* [PATCH for-next 1/6] IB/hfi1: Remove unnecessary initialization from tx request
[not found] ` <20170512161106.1290.45891.stgit-9QXIwq+3FY+1XWohqUldA0EOCMrvLtNR@public.gmane.org>
@ 2017-05-12 16:19 ` Dennis Dalessandro
2017-05-12 16:19 ` [PATCH for-next 2/6] IB/hfi1: Remove unused mk_qpn function Dennis Dalessandro
` (5 subsequent siblings)
6 siblings, 0 replies; 8+ messages in thread
From: Dennis Dalessandro @ 2017-05-12 16:19 UTC (permalink / raw)
To: dledford-H+wXaHxf7aLQT0dZR+AlfA
Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA, Mike Marciniszyn,
Sebastian Sanchez
From: Sebastian Sanchez <sebastian.sanchez-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
The tx request is unnecessarily initialized in the hot
code path with memset(), however, there's no need to do
this as most fields are initialized later on. this
initialization shows to be costly in the profile.
Remove unnecessary initialization from tx request and make
sure all variables are initialized properly.
Reviewed-by: Mike Marciniszyn <mike.marciniszyn-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Sebastian Sanchez <sebastian.sanchez-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
drivers/infiniband/hw/hfi1/user_sdma.c | 12 +++++++++++-
1 files changed, 11 insertions(+), 1 deletions(-)
diff --git a/drivers/infiniband/hw/hfi1/user_sdma.c b/drivers/infiniband/hw/hfi1/user_sdma.c
index d55339f..16fd519 100644
--- a/drivers/infiniband/hw/hfi1/user_sdma.c
+++ b/drivers/infiniband/hw/hfi1/user_sdma.c
@@ -607,12 +607,19 @@ int hfi1_user_sdma_process_request(struct hfi1_filedata *fd,
hfi1_cdbg(SDMA, "[%u:%u:%u] Using req/comp entry %u\n", dd->unit,
uctxt->ctxt, fd->subctxt, info.comp_idx);
req = pq->reqs + info.comp_idx;
- memset(req, 0, sizeof(*req));
req->data_iovs = req_iovcnt(info.ctrl) - 1; /* subtract header vector */
+ req->data_len = 0;
req->pq = pq;
req->cq = cq;
req->status = -1;
req->ahg_idx = -1;
+ req->iov_idx = 0;
+ req->sent = 0;
+ req->seqnum = 0;
+ req->seqcomp = 0;
+ req->seqsubmitted = 0;
+ req->flags = 0;
+ req->tids = NULL;
INIT_LIST_HEAD(&req->txps);
memcpy(&req->info, &info, sizeof(info));
@@ -701,12 +708,14 @@ int hfi1_user_sdma_process_request(struct hfi1_filedata *fd,
/* Save all the IO vector structures */
for (i = 0; i < req->data_iovs; i++) {
+ req->iovs[i].offset = 0;
INIT_LIST_HEAD(&req->iovs[i].list);
memcpy(&req->iovs[i].iov,
iovec + idx++,
sizeof(req->iovs[i].iov));
ret = pin_vector_pages(req, &req->iovs[i]);
if (ret) {
+ req->data_iovs = i;
req->status = ret;
goto free_req;
}
@@ -749,6 +758,7 @@ int hfi1_user_sdma_process_request(struct hfi1_filedata *fd,
}
req->tids = tmp;
req->n_tids = ntids;
+ req->tididx = 0;
idx++;
}
--
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] 8+ messages in thread* [PATCH for-next 2/6] IB/hfi1: Remove unused mk_qpn function
[not found] ` <20170512161106.1290.45891.stgit-9QXIwq+3FY+1XWohqUldA0EOCMrvLtNR@public.gmane.org>
2017-05-12 16:19 ` [PATCH for-next 1/6] IB/hfi1: Remove unnecessary initialization from tx request Dennis Dalessandro
@ 2017-05-12 16:19 ` Dennis Dalessandro
2017-05-12 16:19 ` [PATCH for-next 3/6] IB/hfi1: Add functions to parse BTH/IB headers Dennis Dalessandro
` (4 subsequent siblings)
6 siblings, 0 replies; 8+ messages in thread
From: Dennis Dalessandro @ 2017-05-12 16:19 UTC (permalink / raw)
To: dledford-H+wXaHxf7aLQT0dZR+AlfA
Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA, Ira Weiny
From: Ira Weiny <ira.weiny-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Leftover function that is not used. Remove it.
Reviewed-by: Dennis Dalessandro <dennis.dalessandro-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Ira Weiny <ira.weiny-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
drivers/infiniband/hw/hfi1/qp.c | 6 ------
1 files changed, 0 insertions(+), 6 deletions(-)
diff --git a/drivers/infiniband/hw/hfi1/qp.c b/drivers/infiniband/hw/hfi1/qp.c
index 650305c..e91be05 100644
--- a/drivers/infiniband/hw/hfi1/qp.c
+++ b/drivers/infiniband/hw/hfi1/qp.c
@@ -73,12 +73,6 @@ static int iowait_sleep(
static void iowait_sdma_drained(struct iowait *wait);
static void qp_pio_drain(struct rvt_qp *qp);
-static inline unsigned mk_qpn(struct rvt_qpn_table *qpt,
- struct rvt_qpn_map *map, unsigned off)
-{
- return (map - qpt->map) * RVT_BITS_PER_PAGE + off;
-}
-
const struct rvt_operation_params hfi1_post_parms[RVT_OPERATION_MAX] = {
[IB_WR_RDMA_WRITE] = {
.length = sizeof(struct ib_rdma_wr),
--
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] 8+ messages in thread* [PATCH for-next 3/6] IB/hfi1: Add functions to parse BTH/IB headers
[not found] ` <20170512161106.1290.45891.stgit-9QXIwq+3FY+1XWohqUldA0EOCMrvLtNR@public.gmane.org>
2017-05-12 16:19 ` [PATCH for-next 1/6] IB/hfi1: Remove unnecessary initialization from tx request Dennis Dalessandro
2017-05-12 16:19 ` [PATCH for-next 2/6] IB/hfi1: Remove unused mk_qpn function Dennis Dalessandro
@ 2017-05-12 16:19 ` Dennis Dalessandro
2017-05-12 16:20 ` [PATCH for-next 4/6] IB/hfi1: Separate input/output header tracing Dennis Dalessandro
` (3 subsequent siblings)
6 siblings, 0 replies; 8+ messages in thread
From: Dennis Dalessandro @ 2017-05-12 16:19 UTC (permalink / raw)
To: dledford-H+wXaHxf7aLQT0dZR+AlfA
Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA, Don Hiatt, Brian Welty,
Dasaratharaman Chandramouli
From: Don Hiatt <don.hiatt-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Improve code readablity by adding inline functions
to read specific BTH/IB fields without knowledge of
byte offsets.
Reviewed-by: Brian Welty <brian.welty-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Reviewed-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Reviewed-by: Dennis Dalessandro <dennis.dalessandro-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Don Hiatt <don.hiatt-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
drivers/infiniband/hw/hfi1/driver.c | 6 +--
drivers/infiniband/hw/hfi1/rc.c | 8 ++-
drivers/infiniband/hw/hfi1/uc.c | 2 -
drivers/infiniband/hw/hfi1/ud.c | 6 +--
drivers/infiniband/hw/hfi1/verbs.c | 6 +--
include/rdma/ib_hdrs.h | 84 +++++++++++++++++++++++++++++++++++
include/rdma/ib_verbs.h | 2 +
include/rdma/rdmavt_qp.h | 2 -
8 files changed, 101 insertions(+), 15 deletions(-)
diff --git a/drivers/infiniband/hw/hfi1/driver.c b/drivers/infiniband/hw/hfi1/driver.c
index a50870e..0583479 100644
--- a/drivers/infiniband/hw/hfi1/driver.c
+++ b/drivers/infiniband/hw/hfi1/driver.c
@@ -286,7 +286,7 @@ static void rcv_hdrerr(struct hfi1_ctxtdata *rcd, struct hfi1_pportdata *ppd,
goto drop;
}
/* Get the destination QP number. */
- qp_num = be32_to_cpu(ohdr->bth[1]) & RVT_QPN_MASK;
+ qp_num = ib_bth_get_qpn(ohdr);
if (lid < be16_to_cpu(IB_MULTICAST_LID_BASE)) {
struct rvt_qp *qp;
unsigned long flags;
@@ -438,7 +438,7 @@ void hfi1_process_ecn_slowpath(struct rvt_qp *qp, struct hfi1_packet *pkt,
case IB_QPT_GSI:
case IB_QPT_UD:
rlid = ib_get_slid(hdr);
- rqpn = be32_to_cpu(ohdr->u.ud.deth[1]) & RVT_QPN_MASK;
+ rqpn = ib_get_sqpn(ohdr);
svc_type = IB_CC_SVCTYPE_UD;
is_mcast = (dlid > be16_to_cpu(IB_MULTICAST_LID_BASE)) &&
(dlid != be16_to_cpu(IB_LID_PERMISSIVE));
@@ -461,7 +461,7 @@ void hfi1_process_ecn_slowpath(struct rvt_qp *qp, struct hfi1_packet *pkt,
bth1 = be32_to_cpu(ohdr->bth[1]);
if (do_cnp && (bth1 & IB_FECN_SMASK)) {
- u16 pkey = (u16)be32_to_cpu(ohdr->bth[0]);
+ u16 pkey = ib_bth_get_pkey(ohdr);
return_cnp(ibp, qp, rqpn, pkey, dlid, rlid, sc, grh);
}
diff --git a/drivers/infiniband/hw/hfi1/rc.c b/drivers/infiniband/hw/hfi1/rc.c
index 1080778..66e6843 100644
--- a/drivers/infiniband/hw/hfi1/rc.c
+++ b/drivers/infiniband/hw/hfi1/rc.c
@@ -765,7 +765,7 @@ void hfi1_send_rc_ack(struct hfi1_ctxtdata *rcd, struct rvt_qp *qp,
ohdr->u.aeth = rvt_compute_aeth(qp);
sc5 = ibp->sl_to_sc[rdma_ah_get_sl(&qp->remote_ah_attr)];
/* set PBC_DC_INFO bit (aka SC[4]) in pbc_flags */
- pbc_flags |= ((!!(sc5 & 0x10)) << PBC_DC_INFO_SHIFT);
+ pbc_flags |= (ib_is_sc5(sc5) << PBC_DC_INFO_SHIFT);
lrh0 |= (sc5 & 0xf) << 12 | (rdma_ah_get_sl(&qp->remote_ah_attr)
& 0xf) << 4;
hdr.lrh[0] = cpu_to_be16(lrh0);
@@ -1009,7 +1009,7 @@ void hfi1_rc_send_complete(struct rvt_qp *qp, struct ib_header *hdr)
return;
}
- psn = be32_to_cpu(ohdr->bth[2]);
+ psn = ib_bth_get_psn(ohdr);
reset_sending_psn(qp, psn);
/*
@@ -1943,7 +1943,7 @@ void hfi1_rc_rcv(struct hfi1_packet *packet)
is_fecn = process_ecn(qp, packet, false);
- psn = be32_to_cpu(ohdr->bth[2]);
+ psn = ib_bth_get_psn(ohdr);
opcode = ib_bth_get_opcode(ohdr);
/*
@@ -2388,7 +2388,7 @@ void hfi1_rc_hdrerr(
if (hfi1_ruc_check_hdr(ibp, hdr, has_grh, qp, bth0))
return;
- psn = be32_to_cpu(ohdr->bth[2]);
+ psn = ib_bth_get_psn(ohdr);
opcode = ib_bth_get_opcode(ohdr);
/* Only deal with RDMA Writes for now */
diff --git a/drivers/infiniband/hw/hfi1/uc.c b/drivers/infiniband/hw/hfi1/uc.c
index 5da1e45..2a5650f 100644
--- a/drivers/infiniband/hw/hfi1/uc.c
+++ b/drivers/infiniband/hw/hfi1/uc.c
@@ -319,7 +319,7 @@ void hfi1_uc_rcv(struct hfi1_packet *packet)
process_ecn(qp, packet, true);
- psn = be32_to_cpu(ohdr->bth[2]);
+ psn = ib_bth_get_psn(ohdr);
opcode = ib_bth_get_opcode(ohdr);
/* Compare the PSN verses the expected PSN. */
diff --git a/drivers/infiniband/hw/hfi1/ud.c b/drivers/infiniband/hw/hfi1/ud.c
index 6a4e95c..49fe179 100644
--- a/drivers/infiniband/hw/hfi1/ud.c
+++ b/drivers/infiniband/hw/hfi1/ud.c
@@ -549,7 +549,7 @@ void return_cnp(struct hfi1_ibport *ibp, struct rvt_qp *qp, u32 remote_qpn,
hdr.lrh[3] = cpu_to_be16(slid);
plen = 2 /* PBC */ + hwords;
- pbc_flags |= (!!(sc5 & 0x10)) << PBC_DC_INFO_SHIFT;
+ pbc_flags |= (ib_is_sc5(sc5) << PBC_DC_INFO_SHIFT);
vl = sc_to_vlt(ppd->dd, sc5);
pbc = create_pbc(ppd, pbc_flags, qp->srate_mbps, vl, plen);
if (ctxt) {
@@ -689,8 +689,8 @@ void hfi1_ud_rcv(struct hfi1_packet *packet)
