Skip to contentSkip to conversation

net: ks8851: Unify KS8851 SPI and MLL drivers

PATCH Superseded Marek Vasut 3 participants 31 messages Mar 28, 2020 – Apr 16, 2020

Subsystems (2)

v3 of 7 Superseded Latest is v7 →

lore
Share & tools
lore
#0 lore raw

v3 cover net: ks8851: Unify KS8851 SPI and MLL drivers

The KS8851SNL/SNLI and KS8851-16MLL/MLLI/MLLU are very much the same pieces of silicon, except the former has an SPI interface, while the later has a parallel bus interface. Thus far, Linux has two separate drivers for each and they are diverging considerably.

This series unifies them into a single driver with small SPI and parallel bus specific parts. The approach here is to first separate out the SPI specific parts into a separate file, then add parallel bus accessors in another separate file and then finally remove the old parallel bus driver. The reason for replacing the old parallel bus driver is because the SPI bus driver is much higher quality.

NOTE: This series depends on "net: ks8851-ml: Fix IO operations, again"

NOTE: The performance regression on KS8851-16MLL is now fixes, the TX
throughput is back to ~75 Mbit/s , RX is still 50 Mbit/s .

Marek Vasut (18):
net: ks8851: Factor out spi->dev in probe()/remove()
net: ks8851: Rename ndev to netdev in probe
net: ks8851: Replace dev_err() with netdev_err() in IRQ handler
net: ks8851: Pass device node into ks8851_init_mac()
net: ks8851: Use devm_alloc_etherdev()
net: ks8851: Use dev_{get,set}_drvdata()
net: ks8851: Remove ks8851_rdreg32()
net: ks8851: Use 16-bit writes to program MAC address
net: ks8851: Use 16-bit read of RXFC register
net: ks8851: Factor out bus lock handling
net: ks8851: Factor out SKB receive function
net: ks8851: Split out SPI specific entries in struct ks8851_net
net: ks8851: Split out SPI specific code from probe() and remove()
net: ks8851: Factor out TX work flush function
net: ks8851: Permit overridding interrupt enable register
net: ks8851: Separate SPI operations into separate file
net: ks8851: Implement Parallel bus operations
net: ks8851: Remove ks8851_mll.c

 drivers/net/ethernet/micrel/Kconfig           |    2 +
 drivers/net/ethernet/micrel/Makefile          |    2 +
 drivers/net/ethernet/micrel/ks8851.h          |  142 +-
 .../micrel/{ks8851.c => ks8851_common.c}      |  688 ++------
 drivers/net/ethernet/micrel/ks8851_mll.c      | 1393 -----------------
 drivers/net/ethernet/micrel/ks8851_par.c      |  337 ++++
 drivers/net/ethernet/micrel/ks8851_spi.c      |  448 ++++++
 7 files changed, 1027 insertions(+), 1985 deletions(-)
 rename drivers/net/ethernet/micrel/{ks8851.c => ks8851_common.c} (60%)
 delete mode 100644 drivers/net/ethernet/micrel/ks8851_mll.c
 create mode 100644 drivers/net/ethernet/micrel/ks8851_par.c
 create mode 100644 drivers/net/ethernet/micrel/ks8851_spi.c
Cc David S. Miller <[email protected]>
Cc Lukas Wunner <[email protected]>
Cc Petr Stetiar <[email protected]>
Cc YueHaibing <[email protected]>
Signature · 1 line
2.25.1
#1 lore raw

PATCH v3 1/18 net: ks8851: Factor out spi->dev in probe()/remove()

Pull out the spi->dev into one common place in the function instead of having it repeated over and over again. This is done in preparation for unifying ks8851 and ks8851-mll drivers. No functional change.

Cc David S. Miller <[email protected]>
Cc Lukas Wunner <[email protected]>
Cc Petr Stetiar <[email protected]>
Cc YueHaibing <[email protected]>

---
V2: Unwrap function calls that are shorter than 80 chars
V3: No change

---
 drivers/net/ethernet/micrel/ks8851.c | 29 ++++++++++++++--------------
 1 file changed, 14 insertions(+), 15 deletions(-)
drivers/net/ethernet/micrel/ks8851.c +14 -15
index 33305c9c5a62..e32ef9403803 100644
@@ -1413,6 +1413,7 @@ static SIMPLE_DEV_PM_OPS(ks8851_pm_ops, ks8851_suspend, ks8851_resume);
 
 static int ks8851_probe(struct spi_device *spi)
 {
+	struct device *dev = &spi->dev;
 	struct net_device *ndev;
 	struct ks8851_net *ks;
 	int ret;
@@ -1431,8 +1432,7 @@ static int ks8851_probe(struct spi_device *spi)
 	ks->spidev = spi;
 	ks->tx_space = 6144;
 
-	gpio = of_get_named_gpio_flags(spi->dev.of_node, "reset-gpios",
-				       0, NULL);
+	gpio = of_get_named_gpio_flags(dev->of_node, "reset-gpios", 0, NULL);
 	if (gpio == -EPROBE_DEFER) {
 		ret = gpio;
 		goto err_gpio;
@@ -1440,15 +1440,15 @@ static int ks8851_probe(struct spi_device *spi)
 
 	ks->gpio = gpio;
 	if (gpio_is_valid(gpio)) {
-		ret = devm_gpio_request_one(&spi->dev, gpio,
+		ret = devm_gpio_request_one(dev, gpio,
 					    GPIOF_OUT_INIT_LOW, "ks8851_rst_n");
 		if (ret) {
-			dev_err(&spi->dev, "reset gpio request failed\n");
+			dev_err(dev, "reset gpio request failed\n");
 			goto err_gpio;
 		}
 	}
 
-	ks->vdd_io = devm_regulator_get(&spi->dev, "vdd-io");
+	ks->vdd_io = devm_regulator_get(dev, "vdd-io");
 	if (IS_ERR(ks->vdd_io)) {
 		ret = PTR_ERR(ks->vdd_io);
 		goto err_reg_io;
@@ -1456,12 +1456,11 @@ static int ks8851_probe(struct spi_device *spi)
 
 	ret = regulator_enable(ks->vdd_io);
 	if (ret) {
-		dev_err(&spi->dev, "regulator vdd_io enable fail: %d\n",
-			ret);
+		dev_err(dev, "regulator vdd_io enable fail: %d\n", ret);
 		goto err_reg_io;
 	}
 
-	ks->vdd_reg = devm_regulator_get(&spi->dev, "vdd");
+	ks->vdd_reg = devm_regulator_get(dev, "vdd");
 	if (IS_ERR(ks->vdd_reg)) {
 		ret = PTR_ERR(ks->vdd_reg);
 		goto err_reg;
@@ -1469,8 +1468,7 @@ static int ks8851_probe(struct spi_device *spi)
 
 	ret = regulator_enable(ks->vdd_reg);
 	if (ret) {
-		dev_err(&spi->dev, "regulator vdd enable fail: %d\n",
-			ret);
+		dev_err(dev, "regulator vdd enable fail: %d\n", ret);
 		goto err_reg;
 	}
 
@@ -1509,7 +1507,7 @@ static int ks8851_probe(struct spi_device *spi)
 	ks->mii.mdio_read	= ks8851_phy_read;
 	ks->mii.mdio_write	= ks8851_phy_write;
 
-	dev_info(&spi->dev, "message enable is %d\n", msg_enable);
+	dev_info(dev, "message enable is %d\n", msg_enable);
 
 	/* set the default message enable */
 	ks->msg_enable = netif_msg_init(msg_enable, (NETIF_MSG_DRV |
@@ -1519,7 +1517,7 @@ static int ks8851_probe(struct spi_device *spi)
 	skb_queue_head_init(&ks->txq);
 
 	ndev->ethtool_ops = &ks8851_ethtool_ops;
-	SET_NETDEV_DEV(ndev, &spi->dev);
+	SET_NETDEV_DEV(ndev, dev);
 
 	spi_set_drvdata(spi, ks);
 
@@ -1534,7 +1532,7 @@ static int ks8851_probe(struct spi_device *spi)
 	/* simple check for a valid chip being connected to the bus */
 	cider = ks8851_rdreg16(ks, KS_CIDER);
 	if ((cider & ~CIDER_REV_MASK) != CIDER_ID) {
-		dev_err(&spi->dev, "failed to read device ID\n");
+		dev_err(dev, "failed to read device ID\n");
 		ret = -ENODEV;
 		goto err_id;
 	}
@@ -1547,7 +1545,7 @@ static int ks8851_probe(struct spi_device *spi)
 
 	ret = register_netdev(ndev);
 	if (ret) {
-		dev_err(&spi->dev, "failed to register network device\n");
+		dev_err(dev, "failed to register network device\n");
 		goto err_netdev;
 	}
 
@@ -1573,9 +1571,10 @@ static int ks8851_probe(struct spi_device *spi)
 static int ks8851_remove(struct spi_device *spi)
 {
 	struct ks8851_net *priv = spi_get_drvdata(spi);
+	struct device *dev = &spi->dev;
 
 	if (netif_msg_drv(priv))
-		dev_info(&spi->dev, "remove\n");
+		dev_info(dev, "remove\n");
 
 	unregister_netdev(priv->netdev);
 	if (gpio_is_valid(priv->gpio))
Signature · 1 line
2.25.1
#2 lore raw

PATCH v3 2/18 net: ks8851: Rename ndev to netdev in probe

In reply to Marek Vasut

Rename ndev variable to netdev for the sake of consistency.

No functional change.

Cc David S. Miller <[email protected]>
Cc Lukas Wunner <[email protected]>
Cc Petr Stetiar <[email protected]>
Cc YueHaibing <[email protected]>

---
V2: New patch
V3: No change

---
 drivers/net/ethernet/micrel/ks8851.c | 30 ++++++++++++++--------------
 1 file changed, 15 insertions(+), 15 deletions(-)
drivers/net/ethernet/micrel/ks8851.c +15 -15
index e32ef9403803..2b85072993c5 100644
@@ -1414,21 +1414,21 @@ static SIMPLE_DEV_PM_OPS(ks8851_pm_ops, ks8851_suspend, ks8851_resume);
 static int ks8851_probe(struct spi_device *spi)
 {
 	struct device *dev = &spi->dev;
-	struct net_device *ndev;
+	struct net_device *netdev;
 	struct ks8851_net *ks;
 	int ret;
 	unsigned cider;
 	int gpio;
 
-	ndev = alloc_etherdev(sizeof(struct ks8851_net));
-	if (!ndev)
+	netdev = alloc_etherdev(sizeof(struct ks8851_net));
+	if (!netdev)
 		return -ENOMEM;
 
 	spi->bits_per_word = 8;
 
-	ks = netdev_priv(ndev);
+	ks = netdev_priv(netdev);
 
-	ks->netdev = ndev;
+	ks->netdev = netdev;
 	ks->spidev = spi;
 	ks->tx_space = 6144;
 
@@ -1500,7 +1500,7 @@ static int ks8851_probe(struct spi_device *spi)
 	ks->eeprom.register_write = ks8851_eeprom_regwrite;
 
 	/* setup mii state */
-	ks->mii.dev		= ndev;
+	ks->mii.dev		= netdev;
 	ks->mii.phy_id		= 1,
 	ks->mii.phy_id_mask	= 1;
 	ks->mii.reg_num_mask	= 0xf;
@@ -1516,15 +1516,15 @@ static int ks8851_probe(struct spi_device *spi)
 
 	skb_queue_head_init(&ks->txq);
 
-	ndev->ethtool_ops = &ks8851_ethtool_ops;
-	SET_NETDEV_DEV(ndev, dev);
+	netdev->ethtool_ops = &ks8851_ethtool_ops;
+	SET_NETDEV_DEV(netdev, dev);
 
 	spi_set_drvdata(spi, ks);
 
 	netif_carrier_off(ks->netdev);
-	ndev->if_port = IF_PORT_100BASET;
-	ndev->netdev_ops = &ks8851_netdev_ops;
-	ndev->irq = spi->irq;
+	netdev->if_port = IF_PORT_100BASET;
+	netdev->netdev_ops = &ks8851_netdev_ops;
+	netdev->irq = spi->irq;
 
 	/* issue a global soft reset to reset the device. */
 	ks8851_soft_reset(ks, GRR_GSR);
@@ -1543,14 +1543,14 @@ static int ks8851_probe(struct spi_device *spi)
 	ks8851_read_selftest(ks);
 	ks8851_init_mac(ks);
 
-	ret = register_netdev(ndev);
+	ret = register_netdev(netdev);
 	if (ret) {
 		dev_err(dev, "failed to register network device\n");
 		goto err_netdev;
 	}
 
-	netdev_info(ndev, "revision %d, MAC %pM, IRQ %d, %s EEPROM\n",
-		    CIDER_REV_GET(cider), ndev->dev_addr, ndev->irq,
+	netdev_info(netdev, "revision %d, MAC %pM, IRQ %d, %s EEPROM\n",
+		    CIDER_REV_GET(cider), netdev->dev_addr, netdev->irq,
 		    ks->rc_ccr & CCR_EEPROM ? "has" : "no");
 
 	return 0;
@@ -1564,7 +1564,7 @@ static int ks8851_probe(struct spi_device *spi)
 	regulator_disable(ks->vdd_io);
 err_reg_io:
 err_gpio:
-	free_netdev(ndev);
+	free_netdev(netdev);
 	return ret;
 }
 
Signature · 1 line
2.25.1
#3 lore raw

PATCH v3 3/18 net: ks8851: Replace dev_err() with netdev_err() in IRQ handler

In reply to Marek Vasut

Use netdev_err() instead of dev_err() to avoid accessing the spidev->dev in the interrupt handler. This is the only place which uses the spidev in this function, so replace it with netdev_err() to get rid of it. This is done in preparation for unifying the KS8851 SPI and parallel drivers.

