kmod-redhat-bnxt_en/SOURCES/0042-netdrv-bnxt_en-Refactor-TPA-logic.patch

From 746679f876254905a83b0c7ad192d4cc9c149315 Mon Sep 17 00:00:00 2001
From: Jonathan Toppins <jtoppins@redhat.com>
Date: Wed, 2 Oct 2019 18:22:57 -0400
Subject: [PATCH 42/96] [netdrv] bnxt_en: Refactor TPA logic

Message-id: <b6a806b7d4b8e826e928a731708ab23ed16a4e2d.1570027456.git.jtoppins@redhat.com>
Patchwork-id: 276460
O-Subject: [RHEL-8.2 PATCH 35/78] bnxt_en: Refactor TPA logic.
Bugzilla: 1724766
RH-Acked-by: John Linville <linville@redhat.com>
RH-Acked-by: Jarod Wilson <jarod@redhat.com>

Refactor the TPA logic slightly, so that the code can be more easily
extended to support TPA on the new 57500 chips.  In particular, the
logic to get the next aggregation completion is refactored into a
new function bnxt_get_agg() so that this operation is made more
generalized.  This operation will be different on the new chip in TPA
mode.  The logic to recycle the aggregation buffers has a new start
index parameter added for the same purpose.

Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
(cherry picked from commit 4a228a3a5e58e5c05c6ffb5b430e5cb936865a8b)
Bugzilla: 1724766
Build Info: https://brewweb.engineering.redhat.com/brew/taskinfo?taskID=23809532
Tested: build, boot, basic ping
Signed-off-by: Jonathan Toppins <jtoppins@redhat.com>
Signed-off-by: Bruno Meneguele <bmeneg@redhat.com>
---
 drivers/net/ethernet/broadcom/bnxt/bnxt.c | 117 ++++++++++++++++++------------
 1 file changed, 69 insertions(+), 48 deletions(-)

Index: src/drivers/net/ethernet/broadcom/bnxt/bnxt.c
===================================================================
--- src.orig/drivers/net/ethernet/broadcom/bnxt/bnxt.c	2020-02-06 16:23:15.092518825 +0100
+++ src/drivers/net/ethernet/broadcom/bnxt/bnxt.c	2020-02-06 16:23:15.989510591 +0100
@@ -830,8 +830,20 @@
 	return 0;
 }
 
-static void bnxt_reuse_rx_agg_bufs(struct bnxt_cp_ring_info *cpr, u16 cp_cons,
-				   u32 agg_bufs)
+static struct rx_agg_cmp *bnxt_get_agg(struct bnxt *bp,
+				       struct bnxt_cp_ring_info *cpr,
+				       u16 cp_cons, u16 curr)
+{
+	struct rx_agg_cmp *agg;
+
+	cp_cons = RING_CMP(ADV_RAW_CMP(cp_cons, curr));
+	agg = (struct rx_agg_cmp *)
+		&cpr->cp_desc_ring[CP_RING(cp_cons)][CP_IDX(cp_cons)];
+	return agg;
+}
+
+static void bnxt_reuse_rx_agg_bufs(struct bnxt_cp_ring_info *cpr, u16 idx,
+				   u16 start, u32 agg_bufs, bool tpa)
 {
 	struct bnxt_napi *bnapi = cpr->bnapi;
 	struct bnxt *bp = bnapi->bp;
@@ -847,8 +859,7 @@
 		struct rx_bd *prod_bd;
 		struct page *page;
 
-		agg = (struct rx_agg_cmp *)
-			&cpr->cp_desc_ring[CP_RING(cp_cons)][CP_IDX(cp_cons)];
+		agg = bnxt_get_agg(bp, cpr, idx, start + i);
 		cons = agg->rx_agg_cmp_opaque;
 		__clear_bit(cons, rxr->rx_agg_bmap);
 