u16 slid;
u8 extra_bytes;
- qkey = be32_to_cpu(ohdr->u.ud.deth[0]);
- src_qp = be32_to_cpu(ohdr->u.ud.deth[1]) & RVT_QPN_MASK;
+ qkey = ib_get_qkey(ohdr);
+ src_qp = ib_get_sqpn(ohdr);
dlid = ib_get_dlid(hdr);
bth1 = be32_to_cpu(ohdr->bth[1]);
slid = ib_get_slid(hdr);
diff --git a/drivers/infiniband/hw/hfi1/verbs.c b/drivers/infiniband/hw/hfi1/verbs.c
index 90e7b77..128d291 100644
--- a/drivers/infiniband/hw/hfi1/verbs.c
+++ b/drivers/infiniband/hw/hfi1/verbs.c
@@ -595,7 +595,7 @@ void hfi1_ib_rcv(struct hfi1_packet *packet)
inc_opstats(tlen, &rcd->opstats->stats[opcode]);
/* Get the destination QP number. */
- qp_num = be32_to_cpu(packet->ohdr->bth[1]) & RVT_QPN_MASK;
+ qp_num = ib_bth_get_qpn(packet->ohdr);
lid = ib_get_dlid(hdr);
if (unlikely((lid >= be16_to_cpu(IB_MULTICAST_LID_BASE)) &&
(lid != be16_to_cpu(IB_LID_PERMISSIVE)))) {
@@ -863,7 +863,7 @@ int hfi1_verbs_send_dma(struct rvt_qp *qp, struct hfi1_pkt_state *ps,
/* No vl15 here */
/* set PBC_DC_INFO bit (aka SC[4]) in pbc_flags */
- pbc |= (!!(sc5 & 0x10)) << PBC_DC_INFO_SHIFT;
+ pbc |= (ib_is_sc5(sc5) << PBC_DC_INFO_SHIFT);
if (unlikely(hfi1_dbg_fault_opcode(qp, opcode, false)))
pbc = hfi1_fault_tx(qp, opcode, pbc);
@@ -999,7 +999,7 @@ int hfi1_verbs_send_pio(struct rvt_qp *qp, struct hfi1_pkt_state *ps,
u8 opcode = get_opcode(&tx->phdr.hdr);
/* set PBC_DC_INFO bit (aka SC[4]) in pbc_flags */
- pbc |= (!!(sc5 & 0x10)) << PBC_DC_INFO_SHIFT;
+ pbc |= (ib_is_sc5(sc5) << PBC_DC_INFO_SHIFT);
if (unlikely(hfi1_dbg_fault_opcode(qp, opcode, false)))
pbc = hfi1_fault_tx(qp, opcode, pbc);
pbc = create_pbc(ppd, pbc, qp->srate_mbps, vl, plen);
diff --git a/include/rdma/ib_hdrs.h b/include/rdma/ib_hdrs.h
index 5519f31..c124d51 100644
--- a/include/rdma/ib_hdrs.h
+++ b/include/rdma/ib_hdrs.h
@@ -193,8 +193,12 @@ static inline void put_ib_ateth_compare(u64 val, struct ib_atomic_eth *ateth)
#define IB_LNH_MASK 3
#define IB_SC_MASK 0xf
#define IB_SC_SHIFT 12
+#define IB_SC5_MASK 0x10
#define IB_SL_MASK 0xf
#define IB_SL_SHIFT 4
+#define IB_SL_SHIFT 4
+#define IB_LVER_MASK 0xf
+#define IB_LVER_SHIFT 8
static inline u8 ib_get_lnh(struct ib_header *hdr)
{
@@ -206,6 +210,11 @@ static inline u8 ib_get_sc(struct ib_header *hdr)
return ((be16_to_cpu(hdr->lrh[0]) >> IB_SC_SHIFT) & IB_SC_MASK);
}
+static inline bool ib_is_sc5(u16 sc5)
+{
+ return !!(sc5 & IB_SC5_MASK);
+}
+
static inline u8 ib_get_sl(struct ib_header *hdr)
{
return ((be16_to_cpu(hdr->lrh[0]) >> IB_SL_SHIFT) & IB_SL_MASK);
@@ -221,6 +230,27 @@ static inline u16 ib_get_slid(struct ib_header *hdr)
return (be16_to_cpu(hdr->lrh[3]));
}
+static inline u8 ib_get_lver(struct ib_header *hdr)
+{
+ return (u8)((be16_to_cpu(hdr->lrh[0]) >> IB_LVER_SHIFT) &
+ IB_LVER_MASK);
+}
+
+static inline u16 ib_get_len(struct ib_header *hdr)
+{
+ return (u16)(be16_to_cpu(hdr->lrh[2]));
+}
+
+static inline u32 ib_get_qkey(struct ib_other_headers *ohdr)
+{
+ return be32_to_cpu(ohdr->u.ud.deth[0]);
+}
+
+static inline u32 ib_get_sqpn(struct ib_other_headers *ohdr)
+{
+ return ((be32_to_cpu(ohdr->u.ud.deth[1])) & IB_QPN_MASK);
+}
+
/*
* BTH
*/
@@ -229,6 +259,14 @@ static inline u16 ib_get_slid(struct ib_header *hdr)
#define IB_BTH_PAD_MASK 3
#define IB_BTH_PKEY_MASK 0xffff
#define IB_BTH_PAD_SHIFT 20
+#define IB_BTH_A_MASK 1
+#define IB_BTH_A_SHIFT 31
+#define IB_BTH_M_MASK 1
+#define IB_BTH_M_SHIFT 22
+#define IB_BTH_SE_MASK 1
+#define IB_BTH_SE_SHIFT 23
+#define IB_BTH_TVER_MASK 0xf
+#define IB_BTH_TVER_SHIFT 16
static inline u8 ib_bth_get_pad(struct ib_other_headers *ohdr)
{
@@ -247,4 +285,50 @@ static inline u8 ib_bth_get_opcode(struct ib_other_headers *ohdr)
IB_BTH_OPCODE_MASK);
}
+static inline u8 ib_bth_get_ackreq(struct ib_other_headers *ohdr)
+{
+ return (u8)((be32_to_cpu(ohdr->bth[2]) >> IB_BTH_A_SHIFT) &
+ IB_BTH_A_MASK);
+}
+
+static inline u8 ib_bth_get_migreq(struct ib_other_headers *ohdr)
+{
+ return (u8)((be32_to_cpu(ohdr->bth[0]) >> IB_BTH_M_SHIFT) &
+ IB_BTH_M_MASK);
+}
+
+static inline u8 ib_bth_get_se(struct ib_other_headers *ohdr)
+{
+ return (u8)((be32_to_cpu(ohdr->bth[0]) >> IB_BTH_SE_SHIFT) &
+ IB_BTH_SE_MASK);
+}
+
+static inline u32 ib_bth_get_psn(struct ib_other_headers *ohdr)
+{
+ return (u32)(be32_to_cpu(ohdr->bth[2]));
+}
+
+static inline u32 ib_bth_get_qpn(struct ib_other_headers *ohdr)
+{
+ return (u32)((be32_to_cpu(ohdr->bth[1])) & IB_QPN_MASK);
+}
+
+static inline u8 ib_bth_get_becn(struct ib_other_headers *ohdr)
+{
+ return (u8)((be32_to_cpu(ohdr->bth[1]) >> IB_BECN_SHIFT) &
+ IB_BECN_MASK);
+}
+
+static inline u8 ib_bth_get_fecn(struct ib_other_headers *ohdr)
+{
+ return (u8)((be32_to_cpu(ohdr->bth[1]) >> IB_FECN_SHIFT) &
+ IB_FECN_MASK);
+}
+
+static inline u8 ib_bth_get_tver(struct ib_other_headers *ohdr)
+{
+ return (u8)((be32_to_cpu(ohdr->bth[0]) >> IB_BTH_TVER_SHIFT) &
+ IB_BTH_TVER_MASK);
+}
+
#endif /* IB_HDRS_H */
diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index f0cb490..785e927 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -663,6 +663,8 @@ struct ib_grh {
};
};
+#define IB_QPN_MASK 0xFFFFFF
+
enum {
IB_MULTICAST_QPN = 0xffffff
};
diff --git a/include/rdma/rdmavt_qp.h b/include/rdma/rdmavt_qp.h
index be6472e..13f43b3 100644
--- a/include/rdma/rdmavt_qp.h
+++ b/include/rdma/rdmavt_qp.h
@@ -396,7 +396,7 @@ struct rvt_srq {
#define RVT_QPNMAP_ENTRIES (RVT_QPN_MAX / PAGE_SIZE / BITS_PER_BYTE)
#define RVT_BITS_PER_PAGE (PAGE_SIZE * BITS_PER_BYTE)
#define RVT_BITS_PER_PAGE_MASK (RVT_BITS_PER_PAGE - 1)
-#define RVT_QPN_MASK 0xFFFFFF
+#define RVT_QPN_MASK IB_QPN_MASK
/*
* QPN-map pages start out as NULL, they get allocated upon
--
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] 8+ messages in thread* [PATCH for-next 4/6] IB/hfi1: Separate input/output header tracing
[not found] ` <20170512161106.1290.45891.stgit-9QXIwq+3FY+1XWohqUldA0EOCMrvLtNR@public.gmane.org>
` (2 preceding siblings ...)
2017-05-12 16:19 ` [PATCH for-next 3/6] IB/hfi1: Add functions to parse BTH/IB headers Dennis Dalessandro
@ 2017-05-12 16:20 ` Dennis Dalessandro
2017-05-12 16:20 ` [PATCH for-next 5/6] IB/hfi1: Setup common IB fields in hfi1_packet struct Dennis Dalessandro
` (2 subsequent siblings)
6 siblings, 0 replies; 8+ messages in thread
From: Dennis Dalessandro @ 2017-05-12 16:20 UTC (permalink / raw)
To: dledford-H+wXaHxf7aLQT0dZR+AlfA
Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA, Don Hiatt,
Dasaratharaman Chandramouli
From: Don Hiatt <don.hiatt-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Calls to trace incoming packets will now receive the packet
context as parameter. This enables trace support for future
packet types.
Header trace output is in the format <field>:<value>
which makes parsing easier.
input_ibhdr trace before change:
<idle>-0 [001] d.h. 5904.250925: input_ibhdr: [0000:05:00.0] vl 0
lver 0 sl 0 lnh 2,LRH_BTH dlid 0002 len 18 slid 0001 op
0x64,UD_SEND_ONLY se 0 m 0 pad 0 tver 0 pkey 0xffff f 0 b 0 qpn 0x000001
a 0 psn 0x000001b2 deth qkey 0x80010000 sqpn 0x000001
input_ibhdr trace after change:
<idle>-0 [001] d.h. 6655.714488: input_ibhdr: [0000:05:00.0] (IB)
len:124 sc:0 dlid:0x0001 slid:0x0002 lnh:2,LRH_BTH lver:0 sl:0 age:0
becn:0 fecn:0 l4:0 rc:0 entropy:0 op:0x64,UD_SEND_ONLY se:0 m:0 pad:0
tver:0 pkey:0x7fff f:0 b:0 qpn:0x000001 a:0 psn:0x00000036 hlen:8 deth
qkey:0x80010000 sqpn:0x000001
Reviewed-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Reviewed-by: Dennis Dalessandro <dennis.dalessandro-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Don Hiatt <don.hiatt-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
drivers/infiniband/hw/hfi1/hfi.h | 49 ----
drivers/infiniband/hw/hfi1/rc.c | 3
drivers/infiniband/hw/hfi1/trace.c | 58 +++++
drivers/infiniband/hw/hfi1/trace_ibhdrs.h | 322 +++++++++++++++++++----------
drivers/infiniband/hw/hfi1/trace_rx.h | 9 +
drivers/infiniband/hw/hfi1/verbs.c | 7 -
6 files changed, 279 insertions(+), 169 deletions(-)
diff --git a/drivers/infiniband/hw/hfi1/hfi.h b/drivers/infiniband/hw/hfi1/hfi.h
index 0f0abfd..2fb0a54 100644
--- a/drivers/infiniband/hw/hfi1/hfi.h
+++ b/drivers/infiniband/hw/hfi1/hfi.h
@@ -2087,53 +2087,4 @@ static inline bool is_integrated(struct hfi1_devdata *dd)
#define DD_DEV_ENTRY(dd) __string(dev, dev_name(&(dd)->pcidev->dev))
#define DD_DEV_ASSIGN(dd) __assign_str(dev, dev_name(&(dd)->pcidev->dev))
-
-#define packettype_name(etype) { RHF_RCV_TYPE_##etype, #etype }
-#define show_packettype(etype) \
-__print_symbolic(etype, \
- packettype_name(EXPECTED), \
- packettype_name(EAGER), \
- packettype_name(IB), \
- packettype_name(ERROR), \
- packettype_name(BYPASS))
-
-#define ib_opcode_name(opcode) { IB_OPCODE_##opcode, #opcode }
-#define show_ib_opcode(opcode) \
-__print_symbolic(opcode, \
- ib_opcode_name(RC_SEND_FIRST), \
- ib_opcode_name(RC_SEND_MIDDLE), \
- ib_opcode_name(RC_SEND_LAST), \
- ib_opcode_name(RC_SEND_LAST_WITH_IMMEDIATE), \
- ib_opcode_name(RC_SEND_ONLY), \
- ib_opcode_name(RC_SEND_ONLY_WITH_IMMEDIATE), \
- ib_opcode_name(RC_RDMA_WRITE_FIRST), \
- ib_opcode_name(RC_RDMA_WRITE_MIDDLE), \
- ib_opcode_name(RC_RDMA_WRITE_LAST), \
- ib_opcode_name(RC_RDMA_WRITE_LAST_WITH_IMMEDIATE), \
- ib_opcode_name(RC_RDMA_WRITE_ONLY), \
- ib_opcode_name(RC_RDMA_WRITE_ONLY_WITH_IMMEDIATE), \
- ib_opcode_name(RC_RDMA_READ_REQUEST), \
- ib_opcode_name(RC_RDMA_READ_RESPONSE_FIRST), \
- ib_opcode_name(RC_RDMA_READ_RESPONSE_MIDDLE), \
- ib_opcode_name(RC_RDMA_READ_RESPONSE_LAST), \
- ib_opcode_name(RC_RDMA_READ_RESPONSE_ONLY), \
- ib_opcode_name(RC_ACKNOWLEDGE), \
- ib_opcode_name(RC_ATOMIC_ACKNOWLEDGE), \
- ib_opcode_name(RC_COMPARE_SWAP), \
- ib_opcode_name(RC_FETCH_ADD), \
- ib_opcode_name(UC_SEND_FIRST), \
- ib_opcode_name(UC_SEND_MIDDLE), \
- ib_opcode_name(UC_SEND_LAST), \
- ib_opcode_name(UC_SEND_LAST_WITH_IMMEDIATE), \
