64 dma_region_handle);
 void mana_register_debugfs(void);
 void mana_unregister_debugfs(void);
 
+int mana_rdma_service_event(struct gdma_context *gc, enum gdma_service_type event);
+
 #endif /* _GDMA_H */
diff --git a/include/net/mana/hw_channel.h b/include/net/mana/hw_channel.h
index 158b125692c2..83cf93338eb3 100644
--- a/include/net/mana/hw_channel.h
+++ b/include/net/mana/hw_channel.h
@@ -49,6 +49,15 @@ union hwc_init_type_data {
 	};
 }; /* HW DATA */
 
+union hwc_init_soc_service_type {
+	u32 as_uint32;
+
+	struct {
+		u32 value	: 28;
+		u32 type	:  4;
+	};
+}; /* HW DATA */
+
 struct hwc_rx_oob {
 	u32 type	: 6;
 	u32 eom		: 1;

From 6883b680e703c6b2efddb4e7a8d891ce1803d06b Mon Sep 17 00:00:00 2001
From: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com>
Date: Sat, 10 May 2025 19:10:36 +0900
Subject: [PATCH 45/57] RDMA/iwcm: Fix use-after-free of work objects after
 cm_id destruction

The commit 59c68ac31e15 ("iw_cm: free cm_id resources on the last
deref") simplified cm_id resource management by freeing cm_id once all
references to the cm_id were removed. The references are removed either
upon completion of iw_cm event handlers or when the application destroys
the cm_id. This commit introduced the use-after-free condition where
cm_id_private object could still be in use by event handler works during
the destruction of cm_id. The commit aee2424246f9 ("RDMA/iwcm: Fix a
use-after-free related to destroying CM IDs") addressed this use-after-
free by flushing all pending works at the cm_id destruction.

However, still another use-after-free possibility remained. It happens
with the work objects allocated for each cm_id_priv within
alloc_work_entries() during cm_id creation, and subsequently freed in
dealloc_work_entries() once all references to the cm_id are removed.
If the cm_id's last reference is decremented in the event handler work,
the work object for the work itself gets removed, and causes the use-
after-free BUG below:

  BUG: KASAN: slab-use-after-free in __pwq_activate_work+0x1ff/0x250
  Read of size 8 at addr ffff88811f9cf800 by task kworker/u16:1/147091

  CPU: 2 UID: 0 PID: 147091 Comm: kworker/u16:1 Not tainted 6.15.0-rc2+ #27 PREEMPT(voluntary)
  Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.16.3-3.fc41 04/01/2014
  Workqueue:  0x0 (iw_cm_wq)
  Call Trace:
   <TASK>
   dump_stack_lvl+0x6a/0x90
   print_report+0x174/0x554
   ? __virt_addr_valid+0x208/0x430
   ? __pwq_activate_work+0x1ff/0x250
   kasan_report+0xae/0x170
   ? __pwq_activate_work+0x1ff/0x250
   __pwq_activate_work+0x1ff/0x250
   pwq_dec_nr_in_flight+0x8c5/0xfb0
   process_one_work+0xc11/0x1460
   ? __pfx_process_one_work+0x10/0x10
   ? assign_work+0x16c/0x240
   worker_thread+0x5ef/0xfd0
   ? __pfx_worker_thread+0x10/0x10
   kthread+0x3b0/0x770
   ? __pfx_kthread+0x10/0x10
   ? rcu_is_watching+0x11/0xb0
   ? _raw_spin_unlock_irq+0x24/0x50
   ? rcu_is_watching+0x11/0xb0
   ? __pfx_kthread+0x10/0x10
   ret_from_fork+0x30/0x70
   ? __pfx_kthread+0x10/0x10
   ret_from_fork_asm+0x1a/0x30
   </TASK>

  Allocated by task 147416:
   kasan_save_stack+0x2c/0x50
   kasan_save_track+0x10/0x30
   __kasan_kmalloc+0xa6/0xb0
   alloc_work_entries+0xa9/0x260 [iw_cm]
   iw_cm_connect+0x23/0x4a0 [iw_cm]
   rdma_connect_locked+0xbfd/0x1920 [rdma_cm]
   nvme_rdma_cm_handler+0x8e5/0x1b60 [nvme_rdma]
   cma_cm_event_handler+0xae/0x320 [rdma_cm]
   cma_work_handler+0x106/0x1b0 [rdma_cm]
   process_one_work+0x84f/0x1460
   worker_thread+0x5ef/0xfd0
   kthread+0x3b0/0x770
   ret_from_fork+0x30/0x70
   ret_from_fork_asm+0x1a/0x30

  Freed by task 147091:
   kasan_save_stack+0x2c/0x50
   kasan_save_track+0x10/0x30
   kasan_save_free_info+0x37/0x60
   __kasan_slab_free+0x4b/0x70
   kfree+0x13a/0x4b0
   dealloc_work_entries+0x125/0x1f0 [iw_cm]
   iwcm_deref_id+0x6f/0xa0 [iw_cm]
   cm_work_handler+0x136/0x1ba0 [iw_cm]
   process_one_work+0x84f/0x1460
   worker_thread+0x5ef/0xfd0
   kthread+0x3b0/0x770
   ret_from_fork+0x30/0x70
   ret_from_fork_asm+0x1a/0x30

  Last potentially related work creation:
   kasan_save_stack+0x2c/0x50
   kasan_record_aux_stack+0xa3/0xb0
   __queue_work+0x2ff/0x1390
   queue_work_on+0x67/0xc0
   cm_event_handler+0x46a/0x820 [iw_cm]
   siw_cm_upcall+0x330/0x650 [siw]
   siw_cm_work_handler+0x6b9/0x2b20 [siw]
   process_one_work+0x84f/0x1460
   worker_thread+0x5ef/0xfd0
   kthread+0x3b0/0x770
   ret_from_fork+0x30/0x70
   ret_from_fork_asm+0x1a/0x30

This BUG is reproducible by repeating the blktests test case nvme/061
for the rdma transport and the siw driver.

To avoid the use-after-free of cm_id_private work objects, ensure that
the last reference to the cm_id is decremented not in the event handler
works, but in the cm_id destruction context. For that purpose, move
iwcm_deref_id() call from destroy_cm_id() to the callers of
destroy_cm_id(). In iw_destroy_cm_id(), call iwcm_deref_id() after
flushing the pending works.

During the fix work, I noticed that iw_destroy_cm_id() is called from
cm_work_handler() and process_event() context. However, the comment of
iw_destroy_cm_id() notes that the function "cannot be called by the
event thread". Drop the false comment.

Closes: https://lore.kernel.org/linux-rdma/r5676e754sv35aq7cdsqrlnvyhiq5zktteaurl7vmfih35efko@z6lay7uypy3c/
Fixes: 59c68ac31e15 ("iw_cm: free cm_id resources on the last deref")
Cc: stable@vger.kernel.org
Signed-off-by: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com>
Link: https://patch.msgid.link/20250510101036.1756439-1-shinichiro.kawasaki@wdc.com
Reviewed-by: Zhu Yanjun <yanjun.zhu@linux.dev>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/core/iwcm.c | 29 +++++++++++++++--------------
 1 file changed, 15 insertions(+), 14 deletions(-)

diff --git a/drivers/infiniband/core/iwcm.c b/drivers/infiniband/core/iwcm.c
index f4486cbd8f45..62410578dec3 100644
--- a/drivers/infiniband/core/iwcm.c
+++ b/drivers/infiniband/core/iwcm.c
@@ -368,12 +368,9 @@ EXPORT_SYMBOL(iw_cm_disconnect);
 /*
  * CM_ID <-- DESTROYING
  *
- * Clean up all resources associated with the connection and release
- * the initial reference taken by iw_create_cm_id.
- *
- * Returns true if and only if the last cm_id_priv reference has been dropped.
+ * Clean up all resources associated with the connection.
  */
-static bool destroy_cm_id(struct iw_cm_id *cm_id)
+static void destroy_cm_id(struct iw_cm_id *cm_id)
 {
 	struct iwcm_id_private *cm_id_priv;
 	struct ib_qp *qp;
@@ -442,20 +439,22 @@ static bool destroy_cm_id(struct iw_cm_id *cm_id)
 		iwpm_remove_mapinfo(&cm_id->local_addr, &cm_id->m_local_addr);
 		iwpm_remove_mapping(&cm_id->local_addr, RDMA_NL_IWCM);
 	}
-
-	return iwcm_deref_id(cm_id_priv);
 }
 
