aboutsummaryrefslogtreecommitdiff
path: root/drivers/infiniband/hw/ipath/ipath_uc.c
diff options
context:
space:
mode:
Diffstat (limited to 'drivers/infiniband/hw/ipath/ipath_uc.c')
-rw-r--r--drivers/infiniband/hw/ipath/ipath_uc.c298
1 files changed, 100 insertions, 198 deletions
diff --git a/drivers/infiniband/hw/ipath/ipath_uc.c b/drivers/infiniband/hw/ipath/ipath_uc.c
index 0d6dbc0a541..22e60998f1a 100644
--- a/drivers/infiniband/hw/ipath/ipath_uc.c
+++ b/drivers/infiniband/hw/ipath/ipath_uc.c
@@ -1,4 +1,5 @@
/*
+ * Copyright (c) 2006, 2007, 2008 QLogic Corporation. All rights reserved.
* Copyright (c) 2005, 2006 PathScale, Inc. All rights reserved.
*
* This software is available to you under a choice of one of two
@@ -31,127 +32,65 @@
*/
#include "ipath_verbs.h"
-#include "ips_common.h"
+#include "ipath_kernel.h"
/* cut down ridiculously long IB macro names */
#define OP(x) IB_OPCODE_UC_##x
-static void complete_last_send(struct ipath_qp *qp, struct ipath_swqe *wqe,
- struct ib_wc *wc)
-{
- if (++qp->s_last == qp->s_size)
- qp->s_last = 0;
- if (!test_bit(IPATH_S_SIGNAL_REQ_WR, &qp->s_flags) ||
- (wqe->wr.send_flags & IB_SEND_SIGNALED)) {
- wc->wr_id = wqe->wr.wr_id;
- wc->status = IB_WC_SUCCESS;
- wc->opcode = ib_ipath_wc_opcode[wqe->wr.opcode];
- wc->vendor_err = 0;
- wc->byte_len = wqe->length;
- wc->qp_num = qp->ibqp.qp_num;
- wc->src_qp = qp->remote_qpn;
- wc->pkey_index = 0;
- wc->slid = qp->remote_ah_attr.dlid;
- wc->sl = qp->remote_ah_attr.sl;
- wc->dlid_path_bits = 0;
- wc->port_num = 0;
- ipath_cq_enter(to_icq(qp->ibqp.send_cq), wc, 0);
- }
- wqe = get_swqe_ptr(qp, qp->s_last);
-}
-
/**
- * ipath_do_uc_send - do a send on a UC queue
- * @data: contains a pointer to the QP to send on
+ * ipath_make_uc_req - construct a request packet (SEND, RDMA write)
+ * @qp: a pointer to the QP
*
- * Process entries in the send work queue until the queue is exhausted.
- * Only allow one CPU to send a packet per QP (tasklet).
- * Otherwise, after we drop the QP lock, two threads could send
- * packets out of order.
- * This is similar to ipath_do_rc_send() below except we don't have
- * timeouts or resends.
+ * Return 1 if constructed; otherwise, return 0.