- ib_opcode_name(UC_SEND_ONLY), \
- ib_opcode_name(UC_SEND_ONLY_WITH_IMMEDIATE), \
- ib_opcode_name(UC_RDMA_WRITE_FIRST), \
- ib_opcode_name(UC_RDMA_WRITE_MIDDLE), \
- ib_opcode_name(UC_RDMA_WRITE_LAST), \
- ib_opcode_name(UC_RDMA_WRITE_LAST_WITH_IMMEDIATE), \
- ib_opcode_name(UC_RDMA_WRITE_ONLY), \
- ib_opcode_name(UC_RDMA_WRITE_ONLY_WITH_IMMEDIATE), \
- ib_opcode_name(UD_SEND_ONLY), \
- ib_opcode_name(UD_SEND_ONLY_WITH_IMMEDIATE), \
- ib_opcode_name(CNP))
#endif /* _HFI1_KERNEL_H */
diff --git a/drivers/infiniband/hw/hfi1/rc.c b/drivers/infiniband/hw/hfi1/rc.c
index 66e6843..b443c1e 100644
--- a/drivers/infiniband/hw/hfi1/rc.c
+++ b/drivers/infiniband/hw/hfi1/rc.c
@@ -798,7 +798,8 @@ void hfi1_send_rc_ack(struct hfi1_ctxtdata *rcd, struct rvt_qp *qp,
goto queue_ack;
}
- trace_ack_output_ibhdr(dd_from_ibdev(qp->ibqp.device), &hdr);
+ trace_ack_output_ibhdr(dd_from_ibdev(qp->ibqp.device),
+ &hdr, ib_is_sc5(sc5));
/* write the pbc and data */
ppd->dd->pio_inline_send(ppd->dd, pbuf, pbc, &hdr, hwords);
diff --git a/drivers/infiniband/hw/hfi1/trace.c b/drivers/infiniband/hw/hfi1/trace.c
index eafae48..b80b74d 100644
--- a/drivers/infiniband/hw/hfi1/trace.c
+++ b/drivers/infiniband/hw/hfi1/trace.c
@@ -47,7 +47,7 @@
#define CREATE_TRACE_POINTS
#include "trace.h"
-u8 ibhdr_exhdr_len(struct ib_header *hdr)
+u8 hfi1_trace_ib_hdr_len(struct ib_header *hdr)
{
struct ib_other_headers *ohdr;
u8 opcode;
@@ -61,13 +61,18 @@ u8 ibhdr_exhdr_len(struct ib_header *hdr)
0 : hdr_len_by_opcode[opcode] - (12 + 8);
}
-#define IMM_PRN "imm %d"
-#define RETH_PRN "reth vaddr 0x%.16llx rkey 0x%.8x dlen 0x%.8x"
-#define AETH_PRN "aeth syn 0x%.2x %s msn 0x%.8x"
-#define DETH_PRN "deth qkey 0x%.8x sqpn 0x%.6x"
-#define IETH_PRN "ieth rkey 0x%.8x"
-#define ATOMICACKETH_PRN "origdata %llx"
-#define ATOMICETH_PRN "vaddr 0x%llx rkey 0x%.8x sdata %llx cdata %llx"
+const char *hfi1_trace_get_packet_str(struct hfi1_packet *packet)
+{
+ return "IB";
+}
+
+#define IMM_PRN "imm:%d"
+#define RETH_PRN "reth vaddr:0x%.16llx rkey:0x%.8x dlen:0x%.8x"
+#define AETH_PRN "aeth syn:0x%.2x %s msn:0x%.8x"
+#define DETH_PRN "deth qkey:0x%.8x sqpn:0x%.6x"
+#define IETH_PRN "ieth rkey:0x%.8x"
+#define ATOMICACKETH_PRN "origdata:%llx"
+#define ATOMICETH_PRN "vaddr:0x%llx rkey:0x%.8x sdata:%llx cdata:%llx"
#define OP(transport, op) IB_OPCODE_## transport ## _ ## op
@@ -84,6 +89,43 @@ u8 ibhdr_exhdr_len(struct ib_header *hdr)
return "";
}
+void hfi1_trace_parse_bth(struct ib_other_headers *ohdr,
+ u8 *ack, u8 *becn, u8 *fecn, u8 *mig,
+ u8 *se, u8 *pad, u8 *opcode, u8 *tver,
+ u16 *pkey, u32 *psn, u32 *qpn)
+{
+ *ack = ib_bth_get_ackreq(ohdr);
+ *becn = ib_bth_get_becn(ohdr);
+ *fecn = ib_bth_get_fecn(ohdr);
+ *mig = ib_bth_get_migreq(ohdr);
+ *se = ib_bth_get_se(ohdr);
+ *pad = ib_bth_get_pad(ohdr);
+ *opcode = ib_bth_get_opcode(ohdr);
+ *tver = ib_bth_get_tver(ohdr);
+ *pkey = ib_bth_get_pkey(ohdr);
+ *psn = ib_bth_get_psn(ohdr);
+ *qpn = ib_bth_get_qpn(ohdr);
+}
+
+void hfi1_trace_parse_9b_hdr(struct ib_header *hdr, bool sc5,
+ struct ib_other_headers **ohdr,
+ u8 *lnh, u8 *lver, u8 *sl, u8 *sc,
+ u16 *len, u32 *dlid, u32 *slid)
+{
+ *lnh = ib_get_lnh(hdr);
+ *lver = ib_get_lver(hdr);
+ *sl = ib_get_sl(hdr);
+ *sc = ib_get_sc(hdr) | (sc5 << 4);
+ *len = ib_get_len(hdr);
+ *dlid = ib_get_dlid(hdr);
+ *slid = ib_get_slid(hdr);
+
+ if (*lnh == HFI1_LRH_BTH)
+ *ohdr = &hdr->u.oth;
+ else
+ *ohdr = &hdr->u.l.oth;
+}
+
const char *parse_everbs_hdrs(
struct trace_seq *p,
u8 opcode,
diff --git a/drivers/infiniband/hw/hfi1/trace_ibhdrs.h b/drivers/infiniband/hw/hfi1/trace_ibhdrs.h
index 090f6b5..0f2d2da 100644
--- a/drivers/infiniband/hw/hfi1/trace_ibhdrs.h
+++ b/drivers/infiniband/hw/hfi1/trace_ibhdrs.h
@@ -55,8 +55,57 @@
#undef TRACE_SYSTEM
#define TRACE_SYSTEM hfi1_ibhdrs
-u8 ibhdr_exhdr_len(struct ib_header *hdr);
+#define ib_opcode_name(opcode) { IB_OPCODE_##opcode, #opcode }
+#define show_ib_opcode(opcode) \
+__print_symbolic(opcode, \
+ ib_opcode_name(RC_SEND_FIRST), \
+ ib_opcode_name(RC_SEND_MIDDLE), \
+ ib_opcode_name(RC_SEND_LAST), \
+ ib_opcode_name(RC_SEND_LAST_WITH_IMMEDIATE), \
+ ib_opcode_name(RC_SEND_ONLY), \
+ ib_opcode_name(RC_SEND_ONLY_WITH_IMMEDIATE), \
+ ib_opcode_name(RC_RDMA_WRITE_FIRST), \
+ ib_opcode_name(RC_RDMA_WRITE_MIDDLE), \
+ ib_opcode_name(RC_RDMA_WRITE_LAST), \
+ ib_opcode_name(RC_RDMA_WRITE_LAST_WITH_IMMEDIATE), \
+ ib_opcode_name(RC_RDMA_WRITE_ONLY), \
+ ib_opcode_name(RC_RDMA_WRITE_ONLY_WITH_IMMEDIATE), \
+ ib_opcode_name(RC_RDMA_READ_REQUEST), \
+ ib_opcode_name(RC_RDMA_READ_RESPONSE_FIRST), \
+ ib_opcode_name(RC_RDMA_READ_RESPONSE_MIDDLE), \
+ ib_opcode_name(RC_RDMA_READ_RESPONSE_LAST), \
+ ib_opcode_name(RC_RDMA_READ_RESPONSE_ONLY), \
+ ib_opcode_name(RC_ACKNOWLEDGE), \
+ ib_opcode_name(RC_ATOMIC_ACKNOWLEDGE), \
+ ib_opcode_name(RC_COMPARE_SWAP), \
+ ib_opcode_name(RC_FETCH_ADD), \
+ ib_opcode_name(UC_SEND_FIRST), \
+ ib_opcode_name(UC_SEND_MIDDLE), \
+ ib_opcode_name(UC_SEND_LAST), \
+ ib_opcode_name(UC_SEND_LAST_WITH_IMMEDIATE), \
+ ib_opcode_name(UC_SEND_ONLY), \
+ ib_opcode_name(UC_SEND_ONLY_WITH_IMMEDIATE), \
+ ib_opcode_name(UC_RDMA_WRITE_FIRST), \
+ ib_opcode_name(UC_RDMA_WRITE_MIDDLE), \
+ ib_opcode_name(UC_RDMA_WRITE_LAST), \
+ ib_opcode_name(UC_RDMA_WRITE_LAST_WITH_IMMEDIATE), \
+ ib_opcode_name(UC_RDMA_WRITE_ONLY), \
+ ib_opcode_name(UC_RDMA_WRITE_ONLY_WITH_IMMEDIATE), \
+ ib_opcode_name(UD_SEND_ONLY), \
+ ib_opcode_name(UD_SEND_ONLY_WITH_IMMEDIATE), \
+ ib_opcode_name(CNP))
+
const char *parse_everbs_hdrs(struct trace_seq *p, u8 opcode, void *ehdrs);
+u8 hfi1_trace_ib_hdr_len(struct ib_header *hdr);
+const char *hfi1_trace_get_packet_str(struct hfi1_packet *packet);
+void hfi1_trace_parse_bth(struct ib_other_headers *ohdr,
+ u8 *ack, u8 *becn, u8 *fecn, u8 *mig,
+ u8 *se, u8 *pad, u8 *opcode, u8 *tver,
+ u16 *pkey, u32 *psn, u32 *qpn);
+void hfi1_trace_parse_9b_hdr(struct ib_header *hdr, bool sc5,
+ struct ib_other_headers **ohdr,
+ u8 *lnh, u8 *lver, u8 *sl, u8 *sc,
+ u16 *len, u32 *dlid, u32 *slid);
#define __parse_ib_ehdrs(op, ehdrs) parse_everbs_hdrs(p, op, ehdrs)
@@ -66,139 +115,198 @@
lrh_name(LRH_BTH), \
lrh_name(LRH_GRH))
-#define LRH_PRN "vl %d lver %d sl %d lnh %d,%s dlid %.4x len %d slid %.4x"
+#define LRH_PRN "len:%d sc:%d dlid:0x%.4x slid:0x%.4x"
+#define LRH_9B_PRN "lnh:%d,%s lver:%d sl:%d "
#define BTH_PRN \
- "op 0x%.2x,%s se %d m %d pad %d tver %d pkey 0x%.4x " \
- "f %d b %d qpn 0x%.6x a %d psn 0x%.8x"
-#define EHDR_PRN "%s"
+ "op:0x%.2x,%s se:%d m:%d pad:%d tver:%d pkey:0x%.4x " \
+ "f:%d b:%d qpn:0x%.6x a:%d psn:0x%.8x"
+#define EHDR_PRN "hlen:%d %s"
-DECLARE_EVENT_CLASS(hfi1_ibhdr_template,
+DECLARE_EVENT_CLASS(hfi1_input_ibhdr_template,
TP_PROTO(struct hfi1_devdata *dd,
- struct ib_header *hdr),
- TP_ARGS(dd, hdr),
+ struct hfi1_packet *packet,
+ bool sc5),
+ TP_ARGS(dd, packet, sc5),
TP_STRUCT__entry(
DD_DEV_ENTRY(dd)
- /* LRH */
- __field(u8, vl)
+ __field(u8, lnh)
__field(u8, lver)
__field(u8, sl)
+ __field(u16, len)
+ __field(u32, dlid)
+ __field(u8, sc)
+ __field(u32, slid)
+ __field(u8, opcode)
+ __field(u8, se)
+ __field(u8, mig)
+ __field(u8, pad)
+ __field(u8, tver)
+ __field(u16, pkey)
+ __field(u8, fecn)
+ __field(u8, becn)
+ __field(u32, qpn)
+ __field(u8, ack)
+ __field(u32, psn)
+ /* extended headers */
+ __dynamic_array(u8, ehdrs,
+ hfi1_trace_ib_hdr_len(packet->hdr))
+ ),
+ TP_fast_assign(
+ struct ib_other_headers *ohdr;
+
+ DD_DEV_ASSIGN(dd);
+
+ hfi1_trace_parse_9b_hdr(packet->hdr, sc5,
+ &ohdr,
+ &__entry->lnh,
+ &__entry->lver,
+ &__entry->sl,
+ &__entry->sc,
+ &__entry->len,
+ &__entry->dlid,
+ &__entry->slid);
+
+ hfi1_trace_parse_bth(ohdr, &__entry->ack,
+ &__entry->becn, &__entry->fecn,
+ &__entry->mig, &__entry->se,
+ &__entry->pad, &__entry->opcode,
+ &__entry->tver, &__entry->pkey,
+ &__entry->psn, &__entry->qpn);
+ /* extended headers */
+ memcpy(__get_dynamic_array(ehdrs), &ohdr->u,
+ __get_dynamic_array_len(ehdrs));
+ ),
+ TP_printk("[%s] (IB) " LRH_PRN " " LRH_9B_PRN " "
+ BTH_PRN " " EHDR_PRN,
+ __get_str(dev),
+ __entry->len,
+ __entry->sc,
+ __entry->dlid,
+ __entry->slid,
+ __entry->lnh, show_lnh(__entry->lnh),
+ __entry->lver,
+ __entry->sl,
+ /* BTH */
+ __entry->opcode, show_ib_opcode(__entry->opcode),
+ __entry->se,
+ __entry->mig,
+ __entry->pad,
+ __entry->tver,
+ __entry->pkey,
+ __entry->fecn,
+ __entry->becn,
+ __entry->qpn,
+ __entry->ack,
+ __entry->psn,
+ /* extended headers */
+ __get_dynamic_array_len(ehdrs),
+ __parse_ib_ehdrs(
+ __entry->opcode,
+ (void *)__get_dynamic_array(ehdrs))
+ )
+);
+
+DEFINE_EVENT(hfi1_input_ibhdr_template, input_ibhdr,
+ TP_PROTO(struct hfi1_devdata *dd,
+ struct hfi1_packet *packet, bool sc5),
+ TP_ARGS(dd, packet, sc5));
+
+DECLARE_EVENT_CLASS(hfi1_output_ibhdr_template,
+ TP_PROTO(struct hfi1_devdata *dd,
+ struct ib_header *hdr,
+ bool sc5),
+ TP_ARGS(dd, hdr, sc5),
+ TP_STRUCT__entry(
+ DD_DEV_ENTRY(dd)
__field(u8, lnh)
- __field(u16, dlid)
+ __field(u8, lver)
+ __field(u8, sl)
__field(u16, len)
- __field(u16, slid)
- /* BTH */
+ __field(u32, dlid)
+ __field(u8, sc)
+ __field(u32, slid)
__field(u8, opcode)
__field(u8, se)
- __field(u8, m)
+ __field(u8, mig)
__field(u8, pad)
__field(u8, tver)
__field(u16, pkey)
- __field(u8, f)
- __field(u8, b)
+ __field(u8, fecn)
+ __field(u8, becn)
__field(u32, qpn)
- __field(u8, a)
+ __field(u8, ack)
__field(u32, psn)
/* extended headers */
- __dynamic_array(u8, ehdrs, ibhdr_exhdr_len(hdr))
+ __dynamic_array(u8, ehdrs,
+ hfi1_trace_ib_hdr_len(hdr))
),
- TP_fast_assign(
+ TP_fast_assign(
struct ib_other_headers *ohdr;
DD_DEV_ASSIGN(dd);
- /* LRH */
- __entry->vl =
- (u8)(be16_to_cpu(hdr->lrh[0]) >> 12);
- __entry->lver =
- (u8)(be16_to_cpu(hdr->lrh[0]) >> 8) & 0xf;
- __entry->sl =
- (u8)(be16_to_cpu(hdr->lrh[0]) >> 4) & 0xf;
- __entry->lnh =