Cc David S. Miller <[email protected]>
Cc Lukas Wunner <[email protected]>
Cc Petr Stetiar <[email protected]>
Cc YueHaibing <[email protected]>

---
V2: Add RB from Andrew
V3: No change

---
 drivers/net/ethernet/micrel/ks8851.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/net/ethernet/micrel/ks8851.c +1 -1
index 2b85072993c5..0088df970ad6 100644
@@ -631,7 +631,7 @@ static irqreturn_t ks8851_irq(int irq, void *_ks)
 		handled |= IRQ_RXI;
 
 	if (status & IRQ_SPIBEI) {
-		dev_err(&ks->spidev->dev, "%s: spi bus error\n", __func__);
+		netdev_err(ks->netdev, "%s: spi bus error\n", __func__);
 		handled |= IRQ_SPIBEI;
 	}
 
Signature · 1 line
2.25.1
#4 lore raw

PATCH v3 4/18 net: ks8851: Pass device node into ks8851_init_mac()

In reply to Marek Vasut

Since the driver probe function already has a struct device *dev pointer and can easily derive of_node pointer from it, pass the of_node pointer as a parameter to ks8851_init_mac() to avoid fishing it out from ks->spidev. This is the only reference to spidev in the function, so get rid of it. This is done in preparation for unifying the KS8851 SPI and parallel bus drivers.

No functional change.

Cc David S. Miller <[email protected]>
Cc Lukas Wunner <[email protected]>
Cc Petr Stetiar <[email protected]>
Cc YueHaibing <[email protected]>

---
V2: Pass in of_node instead of the entire device pointer
V3: No change

---
 drivers/net/ethernet/micrel/ks8851.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)
drivers/net/ethernet/micrel/ks8851.c +4 -3
index 0088df970ad6..942e694c750a 100644
@@ -409,6 +409,7 @@ static void ks8851_read_mac_addr(struct net_device *dev)
 /**
  * ks8851_init_mac - initialise the mac address
  * @ks: The device structure
+ * @ddev: The device structure pointer
  *
  * Get or create the initial mac address for the device and then set that
  * into the station address register. A mac address supplied in the device
@@ -416,12 +417,12 @@ static void ks8851_read_mac_addr(struct net_device *dev)
  * we try that. If no valid mac address is found we use eth_random_addr()
  * to create a new one.
  */
-static void ks8851_init_mac(struct ks8851_net *ks)
+static void ks8851_init_mac(struct ks8851_net *ks, struct device_node *np)
 {
 	struct net_device *dev = ks->netdev;
 	const u8 *mac_addr;
 
-	mac_addr = of_get_mac_address(ks->spidev->dev.of_node);
+	mac_addr = of_get_mac_address(np);
 	if (!IS_ERR(mac_addr)) {
 		ether_addr_copy(dev->dev_addr, mac_addr);
 		ks8851_write_mac_addr(dev);
@@ -1541,7 +1542,7 @@ static int ks8851_probe(struct spi_device *spi)
 	ks->rc_ccr = ks8851_rdreg16(ks, KS_CCR);
 
 	ks8851_read_selftest(ks);
-	ks8851_init_mac(ks);
+	ks8851_init_mac(ks, dev->of_node);
 
 	ret = register_netdev(netdev);
 	if (ret) {
Signature · 1 line
2.25.1
#5 lore raw

PATCH v3 5/18 net: ks8851: Use devm_alloc_etherdev()

In reply to Marek Vasut

Use device managed version of alloc_etherdev() to simplify the code. No functional change intended.

Cc David S. Miller <[email protected]>
Cc Lukas Wunner <[email protected]>
Cc Petr Stetiar <[email protected]>
Cc YueHaibing <[email protected]>

---
V2: Add RB from Andrew
V3: No change

---
 drivers/net/ethernet/micrel/ks8851.c | 13 ++++---------
 1 file changed, 4 insertions(+), 9 deletions(-)
drivers/net/ethernet/micrel/ks8851.c +4 -9
index 942e694c750a..f0b70b79e7ed 100644
@@ -1421,7 +1421,7 @@ static int ks8851_probe(struct spi_device *spi)
 	unsigned cider;
 	int gpio;
 
-	netdev = alloc_etherdev(sizeof(struct ks8851_net));
+	netdev = devm_alloc_etherdev(dev, sizeof(struct ks8851_net));
 	if (!netdev)
 		return -ENOMEM;
 
@@ -1434,10 +1434,8 @@ static int ks8851_probe(struct spi_device *spi)
 	ks->tx_space = 6144;
 
 	gpio = of_get_named_gpio_flags(dev->of_node, "reset-gpios", 0, NULL);
-	if (gpio == -EPROBE_DEFER) {
-		ret = gpio;
-		goto err_gpio;
-	}
+	if (gpio == -EPROBE_DEFER)
+		return gpio;
 
 	ks->gpio = gpio;
 	if (gpio_is_valid(gpio)) {
@@ -1445,7 +1443,7 @@ static int ks8851_probe(struct spi_device *spi)
 					    GPIOF_OUT_INIT_LOW, "ks8851_rst_n");
 		if (ret) {
 			dev_err(dev, "reset gpio request failed\n");
-			goto err_gpio;
+			return ret;
 		}
 	}
 
@@ -1564,8 +1562,6 @@ static int ks8851_probe(struct spi_device *spi)
 err_reg:
 	regulator_disable(ks->vdd_io);
 err_reg_io:
-err_gpio:
-	free_netdev(netdev);
 	return ret;
 }
 
@@ -1582,7 +1578,6 @@ static int ks8851_remove(struct spi_device *spi)
 		gpio_set_value(priv->gpio, 0);
 	regulator_disable(priv->vdd_reg);
 	regulator_disable(priv->vdd_io);
-	free_netdev(priv->netdev);
 
 	return 0;
 }
Signature · 1 line
2.25.1
#6 lore raw

PATCH v3 6/18 net: ks8851: Use dev_{get,set}_drvdata()

In reply to Marek Vasut

Replace spi_{get,set}_drvdata() with dev_{get,set}_drvdata(), which works for both SPI and platform drivers. This is done in preparation for unifying the KS8851 SPI and parallel bus drivers.

There should be no functional change.

Cc David S. Miller <[email protected]>
Cc Lukas Wunner <[email protected]>
Cc Petr Stetiar <[email protected]>
Cc YueHaibing <[email protected]>

---
V2: Reverse xmas tree.
V3: No change

---
 drivers/net/ethernet/micrel/ks8851.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
drivers/net/ethernet/micrel/ks8851.c +4 -2
index f0b70b79e7ed..c2f381a7b3f3 100644
@@ -1518,7 +1518,7 @@ static int ks8851_probe(struct spi_device *spi)
 	netdev->ethtool_ops = &ks8851_ethtool_ops;
 	SET_NETDEV_DEV(netdev, dev);
 
-	spi_set_drvdata(spi, ks);
+	dev_set_drvdata(dev, ks);
 
 	netif_carrier_off(ks->netdev);
 	netdev->if_port = IF_PORT_100BASET;
@@ -1567,8 +1567,10 @@ static int ks8851_probe(struct spi_device *spi)
 
 static int ks8851_remove(struct spi_device *spi)
 {
-	struct ks8851_net *priv = spi_get_drvdata(spi);
 	struct device *dev = &spi->dev;
+	struct ks8851_net *priv;
+
+	priv = dev_get_drvdata(dev);
 
 	if (netif_msg_drv(priv))
 		dev_info(dev, "remove\n");
Signature · 1 line
2.25.1
#7 lore raw

PATCH v3 7/18 net: ks8851: Remove ks8851_rdreg32()

In reply to Marek Vasut

The ks8851_rdreg32() is used only in one place, to read two registers using a single read. To make it easier to support 16-bit accesses via parallel bus later on, replace this single read with two 16-bit reads from each of the registers and drop the ks8851_rdreg32() altogether.

If this has noticeable performance impact on the SPI variant of KS8851, then we should consider using regmap to abstract the SPI and parallel bus options and in case of SPI, permit regmap to merge register reads of neighboring registers into single, longer, read.

Cc David S. Miller <[email protected]>
Cc Lukas Wunner <[email protected]>
Cc Petr Stetiar <[email protected]>
Cc YueHaibing <[email protected]>

---
NOTE: This might need to be reinstated if it limits performance
of the SPI part
V2: No change
V3: No change

---
 drivers/net/ethernet/micrel/ks8851.c | 25 ++-----------------------
 1 file changed, 2 insertions(+), 23 deletions(-)
drivers/net/ethernet/micrel/ks8851.c +2 -23
index c2f381a7b3f3..8f1cc05dc3c8 100644
@@ -296,25 +296,6 @@ static unsigned ks8851_rdreg16(struct ks8851_net *ks, unsigned reg)
 	return le16_to_cpu(rx);
 }
 
-/**
- * ks8851_rdreg32 - read 32 bit register from device
- * @ks: The chip information
- * @reg: The register address
- *
- * Read a 32bit register from the chip.
- *
- * Note, this read requires the address be aligned to 4 bytes.
-*/
-static unsigned ks8851_rdreg32(struct ks8851_net *ks, unsigned reg)
-{
-	__le32 rx = 0;
-
-	WARN_ON(reg & 3);
-
-	ks8851_rdreg(ks, MK_OP(0xf, reg), (u8 *)&rx, 4);
-	return le32_to_cpu(rx);
-}
-
 /**
  * ks8851_soft_reset - issue one of the soft reset to the device
  * @ks: The device state.
@@ -508,7 +489,6 @@ static void ks8851_rx_pkts(struct ks8851_net *ks)
 	unsigned rxfc;
 	unsigned rxlen;
 	unsigned rxstat;
-	u32 rxh;
 	u8 *rxpkt;
 
 	rxfc = ks8851_rdreg8(ks, KS_RXFC);
@@ -527,9 +507,8 @@ static void ks8851_rx_pkts(struct ks8851_net *ks)
 	 */
 
 	for (; rxfc != 0; rxfc--) {
-		rxh = ks8851_rdreg32(ks, KS_RXFHSR);
-		rxstat = rxh & 0xffff;
-		rxlen = (rxh >> 16) & 0xfff;
+		rxstat = ks8851_rdreg16(ks, KS_RXFHSR);
+		rxlen = ks8851_rdreg16(ks, KS_RXFHBCR) & RXFHBCR_CNT_MASK;
 
 		netif_dbg(ks, rx_status, ks->netdev,
 			  "rx: stat 0x%04x, len 0x%04x\n", rxstat, rxlen);
Signature · 1 line
2.25.1
#8 lore raw

PATCH v3 8/18 net: ks8851: Use 16-bit writes to program MAC address

In reply to Marek Vasut

On the SPI variant of KS8851, the MAC address can be programmed with either 8/16/32-bit writes. To make it easier to support the 16-bit parallel option of KS8851 too, switch both the MAC address programming and readout to 16-bit operations.

Remove ks8851_wrreg8() as it is not used anywhere anymore.

There should be no functional change.

