From: Potnuri Bharat Teja <bharat@chelsio.com>
Date: Fri, 3 Aug 2018 18:26:47 +0530
Subject: iw_cxgb4: pass window scale in flowc work request
Patch-mainline: v4.19-rc1
Git-commit: 2e51e45cf613491a2bd9d757f04e36d8617be5ac
References: bsc#1104276 FATE#325935

This will allow FW to not send more data to TP (which would then need to
be buffered). Pass the negotiated TCP window scale to FW in the FLOWC WR.

Also refactor send_flowc() a bit to clean it up.

Signed-off-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Potnuri Bharat Teja <bharat@chelsio.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Acked-by: Thomas Bogendoerfer <tbogendoerfer@suse.de>
---
 drivers/infiniband/hw/cxgb4/cm.c       |   46 ++++++++++++++++-----------------
 drivers/infiniband/hw/cxgb4/iw_cxgb4.h |    6 +++-
 2 files changed, 28 insertions(+), 24 deletions(-)

--- a/drivers/infiniband/hw/cxgb4/cm.c
+++ b/drivers/infiniband/hw/cxgb4/cm.c
@@ -589,24 +589,29 @@ static int send_flowc(struct c4iw_ep *ep
 {
 	struct fw_flowc_wr *flowc;
 	struct sk_buff *skb = skb_dequeue(&ep->com.ep_skb_list);
-	int i;
 	u16 vlan = ep->l2t->vlan;
 	int nparams;
+	int flowclen, flowclen16;
 
 	if (WARN_ON(!skb))
 		return -ENOMEM;
 
 	if (vlan == CPL_L2T_VLAN_NONE)
-		nparams = 8;
-	else
 		nparams = 9;
+	else
+		nparams = 10;
+
+	flowclen = offsetof(struct fw_flowc_wr, mnemval[nparams]);
+	flowclen16 = DIV_ROUND_UP(flowclen, 16);
+	flowclen = flowclen16 * 16;
 
-	flowc = __skb_put(skb, FLOWC_LEN);
+	flowc = __skb_put(skb, flowclen);
+	memset(flowc, 0, flowclen);
 
 	flowc->op_to_nparams = cpu_to_be32(FW_WR_OP_V(FW_FLOWC_WR) |
 					   FW_FLOWC_WR_NPARAMS_V(nparams));
-	flowc->flowid_len16 = cpu_to_be32(FW_WR_LEN16_V(DIV_ROUND_UP(FLOWC_LEN,
-					  16)) | FW_WR_FLOWID_V(ep->hwtid));
+	flowc->flowid_len16 = cpu_to_be32(FW_WR_LEN16_V(flowclen16) |
+					  FW_WR_FLOWID_V(ep->hwtid));
 
 	flowc->mnemval[0].mnemonic = FW_FLOWC_MNEM_PFNVFN;
 	flowc->mnemval[0].val = cpu_to_be32(FW_PFVF_CMD_PFN_V
@@ -625,21 +630,13 @@ static int send_flowc(struct c4iw_ep *ep
 	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);
-	if (nparams == 9) {
+	flowc->mnemval[8].mnemonic = FW_FLOWC_MNEM_RCV_SCALE;
+	flowc->mnemval[8].val = cpu_to_be32(ep->snd_wscale);
+	if (nparams == 10) {
 		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;
-		flowc->mnemval[i].r4[2] = 0;
+		flowc->mnemval[9].mnemonic = FW_FLOWC_MNEM_SCHEDCLASS;
+		flowc->mnemval[9].val = cpu_to_be32(pri);
 	}
 
 	set_wr_txq(skb, CPL_PRIORITY_DATA, ep->txq_idx);
@@ -1178,6 +1175,7 @@ static int act_establish(struct c4iw_dev
 {
 	struct c4iw_ep *ep;
 	struct cpl_act_establish *req = cplhdr(skb);
+	unsigned short tcp_opt = ntohs(req->tcp_opt);
 	unsigned int tid = GET_TID(req);
 	unsigned int atid = TID_TID_G(ntohl(req->tos_atid));
 	struct tid_info *t = dev->rdev.lldi.tids;
@@ -1198,8 +1196,9 @@ static int act_establish(struct c4iw_dev
 
 	ep->snd_seq = be32_to_cpu(req->snd_isn);
 	ep->rcv_seq = be32_to_cpu(req->rcv_isn);
+	ep->snd_wscale = TCPOPT_SND_WSCALE_G(tcp_opt);
 
-	set_emss(ep, ntohs(req->tcp_opt));
+	set_emss(ep, tcp_opt);
 
 	/* dealloc the atid */
 	remove_handle(ep->com.dev, &ep->com.dev->atid_idr, atid);
@@ -2632,16 +2631,17 @@ static int pass_establish(struct c4iw_de
 	struct cpl_pass_establish *req = cplhdr(skb);
 	unsigned int tid = GET_TID(req);
 	int ret;
+	u16 tcp_opt = ntohs(req->tcp_opt);
 
 	ep = get_ep_from_tid(dev, tid);
 	pr_debug("ep %p tid %u\n", ep, ep->hwtid);
 	ep->snd_seq = be32_to_cpu(req->snd_isn);
 	ep->rcv_seq = be32_to_cpu(req->rcv_isn);
+	ep->snd_wscale = TCPOPT_SND_WSCALE_G(tcp_opt);
 
-	pr_debug("ep %p hwtid %u tcp_opt 0x%02x\n", ep, tid,
-		 ntohs(req->tcp_opt));
+	pr_debug("ep %p hwtid %u tcp_opt 0x%02x\n", ep, tid, tcp_opt);
 
-	set_emss(ep, ntohs(req->tcp_opt));
+	set_emss(ep, tcp_opt);
 
 	dst_confirm(ep->dst);
 	mutex_lock(&ep->com.mutex);
--- a/drivers/infiniband/hw/cxgb4/iw_cxgb4.h
+++ b/drivers/infiniband/hw/cxgb4/iw_cxgb4.h
@@ -911,7 +911,10 @@ enum conn_pre_alloc_buffers {
 	CN_MAX_CON_BUF
 };
 
-#define FLOWC_LEN 80
+enum {
+	FLOWC_LEN = offsetof(struct fw_flowc_wr, mnemval[FW_FLOWC_MNEM_MAX])
+};
+
 union cpl_wr_size {
 	struct cpl_abort_req abrt_req;
 	struct cpl_abort_rpl abrt_rpl;
@@ -978,6 +981,7 @@ struct c4iw_ep {
 	unsigned int retry_count;
 	int snd_win;
 	int rcv_win;
+	u32 snd_wscale;
 	struct c4iw_ep_stats stats;
 };