Merge branch 'cxgb4-tos'

Hariprasad Shenai says:

====================
Add TOS support and some cleanup

This series adds TOS support for iWARP and also does some cleanup to make
code more readable. Patch series is created against infiniband tree and
includes patches on iw_cxgb4 and cxgb4 driver.

We have included all the maintainers of respective drivers. Kindly review
the change and let us know in case of any review comments.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

4 files changed, 80 insertions, 58 deletions

diff --git a/drivers/infiniband/hw/cxgb4/cm.c b/drivers/infiniband/hw/cxgb4/cm.c
index cd2ff5f9518a..8cd7b5e4b03d 100644
--- a/drivers/infiniband/hw/cxgb4/cm.c
+++ b/drivers/infiniband/hw/cxgb4/cm.c
@@ -485,12 +485,19 @@ static void send_flowc(struct c4iw_ep *ep, struct sk_buff *skb)
 	unsigned int flowclen = 80;
 	struct fw_flowc_wr *flowc;
 	int i;
+	u16 vlan = ep->l2t->vlan;
+	int nparams;
+
+	if (vlan == CPL_L2T_VLAN_NONE)
+		nparams = 8;
+	else
+		nparams = 9;
 
 	skb = get_skb(skb, flowclen, GFP_KERNEL);
 	flowc = (struct fw_flowc_wr *)__skb_put(skb, flowclen);
 
 	flowc->op_to_nparams = cpu_to_be32(FW_WR_OP_V(FW_FLOWC_WR) |
-					   FW_FLOWC_WR_NPARAMS_V(8));
+					   FW_FLOWC_WR_NPARAMS_V(nparams));
 	flowc->flowid_len16 = cpu_to_be32(FW_WR_LEN16_V(DIV_ROUND_UP(flowclen,
 					  16)) | FW_WR_FLOWID_V(ep->hwtid));
 
@@ -511,9 +518,17 @@ static void send_flowc(struct c4iw_ep *ep, struct sk_buff *skb)
 	flowc->mnemval[6].val = cpu_to_be32(ep->snd_win);
 	flowc->mnemval[7].mnemonic = FW_FLOWC_MNEM_MSS;
 	flowc->mnemval[7].val = cpu_to_be32(ep->emss);
-	/* Pad WR to 16 byte boundary */
-	flowc->mnemval[8].mnemonic = 0;
-	flowc->mnemval[8].val = 0;
+	if (nparams == 9) {
+		u16 pri;
+
+		pri = (vlan & VLAN_PRIO_MASK) >> VLAN_PRIO_SHIFT;
+		flowc->mnemval[8].mnemonic = FW_FLOWC_MNEM_SCHEDCLASS;
+		flowc->mnemval[8].val = cpu_to_be32(pri);
+	} else {
+		/* Pad WR to 16 byte boundary */
+		flowc->mnemval[8].mnemonic = 0;
+		flowc->mnemval[8].val = 0;
+	}
 	for (i = 0; i < 9; i++) {
 		flowc->mnemval[i].r4[0] = 0;
 		flowc->mnemval[i].r4[1] = 0;
@@ -710,7 +725,7 @@ static int send_connect(struct c4iw_ep *ep)
 	       L2T_IDX_V(ep->l2t->idx) |
 	       TX_CHAN_V(ep->tx_chan) |
 	       SMAC_SEL_V(ep->smac_idx) |
-	       DSCP_V(ep->tos) |
+	       DSCP_V(ep->tos >> 2) |
 	       ULP_MODE_V(ULP_MODE_TCPDDP) |
 	       RCV_BUFSIZ_V(win);
 	opt2 = RX_CHANNEL_V(0) |
@@ -1864,7 +1879,7 @@ static void send_fw_act_open_req(struct c4iw_ep *ep, unsigned int atid)
 		L2T_IDX_V(ep->l2t->idx) |
 		TX_CHAN_V(ep->tx_chan) |
 		SMAC_SEL_V(ep->smac_idx) |
-		DSCP_V(ep->tos) |
+		DSCP_V(ep->tos >> 2) |
 		ULP_MODE_V(ULP_MODE_TCPDDP) |
 		RCV_BUFSIZ_V(win));
 	req->tcb.opt2 = (__force __be32) (PACE_V(1) |
@@ -1928,7 +1943,7 @@ static void set_tcp_window(struct c4iw_ep *ep, struct port_info *pi)
 