 /*
- * This function is only called by the application thread and cannot
- * be called by the event thread. The function will wait for all
- * references to be released on the cm_id and then kfree the cm_id
- * object.
+ * Destroy cm_id. If the cm_id still has other references, wait for all
+ * references to be released on the cm_id and then release the initial
+ * reference taken by iw_create_cm_id.
  */
 void iw_destroy_cm_id(struct iw_cm_id *cm_id)
 {
-	if (!destroy_cm_id(cm_id))
+	struct iwcm_id_private *cm_id_priv;
+
+	cm_id_priv = container_of(cm_id, struct iwcm_id_private, id);
+	destroy_cm_id(cm_id);
+	if (refcount_read(&cm_id_priv->refcount) > 1)
 		flush_workqueue(iwcm_wq);
+	iwcm_deref_id(cm_id_priv);
 }
 EXPORT_SYMBOL(iw_destroy_cm_id);
 
@@ -1035,8 +1034,10 @@ static void cm_work_handler(struct work_struct *_work)
 
 		if (!test_bit(IWCM_F_DROP_EVENTS, &cm_id_priv->flags)) {
 			ret = process_event(cm_id_priv, &levent);
-			if (ret)
-				WARN_ON_ONCE(destroy_cm_id(&cm_id_priv->id));
+			if (ret) {
+				destroy_cm_id(&cm_id_priv->id);
+				WARN_ON_ONCE(iwcm_deref_id(cm_id_priv));
+			}
 		} else
 			pr_debug("dropping event %d\n", levent.event);
 		if (iwcm_deref_id(cm_id_priv))

From 972db388d40ded1a5ef8ce09d92ef1e2b9e40f07 Mon Sep 17 00:00:00 2001
From: Yishai Hadas <yishaih@nvidia.com>
Date: Tue, 13 May 2025 14:02:40 +0300
Subject: [PATCH 46/57] RDMA/mlx5: Remove the redundant MLX5_IB_STAGE_UAR stage

The MLX5_IB_STAGE_UAR stage in the RDMA driver is redundant and should
be removed.

Responsibility for initializing the device's UAR pointer
(mdev->priv.uar) lies with mlx5_core, which already sets it during the
mlx5_load() process.

At present, the RDMA UAR stage overwrites this pointer, which was
correctly initialized by mlx5_core, creating the risk of inconsistency.

Ownership and management of the UAR pointer should remain exclusively
within mlx5_core.

In the current upstream code, we luckily receive the same pointer, since
mlx5_get_uars_page() still finds available BF registers for that UAR,
allowing it to be shared.

However, future changes in mlx5_core may expose this flaw.
For instance, if mlx5_alloc_bfreg() is invoked twice before the RDMA UAR
stage runs, the RDMA driver may overwrite the UAR allocated by
mlx5_core.

This could lead to real bugs. For example, if mlx5_ib is unloaded
(rmmod), it might free the UAR, leaving mlx5_core with a dangling
reference to an invalid UAR.

Signed-off-by: Yishai Hadas <yishaih@nvidia.com>
Reviewed-by: Fan Li <fanl@nvidia.com>
Link: https://patch.msgid.link/feaa84ec6f20468b4935c439923e9266122a93d0.1747134130.git.leon@kernel.org
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/hw/mlx5/main.c    | 17 -----------------
 drivers/infiniband/hw/mlx5/mlx5_ib.h |  1 -
 2 files changed, 18 deletions(-)

diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index d07cacaa0abd..32ad066c3c4a 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -4422,17 +4422,6 @@ static void mlx5_ib_stage_cong_debugfs_cleanup(struct mlx5_ib_dev *dev)
 				     mlx5_core_native_port_num(dev->mdev) - 1);
 }
 
-static int mlx5_ib_stage_uar_init(struct mlx5_ib_dev *dev)
-{
-	dev->mdev->priv.uar = mlx5_get_uars_page(dev->mdev);
-	return PTR_ERR_OR_ZERO(dev->mdev->priv.uar);
-}
-
-static void mlx5_ib_stage_uar_cleanup(struct mlx5_ib_dev *dev)
-{
-	mlx5_put_uars_page(dev->mdev, dev->mdev->priv.uar);
-}
-
 static int mlx5_ib_stage_bfrag_init(struct mlx5_ib_dev *dev)
 {
 	int err;
@@ -4662,9 +4651,6 @@ static const struct mlx5_ib_profile pf_profile = {
 	STAGE_CREATE(MLX5_IB_STAGE_CONG_DEBUGFS,
 		     mlx5_ib_stage_cong_debugfs_init,
 		     mlx5_ib_stage_cong_debugfs_cleanup),
-	STAGE_CREATE(MLX5_IB_STAGE_UAR,
-		     mlx5_ib_stage_uar_init,
-		     mlx5_ib_stage_uar_cleanup),
 	STAGE_CREATE(MLX5_IB_STAGE_BFREG,
 		     mlx5_ib_stage_bfrag_init,
 		     mlx5_ib_stage_bfrag_cleanup),
@@ -4722,9 +4708,6 @@ const struct mlx5_ib_profile raw_eth_profile = {
 	STAGE_CREATE(MLX5_IB_STAGE_CONG_DEBUGFS,
 		     mlx5_ib_stage_cong_debugfs_init,
 		     mlx5_ib_stage_cong_debugfs_cleanup),
-	STAGE_CREATE(MLX5_IB_STAGE_UAR,
-		     mlx5_ib_stage_uar_init,
-		     mlx5_ib_stage_uar_cleanup),
 	STAGE_CREATE(MLX5_IB_STAGE_BFREG,
 		     mlx5_ib_stage_bfrag_init,
 		     mlx5_ib_stage_bfrag_cleanup),
diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h
index 27fbf5f3a2d6..fde859d207ae 100644
--- a/drivers/infiniband/hw/mlx5/mlx5_ib.h
+++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h
@@ -1006,7 +1006,6 @@ enum mlx5_ib_stages {
 	MLX5_IB_STAGE_ODP,
 	MLX5_IB_STAGE_COUNTERS,
 	MLX5_IB_STAGE_CONG_DEBUGFS,
-	MLX5_IB_STAGE_UAR,
 	MLX5_IB_STAGE_BFREG,
 	MLX5_IB_STAGE_PRE_IB_REG_UMR,
 	MLX5_IB_STAGE_WHITELIST_UID,

From d00d16bcbc2553a3ac9acccf2d6444cda5502adf Mon Sep 17 00:00:00 2001
From: Patrisious Haddad <phaddad@nvidia.com>
Date: Tue, 13 May 2025 14:03:41 +0300
Subject: [PATCH 47/57] RDMA/mlx5: Add support for 200Gbps per lane speeds

Add support for 200Gbps per lane speeds speed when querying PTYS and
report it back correctly when needed.

Signed-off-by: Patrisious Haddad <phaddad@nvidia.com>
Reviewed-by: Maor Gottlieb <maorg@nvidia.com>
Link: https://patch.msgid.link/b842d2f523e9b82e221378c444ebd5860d612959.1747134197.git.leon@kernel.org
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/hw/mlx5/main.c | 12 ++++++++++++
 1 file changed, 12 insertions(+)

diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 32ad066c3c4a..ce7610740412 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -485,6 +485,10 @@ static int translate_eth_ext_proto_oper(u32 eth_proto_oper, u16 *active_speed,
 		*active_width = IB_WIDTH_2X;
 		*active_speed = IB_SPEED_NDR;
 		break;
+	case MLX5E_PROT_MASK(MLX5E_200GAUI_1_200GBASE_CR1_KR1):
+		*active_width = IB_WIDTH_1X;
+		*active_speed = IB_SPEED_XDR;
+		break;
 	case MLX5E_PROT_MASK(MLX5E_400GAUI_8_400GBASE_CR8):
 		*active_width = IB_WIDTH_8X;
 		*active_speed = IB_SPEED_HDR;
@@ -493,10 +497,18 @@ static int translate_eth_ext_proto_oper(u32 eth_proto_oper, u16 *active_speed,
 		*active_width = IB_WIDTH_4X;
 		*active_speed = IB_SPEED_NDR;
 		break;
+	case MLX5E_PROT_MASK(MLX5E_400GAUI_2_400GBASE_CR2_KR2):
+		*active_width = IB_WIDTH_2X;
+		*active_speed = IB_SPEED_XDR;
+		break;
 	case MLX5E_PROT_MASK(MLX5E_800GAUI_8_800GBASE_CR8_KR8):
 		*active_width = IB_WIDTH_8X;
 		*active_speed = IB_SPEED_NDR;
 		break;
+	case MLX5E_PROT_MASK(MLX5E_800GAUI_4_800GBASE_CR4_KR4):
+		*active_width = IB_WIDTH_4X;
+		*active_speed = IB_SPEED_XDR;
+		break;
 	default:
 		return -EINVAL;
 	}

From 58d7a965bb2b014d467445d38cdb07099b1f0f77 Mon Sep 17 00:00:00 2001
From: Gautam R A <gautam-r.a@broadcom.com>
Date: Tue, 20 May 2025 09:29:07 +0530
Subject: [PATCH 48/57] RDMA/bnxt_re: Fix incorrect display of inactivity_cp in
 debugfs output

The inactivity_cp parameter in debugfs was not being read or
written correctly, resulting in "Invalid argument" errors.

