[PATCH rdma-next 0/3] bnxt_re driver update

[PATCH rdma-next 0/3] bnxt_re driver update

[Kalesh AP]

This patchset adds support for 2G page size in the bnxt_re driver.
There are couple of cosmetic changes as well.

Kalesh AP (2):
  RDMA/bnxt_re: Fix size of uverbs_copy_to() in
    BNXT_RE_METHOD_GET_TOGGLE_MEM
  RDMA/bnxt_re: Use macro instead of hard coded value

Selvin Xavier (1):
  RDMA/bnxt_re: Support 2G message size

 drivers/infiniband/hw/bnxt_re/ib_verbs.c |  2 +-
 drivers/infiniband/hw/bnxt_re/qplib_fp.c | 28 ++++++++++++------------
 drivers/infiniband/hw/bnxt_re/qplib_fp.h |  3 +++
 drivers/infiniband/hw/bnxt_re/qplib_sp.c |  2 +-
 drivers/infiniband/hw/bnxt_re/qplib_sp.h |  1 +
 5 files changed, 20 insertions(+), 16 deletions(-)

-- 
2.43.5

[PATCH rdma-next 1/3] RDMA/bnxt_re: Fix size of uverbs_copy_to() in BNXT_RE_METHOD_GET_TOGGLE_MEM

[Kalesh AP]

Since both "length" and "offset" are of type u32, there is
no functional issue here.

Reviewed-by: Saravanan Vajravel <saravanan.vajravel@broadcom.com>
Signed-off-by: Shravya KN <shravya.k-n@broadcom.com>
Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
---
 drivers/infiniband/hw/bnxt_re/ib_verbs.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/infiniband/hw/bnxt_re/ib_verbs.c b/drivers/infiniband/hw/bnxt_re/ib_verbs.c
index 063801384b2b..3a627acb82ce 100644
--- a/drivers/infiniband/hw/bnxt_re/ib_verbs.c
+++ b/drivers/infiniband/hw/bnxt_re/ib_verbs.c
@@ -4738,7 +4738,7 @@ static int UVERBS_HANDLER(BNXT_RE_METHOD_GET_TOGGLE_MEM)(struct uverbs_attr_bund
 		return err;
 
 	err = uverbs_copy_to(attrs, BNXT_RE_TOGGLE_MEM_MMAP_OFFSET,
-			     &offset, sizeof(length));
+			     &offset, sizeof(offset));
 	if (err)
 		return err;
 
-- 
2.43.5

[PATCH rdma-next 2/3] RDMA/bnxt_re: Support 2G message size

[Kalesh AP]

From: Selvin Xavier <selvin.xavier@broadcom.com>

bnxt_qplib_put_sges is calculating the length in
a signed int. So handling the 2G message size
is not working since it is considered as negative.

Use a unsigned number to calculate the total message
length. As per the spec, IB message size shall be
between zero and 2^31 bytes (inclusive). So adding
a check for 2G message size.

Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com>
Signed-off-by: Shravya KN <shravya.k-n@broadcom.com>
Reviewed-by: Saravanan Vajravel <saravanan.vajravel@broadcom.com>
Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
---
 drivers/infiniband/hw/bnxt_re/qplib_fp.c | 28 ++++++++++++------------
 drivers/infiniband/hw/bnxt_re/qplib_fp.h |  3 +++
 2 files changed, 17 insertions(+), 14 deletions(-)

diff --git a/drivers/infiniband/hw/bnxt_re/qplib_fp.c b/drivers/infiniband/hw/bnxt_re/qplib_fp.c
index be34c605d516..dfe3177123e5 100644
--- a/drivers/infiniband/hw/bnxt_re/qplib_fp.c
+++ b/drivers/infiniband/hw/bnxt_re/qplib_fp.c
@@ -1750,9 +1750,9 @@ static void bnxt_qplib_fill_psn_search(struct bnxt_qplib_qp *qp,
 	}
 }
 
-static int bnxt_qplib_put_inline(struct bnxt_qplib_qp *qp,
-				 struct bnxt_qplib_swqe *wqe,
-				 u16 *idx)
+static unsigned int bnxt_qplib_put_inline(struct bnxt_qplib_qp *qp,
+					  struct bnxt_qplib_swqe *wqe,
+					  u32 *idx)
 {
 	struct bnxt_qplib_hwq *hwq;
 	int len, t_len, offt;
@@ -1769,7 +1769,7 @@ static int bnxt_qplib_put_inline(struct bnxt_qplib_qp *qp,
 		il_src = (void *)wqe->sg_list[indx].addr;
 		t_len += len;
 		if (t_len > qp->max_inline_data)
-			return -ENOMEM;
+			return BNXT_RE_INVAL_MSG_SIZE;
 		while (len) {
 			if (pull_dst) {
 				pull_dst = false;
@@ -1795,9 +1795,9 @@ static int bnxt_qplib_put_inline(struct bnxt_qplib_qp *qp,
 	return t_len;
 }
 
