[PATCH rdma 0/3] RDMA/qedr: User-to-kernel space patches

[PATCH rdma 0/3] RDMA/qedr: User-to-kernel space patches

[Ram Amrani]

The first patch is a fix that prevents memory overrun.
Thanks Jason.

The other two add new parameters that are passed to rdma-core.

Ram Amrani (3):
  RDMA/qedr: Prevent memory overrun in verbs' user responses
  RDMA/qedr: notify user application if DPM is supported
  RDMA/qedr: notify user application of supported WIDs

 drivers/infiniband/hw/qedr/main.c  |  1 +
 drivers/infiniband/hw/qedr/qedr.h  |  2 ++
 drivers/infiniband/hw/qedr/verbs.c | 19 +++++++++++++++----
 include/uapi/rdma/qedr-abi.h       |  3 +++
 4 files changed, 21 insertions(+), 4 deletions(-)

-- 
1.8.3.1

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH rdma 1/3] RDMA/qedr: Prevent memory overrun in verbs' user responses

[Ram Amrani]

Wrap ib_copy_to_udata with a function that ensures that the data
being copied over to user space isn't longer than the allowed.

Fixes: cecbcddf6461 ("qedr: Add support for QP verbs")
Fixes: a7efd7773e31 ("qedr: Add support for PD,PKEY and CQ verbs")
Fixes: ac1b36e55a51 ("qedr: Add support for user context verbs")
Signed-off-by: Ram Amrani <Ram.Amrani-YGCgFSpz5w/QT0dZR+AlfA@public.gmane.org>
---
 drivers/infiniband/hw/qedr/verbs.c | 16 ++++++++++++----
 1 file changed, 12 insertions(+), 4 deletions(-)

diff --git a/drivers/infiniband/hw/qedr/verbs.c b/drivers/infiniband/hw/qedr/verbs.c
index 17685cf..3ec861c 100644
--- a/drivers/infiniband/hw/qedr/verbs.c
+++ b/drivers/infiniband/hw/qedr/verbs.c
@@ -53,6 +53,14 @@
 
 #define DB_ADDR_SHIFT(addr)		((addr) << DB_PWM_ADDR_OFFSET_SHIFT)
 
+static inline int qedr_ib_copy_to_udata(struct ib_udata *udata, void *src,
+					size_t len)
+{
+	size_t min_len = min_t(size_t, len, udata->outlen);
+
+	return ib_copy_to_udata(udata, src, min_len);
+}
+
 int qedr_query_pkey(struct ib_device *ibdev, u8 port, u16 index, u16 *pkey)
 {
 	if (index > QEDR_ROCE_PKEY_TABLE_LEN)
@@ -378,7 +386,7 @@ struct ib_ucontext *qedr_alloc_ucontext(struct ib_device *ibdev,
 	uresp.sges_per_srq_wr = dev->attr.max_srq_sge;
 	uresp.max_cqes = QEDR_MAX_CQES;
 
-	rc = ib_copy_to_udata(udata, &uresp, sizeof(uresp));
+	rc = qedr_ib_copy_to_udata(udata, &uresp, sizeof(uresp));
 	if (rc)
 		goto err;
 
@@ -499,7 +507,7 @@ struct ib_pd *qedr_alloc_pd(struct ib_device *ibdev,
 
 		uresp.pd_id = pd_id;
 
-		rc = ib_copy_to_udata(udata, &uresp, sizeof(uresp));
+		rc = qedr_ib_copy_to_udata(udata, &uresp, sizeof(uresp));
 		if (rc) {
 			DP_ERR(dev, "copy error pd_id=0x%x.\n", pd_id);
 			dev->ops->rdma_dealloc_pd(dev->rdma_ctx, pd_id);
@@ -722,7 +730,7 @@ static int qedr_copy_cq_uresp(struct qedr_dev *dev,
 	uresp.db_offset = DB_ADDR_SHIFT(DQ_PWM_OFFSET_UCM_RDMA_CQ_CONS_32BIT);
 	uresp.icid = cq->icid;
 
-	rc = ib_copy_to_udata(udata, &uresp, sizeof(uresp));
+	rc = qedr_ib_copy_to_udata(udata, &uresp, sizeof(uresp));
 	if (rc)
 		DP_ERR(dev, "copy error cqid=0x%x.\n", cq->icid);
 