Fixed this by ensuring proper mapping of inactivity_cp in
both the map_cc_config_offset_gen0_ext0 and
bnxt_re_fill_gen0_ext0() functions.

Fixes: 656dff55da19 ("RDMA/bnxt_re: Congestion control settings using debugfs hook")
Signed-off-by: Gautam R A <gautam-r.a@broadcom.com>
Link: https://patch.msgid.link/20250520035910.1061918-2-kalesh-anakkur.purayil@broadcom.com
Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/hw/bnxt_re/debugfs.c | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/drivers/infiniband/hw/bnxt_re/debugfs.c b/drivers/infiniband/hw/bnxt_re/debugfs.c
index af91d16c3c77..a3aad6c3dbec 100644
--- a/drivers/infiniband/hw/bnxt_re/debugfs.c
+++ b/drivers/infiniband/hw/bnxt_re/debugfs.c
@@ -170,6 +170,9 @@ static int map_cc_config_offset_gen0_ext0(u32 offset, struct bnxt_qplib_cc_param
 	case CMDQ_MODIFY_ROCE_CC_MODIFY_MASK_TCP_CP:
 		*val =  ccparam->tcp_cp;
 		break;
+	case CMDQ_MODIFY_ROCE_CC_MODIFY_MASK_INACTIVITY_CP:
+		*val = ccparam->inact_th;
+		break;
 	default:
 		return -EINVAL;
 	}
@@ -247,7 +250,9 @@ static void bnxt_re_fill_gen0_ext0(struct bnxt_qplib_cc_param *ccparam, u32 offs
 		ccparam->tcp_cp = val;
 		break;
 	case CMDQ_MODIFY_ROCE_CC_MODIFY_MASK_TX_QUEUE:
+		break;
 	case CMDQ_MODIFY_ROCE_CC_MODIFY_MASK_INACTIVITY_CP:
+		ccparam->inact_th = val;
 		break;
 	case CMDQ_MODIFY_ROCE_CC_MODIFY_MASK_TIME_PER_PHASE:
 		ccparam->time_pph = val;

From e3d57a00d4d1f36689e9eab80b60d5024361efec Mon Sep 17 00:00:00 2001
From: Gautam R A <gautam-r.a@broadcom.com>
Date: Tue, 20 May 2025 09:29:08 +0530
Subject: [PATCH 49/57] RDMA/bnxt_re: Fix missing error handling for tx_queue

bnxt_re_fill_gen0_ext0() did not return an error when
attempting to modify CMDQ_MODIFY_ROCE_CC_MODIFY_MASK_TX_QUEUE,
leading to silent failures.

Fixed this by returning -EOPNOTSUPP for tx_queue modifications and
ensuring proper error propagation in bnxt_re_configure_cc().

Fixes: 656dff55da19 ("RDMA/bnxt_re: Congestion control settings using debugfs hook")
Signed-off-by: Gautam R A <gautam-r.a@broadcom.com>
Link: https://patch.msgid.link/20250520035910.1061918-3-kalesh-anakkur.purayil@broadcom.com
Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/hw/bnxt_re/debugfs.c | 15 +++++++++------
 1 file changed, 9 insertions(+), 6 deletions(-)

diff --git a/drivers/infiniband/hw/bnxt_re/debugfs.c b/drivers/infiniband/hw/bnxt_re/debugfs.c
index a3aad6c3dbec..9f6392155d91 100644
--- a/drivers/infiniband/hw/bnxt_re/debugfs.c
+++ b/drivers/infiniband/hw/bnxt_re/debugfs.c
@@ -206,7 +206,7 @@ static ssize_t bnxt_re_cc_config_get(struct file *filp, char __user *buffer,
 	return simple_read_from_buffer(buffer, usr_buf_len, ppos, (u8 *)(buf), rc);
 }
 
-static void bnxt_re_fill_gen0_ext0(struct bnxt_qplib_cc_param *ccparam, u32 offset, u32 val)
+static int bnxt_re_fill_gen0_ext0(struct bnxt_qplib_cc_param *ccparam, u32 offset, u32 val)
 {
 	u32 modify_mask;
 
@@ -250,7 +250,7 @@ static void bnxt_re_fill_gen0_ext0(struct bnxt_qplib_cc_param *ccparam, u32 offs
 		ccparam->tcp_cp = val;
 		break;
 	case CMDQ_MODIFY_ROCE_CC_MODIFY_MASK_TX_QUEUE:
-		break;
+		return -EOPNOTSUPP;
 	case CMDQ_MODIFY_ROCE_CC_MODIFY_MASK_INACTIVITY_CP:
 		ccparam->inact_th = val;
 		break;
@@ -263,18 +263,21 @@ static void bnxt_re_fill_gen0_ext0(struct bnxt_qplib_cc_param *ccparam, u32 offs
 	}
 
 	ccparam->mask = modify_mask;
+	return 0;
 }
 
 static int bnxt_re_configure_cc(struct bnxt_re_dev *rdev, u32 gen_ext, u32 offset, u32 val)
 {
 	struct bnxt_qplib_cc_param ccparam = { };
+	int rc;
 
-	/* Supporting only Gen 0 now */
-	if (gen_ext == CC_CONFIG_GEN0_EXT0)
-		bnxt_re_fill_gen0_ext0(&ccparam, offset, val);
-	else
+	if (gen_ext != CC_CONFIG_GEN0_EXT0)
 		return -EINVAL;
 
+	rc = bnxt_re_fill_gen0_ext0(&ccparam, offset, val);
+	if (rc)
+		return rc;
+
 	bnxt_qplib_modify_cc(&rdev->qplib_res, &ccparam);
 	return 0;
 }

From 990b5c07f677a0b633b41130a70771337c18343e Mon Sep 17 00:00:00 2001
From: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Date: Tue, 20 May 2025 09:29:09 +0530
Subject: [PATCH 50/57] RDMA/bnxt_re: Fix return code of bnxt_re_configure_cc

Driver currently supports modifying GEN0_EXT0 CC parameters
through debugfs hook.

Fixed to return -EOPNOTSUPP instead of -EINVAL in bnxt_re_configure_cc()
when the user tries to modify any other CC parameters.

Fixes: 656dff55da19 ("RDMA/bnxt_re: Congestion control settings using debugfs hook")
Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Link: https://patch.msgid.link/20250520035910.1061918-4-kalesh-anakkur.purayil@broadcom.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/hw/bnxt_re/debugfs.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/infiniband/hw/bnxt_re/debugfs.c b/drivers/infiniband/hw/bnxt_re/debugfs.c
index 9f6392155d91..e632f1661b92 100644
--- a/drivers/infiniband/hw/bnxt_re/debugfs.c
+++ b/drivers/infiniband/hw/bnxt_re/debugfs.c
@@ -272,7 +272,7 @@ static int bnxt_re_configure_cc(struct bnxt_re_dev *rdev, u32 gen_ext, u32 offse
 	int rc;
 
 	if (gen_ext != CC_CONFIG_GEN0_EXT0)
-		return -EINVAL;
+		return -EOPNOTSUPP;
 
 	rc = bnxt_re_fill_gen0_ext0(&ccparam, offset, val);
 	if (rc)

From 0b261d7c1cd32dc93cbc92425fb55e67b24c6638 Mon Sep 17 00:00:00 2001
From: Leon Romanovsky <leonro@nvidia.com>
Date: Thu, 22 May 2025 14:36:18 +0300
Subject: [PATCH 51/57] RDMA/rxe: Break endless pagefault loop for RO pages

RO pages has "perm" equal to 0, that caused to the situation
where such pages were marked as needed to have fault and caused
to infinite loop.