*/
-void ipath_do_uc_send(unsigned long data)
+int ipath_make_uc_req(struct ipath_qp *qp)
{
- struct ipath_qp *qp = (struct ipath_qp *)data;
- struct ipath_ibdev *dev = to_idev(qp->ibqp.device);
+ struct ipath_other_headers *ohdr;
struct ipath_swqe *wqe;
unsigned long flags;
- u16 lrh0;
u32 hwords;
- u32 nwords;
- u32 extra_bytes;
u32 bth0;
- u32 bth2;
- u32 pmtu = ib_mtu_enum_to_int(qp->path_mtu);
u32 len;
- struct ipath_other_headers *ohdr;
- struct ib_wc wc;
+ u32 pmtu = ib_mtu_enum_to_int(qp->path_mtu);
+ int ret = 0;
- if (test_and_set_bit(IPATH_S_BUSY, &qp->s_flags))
- goto bail;
+ spin_lock_irqsave(&qp->s_lock, flags);
- if (unlikely(qp->remote_ah_attr.dlid ==
- ipath_layer_get_lid(dev->dd))) {
- /* Pass in an uninitialized ib_wc to save stack space. */
- ipath_ruc_loopback(qp, &wc);
- clear_bit(IPATH_S_BUSY, &qp->s_flags);
- goto bail;
+ if (!(ib_ipath_state_ops[qp->state] & IPATH_PROCESS_SEND_OK)) {
+ if (!(ib_ipath_state_ops[qp->state] & IPATH_FLUSH_SEND))
+ goto bail;
+ /* We are in the error state, flush the work request. */
+ if (qp->s_last == qp->s_head)
+ goto bail;
+ /* If DMAs are in progress, we can't flush immediately. */
+ if (atomic_read(&qp->s_dma_busy)) {
+ qp->s_flags |= IPATH_S_WAIT_DMA;
+ goto bail;
+ }
+ wqe = get_swqe_ptr(qp, qp->s_last);
+ ipath_send_complete(qp, wqe, IB_WC_WR_FLUSH_ERR);
+ goto done;
}
ohdr = &qp->s_hdr.u.oth;
if (qp->remote_ah_attr.ah_flags & IB_AH_GRH)
ohdr = &qp->s_hdr.u.l.oth;
-again:
- /* Check for a constructed packet to be sent. */
- if (qp->s_hdrwords != 0) {
- /*
- * If no PIO bufs are available, return.
- * An interrupt will call ipath_ib_piobufavail()
- * when one is available.
- */
- if (ipath_verbs_send(dev->dd, qp->s_hdrwords,
- (u32 *) &qp->s_hdr,
- qp->s_cur_size,
- qp->s_cur_sge)) {
- ipath_no_bufs_available(qp, dev);
- goto bail;
- }
- dev->n_unicast_xmit++;
- /* Record that we sent the packet and s_hdr is empty. */
- qp->s_hdrwords = 0;
- }
-
- lrh0 = IPS_LRH_BTH;
/* header size in 32-bit words LRH+BTH = (8+12)/4. */
hwords = 5;
+ bth0 = 1 << 22; /* Set M bit */
- /*
- * The lock is needed to synchronize between
- * setting qp->s_ack_state and post_send().
- */
- spin_lock_irqsave(&qp->s_lock, flags);
-
- if (!(ib_ipath_state_ops[qp->state] & IPATH_PROCESS_SEND_OK))
- goto done;
-
- bth0 = ipath_layer_get_pkey(dev->dd, qp->s_pkey_index);
-
- /* Send a request. */
- wqe = get_swqe_ptr(qp, qp->s_last);
+ /* Get the next send request. */
+ wqe = get_swqe_ptr(qp, qp->s_cur);
+ qp->s_wqe = NULL;
switch (qp->s_state) {
default:
- /*
- * Signal the completion of the last send (if there is
- * one).
- */
- if (qp->s_last != qp->s_tail)
- complete_last_send(qp, wqe, &wc);
-
+ if (!(ib_ipath_state_ops[qp->state] &
+ IPATH_PROCESS_NEXT_SEND_OK))
+ goto bail;
/* Check if send work queue is empty. */
- if (qp->s_tail == qp->s_head)
- goto done;
+ if (qp->s_cur == qp->s_head)
+ goto bail;
/*
* Start a new request.