- (u8)(be16_to_cpu(hdr->lrh[0]) & 3);
- __entry->dlid =
- be16_to_cpu(hdr->lrh[1]);
- /* allow for larger len */
- __entry->len =
- be16_to_cpu(hdr->lrh[2]);
- __entry->slid =
- be16_to_cpu(hdr->lrh[3]);
- /* BTH */
- if (__entry->lnh == HFI1_LRH_BTH)
- ohdr = &hdr->u.oth;
- else
- ohdr = &hdr->u.l.oth;
- __entry->opcode =
- (be32_to_cpu(ohdr->bth[0]) >> 24) & 0xff;
- __entry->se =
- (be32_to_cpu(ohdr->bth[0]) >> 23) & 1;
- __entry->m =
- (be32_to_cpu(ohdr->bth[0]) >> 22) & 1;
- __entry->pad =
- (be32_to_cpu(ohdr->bth[0]) >> 20) & 3;
- __entry->tver =
- (be32_to_cpu(ohdr->bth[0]) >> 16) & 0xf;
- __entry->pkey =
- be32_to_cpu(ohdr->bth[0]) & 0xffff;
- __entry->f =
- (be32_to_cpu(ohdr->bth[1]) >> IB_FECN_SHIFT) &
- IB_FECN_MASK;
- __entry->b =
- (be32_to_cpu(ohdr->bth[1]) >> IB_BECN_SHIFT) &
- IB_BECN_MASK;
- __entry->qpn =
- be32_to_cpu(ohdr->bth[1]) & RVT_QPN_MASK;
- __entry->a =
- (be32_to_cpu(ohdr->bth[2]) >> 31) & 1;
- /* allow for larger PSN */
- __entry->psn =
- be32_to_cpu(ohdr->bth[2]) & 0x7fffffff;
+
+ hfi1_trace_parse_9b_hdr(hdr, sc5,
+ &ohdr, &__entry->lnh,
+ &__entry->lver, &__entry->sl,
+ &__entry->sc, &__entry->len,
+ &__entry->dlid, &__entry->slid);
+
+ hfi1_trace_parse_bth(ohdr, &__entry->ack,
+ &__entry->becn, &__entry->fecn,
+ &__entry->mig, &__entry->se,
+ &__entry->pad, &__entry->opcode,
+ &__entry->tver, &__entry->pkey,
+ &__entry->psn, &__entry->qpn);
+
/* extended headers */
- memcpy(__get_dynamic_array(ehdrs), &ohdr->u,
- ibhdr_exhdr_len(hdr));
- ),
- TP_printk("[%s] " LRH_PRN " " BTH_PRN " " EHDR_PRN,
- __get_str(dev),
- /* LRH */
- __entry->vl,
- __entry->lver,
- __entry->sl,
- __entry->lnh, show_lnh(__entry->lnh),
- __entry->dlid,
- __entry->len,
- __entry->slid,
- /* BTH */
- __entry->opcode, show_ib_opcode(__entry->opcode),
- __entry->se,
- __entry->m,
- __entry->pad,
- __entry->tver,
- __entry->pkey,
- __entry->f,
- __entry->b,
- __entry->qpn,
- __entry->a,
- __entry->psn,
- /* extended headers */
- __parse_ib_ehdrs(
- __entry->opcode,
- (void *)__get_dynamic_array(ehdrs))
- )
+ memcpy(__get_dynamic_array(ehdrs),
+ &ohdr->u, __get_dynamic_array_len(ehdrs));
+ ),
+ TP_printk("[%s] (IB) " LRH_PRN " " LRH_9B_PRN " "
+ BTH_PRN " " EHDR_PRN,
+ __get_str(dev),
+ __entry->len,
+ __entry->sc,
+ __entry->dlid,
+ __entry->slid,
+ __entry->lnh, show_lnh(__entry->lnh),
+ __entry->lver,
+ __entry->sl,
+ /* BTH */
+ __entry->opcode, show_ib_opcode(__entry->opcode),
+ __entry->se,
+ __entry->mig,
+ __entry->pad,
+ __entry->tver,
+ __entry->pkey,
+ __entry->fecn,
+ __entry->becn,
+ __entry->qpn,
+ __entry->ack,
+ __entry->psn,
+ /* extended headers */
+ __get_dynamic_array_len(ehdrs),
+ __parse_ib_ehdrs(
+ __entry->opcode,
+ (void *)__get_dynamic_array(ehdrs))
+ )
);
-DEFINE_EVENT(hfi1_ibhdr_template, input_ibhdr,
- TP_PROTO(struct hfi1_devdata *dd, struct ib_header *hdr),
- TP_ARGS(dd, hdr));
+DEFINE_EVENT(hfi1_output_ibhdr_template, pio_output_ibhdr,
+ TP_PROTO(struct hfi1_devdata *dd,
+ struct ib_header *hdr, bool sc5),
+ TP_ARGS(dd, hdr, sc5));
-DEFINE_EVENT(hfi1_ibhdr_template, pio_output_ibhdr,
- TP_PROTO(struct hfi1_devdata *dd, struct ib_header *hdr),
- TP_ARGS(dd, hdr));
+DEFINE_EVENT(hfi1_output_ibhdr_template, ack_output_ibhdr,
+ TP_PROTO(struct hfi1_devdata *dd,
+ struct ib_header *hdr, bool sc5),
+ TP_ARGS(dd, hdr, sc5));
-DEFINE_EVENT(hfi1_ibhdr_template, ack_output_ibhdr,
- TP_PROTO(struct hfi1_devdata *dd, struct ib_header *hdr),
- TP_ARGS(dd, hdr));
+DEFINE_EVENT(hfi1_output_ibhdr_template, sdma_output_ibhdr,
+ TP_PROTO(struct hfi1_devdata *dd,
+ struct ib_header *hdr, bool sc5),
+ TP_ARGS(dd, hdr, sc5));
-DEFINE_EVENT(hfi1_ibhdr_template, sdma_output_ibhdr,
- TP_PROTO(struct hfi1_devdata *dd, struct ib_header *hdr),
- TP_ARGS(dd, hdr));
#endif /* __HFI1_TRACE_IBHDRS_H */
diff --git a/drivers/infiniband/hw/hfi1/trace_rx.h b/drivers/infiniband/hw/hfi1/trace_rx.h
index f77e59f..05fc6d6 100644
--- a/drivers/infiniband/hw/hfi1/trace_rx.h
+++ b/drivers/infiniband/hw/hfi1/trace_rx.h
@@ -55,6 +55,15 @@
#undef TRACE_SYSTEM
#define TRACE_SYSTEM hfi1_rx
+#define packettype_name(etype) { RHF_RCV_TYPE_##etype, #etype }
+#define show_packettype(etype) \
+__print_symbolic(etype, \
+ packettype_name(EXPECTED), \
+ packettype_name(EAGER), \
+ packettype_name(IB), \
+ packettype_name(ERROR), \
+ packettype_name(BYPASS))
+
TRACE_EVENT(hfi1_rcvhdr,
TP_PROTO(struct hfi1_devdata *dd,
u32 ctxt,
diff --git a/drivers/infiniband/hw/hfi1/verbs.c b/drivers/infiniband/hw/hfi1/verbs.c
index 128d291..5f4be35 100644
--- a/drivers/infiniband/hw/hfi1/verbs.c
+++ b/drivers/infiniband/hw/hfi1/verbs.c
@@ -589,8 +589,7 @@ void hfi1_ib_rcv(struct hfi1_packet *packet)
goto drop;
}
- trace_input_ibhdr(rcd->dd, hdr);
-
+ trace_input_ibhdr(rcd->dd, packet, !!(packet->rhf & RHF_DC_INFO_SMASK));
opcode = ib_bth_get_opcode(packet->ohdr);
inc_opstats(tlen, &rcd->opstats->stats[opcode]);
@@ -885,7 +884,7 @@ int hfi1_verbs_send_dma(struct rvt_qp *qp, struct hfi1_pkt_state *ps,
return ret;
}
trace_sdma_output_ibhdr(dd_from_ibdev(qp->ibqp.device),
- &ps->s_txreq->phdr.hdr);
+ &ps->s_txreq->phdr.hdr, ib_is_sc5(sc5));
return ret;
bail_ecomm:
@@ -1058,7 +1057,7 @@ int hfi1_verbs_send_pio(struct rvt_qp *qp, struct hfi1_pkt_state *ps,
}
trace_pio_output_ibhdr(dd_from_ibdev(qp->ibqp.device),
- &ps->s_txreq->phdr.hdr);
+ &ps->s_txreq->phdr.hdr, ib_is_sc5(sc5));
pio_bail:
if (qp->s_wqe) {
--
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] 8+ messages in thread* [PATCH for-next 5/6] IB/hfi1: Setup common IB fields in hfi1_packet struct
[not found] ` <20170512161106.1290.45891.stgit-9QXIwq+3FY+1XWohqUldA0EOCMrvLtNR@public.gmane.org>
` (3 preceding siblings ...)
2017-05-12 16:20 ` [PATCH for-next 4/6] IB/hfi1: Separate input/output header tracing Dennis Dalessandro
@ 2017-05-12 16:20 ` Dennis Dalessandro
2017-05-12 16:20 ` [PATCH for-next 6/6] IB/rdmavt: Compress adjacent SGEs in rvt_lkey_ok() Dennis Dalessandro
2017-06-27 20:59 ` [PATCH for-next 0/6] IB/hfi1,qib: Cleanups and updates for-next Doug Ledford
6 siblings, 0 replies; 8+ messages in thread
From: Dennis Dalessandro @ 2017-05-12 16:20 UTC (permalink / raw)
To: dledford-H+wXaHxf7aLQT0dZR+AlfA
Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA, Don Hiatt, Brian Welty,
Dasaratharaman Chandramouli
From: Don Hiatt <don.hiatt-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
We move many common IB fields into the hfi1_packet structure and
set them up in a single function. This allows us to set the fields
in a single place and not deal with them throughout the driver.
Reviewed-by: Brian Welty <brian.welty-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Reviewed-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Reviewed-by: Dennis Dalessandro <dennis.dalessandro-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Don Hiatt <don.hiatt-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
drivers/infiniband/hw/hfi1/chip.c | 9 --
drivers/infiniband/hw/hfi1/chip.h | 2
drivers/infiniband/hw/hfi1/common.h | 1
drivers/infiniband/hw/hfi1/driver.c | 158 +++++++++++++++++++++++++----------
drivers/infiniband/hw/hfi1/hfi.h | 21 ++++-
drivers/infiniband/hw/hfi1/rc.c | 33 ++-----
drivers/infiniband/hw/hfi1/ruc.c | 89 +++++++++-----------
drivers/infiniband/hw/hfi1/uc.c | 16 +---
drivers/infiniband/hw/hfi1/ud.c | 23 ++---
drivers/infiniband/hw/hfi1/verbs.c | 89 ++++++++------------
drivers/infiniband/hw/hfi1/verbs.h | 6 -
11 files changed, 236 insertions(+), 211 deletions(-)
diff --git a/drivers/infiniband/hw/hfi1/chip.c b/drivers/infiniband/hw/hfi1/chip.c
index bbf7759..a4fb116 100644
--- a/drivers/infiniband/hw/hfi1/chip.c
+++ b/drivers/infiniband/hw/hfi1/chip.c
@@ -9810,15 +9810,6 @@ void hfi1_clear_tids(struct hfi1_ctxtdata *rcd)
hfi1_put_tid(dd, i, PT_INVALID, 0, 0);
}
-struct ib_header *hfi1_get_msgheader(
- struct hfi1_devdata *dd, __le32 *rhf_addr)
-{
- u32 offset = rhf_hdrq_offset(rhf_to_cpu(rhf_addr));
-
- return (struct ib_header *)
- (rhf_addr - dd->rhf_offset + offset);
-}
-
static const char * const ib_cfg_name_strings[] = {
"HFI1_IB_CFG_LIDLMC",
"HFI1_IB_CFG_LWID_DG_ENB",
diff --git a/drivers/infiniband/hw/hfi1/chip.h b/drivers/infiniband/hw/hfi1/chip.h
index cbe455d..0b4f418 100644
--- a/drivers/infiniband/hw/hfi1/chip.h
+++ b/drivers/infiniband/hw/hfi1/chip.h
@@ -1347,8 +1347,6 @@ enum {
u64 get_all_cpu_total(u64 __percpu *cntr);
void hfi1_start_cleanup(struct hfi1_devdata *dd);
void hfi1_clear_tids(struct hfi1_ctxtdata *rcd);
-struct ib_header *hfi1_get_msgheader(
- struct hfi1_devdata *dd, __le32 *rhf_addr);
void hfi1_init_ctxt(struct send_context *sc);
void hfi1_put_tid(struct hfi1_devdata *dd, u32 index,
u32 type, unsigned long pa, u16 order);
diff --git a/drivers/infiniband/hw/hfi1/common.h b/drivers/infiniband/hw/hfi1/common.h
index 995d62c..ba9ab97 100644
--- a/drivers/infiniband/hw/hfi1/common.h
+++ b/drivers/infiniband/hw/hfi1/common.h
@@ -325,6 +325,7 @@ struct diag_pkt {
#define HFI1_LRH_BTH 0x0002 /* 1. word of IB LRH - next header: BTH */
/* misc. */
+#define SC15_PACKET 0xF
#define SIZE_OF_CRC 1
#define LIM_MGMT_P_KEY 0x7FFF
diff --git a/drivers/infiniband/hw/hfi1/driver.c b/drivers/infiniband/hw/hfi1/driver.c
index 0583479..2a1022e 100644
--- a/drivers/infiniband/hw/hfi1/driver.c
+++ b/drivers/infiniband/hw/hfi1/driver.c
@@ -224,6 +224,20 @@ int hfi1_count_active_units(void)
(offset * RCV_BUF_BLOCK_SIZE));
}
+static inline void *hfi1_get_header(struct hfi1_devdata *dd,
+ __le32 *rhf_addr)
+{
+ u32 offset = rhf_hdrq_offset(rhf_to_cpu(rhf_addr));
+
+ return (void *)(rhf_addr - dd->rhf_offset + offset);
+}
+
+static inline struct ib_header *hfi1_get_msgheader(struct hfi1_devdata *dd,
+ __le32 *rhf_addr)
+{
+ return (struct ib_header *)hfi1_get_header(dd, rhf_addr);
+}
+
/*
* Validate and encode the a given RcvArray Buffer size.