-static u32 bnxt_qplib_put_sges(struct bnxt_qplib_hwq *hwq,
-			       struct bnxt_qplib_sge *ssge,
-			       u16 nsge, u16 *idx)
+static unsigned int bnxt_qplib_put_sges(struct bnxt_qplib_hwq *hwq,
+					struct bnxt_qplib_sge *ssge,
+					u32 nsge, u32 *idx)
 {
 	struct sq_sge *dsge;
 	int indx, len = 0;
@@ -1878,14 +1878,12 @@ int bnxt_qplib_post_send(struct bnxt_qplib_qp *qp,
 	struct bnxt_qplib_hwq *hwq;
 	struct bnxt_qplib_swq *swq;
 	bool sch_handler = false;
+	u32 wqe_idx, slots, idx;
 	u16 wqe_sz, qdf = 0;
 	bool msn_update;
 	void *base_hdr;
 	void *ext_hdr;
 	__le32 temp32;
-	u32 wqe_idx;
-	u32 slots;
-	u16 idx;
 
 	hwq = &sq->hwq;
 	if (qp->state != CMDQ_MODIFY_QP_NEW_STATE_RTS &&
@@ -1937,8 +1935,10 @@ int bnxt_qplib_post_send(struct bnxt_qplib_qp *qp,
 	else
 		data_len = bnxt_qplib_put_sges(hwq, wqe->sg_list, wqe->num_sge,
 					       &idx);
-	if (data_len < 0)
-		goto queue_err;
+	if (data_len > BNXT_RE_MAX_MSG_SIZE) {
+		rc = -EINVAL;
+		goto done;
+	}
 	/* Make sure we update MSN table only for wired wqes */
 	msn_update = true;
 	/* Specifics */
@@ -2139,8 +2139,8 @@ int bnxt_qplib_post_recv(struct bnxt_qplib_qp *qp,
 	struct bnxt_qplib_hwq *hwq;
 	struct bnxt_qplib_swq *swq;
 	bool sch_handler = false;
-	u16 wqe_sz, idx;
-	u32 wqe_idx;
+	u32 wqe_idx, idx;
+	u16 wqe_sz;
 	int rc = 0;
 
 	hwq = &rq->hwq;
diff --git a/drivers/infiniband/hw/bnxt_re/qplib_fp.h b/drivers/infiniband/hw/bnxt_re/qplib_fp.h
index 0d9487c889ff..ab125f1d949e 100644
--- a/drivers/infiniband/hw/bnxt_re/qplib_fp.h
+++ b/drivers/infiniband/hw/bnxt_re/qplib_fp.h
@@ -346,6 +346,9 @@ struct bnxt_qplib_qp {
 	u8				tos_dscp;
 };
 
+#define BNXT_RE_MAX_MSG_SIZE	0x80000000
+#define BNXT_RE_INVAL_MSG_SIZE	0xFFFFFFFF
+
 #define BNXT_QPLIB_MAX_CQE_ENTRY_SIZE	sizeof(struct cq_base)
 
 #define CQE_CNT_PER_PG		(PAGE_SIZE / BNXT_QPLIB_MAX_CQE_ENTRY_SIZE)
-- 
2.43.5

[PATCH rdma-next 3/3] RDMA/bnxt_re: Use macro instead of hard coded value

[Kalesh AP]

1. Defined a macro for the hard coded value.
2. "access" field in the request structure is of type "u8".
   Updated the mask accordingly.

Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Signed-off-by: Shravya KN <shravya.k-n@broadcom.com>
Reviewed-by: Hongguang Gao <hongguang.gao@broadcom.com>
---
 drivers/infiniband/hw/bnxt_re/qplib_sp.c | 2 +-
 drivers/infiniband/hw/bnxt_re/qplib_sp.h | 1 +
 2 files changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/infiniband/hw/bnxt_re/qplib_sp.c b/drivers/infiniband/hw/bnxt_re/qplib_sp.c
index 9efd32a3dc55..68981399598d 100644
--- a/drivers/infiniband/hw/bnxt_re/qplib_sp.c
+++ b/drivers/infiniband/hw/bnxt_re/qplib_sp.c
@@ -674,7 +674,7 @@ int bnxt_qplib_reg_mr(struct bnxt_qplib_res *res, struct bnxt_qplib_mrw *mr,
 	req.log2_pbl_pg_size = cpu_to_le16(((ilog2(PAGE_SIZE) <<
 				 CMDQ_REGISTER_MR_LOG2_PBL_PG_SIZE_SFT) &
 				CMDQ_REGISTER_MR_LOG2_PBL_PG_SIZE_MASK));
-	req.access = (mr->access_flags & 0xFFFF);
+	req.access = (mr->access_flags & BNXT_QPLIB_MR_ACCESS_MASK);
 	req.va = cpu_to_le64(mr->va);
 	req.key = cpu_to_le32(mr->lkey);
 	if (_is_alloc_mr_unified(res->dattr->dev_cap_flags))
diff --git a/drivers/infiniband/hw/bnxt_re/qplib_sp.h b/drivers/infiniband/hw/bnxt_re/qplib_sp.h
index e626b05038a1..09faf4a1e849 100644
--- a/drivers/infiniband/hw/bnxt_re/qplib_sp.h
+++ b/drivers/infiniband/hw/bnxt_re/qplib_sp.h
@@ -111,6 +111,7 @@ struct bnxt_qplib_mrw {
 	struct bnxt_qplib_pd		*pd;
 	int				type;
 	u32				access_flags;
+#define BNXT_QPLIB_MR_ACCESS_MASK	0xFF
 #define BNXT_QPLIB_FR_PMR		0x80000000
 	u32				lkey;
 	u32				rkey;
-- 
2.43.5

Re: [PATCH rdma-next 2/3] RDMA/bnxt_re: Support 2G message size

[Leon Romanovsky]

On Fri, Jul 04, 2025 at 10:08:56AM +0530, Kalesh AP wrote:
> From: Selvin Xavier <selvin.xavier@broadcom.com>
> 
> bnxt_qplib_put_sges is calculating the length in
> a signed int. So handling the 2G message size
> is not working since it is considered as negative.
> 
> Use a unsigned number to calculate the total message
> length. As per the spec, IB message size shall be
> between zero and 2^31 bytes (inclusive). So adding
> a check for 2G message size.

You can easily use opto 80 chars per-line in commit messages, there is
no need to limit yourself to 50 chars.

Thanks

Re: [PATCH rdma-next 0/3] bnxt_re driver update

[Leon Romanovsky]

On Fri, 04 Jul 2025 10:08:54 +0530, Kalesh AP wrote:
> This patchset adds support for 2G page size in the bnxt_re driver.
> There are couple of cosmetic changes as well.
> 
> Kalesh AP (2):
>   RDMA/bnxt_re: Fix size of uverbs_copy_to() in
>     BNXT_RE_METHOD_GET_TOGGLE_MEM
>   RDMA/bnxt_re: Use macro instead of hard coded value
> 
> [...]

Applied, thanks!

[1/3] RDMA/bnxt_re: Fix size of uverbs_copy_to() in BNXT_RE_METHOD_GET_TOGGLE_MEM
      https://git.kernel.org/rdma/rdma/c/8ba2e1eb6b3ec2
[2/3] RDMA/bnxt_re: Support 2G message size
      https://git.kernel.org/rdma/rdma/c/c6acd006f110a0
[3/3] RDMA/bnxt_re: Use macro instead of hard coded value
      https://git.kernel.org/rdma/rdma/c/bdc5726cdd454f

Best regards,
-- 
Leon Romanovsky <leon@kernel.org>

Re: [PATCH rdma-next 2/3] RDMA/bnxt_re: Support 2G message size

[Kalesh Anakkur Purayil]

[-- Attachment #1.1: Type: text/plain, Size: 829 bytes --]

On Sun, Jul 6, 2025 at 1:07 PM Leon Romanovsky <leon@kernel.org> wrote:

> On Fri, Jul 04, 2025 at 10:08:56AM +0530, Kalesh AP wrote:
> > From: Selvin Xavier <selvin.xavier@broadcom.com>
> >
> > bnxt_qplib_put_sges is calculating the length in
> > a signed int. So handling the 2G message size
> > is not working since it is considered as negative.
> >
> > Use a unsigned number to calculate the total message
> > length. As per the spec, IB message size shall be
> > between zero and 2^31 bytes (inclusive). So adding
> > a check for 2G message size.
>
> You can easily use opto 80 chars per-line in commit messages, there is
> no need to limit yourself to 50 chars.
> Sure, thank you Leon. I usually format upto 80 characters per line, not
> sure how I missed it.
> Thanks
>


-- 
Regards,
Kalesh AP

[-- Attachment #1.2: Type: text/html, Size: 1437 bytes --]

[-- Attachment #2: S/MIME Cryptographic Signature --]
[-- Type: application/pkcs7-signature, Size: 4226 bytes --]