@@ -876,7 +887,6 @@
 
 		prod = NEXT_RX_AGG(prod);
 		sw_prod = NEXT_RX_AGG(sw_prod);
-		cp_cons = NEXT_CMP(cp_cons);
 	}
 	rxr->rx_agg_prod = prod;
 	rxr->rx_sw_agg_prod = sw_prod;
@@ -959,8 +969,8 @@
 
 static struct sk_buff *bnxt_rx_pages(struct bnxt *bp,
 				     struct bnxt_cp_ring_info *cpr,
-				     struct sk_buff *skb, u16 cp_cons,
-				     u32 agg_bufs)
+				     struct sk_buff *skb, u16 idx,
+				     u32 agg_bufs, bool tpa)
 {
 	struct bnxt_napi *bnapi = cpr->bnapi;
 	struct pci_dev *pdev = bp->pdev;
@@ -975,8 +985,7 @@
 		struct page *page;
 		dma_addr_t mapping;
 
-		agg = (struct rx_agg_cmp *)
-			&cpr->cp_desc_ring[CP_RING(cp_cons)][CP_IDX(cp_cons)];
+		agg = bnxt_get_agg(bp, cpr, idx, i);
 		cons = agg->rx_agg_cmp_opaque;
 		frag_len = (le32_to_cpu(agg->rx_agg_cmp_len_flags_type) &
 			    RX_AGG_CMP_LEN) >> RX_AGG_CMP_LEN_SHIFT;
@@ -1010,7 +1019,7 @@
 			 * allocated already.
 			 */
 			rxr->rx_agg_prod = prod;
-			bnxt_reuse_rx_agg_bufs(cpr, cp_cons, agg_bufs - i);
+			bnxt_reuse_rx_agg_bufs(cpr, idx, i, agg_bufs - i, tpa);
 			return NULL;
 		}
 
@@ -1023,7 +1032,6 @@
 		skb->truesize += PAGE_SIZE;
 
 		prod = NEXT_RX_AGG(prod);
-		cp_cons = NEXT_CMP(cp_cons);
 	}
 	rxr->rx_agg_prod = prod;
 	return skb;
@@ -1083,9 +1091,7 @@
 	} else if (cmp_type == CMP_TYPE_RX_L2_TPA_END_CMP) {
 		struct rx_tpa_end_cmp *tpa_end = cmp;
 
-		agg_bufs = (le32_to_cpu(tpa_end->rx_tpa_end_cmp_misc_v1) &
-			    RX_TPA_END_CMP_AGG_BUFS) >>
-			   RX_TPA_END_CMP_AGG_BUFS_SHIFT;
+		agg_bufs = TPA_END_AGG_BUFS(tpa_end);
 	}
 
 	if (agg_bufs) {
@@ -1197,11 +1203,10 @@
 	cons_rx_buf->data = NULL;
 }
 
-static void bnxt_abort_tpa(struct bnxt_cp_ring_info *cpr, u16 cp_cons,
-			   u32 agg_bufs)
+static void bnxt_abort_tpa(struct bnxt_cp_ring_info *cpr, u16 idx, u32 agg_bufs)
 {
 	if (agg_bufs)
-		bnxt_reuse_rx_agg_bufs(cpr, cp_cons, agg_bufs);
+		bnxt_reuse_rx_agg_bufs(cpr, idx, 0, agg_bufs, true);
 }
 
 static struct sk_buff *bnxt_gro_func_5731x(struct bnxt_tpa_info *tpa_info,
@@ -1373,9 +1378,7 @@
 	skb_shinfo(skb)->gso_size =
 		le32_to_cpu(tpa_end1->rx_tpa_end_cmp_seg_len);
 	skb_shinfo(skb)->gso_type = tpa_info->gso_type;
-	payload_off = (le32_to_cpu(tpa_end->rx_tpa_end_cmp_misc_v1) &
-		       RX_TPA_END_CMP_PAYLOAD_OFFSET) >>
-		      RX_TPA_END_CMP_PAYLOAD_OFFSET_SHIFT;
+	payload_off = TPA_END_PAYLOAD_OFF(tpa_end);
 	skb = bp->gro_func(tpa_info, payload_off, TPA_END_GRO_TS(tpa_end), skb);
 	if (likely(skb))
 		tcp_gro_complete(skb);
@@ -1405,11 +1408,11 @@
 	struct bnxt_rx_ring_info *rxr = bnapi->rx_ring;
 	u8 agg_id = TPA_END_AGG_ID(tpa_end);
 	u8 *data_ptr, agg_bufs;
-	u16 cp_cons = RING_CMP(*raw_cons);
 	unsigned int len;
 	struct bnxt_tpa_info *tpa_info;
 	dma_addr_t mapping;
 	struct sk_buff *skb;
+	u16 idx = 0;
 	void *data;
 