Fixes: eedd5b1276e7 ("RDMA/umem: Store ODP access mask information in PFN")
Reported-by: Daisuke Matsuda <dskmtsd@gmail.com>
Closes: https://lore.kernel.org/all/3016329a-4edd-4550-862f-b298a1b79a39@gmail.com/
Link: https://patch.msgid.link/096fab178d48ed86942ee22eafe9be98e29092aa.1747913377.git.leonro@nvidia.com
Tested-by: Daisuke Matsuda <dskmtsd@gmail.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
---
 drivers/infiniband/sw/rxe/rxe_odp.c | 14 +++++---------
 1 file changed, 5 insertions(+), 9 deletions(-)

diff --git a/drivers/infiniband/sw/rxe/rxe_odp.c b/drivers/infiniband/sw/rxe/rxe_odp.c
index a1416626f61a..dbc5a5600eb7 100644
--- a/drivers/infiniband/sw/rxe/rxe_odp.c
+++ b/drivers/infiniband/sw/rxe/rxe_odp.c
@@ -124,8 +124,8 @@ int rxe_odp_mr_init_user(struct rxe_dev *rxe, u64 start, u64 length,
 	return err;
 }
 
-static inline bool rxe_check_pagefault(struct ib_umem_odp *umem_odp,
-				       u64 iova, int length, u32 perm)
+static inline bool rxe_check_pagefault(struct ib_umem_odp *umem_odp, u64 iova,
+				       int length)
 {
 	bool need_fault = false;
 	u64 addr;
@@ -137,7 +137,7 @@ static inline bool rxe_check_pagefault(struct ib_umem_odp *umem_odp,
 	while (addr < iova + length) {
 		idx = (addr - ib_umem_start(umem_odp)) >> umem_odp->page_shift;
 
-		if (!(umem_odp->map.pfn_list[idx] & perm)) {
+		if (!(umem_odp->map.pfn_list[idx] & HMM_PFN_VALID)) {
 			need_fault = true;
 			break;
 		}
@@ -161,18 +161,14 @@ static int rxe_odp_map_range_and_lock(struct rxe_mr *mr, u64 iova, int length, u
 {
 	struct ib_umem_odp *umem_odp = to_ib_umem_odp(mr->umem);
 	bool need_fault;
-	u64 perm = 0;
 	int err;
 
 	if (unlikely(length < 1))
 		return -EINVAL;
 
-	if (!(flags & RXE_PAGEFAULT_RDONLY))
-		perm |= HMM_PFN_WRITE;
-
 	mutex_lock(&umem_odp->umem_mutex);
 
-	need_fault = rxe_check_pagefault(umem_odp, iova, length, perm);
+	need_fault = rxe_check_pagefault(umem_odp, iova, length);
 	if (need_fault) {
 		mutex_unlock(&umem_odp->umem_mutex);
 
@@ -182,7 +178,7 @@ static int rxe_odp_map_range_and_lock(struct rxe_mr *mr, u64 iova, int length, u
 		if (err < 0)
 			return err;
 
-		need_fault = rxe_check_pagefault(umem_odp, iova, length, perm);
+		need_fault = rxe_check_pagefault(umem_odp, iova, length);
 		if (need_fault)
 			return -EFAULT;
 	}

From 259e9bd07c57b0fd8588f68e9adbe2c41170f4fe Mon Sep 17 00:00:00 2001
From: Daisuke Matsuda <dskmtsd@gmail.com>
Date: Sat, 24 May 2025 14:43:28 +0000
Subject: [PATCH 52/57] RDMA/core: Avoid hmm_dma_map_alloc() for virtual DMA
 devices

Drivers such as rxe, which use virtual DMA, must not call into the DMA
mapping core since they lack physical DMA capabilities. Otherwise, a NULL
pointer dereference is observed as shown below. This patch ensures the RDMA
core handles virtual and physical DMA paths appropriately.

This fixes the following kernel oops:

 BUG: kernel NULL pointer dereference, address: 00000000000002fc
 #PF: supervisor read access in kernel mode
 #PF: error_code(0x0000) - not-present page
 PGD 1028eb067 P4D 1028eb067 PUD 105da0067 PMD 0
 Oops: Oops: 0000 [#1] SMP NOPTI
 CPU: 3 UID: 1000 PID: 1854 Comm: python3 Tainted: G        W           6.15.0-rc1+ #11 PREEMPT(voluntary)
 Tainted: [W]=WARN
 Hardware name: Trigkey Key N/Key N, BIOS KEYN101 09/02/2024
 RIP: 0010:hmm_dma_map_alloc+0x25/0x100
 Code: 90 90 90 90 90 0f 1f 44 00 00 55 48 89 e5 41 57 41 56 49 89 d6 49 c1 e6 0c 41 55 41 54 53 49 39 ce 0f 82 c6 00 00 00 49 89 fc <f6> 87 fc 02 00 00 20 0f 84 af 00 00 00 49 89 f5 48 89 d3 49 89 cf
 RSP: 0018:ffffd3d3420eb830 EFLAGS: 00010246
 RAX: 0000000000001000 RBX: ffff8b727c7f7400 RCX: 0000000000001000
 RDX: 0000000000000001 RSI: ffff8b727c7f74b0 RDI: 0000000000000000
 RBP: ffffd3d3420eb858 R08: 0000000000000000 R09: 0000000000000000
 R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000
 R13: 00007262a622a000 R14: 0000000000001000 R15: ffff8b727c7f74b0
 FS:  00007262a62a1080(0000) GS:ffff8b762ac3e000(0000) knlGS:0000000000000000
 CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
 CR2: 00000000000002fc CR3: 000000010a1f0004 CR4: 0000000000f72ef0
 PKRU: 55555554
 Call Trace:
  <TASK>
  ib_init_umem_odp+0xb6/0x110 [ib_uverbs]
  ib_umem_odp_get+0xf0/0x150 [ib_uverbs]
  rxe_odp_mr_init_user+0x71/0x170 [rdma_rxe]
  rxe_reg_user_mr+0x217/0x2e0 [rdma_rxe]
  ib_uverbs_reg_mr+0x19e/0x2e0 [ib_uverbs]
  ib_uverbs_handler_UVERBS_METHOD_INVOKE_WRITE+0xd9/0x150 [ib_uverbs]
  ib_uverbs_cmd_verbs+0xd19/0xee0 [ib_uverbs]
  ? mmap_region+0x63/0xd0
  ? __pfx_ib_uverbs_handler_UVERBS_METHOD_INVOKE_WRITE+0x10/0x10 [ib_uverbs]
  ib_uverbs_ioctl+0xba/0x130 [ib_uverbs]
  __x64_sys_ioctl+0xa4/0xe0
  x64_sys_call+0x1178/0x2660
  do_syscall_64+0x7e/0x170
  ? syscall_exit_to_user_mode+0x4e/0x250
  ? do_syscall_64+0x8a/0x170
  ? do_syscall_64+0x8a/0x170
  ? syscall_exit_to_user_mode+0x4e/0x250
  ? do_syscall_64+0x8a/0x170
  ? syscall_exit_to_user_mode+0x4e/0x250
  ? do_syscall_64+0x8a/0x170
  ? do_user_addr_fault+0x1d2/0x8d0
  ? irqentry_exit_to_user_mode+0x43/0x250
  ? irqentry_exit+0x43/0x50
  ? exc_page_fault+0x93/0x1d0
  entry_SYSCALL_64_after_hwframe+0x76/0x7e
 RIP: 0033:0x7262a6124ded
 Code: 04 25 28 00 00 00 48 89 45 c8 31 c0 48 8d 45 10 c7 45 b0 10 00 00 00 48 89 45 b8 48 8d 45 d0 48 89 45 c0 b8 10 00 00 00 0f 05 <89> c2 3d 00 f0 ff ff 77 1a 48 8b 45 c8 64 48 2b 04 25 28 00 00 00
 RSP: 002b:00007fffd08c3960 EFLAGS: 00000246 ORIG_RAX: 0000000000000010
 RAX: ffffffffffffffda RBX: 00007fffd08c39f0 RCX: 00007262a6124ded
 RDX: 00007fffd08c3a10 RSI: 00000000c0181b01 RDI: 0000000000000007
 RBP: 00007fffd08c39b0 R08: 0000000014107820 R09: 00007fffd08c3b44
 R10: 000000000000000c R11: 0000000000000246 R12: 00007fffd08c3b44
 R13: 000000000000000c R14: 00007fffd08c3b58 R15: 0000000014107960
  </TASK>