*/
@@ -174,11 +113,14 @@ again:
qp->s_state =
OP(SEND_ONLY_WITH_IMMEDIATE);
/* Immediate data comes after the BTH */
- ohdr->u.imm_data = wqe->wr.imm_data;
+ ohdr->u.imm_data = wqe->wr.ex.imm_data;
hwords += 1;
}
if (wqe->wr.send_flags & IB_SEND_SOLICITED)
bth0 |= 1 << 23;
+ qp->s_wqe = wqe;
+ if (++qp->s_cur >= qp->s_size)
+ qp->s_cur = 0;
break;
case IB_WR_RDMA_WRITE:
@@ -200,18 +142,19 @@ again:
qp->s_state =
OP(RDMA_WRITE_ONLY_WITH_IMMEDIATE);
/* Immediate data comes after the RETH */
- ohdr->u.rc.imm_data = wqe->wr.imm_data;
+ ohdr->u.rc.imm_data = wqe->wr.ex.imm_data;
hwords += 1;
if (wqe->wr.send_flags & IB_SEND_SOLICITED)
bth0 |= 1 << 23;
}
+ qp->s_wqe = wqe;
+ if (++qp->s_cur >= qp->s_size)
+ qp->s_cur = 0;
break;
default:
- goto done;
+ goto bail;
}
- if (++qp->s_tail >= qp->s_size)
- qp->s_tail = 0;
break;
case OP(SEND_FIRST):
@@ -228,11 +171,14 @@ again:
else {
qp->s_state = OP(SEND_LAST_WITH_IMMEDIATE);
/* Immediate data comes after the BTH */
- ohdr->u.imm_data = wqe->wr.imm_data;
+ ohdr->u.imm_data = wqe->wr.ex.imm_data;
hwords += 1;
}
if (wqe->wr.send_flags & IB_SEND_SOLICITED)
bth0 |= 1 << 23;
+ qp->s_wqe = wqe;
+ if (++qp->s_cur >= qp->s_size)
+ qp->s_cur = 0;
break;
case OP(RDMA_WRITE_FIRST):
@@ -250,68 +196,32 @@ again:
qp->s_state =
OP(RDMA_WRITE_LAST_WITH_IMMEDIATE);
/* Immediate data comes after the BTH */
- ohdr->u.imm_data = wqe->wr.imm_data;
+ ohdr->u.imm_data = wqe->wr.ex.imm_data;
hwords += 1;
if (wqe->wr.send_flags & IB_SEND_SOLICITED)
bth0 |= 1 << 23;
}
+ qp->s_wqe = wqe;
+ if (++qp->s_cur >= qp->s_size)
+ qp->s_cur = 0;
break;
}
- bth2 = qp->s_next_psn++ & IPS_PSN_MASK;
qp->s_len -= len;
- bth0 |= qp->s_state << 24;
-
- spin_unlock_irqrestore(&qp->s_lock, flags);
-
- /* Construct the header. */
- extra_bytes = (4 - len) & 3;
- nwords = (len + extra_bytes) >> 2;
- if (unlikely(qp->remote_ah_attr.ah_flags & IB_AH_GRH)) {
- /* Header size in 32-bit words. */
- hwords += 10;
- lrh0 = IPS_LRH_GRH;
- qp->s_hdr.u.l.grh.version_tclass_flow =
- cpu_to_be32((6 << 28) |
- (qp->remote_ah_attr.grh.traffic_class
- << 20) |
- qp->remote_ah_attr.grh.flow_label);
- qp->s_hdr.u.l.grh.paylen =
- cpu_to_be16(((hwords - 12) + nwords +
- SIZE_OF_CRC) << 2);
- /* next_hdr is defined by C8-7 in ch. 8.4.1 */
- qp->s_hdr.u.l.grh.next_hdr = 0x1B;
- qp->s_hdr.u.l.grh.hop_limit =
- qp->remote_ah_attr.grh.hop_limit;
- /* The SGID is 32-bit aligned. */
- qp->s_hdr.u.l.grh.sgid.global.subnet_prefix =
- dev->gid_prefix;