* The function will check whether the given size falls within
@@ -249,7 +263,8 @@ static void rcv_hdrerr(struct hfi1_ctxtdata *rcd, struct hfi1_pportdata *ppd,
{
struct ib_header *rhdr = packet->hdr;
u32 rte = rhf_rcv_type_err(packet->rhf);
- int lnh = ib_get_lnh(rhdr);
+ u8 lnh = ib_get_lnh(rhdr);
+ bool has_grh = false;
struct hfi1_ibport *ibp = rcd_to_iport(rcd);
struct hfi1_devdata *dd = ppd->dd;
struct rvt_dev_info *rdi = &dd->verbs_dev.rdi;
@@ -257,37 +272,42 @@ static void rcv_hdrerr(struct hfi1_ctxtdata *rcd, struct hfi1_pportdata *ppd,
if (packet->rhf & (RHF_VCRC_ERR | RHF_ICRC_ERR))
return;
+ if (lnh == HFI1_LRH_BTH) {
+ packet->ohdr = &rhdr->u.oth;
+ } else if (lnh == HFI1_LRH_GRH) {
+ has_grh = true;
+ packet->ohdr = &rhdr->u.l.oth;
+ packet->grh = &rhdr->u.l.grh;
+ } else {
+ goto drop;
+ }
+
if (packet->rhf & RHF_TID_ERR) {
/* For TIDERR and RC QPs preemptively schedule a NAK */
- struct ib_other_headers *ohdr = NULL;
u32 tlen = rhf_pkt_len(packet->rhf); /* in bytes */
- u16 lid = ib_get_dlid(rhdr);
+ u32 dlid = ib_get_dlid(rhdr);
u32 qp_num;
- u32 rcv_flags = 0;
+ u32 mlid_base = be16_to_cpu(IB_MULTICAST_LID_BASE);
/* Sanity check packet */
if (tlen < 24)
goto drop;
/* Check for GRH */
- if (lnh == HFI1_LRH_BTH) {
- ohdr = &rhdr->u.oth;
- } else if (lnh == HFI1_LRH_GRH) {
+ if (has_grh) {
u32 vtf;
+ struct ib_grh *grh = packet->grh;
- ohdr = &rhdr->u.l.oth;
- if (rhdr->u.l.grh.next_hdr != IB_GRH_NEXT_HDR)
+ if (grh->next_hdr != IB_GRH_NEXT_HDR)
goto drop;
- vtf = be32_to_cpu(rhdr->u.l.grh.version_tclass_flow);
+ vtf = be32_to_cpu(grh->version_tclass_flow);
if ((vtf >> IB_GRH_VERSION_SHIFT) != IB_GRH_VERSION)
goto drop;
- rcv_flags |= HFI1_HAS_GRH;
- } else {
- goto drop;
}
+
/* Get the destination QP number. */
- qp_num = ib_bth_get_qpn(ohdr);
- if (lid < be16_to_cpu(IB_MULTICAST_LID_BASE)) {
+ qp_num = ib_bth_get_qpn(packet->ohdr);
+ if (dlid < mlid_base) {
struct rvt_qp *qp;
unsigned long flags;
@@ -312,11 +332,7 @@ static void rcv_hdrerr(struct hfi1_ctxtdata *rcd, struct hfi1_pportdata *ppd,
switch (qp->ibqp.qp_type) {
case IB_QPT_RC:
- hfi1_rc_hdrerr(
- rcd,
- rhdr,
- rcv_flags,
- qp);
+ hfi1_rc_hdrerr(rcd, packet, qp);
break;
default:
/* For now don't handle any other QP types */
@@ -332,9 +348,8 @@ static void rcv_hdrerr(struct hfi1_ctxtdata *rcd, struct hfi1_pportdata *ppd,
switch (rte) {
case RHF_RTE_ERROR_OP_CODE_ERR:
{
- u32 opcode;
void *ebuf = NULL;
- __be32 *bth = NULL;
+ u8 opcode;
if (rhf_use_egr_bfr(packet->rhf))
ebuf = packet->ebuf;
@@ -342,16 +357,7 @@ static void rcv_hdrerr(struct hfi1_ctxtdata *rcd, struct hfi1_pportdata *ppd,
if (!ebuf)
goto drop; /* this should never happen */
- if (lnh == HFI1_LRH_BTH)
- bth = (__be32 *)ebuf;
- else if (lnh == HFI1_LRH_GRH)
- bth = (__be32 *)((char *)ebuf + sizeof(struct ib_grh));
- else
- goto drop;
-
- opcode = be32_to_cpu(bth[0]) >> 24;
- opcode &= 0xff;
-
+ opcode = ib_bth_get_opcode(packet->ohdr);
if (opcode == IB_OPCODE_CNP) {
/*
* Only in pre-B0 h/w is the CNP_OPCODE handled
@@ -365,7 +371,7 @@ static void rcv_hdrerr(struct hfi1_ctxtdata *rcd, struct hfi1_pportdata *ppd,
sc5 = hfi1_9B_get_sc5(rhdr, packet->rhf);
sl = ibp->sc_to_sl[sc5];
- lqpn = be32_to_cpu(bth[1]) & RVT_QPN_MASK;
+ lqpn = ib_bth_get_qpn(packet->ohdr);
rcu_read_lock();
qp = rvt_lookup_qpn(rdi, &ibp->rvp, lqpn);
if (!qp) {
@@ -415,7 +421,6 @@ static inline void init_packet(struct hfi1_ctxtdata *rcd,
packet->rhf = rhf_to_cpu(packet->rhf_addr);
packet->rhqoff = rcd->head;
packet->numpkt = 0;
- packet->rcv_flags = 0;
}
void hfi1_process_ecn_slowpath(struct rvt_qp *qp, struct hfi1_packet *pkt,
@@ -424,15 +429,12 @@ void hfi1_process_ecn_slowpath(struct rvt_qp *qp, struct hfi1_packet *pkt,
struct hfi1_ibport *ibp = to_iport(qp->ibqp.device, qp->port_num);
struct ib_header *hdr = pkt->hdr;
struct ib_other_headers *ohdr = pkt->ohdr;
- struct ib_grh *grh = NULL;
+ struct ib_grh *grh = pkt->grh;
u32 rqpn = 0, bth1;
u16 rlid, dlid = ib_get_dlid(hdr);
u8 sc, svc_type;
bool is_mcast = false;
- if (pkt->rcv_flags & HFI1_HAS_GRH)
- grh = &hdr->u.l.grh;
-
switch (qp->ibqp.qp_type) {
case IB_QPT_SMI:
case IB_QPT_GSI:
@@ -591,9 +593,10 @@ static void __prescan_rxq(struct hfi1_packet *packet)
if (lnh == HFI1_LRH_BTH) {
packet->ohdr = &hdr->u.oth;
+ packet->grh = NULL;
} else if (lnh == HFI1_LRH_GRH) {
packet->ohdr = &hdr->u.l.oth;
- packet->rcv_flags |= HFI1_HAS_GRH;
+ packet->grh = &hdr->u.l.grh;
} else {
goto next; /* just in case */
}
@@ -698,10 +701,9 @@ static inline int process_rcv_packet(struct hfi1_packet *packet, int thread)
{
int ret;
- packet->hdr = hfi1_get_msgheader(packet->rcd->dd,
- packet->rhf_addr);
- packet->hlen = (u8 *)packet->rhf_addr - (u8 *)packet->hdr;
packet->etype = rhf_rcv_type(packet->rhf);
+
+ packet->hlen = (u8 *)packet->rhf_addr - (u8 *)packet->hdr;
/* total length */
packet->tlen = rhf_pkt_len(packet->rhf); /* in bytes */
/* retrieve eager buffer details */
@@ -759,7 +761,7 @@ static inline void process_rcv_update(int last, struct hfi1_packet *packet)
packet->etail, 0, 0);
packet->updegr = 0;
}
- packet->rcv_flags = 0;
+ packet->grh = NULL;
}
static inline void finish_packet(struct hfi1_packet *packet)
@@ -896,12 +898,15 @@ static inline int set_armed_to_active(struct hfi1_ctxtdata *rcd,
struct hfi1_devdata *dd)
{
struct work_struct *lsaw = &rcd->ppd->linkstate_active_work;
- struct ib_header *hdr = hfi1_get_msgheader(packet->rcd->dd,
- packet->rhf_addr);
u8 etype = rhf_rcv_type(packet->rhf);
+ u8 sc = SC15_PACKET;
- if (etype == RHF_RCV_TYPE_IB &&
- hfi1_9B_get_sc5(hdr, packet->rhf) != 0xf) {
+ if (etype == RHF_RCV_TYPE_IB) {
+ struct ib_header *hdr = hfi1_get_msgheader(packet->rcd->dd,
+ packet->rhf_addr);
+ sc = hfi1_9B_get_sc5(hdr, packet->rhf);
+ }
+ if (sc != SC15_PACKET) {
int hwstate = read_logical_state(dd);
if (hwstate != LSTATE_ACTIVE) {
@@ -1321,6 +1326,58 @@ int hfi1_reset_device(int unit)
return ret;
}
+static inline void hfi1_setup_ib_header(struct hfi1_packet *packet)
+{
+ packet->hdr = (struct hfi1_ib_message_header *)
+ hfi1_get_msgheader(packet->rcd->dd,
+ packet->rhf_addr);
+ packet->hlen = (u8 *)packet->rhf_addr - (u8 *)packet->hdr;
+}
+
+static int hfi1_setup_9B_packet(struct hfi1_packet *packet)
+{
+ struct hfi1_ibport *ibp = rcd_to_iport(packet->rcd);
+ struct ib_header *hdr;
+ u8 lnh;
+
+ hfi1_setup_ib_header(packet);
+ hdr = packet->hdr;
+
+ lnh = ib_get_lnh(hdr);
+ if (lnh == HFI1_LRH_BTH) {
+ packet->ohdr = &hdr->u.oth;
+ packet->grh = NULL;
+ } else if (lnh == HFI1_LRH_GRH) {
+ u32 vtf;
+
+ packet->ohdr = &hdr->u.l.oth;
+ packet->grh = &hdr->u.l.grh;
+ if (packet->grh->next_hdr != IB_GRH_NEXT_HDR)
+ goto drop;
+ vtf = be32_to_cpu(packet->grh->version_tclass_flow);
+ if ((vtf >> IB_GRH_VERSION_SHIFT) != IB_GRH_VERSION)
+ goto drop;
+ } else {
+ goto drop;
+ }
+
+ /* Query commonly used fields from packet header */
+ packet->opcode = ib_bth_get_opcode(packet->ohdr);
+ packet->slid = ib_get_slid(hdr);
+ packet->dlid = ib_get_dlid(hdr);
+ packet->sl = ib_get_sl(hdr);
+ packet->sc = hfi1_9B_get_sc5(hdr, packet->rhf);
+ packet->pad = ib_bth_get_pad(packet->ohdr);
+ packet->extra_byte = 0;
+ packet->fecn = ib_bth_get_fecn(packet->ohdr);
+ packet->becn = ib_bth_get_becn(packet->ohdr);
+
+ return 0;
+drop:
+ ibp->rvp.n_pkt_drops++;
+ return -EINVAL;
+}
+
void handle_eflags(struct hfi1_packet *packet)
{
struct hfi1_ctxtdata *rcd = packet->rcd;
@@ -1351,6 +1408,9 @@ int process_receive_ib(struct hfi1_packet *packet)
if (unlikely(hfi1_dbg_fault_packet(packet)))
return RHF_RCV_CONTINUE;
+ if (hfi1_setup_9B_packet(packet))
+ return RHF_RCV_CONTINUE;
+
trace_hfi1_rcvhdr(packet->rcd->ppd->dd,
packet->rcd->ctxt,
rhf_err_flags(packet->rhf),
@@ -1422,6 +1482,7 @@ int process_receive_error(struct hfi1_packet *packet)
rhf_rcv_type_err(packet->rhf) == 3))
return RHF_RCV_CONTINUE;
+ hfi1_setup_ib_header(packet);
handle_eflags(packet);
if (unlikely(rhf_err_flags(packet->rhf)))
@@ -1435,6 +1496,8 @@ int kdeth_process_expected(struct hfi1_packet *packet)
{
if (unlikely(hfi1_dbg_fault_packet(packet)))
return RHF_RCV_CONTINUE;
+
+ hfi1_setup_ib_header(packet);
if (unlikely(rhf_err_flags(packet->rhf)))
handle_eflags(packet);
@@ -1445,6 +1508,7 @@ int kdeth_process_expected(struct hfi1_packet *packet)
int kdeth_process_eager(struct hfi1_packet *packet)
{
+ hfi1_setup_ib_header(packet);
if (unlikely(rhf_err_flags(packet->rhf)))
handle_eflags(packet);
if (unlikely(hfi1_dbg_fault_packet(packet)))
diff --git a/drivers/infiniband/hw/hfi1/hfi.h b/drivers/infiniband/hw/hfi1/hfi.h
index 2fb0a54..cc69b51 100644
--- a/drivers/infiniband/hw/hfi1/hfi.h
+++ b/drivers/infiniband/hw/hfi1/hfi.h
@@ -356,17 +356,26 @@ struct hfi1_packet {
__le32 *rhf_addr;
struct rvt_qp *qp;
struct ib_other_headers *ohdr;
+ struct ib_grh *grh;
u64 rhf;
u32 maxcnt;
u32 rhqoff;
+ u32 dlid;
+ u32 slid;
u16 tlen;
s16 etail;
u8 hlen;
u8 numpkt;
u8 rsize;
u8 updegr;
- u8 rcv_flags;
u8 etype;
+ u8 extra_byte;
+ u8 pad;
+ u8 sc;
+ u8 sl;
+ u8 opcode;
+ bool becn;
+ bool fecn;
};
struct rvt_sge_state;
@@ -2087,4 +2096,14 @@ static inline bool is_integrated(struct hfi1_devdata *dd)
#define DD_DEV_ENTRY(dd) __string(dev, dev_name(&(dd)->pcidev->dev))
#define DD_DEV_ASSIGN(dd) __assign_str(dev, dev_name(&(dd)->pcidev->dev))
+
+/*
+ * hfi1_check_mcast- Check if the given lid is
+ * in the IB multicast range.