Cc David S. Miller <[email protected]>
Cc Lukas Wunner <[email protected]>
Cc Petr Stetiar <[email protected]>
Cc YueHaibing <[email protected]>

---
V2: Get rid of the KS_MAR(i + 1) by adjusting KS_MAR(x) macro
V3: Swap MARx MSB and LSB in ks8851_read_mac_addr()

---
 drivers/net/ethernet/micrel/ks8851.c | 47 ++++++++--------------------
 drivers/net/ethernet/micrel/ks8851.h |  2 +-
 2 files changed, 14 insertions(+), 35 deletions(-)
drivers/net/ethernet/micrel/ks8851.c +13 -34
index 8f1cc05dc3c8..bb2f0dc8214d 100644
@@ -185,36 +185,6 @@ static void ks8851_wrreg16(struct ks8851_net *ks, unsigned reg, unsigned val)
 		netdev_err(ks->netdev, "spi_sync() failed\n");
 }
 
-/**
- * ks8851_wrreg8 - write 8bit register value to chip
- * @ks: The chip state
- * @reg: The register address
- * @val: The value to write
- *
- * Issue a write to put the value @val into the register specified in @reg.
- */
-static void ks8851_wrreg8(struct ks8851_net *ks, unsigned reg, unsigned val)
-{
-	struct spi_transfer *xfer = &ks->spi_xfer1;
-	struct spi_message *msg = &ks->spi_msg1;
-	__le16 txb[2];
-	int ret;
-	int bit;
-
-	bit = 1 << (reg & 3);
-
-	txb[0] = cpu_to_le16(MK_OP(bit, reg) | KS_SPIOP_WR);
-	txb[1] = val;
-
-	xfer->tx_buf = txb;
-	xfer->rx_buf = NULL;
-	xfer->len = 3;
-
-	ret = spi_sync(ks->spidev, msg);
-	if (ret < 0)
-		netdev_err(ks->netdev, "spi_sync() failed\n");
-}
-
 /**
  * ks8851_rdreg - issue read register command and return the data
  * @ks: The device state
@@ -349,6 +319,7 @@ static void ks8851_set_powermode(struct ks8851_net *ks, unsigned pwrmode)
 static int ks8851_write_mac_addr(struct net_device *dev)
 {
 	struct ks8851_net *ks = netdev_priv(dev);
+	u16 val;
 	int i;
 
 	mutex_lock(&ks->lock);
@@ -358,8 +329,12 @@ static int ks8851_write_mac_addr(struct net_device *dev)
 	 * the first write to the MAC address does not take effect.
 	 */
 	ks8851_set_powermode(ks, PMECR_PM_NORMAL);
-	for (i = 0; i < ETH_ALEN; i++)
-		ks8851_wrreg8(ks, KS_MAR(i), dev->dev_addr[i]);
+
+	for (i = 0; i < ETH_ALEN; i += 2) {
+		val = (dev->dev_addr[i] << 8) | dev->dev_addr[i + 1];
+		ks8851_wrreg16(ks, KS_MAR(i), val);
+	}
+
 	if (!netif_running(dev))
 		ks8851_set_powermode(ks, PMECR_PM_SOFTDOWN);
 
@@ -377,12 +352,16 @@ static int ks8851_write_mac_addr(struct net_device *dev)
 static void ks8851_read_mac_addr(struct net_device *dev)
 {
 	struct ks8851_net *ks = netdev_priv(dev);
+	u16 reg;
 	int i;
 
 	mutex_lock(&ks->lock);
 
-	for (i = 0; i < ETH_ALEN; i++)
-		dev->dev_addr[i] = ks8851_rdreg8(ks, KS_MAR(i));
+	for (i = 0; i < ETH_ALEN; i += 2) {
+		reg = ks8851_rdreg16(ks, KS_MAR(i));
+		dev->dev_addr[i] = reg >> 8;
+		dev->dev_addr[i + 1] = reg & 0xff;
+	}
 
 	mutex_unlock(&ks->lock);
 }
drivers/net/ethernet/micrel/ks8851.h +1 -1
index 8f834aef8e32..f210d18a10b5 100644
@@ -19,7 +19,7 @@
 #define CCR_32PIN				(1 << 0)    /* KSZ8851SNL    */
 
 /* MAC address registers */
-#define KS_MAR(_m)				(0x15 - (_m))
+#define KS_MAR(_m)				(0x14 - (_m))
 #define KS_MARL					0x10
 #define KS_MARM					0x12
 #define KS_MARH					0x14
Signature · 1 line
2.25.1
#9 lore raw

PATCH v3 9/18 net: ks8851: Use 16-bit read of RXFC register

In reply to Marek Vasut

The RXFC register is the only one being read using 8-bit accessors. To make it easier to support the 16-bit accesses used by the parallel bus variant of KS8851, use 16-bit accessor to read RXFC register as well as neighboring RXFCTR register.

Remove ks8851_rdreg8() as it is not used anywhere anymore.

There should be no functional change.

Cc David S. Miller <[email protected]>
Cc Lukas Wunner <[email protected]>
Cc Petr Stetiar <[email protected]>
Cc YueHaibing <[email protected]>

---
NOTE: This might need to be revisited if it limits performance
of the SPI part
V2: No change
V3: No change

---
 drivers/net/ethernet/micrel/ks8851.c | 17 +----------------
 1 file changed, 1 insertion(+), 16 deletions(-)
drivers/net/ethernet/micrel/ks8851.c +1 -16
index bb2f0dc8214d..b9820068c839 100644
@@ -236,21 +236,6 @@ static void ks8851_rdreg(struct ks8851_net *ks, unsigned op,
 		memcpy(rxb, trx + 2, rxl);
 }
 
-/**
- * ks8851_rdreg8 - read 8 bit register from device
- * @ks: The chip information
- * @reg: The register address
- *
- * Read a 8bit register from the chip, returning the result
-*/
-static unsigned ks8851_rdreg8(struct ks8851_net *ks, unsigned reg)
-{
-	u8 rxb[1];
-
-	ks8851_rdreg(ks, MK_OP(1 << (reg & 3), reg), rxb, 1);
-	return rxb[0];
-}
-
 /**
  * ks8851_rdreg16 - read 16 bit register from device
  * @ks: The chip information
@@ -470,7 +455,7 @@ static void ks8851_rx_pkts(struct ks8851_net *ks)
 	unsigned rxstat;
 	u8 *rxpkt;
 
-	rxfc = ks8851_rdreg8(ks, KS_RXFC);
+	rxfc = (ks8851_rdreg16(ks, KS_RXFCTR) >> 8) & 0xff;
 
 	netif_dbg(ks, rx_status, ks->netdev,
 		  "%s: %d packets\n", __func__, rxfc);
Signature · 1 line
2.25.1
#10 lore raw

PATCH v3 10/18 net: ks8851: Factor out bus lock handling

In reply to Marek Vasut

Pull out bus access locking code into separate functions, this is done in preparation for unifying the driver with the parallel bus one. The parallel bus driver does not need heavy mutex locking of the bus and works better with spinlocks, hence prepare these locking functions to be overridden then.

Cc David S. Miller <[email protected]>
Cc Lukas Wunner <[email protected]>
Cc Petr Stetiar <[email protected]>
Cc YueHaibing <[email protected]>

---
V3: New patch

---
 drivers/net/ethernet/micrel/ks8851.c | 97 +++++++++++++++++++---------
 1 file changed, 68 insertions(+), 29 deletions(-)
drivers/net/ethernet/micrel/ks8851.c +68 -29
index b9820068c839..7a91d30472f2 100644
@@ -151,6 +151,30 @@ static int msg_enable;
 /* turn register number and byte-enable mask into data for start of packet */
 #define MK_OP(_byteen, _reg) (BYTE_EN(_byteen) | (_reg)  << (8+2) | (_reg) >> 6)
 
+/**
+ * ks8851_lock - register access lock
+ * @ks: The chip state
+ * @flags: Spinlock flags
+ *
+ * Claim chip register access lock
+ */
+static void ks8851_lock(struct ks8851_net *ks, unsigned long *flags)
+{
+	mutex_lock(&ks->lock);
+}
+
+/**
+ * ks8851_unlock - register access unlock
+ * @ks: The chip state
+ * @flags: Spinlock flags
+ *
+ * Release chip register access lock
+ */
+static void ks8851_unlock(struct ks8851_net *ks, unsigned long *flags)
+{
+	mutex_unlock(&ks->lock);
+}
+
 /* SPI register read/write calls.
  *
  * All these calls issue SPI transactions to access the chip's registers. They
@@ -304,10 +328,11 @@ static void ks8851_set_powermode(struct ks8851_net *ks, unsigned pwrmode)
 static int ks8851_write_mac_addr(struct net_device *dev)
 {
 	struct ks8851_net *ks = netdev_priv(dev);
+	unsigned long flags;
 	u16 val;
 	int i;
 
-	mutex_lock(&ks->lock);
+	ks8851_lock(ks, &flags);
 
 	/*
 	 * Wake up chip in case it was powered off when stopped; otherwise,
@@ -323,7 +348,7 @@ static int ks8851_write_mac_addr(struct net_device *dev)
 	if (!netif_running(dev))
 		ks8851_set_powermode(ks, PMECR_PM_SOFTDOWN);
 
-	mutex_unlock(&ks->lock);
+	ks8851_unlock(ks, &flags);
 
 	return 0;
 }
@@ -337,10 +362,11 @@ static int ks8851_write_mac_addr(struct net_device *dev)
 static void ks8851_read_mac_addr(struct net_device *dev)
 {
 	struct ks8851_net *ks = netdev_priv(dev);
+	unsigned long flags;
 	u16 reg;
 	int i;
 
-	mutex_lock(&ks->lock);
+	ks8851_lock(ks, &flags);
 
 	for (i = 0; i < ETH_ALEN; i += 2) {
 		reg = ks8851_rdreg16(ks, KS_MAR(i));
@@ -348,7 +374,7 @@ static void ks8851_read_mac_addr(struct net_device *dev)
 		dev->dev_addr[i + 1] = reg & 0xff;
 	}
 
-	mutex_unlock(&ks->lock);
+	ks8851_unlock(ks, &flags);
 }
 
 /**
@@ -534,10 +560,11 @@ static void ks8851_rx_pkts(struct ks8851_net *ks)
 static irqreturn_t ks8851_irq(int irq, void *_ks)
 {
 	struct ks8851_net *ks = _ks;
-	unsigned status;
 	unsigned handled = 0;
+	unsigned long flags;
+	unsigned status;
 
-	mutex_lock(&ks->lock);
+	ks8851_lock(ks, &flags);
 
 	status = ks8851_rdreg16(ks, KS_ISR);
 
@@ -606,7 +633,7 @@ static irqreturn_t ks8851_irq(int irq, void *_ks)
 		ks8851_wrreg16(ks, KS_RXCR1, rxc->rxcr1);
 	}
 
-	mutex_unlock(&ks->lock);
+	ks8851_unlock(ks, &flags);
 
 	if (status & IRQ_LCI)
 		mii_check_link(&ks->mii);
@@ -700,10 +727,11 @@ static void ks8851_done_tx(struct ks8851_net *ks, struct sk_buff *txb)
 static void ks8851_tx_work(struct work_struct *work)
 {
 	struct ks8851_net *ks = container_of(work, struct ks8851_net, tx_work);
+	unsigned long flags;
 	struct sk_buff *txb;
 	bool last = skb_queue_empty(&ks->txq);
 
-	mutex_lock(&ks->lock);
+	ks8851_lock(ks, &flags);
 
 	while (!last) {
 		txb = skb_dequeue(&ks->txq);
@@ -719,7 +747,7 @@ static void ks8851_tx_work(struct work_struct *work)
 		}
 	}
 
-	mutex_unlock(&ks->lock);
+	ks8851_unlock(ks, &flags);
 }
 
 /**
@@ -732,6 +760,7 @@ static void ks8851_tx_work(struct work_struct *work)
 static int ks8851_net_open(struct net_device *dev)
 {
 	struct ks8851_net *ks = netdev_priv(dev);
+	unsigned long flags;
 	int ret;
 
 	ret = request_threaded_irq(dev->irq, NULL, ks8851_irq,
@@ -744,7 +773,7 @@ static int ks8851_net_open(struct net_device *dev)
 
 	/* lock the card, even if we may not actually be doing anything
 	 * else at the moment */
-	mutex_lock(&ks->lock);
+	ks8851_lock(ks, &flags);
 
 	netif_dbg(ks, ifup, ks->netdev, "opening\n");
 
@@ -804,7 +833,7 @@ static int ks8851_net_open(struct net_device *dev)
 
 	netif_dbg(ks, ifup, ks->netdev, "network device up\n");
 
-	mutex_unlock(&ks->lock);
+	ks8851_unlock(ks, &flags);
 	mii_check_link(&ks->mii);
 	return 0;
 }
@@ -820,22 +849,23 @@ static int ks8851_net_open(struct net_device *dev)
 static int ks8851_net_stop(struct net_device *dev)
 {
 	struct ks8851_net *ks = netdev_priv(dev);
+	unsigned long flags;
 
 	netif_info(ks, ifdown, dev, "shutting down\n");
 
 	netif_stop_queue(dev);
 
-	mutex_lock(&ks->lock);
+	ks8851_lock(ks, &flags);
 	/* turn off the IRQs and ack any outstanding */
 	ks8851_wrreg16(ks, KS_IER, 0x0000);
 	ks8851_wrreg16(ks, KS_ISR, 0xffff);
-	mutex_unlock(&ks->lock);
+	ks8851_unlock(ks, &flags);
 
 	/* stop any outstanding work */
 	flush_work(&ks->tx_work);
 	flush_work(&ks->rxctrl_work);
 
-	mutex_lock(&ks->lock);
+	ks8851_lock(ks, &flags);
 	/* shutdown RX process */
 	ks8851_wrreg16(ks, KS_RXCR1, 0x0000);
 