Fixes: 1efe8c0670d6 ("RDMA/core: Convert UMEM ODP DMA mapping to caching IOVA and page linkage")
Closes: https://lore.kernel.org/all/3e8f343f-7d66-4f7a-9f08-3910623e322f@gmail.com/
Signed-off-by: Daisuke Matsuda <dskmtsd@gmail.com>
Link: https://patch.msgid.link/20250524144328.4361-1-dskmtsd@gmail.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/core/umem_odp.c | 31 ++++++++++++++++++++++++------
 mm/hmm.c                           |  3 +--
 2 files changed, 26 insertions(+), 8 deletions(-)

diff --git a/drivers/infiniband/core/umem_odp.c b/drivers/infiniband/core/umem_odp.c
index 51d518989914..c752ae9fad6c 100644
--- a/drivers/infiniband/core/umem_odp.c
+++ b/drivers/infiniband/core/umem_odp.c
@@ -60,9 +60,11 @@ static int ib_init_umem_odp(struct ib_umem_odp *umem_odp,
 {
 	struct ib_device *dev = umem_odp->umem.ibdev;
 	size_t page_size = 1UL << umem_odp->page_shift;
+	struct hmm_dma_map *map;
 	unsigned long start;
 	unsigned long end;
-	int ret;
+	size_t nr_entries;
+	int ret = 0;
 
 	umem_odp->umem.is_odp = 1;
 	mutex_init(&umem_odp->umem_mutex);
@@ -75,9 +77,20 @@ static int ib_init_umem_odp(struct ib_umem_odp *umem_odp,
 	if (unlikely(end < page_size))
 		return -EOVERFLOW;
 
-	ret = hmm_dma_map_alloc(dev->dma_device, &umem_odp->map,
-				(end - start) >> PAGE_SHIFT,
-				1 << umem_odp->page_shift);
+	nr_entries = (end - start) >> PAGE_SHIFT;
+	if (!(nr_entries * PAGE_SIZE / page_size))
+		return -EINVAL;
+
+	map = &umem_odp->map;
+	if (ib_uses_virt_dma(dev)) {
+		map->pfn_list = kvcalloc(nr_entries, sizeof(*map->pfn_list),
+					 GFP_KERNEL | __GFP_NOWARN);
+		if (!map->pfn_list)
+			ret = -ENOMEM;
+	} else
+		ret = hmm_dma_map_alloc(dev->dma_device, map,
+					(end - start) >> PAGE_SHIFT,
+					1 << umem_odp->page_shift);
 	if (ret)
 		return ret;
 
@@ -90,7 +103,10 @@ static int ib_init_umem_odp(struct ib_umem_odp *umem_odp,
 	return 0;
 
 out_free_map:
-	hmm_dma_map_free(dev->dma_device, &umem_odp->map);
+	if (ib_uses_virt_dma(dev))
+		kfree(map->pfn_list);
+	else
+		hmm_dma_map_free(dev->dma_device, map);
 	return ret;
 }
 
@@ -259,7 +275,10 @@ static void ib_umem_odp_free(struct ib_umem_odp *umem_odp)
 				    ib_umem_end(umem_odp));
 	mutex_unlock(&umem_odp->umem_mutex);
 	mmu_interval_notifier_remove(&umem_odp->notifier);
-	hmm_dma_map_free(dev->dma_device, &umem_odp->map);
+	if (ib_uses_virt_dma(dev))
+		kfree(umem_odp->map.pfn_list);
+	else
+		hmm_dma_map_free(dev->dma_device, &umem_odp->map);
 }
 
 void ib_umem_odp_release(struct ib_umem_odp *umem_odp)
diff --git a/mm/hmm.c b/mm/hmm.c
index a8bf097677f3..feac86196a65 100644
--- a/mm/hmm.c
+++ b/mm/hmm.c
@@ -640,8 +640,7 @@ int hmm_dma_map_alloc(struct device *dev, struct hmm_dma_map *map,
 	bool dma_need_sync = false;
 	bool use_iova;
 
-	if (!(nr_entries * PAGE_SIZE / dma_entry_size))
-		return -EINVAL;
+	WARN_ON_ONCE(!(nr_entries * PAGE_SIZE / dma_entry_size));
 
 	/*
 	 * The HMM API violates our normal DMA buffer ownership rules and can't

From 45611fe821af0e09cb9dc2dfd2418b2e1e1831f3 Mon Sep 17 00:00:00 2001
From: Vlad Dumitrescu <vdumitrescu@nvidia.com>
Date: Wed, 21 May 2025 14:34:17 +0300
Subject: [PATCH 53/57] IB/cm: Remove dead code and adjust naming

Drop ib_send_cm_mra parameters which are always constant. Remove branch
which is never taken. Adjust name to ib_prepare_cm_mra, which better
reflects its functionality - no MRA is actually sent. Adjust name of
related tracepoints. Push setting of the constant service timeout to
cm.c and drop IB_CM_MRA_FLAG_DELAY.

Signed-off-by: Vlad Dumitrescu <vdumitrescu@nvidia.com>
Reviewed-by: Sean Hefty <shefty@nvidia.com>
Link: https://patch.msgid.link/cdd2a237acf2b495c19ce02e4b1c42c41c6751c2.1747827207.git.leon@kernel.org
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/core/cm.c        | 59 +++++------------------------
 drivers/infiniband/core/cm_trace.h  |  2 +-
 drivers/infiniband/core/cma.c       |  9 ++---
 drivers/infiniband/core/cma_trace.h |  2 +-
 include/rdma/ib_cm.h                | 17 ++-------
 5 files changed, 19 insertions(+), 70 deletions(-)

diff --git a/drivers/infiniband/core/cm.c b/drivers/infiniband/core/cm.c
index e64cbd034a2a..8670e58675c6 100644
--- a/drivers/infiniband/core/cm.c
+++ b/drivers/infiniband/core/cm.c
@@ -36,6 +36,7 @@ MODULE_LICENSE("Dual BSD/GPL");
 
 #define CM_DESTROY_ID_WAIT_TIMEOUT 10000 /* msecs */
 #define CM_DIRECT_RETRY_CTX ((void *) 1UL)
+#define CM_MRA_SETTING 24 /* 4.096us * 2^24 = ~68.7 seconds */
 
 static const char * const ibcm_rej_reason_strs[] = {
 	[IB_CM_REJ_NO_QP]			= "no QP",
@@ -241,7 +242,6 @@ struct cm_id_private {
 	u8 initiator_depth;
 	u8 retry_count;
 	u8 rnr_retry_count;
-	u8 service_timeout;
 	u8 target_ack_delay;
 
 	struct list_head work_list;
@@ -1872,7 +1872,7 @@ static void cm_process_work(struct cm_id_private *cm_id_priv,
 
 static void cm_format_mra(struct cm_mra_msg *mra_msg,
 			  struct cm_id_private *cm_id_priv,
-			  enum cm_msg_response msg_mraed, u8 service_timeout,
+			  enum cm_msg_response msg_mraed,
 			  const void *private_data, u8 private_data_len)
 {
 	cm_format_mad_hdr(&mra_msg->hdr, CM_MRA_ATTR_ID, cm_id_priv->tid);
@@ -1881,7 +1881,7 @@ static void cm_format_mra(struct cm_mra_msg *mra_msg,
 		be32_to_cpu(cm_id_priv->id.local_id));
 	IBA_SET(CM_MRA_REMOTE_COMM_ID, mra_msg,
 		be32_to_cpu(cm_id_priv->id.remote_id));
-	IBA_SET(CM_MRA_SERVICE_TIMEOUT, mra_msg, service_timeout);
+	IBA_SET(CM_MRA_SERVICE_TIMEOUT, mra_msg, CM_MRA_SETTING);
 
 	if (private_data && private_data_len)
 		IBA_SET_MEM(CM_MRA_PRIVATE_DATA, mra_msg, private_data,
@@ -1960,7 +1960,7 @@ static void cm_dup_req_handler(struct cm_work *work,
 	switch (cm_id_priv->id.state) {
 	case IB_CM_MRA_REQ_SENT:
 		cm_format_mra((struct cm_mra_msg *) msg->mad, cm_id_priv,
-			      CM_MSG_RESPONSE_REQ, cm_id_priv->service_timeout,
+			      CM_MSG_RESPONSE_REQ,
 			      cm_id_priv->private_data,
 			      cm_id_priv->private_data_len);
 		break;
@@ -2454,7 +2454,7 @@ static void cm_dup_rep_handler(struct cm_work *work)
 			      cm_id_priv->private_data_len);
 	else if (cm_id_priv->id.state == IB_CM_MRA_REP_SENT)
 		cm_format_mra((struct cm_mra_msg *) msg->mad, cm_id_priv,
-			      CM_MSG_RESPONSE_REP, cm_id_priv->service_timeout,
+			      CM_MSG_RESPONSE_REP,
 			      cm_id_priv->private_data,
 			      cm_id_priv->private_data_len);
 	else
@@ -3094,26 +3094,13 @@ out:
 	return -EINVAL;
 }
 