- qp->s_hdr.u.l.grh.sgid.global.interface_id =
- ipath_layer_get_guid(dev->dd);
- qp->s_hdr.u.l.grh.dgid = qp->remote_ah_attr.grh.dgid;
- }
qp->s_hdrwords = hwords;
qp->s_cur_sge = &qp->s_sge;
qp->s_cur_size = len;
- lrh0 |= qp->remote_ah_attr.sl << 4;
- qp->s_hdr.lrh[0] = cpu_to_be16(lrh0);
- /* DEST LID */
- qp->s_hdr.lrh[1] = cpu_to_be16(qp->remote_ah_attr.dlid);
- qp->s_hdr.lrh[2] = cpu_to_be16(hwords + nwords + SIZE_OF_CRC);
- qp->s_hdr.lrh[3] = cpu_to_be16(ipath_layer_get_lid(dev->dd));
- bth0 |= extra_bytes << 20;
- ohdr->bth[0] = cpu_to_be32(bth0);
- ohdr->bth[1] = cpu_to_be32(qp->remote_qpn);
- ohdr->bth[2] = cpu_to_be32(bth2);
-
- /* Check for more work to do. */
- goto again;
-
+ ipath_make_ruc_header(to_idev(qp->ibqp.device),
+ qp, ohdr, bth0 | (qp->s_state << 24),
+ qp->s_next_psn++ & IPATH_PSN_MASK);
done:
- spin_unlock_irqrestore(&qp->s_lock, flags);
- clear_bit(IPATH_S_BUSY, &qp->s_flags);
+ ret = 1;
+ goto unlock;
bail:
- return;
+ qp->s_flags &= ~IPATH_S_BUSY;
+unlock:
+ spin_unlock_irqrestore(&qp->s_lock, flags);
+ return ret;
}
/**
@@ -335,12 +245,15 @@ void ipath_uc_rcv(struct ipath_ibdev *dev, struct ipath_ib_header *hdr,
u32 hdrsize;
u32 psn;
u32 pad;
- unsigned long flags;
struct ib_wc wc;
u32 pmtu = ib_mtu_enum_to_int(qp->path_mtu);
struct ib_reth *reth;
int header_in_data;
+ /* Validate the SLID. See Ch. 9.6.1.5 */
+ if (unlikely(be16_to_cpu(hdr->lrh[3]) != qp->remote_ah_attr.dlid))
+ goto done;
+
/* Check for GRH */
if (!has_grh) {
ohdr = &hdr->u.oth;
@@ -356,8 +269,7 @@ void ipath_uc_rcv(struct ipath_ibdev *dev, struct ipath_ib_header *hdr,
* size to 56 bytes so the last 4 bytes of
* the BTH header (PSN) is in the data buffer.
*/
- header_in_data =
- ipath_layer_get_rcvhdrentsize(dev->dd) == 16;
+ header_in_data = dev->dd->ipath_rcvhdrentsize == 16;
if (header_in_data) {
psn = be32_to_cpu(((__be32 *) data)[0]);
data += sizeof(__be32);
@@ -370,10 +282,7 @@ void ipath_uc_rcv(struct ipath_ibdev *dev, struct ipath_ib_header *hdr,
*/
opcode = be32_to_cpu(ohdr->bth[0]) >> 24;
- wc.imm_data = 0;
- wc.wc_flags = 0;
-
- spin_lock_irqsave(&qp->r_rq.lock, flags);
+ memset(&wc, 0, sizeof wc);
/* Compare the PSN verses the expected PSN. */
if (unlikely(ipath_cmp24(psn, qp->r_psn) != 0)) {
@@ -436,15 +345,15 @@ void ipath_uc_rcv(struct ipath_ibdev *dev, struct ipath_ib_header *hdr,
case OP(SEND_ONLY):
case OP(SEND_ONLY_WITH_IMMEDIATE):
send_first:
- if (qp->r_reuse_sge) {
- qp->r_reuse_sge = 0;
- qp->r_sge = qp->s_rdma_sge;