 static int import_ep(struct c4iw_ep *ep, int iptype, __u8 *peer_ip,
 		     struct dst_entry *dst, struct c4iw_dev *cdev,
-		     bool clear_mpa_v1, enum chip_type adapter_type)
+		     bool clear_mpa_v1, enum chip_type adapter_type, u8 tos)
 {
 	struct neighbour *n;
 	int err, step;
@@ -1958,7 +1973,7 @@ static int import_ep(struct c4iw_ep *ep, int iptype, __u8 *peer_ip,
 			goto out;
 		}
 		ep->l2t = cxgb4_l2t_get(cdev->rdev.lldi.l2t,
-					n, pdev, 0);
+					n, pdev, rt_tos2priority(tos));
 		if (!ep->l2t)
 			goto out;
 		ep->mtu = pdev->mtu;
@@ -2041,7 +2056,7 @@ static int c4iw_reconnect(struct c4iw_ep *ep)
 	if (ep->com.cm_id->local_addr.ss_family == AF_INET) {
 		ep->dst = find_route(ep->com.dev, laddr->sin_addr.s_addr,
 				     raddr->sin_addr.s_addr, laddr->sin_port,
-				     raddr->sin_port, 0);
+				     raddr->sin_port, ep->com.cm_id->tos);
 		iptype = 4;
 		ra = (__u8 *)&raddr->sin_addr;
 	} else {
@@ -2058,7 +2073,8 @@ static int c4iw_reconnect(struct c4iw_ep *ep)
 		goto fail3;
 	}
 	err = import_ep(ep, iptype, ra, ep->dst, ep->com.dev, false,
-			ep->com.dev->rdev.lldi.adapter_type);
+			ep->com.dev->rdev.lldi.adapter_type,
+			ep->com.cm_id->tos);
 	if (err) {
 		pr_err("%s - cannot alloc l2e.\n", __func__);
 		goto fail4;
@@ -2069,7 +2085,7 @@ static int c4iw_reconnect(struct c4iw_ep *ep)
 	     ep->l2t->idx);
 
 	state_set(&ep->com, CONNECTING);
-	ep->tos = 0;
+	ep->tos = ep->com.cm_id->tos;
 
 	/* send connect request to rnic */
 	err = send_connect(ep);
@@ -2391,6 +2407,7 @@ static int pass_accept_req(struct c4iw_dev *dev, struct sk_buff *skb)
 	u16 peer_mss = ntohs(req->tcpopt.mss);
 	int iptype;
 	unsigned short hdrs;
+	u8 tos = PASS_OPEN_TOS_G(ntohl(req->tos_stid));
 
 	parent_ep = lookup_stid(t, stid);
 	if (!parent_ep) {
@@ -2399,8 +2416,7 @@ static int pass_accept_req(struct c4iw_dev *dev, struct sk_buff *skb)
 	}
 
 	if (state_read(&parent_ep->com) != LISTEN) {
-		printk(KERN_ERR "%s - listening ep not in LISTEN\n",
-		       __func__);
+		PDBG("%s - listening ep not in LISTEN\n", __func__);
 		goto reject;
 	}
 