+ */
+static inline bool hfi1_check_mcast(u16 lid)
+{
+ return ((lid >= be16_to_cpu(IB_MULTICAST_LID_BASE)) &&
+ (lid != be16_to_cpu(IB_LID_PERMISSIVE)));
+}
#endif /* _HFI1_KERNEL_H */
diff --git a/drivers/infiniband/hw/hfi1/rc.c b/drivers/infiniband/hw/hfi1/rc.c
index b443c1e..baa67bf 100644
--- a/drivers/infiniband/hw/hfi1/rc.c
+++ b/drivers/infiniband/hw/hfi1/rc.c
@@ -1916,17 +1916,16 @@ void process_becn(struct hfi1_pportdata *ppd, u8 sl, u16 rlid, u32 lqpn,
void hfi1_rc_rcv(struct hfi1_packet *packet)
{
struct hfi1_ctxtdata *rcd = packet->rcd;
- struct ib_header *hdr = packet->hdr;
- u32 rcv_flags = packet->rcv_flags;
void *data = packet->ebuf;
u32 tlen = packet->tlen;
struct rvt_qp *qp = packet->qp;
struct hfi1_ibport *ibp = rcd_to_iport(rcd);
struct ib_other_headers *ohdr = packet->ohdr;
- u32 bth0, opcode;
+ u32 bth0;
+ u32 opcode = packet->opcode;
u32 hdrsize = packet->hlen;
u32 psn;
- u32 pad;
+ u32 pad = packet->pad;
struct ib_wc wc;
u32 pmtu = qp->pmtu;
int diff;
@@ -1938,14 +1937,13 @@ void hfi1_rc_rcv(struct hfi1_packet *packet)
u32 rkey;
lockdep_assert_held(&qp->r_lock);
+
bth0 = be32_to_cpu(ohdr->bth[0]);
- if (hfi1_ruc_check_hdr(ibp, hdr, rcv_flags & HFI1_HAS_GRH, qp, bth0))
+ if (hfi1_ruc_check_hdr(ibp, packet))
return;
is_fecn = process_ecn(qp, packet, false);
-
psn = ib_bth_get_psn(ohdr);
- opcode = ib_bth_get_opcode(ohdr);
/*
* Process responses (ACKs) before anything else. Note that the
@@ -2075,8 +2073,6 @@ void hfi1_rc_rcv(struct hfi1_packet *packet)
wc.wc_flags = 0;
wc.ex.imm_data = 0;
send_last:
- /* Get the number of bytes the message was padded by. */
- pad = ib_bth_get_pad(ohdr);
/* Check for invalid length. */
/* LAST len should be >= 1 */
if (unlikely(tlen < (hdrsize + pad + 4)))
@@ -2369,28 +2365,19 @@ void hfi1_rc_rcv(struct hfi1_packet *packet)
void hfi1_rc_hdrerr(
struct hfi1_ctxtdata *rcd,
- struct ib_header *hdr,
- u32 rcv_flags,
+ struct hfi1_packet *packet,
struct rvt_qp *qp)
{
- int has_grh = rcv_flags & HFI1_HAS_GRH;
- struct ib_other_headers *ohdr;
struct hfi1_ibport *ibp = rcd_to_iport(rcd);
int diff;
u32 opcode;
- u32 psn, bth0;
-
- /* Check for GRH */
- ohdr = &hdr->u.oth;
- if (has_grh)
- ohdr = &hdr->u.l.oth;
+ u32 psn;
- bth0 = be32_to_cpu(ohdr->bth[0]);
- if (hfi1_ruc_check_hdr(ibp, hdr, has_grh, qp, bth0))
+ if (hfi1_ruc_check_hdr(ibp, packet))
return;
- psn = ib_bth_get_psn(ohdr);
- opcode = ib_bth_get_opcode(ohdr);
+ psn = ib_bth_get_psn(packet->ohdr);
+ opcode = ib_bth_get_opcode(packet->ohdr);
/* Only deal with RDMA Writes for now */
if (opcode < IB_OPCODE_RC_RDMA_READ_RESPONSE_FIRST) {
diff --git a/drivers/infiniband/hw/hfi1/ruc.c b/drivers/infiniband/hw/hfi1/ruc.c
index 3a17dab..9cc9c7b 100644
--- a/drivers/infiniband/hw/hfi1/ruc.c
+++ b/drivers/infiniband/hw/hfi1/ruc.c
@@ -214,100 +214,95 @@ static int gid_ok(union ib_gid *gid, __be64 gid_prefix, __be64 id)
*
* The s_lock will be acquired around the hfi1_migrate_qp() call.
*/
-int hfi1_ruc_check_hdr(struct hfi1_ibport *ibp, struct ib_header *hdr,
- int has_grh, struct rvt_qp *qp, u32 bth0)
+int hfi1_ruc_check_hdr(struct hfi1_ibport *ibp, struct hfi1_packet *packet)
{
__be64 guid;
unsigned long flags;
+ struct rvt_qp *qp = packet->qp;
u8 sc5 = ibp->sl_to_sc[rdma_ah_get_sl(&qp->remote_ah_attr)];
-
- if (qp->s_mig_state == IB_MIG_ARMED && (bth0 & IB_BTH_MIG_REQ)) {
- if (!has_grh) {
+ u32 dlid = packet->dlid;
+ u32 slid = packet->slid;
+ u32 sl = packet->sl;
+ int migrated;
+ u32 bth0, bth1;
+
+ bth0 = be32_to_cpu(packet->ohdr->bth[0]);
+ bth1 = be32_to_cpu(packet->ohdr->bth[1]);
+ migrated = bth0 & IB_BTH_MIG_REQ;
+
+ if (qp->s_mig_state == IB_MIG_ARMED && migrated) {
+ if (!packet->grh) {
if (rdma_ah_get_ah_flags(&qp->alt_ah_attr) &
IB_AH_GRH)
- goto err;
+ return 1;
} else {
const struct ib_global_route *grh;
if (!(rdma_ah_get_ah_flags(&qp->alt_ah_attr) &
IB_AH_GRH))
- goto err;
+ return 1;
grh = rdma_ah_read_grh(&qp->alt_ah_attr);
guid = get_sguid(ibp, grh->sgid_index);
- if (!gid_ok(&hdr->u.l.grh.dgid, ibp->rvp.gid_prefix,
+ if (!gid_ok(&packet->grh->dgid, ibp->rvp.gid_prefix,
guid))
- goto err;
+ return 1;
if (!gid_ok(
- &hdr->u.l.grh.sgid,
+ &packet->grh->sgid,
grh->dgid.global.subnet_prefix,
grh->dgid.global.interface_id))
- goto err;
+ return 1;
}
- if (unlikely(rcv_pkey_check(ppd_from_ibp(ibp), (u16)bth0, sc5,
- ib_get_slid(hdr)))) {
- hfi1_bad_pqkey(ibp, OPA_TRAP_BAD_P_KEY,
- (u16)bth0,
- ib_get_sl(hdr),
- 0, qp->ibqp.qp_num,
- ib_get_slid(hdr),
- ib_get_dlid(hdr));
- goto err;
+ if (unlikely(rcv_pkey_check(ppd_from_ibp(ibp), (u16)bth0,
+ sc5, slid))) {
+ hfi1_bad_pqkey(ibp, OPA_TRAP_BAD_P_KEY, (u16)bth0, sl,
+ 0, qp->ibqp.qp_num, slid, dlid);
+ return 1;
}
/* Validate the SLID. See Ch. 9.6.1.5 and 17.2.8 */
- if (ib_get_slid(hdr) !=
- rdma_ah_get_dlid(&qp->alt_ah_attr) ||
+ if (slid != rdma_ah_get_dlid(&qp->alt_ah_attr) ||
ppd_from_ibp(ibp)->port !=
rdma_ah_get_port_num(&qp->alt_ah_attr))
- goto err;
+ return 1;
spin_lock_irqsave(&qp->s_lock, flags);
hfi1_migrate_qp(qp);
spin_unlock_irqrestore(&qp->s_lock, flags);
} else {
- if (!has_grh) {
+ if (!packet->grh) {
if (rdma_ah_get_ah_flags(&qp->remote_ah_attr) &
IB_AH_GRH)
- goto err;
+ return 1;
} else {
const struct ib_global_route *grh;
if (!(rdma_ah_get_ah_flags(&qp->remote_ah_attr) &
IB_AH_GRH))
- goto err;
+ return 1;
grh = rdma_ah_read_grh(&qp->remote_ah_attr);
guid = get_sguid(ibp, grh->sgid_index);
- if (!gid_ok(&hdr->u.l.grh.dgid, ibp->rvp.gid_prefix,
+ if (!gid_ok(&packet->grh->dgid, ibp->rvp.gid_prefix,
guid))
- goto err;
+ return 1;
if (!gid_ok(
- &hdr->u.l.grh.sgid,
+ &packet->grh->sgid,
grh->dgid.global.subnet_prefix,
grh->dgid.global.interface_id))
- goto err;
+ return 1;
}
- if (unlikely(rcv_pkey_check(ppd_from_ibp(ibp), (u16)bth0, sc5,
- ib_get_slid(hdr)))) {
- hfi1_bad_pqkey(ibp, OPA_TRAP_BAD_P_KEY,
- (u16)bth0,
- ib_get_sl(hdr),
- 0, qp->ibqp.qp_num,
- ib_get_slid(hdr),
- ib_get_dlid(hdr));
- goto err;
+ if (unlikely(rcv_pkey_check(ppd_from_ibp(ibp), (u16)bth0,
+ sc5, slid))) {
+ hfi1_bad_pqkey(ibp, OPA_TRAP_BAD_P_KEY, (u16)bth0, sl,
+ 0, qp->ibqp.qp_num, slid, dlid);
+ return 1;
}
/* Validate the SLID. See Ch. 9.6.1.5 */
- if (ib_get_slid(hdr) !=
- rdma_ah_get_dlid(&qp->remote_ah_attr) ||
+ if ((slid != rdma_ah_get_dlid(&qp->remote_ah_attr)) ||
ppd_from_ibp(ibp)->port != qp->port_num)
- goto err;
- if (qp->s_mig_state == IB_MIG_REARM &&
- !(bth0 & IB_BTH_MIG_REQ))
+ return 1;
+ if (qp->s_mig_state == IB_MIG_REARM && !migrated)
qp->s_mig_state = IB_MIG_ARMED;
}
return 0;
-
-err:
- return 1;
}
/**
diff --git a/drivers/infiniband/hw/hfi1/uc.c b/drivers/infiniband/hw/hfi1/uc.c
index 2a5650f..76c2451 100644
--- a/drivers/infiniband/hw/hfi1/uc.c
+++ b/drivers/infiniband/hw/hfi1/uc.c
@@ -297,31 +297,25 @@ int hfi1_make_uc_req(struct rvt_qp *qp, struct hfi1_pkt_state *ps)
void hfi1_uc_rcv(struct hfi1_packet *packet)
{
struct hfi1_ibport *ibp = rcd_to_iport(packet->rcd);
- struct ib_header *hdr = packet->hdr;
- u32 rcv_flags = packet->rcv_flags;
void *data = packet->ebuf;
u32 tlen = packet->tlen;
struct rvt_qp *qp = packet->qp;
struct ib_other_headers *ohdr = packet->ohdr;
- u32 bth0, opcode;
+ u32 opcode = packet->opcode;
u32 hdrsize = packet->hlen;
u32 psn;
- u32 pad;
+ u32 pad = packet->pad;
struct ib_wc wc;
u32 pmtu = qp->pmtu;
struct ib_reth *reth;
- int has_grh = rcv_flags & HFI1_HAS_GRH;
int ret;
- bth0 = be32_to_cpu(ohdr->bth[0]);
- if (hfi1_ruc_check_hdr(ibp, hdr, has_grh, qp, bth0))
+ if (hfi1_ruc_check_hdr(ibp, packet))
return;
process_ecn(qp, packet, true);
psn = ib_bth_get_psn(ohdr);
- opcode = ib_bth_get_opcode(ohdr);
-
/* Compare the PSN verses the expected PSN. */
if (unlikely(cmp_psn(psn, qp->r_psn) != 0)) {
/*
@@ -432,8 +426,6 @@ void hfi1_uc_rcv(struct hfi1_packet *packet)
wc.ex.imm_data = 0;
wc.wc_flags = 0;
send_last:
- /* Get the number of bytes the message was padded by. */
- pad = ib_bth_get_pad(ohdr);
/* Check for invalid length. */
/* LAST len should be >= 1 */
if (unlikely(tlen < (hdrsize + pad + 4)))
@@ -527,8 +519,6 @@ void hfi1_uc_rcv(struct hfi1_packet *packet)
rdma_last_imm:
wc.wc_flags = IB_WC_WITH_IMM;
- /* Get the number of bytes the message was padded by. */
- pad = ib_bth_get_pad(ohdr);
/* Check for invalid length. */
/* LAST len should be >= 1 */
if (unlikely(tlen < (hdrsize + pad + 4)))
diff --git a/drivers/infiniband/hw/hfi1/ud.c b/drivers/infiniband/hw/hfi1/ud.c
index 49fe179..c995aa5 100644
--- a/drivers/infiniband/hw/hfi1/ud.c
+++ b/drivers/infiniband/hw/hfi1/ud.c
@@ -668,36 +668,31 @@ static int opa_smp_check(struct hfi1_ibport *ibp, u16 pkey, u8 sc5,
void hfi1_ud_rcv(struct hfi1_packet *packet)
{
struct ib_other_headers *ohdr = packet->ohdr;
- int opcode;
u32 hdrsize = packet->hlen;
struct ib_wc wc;
u32 qkey;
u32 src_qp;
- u16 dlid, pkey;
+ u16 pkey;
int mgmt_pkey_idx = -1;
struct hfi1_ibport *ibp = rcd_to_iport(packet->rcd);
struct hfi1_pportdata *ppd = ppd_from_ibp(ibp);
struct ib_header *hdr = packet->hdr;
- u32 rcv_flags = packet->rcv_flags;
void *data = packet->ebuf;
u32 tlen = packet->tlen;
struct rvt_qp *qp = packet->qp;
- bool has_grh = rcv_flags & HFI1_HAS_GRH;
u8 sc5 = hfi1_9B_get_sc5(hdr, packet->rhf);
u32 bth1;
- u8 sl_from_sc, sl;
- u16 slid;
- u8 extra_bytes;
+ u8 sl_from_sc;
+ u8 extra_bytes = packet->pad;
+ u8 opcode = packet->opcode;
+ u8 sl = packet->sl;
+ u32 dlid = packet->dlid;
+ u32 slid = packet->slid;
+ bth1 = be32_to_cpu(ohdr->bth[1]);
qkey = ib_get_qkey(ohdr);
src_qp = ib_get_sqpn(ohdr);
- dlid = ib_get_dlid(hdr);
- bth1 = be32_to_cpu(ohdr->bth[1]);
- slid = ib_get_slid(hdr);
pkey = ib_bth_get_pkey(ohdr);
- opcode = ib_bth_get_opcode(ohdr);
- sl = ib_get_sl(hdr);
- extra_bytes = ib_bth_get_pad(ohdr);
extra_bytes += (SIZE_OF_CRC << 2);
sl_from_sc = ibp->sc_to_sl[sc5];
@@ -811,7 +806,7 @@ void hfi1_ud_rcv(struct hfi1_packet *packet)
qp->r_flags |= RVT_R_REUSE_SGE;
goto drop;
}
- if (has_grh) {
+ if (packet->grh) {
hfi1_copy_sge(&qp->r_sge, &hdr->u.l.grh,
sizeof(struct ib_grh), true, false);
wc.wc_flags |= IB_WC_GRH;
diff --git a/drivers/infiniband/hw/hfi1/verbs.c b/drivers/infiniband/hw/hfi1/verbs.c
index 5f4be35..af54d3f 100644
--- a/drivers/infiniband/hw/hfi1/verbs.c
+++ b/drivers/infiniband/hw/hfi1/verbs.c
@@ -508,13 +508,14 @@ void hfi1_copy_sge(
/*
* Make sure the QP is ready and able to accept the given opcode.