@@ -1228,7 +1236,7 @@ static int qedr_copy_qp_uresp(struct qedr_dev *dev,
 	uresp.atomic_supported = dev->atomic_cap != IB_ATOMIC_NONE;
 	uresp.qp_id = qp->qp_id;
 
-	rc = ib_copy_to_udata(udata, &uresp, sizeof(uresp));
+	rc = qedr_ib_copy_to_udata(udata, &uresp, sizeof(uresp));
 	if (rc)
 		DP_ERR(dev,
 		       "create qp: failed a copy to user space with qp icid=0x%x.\n",
-- 
1.8.3.1

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH rdma 1/3] RDMA/qedr: Prevent memory overrun in verbs' user responses

[Doug Ledford]

[-- Attachment #1.1: Type: text/plain, Size: 683 bytes --]

On 6/26/2017 12:05 PM, Ram Amrani wrote:
> Wrap ib_copy_to_udata with a function that ensures that the data
> being copied over to user space isn't longer than the allowed.
> 
> Fixes: cecbcddf6461 ("qedr: Add support for QP verbs")
> Fixes: a7efd7773e31 ("qedr: Add support for PD,PKEY and CQ verbs")
> Fixes: ac1b36e55a51 ("qedr: Add support for user context verbs")
> Signed-off-by: Ram Amrani <Ram.Amrani-YGCgFSpz5w/QT0dZR+AlfA@public.gmane.org>

This was accepted into 4.13-rc, thanks.

-- 
Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
    GPG Key ID: B826A3330E572FDD
    Key fingerprint = AE6B 1BDA 122B 23B4 265B  1274 B826 A333 0E57 2FDD


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 884 bytes --]

[PATCH rdma 2/3] RDMA/qedr: notify user application if DPM is supported

[Ram Amrani]

Direct Packet Mode support may be disabled, e.g, due to limited
resources. Notifying the user application prevents wasting cycles
on attempting to send these kind of packets.

Signed-off-by: Ram Amrani <Ram.Amrani-YGCgFSpz5w/QT0dZR+AlfA@public.gmane.org>
---
 drivers/infiniband/hw/qedr/main.c  | 1 +
 drivers/infiniband/hw/qedr/qedr.h  | 2 ++
 drivers/infiniband/hw/qedr/verbs.c | 1 +
 include/uapi/rdma/qedr-abi.h       | 1 +
 4 files changed, 5 insertions(+)

diff --git a/drivers/infiniband/hw/qedr/main.c b/drivers/infiniband/hw/qedr/main.c
index 6a72095..d228718 100644
--- a/drivers/infiniband/hw/qedr/main.c
+++ b/drivers/infiniband/hw/qedr/main.c
@@ -778,6 +778,7 @@ static struct qedr_dev *qedr_add(struct qed_dev *cdev, struct pci_dev *pdev,
 	if (rc)
 		goto init_err;
 
+	dev->user_dpm_enabled = dev_info.user_dpm_enabled;
 	dev->num_hwfns = dev_info.common.num_hwfns;
 	dev->rdma_ctx = dev->ops->rdma_get_rdma_ctx(cdev);
 
diff --git a/drivers/infiniband/hw/qedr/qedr.h b/drivers/infiniband/hw/qedr/qedr.h
index aa08c76..42af9b6 100644
--- a/drivers/infiniband/hw/qedr/qedr.h
+++ b/drivers/infiniband/hw/qedr/qedr.h
@@ -158,6 +158,8 @@ struct qedr_dev {
 	struct qedr_qp		*gsi_qp;
 
 	unsigned long enet_state;
+
+	u8 user_dpm_enabled;
 };
 
 #define QEDR_MAX_SQ_PBL			(0x8000)
diff --git a/drivers/infiniband/hw/qedr/verbs.c b/drivers/infiniband/hw/qedr/verbs.c
index 3ec861c..d655364 100644
--- a/drivers/infiniband/hw/qedr/verbs.c
+++ b/drivers/infiniband/hw/qedr/verbs.c
@@ -376,6 +376,7 @@ struct ib_ucontext *qedr_alloc_ucontext(struct ib_device *ibdev,
 
 	memset(&uresp, 0, sizeof(uresp));
 