 	if (unlikely(bnapi->in_reset)) {
@@ -1427,19 +1430,19 @@
 	len = tpa_info->len;
 	mapping = tpa_info->mapping;
 
-	agg_bufs = (le32_to_cpu(tpa_end->rx_tpa_end_cmp_misc_v1) &
-		    RX_TPA_END_CMP_AGG_BUFS) >> RX_TPA_END_CMP_AGG_BUFS_SHIFT;
+	agg_bufs = TPA_END_AGG_BUFS(tpa_end);
 
 	if (agg_bufs) {
+		idx = RING_CMP(*raw_cons);
 		if (!bnxt_agg_bufs_valid(bp, cpr, agg_bufs, raw_cons))
 			return ERR_PTR(-EBUSY);
 
 		*event |= BNXT_AGG_EVENT;
-		cp_cons = NEXT_CMP(cp_cons);
+		idx = NEXT_CMP(idx);
 	}
 
 	if (unlikely(agg_bufs > MAX_SKB_FRAGS || TPA_END_ERRORS(tpa_end1))) {
-		bnxt_abort_tpa(cpr, cp_cons, agg_bufs);
+		bnxt_abort_tpa(cpr, idx, agg_bufs);
 		if (agg_bufs > MAX_SKB_FRAGS)
 			netdev_warn(bp->dev, "TPA frags %d exceeded MAX_SKB_FRAGS %d\n",
 				    agg_bufs, (int)MAX_SKB_FRAGS);
@@ -1449,7 +1452,7 @@
 	if (len <= bp->rx_copy_thresh) {
 		skb = bnxt_copy_skb(bnapi, data_ptr, len, mapping);
 		if (!skb) {
-			bnxt_abort_tpa(cpr, cp_cons, agg_bufs);
+			bnxt_abort_tpa(cpr, idx, agg_bufs);
 			return NULL;
 		}
 	} else {
@@ -1458,7 +1461,7 @@
 
 		new_data = __bnxt_alloc_rx_data(bp, &new_mapping, GFP_ATOMIC);
 		if (!new_data) {
-			bnxt_abort_tpa(cpr, cp_cons, agg_bufs);
+			bnxt_abort_tpa(cpr, idx, agg_bufs);
 			return NULL;
 		}
 
@@ -1473,7 +1476,7 @@
 
 		if (!skb) {
 			kfree(data);
-			bnxt_abort_tpa(cpr, cp_cons, agg_bufs);
+			bnxt_abort_tpa(cpr, idx, agg_bufs);
 			return NULL;
 		}
 		skb_reserve(skb, bp->rx_offset);
@@ -1481,7 +1484,7 @@
 	}
 
 	if (agg_bufs) {
-		skb = bnxt_rx_pages(bp, cpr, skb, cp_cons, agg_bufs);
+		skb = bnxt_rx_pages(bp, cpr, skb, idx, agg_bufs, true);
 		if (!skb) {
 			/* Page reuse already handled by bnxt_rx_pages(). */
 			return NULL;
@@ -1625,7 +1628,8 @@
 
 		bnxt_reuse_rx_data(rxr, cons, data);
 		if (agg_bufs)
-			bnxt_reuse_rx_agg_bufs(cpr, cp_cons, agg_bufs);
+			bnxt_reuse_rx_agg_bufs(cpr, cp_cons, 0, agg_bufs,
+					       false);
 