-int ib_send_cm_mra(struct ib_cm_id *cm_id,
-		   u8 service_timeout,
-		   const void *private_data,
-		   u8 private_data_len)
+int ib_prepare_cm_mra(struct ib_cm_id *cm_id)
 {
 	struct cm_id_private *cm_id_priv;
-	struct ib_mad_send_buf *msg;
 	enum ib_cm_state cm_state;
 	enum ib_cm_lap_state lap_state;
-	enum cm_msg_response msg_response;
-	void *data;
 	unsigned long flags;
-	int ret;
-
-	if (private_data && private_data_len > IB_CM_MRA_PRIVATE_DATA_SIZE)
-		return -EINVAL;
-
-	data = cm_copy_private_data(private_data, private_data_len);
-	if (IS_ERR(data))
-		return PTR_ERR(data);
+	int ret = 0;
 
 	cm_id_priv = container_of(cm_id, struct cm_id_private, id);
 
@@ -3122,58 +3109,33 @@ int ib_send_cm_mra(struct ib_cm_id *cm_id,
 	case IB_CM_REQ_RCVD:
 		cm_state = IB_CM_MRA_REQ_SENT;
 		lap_state = cm_id->lap_state;
-		msg_response = CM_MSG_RESPONSE_REQ;
 		break;
 	case IB_CM_REP_RCVD:
 		cm_state = IB_CM_MRA_REP_SENT;
 		lap_state = cm_id->lap_state;
-		msg_response = CM_MSG_RESPONSE_REP;
 		break;
 	case IB_CM_ESTABLISHED:
 		if (cm_id->lap_state == IB_CM_LAP_RCVD) {
 			cm_state = cm_id->state;
 			lap_state = IB_CM_MRA_LAP_SENT;
-			msg_response = CM_MSG_RESPONSE_OTHER;
 			break;
 		}
 		fallthrough;
 	default:
-		trace_icm_send_mra_unknown_err(&cm_id_priv->id);
+		trace_icm_prepare_mra_unknown_err(&cm_id_priv->id);
 		ret = -EINVAL;
 		goto error_unlock;
 	}
 
-	if (!(service_timeout & IB_CM_MRA_FLAG_DELAY)) {
-		msg = cm_alloc_msg(cm_id_priv);
-		if (IS_ERR(msg)) {
-			ret = PTR_ERR(msg);
-			goto error_unlock;
-		}
-
-		cm_format_mra((struct cm_mra_msg *) msg->mad, cm_id_priv,
-			      msg_response, service_timeout,
-			      private_data, private_data_len);
-		trace_icm_send_mra(cm_id);
-		ret = ib_post_send_mad(msg, NULL);
-		if (ret)
-			goto error_free_msg;
-	}
-
 	cm_id->state = cm_state;
 	cm_id->lap_state = lap_state;
-	cm_id_priv->service_timeout = service_timeout;
-	cm_set_private_data(cm_id_priv, data, private_data_len);
-	spin_unlock_irqrestore(&cm_id_priv->lock, flags);
-	return 0;
+	cm_set_private_data(cm_id_priv, NULL, 0);
 
-error_free_msg:
-	cm_free_msg(msg);
 error_unlock:
 	spin_unlock_irqrestore(&cm_id_priv->lock, flags);
-	kfree(data);
 	return ret;
 }
-EXPORT_SYMBOL(ib_send_cm_mra);
+EXPORT_SYMBOL(ib_prepare_cm_mra);
 
 static struct cm_id_private *cm_acquire_mraed_id(struct cm_mra_msg *mra_msg)
 {
@@ -3377,7 +3339,6 @@ static int cm_lap_handler(struct cm_work *work)
 
 		cm_format_mra((struct cm_mra_msg *) msg->mad, cm_id_priv,
 			      CM_MSG_RESPONSE_OTHER,
-			      cm_id_priv->service_timeout,
 			      cm_id_priv->private_data,
 			      cm_id_priv->private_data_len);
 		spin_unlock_irq(&cm_id_priv->lock);
diff --git a/drivers/infiniband/core/cm_trace.h b/drivers/infiniband/core/cm_trace.h
index 944d9071245d..4a4987da69d4 100644
--- a/drivers/infiniband/core/cm_trace.h
+++ b/drivers/infiniband/core/cm_trace.h
@@ -229,7 +229,7 @@ DEFINE_CM_ERR_EVENT(send_drep);
 DEFINE_CM_ERR_EVENT(dreq_unknown);
 DEFINE_CM_ERR_EVENT(send_unknown_rej);
 DEFINE_CM_ERR_EVENT(rej_unknown);
-DEFINE_CM_ERR_EVENT(send_mra_unknown);
+DEFINE_CM_ERR_EVENT(prepare_mra_unknown);
 DEFINE_CM_ERR_EVENT(mra_unknown);
 DEFINE_CM_ERR_EVENT(qp_init);
 DEFINE_CM_ERR_EVENT(qp_rtr);
diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c
index cd7df623161d..ccd0d9cf1dda 100644
--- a/drivers/infiniband/core/cma.c
+++ b/drivers/infiniband/core/cma.c
@@ -46,7 +46,6 @@ MODULE_LICENSE("Dual BSD/GPL");
 
 #define CMA_CM_RESPONSE_TIMEOUT 20
 #define CMA_MAX_CM_RETRIES 15
-#define CMA_CM_MRA_SETTING (IB_CM_MRA_FLAG_DELAY | 24)
 #define CMA_IBOE_PACKET_LIFETIME 16
 #define CMA_PREFERRED_ROCE_GID_TYPE IB_GID_TYPE_ROCE_UDP_ENCAP
 
@@ -2198,8 +2197,8 @@ static int cma_ib_handler(struct ib_cm_id *cm_id,
 	case IB_CM_REP_RECEIVED:
 		if (state == RDMA_CM_CONNECT &&
 		    (id_priv->id.qp_type != IB_QPT_UD)) {
-			trace_cm_send_mra(id_priv);
-			ib_send_cm_mra(cm_id, CMA_CM_MRA_SETTING, NULL, 0);
+			trace_cm_prepare_mra(id_priv);
+			ib_prepare_cm_mra(cm_id);
 		}
 		if (id_priv->id.qp) {
 			event.status = cma_rep_recv(id_priv);
@@ -2460,8 +2459,8 @@ static int cma_ib_req_handler(struct ib_cm_id *cm_id,
 
 	if (READ_ONCE(conn_id->state) == RDMA_CM_CONNECT &&
 	    conn_id->id.qp_type != IB_QPT_UD) {
-		trace_cm_send_mra(cm_id->context);
-		ib_send_cm_mra(cm_id, CMA_CM_MRA_SETTING, NULL, 0);
+		trace_cm_prepare_mra(cm_id->context);
+		ib_prepare_cm_mra(cm_id);
 	}
 	mutex_unlock(&conn_id->handler_mutex);
 
diff --git a/drivers/infiniband/core/cma_trace.h b/drivers/infiniband/core/cma_trace.h
index dc622f3778be..3456d5f3aa47 100644
--- a/drivers/infiniband/core/cma_trace.h
+++ b/drivers/infiniband/core/cma_trace.h
@@ -55,7 +55,7 @@ DECLARE_EVENT_CLASS(cma_fsm_class,
 
 DEFINE_CMA_FSM_EVENT(send_rtu);
 DEFINE_CMA_FSM_EVENT(send_rej);
-DEFINE_CMA_FSM_EVENT(send_mra);
+DEFINE_CMA_FSM_EVENT(prepare_mra);
 DEFINE_CMA_FSM_EVENT(send_sidr_req);
 DEFINE_CMA_FSM_EVENT(send_sidr_rep);
 DEFINE_CMA_FSM_EVENT(disconnect);
diff --git a/include/rdma/ib_cm.h b/include/rdma/ib_cm.h
index a2ac62b4a6cf..1fa3786f82f4 100644
--- a/include/rdma/ib_cm.h
+++ b/include/rdma/ib_cm.h
@@ -480,23 +480,12 @@ int ib_send_cm_rej(struct ib_cm_id *cm_id,
 		   const void *private_data,
 		   u8 private_data_len);
 