@@ -2415,7 +2431,7 @@ static int pass_accept_req(struct c4iw_dev *dev, struct sk_buff *skb)
 		     ntohs(peer_port), peer_mss);
 		dst = find_route(dev, *(__be32 *)local_ip, *(__be32 *)peer_ip,
 				 local_port, peer_port,
-				 PASS_OPEN_TOS_G(ntohl(req->tos_stid)));
+				 tos);
 	} else {
 		PDBG("%s parent ep %p hwtid %u laddr %pI6 raddr %pI6 lport %d rport %d peer_mss %d\n"
 		     , __func__, parent_ep, hwtid,
@@ -2441,7 +2457,7 @@ static int pass_accept_req(struct c4iw_dev *dev, struct sk_buff *skb)
 	}
 
 	err = import_ep(child_ep, iptype, peer_ip, dst, dev, false,
-			parent_ep->com.dev->rdev.lldi.adapter_type);
+			parent_ep->com.dev->rdev.lldi.adapter_type, tos);
 	if (err) {
 		printk(KERN_ERR MOD "%s - failed to allocate l2t entry!\n",
 		       __func__);
@@ -2509,7 +2525,7 @@ static int pass_accept_req(struct c4iw_dev *dev, struct sk_buff *skb)
 
 	c4iw_get_ep(&parent_ep->com);
 	child_ep->parent_ep = parent_ep;
-	child_ep->tos = PASS_OPEN_TOS_G(ntohl(req->tos_stid));
+	child_ep->tos = tos;
 	child_ep->dst = dst;
 	child_ep->hwtid = hwtid;
 
@@ -3203,7 +3219,7 @@ int c4iw_connect(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param)
 		     ra, ntohs(raddr->sin_port));
 		ep->dst = find_route(dev, laddr->sin_addr.s_addr,
 				     raddr->sin_addr.s_addr, laddr->sin_port,
-				     raddr->sin_port, 0);
+				     raddr->sin_port, cm_id->tos);
 	} else {
 		iptype = 6;
 		ra = (__u8 *)&raddr6->sin6_addr;
@@ -3234,7 +3250,7 @@ int c4iw_connect(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param)
 	}
 
 	err = import_ep(ep, iptype, ra, ep->dst, ep->com.dev, true,
-			ep->com.dev->rdev.lldi.adapter_type);
+			ep->com.dev->rdev.lldi.adapter_type, cm_id->tos);
 	if (err) {
 		printk(KERN_ERR MOD "%s - cannot alloc l2e.\n", __func__);
 		goto fail3;
@@ -3245,7 +3261,7 @@ int c4iw_connect(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param)
 		ep->l2t->idx);
 
 	state_set(&ep->com, CONNECTING);
-	ep->tos = 0;
+	ep->tos = cm_id->tos;
 
 	/* send connect request to rnic */
 	err = send_connect(ep);
diff --git a/drivers/infiniband/hw/cxgb4/qp.c b/drivers/infiniband/hw/cxgb4/qp.c
index e99345eb875a..dadf5f1855b3 100644
--- a/drivers/infiniband/hw/cxgb4/qp.c
+++ b/drivers/infiniband/hw/cxgb4/qp.c
@@ -1621,7 +1621,8 @@ struct ib_qp *c4iw_create_qp(struct ib_pd *pd, struct ib_qp_init_attr *attrs,
 	unsigned int sqsize, rqsize;
 	struct c4iw_ucontext *ucontext;
 	int ret;
-	struct c4iw_mm_entry *mm1, *mm2, *mm3, *mm4, *mm5 = NULL;
+	struct c4iw_mm_entry *sq_key_mm, *rq_key_mm = NULL, *sq_db_key_mm;
+	struct c4iw_mm_entry *rq_db_key_mm = NULL, *ma_sync_key_mm = NULL;
 
 	PDBG("%s ib_pd %p\n", __func__, pd);
 