@@ -844,7 +874,7 @@ static int ks8851_net_stop(struct net_device *dev)
 
 	/* set powermode to soft power down to save power */
 	ks8851_set_powermode(ks, PMECR_PM_SOFTDOWN);
-	mutex_unlock(&ks->lock);
+	ks8851_unlock(ks, &flags);
 
 	/* ensure any queued tx buffers are dumped */
 	while (!skb_queue_empty(&ks->txq)) {
@@ -916,13 +946,14 @@ static netdev_tx_t ks8851_start_xmit(struct sk_buff *skb,
 static void ks8851_rxctrl_work(struct work_struct *work)
 {
 	struct ks8851_net *ks = container_of(work, struct ks8851_net, rxctrl_work);
+	unsigned long flags;
 
-	mutex_lock(&ks->lock);
+	ks8851_lock(ks, &flags);
 
 	/* need to shutdown RXQ before modifying filter parameters */
 	ks8851_wrreg16(ks, KS_RXCR1, 0x00);
 
-	mutex_unlock(&ks->lock);
+	ks8851_unlock(ks, &flags);
 }
 
 static void ks8851_set_rx_mode(struct net_device *dev)
@@ -1104,11 +1135,6 @@ static void ks8851_eeprom_regwrite(struct eeprom_93cx6 *ee)
  */
 static int ks8851_eeprom_claim(struct ks8851_net *ks)
 {
-	if (!(ks->rc_ccr & CCR_EEPROM))
-		return -ENOENT;
-
-	mutex_lock(&ks->lock);
-
 	/* start with clock low, cs high */
 	ks8851_wrreg16(ks, KS_EEPCR, EEPCR_EESA | EEPCR_EECS);
 	return 0;
@@ -1125,7 +1151,6 @@ static void ks8851_eeprom_release(struct ks8851_net *ks)
 	unsigned val = ks8851_rdreg16(ks, KS_EEPCR);
 
 	ks8851_wrreg16(ks, KS_EEPCR, val & ~EEPCR_EESA);
-	mutex_unlock(&ks->lock);
 }
 
 #define KS_EEPROM_MAGIC (0x00008851)
@@ -1135,6 +1160,7 @@ static int ks8851_set_eeprom(struct net_device *dev,
 {
 	struct ks8851_net *ks = netdev_priv(dev);
 	int offset = ee->offset;
+	unsigned long flags;
 	int len = ee->len;
 	u16 tmp;
 
@@ -1145,9 +1171,13 @@ static int ks8851_set_eeprom(struct net_device *dev,
 	if (ee->magic != KS_EEPROM_MAGIC)
 		return -EINVAL;
 
-	if (ks8851_eeprom_claim(ks))
+	if (!(ks->rc_ccr & CCR_EEPROM))
 		return -ENOENT;
 
+	ks8851_lock(ks, &flags);
+
+	ks8851_eeprom_claim(ks);
+
 	eeprom_93cx6_wren(&ks->eeprom, true);
 
 	/* ethtool currently only supports writing bytes, which means
@@ -1167,6 +1197,7 @@ static int ks8851_set_eeprom(struct net_device *dev,
 	eeprom_93cx6_wren(&ks->eeprom, false);
 
 	ks8851_eeprom_release(ks);
+	ks8851_unlock(ks, &flags);
 
 	return 0;
 }
@@ -1176,19 +1207,25 @@ static int ks8851_get_eeprom(struct net_device *dev,
 {
 	struct ks8851_net *ks = netdev_priv(dev);
 	int offset = ee->offset;
+	unsigned long flags;
 	int len = ee->len;
 
 	/* must be 2 byte aligned */
 	if (len & 1 || offset & 1)
 		return -EINVAL;
 
-	if (ks8851_eeprom_claim(ks))
+	if (!(ks->rc_ccr & CCR_EEPROM))
 		return -ENOENT;
 
+	ks8851_lock(ks, &flags);
+
+	ks8851_eeprom_claim(ks);
+
 	ee->magic = KS_EEPROM_MAGIC;
 
 	eeprom_93cx6_multiread(&ks->eeprom, offset/2, (__le16 *)data, len/2);
 	ks8851_eeprom_release(ks);
+	ks8851_unlock(ks, &flags);
 
 	return 0;
 }
@@ -1262,6 +1299,7 @@ static int ks8851_phy_reg(int reg)
 static int ks8851_phy_read(struct net_device *dev, int phy_addr, int reg)
 {
 	struct ks8851_net *ks = netdev_priv(dev);
+	unsigned long flags;
 	int ksreg;
 	int result;
 
@@ -1269,9 +1307,9 @@ static int ks8851_phy_read(struct net_device *dev, int phy_addr, int reg)
 	if (!ksreg)
 		return 0x0;	/* no error return allowed, so use zero */
 
-	mutex_lock(&ks->lock);
+	ks8851_lock(ks, &flags);
 	result = ks8851_rdreg16(ks, ksreg);
-	mutex_unlock(&ks->lock);
+	ks8851_unlock(ks, &flags);
 
 	return result;
 }
@@ -1280,13 +1318,14 @@ static void ks8851_phy_write(struct net_device *dev,
 			     int phy, int reg, int value)
 {
 	struct ks8851_net *ks = netdev_priv(dev);
+	unsigned long flags;
 	int ksreg;
 
 	ksreg = ks8851_phy_reg(reg);
 	if (ksreg) {
-		mutex_lock(&ks->lock);
+		ks8851_lock(ks, &flags);
 		ks8851_wrreg16(ks, ksreg, value);
-		mutex_unlock(&ks->lock);
+		ks8851_unlock(ks, &flags);
 	}
 }
 
Signature · 1 line
2.25.1
#11 lore raw

PATCH v3 11/18 net: ks8851: Factor out SKB receive function

In reply to Marek Vasut

Factor out this netif_rx_ni(), so it could be overridden by the parallel bus variant of the KS8851 driver.

Cc David S. Miller <[email protected]>
Cc Lukas Wunner <[email protected]>
Cc Petr Stetiar <[email protected]>
Cc YueHaibing <[email protected]>

---
V3: New patch

---
 drivers/net/ethernet/micrel/ks8851.c | 11 ++++++++++-
 1 file changed, 10 insertions(+), 1 deletion(-)
drivers/net/ethernet/micrel/ks8851.c +10 -1
index 7a91d30472f2..fe09795265b9 100644
@@ -465,6 +465,15 @@ static void ks8851_dbg_dumpkkt(struct ks8851_net *ks, u8 *rxpkt)
 		   rxpkt[12], rxpkt[13], rxpkt[14], rxpkt[15]);
 }
 
+/**
+ * ks8851_rx_skb - receive skbuff
+ * @skb: The skbuff
+ */
+static void ks8851_rx_skb(struct sk_buff *skb)
+{
+	netif_rx_ni(skb);
+}
+
 /**
  * ks8851_rx_pkts - receive packets from the host
  * @ks: The device information.
@@ -533,7 +542,7 @@ static void ks8851_rx_pkts(struct ks8851_net *ks)
 					ks8851_dbg_dumpkkt(ks, rxpkt);
 
 				skb->protocol = eth_type_trans(skb, ks->netdev);
-				netif_rx_ni(skb);
+				ks8851_rx_skb(skb);
 
 				ks->netdev->stats.rx_packets++;
 				ks->netdev->stats.rx_bytes += rxlen;
Signature · 1 line
2.25.1
#12 lore raw

PATCH v3 12/18 net: ks8851: Split out SPI specific entries in struct ks8851_net

In reply to Marek Vasut

Add a new struct ks8851_net_spi, which embeds the original struct ks8851_net and contains the entries specific only to the SPI variant of KS8851.

There should be no functional change.

Cc David S. Miller <[email protected]>
Cc Lukas Wunner <[email protected]>
Cc Petr Stetiar <[email protected]>
Cc YueHaibing <[email protected]>

---
V2: - Drop 'must be first' comment on the embedded struct ks8851_net
- Remove moved kerneldoc entries for struct ks8851_net
V3: - Split out also tx_work and lock

---
 drivers/net/ethernet/micrel/ks8851.c | 128 +++++++++++++++++----------
 1 file changed, 79 insertions(+), 49 deletions(-)
drivers/net/ethernet/micrel/ks8851.c +79 -49
index fe09795265b9..1487ca135658 100644
@@ -64,16 +64,11 @@ union ks8851_tx_hdr {
 /**
  * struct ks8851_net - KS8851 driver private data
  * @netdev: The network device we're bound to
- * @spidev: The spi device we're bound to.
- * @lock: Lock to ensure that the device is not accessed when busy.
  * @statelock: Lock on this structure for tx list.
  * @mii: The MII state information for the mii calls.
  * @rxctrl: RX settings for @rxctrl_work.
- * @tx_work: Work queue for tx packets
  * @rxctrl_work: Work queue for updating RX mode and multicast lists
  * @txq: Queue of packets for transmission.
- * @spi_msg1: pre-setup SPI transfer with one message, @spi_xfer1.
- * @spi_msg2: pre-setup SPI transfer with two messages, @spi_xfer2.
  * @txh: Space for generating packet TX header in DMA-able data
  * @rxd: Space for receiving SPI data, in DMA-able space.
  * @txd: Space for transmitting SPI data, in DMA-able space.
@@ -87,11 +82,6 @@ union ks8851_tx_hdr {
  * @vdd_io: Optional digital power supply for IO
  * @gpio: Optional reset_n gpio
  *
- * The @lock ensures that the chip is protected when certain operations are
- * in progress. When the read or write packet transfer is in progress, most
- * of the chip registers are not ccessible until the transfer is finished and
- * the DMA has been de-asserted.
- *
  * The @statelock is used to protect information in the structure which may
  * need to be accessed via several sources, such as the network driver layer
  * or one of the work queues.
@@ -102,8 +92,6 @@ union ks8851_tx_hdr {
  */
 struct ks8851_net {
 	struct net_device	*netdev;
-	struct spi_device	*spidev;
-	struct mutex		lock;
 	spinlock_t		statelock;
 
 	union ks8851_tx_hdr	txh ____cacheline_aligned;
@@ -121,22 +109,43 @@ struct ks8851_net {
 	struct mii_if_info	mii;
 	struct ks8851_rxctrl	rxctrl;
 
-	struct work_struct	tx_work;
 	struct work_struct	rxctrl_work;
 
 	struct sk_buff_head	txq;
 
-	struct spi_message	spi_msg1;
-	struct spi_message	spi_msg2;
-	struct spi_transfer	spi_xfer1;
-	struct spi_transfer	spi_xfer2[2];
-
 	struct eeprom_93cx6	eeprom;
 	struct regulator	*vdd_reg;
 	struct regulator	*vdd_io;
 	int			gpio;
 };
 
+/**
+ * struct ks8851_net_spi - KS8851 SPI driver private data
+ * @ks8851: KS8851 driver common private data
+ * @lock: Lock to ensure that the device is not accessed when busy.
+ * @tx_work: Work queue for tx packets
+ * @spidev: The spi device we're bound to.
+ * @spi_msg1: pre-setup SPI transfer with one message, @spi_xfer1.
+ * @spi_msg2: pre-setup SPI transfer with two messages, @spi_xfer2.
+ *
+ * The @lock ensures that the chip is protected when certain operations are
+ * in progress. When the read or write packet transfer is in progress, most
+ * of the chip registers are not ccessible until the transfer is finished and
+ * the DMA has been de-asserted.
+ */
+struct ks8851_net_spi {
+	struct ks8851_net	ks8851;
+	struct mutex		lock;
+	struct work_struct	tx_work;
+	struct spi_device	*spidev;
+	struct spi_message	spi_msg1;
+	struct spi_message	spi_msg2;
+	struct spi_transfer	spi_xfer1;
+	struct spi_transfer	spi_xfer2[2];
+};
+
+#define to_ks8851_spi(ks) container_of((ks), struct ks8851_net_spi, ks8851)
+
 static int msg_enable;
 
 /* SPI frame opcodes */
@@ -160,7 +169,9 @@ static int msg_enable;
  */
 static void ks8851_lock(struct ks8851_net *ks, unsigned long *flags)
 {
-	mutex_lock(&ks->lock);
+	struct ks8851_net_spi *kss = to_ks8851_spi(ks);
+
+	mutex_lock(&kss->lock);
 }
 
 /**
@@ -172,7 +183,9 @@ static void ks8851_lock(struct ks8851_net *ks, unsigned long *flags)
  */
 static void ks8851_unlock(struct ks8851_net *ks, unsigned long *flags)
 {
-	mutex_unlock(&ks->lock);
+	struct ks8851_net_spi *kss = to_ks8851_spi(ks);
+
+	mutex_unlock(&kss->lock);
 }
 