+ if (qp->r_flags & IPATH_R_REUSE_SGE) {
+ qp->r_flags &= ~IPATH_R_REUSE_SGE;
+ qp->r_sge = qp->s_rdma_read_sge;
} else if (!ipath_get_rwqe(qp, 0)) {
dev->n_pkt_drops++;
goto done;
}
/* Save the WQE so we can reuse it in case of an error. */
- qp->s_rdma_sge = qp->r_sge;
+ qp->s_rdma_read_sge = qp->r_sge;
qp->r_rcv_len = 0;
if (opcode == OP(SEND_ONLY))
goto send_last;
@@ -454,13 +363,13 @@ void ipath_uc_rcv(struct ipath_ibdev *dev, struct ipath_ib_header *hdr,
case OP(SEND_MIDDLE):
/* Check for invalid length PMTU or posted rwqe len. */
if (unlikely(tlen != (hdrsize + pmtu + 4))) {
- qp->r_reuse_sge = 1;
+ qp->r_flags |= IPATH_R_REUSE_SGE;
dev->n_pkt_drops++;
goto done;
}
qp->r_rcv_len += pmtu;
if (unlikely(qp->r_rcv_len > qp->r_len)) {
- qp->r_reuse_sge = 1;
+ qp->r_flags |= IPATH_R_REUSE_SGE;
dev->n_pkt_drops++;
goto done;
}
@@ -470,11 +379,11 @@ void ipath_uc_rcv(struct ipath_ibdev *dev, struct ipath_ib_header *hdr,
case OP(SEND_LAST_WITH_IMMEDIATE):
send_last_imm:
if (header_in_data) {
- wc.imm_data = *(__be32 *) data;
+ wc.ex.imm_data = *(__be32 *) data;
data += sizeof(__be32);
} else {
/* Immediate data comes after BTH */
- wc.imm_data = ohdr->u.imm_data;
+ wc.ex.imm_data = ohdr->u.imm_data;
}
hdrsize += 4;
wc.wc_flags = IB_WC_WITH_IMM;
@@ -486,7 +395,7 @@ void ipath_uc_rcv(struct ipath_ibdev *dev, struct ipath_ib_header *hdr,
/* Check for invalid length. */
/* XXX LAST len should be >= 1 */
if (unlikely(tlen < (hdrsize + pad + 4))) {
- qp->r_reuse_sge = 1;
+ qp->r_flags |= IPATH_R_REUSE_SGE;
dev->n_pkt_drops++;
goto done;
}
@@ -494,28 +403,23 @@ void ipath_uc_rcv(struct ipath_ibdev *dev, struct ipath_ib_header *hdr,
tlen -= (hdrsize + pad + 4);
wc.byte_len = tlen + qp->r_rcv_len;
if (unlikely(wc.byte_len > qp->r_len)) {
- qp->r_reuse_sge = 1;
+ qp->r_flags |= IPATH_R_REUSE_SGE;
dev->n_pkt_drops++;
goto done;
}
- /* XXX Need to free SGEs */
+ wc.opcode = IB_WC_RECV;
last_imm:
ipath_copy_sge(&qp->r_sge, data, tlen);
wc.wr_id = qp->r_wr_id;
wc.status = IB_WC_SUCCESS;
- wc.opcode = IB_WC_RECV;
- wc.vendor_err = 0;
- wc.qp_num = qp->ibqp.qp_num;
+ wc.qp = &qp->ibqp;
wc.src_qp = qp->remote_qpn;
- wc.pkey_index = 0;
wc.slid = qp->remote_ah_attr.dlid;
wc.sl = qp->remote_ah_attr.sl;
- wc.dlid_path_bits = 0;
- wc.port_num = 0;
/* Signal completion event if the solicited bit is set. */
ipath_cq_enter(to_icq(qp->ibqp.recv_cq), &wc,
(ohdr->bth[0] &
- __constant_cpu_to_be32(1 << 23)) != 0);
+ cpu_to_be32(1 << 23)) != 0);