@@ -1706,29 +1707,30 @@ struct ib_qp *c4iw_create_qp(struct ib_pd *pd, struct ib_qp_init_attr *attrs,
 		goto err2;
 
 	if (udata) {
-		mm1 = kmalloc(sizeof *mm1, GFP_KERNEL);
-		if (!mm1) {
+		sq_key_mm = kmalloc(sizeof(*sq_key_mm), GFP_KERNEL);
+		if (!sq_key_mm) {
 			ret = -ENOMEM;
 			goto err3;
 		}
-		mm2 = kmalloc(sizeof *mm2, GFP_KERNEL);
-		if (!mm2) {
+		rq_key_mm = kmalloc(sizeof(*rq_key_mm), GFP_KERNEL);
+		if (!rq_key_mm) {
 			ret = -ENOMEM;
 			goto err4;
 		}
-		mm3 = kmalloc(sizeof *mm3, GFP_KERNEL);
-		if (!mm3) {
+		sq_db_key_mm = kmalloc(sizeof(*sq_db_key_mm), GFP_KERNEL);
+		if (!sq_db_key_mm) {
 			ret = -ENOMEM;
 			goto err5;
 		}
-		mm4 = kmalloc(sizeof *mm4, GFP_KERNEL);
-		if (!mm4) {
+		rq_db_key_mm = kmalloc(sizeof(*rq_db_key_mm), GFP_KERNEL);
+		if (!rq_db_key_mm) {
 			ret = -ENOMEM;
 			goto err6;
 		}
 		if (t4_sq_onchip(&qhp->wq.sq)) {
-			mm5 = kmalloc(sizeof *mm5, GFP_KERNEL);
-			if (!mm5) {
+			ma_sync_key_mm = kmalloc(sizeof(*ma_sync_key_mm),
+						 GFP_KERNEL);
+			if (!ma_sync_key_mm) {
 				ret = -ENOMEM;
 				goto err7;
 			}
@@ -1743,7 +1745,7 @@ struct ib_qp *c4iw_create_qp(struct ib_pd *pd, struct ib_qp_init_attr *attrs,
 		uresp.rq_size = qhp->wq.rq.size;
 		uresp.rq_memsize = qhp->wq.rq.memsize;
 		spin_lock(&ucontext->mmap_lock);
-		if (mm5) {
+		if (ma_sync_key_mm) {
 			uresp.ma_sync_key = ucontext->key;
 			ucontext->key += PAGE_SIZE;
 		} else {
@@ -1761,28 +1763,29 @@ struct ib_qp *c4iw_create_qp(struct ib_pd *pd, struct ib_qp_init_attr *attrs,
 		ret = ib_copy_to_udata(udata, &uresp, sizeof uresp);
 		if (ret)
 			goto err8;
-		mm1->key = uresp.sq_key;
-		mm1->addr = qhp->wq.sq.phys_addr;
-		mm1->len = PAGE_ALIGN(qhp->wq.sq.memsize);
-		insert_mmap(ucontext, mm1);
-		mm2->key = uresp.rq_key;
-		mm2->addr = virt_to_phys(qhp->wq.rq.queue);
-		mm2->len = PAGE_ALIGN(qhp->wq.rq.memsize);
-		insert_mmap(ucontext, mm2);
-		mm3->key = uresp.sq_db_gts_key;
-		mm3->addr = (__force unsigned long)qhp->wq.sq.bar2_pa;
-		mm3->len = PAGE_SIZE;
-		insert_mmap(ucontext, mm3);
-		mm4->key = uresp.rq_db_gts_key;
-		mm4->addr = (__force unsigned long)qhp->wq.rq.bar2_pa;
-		mm4->len = PAGE_SIZE;
-		insert_mmap(ucontext, mm4);
-		if (mm5) {
-			mm5->key = uresp.ma_sync_key;
-			mm5->addr = (pci_resource_start(rhp->rdev.lldi.pdev, 0)
-				    + PCIE_MA_SYNC_A) & PAGE_MASK;
-			mm5->len = PAGE_SIZE;
-			insert_mmap(ucontext, mm5);
