Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
From:   Ivan Khoronzhuk <ivan.khoronzhuk@linaro.org>
To:     grygorii.strashko@ti.com, davem@davemloft.net
Cc:     corbet@lwn.net, akpm@linux-foundation.org, netdev@vger.kernel.org,
        linux-doc@vger.kernel.org, linux-kernel@vger.kernel.org,
        linux-omap@vger.kernel.org, vinicius.gomes@intel.com,
        henrik@austad.us, jesus.sanchez-palencia@intel.com,
        Ivan Khoronzhuk <ivan.khoronzhuk@linaro.org>
Subject: [RFC PATCH 1/6] net: ethernet: ti: cpsw: use cpdma channels in backward order for txq
Date:   Sat, 19 May 2018 00:15:05 +0300
Message-Id: <20180518211510.13341-2-ivan.khoronzhuk@linaro.org>
In-Reply-To: <20180518211510.13341-1-ivan.khoronzhuk@linaro.org>
References: <20180518211510.13341-1-ivan.khoronzhuk@linaro.org>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

The cpdma channel highest priority is from hi to lo number.
The driver has limited number of descriptors that are shared between
number of cpdma channels. Number of queues can be tuned with ethtool,
that allows to not spend descriptors on not needed cpdma channels.
In AVB usually only 2 tx queues can be enough with rate limitation.
The rate limitation can be used only for hi priority queues. Thus, to
use only 2 queues the 8 has to be created. It's wasteful.

So, in order to allow using only needed number of rate limited
tx queues, save resources, and be able to set rate limitation for
them, let assign tx cpdma channels in backward order to queues.

Signed-off-by: Ivan Khoronzhuk <ivan.khoronzhuk@linaro.org>
---
 drivers/net/ethernet/ti/cpsw.c | 14 ++++++++------
 1 file changed, 8 insertions(+), 6 deletions(-)

diff --git a/drivers/net/ethernet/ti/cpsw.c b/drivers/net/ethernet/ti/cpsw.c
index a7285dddfd29..9bd615da04d3 100644
--- a/drivers/net/ethernet/ti/cpsw.c
+++ b/drivers/net/ethernet/ti/cpsw.c
@@ -967,8 +967,8 @@ static int cpsw_tx_mq_poll(struct napi_struct *napi_tx, int budget)
 
 	/* process every unprocessed channel */
 	ch_map = cpdma_ctrl_txchs_state(cpsw->dma);
-	for (ch = 0, num_tx = 0; ch_map; ch_map >>= 1, ch++) {
-		if (!(ch_map & 0x01))
+	for (ch = 0, num_tx = 0; ch_map & 0xff; ch_map <<= 1, ch++) {
+		if (!(ch_map & 0x80))
 			continue;
 
 		txv = &cpsw->txv[ch];
@@ -2431,7 +2431,7 @@ static int cpsw_update_channels_res(struct cpsw_priv *priv, int ch_num, int rx)
 	void (*handler)(void *, int, int);
 	struct netdev_queue *queue;
 	struct cpsw_vector *vec;
-	int ret, *ch;
+	int ret, *ch, vch;
 
 	if (rx) {
 		ch = &cpsw->rx_ch_num;
@@ -2444,7 +2444,8 @@ static int cpsw_update_channels_res(struct cpsw_priv *priv, int ch_num, int rx)
 	}
 
 	while (*ch < ch_num) {
-		vec[*ch].ch = cpdma_chan_create(cpsw->dma, *ch, handler, rx);
+		vch = rx ? *ch : 7 - *ch;
+		vec[*ch].ch = cpdma_chan_create(cpsw->dma, vch, handler, rx);
 		queue = netdev_get_tx_queue(priv->ndev, *ch);
 		queue->tx_maxrate = 0;
 
@@ -2980,7 +2981,7 @@ static int cpsw_probe(struct platform_device *pdev)
 	u32 slave_offset, sliver_offset, slave_size;
 	const struct soc_device_attribute *soc;
 	struct cpsw_common		*cpsw;
-	int ret = 0, i;
+	int ret = 0, i, ch;
 	int irq;
 
 	cpsw = devm_kzalloc(&pdev->dev, sizeof(struct cpsw_common), GFP_KERNEL);
@@ -3155,7 +3156,8 @@ static int cpsw_probe(struct platform_device *pdev)
 	if (soc)
 		cpsw->quirk_irq = 1;
 
-	cpsw->txv[0].ch = cpdma_chan_create(cpsw->dma, 0, cpsw_tx_handler, 0);
+	ch = cpsw->quirk_irq ? 0 : 7;
+	cpsw->txv[0].ch = cpdma_chan_create(cpsw->dma, ch, cpsw_tx_handler, 0);
 	if (IS_ERR(cpsw->txv[0].ch)) {
 		dev_err(priv->dev, "error initializing tx dma channel\n");
 		ret = PTR_ERR(cpsw->txv[0].ch);
-- 
2.17.0