-#define IB_CM_MRA_FLAG_DELAY 0x80  /* Send MRA only after a duplicate msg */
-
 /**
- * ib_send_cm_mra - Sends a message receipt acknowledgement to a connection
- *   message.
+ * ib_prepare_cm_mra - Prepares to send a message receipt acknowledgment to a
+     connection message in case duplicates are received.
  * @cm_id: Connection identifier associated with the connection message.
- * @service_timeout: The lower 5-bits specify the maximum time required for
- *   the sender to reply to the connection message.  The upper 3-bits
- *   specify additional control flags.
- * @private_data: Optional user-defined private data sent with the
- *   message receipt acknowledgement.
- * @private_data_len: Size of the private data buffer, in bytes.
  */
-int ib_send_cm_mra(struct ib_cm_id *cm_id,
-		   u8 service_timeout,
-		   const void *private_data,
-		   u8 private_data_len);
+int ib_prepare_cm_mra(struct ib_cm_id *cm_id);
 
 /**
  * ib_cm_init_qp_attr - Initializes the QP attributes for use in transitioning

From e91fb8b9d0edec86a1ef26490bc80af96210863d Mon Sep 17 00:00:00 2001
From: Leon Romanovsky <leonro@nvidia.com>
Date: Wed, 21 May 2025 14:34:58 +0300
Subject: [PATCH 54/57] RDMA/mlx5: Avoid flexible array warning

The following warning is reported by sparse tool:
drivers/infiniband/hw/mlx5/fs.c:1664:26: warning: array of flexible
structures

Avoid it by simply splitting array into two separate structs.

Link: https://patch.msgid.link/7b891b96a9fc053d01284c184d25ae98d35db2d4.1747827041.git.leon@kernel.org
Reviewed-by: Zhu Yanjun <yanjun.zhu@linux.dev>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
---
 drivers/infiniband/hw/mlx5/fs.c | 58 ++++++++++++---------------------
 1 file changed, 21 insertions(+), 37 deletions(-)

diff --git a/drivers/infiniband/hw/mlx5/fs.c b/drivers/infiniband/hw/mlx5/fs.c
index 251246c73b33..1d67f0ce6b59 100644
--- a/drivers/infiniband/hw/mlx5/fs.c
+++ b/drivers/infiniband/hw/mlx5/fs.c
@@ -1645,11 +1645,6 @@ static struct mlx5_ib_flow_handler *create_flow_rule(struct mlx5_ib_dev *dev,
 	return _create_flow_rule(dev, ft_prio, flow_attr, dst, 0, NULL);
 }
 
-enum {
-	LEFTOVERS_MC,
-	LEFTOVERS_UC,
-};
-
 static struct mlx5_ib_flow_handler *create_leftovers_rule(struct mlx5_ib_dev *dev,
 							  struct mlx5_ib_flow_prio *ft_prio,
 							  struct ib_flow_attr *flow_attr,
@@ -1659,43 +1654,32 @@ static struct mlx5_ib_flow_handler *create_leftovers_rule(struct mlx5_ib_dev *de
 	struct mlx5_ib_flow_handler *handler = NULL;
 
 	static struct {
-		struct ib_flow_attr	flow_attr;
 		struct ib_flow_spec_eth eth_flow;
-	} leftovers_specs[] = {
-		[LEFTOVERS_MC] = {
-			.flow_attr = {
-				.num_of_specs = 1,
-				.size = sizeof(leftovers_specs[0])
-			},
-			.eth_flow = {
-				.type = IB_FLOW_SPEC_ETH,
-				.size = sizeof(struct ib_flow_spec_eth),
-				.mask = {.dst_mac = {0x1} },
-				.val =  {.dst_mac = {0x1} }
-			}
-		},
-		[LEFTOVERS_UC] = {
-			.flow_attr = {
-				.num_of_specs = 1,
-				.size = sizeof(leftovers_specs[0])
-			},
-			.eth_flow = {
-				.type = IB_FLOW_SPEC_ETH,
-				.size = sizeof(struct ib_flow_spec_eth),
-				.mask = {.dst_mac = {0x1} },
-				.val = {.dst_mac = {} }
-			}
-		}
-	};
+		struct ib_flow_attr	flow_attr;
+	} leftovers_wc = { .flow_attr = { .num_of_specs = 1,
+					  .size = sizeof(leftovers_wc) },
+			   .eth_flow = {
+				   .type = IB_FLOW_SPEC_ETH,
+				   .size = sizeof(struct ib_flow_spec_eth),
+				   .mask = { .dst_mac = { 0x1 } },
+				   .val = { .dst_mac = { 0x1 } } } };
 
-	handler = create_flow_rule(dev, ft_prio,
-				   &leftovers_specs[LEFTOVERS_MC].flow_attr,
-				   dst);
+	static struct {
+		struct ib_flow_spec_eth eth_flow;
+		struct ib_flow_attr	flow_attr;
+	} leftovers_uc = { .flow_attr = { .num_of_specs = 1,
+					  .size = sizeof(leftovers_uc) },
+			   .eth_flow = {
+				   .type = IB_FLOW_SPEC_ETH,
+				   .size = sizeof(struct ib_flow_spec_eth),
+				   .mask = { .dst_mac = { 0x1 } },
+				   .val = { .dst_mac = {} } } };
+
+	handler = create_flow_rule(dev, ft_prio, &leftovers_wc.flow_attr, dst);
 	if (!IS_ERR(handler) &&
 	    flow_attr->type == IB_FLOW_ATTR_ALL_DEFAULT) {
 		handler_ucast = create_flow_rule(dev, ft_prio,
-						 &leftovers_specs[LEFTOVERS_UC].flow_attr,
-						 dst);
+						 &leftovers_uc.flow_attr, dst);
 		if (IS_ERR(handler_ucast)) {
 			mlx5_del_flow_rules(handler->rule);
 			ft_prio->refcount--;

From 6e7272afc9ab556be01de19b54bea3a1f9a9668f Mon Sep 17 00:00:00 2001
From: Junxian Huang <huangjunxian6@hisilicon.com>
Date: Fri, 23 May 2025 10:34:33 +0800
Subject: [PATCH 55/57] RDMA/hns: Fix endian issue in trace events

Avoid using __le32 directly in trace events to fix sparse complains:

drivers/infiniband/hw/hns/./hns_roce_trace.h:48:1: sparse: sparse: cast to restricted __le32
drivers/infiniband/hw/hns/./hns_roce_trace.h:48:1: sparse: sparse: restricted __le32 degrades to integer
drivers/infiniband/hw/hns/./hns_roce_trace.h:48:1: sparse: sparse: restricted __le32 degrades to integer
drivers/infiniband/hw/hns/./hns_roce_trace.h:90:1: sparse: sparse: cast to restricted __le32
drivers/infiniband/hw/hns/./hns_roce_trace.h:90:1: sparse: sparse: restricted __le32 degrades to integer
drivers/infiniband/hw/hns/./hns_roce_trace.h:90:1: sparse: sparse: restricted __le32 degrades to integer
drivers/infiniband/hw/hns/./hns_roce_trace.h:173:1: sparse: sparse: cast to restricted __le32
drivers/infiniband/hw/hns/./hns_roce_trace.h:173:1: sparse: sparse: restricted __le32 degrades to integer
drivers/infiniband/hw/hns/./hns_roce_trace.h:173:1: sparse: sparse: restricted __le32 degrades to integer

Fixes: 6c98c8670806 ("RDMA/hns: Add trace for WQE dumping")
Fixes: 1e63e2f96613 ("RDMA/hns: Add trace for AEQE dumping")
Fixes: 6bd18dabf1c9 ("RDMA/hns: Add trace for CMDQ dumping")
Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202505170327.TNOpreil-lkp@intel.com/
Signed-off-by: Junxian Huang <huangjunxian6@hisilicon.com>
Link: https://patch.msgid.link/20250523023433.2171003-1-huangjunxian6@hisilicon.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/hw/hns/hns_roce_trace.h | 15 +++++++++------
 1 file changed, 9 insertions(+), 6 deletions(-)

diff --git a/drivers/infiniband/hw/hns/hns_roce_trace.h b/drivers/infiniband/hw/hns/hns_roce_trace.h
index 23cbdbaeffaa..59ceb591b3a1 100644
--- a/drivers/infiniband/hw/hns/hns_roce_trace.h
+++ b/drivers/infiniband/hw/hns/hns_roce_trace.h
@@ -52,7 +52,7 @@ DECLARE_EVENT_CLASS(wqe_template,
 
