* [PATCH 1/3] RDMA: Mark imm_data as be32 in the verbs uapi header
[not found] ` <20180111214307.501-1-jgg-uk2M96/98Pc@public.gmane.org>
@ 2018-01-11 21:43 ` Jason Gunthorpe
2018-01-11 21:43 ` [PATCH 2/3] RDMA/hns: Fix endian problems around imm_data and rkey Jason Gunthorpe
` (2 subsequent siblings)
3 siblings, 0 replies; 7+ messages in thread
From: Jason Gunthorpe @ 2018-01-11 21:43 UTC (permalink / raw)
To: linux-rdma-u79uwXL29TY76Z2rM5mHXA; +Cc: Jason Gunthorpe
From: Jason Gunthorpe <jgg-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
This matches what the userspace copy of this header has been doing
for a while. imm_data is an opaque 4 byte array carried over the network,
and invalidate_rkey is in CPU byte order.
Signed-off-by: Jason Gunthorpe <jgg-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
---
drivers/infiniband/core/uverbs_cmd.c | 2 +-
drivers/infiniband/sw/rdmavt/cq.c | 3 +--
drivers/infiniband/sw/rxe/rxe_resp.c | 3 +--
include/uapi/rdma/ib_user_verbs.h | 4 ++--
4 files changed, 5 insertions(+), 7 deletions(-)
diff --git a/drivers/infiniband/core/uverbs_cmd.c b/drivers/infiniband/core/uverbs_cmd.c
index c216d98bb816a0..4ddd61d905075c 100644
--- a/drivers/infiniband/core/uverbs_cmd.c
+++ b/drivers/infiniband/core/uverbs_cmd.c
@@ -1196,7 +1196,7 @@ static int copy_wc_to_user(struct ib_device *ib_dev, void __user *dest,
tmp.opcode = wc->opcode;
tmp.vendor_err = wc->vendor_err;
tmp.byte_len = wc->byte_len;
- tmp.ex.imm_data = (__u32 __force) wc->ex.imm_data;
+ tmp.ex.imm_data = wc->ex.imm_data;
tmp.qp_num = wc->qp->qp_num;
tmp.src_qp = wc->src_qp;
tmp.wc_flags = wc->wc_flags;
diff --git a/drivers/infiniband/sw/rdmavt/cq.c b/drivers/infiniband/sw/rdmavt/cq.c
index 61fca14c2598be..fb52b669bfce5c 100644
--- a/drivers/infiniband/sw/rdmavt/cq.c
+++ b/drivers/infiniband/sw/rdmavt/cq.c
@@ -101,8 +101,7 @@ void rvt_cq_enter(struct rvt_cq *cq, struct ib_wc *entry, bool solicited)
wc->uqueue[head].opcode = entry->opcode;
wc->uqueue[head].vendor_err = entry->vendor_err;
wc->uqueue[head].byte_len = entry->byte_len;
- wc->uqueue[head].ex.imm_data =
- (__u32 __force)entry->ex.imm_data;
+ wc->uqueue[head].ex.imm_data = entry->ex.imm_data;
wc->uqueue[head].qp_num = entry->qp->qp_num;
wc->uqueue[head].src_qp = entry->src_qp;
wc->uqueue[head].wc_flags = entry->wc_flags;
diff --git a/drivers/infiniband/sw/rxe/rxe_resp.c b/drivers/infiniband/sw/rxe/rxe_resp.c
index 4240866a533192..f07324f2cde207 100644
--- a/drivers/infiniband/sw/rxe/rxe_resp.c
+++ b/drivers/infiniband/sw/rxe/rxe_resp.c
@@ -863,8 +863,7 @@ static enum resp_states do_complete(struct rxe_qp *qp,
if (pkt->mask & RXE_IMMDT_MASK) {
uwc->wc_flags |= IB_WC_WITH_IMM;
- uwc->ex.imm_data =
- (__u32 __force)immdt_imm(pkt);
+ uwc->ex.imm_data = immdt_imm(pkt);
}
if (pkt->mask & RXE_IETH_MASK) {
diff --git a/include/uapi/rdma/ib_user_verbs.h b/include/uapi/rdma/ib_user_verbs.h
index 7e11bb8651b641..fd035641cf41ce 100644
--- a/include/uapi/rdma/ib_user_verbs.h
+++ b/include/uapi/rdma/ib_user_verbs.h
@@ -449,7 +449,7 @@ struct ib_uverbs_wc {
__u32 vendor_err;
__u32 byte_len;
union {
- __u32 imm_data;
+ __be32 imm_data;
__u32 invalidate_rkey;
} ex;
__u32 qp_num;
@@ -765,7 +765,7 @@ struct ib_uverbs_send_wr {
__u32 opcode;
__u32 send_flags;
union {
- __u32 imm_data;
+ __be32 imm_data;
__u32 invalidate_rkey;
} ex;
union {
--
2.15.1
--
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
^ permalink raw reply related [flat|nested] 7+ messages in thread* [PATCH 2/3] RDMA/hns: Fix endian problems around imm_data and rkey
[not found] ` <20180111214307.501-1-jgg-uk2M96/98Pc@public.gmane.org>
2018-01-11 21:43 ` [PATCH 1/3] RDMA: Mark imm_data as be32 in the verbs " Jason Gunthorpe
@ 2018-01-11 21:43 ` Jason Gunthorpe
[not found] ` <20180111214307.501-3-jgg-uk2M96/98Pc@public.gmane.org>
2018-01-11 21:43 ` [PATCH 3/3] RDMA/qedr: Fix endian problems around imm_data Jason Gunthorpe
2018-01-15 23:06 ` [PATCH 0/3] Annotate imm_data as be32 in the uapi header Jason Gunthorpe
3 siblings, 1 reply; 7+ messages in thread
From: Jason Gunthorpe @ 2018-01-11 21:43 UTC (permalink / raw)
To: linux-rdma-u79uwXL29TY76Z2rM5mHXA
Cc: Jason Gunthorpe, Lijun Ou, Wei Hu(Xavier)
From: Jason Gunthorpe <jgg-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
This matches the changes made recently to the userspace hns
driver when it was made sparse clean.