 		rc = -EIO;
 		if (rx_err & RX_CMPL_ERRORS_BUFFER_ERROR_MASK) {
@@ -1648,7 +1652,8 @@
 		bnxt_reuse_rx_data(rxr, cons, data);
 		if (!skb) {
 			if (agg_bufs)
-				bnxt_reuse_rx_agg_bufs(cpr, cp_cons, agg_bufs);
+				bnxt_reuse_rx_agg_bufs(cpr, cp_cons, 0,
+						       agg_bufs, false);
 			rc = -ENOMEM;
 			goto next_rx;
 		}
@@ -1668,7 +1673,7 @@
 	}
 
 	if (agg_bufs) {
-		skb = bnxt_rx_pages(bp, cpr, skb, cp_cons, agg_bufs);
+		skb = bnxt_rx_pages(bp, cpr, skb, cp_cons, agg_bufs, false);
 		if (!skb) {
 			rc = -ENOMEM;
 			goto next_rx;
@@ -2486,6 +2491,33 @@
 	return 0;
 }
 
+static void bnxt_free_tpa_info(struct bnxt *bp)
+{
+	int i;
+
+	for (i = 0; i < bp->rx_nr_rings; i++) {
+		struct bnxt_rx_ring_info *rxr = &bp->rx_ring[i];
+
+		kfree(rxr->rx_tpa);
+		rxr->rx_tpa = NULL;
+	}
+}
+
+static int bnxt_alloc_tpa_info(struct bnxt *bp)
+{
+	int i;
+
+	for (i = 0; i < bp->rx_nr_rings; i++) {
+		struct bnxt_rx_ring_info *rxr = &bp->rx_ring[i];
+
+		rxr->rx_tpa = kcalloc(MAX_TPA, sizeof(struct bnxt_tpa_info),
+				      GFP_KERNEL);
+		if (!rxr->rx_tpa)
+			return -ENOMEM;
+	}
+	return 0;
+}
+
 static void bnxt_free_rx_rings(struct bnxt *bp)
 {
 	int i;
@@ -2493,6 +2525,7 @@
 	if (!bp->rx_ring)
 		return;
 
+	bnxt_free_tpa_info(bp);
 	for (i = 0; i < bp->rx_nr_rings; i++) {
 		struct bnxt_rx_ring_info *rxr = &bp->rx_ring[i];
 		struct bnxt_ring_struct *ring;
@@ -2506,9 +2539,6 @@
 		page_pool_destroy(rxr->page_pool);
 		rxr->page_pool = NULL;
 
-		kfree(rxr->rx_tpa);
-		rxr->rx_tpa = NULL;
-
 		kfree(rxr->rx_agg_bmap);
 		rxr->rx_agg_bmap = NULL;
 
@@ -2542,7 +2572,7 @@
 
 static int bnxt_alloc_rx_rings(struct bnxt *bp)
 {
-	int i, rc, agg_rings = 0, tpa_rings = 0;
+	int i, rc = 0, agg_rings = 0;
 
 	if (!bp->rx_ring)
 		return -ENOMEM;
@@ -2550,9 +2580,6 @@
 	if (bp->flags & BNXT_FLAG_AGG_RINGS)
 		agg_rings = 1;
 
-	if (bp->flags & BNXT_FLAG_TPA)
-		tpa_rings = 1;
-
 	for (i = 0; i < bp->rx_nr_rings; i++) {
 		struct bnxt_rx_ring_info *rxr = &bp->rx_ring[i];
 		struct bnxt_ring_struct *ring;
@@ -2594,17 +2621,11 @@
 			rxr->rx_agg_bmap = kzalloc(mem_size, GFP_KERNEL);
 			if (!rxr->rx_agg_bmap)
 				return -ENOMEM;
-
-			if (tpa_rings) {
-				rxr->rx_tpa = kcalloc(MAX_TPA,
-						sizeof(struct bnxt_tpa_info),
-						GFP_KERNEL);
-				if (!rxr->rx_tpa)
-					return -ENOMEM;
-			}
 		}
 	}
-	return 0;
+	if (bp->flags & BNXT_FLAG_TPA)
+		rc = bnxt_alloc_tpa_info(bp);
+	return rc;
 }
 
 static void bnxt_free_tx_rings(struct bnxt *bp)