+		sq_key_mm->key = uresp.sq_key;
+		sq_key_mm->addr = qhp->wq.sq.phys_addr;
+		sq_key_mm->len = PAGE_ALIGN(qhp->wq.sq.memsize);
+		insert_mmap(ucontext, sq_key_mm);
+		rq_key_mm->key = uresp.rq_key;
+		rq_key_mm->addr = virt_to_phys(qhp->wq.rq.queue);
+		rq_key_mm->len = PAGE_ALIGN(qhp->wq.rq.memsize);
+		insert_mmap(ucontext, rq_key_mm);
+		sq_db_key_mm->key = uresp.sq_db_gts_key;
+		sq_db_key_mm->addr = (u64)(unsigned long)qhp->wq.sq.bar2_pa;
+		sq_db_key_mm->len = PAGE_SIZE;
+		insert_mmap(ucontext, sq_db_key_mm);
+		rq_db_key_mm->key = uresp.rq_db_gts_key;
+		rq_db_key_mm->addr = (u64)(unsigned long)qhp->wq.rq.bar2_pa;
+		rq_db_key_mm->len = PAGE_SIZE;
+		insert_mmap(ucontext, rq_db_key_mm);
+		if (ma_sync_key_mm) {
+			ma_sync_key_mm->key = uresp.ma_sync_key;
+			ma_sync_key_mm->addr =
+				(pci_resource_start(rhp->rdev.lldi.pdev, 0) +
+				PCIE_MA_SYNC_A) & PAGE_MASK;
+			ma_sync_key_mm->len = PAGE_SIZE;
+			insert_mmap(ucontext, ma_sync_key_mm);
 		}
 	}
 	qhp->ibqp.qp_num = qhp->wq.sq.qid;
@@ -1795,15 +1798,15 @@ struct ib_qp *c4iw_create_qp(struct ib_pd *pd, struct ib_qp_init_attr *attrs,
 	     qhp->wq.rq.memsize, attrs->cap.max_recv_wr);
 	return &qhp->ibqp;
 err8:
-	kfree(mm5);
+	kfree(ma_sync_key_mm);
 err7:
-	kfree(mm4);
+	kfree(rq_db_key_mm);
 err6:
-	kfree(mm3);
+	kfree(sq_db_key_mm);
 err5:
-	kfree(mm2);
+	kfree(rq_key_mm);
 err4:
-	kfree(mm1);
+	kfree(sq_key_mm);
 err3:
 	remove_handle(rhp, &rhp->qpidr, qhp->wq.sq.qid);
 err2:
diff --git a/drivers/net/ethernet/chelsio/cxgb4/t4_msg.h b/drivers/net/ethernet/chelsio/cxgb4/t4_msg.h
index a072d341e205..1d2d1da40c80 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/t4_msg.h
+++ b/drivers/net/ethernet/chelsio/cxgb4/t4_msg.h
@@ -1021,6 +1021,8 @@ struct cpl_l2t_write_req {
 #define L2T_W_NOREPLY_V(x) ((x) << L2T_W_NOREPLY_S)
 #define L2T_W_NOREPLY_F    L2T_W_NOREPLY_V(1U)
 
+#define CPL_L2T_VLAN_NONE 0xfff
+
 struct cpl_l2t_write_rpl {
 	union opcode_tid ot;
 	u8 status;
diff --git a/drivers/net/ethernet/chelsio/cxgb4/t4fw_api.h b/drivers/net/ethernet/chelsio/cxgb4/t4fw_api.h
index a32de30ea663..c8661c77b4e3 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/t4fw_api.h
+++ b/drivers/net/ethernet/chelsio/cxgb4/t4fw_api.h
@@ -561,6 +561,7 @@ enum fw_flowc_mnem {
 	FW_FLOWC_MNEM_SNDBUF,
 	FW_FLOWC_MNEM_MSS,
 	FW_FLOWC_MNEM_TXDATAPLEN_MAX,
+	FW_FLOWC_MNEM_SCHEDCLASS = 11,
 };
 
 struct fw_flowc_mnemval {