See rdma-core commit bffd380cfe56 ("libhns: Make the provider sparse
clean")
wc->imm_data is not used in the kernel so this change has no practical
impact.
Signed-off-by: Jason Gunthorpe <jgg-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
---
drivers/infiniband/hw/hns/hns_roce_hw_v1.c | 7 ++++---
drivers/infiniband/hw/hns/hns_roce_hw_v2.c | 6 +++---
drivers/infiniband/hw/hns/hns_roce_hw_v2.h | 5 ++++-
3 files changed, 11 insertions(+), 7 deletions(-)
diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v1.c b/drivers/infiniband/hw/hns/hns_roce_hw_v1.c
index 939355ede14ab7..833a305085efe1 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v1.c
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v1.c
@@ -2311,15 +2311,16 @@ static int hns_roce_v1_poll_one(struct hns_roce_cq *hr_cq,
case HNS_ROCE_OPCODE_RDMA_WITH_IMM_RECEIVE:
wc->opcode = IB_WC_RECV_RDMA_WITH_IMM;
wc->wc_flags = IB_WC_WITH_IMM;
- wc->ex.imm_data = le32_to_cpu(cqe->immediate_data);
+ wc->ex.imm_data =
+ cpu_to_be32(le32_to_cpu(cqe->immediate_data));
break;
case HNS_ROCE_OPCODE_SEND_DATA_RECEIVE:
if (roce_get_bit(cqe->cqe_byte_4,
CQE_BYTE_4_IMM_INDICATOR_S)) {
wc->opcode = IB_WC_RECV;
wc->wc_flags = IB_WC_WITH_IMM;
- wc->ex.imm_data = le32_to_cpu(
- cqe->immediate_data);
+ wc->ex.imm_data = cpu_to_be32(
+ le32_to_cpu(cqe->immediate_data));
} else {
wc->opcode = IB_WC_RECV;
wc->wc_flags = 0;
diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
index 2ca35e341d0912..34f13c3cb088c7 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
@@ -1709,7 +1709,7 @@ static int hns_roce_v2_poll_one(struct hns_roce_cq *hr_cq,
case HNS_ROCE_V2_OPCODE_RDMA_WRITE_IMM:
wc->opcode = IB_WC_RECV_RDMA_WITH_IMM;
wc->wc_flags = IB_WC_WITH_IMM;
- wc->ex.imm_data = le32_to_cpu(cqe->rkey_immtdata);
+ wc->ex.imm_data = cqe->immtdata;
break;
case HNS_ROCE_V2_OPCODE_SEND:
wc->opcode = IB_WC_RECV;
@@ -1718,12 +1718,12 @@ static int hns_roce_v2_poll_one(struct hns_roce_cq *hr_cq,
case HNS_ROCE_V2_OPCODE_SEND_WITH_IMM:
wc->opcode = IB_WC_RECV;
wc->wc_flags = IB_WC_WITH_IMM;
- wc->ex.imm_data = le32_to_cpu(cqe->rkey_immtdata);
+ wc->ex.imm_data = cqe->immtdata;
break;
case HNS_ROCE_V2_OPCODE_SEND_WITH_INV:
wc->opcode = IB_WC_RECV;
wc->wc_flags = IB_WC_WITH_INVALIDATE;
- wc->ex.invalidate_rkey = cqe->rkey_immtdata;
+ wc->ex.invalidate_rkey = le32_to_cpu(cqe->rkey);
break;
default:
wc->status = IB_WC_GENERAL_ERR;
diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.h b/drivers/infiniband/hw/hns/hns_roce_hw_v2.h
index 463edab9b7193f..cbf61ad2539c78 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.h
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.h
@@ -762,7 +762,10 @@ struct hns_roce_v2_qp_context {
struct hns_roce_v2_cqe {
u32 byte_4;
- u32 rkey_immtdata;
+ union {
+ __le32 rkey;
+ __be32 immtdata;
+ };
u32 byte_12;
u32 byte_16;
u32 byte_cnt;
--
2.15.1
--
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
^ permalink raw reply related [flat|nested] 7+ messages in thread* [PATCH 3/3] RDMA/qedr: Fix endian problems around imm_data
[not found] ` <20180111214307.501-1-jgg-uk2M96/98Pc@public.gmane.org>
2018-01-11 21:43 ` [PATCH 1/3] RDMA: Mark imm_data as be32 in the verbs " Jason Gunthorpe