 /* SPI register read/write calls.
@@ -192,8 +205,9 @@ static void ks8851_unlock(struct ks8851_net *ks, unsigned long *flags)
  */
 static void ks8851_wrreg16(struct ks8851_net *ks, unsigned reg, unsigned val)
 {
-	struct spi_transfer *xfer = &ks->spi_xfer1;
-	struct spi_message *msg = &ks->spi_msg1;
+	struct ks8851_net_spi *kss = to_ks8851_spi(ks);
+	struct spi_transfer *xfer = &kss->spi_xfer1;
+	struct spi_message *msg = &kss->spi_msg1;
 	__le16 txb[2];
 	int ret;
 
@@ -204,7 +218,7 @@ static void ks8851_wrreg16(struct ks8851_net *ks, unsigned reg, unsigned val)
 	xfer->rx_buf = NULL;
 	xfer->len = 4;
 
-	ret = spi_sync(ks->spidev, msg);
+	ret = spi_sync(kss->spidev, msg);
 	if (ret < 0)
 		netdev_err(ks->netdev, "spi_sync() failed\n");
 }
@@ -222,6 +236,7 @@ static void ks8851_wrreg16(struct ks8851_net *ks, unsigned reg, unsigned val)
 static void ks8851_rdreg(struct ks8851_net *ks, unsigned op,
 			 u8 *rxb, unsigned rxl)
 {
+	struct ks8851_net_spi *kss = to_ks8851_spi(ks);
 	struct spi_transfer *xfer;
 	struct spi_message *msg;
 	__le16 *txb = (__le16 *)ks->txd;
@@ -230,9 +245,9 @@ static void ks8851_rdreg(struct ks8851_net *ks, unsigned op,
 
 	txb[0] = cpu_to_le16(op | KS_SPIOP_RD);
 
-	if (ks->spidev->master->flags & SPI_MASTER_HALF_DUPLEX) {
-		msg = &ks->spi_msg2;
-		xfer = ks->spi_xfer2;
+	if (kss->spidev->master->flags & SPI_MASTER_HALF_DUPLEX) {
+		msg = &kss->spi_msg2;
+		xfer = kss->spi_xfer2;
 
 		xfer->tx_buf = txb;
 		xfer->rx_buf = NULL;
@@ -243,18 +258,18 @@ static void ks8851_rdreg(struct ks8851_net *ks, unsigned op,
 		xfer->rx_buf = trx;
 		xfer->len = rxl;
 	} else {
-		msg = &ks->spi_msg1;
-		xfer = &ks->spi_xfer1;
+		msg = &kss->spi_msg1;
+		xfer = &kss->spi_xfer1;
 
 		xfer->tx_buf = txb;
 		xfer->rx_buf = trx;
 		xfer->len = rxl + 2;
 	}
 
-	ret = spi_sync(ks->spidev, msg);
+	ret = spi_sync(kss->spidev, msg);
 	if (ret < 0)
 		netdev_err(ks->netdev, "read: spi_sync() failed\n");
-	else if (ks->spidev->master->flags & SPI_MASTER_HALF_DUPLEX)
+	else if (kss->spidev->master->flags & SPI_MASTER_HALF_DUPLEX)
 		memcpy(rxb, trx, rxl);
 	else
 		memcpy(rxb, trx + 2, rxl);
@@ -424,8 +439,9 @@ static void ks8851_init_mac(struct ks8851_net *ks, struct device_node *np)
  */
 static void ks8851_rdfifo(struct ks8851_net *ks, u8 *buff, unsigned len)
 {
-	struct spi_transfer *xfer = ks->spi_xfer2;
-	struct spi_message *msg = &ks->spi_msg2;
+	struct ks8851_net_spi *kss = to_ks8851_spi(ks);
+	struct spi_transfer *xfer = kss->spi_xfer2;
+	struct spi_message *msg = &kss->spi_msg2;
 	u8 txb[1];
 	int ret;
 
@@ -444,7 +460,7 @@ static void ks8851_rdfifo(struct ks8851_net *ks, u8 *buff, unsigned len)
 	xfer->tx_buf = NULL;
 	xfer->len = len;
 
-	ret = spi_sync(ks->spidev, msg);
+	ret = spi_sync(kss->spidev, msg);
 	if (ret < 0)
 		netdev_err(ks->netdev, "%s: spi_sync() failed\n", __func__);
 }
@@ -678,8 +694,9 @@ static inline unsigned calc_txlen(unsigned len)
  */
 static void ks8851_wrpkt(struct ks8851_net *ks, struct sk_buff *txp, bool irq)
 {
-	struct spi_transfer *xfer = ks->spi_xfer2;
-	struct spi_message *msg = &ks->spi_msg2;
+	struct ks8851_net_spi *kss = to_ks8851_spi(ks);
+	struct spi_transfer *xfer = kss->spi_xfer2;
+	struct spi_message *msg = &kss->spi_msg2;
 	unsigned fid = 0;
 	int ret;
 
@@ -706,7 +723,7 @@ static void ks8851_wrpkt(struct ks8851_net *ks, struct sk_buff *txp, bool irq)
 	xfer->rx_buf = NULL;
 	xfer->len = ALIGN(txp->len, 4);
 
-	ret = spi_sync(ks->spidev, msg);
+	ret = spi_sync(kss->spidev, msg);
 	if (ret < 0)
 		netdev_err(ks->netdev, "%s: spi_sync() failed\n", __func__);
 }
@@ -735,10 +752,15 @@ static void ks8851_done_tx(struct ks8851_net *ks, struct sk_buff *txb)
  */
 static void ks8851_tx_work(struct work_struct *work)
 {
-	struct ks8851_net *ks = container_of(work, struct ks8851_net, tx_work);
+	struct ks8851_net_spi *kss;
+	struct ks8851_net *ks;
 	unsigned long flags;
 	struct sk_buff *txb;
-	bool last = skb_queue_empty(&ks->txq);
+	bool last;
+
+	kss = container_of(work, struct ks8851_net_spi, tx_work);
+	ks = &kss->ks8851;
+	last = skb_queue_empty(&ks->txq);
 
 	ks8851_lock(ks, &flags);
 
@@ -858,8 +880,11 @@ static int ks8851_net_open(struct net_device *dev)
 static int ks8851_net_stop(struct net_device *dev)
 {
 	struct ks8851_net *ks = netdev_priv(dev);
+	struct ks8851_net_spi *kss;
 	unsigned long flags;
 
+	kss = to_ks8851_spi(ks);
+
 	netif_info(ks, ifdown, dev, "shutting down\n");
 
 	netif_stop_queue(dev);
@@ -871,7 +896,7 @@ static int ks8851_net_stop(struct net_device *dev)
 	ks8851_unlock(ks, &flags);
 
 	/* stop any outstanding work */
-	flush_work(&ks->tx_work);
+	flush_work(&kss->tx_work);
 	flush_work(&ks->rxctrl_work);
 
 	ks8851_lock(ks, &flags);
@@ -919,6 +944,9 @@ static netdev_tx_t ks8851_start_xmit(struct sk_buff *skb,
 	struct ks8851_net *ks = netdev_priv(dev);
 	unsigned needed = calc_txlen(skb->len);
 	netdev_tx_t ret = NETDEV_TX_OK;
+	struct ks8851_net_spi *kss;
+
+	kss = to_ks8851_spi(ks);
 
 	netif_dbg(ks, tx_queued, ks->netdev,
 		  "%s: skb %p, %d@%p\n", __func__, skb, skb->len, skb->data);
@@ -934,7 +962,7 @@ static netdev_tx_t ks8851_start_xmit(struct sk_buff *skb,
 	}
 
 	spin_unlock(&ks->statelock);
-	schedule_work(&ks->tx_work);
+	schedule_work(&kss->tx_work);
 
 	return ret;
 }
@@ -1406,22 +1434,24 @@ static SIMPLE_DEV_PM_OPS(ks8851_pm_ops, ks8851_suspend, ks8851_resume);
 static int ks8851_probe(struct spi_device *spi)
 {
 	struct device *dev = &spi->dev;
+	struct ks8851_net_spi *kss;
 	struct net_device *netdev;
 	struct ks8851_net *ks;
 	int ret;
 	unsigned cider;
 	int gpio;
 
-	netdev = devm_alloc_etherdev(dev, sizeof(struct ks8851_net));
+	netdev = devm_alloc_etherdev(dev, sizeof(struct ks8851_net_spi));
 	if (!netdev)
 		return -ENOMEM;
 
 	spi->bits_per_word = 8;
 
 	ks = netdev_priv(netdev);
+	kss = to_ks8851_spi(ks);
 
 	ks->netdev = netdev;
-	ks->spidev = spi;
+	kss->spidev = spi;
 	ks->tx_space = 6144;
 
 	gpio = of_get_named_gpio_flags(dev->of_node, "reset-gpios", 0, NULL);
@@ -1467,20 +1497,20 @@ static int ks8851_probe(struct spi_device *spi)
 		gpio_set_value(gpio, 1);
 	}
 
-	mutex_init(&ks->lock);
+	mutex_init(&kss->lock);
 	spin_lock_init(&ks->statelock);
 
-	INIT_WORK(&ks->tx_work, ks8851_tx_work);
+	INIT_WORK(&kss->tx_work, ks8851_tx_work);
 	INIT_WORK(&ks->rxctrl_work, ks8851_rxctrl_work);
 
 	/* initialise pre-made spi transfer messages */
 
-	spi_message_init(&ks->spi_msg1);
-	spi_message_add_tail(&ks->spi_xfer1, &ks->spi_msg1);
+	spi_message_init(&kss->spi_msg1);
+	spi_message_add_tail(&kss->spi_xfer1, &kss->spi_msg1);
 
-	spi_message_init(&ks->spi_msg2);
-	spi_message_add_tail(&ks->spi_xfer2[0], &ks->spi_msg2);
-	spi_message_add_tail(&ks->spi_xfer2[1], &ks->spi_msg2);
+	spi_message_init(&kss->spi_msg2);
+	spi_message_add_tail(&kss->spi_xfer2[0], &kss->spi_msg2);
+	spi_message_add_tail(&kss->spi_xfer2[1], &kss->spi_msg2);
 
 	/* setup EEPROM state */
 
Signature · 1 line
2.25.1
#13 lore raw

PATCH v3 13/18 net: ks8851: Split out SPI specific code from probe() and remove()

In reply to Marek Vasut

Factor out common code into ks8851_probe_common() and ks8851_remove_common() to permit both SPI and parallel bus driver variants to use the common code path for both probing and removal.

There should be no functional change.

Cc David S. Miller <[email protected]>
Cc Lukas Wunner <[email protected]>
Cc Petr Stetiar <[email protected]>
Cc YueHaibing <[email protected]>

---
V2: - Add RB from Andrew
- Rework on top of locking patches, drop RB
V3: No change

---
 drivers/net/ethernet/micrel/ks8851.c | 77 ++++++++++++++++------------
 1 file changed, 43 insertions(+), 34 deletions(-)
drivers/net/ethernet/micrel/ks8851.c +43 -34
index 1487ca135658..dc7ff857b53c 100644
@@ -1431,27 +1431,14 @@ static int ks8851_resume(struct device *dev)
 
 static SIMPLE_DEV_PM_OPS(ks8851_pm_ops, ks8851_suspend, ks8851_resume);
 
-static int ks8851_probe(struct spi_device *spi)
+static int ks8851_probe_common(struct net_device *netdev, struct device *dev)
 {
-	struct device *dev = &spi->dev;
-	struct ks8851_net_spi *kss;
-	struct net_device *netdev;
-	struct ks8851_net *ks;
-	int ret;
+	struct ks8851_net *ks = netdev_priv(netdev);
 	unsigned cider;
 	int gpio;
-
-	netdev = devm_alloc_etherdev(dev, sizeof(struct ks8851_net_spi));
-	if (!netdev)
-		return -ENOMEM;
-
-	spi->bits_per_word = 8;
-
-	ks = netdev_priv(netdev);
-	kss = to_ks8851_spi(ks);
+	int ret;
 
 	ks->netdev = netdev;
-	kss->spidev = spi;
 	ks->tx_space = 6144;
 
 	gpio = of_get_named_gpio_flags(dev->of_node, "reset-gpios", 0, NULL);
@@ -1497,23 +1484,11 @@ static int ks8851_probe(struct spi_device *spi)
 		gpio_set_value(gpio, 1);
 	}
 
-	mutex_init(&kss->lock);
 	spin_lock_init(&ks->statelock);
 
-	INIT_WORK(&kss->tx_work, ks8851_tx_work);
 	INIT_WORK(&ks->rxctrl_work, ks8851_rxctrl_work);
 