*/
-static inline opcode_handler qp_ok(int opcode, struct hfi1_packet *packet)
+static inline opcode_handler qp_ok(struct hfi1_packet *packet)
{
if (!(ib_rvt_state_ops[packet->qp->state] & RVT_PROCESS_RECV_OK))
return NULL;
- if (((opcode & RVT_OPCODE_QP_MASK) == packet->qp->allowed_ops) ||
- (opcode == IB_OPCODE_CNP))
- return opcode_handler_tbl[opcode];
+ if (((packet->opcode & RVT_OPCODE_QP_MASK) ==
+ packet->qp->allowed_ops) ||
+ (packet->opcode == IB_OPCODE_CNP))
+ return opcode_handler_tbl[packet->opcode];
return NULL;
}
@@ -548,68 +549,34 @@ static u64 hfi1_fault_tx(struct rvt_qp *qp, u8 opcode, u64 pbc)
return pbc;
}
-/**
- * hfi1_ib_rcv - process an incoming packet
- * @packet: data packet information
- *
- * This is called to process an incoming packet at interrupt level.
- *
- * Tlen is the length of the header + data + CRC in bytes.
- */
-void hfi1_ib_rcv(struct hfi1_packet *packet)
+static inline void hfi1_handle_packet(struct hfi1_packet *packet,
+ bool is_mcast)
{
+ u32 qp_num;
struct hfi1_ctxtdata *rcd = packet->rcd;
- struct ib_header *hdr = packet->hdr;
- u32 tlen = packet->tlen;
struct hfi1_pportdata *ppd = rcd->ppd;
struct hfi1_ibport *ibp = rcd_to_iport(rcd);
struct rvt_dev_info *rdi = &ppd->dd->verbs_dev.rdi;
opcode_handler packet_handler;
unsigned long flags;
- u32 qp_num;
- int lnh;
- u8 opcode;
- u16 lid;
-
- /* Check for GRH */
- lnh = ib_get_lnh(hdr);
- if (lnh == HFI1_LRH_BTH) {
- packet->ohdr = &hdr->u.oth;
- } else if (lnh == HFI1_LRH_GRH) {
- u32 vtf;
-
- packet->ohdr = &hdr->u.l.oth;
- if (hdr->u.l.grh.next_hdr != IB_GRH_NEXT_HDR)
- goto drop;
- vtf = be32_to_cpu(hdr->u.l.grh.version_tclass_flow);
- if ((vtf >> IB_GRH_VERSION_SHIFT) != IB_GRH_VERSION)
- goto drop;
- packet->rcv_flags |= HFI1_HAS_GRH;
- } else {
- goto drop;
- }
- trace_input_ibhdr(rcd->dd, packet, !!(packet->rhf & RHF_DC_INFO_SMASK));
- opcode = ib_bth_get_opcode(packet->ohdr);
- inc_opstats(tlen, &rcd->opstats->stats[opcode]);
+ inc_opstats(packet->tlen, &rcd->opstats->stats[packet->opcode]);
- /* Get the destination QP number. */
- qp_num = ib_bth_get_qpn(packet->ohdr);
- lid = ib_get_dlid(hdr);
- if (unlikely((lid >= be16_to_cpu(IB_MULTICAST_LID_BASE)) &&
- (lid != be16_to_cpu(IB_LID_PERMISSIVE)))) {
+ if (unlikely(is_mcast)) {
struct rvt_mcast *mcast;
struct rvt_mcast_qp *p;
- if (lnh != HFI1_LRH_GRH)
+ if (!packet->grh)
goto drop;
- mcast = rvt_mcast_find(&ibp->rvp, &hdr->u.l.grh.dgid, lid);
+ mcast = rvt_mcast_find(&ibp->rvp,
+ &packet->grh->dgid,
+ packet->dlid);
if (!mcast)
goto drop;
list_for_each_entry_rcu(p, &mcast->qp_list, list) {
packet->qp = p->qp;
spin_lock_irqsave(&packet->qp->r_lock, flags);
- packet_handler = qp_ok(opcode, packet);
+ packet_handler = qp_ok(packet);
if (likely(packet_handler))
packet_handler(packet);
else
@@ -623,19 +590,21 @@ void hfi1_ib_rcv(struct hfi1_packet *packet)
if (atomic_dec_return(&mcast->refcount) <= 1)
wake_up(&mcast->wait);
} else {
+ /* Get the destination QP number. */
+ qp_num = ib_bth_get_qpn(packet->ohdr);
rcu_read_lock();
packet->qp = rvt_lookup_qpn(rdi, &ibp->rvp, qp_num);
if (!packet->qp) {
rcu_read_unlock();
goto drop;
}
- if (unlikely(hfi1_dbg_fault_opcode(packet->qp, opcode,
+ if (unlikely(hfi1_dbg_fault_opcode(packet->qp, packet->opcode,
true))) {
rcu_read_unlock();
goto drop;
}
spin_lock_irqsave(&packet->qp->r_lock, flags);
- packet_handler = qp_ok(opcode, packet);
+ packet_handler = qp_ok(packet);
if (likely(packet_handler))
packet_handler(packet);
else
@@ -644,11 +613,29 @@ void hfi1_ib_rcv(struct hfi1_packet *packet)
rcu_read_unlock();
}
return;
-
drop:
ibp->rvp.n_pkt_drops++;
}
+/**
+ * hfi1_ib_rcv - process an incoming packet
+ * @packet: data packet information
+ *
+ * This is called to process an incoming packet at interrupt level.
+ */
+void hfi1_ib_rcv(struct hfi1_packet *packet)
+{
+ struct hfi1_ctxtdata *rcd = packet->rcd;
+ bool is_mcast = false;
+
+ if (unlikely(hfi1_check_mcast(packet->dlid)))
+ is_mcast = true;
+
+ trace_input_ibhdr(rcd->dd, packet,
+ !!(packet->rhf & RHF_DC_INFO_SMASK));
+ hfi1_handle_packet(packet, is_mcast);
+}
+
/*
* This is called from a timer to check for QPs
* which need kernel memory in order to send a packet.
diff --git a/drivers/infiniband/hw/hfi1/verbs.h b/drivers/infiniband/hw/hfi1/verbs.h
index cd635d0..17b38cd 100644
--- a/drivers/infiniband/hw/hfi1/verbs.h
+++ b/drivers/infiniband/hw/hfi1/verbs.h
@@ -307,8 +307,7 @@ void hfi1_copy_sge(struct rvt_sge_state *ss, void *data, u32 length,
void hfi1_rc_hdrerr(
struct hfi1_ctxtdata *rcd,
- struct ib_header *hdr,
- u32 rcv_flags,
+ struct hfi1_packet *packet,
struct rvt_qp *qp);
u8 ah_to_sc(struct ib_device *ibdev, struct rdma_ah_attr *ah_attr);
@@ -346,8 +345,7 @@ static inline u8 get_opcode(struct ib_header *h)
return be32_to_cpu(h->u.l.oth.bth[0]) >> 24;
}
-int hfi1_ruc_check_hdr(struct hfi1_ibport *ibp, struct ib_header *hdr,
- int has_grh, struct rvt_qp *qp, u32 bth0);
+int hfi1_ruc_check_hdr(struct hfi1_ibport *ibp, struct hfi1_packet *packet);
u32 hfi1_make_grh(struct hfi1_ibport *ibp, struct ib_grh *hdr,
const struct ib_global_route *grh, u32 hwords, u32 nwords);
--
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] 8+ messages in thread* [PATCH for-next 6/6] IB/rdmavt: Compress adjacent SGEs in rvt_lkey_ok()
[not found] ` <20170512161106.1290.45891.stgit-9QXIwq+3FY+1XWohqUldA0EOCMrvLtNR@public.gmane.org>
` (4 preceding siblings ...)
2017-05-12 16:20 ` [PATCH for-next 5/6] IB/hfi1: Setup common IB fields in hfi1_packet struct Dennis Dalessandro
@ 2017-05-12 16:20 ` Dennis Dalessandro
2017-06-27 20:59 ` [PATCH for-next 0/6] IB/hfi1,qib: Cleanups and updates for-next Doug Ledford
6 siblings, 0 replies; 8+ messages in thread
From: Dennis Dalessandro @ 2017-05-12 16:20 UTC (permalink / raw)
To: dledford-H+wXaHxf7aLQT0dZR+AlfA
Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA, Brian Welty, Mike Marciniszyn,
Venkata Sandeep Dhanalakota
From: Mike Marciniszyn <mike.marciniszyn-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
SGEs that are contiguous needlessly consume driver dependent TX resources.
The lkey validation logic is enhanced to compress the SGE that ends
up in the send wqe when consecutive addresses are detected.
The lkey validation API used to return 1 (success) or 0 (fail).
The return value is now an -errno, 0 (compressed), or 1 (uncompressed). A
additional argument is added to pass the last SQE for the compression.
Loopback callers always pass a NULL to last_sge since the optimization is
of little benefit in that situation.
Reviewed-by: Dennis Dalessandro <dennis.dalessandro-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Brian Welty <brian.welty-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Venkata Sandeep Dhanalakota <venkata.s.dhanalakota-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
drivers/infiniband/hw/hfi1/ruc.c | 2 +
drivers/infiniband/hw/qib/qib_ruc.c | 2 +
drivers/infiniband/sw/rdmavt/mr.c | 51 ++++++++++++++++++++++----
drivers/infiniband/sw/rdmavt/qp.c | 23 +++++++-----
drivers/infiniband/sw/rdmavt/trace_mr.h | 62 +++++++++++++++++++++++++++++++
drivers/infiniband/sw/rdmavt/trace_tx.h | 11 ++++--
include/rdma/rdma_vt.h | 3 +-
7 files changed, 130 insertions(+), 24 deletions(-)
diff --git a/drivers/infiniband/hw/hfi1/ruc.c b/drivers/infiniband/hw/hfi1/ruc.c
index 9cc9c7b..476fe5d 100644
--- a/drivers/infiniband/hw/hfi1/ruc.c
+++ b/drivers/infiniband/hw/hfi1/ruc.c
@@ -75,7 +75,7 @@ static int init_sge(struct rvt_qp *qp, struct rvt_rwqe *wqe)
continue;
/* Check LKEY */
if (!rvt_lkey_ok(rkt, pd, j ? &ss->sg_list[j - 1] : &ss->sge,
- &wqe->sg_list[i], IB_ACCESS_LOCAL_WRITE))
+ NULL, &wqe->sg_list[i], IB_ACCESS_LOCAL_WRITE))
goto bad_lkey;
qp->r_len += wqe->sg_list[i].length;
j++;
diff --git a/drivers/infiniband/hw/qib/qib_ruc.c b/drivers/infiniband/hw/qib/qib_ruc.c
index bd09de7..88d84cb 100644
--- a/drivers/infiniband/hw/qib/qib_ruc.c
+++ b/drivers/infiniband/hw/qib/qib_ruc.c
@@ -59,7 +59,7 @@ static int qib_init_sge(struct rvt_qp *qp, struct rvt_rwqe *wqe)
continue;
/* Check LKEY */
if (!rvt_lkey_ok(rkt, pd, j ? &ss->sg_list[j - 1] : &ss->sge,
- &wqe->sg_list[i], IB_ACCESS_LOCAL_WRITE))
+ NULL, &wqe->sg_list[i], IB_ACCESS_LOCAL_WRITE))
goto bad_lkey;
qp->r_len += wqe->sg_list[i].length;
j++;
diff --git a/drivers/infiniband/sw/rdmavt/mr.c b/drivers/infiniband/sw/rdmavt/mr.c
index aa5f9ea..ea95672 100644
--- a/drivers/infiniband/sw/rdmavt/mr.c
+++ b/drivers/infiniband/sw/rdmavt/mr.c
@@ -778,23 +778,54 @@ int rvt_dealloc_fmr(struct ib_fmr *ibfmr)
}
/**
+ * rvt_sge_adjacent - is isge compressible
+ * @isge: outgoing internal SGE
+ * @last_sge: last outgoing SGE written
+ * @sge: SGE to check
+ *
+ * If adjacent will update last_sge to add length.