+	uresp.dpm_enabled = dev->user_dpm_enabled;
 	uresp.db_pa = ctx->dpi_phys_addr;
 	uresp.db_size = ctx->dpi_size;
 	uresp.max_send_wr = dev->attr.max_sqe;
diff --git a/include/uapi/rdma/qedr-abi.h b/include/uapi/rdma/qedr-abi.h
index 75c270d..2684004 100644
--- a/include/uapi/rdma/qedr-abi.h
+++ b/include/uapi/rdma/qedr-abi.h
@@ -49,6 +49,7 @@ struct qedr_alloc_ucontext_resp {
 	__u32 sges_per_recv_wr;
 	__u32 sges_per_srq_wr;
 	__u32 max_cqes;
+	__u8 dpm_enabled;
 };
 
 struct qedr_alloc_pd_ureq {
-- 
1.8.3.1

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH rdma 3/3] RDMA/qedr: notify user application of supported WIDs

[Ram Amrani]

The number of supported WIDs, if they are supported at all, can be
limited due to resources. Notifying the user space application the
number of available WIDs allows it to utilize them correctly.

Signed-off-by: Ram Amrani <Ram.Amrani-YGCgFSpz5w/QT0dZR+AlfA@public.gmane.org>
---
 drivers/infiniband/hw/qedr/verbs.c | 2 ++
 include/uapi/rdma/qedr-abi.h       | 2 ++
 2 files changed, 4 insertions(+)

diff --git a/drivers/infiniband/hw/qedr/verbs.c b/drivers/infiniband/hw/qedr/verbs.c
index d655364..b11da86 100644
--- a/drivers/infiniband/hw/qedr/verbs.c
+++ b/drivers/infiniband/hw/qedr/verbs.c
@@ -377,6 +377,8 @@ struct ib_ucontext *qedr_alloc_ucontext(struct ib_device *ibdev,
 	memset(&uresp, 0, sizeof(uresp));
 
 	uresp.dpm_enabled = dev->user_dpm_enabled;
+	uresp.wids_enabled = 1;
+	uresp.wid_count = oparams.wid_count;
 	uresp.db_pa = ctx->dpi_phys_addr;
 	uresp.db_size = ctx->dpi_size;
 	uresp.max_send_wr = dev->attr.max_sqe;
diff --git a/include/uapi/rdma/qedr-abi.h b/include/uapi/rdma/qedr-abi.h
index 2684004..54b6435 100644
--- a/include/uapi/rdma/qedr-abi.h
+++ b/include/uapi/rdma/qedr-abi.h
@@ -50,6 +50,8 @@ struct qedr_alloc_ucontext_resp {
 	__u32 sges_per_srq_wr;
 	__u32 max_cqes;
 	__u8 dpm_enabled;
+	__u8 wids_enabled;
+	__u16 wid_count;
 };
 
 struct qedr_alloc_pd_ureq {
-- 
1.8.3.1

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH rdma 0/3] RDMA/qedr: User-to-kernel space patches

[Doug Ledford]

On Mon, 2017-06-26 at 19:05 +0300, Ram Amrani wrote:
> The first patch is a fix that prevents memory overrun.
> Thanks Jason.
> 
> The other two add new parameters that are passed to rdma-core.
> 
> Ram Amrani (3):
>   RDMA/qedr: Prevent memory overrun in verbs' user responses
>   RDMA/qedr: notify user application if DPM is supported
>   RDMA/qedr: notify user application of supported WIDs
> 
>  drivers/infiniband/hw/qedr/main.c  |  1 +
>  drivers/infiniband/hw/qedr/qedr.h  |  2 ++
>  drivers/infiniband/hw/qedr/verbs.c | 19 +++++++++++++++----
>  include/uapi/rdma/qedr-abi.h       |  3 +++
>  4 files changed, 21 insertions(+), 4 deletions(-)

The first patch was a fix and made it into -rc.  The other two were not
really fixes, but have been applied to -next.  Thanks.

-- 
Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
    GPG KeyID: B826A3330E572FDD
    Key fingerprint = AE6B 1BDA 122B 23B4 265B  1274 B826 A333 0E57 2FDD

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html