-	/* initialise pre-made spi transfer messages */
-
-	spi_message_init(&kss->spi_msg1);
-	spi_message_add_tail(&kss->spi_xfer1, &kss->spi_msg1);
-
-	spi_message_init(&kss->spi_msg2);
-	spi_message_add_tail(&kss->spi_xfer2[0], &kss->spi_msg2);
-	spi_message_add_tail(&kss->spi_xfer2[1], &kss->spi_msg2);
-
 	/* setup EEPROM state */
-
 	ks->eeprom.data = ks;
 	ks->eeprom.width = PCI_EEPROM_WIDTH_93C46;
 	ks->eeprom.register_read = ks8851_eeprom_regread;
@@ -1544,7 +1519,6 @@ static int ks8851_probe(struct spi_device *spi)
 	netif_carrier_off(ks->netdev);
 	netdev->if_port = IF_PORT_100BASET;
 	netdev->netdev_ops = &ks8851_netdev_ops;
-	netdev->irq = spi->irq;
 
 	/* issue a global soft reset to reset the device. */
 	ks8851_soft_reset(ks, GRR_GSR);
@@ -1586,12 +1560,9 @@ static int ks8851_probe(struct spi_device *spi)
 	return ret;
 }
 
-static int ks8851_remove(struct spi_device *spi)
+static int ks8851_remove_common(struct device *dev)
 {
-	struct device *dev = &spi->dev;
-	struct ks8851_net *priv;
-
-	priv = dev_get_drvdata(dev);
+	struct ks8851_net *priv = dev_get_drvdata(dev);
 
 	if (netif_msg_drv(priv))
 		dev_info(dev, "remove\n");
@@ -1605,6 +1576,44 @@ static int ks8851_remove(struct spi_device *spi)
 	return 0;
 }
 
+static int ks8851_probe(struct spi_device *spi)
+{
+	struct device *dev = &spi->dev;
+	struct ks8851_net_spi *kss;
+	struct net_device *netdev;
+	struct ks8851_net *ks;
+
+	netdev = devm_alloc_etherdev(dev, sizeof(struct ks8851_net_spi));
+	if (!netdev)
+		return -ENOMEM;
+
+	spi->bits_per_word = 8;
+
+	ks = netdev_priv(netdev);
+	kss = to_ks8851_spi(ks);
+
+	kss->spidev = spi;
+	mutex_init(&kss->lock);
+	INIT_WORK(&kss->tx_work, ks8851_tx_work);
+
+	/* initialise pre-made spi transfer messages */
+	spi_message_init(&kss->spi_msg1);
+	spi_message_add_tail(&kss->spi_xfer1, &kss->spi_msg1);
+
+	spi_message_init(&kss->spi_msg2);
+	spi_message_add_tail(&kss->spi_xfer2[0], &kss->spi_msg2);
+	spi_message_add_tail(&kss->spi_xfer2[1], &kss->spi_msg2);
+
+	netdev->irq = spi->irq;
+
+	return ks8851_probe_common(netdev, dev);
+}
+
+static int ks8851_remove(struct spi_device *spi)
+{
+	return ks8851_remove_common(&spi->dev);
+}
+
 static const struct of_device_id ks8851_match_table[] = {
 	{ .compatible = "micrel,ks8851" },
 	{ }
Signature · 1 line
2.25.1
#14 lore raw

PATCH v3 14/18 net: ks8851: Factor out TX work flush function

In reply to Marek Vasut

While the SPI version of the KS8851 requires a TX worker thread to pump data via SPI, the parallel bus version can write data into the TX FIFO directly in .ndo_start_xmit, as the parallel bus access is much faster and does not sleep. Factor out this TX work flush part, so it can be overridden by the parallel bus driver.

Cc David S. Miller <[email protected]>
Cc Lukas Wunner <[email protected]>
Cc Petr Stetiar <[email protected]>
Cc YueHaibing <[email protected]>

---
V3: New patch

---
 drivers/net/ethernet/micrel/ks8851.c | 16 ++++++++++++----
 1 file changed, 12 insertions(+), 4 deletions(-)
drivers/net/ethernet/micrel/ks8851.c +12 -4
index dc7ff857b53c..0b60acceeae9 100644
@@ -781,6 +781,17 @@ static void ks8851_tx_work(struct work_struct *work)
 	ks8851_unlock(ks, &flags);
 }
 
+/**
+ * ks8851_flush_tx_work - flush outstanding TX work
+ * @ks: The device state
+ */
+static void ks8851_flush_tx_work(struct ks8851_net *ks)
+{
+	struct ks8851_net_spi *kss = to_ks8851_spi(ks);
+
+	flush_work(&kss->tx_work);
+}
+
 /**
  * ks8851_net_open - open network device
  * @dev: The network device being opened.
@@ -880,11 +891,8 @@ static int ks8851_net_open(struct net_device *dev)
 static int ks8851_net_stop(struct net_device *dev)
 {
 	struct ks8851_net *ks = netdev_priv(dev);
-	struct ks8851_net_spi *kss;
 	unsigned long flags;
 
-	kss = to_ks8851_spi(ks);
-
 	netif_info(ks, ifdown, dev, "shutting down\n");
 
 	netif_stop_queue(dev);
@@ -896,7 +904,7 @@ static int ks8851_net_stop(struct net_device *dev)
 	ks8851_unlock(ks, &flags);
 
 	/* stop any outstanding work */
-	flush_work(&kss->tx_work);
+	ks8851_flush_tx_work(ks);
 	flush_work(&ks->rxctrl_work);
 
 	ks8851_lock(ks, &flags);
Signature · 1 line
2.25.1
#15 lore raw

PATCH v3 15/18 net: ks8851: Permit overridding interrupt enable register

In reply to Marek Vasut

The parallel bus variant does not need to use the TX interrupt at all as it writes the TX FIFO directly with in .ndo_start_xmit, permit the drivers to configure the interrupt enable bits.

Cc David S. Miller <[email protected]>
Cc Lukas Wunner <[email protected]>
Cc Petr Stetiar <[email protected]>
Cc YueHaibing <[email protected]>

---
V3: New patch

---
 drivers/net/ethernet/micrel/ks8851.c | 22 +++++++++++-----------
 1 file changed, 11 insertions(+), 11 deletions(-)
drivers/net/ethernet/micrel/ks8851.c +11 -11
index 0b60acceeae9..d29b52ffbb49 100644
@@ -859,17 +859,8 @@ static int ks8851_net_open(struct net_device *dev)
 	ks8851_wrreg16(ks, KS_RXQCR, ks->rc_rxqcr);
 
 	/* clear then enable interrupts */
-
-#define STD_IRQ (IRQ_LCI |	/* Link Change */	\
-		 IRQ_TXI |	/* TX done */		\
-		 IRQ_RXI |	/* RX done */		\
-		 IRQ_SPIBEI |	/* SPI bus error */	\
-		 IRQ_TXPSI |	/* TX process stop */	\
-		 IRQ_RXPSI)	/* RX process stop */
-
-	ks->rc_ier = STD_IRQ;
-	ks8851_wrreg16(ks, KS_ISR, STD_IRQ);
-	ks8851_wrreg16(ks, KS_IER, STD_IRQ);
+	ks8851_wrreg16(ks, KS_ISR, ks->rc_ier);
+	ks8851_wrreg16(ks, KS_IER, ks->rc_ier);
 
 	netif_start_queue(ks->netdev);
 
@@ -1598,6 +1589,15 @@ static int ks8851_probe(struct spi_device *spi)
 	spi->bits_per_word = 8;
 
 	ks = netdev_priv(netdev);
+
+#define STD_IRQ (IRQ_LCI |	/* Link Change */	\
+		 IRQ_TXI |	/* TX done */		\
+		 IRQ_RXI |	/* RX done */		\
+		 IRQ_SPIBEI |	/* SPI bus error */	\
+		 IRQ_TXPSI |	/* TX process stop */	\
+		 IRQ_RXPSI)	/* RX process stop */
+	ks->rc_ier = STD_IRQ;
+
 	kss = to_ks8851_spi(ks);
 
 	kss->spidev = spi;
Signature · 1 line
2.25.1
#16 lore raw

PATCH v3 16/18 net: ks8851: Separate SPI operations into separate file

In reply to Marek Vasut

Pull all the SPI bus specific code into a separate file, so that it is not mixed with the common code. Rename ks8851.c to ks8851_common.c. The ks8851_common.c is linked with ks8851_spi.c now, so it can call the accessors in the ks8851_spi.c without any pointer indirection.

Cc David S. Miller <[email protected]>
Cc Lukas Wunner <[email protected]>
Cc Petr Stetiar <[email protected]>
Cc YueHaibing <[email protected]>

---
V2: Get rid of the pointer indirection
V3: Rename ks8851.c to ks8851_common.c and adjust Makefile to permit
module to be built

---
 drivers/net/ethernet/micrel/Makefile          |   1 +
 drivers/net/ethernet/micrel/ks8851.h          | 140 +++++
 .../micrel/{ks8851.c => ks8851_common.c}      | 537 +-----------------
 drivers/net/ethernet/micrel/ks8851_spi.c      | 448 +++++++++++++++
 4 files changed, 593 insertions(+), 533 deletions(-)
 rename drivers/net/ethernet/micrel/{ks8851.c => ks8851_common.c} (66%)
 create mode 100644 drivers/net/ethernet/micrel/ks8851_spi.c
drivers/net/ethernet/micrel/Makefile +1 -0
index 6d8ac5527aef..c7a4725c2e95 100644
@@ -5,5 +5,6 @@
 
 obj-$(CONFIG_KS8842) += ks8842.o
 obj-$(CONFIG_KS8851) += ks8851.o
+ks8851-objs = ks8851_common.o ks8851_spi.o
 obj-$(CONFIG_KS8851_MLL) += ks8851_mll.o
 obj-$(CONFIG_KSZ884X_PCI) += ksz884x.o
drivers/net/ethernet/micrel/ks8851.h +140 -0

149 lines not on this page · raw message

drivers/net/ethernet/micrel/ks8851_common.c renamed from drivers/net/ethernet/micrel/ks8851.c +4 -533

605 lines not on this page · raw message

drivers/net/ethernet/micrel/ks8851_spi.c new file +448 -0

448 lines not on this page · raw message

Signature · 1 line
2.25.1
#17 lore raw

Re: [PATCH V3 16/18] net: ks8851: Separate SPI operations into separate file

I love your patch! Yet something to improve:

Hi Marek,

I love your patch! Yet something to improve:

[auto build test ERROR on net/master]
[also build test ERROR on net-next/master linus/master v5.6-rc7 next-20200327] [if your patch is applied to the wrong git tree, please drop us a note to help improve the system. BTW, we also suggest to use '--base' option to specify the base tree in git format-patch, please see https://stackoverflow.com/a/37406982]

url:    https://github.com/0day-ci/linux/commits/Marek-Vasut/net-ks8851-Unify-KS8851-SPI-and-MLL-drivers/20200328-083415
base:   https://git.kernel.org/pub/scm/linux/kernel/git/davem/net.git 8262e6f9b1034ede34548a04dec4c302d92c9497
config: m68k-allmodconfig (attached as .config)
compiler: m68k-linux-gcc (GCC) 9.2.0

reproduce:
wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross chmod +x ~/bin/make.cross
# save the attached .config to linux build tree
GCC_VERSION=9.2.0 make.cross ARCH=m68k

If you fix the issue, kindly add following tag

All errors (new ones prefixed by >>):

drivers/net/ethernet/micrel/ks8851_spi.c: In function 'ks8851_start_xmit':

>> drivers/net/ethernet/micrel/ks8851_spi.c:360:20: error: implicit declaration of function 'calc_txlen' [-Werror=implicit-function-declaration]
     360 |  unsigned needed = calc_txlen(skb->len);
         |                    ^~~~~~~~~~
>> drivers/net/ethernet/micrel/ks8851_spi.c:377:19: error: 'struct ks8851_net' has no member named 'tx_work'
     377 |  schedule_work(&ks->tx_work);
         |                   ^~

cc1: some warnings being treated as errors

vim +/calc_txlen +360 drivers/net/ethernet/micrel/ks8851_spi.c

343
344 /**
345 * ks8851_start_xmit - transmit packet
346 * @skb: The buffer to transmit
347 * @dev: The device used to transmit the packet.
348 *
349 * Called by the network layer to transmit the @skb. Queue the packet for
350 * the device and schedule the necessary work to transmit the packet when
351 * it is free.
352 *
353 * We do this to firstly avoid sleeping with the network device locked,
354 * and secondly so we can round up more than one packet to transmit which
355 * means we can try and avoid generating too many transmit done interrupts.
356 */
357 netdev_tx_t ks8851_start_xmit(struct sk_buff *skb, struct net_device *dev)
358 {
359 struct ks8851_net *ks = netdev_priv(dev);