+ *
+ * Return: true if isge is adjacent to last sge
+ */
+static inline bool rvt_sge_adjacent(struct rvt_sge *isge,
+ struct rvt_sge *last_sge,
+ struct ib_sge *sge)
+{
+ if (last_sge && sge->lkey == last_sge->mr->lkey &&
+ ((uint64_t)(last_sge->vaddr + last_sge->length) == sge->addr)) {
+ if (sge->lkey) {
+ if (unlikely((sge->addr - last_sge->mr->user_base +
+ sge->length > last_sge->mr->length)))
+ return false; /* overrun, caller will catch */
+ } else {
+ last_sge->length += sge->length;
+ }
+ last_sge->sge_length += sge->length;
+ trace_rvt_sge_adjacent(last_sge, sge);
+ return true;
+ }
+ return false;
+}
+
+/**
* rvt_lkey_ok - check IB SGE for validity and initialize
* @rkt: table containing lkey to check SGE against
* @pd: protection domain
* @isge: outgoing internal SGE
+ * @last_sge: last outgoing SGE written
* @sge: SGE to check
* @acc: access flags
*
* Check the IB SGE for validity and initialize our internal version
* of it.
*
- * Return: 1 if valid and successful, otherwise returns 0.
- *
- * increments the reference count upon success
+ * Increments the reference count when a new sge is stored.
*
+ * Return: 0 if compressed, 1 if added , otherwise returns -errno.
*/
int rvt_lkey_ok(struct rvt_lkey_table *rkt, struct rvt_pd *pd,
- struct rvt_sge *isge, struct ib_sge *sge, int acc)
+ struct rvt_sge *isge, struct rvt_sge *last_sge,
+ struct ib_sge *sge, int acc)
{
struct rvt_mregion *mr;
unsigned n, m;
@@ -804,12 +835,14 @@ int rvt_lkey_ok(struct rvt_lkey_table *rkt, struct rvt_pd *pd,
* We use LKEY == zero for kernel virtual addresses
* (see rvt_get_dma_mr() and dma_virt_ops).
*/
- rcu_read_lock();
if (sge->lkey == 0) {
struct rvt_dev_info *dev = ib_to_rvt(pd->ibpd.device);
if (pd->user)
- goto bail;
+ return -EINVAL;
+ if (rvt_sge_adjacent(isge, last_sge, sge))
+ return 0;
+ rcu_read_lock();
mr = rcu_dereference(dev->dma_mr);
if (!mr)
goto bail;
@@ -824,6 +857,9 @@ int rvt_lkey_ok(struct rvt_lkey_table *rkt, struct rvt_pd *pd,
isge->n = 0;
goto ok;
}
+ if (rvt_sge_adjacent(isge, last_sge, sge))
+ return 0;
+ rcu_read_lock();
mr = rcu_dereference(rkt->table[sge->lkey >> rkt->shift]);
if (!mr)
goto bail;
@@ -874,12 +910,13 @@ int rvt_lkey_ok(struct rvt_lkey_table *rkt, struct rvt_pd *pd,
isge->m = m;
isge->n = n;
ok:
+ trace_rvt_sge_new(isge, sge);
return 1;
bail_unref:
rvt_put_mr(mr);
bail:
rcu_read_unlock();
- return 0;
+ return -EINVAL;
}
EXPORT_SYMBOL(rvt_lkey_ok);
diff --git a/drivers/infiniband/sw/rdmavt/qp.c b/drivers/infiniband/sw/rdmavt/qp.c
index 727e81c..a3dd1e5 100644
--- a/drivers/infiniband/sw/rdmavt/qp.c
+++ b/drivers/infiniband/sw/rdmavt/qp.c
@@ -1646,7 +1646,7 @@ static int rvt_post_one_wr(struct rvt_qp *qp,
struct rvt_pd *pd;
struct rvt_dev_info *rdi = ib_to_rvt(qp->ibqp.device);
u8 log_pmtu;
- int ret;
+ int ret, incr;
size_t cplen;
bool reserved_op;
int local_ops_delayed = 0;
@@ -1719,22 +1719,23 @@ static int rvt_post_one_wr(struct rvt_qp *qp,
wqe->length = 0;
j = 0;
if (wr->num_sge) {
+ struct rvt_sge *last_sge = NULL;
+
acc = wr->opcode >= IB_WR_RDMA_READ ?
IB_ACCESS_LOCAL_WRITE : 0;
for (i = 0; i < wr->num_sge; i++) {
u32 length = wr->sg_list[i].length;
- int ok;
if (length == 0)
continue;
- ok = rvt_lkey_ok(rkt, pd, &wqe->sg_list[j],
- &wr->sg_list[i], acc);
- if (!ok) {
- ret = -EINVAL;
- goto bail_inval_free;
- }
+ incr = rvt_lkey_ok(rkt, pd, &wqe->sg_list[j], last_sge,
+ &wr->sg_list[i], acc);
+ if (unlikely(incr < 0))
+ goto bail_lkey_error;
wqe->length += length;
- j++;
+ if (incr)
+ last_sge = &wqe->sg_list[j];
+ j += incr;
}
wqe->wr.num_sge = j;
}
@@ -1781,12 +1782,14 @@ static int rvt_post_one_wr(struct rvt_qp *qp,
wqe->wr.send_flags &= ~RVT_SEND_RESERVE_USED;
qp->s_avail--;
}
- trace_rvt_post_one_wr(qp, wqe);
+ trace_rvt_post_one_wr(qp, wqe, wr->num_sge);
smp_wmb(); /* see request builders */
qp->s_head = next;
return 0;
+bail_lkey_error:
+ ret = incr;
bail_inval_free:
/* release mr holds */
while (j) {
diff --git a/drivers/infiniband/sw/rdmavt/trace_mr.h b/drivers/infiniband/sw/rdmavt/trace_mr.h
index 3318a6c..976e482 100644
--- a/drivers/infiniband/sw/rdmavt/trace_mr.h
+++ b/drivers/infiniband/sw/rdmavt/trace_mr.h
@@ -103,6 +103,68 @@
TP_PROTO(struct rvt_mregion *mr, u16 m, u16 n, void *v, size_t len),
TP_ARGS(mr, m, n, v, len));
+DECLARE_EVENT_CLASS(
+ rvt_sge_template,
+ TP_PROTO(struct rvt_sge *sge, struct ib_sge *isge),
+ TP_ARGS(sge, isge),
+ TP_STRUCT__entry(
+ RDI_DEV_ENTRY(ib_to_rvt(sge->mr->pd->device))
+ __field(struct rvt_mregion *, mr)
+ __field(struct rvt_sge *, sge)
+ __field(struct ib_sge *, isge)
+ __field(void *, vaddr)
+ __field(u64, ivaddr)
+ __field(u32, lkey)
+ __field(u32, sge_length)
+ __field(u32, length)
+ __field(u32, ilength)
+ __field(int, user)
+ __field(u16, m)
+ __field(u16, n)
+ ),
+ TP_fast_assign(
+ RDI_DEV_ASSIGN(ib_to_rvt(sge->mr->pd->device));
+ __entry->mr = sge->mr;
+ __entry->sge = sge;
+ __entry->isge = isge;
+ __entry->vaddr = sge->vaddr;
+ __entry->ivaddr = isge->addr;
+ __entry->lkey = sge->mr->lkey;
+ __entry->sge_length = sge->sge_length;
+ __entry->length = sge->length;
+ __entry->ilength = isge->length;
+ __entry->m = sge->m;
+ __entry->n = sge->m;
+ __entry->user = ibpd_to_rvtpd(sge->mr->pd)->user;
+ ),
+ TP_printk(
+ "[%s] mr %p sge %p isge %p vaddr %p ivaddr %llx lkey %x sge_length %u length %u ilength %u m %u n %u user %u",
+ __get_str(dev),
+ __entry->mr,
+ __entry->sge,
+ __entry->isge,
+ __entry->vaddr,
+ __entry->ivaddr,
+ __entry->lkey,
+ __entry->sge_length,
+ __entry->length,
+ __entry->ilength,
+ __entry->m,
+ __entry->n,
+ __entry->user
+ )
+);
+
+DEFINE_EVENT(
+ rvt_sge_template, rvt_sge_adjacent,
+ TP_PROTO(struct rvt_sge *sge, struct ib_sge *isge),
+ TP_ARGS(sge, isge));
+
+DEFINE_EVENT(
+ rvt_sge_template, rvt_sge_new,
+ TP_PROTO(struct rvt_sge *sge, struct ib_sge *isge),
+ TP_ARGS(sge, isge));
+
#endif /* __RVT_TRACE_MR_H */
#undef TRACE_INCLUDE_PATH
diff --git a/drivers/infiniband/sw/rdmavt/trace_tx.h b/drivers/infiniband/sw/rdmavt/trace_tx.h
index a613a22..0ef25fc 100644
--- a/drivers/infiniband/sw/rdmavt/trace_tx.h
+++ b/drivers/infiniband/sw/rdmavt/trace_tx.h
@@ -84,12 +84,12 @@
wr_opcode_name(RESERVED10))
#define POS_PRN \
-"[%s] wqe %p wr_id %llx send_flags %x qpn %x qpt %u psn %x lpsn %x ssn %x length %u opcode 0x%.2x,%s size %u avail %u head %u last %u pid %u num_sge %u"
+"[%s] wqe %p wr_id %llx send_flags %x qpn %x qpt %u psn %x lpsn %x ssn %x length %u opcode 0x%.2x,%s size %u avail %u head %u last %u pid %u num_sge %u wr_num_sge %u"
TRACE_EVENT(
rvt_post_one_wr,
- TP_PROTO(struct rvt_qp *qp, struct rvt_swqe *wqe),
- TP_ARGS(qp, wqe),
+ TP_PROTO(struct rvt_qp *qp, struct rvt_swqe *wqe, int wr_num_sge),
+ TP_ARGS(qp, wqe, wr_num_sge),
TP_STRUCT__entry(
RDI_DEV_ENTRY(ib_to_rvt(qp->ibqp.device))
__field(u64, wr_id)
@@ -108,6 +108,7 @@
__field(int, send_flags)
__field(pid_t, pid)
__field(int, num_sge)
+ __field(int, wr_num_sge)
),
TP_fast_assign(
RDI_DEV_ASSIGN(ib_to_rvt(qp->ibqp.device))
@@ -127,6 +128,7 @@
__entry->ssn = wqe->ssn;
__entry->send_flags = wqe->wr.send_flags;
__entry->num_sge = wqe->wr.num_sge;
+ __entry->wr_num_sge = wr_num_sge;
),
TP_printk(
POS_PRN,
@@ -146,7 +148,8 @@
__entry->head,
__entry->last,
__entry->pid,
- __entry->num_sge
+ __entry->num_sge,
+ __entry->wr_num_sge
)
);
diff --git a/include/rdma/rdma_vt.h b/include/rdma/rdma_vt.h
index 4878aaf..d0b9f91 100644
--- a/include/rdma/rdma_vt.h
+++ b/include/rdma/rdma_vt.h
@@ -515,7 +515,8 @@ int rvt_fast_reg_mr(struct rvt_qp *qp, struct ib_mr *ibmr, u32 key,
int rvt_rkey_ok(struct rvt_qp *qp, struct rvt_sge *sge,
u32 len, u64 vaddr, u32 rkey, int acc);
int rvt_lkey_ok(struct rvt_lkey_table *rkt, struct rvt_pd *pd,
- struct rvt_sge *isge, struct ib_sge *sge, int acc);
+ struct rvt_sge *isge, struct rvt_sge *last_sge,
+ struct ib_sge *sge, int acc);
struct rvt_mcast *rvt_mcast_find(struct rvt_ibport *ibp, union ib_gid *mgid,
u16 lid);
--
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] 8+ messages in thread* Re: [PATCH for-next 0/6] IB/hfi1,qib: Cleanups and updates for-next
[not found] ` <20170512161106.1290.45891.stgit-9QXIwq+3FY+1XWohqUldA0EOCMrvLtNR@public.gmane.org>
` (5 preceding siblings ...)
2017-05-12 16:20 ` [PATCH for-next 6/6] IB/rdmavt: Compress adjacent SGEs in rvt_lkey_ok() Dennis Dalessandro
@ 2017-06-27 20:59 ` Doug Ledford
6 siblings, 0 replies; 8+ messages in thread
From: Doug Ledford @ 2017-06-27 20:59 UTC (permalink / raw)
To: Dennis Dalessandro
Cc: Mike Marciniszyn, linux-rdma-u79uwXL29TY76Z2rM5mHXA, Ira Weiny,
Brian Welty, Venkata Sandeep Dhanalakota, Don Hiatt,
Sebastian Sanchez, Dasaratharaman Chandramouli
On Fri, 2017-05-12 at 09:19 -0700, Dennis Dalessandro wrote:
> Hi Doug,
>
> Here are some patches for-next. There are some clean ups to remove
> unused code
> and useless initializations. We also continue cleaning up the code by
> packing
> more fields into the hfi1_packet to keep them in a single place and
> adding
> some enhancements to the input/output header tracing.
>
> These apply on top of the two patches I just sent for RC.
>
> Patches can can also be found in my GitHub repo at:
> https://github.com/ddalessa/kernel/tree/for-4.13
Hi Denny,
Series applied.
--
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
^ permalink raw reply [flat|nested] 8+ messages in thread