break;
case OP(RDMA_WRITE_FIRST):
@@ -535,12 +439,13 @@ void ipath_uc_rcv(struct ipath_ibdev *dev, struct ipath_ib_header *hdr,
if (qp->r_len != 0) {
u32 rkey = be32_to_cpu(reth->rkey);
u64 vaddr = be64_to_cpu(reth->vaddr);
+ int ok;
/* Check rkey */
- if (unlikely(!ipath_rkey_ok(
- dev, &qp->r_sge, qp->r_len,
- vaddr, rkey,
- IB_ACCESS_REMOTE_WRITE))) {
+ ok = ipath_rkey_ok(qp, &qp->r_sge, qp->r_len,
+ vaddr, rkey,
+ IB_ACCESS_REMOTE_WRITE);
+ if (unlikely(!ok)) {
dev->n_pkt_drops++;
goto done;
}
@@ -558,8 +463,7 @@ void ipath_uc_rcv(struct ipath_ibdev *dev, struct ipath_ib_header *hdr,
}
if (opcode == OP(RDMA_WRITE_ONLY))
goto rdma_last;
- else if (opcode ==
- OP(RDMA_WRITE_ONLY_WITH_IMMEDIATE))
+ else if (opcode == OP(RDMA_WRITE_ONLY_WITH_IMMEDIATE))
goto rdma_last_imm;
/* FALLTHROUGH */
case OP(RDMA_WRITE_MIDDLE):
@@ -578,6 +482,16 @@ void ipath_uc_rcv(struct ipath_ibdev *dev, struct ipath_ib_header *hdr,
case OP(RDMA_WRITE_LAST_WITH_IMMEDIATE):
rdma_last_imm:
+ if (header_in_data) {
+ wc.ex.imm_data = *(__be32 *) data;
+ data += sizeof(__be32);
+ } else {
+ /* Immediate data comes after BTH */
+ wc.ex.imm_data = ohdr->u.imm_data;
+ }
+ hdrsize += 4;
+ wc.wc_flags = IB_WC_WITH_IMM;
+
/* Get the number of bytes the message was padded by. */
pad = (be32_to_cpu(ohdr->bth[0]) >> 20) & 3;
/* Check for invalid length. */
@@ -592,22 +506,14 @@ void ipath_uc_rcv(struct ipath_ibdev *dev, struct ipath_ib_header *hdr,
dev->n_pkt_drops++;
goto done;
}
- if (qp->r_reuse_sge) {
- qp->r_reuse_sge = 0;
- } else if (!ipath_get_rwqe(qp, 1)) {
+ if (qp->r_flags & IPATH_R_REUSE_SGE)
+ qp->r_flags &= ~IPATH_R_REUSE_SGE;
+ else if (!ipath_get_rwqe(qp, 1)) {
dev->n_pkt_drops++;
goto done;
}
- if (header_in_data) {
- wc.imm_data = *(__be32 *) data;
- data += sizeof(__be32);
- } else {
- /* Immediate data comes after BTH */
- wc.imm_data = ohdr->u.imm_data;
- }
- hdrsize += 4;
- wc.wc_flags = IB_WC_WITH_IMM;
- wc.byte_len = 0;
+ wc.byte_len = qp->r_len;
+ wc.opcode = IB_WC_RECV_RDMA_WITH_IMM;
goto last_imm;
case OP(RDMA_WRITE_LAST):
@@ -631,15 +537,11 @@ void ipath_uc_rcv(struct ipath_ibdev *dev, struct ipath_ib_header *hdr,
default:
/* Drop packet for unknown opcodes. */
- spin_unlock_irqrestore(&qp->r_rq.lock, flags);
dev->n_pkt_drops++;
- goto bail;
+ goto done;
}
qp->r_psn++;
qp->r_state = opcode;
done:
- spin_unlock_irqrestore(&qp->r_rq.lock, flags);
-
-bail:
return;
}
generated by cgit v1.2.3-70-g09d2 (git 2.46.0) at 2025-12-20 16:17:32 +0000