2018-01-11 21:43 ` [PATCH 2/3] RDMA/hns: Fix endian problems around imm_data and rkey Jason Gunthorpe
@ 2018-01-11 21:43 ` Jason Gunthorpe
[not found] ` <20180111214307.501-4-jgg-uk2M96/98Pc@public.gmane.org>
2018-01-15 23:06 ` [PATCH 0/3] Annotate imm_data as be32 in the uapi header Jason Gunthorpe
3 siblings, 1 reply; 7+ messages in thread
From: Jason Gunthorpe @ 2018-01-11 21:43 UTC (permalink / raw)
To: linux-rdma-u79uwXL29TY76Z2rM5mHXA
Cc: Jason Gunthorpe, Ram Amrani, Michal Kalderon
From: Jason Gunthorpe <jgg-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
The double swap matches what user space rdma-core does to imm_data.
wc->imm_data is not used in the kernel so this change has no practical
impact.
Signed-off-by: Jason Gunthorpe <jgg-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
---
drivers/infiniband/hw/qedr/verbs.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/infiniband/hw/qedr/verbs.c b/drivers/infiniband/hw/qedr/verbs.c
index 3b9c89848d6698..5551120ac6ea52 100644
--- a/drivers/infiniband/hw/qedr/verbs.c
+++ b/drivers/infiniband/hw/qedr/verbs.c
@@ -3039,7 +3039,7 @@ static int __qedr_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr,
swqe->wqe_size = 2;
swqe2 = qed_chain_produce(&qp->sq.pbl);
- swqe->inv_key_or_imm_data = cpu_to_le32(wr->ex.imm_data);
+ swqe->inv_key_or_imm_data = cpu_to_le32(be32_to_cpu(wr->ex.imm_data));
length = qedr_prepare_sq_send_data(dev, qp, swqe, swqe2,
wr, bad_wr);
swqe->length = cpu_to_le32(length);
@@ -3590,7 +3590,7 @@ static inline int qedr_set_ok_cqe_resp_wc(struct rdma_cqe_responder *resp,
wc->byte_len = le32_to_cpu(resp->length);
if (resp->flags & QEDR_RESP_IMM) {
- wc->ex.imm_data = le32_to_cpu(resp->imm_data_or_inv_r_Key);
+ wc->ex.imm_data = cpu_to_be32(le32_to_cpu(resp->imm_data_or_inv_r_Key));
wc->wc_flags |= IB_WC_WITH_IMM;
if (resp->flags & QEDR_RESP_RDMA)
--
2.15.1
--
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
^ permalink raw reply related [flat|nested] 7+ messages in thread* Re: [PATCH 0/3] Annotate imm_data as be32 in the uapi header
[not found] ` <20180111214307.501-1-jgg-uk2M96/98Pc@public.gmane.org>
` (2 preceding siblings ...)
2018-01-11 21:43 ` [PATCH 3/3] RDMA/qedr: Fix endian problems around imm_data Jason Gunthorpe
@ 2018-01-15 23:06 ` Jason Gunthorpe
3 siblings, 0 replies; 7+ messages in thread
From: Jason Gunthorpe @ 2018-01-15 23:06 UTC (permalink / raw)
To: linux-rdma-u79uwXL29TY76Z2rM5mHXA
Cc: Ram Amrani, Michal Kalderon, Lijun Ou, Wei Hu(Xavier)
On Thu, Jan 11, 2018 at 02:43:04PM -0700, Jason Gunthorpe wrote:
> From: Jason Gunthorpe <jgg-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
>
> This matches what the user space version of this struct has had for some time
> and it removed several __force casts from the tree.
>
> hns and qedr are updated to match what their userspace providers are doing.
>
> Jason Gunthorpe (3):
> RDMA: Mark imm_data as be32 in the verbs uapi header
> RDMA/hns: Fix endian problems around imm_data and rkey
> RDMA/qedr: Fix endian problems around imm_data
With driver maintainer acks I applied all of these to for-next
Jason
--
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
^ permalink raw reply [flat|nested] 7+ messages in thread