 > 360		unsigned needed = calc_txlen(skb->len);
   361		netdev_tx_t ret = NETDEV_TX_OK;
   362	
   363		netif_dbg(ks, tx_queued, ks->netdev,
   364			  "%s: skb %p, %d@%p\n", __func__, skb, skb->len, skb->data);
   365	
   366		spin_lock(&ks->statelock);
   367	
   368		if (needed > ks->tx_space) {
   369			netif_stop_queue(dev);
   370			ret = NETDEV_TX_BUSY;
   371		} else {
   372			ks->tx_space -= needed;
   373			skb_queue_tail(&ks->txq, skb);
   374		}
   375	
   376		spin_unlock(&ks->statelock);
 > 377		schedule_work(&ks->tx_work);

378
379 return ret;
380 }
381

---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all(a)lists.01.org

#18 lore raw

Re: [PATCH V3 16/18] net: ks8851: Separate SPI operations into separate file

In reply to Marek Vasut
I love your patch! Yet something to improve:

Hi Marek,

I love your patch! Yet something to improve:

[auto build test ERROR on net/master]
[also build test ERROR on net-next/master linus/master v5.6-rc7 next-20200327] [if your patch is applied to the wrong git tree, please drop us a note to help improve the system. BTW, we also suggest to use '--base' option to specify the base tree in git format-patch, please see https://stackoverflow.com/a/37406982]

url:    https://github.com/0day-ci/linux/commits/Marek-Vasut/net-ks8851-Unify-KS8851-SPI-and-MLL-drivers/20200328-083415
base:   https://git.kernel.org/pub/scm/linux/kernel/git/davem/net.git 8262e6f9b1034ede34548a04dec4c302d92c9497
config: i386-allyesconfig (attached as .config)
compiler: gcc-7 (Debian 7.5.0-5) 7.5.0
reproduce:
        # save the attached .config to linux build tree
        make ARCH=i386 

If you fix the issue, kindly add following tag

All errors (new ones prefixed by >>):

drivers/net/ethernet/micrel/ks8851_spi.c: In function 'ks8851_start_xmit':

>> drivers/net/ethernet/micrel/ks8851_spi.c:360:20: error: implicit declaration of function 'calc_txlen'; did you mean 'alt_rlen'? [-Werror=implicit-function-declaration]
     unsigned needed = calc_txlen(skb->len);
                       ^~~~~~~~~~
                       alt_rlen
   drivers/net/ethernet/micrel/ks8851_spi.c:377:19: error: 'struct ks8851_net' has no member named 'tx_work'
     schedule_work(&ks->tx_work);
                      ^~

cc1: some warnings being treated as errors

vim +360 drivers/net/ethernet/micrel/ks8851_spi.c

343
344 /**
345 * ks8851_start_xmit - transmit packet
346 * @skb: The buffer to transmit
347 * @dev: The device used to transmit the packet.
348 *
349 * Called by the network layer to transmit the @skb. Queue the packet for
350 * the device and schedule the necessary work to transmit the packet when
351 * it is free.
352 *
353 * We do this to firstly avoid sleeping with the network device locked,
354 * and secondly so we can round up more than one packet to transmit which
355 * means we can try and avoid generating too many transmit done interrupts.
356 */
357 netdev_tx_t ks8851_start_xmit(struct sk_buff *skb, struct net_device *dev)
358 {
359 struct ks8851_net *ks = netdev_priv(dev);

 > 360		unsigned needed = calc_txlen(skb->len);

361 netdev_tx_t ret = NETDEV_TX_OK;
362
363 netif_dbg(ks, tx_queued, ks->netdev,
364 "%s: skb %p, %d@%p\n", __func__, skb, skb->len, skb->data);
365
366 spin_lock(&ks->statelock);
367
368 if (needed > ks->tx_space) {
369 netif_stop_queue(dev);
370 ret = NETDEV_TX_BUSY;
371 } else {
372 ks->tx_space -= needed;
373 skb_queue_tail(&ks->txq, skb);
374 }
375
376 spin_unlock(&ks->statelock);
377 schedule_work(&ks->tx_work);
378
379 return ret;
380 }
381

---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all(a)lists.01.org

#19 lore raw

PATCH v3 17/18 net: ks8851: Implement Parallel bus operations

In reply to Marek Vasut

Implement accessors for KS8851-16MLL/MLLI/MLLU parallel bus variant of the KS8851. This is based off the ks8851_mll.c , which is a driver for exactly the same hardware, however the ks8851.c code is much higher quality. Hence, this patch pulls out the relevant information from the ks8851_mll.c on how to access the bus, but uses the common ks8851.c code. To make this patch reviewable, instead of rewriting ks8851_mll.c, ks8851_mll.c is removed in a separate subsequent patch.

Cc David S. Miller <[email protected]>
Cc Lukas Wunner <[email protected]>
Cc Petr Stetiar <[email protected]>
Cc YueHaibing <[email protected]>

---
V2: - Drop 'must be first' comment on the embedded struct ks8851_net
- Add missing Kconfig 'select's to match the KS8851 SPI driver
- Get rid of the pointer indirection
- Repair the endian handling and add endian check
V3: - Adjust Makefile to permit build as a module
- Enable only RX and LCx interrupts

---
 drivers/net/ethernet/micrel/Kconfig      |   2 +
 drivers/net/ethernet/micrel/Makefile     |   1 +
 drivers/net/ethernet/micrel/ks8851_par.c | 337 +++++++++++++++++++++++
 3 files changed, 340 insertions(+)
 create mode 100644 drivers/net/ethernet/micrel/ks8851_par.c
drivers/net/ethernet/micrel/Kconfig +2 -0
index b9c4d48e28e4..09f35209d43d 100644
@@ -38,6 +38,8 @@ config KS8851_MLL
 	tristate "Micrel KS8851 MLL"
 	depends on HAS_IOMEM
 	select MII
+	select CRC32
+	select EEPROM_93CX6
 	---help---
 	  This platform driver is for Micrel KS8851 Address/data bus
 	  multiplexed network chip.
drivers/net/ethernet/micrel/Makefile +1 -0
index c7a4725c2e95..5cc00d22c708 100644
@@ -7,4 +7,5 @@ obj-$(CONFIG_KS8842) += ks8842.o
 obj-$(CONFIG_KS8851) += ks8851.o
 ks8851-objs = ks8851_common.o ks8851_spi.o
 obj-$(CONFIG_KS8851_MLL) += ks8851_mll.o
+ks8851_mll-objs = ks8851_common.o ks8851_par.o
 obj-$(CONFIG_KSZ884X_PCI) += ksz884x.o
drivers/net/ethernet/micrel/ks8851_par.c new file +337 -0

337 lines not on this page · raw message

Signature · 1 line
2.25.1
#20 lore raw

PATCH v3 18/18 net: ks8851: Remove ks8851_mll.c

In reply to Marek Vasut

The ks8851_mll.c is replaced by ks8851_par.c, which is using common code from ks8851.c, just like ks8851_spi.c . Remove this old ad-hoc driver.

Cc David S. Miller <[email protected]>
Cc Lukas Wunner <[email protected]>
Cc Petr Stetiar <[email protected]>
Cc YueHaibing <[email protected]>