 		    TP_STRUCT__entry(__field(unsigned long, qpn)
 				     __field(u32, idx)
-				     __array(__le32, wqe,
+				     __array(u32, wqe,
 					     MAX_WQE_SIZE / sizeof(__le32))
 				     __field(u32, len)
 				     __field(u64, id)
@@ -62,9 +62,10 @@ DECLARE_EVENT_CLASS(wqe_template,
 		    TP_fast_assign(__entry->qpn = qpn;
 				   __entry->idx = idx;
 				   __entry->id = id;
-				   memcpy(__entry->wqe, wqe, len);
 				   __entry->len = len / sizeof(__le32);
 				   __entry->type = type;
+				   for (int i = 0; i < __entry->len; i++)
+					__entry->wqe[i] = le32_to_cpu(((__le32 *)wqe)[i]);
 				   ),
 
 		    TP_printk("%s 0x%lx wqe(0x%x/0x%llx): %s",
@@ -92,14 +93,15 @@ TRACE_EVENT(hns_ae_info,
 	    TP_ARGS(event_type, aeqe, len),
 
 	    TP_STRUCT__entry(__field(int, event_type)
-			     __array(__le32, aeqe,
+			     __array(u32, aeqe,
 				     HNS_ROCE_V3_EQE_SIZE / sizeof(__le32))
 			     __field(u32, len)
 	    ),
 
 	    TP_fast_assign(__entry->event_type = event_type;
 			   __entry->len = len / sizeof(__le32);
-			   memcpy(__entry->aeqe, aeqe, len);
+			   for (int i = 0; i < __entry->len; i++)
+				__entry->aeqe[i] = le32_to_cpu(((__le32 *)aeqe)[i]);
 	    ),
 
 	    TP_printk("event %2d aeqe: %s", __entry->event_type,
@@ -179,14 +181,15 @@ DECLARE_EVENT_CLASS(cmdq,
 				     __field(u16, opcode)
 				     __field(u16, flag)
 				     __field(u16, retval)
-				     __array(__le32, data, 6)
+				     __array(u32, data, 6)
 		    ),
 
 		    TP_fast_assign(__assign_str(dev_name);
 				   __entry->opcode = le16_to_cpu(desc->opcode);
 				   __entry->flag = le16_to_cpu(desc->flag);
 				   __entry->retval = le16_to_cpu(desc->retval);
-				   memcpy(__entry->data, desc->data, 6 * sizeof(__le32));
+				   for (int i = 0; i < 6; i++)
+					__entry->data[i] = le32_to_cpu(desc->data[i]);
 		    ),
 
 		    TP_printk("%s cmdq opcode:0x%x, flag:0x%x, retval:0x%x, data:%s\n",

From 260ce16e579d375ba8f1ac945308343522f98d50 Mon Sep 17 00:00:00 2001
From: Ajit Khaparde <ajit.khaparde@broadcom.com>
Date: Fri, 23 May 2025 13:29:52 +0530
Subject: [PATCH 56/57] RDMA/bnxt_re: Support extended stats for Thor2 VF

The driver currently checks if the user is querying VF RoCE statistics.
It will not send the query_roce_stats_ext HWRM command if it is for a
VF. But Thor2 VF can support extended statistics.
Allow query of extended stats for Thor2 VFs.

Signed-off-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Signed-off-by: Shravya KN <shravya.k-n@broadcom.com>
Link: https://patch.msgid.link/20250523075952.1267827-1-kalesh-anakkur.purayil@broadcom.com
Reviewed-by: Damodharam Ammepalli <damodharam.ammepalli@broadcom.com>
Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
---
 drivers/infiniband/hw/bnxt_re/qplib_fp.c | 2 +-
 drivers/infiniband/hw/bnxt_re/qplib_sp.c | 7 ++++++-
 2 files changed, 7 insertions(+), 2 deletions(-)

diff --git a/drivers/infiniband/hw/bnxt_re/qplib_fp.c b/drivers/infiniband/hw/bnxt_re/qplib_fp.c
index 457eecb99f96..be34c605d516 100644
--- a/drivers/infiniband/hw/bnxt_re/qplib_fp.c
+++ b/drivers/infiniband/hw/bnxt_re/qplib_fp.c
@@ -1113,7 +1113,7 @@ int bnxt_qplib_create_qp(struct bnxt_qplib_res *res, struct bnxt_qplib_qp *qp)
 		qp_flags |= CMDQ_CREATE_QP_QP_FLAGS_FORCE_COMPLETION;
 	if (qp->wqe_mode == BNXT_QPLIB_WQE_MODE_VARIABLE)
 		qp_flags |= CMDQ_CREATE_QP_QP_FLAGS_VARIABLE_SIZED_WQE_ENABLED;
-	if (_is_ext_stats_supported(res->dattr->dev_cap_flags) && !res->is_vf)
+	if (bnxt_ext_stats_supported(res->cctx, res->dattr->dev_cap_flags, res->is_vf))
 		qp_flags |= CMDQ_CREATE_QP_QP_FLAGS_EXT_STATS_ENABLED;
 
 	req.qp_flags = cpu_to_le32(qp_flags);
diff --git a/drivers/infiniband/hw/bnxt_re/qplib_sp.c b/drivers/infiniband/hw/bnxt_re/qplib_sp.c
index f231e886ad9d..9efd32a3dc55 100644
--- a/drivers/infiniband/hw/bnxt_re/qplib_sp.c
+++ b/drivers/infiniband/hw/bnxt_re/qplib_sp.c
@@ -846,7 +846,12 @@ int bnxt_qplib_qext_stat(struct bnxt_qplib_rcfw *rcfw, u32 fid,
 
 	req.resp_size = sbuf.size / BNXT_QPLIB_CMDQE_UNITS;
 	req.resp_addr = cpu_to_le64(sbuf.dma_addr);
-	req.function_id = cpu_to_le32(fid);
+	if (bnxt_qplib_is_chip_gen_p7(rcfw->res->cctx) && rcfw->res->is_vf)
+		req.function_id =
+			cpu_to_le32(CMDQ_QUERY_ROCE_STATS_EXT_VF_VALID |
+				    (fid << CMDQ_QUERY_ROCE_STATS_EXT_VF_NUM_SFT));
+	else
+		req.function_id = cpu_to_le32(fid);
 	req.flags = cpu_to_le16(CMDQ_QUERY_ROCE_STATS_EXT_FLAGS_FUNCTION_ID);
 
 	bnxt_qplib_fill_cmdqmsg(&msg, &req, &resp, &sbuf, sizeof(req),

From 92a251c3df8ea1991cd9fe00f1ab0cfce18d7711 Mon Sep 17 00:00:00 2001
From: Jack Morgenstein <jackm@nvidia.com>
Date: Wed, 21 May 2025 14:36:02 +0300
Subject: [PATCH 57/57] RDMA/cma: Fix hang when cma_netevent_callback fails to
 queue_work

The cited commit fixed a crash when cma_netevent_callback was called for
a cma_id while work on that id from a previous call had not yet started.
The work item was re-initialized in the second call, which corrupted the
work item currently in the work queue.

However, it left a problem when queue_work fails (because the item is
still pending in the work queue from a previous call). In this case,
cma_id_put (which is called in the work handler) is therefore not
called. This results in a userspace process hang (zombie process).

Fix this by calling cma_id_put() if queue_work fails.

Fixes: 45f5dcdd0497 ("RDMA/cma: Fix workqueue crash in cma_netevent_work_handler")
Link: https://patch.msgid.link/r/4f3640b501e48d0166f312a64fdadf72b059bd04.1747827103.git.leon@kernel.org
Signed-off-by: Jack Morgenstein <jackm@nvidia.com>
Signed-off-by: Feng Liu <feliu@nvidia.com>
Reviewed-by: Vlad Dumitrescu <vdumitrescu@nvidia.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
Reviewed-by: Sharath Srinivasan <sharath.srinivasan@oracle.com>
Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
---
 drivers/infiniband/core/cma.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c
index ce942d300f1e..9b471548e7ae 100644
--- a/drivers/infiniband/core/cma.c
+++ b/drivers/infiniband/core/cma.c
@@ -5231,7 +5231,8 @@ static int cma_netevent_callback(struct notifier_block *self,
 			   neigh->ha, ETH_ALEN))
 			continue;
 		cma_id_get(current_id);
-		queue_work(cma_wq, &current_id->id.net_work);
+		if (!queue_work(cma_wq, &current_id->id.net_work))
+			cma_id_put(current_id);
 	}
 out:
 	spin_unlock_irqrestore(&id_table_lock, flags);