---
V2: No change
V3: No change

---
 drivers/net/ethernet/micrel/ks8851_mll.c | 1393 ----------------------
 1 file changed, 1393 deletions(-)
 delete mode 100644 drivers/net/ethernet/micrel/ks8851_mll.c
drivers/net/ethernet/micrel/ks8851_mll.c deleted +0 -1393
deleted file mode 100644
index 45cc840d8e2e..000000000000
@@ -1,1393 +0,0 @@
-// SPDX-License-Identifier: GPL-2.0-only
-/**
- * drivers/net/ethernet/micrel/ks8851_mll.c
- * Copyright (c) 2009 Micrel Inc.
- */
-
-/* Supports:
- * KS8851 16bit MLL chip from Micrel Inc.
- */
-
-#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
-
-#include <linux/interrupt.h>
-#include <linux/module.h>
-#include <linux/kernel.h>
-#include <linux/netdevice.h>
-#include <linux/etherdevice.h>
-#include <linux/ethtool.h>
-#include <linux/cache.h>
-#include <linux/crc32.h>
-#include <linux/crc32poly.h>
-#include <linux/mii.h>
-#include <linux/platform_device.h>
-#include <linux/delay.h>
-#include <linux/slab.h>
-#include <linux/ks8851_mll.h>
-#include <linux/of.h>
-#include <linux/of_device.h>
-#include <linux/of_net.h>
-
-#include "ks8851.h"
-
-#define	DRV_NAME	"ks8851_mll"
-
-static u8 KS_DEFAULT_MAC_ADDRESS[] = { 0x00, 0x10, 0xA1, 0x86, 0x95, 0x11 };
-#define MAX_RECV_FRAMES			255
-#define MAX_BUF_SIZE			2048
-#define TX_BUF_SIZE			2000
-#define RX_BUF_SIZE			2000
-
-#define RXCR1_FILTER_MASK    		(RXCR1_RXINVF | RXCR1_RXAE | \
-					 RXCR1_RXMAFMA | RXCR1_RXPAFMA)
-#define RXQCR_CMD_CNTL                	(RXQCR_RXFCTE|RXQCR_ADRFE)
-
-#define	ENUM_BUS_NONE			0
-#define	ENUM_BUS_8BIT			1
-#define	ENUM_BUS_16BIT			2
-#define	ENUM_BUS_32BIT			3
-
-#define MAX_MCAST_LST			32
-#define HW_MCAST_SIZE			8
-
-/**
- * union ks_tx_hdr - tx header data
- * @txb: The header as bytes
- * @txw: The header as 16bit, little-endian words
- *
- * A dual representation of the tx header data to allow
- * access to individual bytes, and to allow 16bit accesses
- * with 16bit alignment.
- */
-union ks_tx_hdr {
-	u8      txb[4];
-	__le16  txw[2];
-};
-
-/**
- * struct ks_net - KS8851 driver private data
- * @net_device 	: The network device we're bound to
- * @hw_addr	: start address of data register.
- * @hw_addr_cmd	: start address of command register.
- * @txh    	: temporaly buffer to save status/length.
- * @lock	: Lock to ensure that the device is not accessed when busy.
- * @pdev	: Pointer to platform device.
- * @mii		: The MII state information for the mii calls.
- * @frame_head_info   	: frame header information for multi-pkt rx.
- * @statelock	: Lock on this structure for tx list.
- * @msg_enable	: The message flags controlling driver output (see ethtool).
- * @frame_cnt  	: number of frames received.
- * @bus_width  	: i/o bus width.
- * @rc_rxqcr	: Cached copy of KS_RXQCR.
- * @rc_txcr	: Cached copy of KS_TXCR.
- * @rc_ier	: Cached copy of KS_IER.
- * @sharedbus  	: Multipex(addr and data bus) mode indicator.
- * @cmd_reg_cache	: command register cached.
- * @cmd_reg_cache_int	: command register cached. Used in the irq handler.
- * @promiscuous	: promiscuous mode indicator.
- * @all_mcast  	: mutlicast indicator.
- * @mcast_lst_size   	: size of multicast list.
- * @mcast_lst    	: multicast list.
- * @mcast_bits    	: multicast enabed.
- * @mac_addr   		: MAC address assigned to this device.
- * @fid    		: frame id.
- * @extra_byte    	: number of extra byte prepended rx pkt.
- * @enabled    		: indicator this device works.
- *
- * The @lock ensures that the chip is protected when certain operations are
- * in progress. When the read or write packet transfer is in progress, most
- * of the chip registers are not accessible until the transfer is finished and
- * the DMA has been de-asserted.
- *
- * The @statelock is used to protect information in the structure which may
- * need to be accessed via several sources, such as the network driver layer
- * or one of the work queues.
- *
- */
-
-/* Receive multiplex framer header info */
-struct type_frame_head {
-	u16	sts;         /* Frame status */
-	u16	len;         /* Byte count */
-};
-
-struct ks_net {
-	struct net_device	*netdev;
-	void __iomem    	*hw_addr;
-	void __iomem    	*hw_addr_cmd;
-	union ks_tx_hdr		txh ____cacheline_aligned;
-	struct mutex      	lock; /* spinlock to be interrupt safe */
-	struct platform_device *pdev;
-	struct mii_if_info	mii;
-	struct type_frame_head	*frame_head_info;
-	spinlock_t		statelock;
-	u32			msg_enable;
-	u32			frame_cnt;
-	int			bus_width;
-
-	u16			rc_rxqcr;
-	u16			rc_txcr;
-	u16			rc_ier;
-	u16			sharedbus;
-	u16			cmd_reg_cache;
-	u16			cmd_reg_cache_int;
-	u16			promiscuous;
-	u16			all_mcast;
-	u16			mcast_lst_size;
-	u8			mcast_lst[MAX_MCAST_LST][ETH_ALEN];
-	u8			mcast_bits[HW_MCAST_SIZE];
-	u8			mac_addr[6];
-	u8                      fid;
-	u8			extra_byte;
-	u8			enabled;
-};
-
-static int msg_enable;
-
-#define BE3             0x8000      /* Byte Enable 3 */
-#define BE2             0x4000      /* Byte Enable 2 */
-#define BE1             0x2000      /* Byte Enable 1 */
-#define BE0             0x1000      /* Byte Enable 0 */
-
-/* register read/write calls.
- *
- * All these calls issue transactions to access the chip's registers. They
- * all require that the necessary lock is held to prevent accesses when the
- * chip is busy transferring packet data (RX/TX FIFO accesses).
- */
-
-/**
- * ks_check_endian - Check whether endianness of the bus is correct
- * @ks	  : The chip information
- *
- * The KS8851-16MLL EESK pin allows selecting the endianness of the 16bit
- * bus. To maintain optimum performance, the bus endianness should be set
- * such that it matches the endianness of the CPU.
- */
-
-static int ks_check_endian(struct ks_net *ks)
-{
-	u16 cider;
-
-	/*
-	 * Read CIDER register first, however read it the "wrong" way around.
-	 * If the endian strap on the KS8851-16MLL in incorrect and the chip
-	 * is operating in different endianness than the CPU, then the meaning
-	 * of BE[3:0] byte-enable bits is also swapped such that:
-	 *    BE[3,2,1,0] becomes BE[1,0,3,2]
-	 *
-	 * Luckily for us, the byte-enable bits are the top four MSbits of
-	 * the address register and the CIDER register is at offset 0xc0.
-	 * Hence, by reading address 0xc0c0, which is not impacted by endian
-	 * swapping, we assert either BE[3:2] or BE[1:0] while reading the
-	 * CIDER register.
-	 *
-	 * If the bus configuration is correct, reading 0xc0c0 asserts
-	 * BE[3:2] and this read returns 0x0000, because to read register
-	 * with bottom two LSbits of address set to 0, BE[1:0] must be
-	 * asserted.
-	 *
-	 * If the bus configuration is NOT correct, reading 0xc0c0 asserts
-	 * BE[1:0] and this read returns non-zero 0x8872 value.
-	 */
-	iowrite16(BE3 | BE2 | KS_CIDER, ks->hw_addr_cmd);
-	cider = ioread16(ks->hw_addr);
-	if (!cider)
-		return 0;
-
-	netdev_err(ks->netdev, "incorrect EESK endian strap setting\n");
-
-	return -EINVAL;
-}
-
-/**
- * ks_rdreg16 - read 16 bit register from device
- * @ks	  : The chip information
- * @offset: The register address
- *
- * Read a 16bit register from the chip, returning the result
- */
-
-static u16 ks_rdreg16(struct ks_net *ks, int offset)
-{
-	ks->cmd_reg_cache = (u16)offset | ((BE1 | BE0) << (offset & 0x02));
-	iowrite16(ks->cmd_reg_cache, ks->hw_addr_cmd);
-	return ioread16(ks->hw_addr);
-}
-
-/**
- * ks_wrreg16 - write 16bit register value to chip
- * @ks: The chip information
- * @offset: The register address
- * @value: The value to write
- *
- */
-
-static void ks_wrreg16(struct ks_net *ks, int offset, u16 value)
-{
-	ks->cmd_reg_cache = (u16)offset | ((BE1 | BE0) << (offset & 0x02));
-	iowrite16(ks->cmd_reg_cache, ks->hw_addr_cmd);
-	iowrite16(value, ks->hw_addr);
-}
-
-/**
- * ks_inblk - read a block of data from QMU. This is called after sudo DMA mode enabled.
- * @ks: The chip state
- * @wptr: buffer address to save data
- * @len: length in byte to read
- *
- */
-static inline void ks_inblk(struct ks_net *ks, u16 *wptr, u32 len)
-{
-	len >>= 1;
-	while (len--)
-		*wptr++ = (u16)ioread16(ks->hw_addr);
-}
-
-/**
- * ks_outblk - write data to QMU. This is called after sudo DMA mode enabled.
- * @ks: The chip information
- * @wptr: buffer address
- * @len: length in byte to write
- *
- */
-static inline void ks_outblk(struct ks_net *ks, u16 *wptr, u32 len)
-{
-	len >>= 1;
-	while (len--)
-		iowrite16(*wptr++, ks->hw_addr);
-}
-
-static void ks_disable_int(struct ks_net *ks)
-{
-	ks_wrreg16(ks, KS_IER, 0x0000);
-}  /* ks_disable_int */
-
-static void ks_enable_int(struct ks_net *ks)
-{
-	ks_wrreg16(ks, KS_IER, ks->rc_ier);
-}  /* ks_enable_int */
-
-/**
- * ks_tx_fifo_space - return the available hardware buffer size.
- * @ks: The chip information
- *
- */
-static inline u16 ks_tx_fifo_space(struct ks_net *ks)
-{
-	return ks_rdreg16(ks, KS_TXMIR) & 0x1fff;
-}
-
-/**
- * ks_save_cmd_reg - save the command register from the cache.
- * @ks: The chip information
- *
- */
-static inline void ks_save_cmd_reg(struct ks_net *ks)
-{
-	/*ks8851 MLL has a bug to read back the command register.
-	* So rely on software to save the content of command register.
-	*/
-	ks->cmd_reg_cache_int = ks->cmd_reg_cache;
-}
-
-/**
- * ks_restore_cmd_reg - restore the command register from the cache and
- * 	write to hardware register.
- * @ks: The chip information
- *
- */
-static inline void ks_restore_cmd_reg(struct ks_net *ks)
-{
-	ks->cmd_reg_cache = ks->cmd_reg_cache_int;
-	iowrite16(ks->cmd_reg_cache, ks->hw_addr_cmd);
-}
-
-/**
- * ks_set_powermode - set power mode of the device
- * @ks: The chip information
- * @pwrmode: The power mode value to write to KS_PMECR.
- *
- * Change the power mode of the chip.
- */
-static void ks_set_powermode(struct ks_net *ks, unsigned pwrmode)
-{
-	unsigned pmecr;
-
-	netif_dbg(ks, hw, ks->netdev, "setting power mode %d\n", pwrmode);
-
-	ks_rdreg16(ks, KS_GRR);
-	pmecr = ks_rdreg16(ks, KS_PMECR);
-	pmecr &= ~PMECR_PM_MASK;
-	pmecr |= pwrmode;
-
-	ks_wrreg16(ks, KS_PMECR, pmecr);
-}
-
-/**
- * ks_read_config - read chip configuration of bus width.
- * @ks: The chip information
- *
- */
-static void ks_read_config(struct ks_net *ks)
-{
-	u16 reg_data = 0;
-
-	/* Regardless of bus width, 8 bit read should always work.*/
-	reg_data = ks_rdreg16(ks, KS_CCR);
-
-	/* addr/data bus are multiplexed */
-	ks->sharedbus = (reg_data & CCR_SHARED) == CCR_SHARED;
-
-	/* There are garbage data when reading data from QMU,
-	depending on bus-width.
-	*/
-
-	if (reg_data & CCR_8BIT) {
-		ks->bus_width = ENUM_BUS_8BIT;
-		ks->extra_byte = 1;
-	} else if (reg_data & CCR_16BIT) {
-		ks->bus_width = ENUM_BUS_16BIT;
-		ks->extra_byte = 2;
-	} else {
-		ks->bus_width = ENUM_BUS_32BIT;
-		ks->extra_byte = 4;
-	}
-}
-
-/**
- * ks_soft_reset - issue one of the soft reset to the device
- * @ks: The device state.
- * @op: The bit(s) to set in the GRR
- *
- * Issue the relevant soft-reset command to the device's GRR register
- * specified by @op.
- *
- * Note, the delays are in there as a caution to ensure that the reset
- * has time to take effect and then complete. Since the datasheet does
- * not currently specify the exact sequence, we have chosen something
- * that seems to work with our device.
- */
-static void ks_soft_reset(struct ks_net *ks, unsigned op)
-{
-	/* Disable interrupt first */
-	ks_wrreg16(ks, KS_IER, 0x0000);
-	ks_wrreg16(ks, KS_GRR, op);
-	mdelay(10);	/* wait a short time to effect reset */
-	ks_wrreg16(ks, KS_GRR, 0);
-	mdelay(1);	/* wait for condition to clear */
-}
-
-
-static void ks_enable_qmu(struct ks_net *ks)
-{
-	u16 w;
-
-	w = ks_rdreg16(ks, KS_TXCR);
-	/* Enables QMU Transmit (TXCR). */
-	ks_wrreg16(ks, KS_TXCR, w | TXCR_TXE);
-
-	/*
-	 * RX Frame Count Threshold Enable and Auto-Dequeue RXQ Frame
-	 * Enable
-	 */
-
-	w = ks_rdreg16(ks, KS_RXQCR);
-	ks_wrreg16(ks, KS_RXQCR, w | RXQCR_RXFCTE);
-
-	/* Enables QMU Receive (RXCR1). */
-	w = ks_rdreg16(ks, KS_RXCR1);
-	ks_wrreg16(ks, KS_RXCR1, w | RXCR1_RXE);
-	ks->enabled = true;
-}  /* ks_enable_qmu */
-
-static void ks_disable_qmu(struct ks_net *ks)
-{
-	u16	w;
-
-	w = ks_rdreg16(ks, KS_TXCR);
-
-	/* Disables QMU Transmit (TXCR). */
-	w  &= ~TXCR_TXE;
-	ks_wrreg16(ks, KS_TXCR, w);
-
-	/* Disables QMU Receive (RXCR1). */
-	w = ks_rdreg16(ks, KS_RXCR1);
-	w &= ~RXCR1_RXE ;
-	ks_wrreg16(ks, KS_RXCR1, w);
-
-	ks->enabled = false;
-
-}  /* ks_disable_qmu */
-
-/**
- * ks_read_qmu - read 1 pkt data from the QMU.
- * @ks: The chip information
- * @buf: buffer address to save 1 pkt
- * @len: Pkt length
- * Here is the sequence to read 1 pkt:
- *	1. set sudo DMA mode
- *	2. read prepend data
- *	3. read pkt data
- *	4. reset sudo DMA Mode
- */
-static inline void ks_read_qmu(struct ks_net *ks, u16 *buf, u32 len)
-{
-	u32 r =  ks->extra_byte & 0x1 ;
-	u32 w = ks->extra_byte - r;
-
-	/* 1. set sudo DMA mode */
-	ks_wrreg16(ks, KS_RXFDPR, RXFDPR_RXFPAI);
-	ks_wrreg16(ks, KS_RXQCR, ks->rc_rxqcr | RXQCR_SDA);
-
-	/* 2. read prepend data */
-	/**
-	 * read 4 + extra bytes and discard them.
-	 * extra bytes for dummy, 2 for status, 2 for len
-	 */
-
-	/* use likely(r) for 8 bit access for performance */
-	if (unlikely(r))
-		ioread8(ks->hw_addr);
-	ks_inblk(ks, buf, w + 2 + 2);
-
-	/* 3. read pkt data */
-	ks_inblk(ks, buf, ALIGN(len, 4));
-
-	/* 4. reset sudo DMA Mode */
-	ks_wrreg16(ks, KS_RXQCR, ks->rc_rxqcr);
-}
-
-/**
- * ks_rcv - read multiple pkts data from the QMU.
- * @ks: The chip information
- * @netdev: The network device being opened.
- *
- * Read all of header information before reading pkt content.
- * It is not allowed only port of pkts in QMU after issuing
- * interrupt ack.
- */
-static void ks_rcv(struct ks_net *ks, struct net_device *netdev)
-{
-	u32	i;
-	struct type_frame_head *frame_hdr = ks->frame_head_info;
-	struct sk_buff *skb;
-
-	ks->frame_cnt = ks_rdreg16(ks, KS_RXFCTR) >> 8;
-
-	/* read all header information */
-	for (i = 0; i < ks->frame_cnt; i++) {
-		/* Checking Received packet status */
-		frame_hdr->sts = ks_rdreg16(ks, KS_RXFHSR);
-		/* Get packet len from hardware */
-		frame_hdr->len = ks_rdreg16(ks, KS_RXFHBCR);
-		frame_hdr++;
-	}
-
-	frame_hdr = ks->frame_head_info;
-	while (ks->frame_cnt--) {
-		if (unlikely(!(frame_hdr->sts & RXFSHR_RXFV) ||
-			     frame_hdr->len >= RX_BUF_SIZE ||
-			     frame_hdr->len <= 0)) {
-
-			/* discard an invalid packet */
-			ks_wrreg16(ks, KS_RXQCR, (ks->rc_rxqcr | RXQCR_RRXEF));
-			netdev->stats.rx_dropped++;
-			if (!(frame_hdr->sts & RXFSHR_RXFV))
-				netdev->stats.rx_frame_errors++;
-			else
-				netdev->stats.rx_length_errors++;
-			frame_hdr++;
-			continue;
-		}
-
-		skb = netdev_alloc_skb(netdev, frame_hdr->len + 16);
-		if (likely(skb)) {
-			skb_reserve(skb, 2);
-			/* read data block including CRC 4 bytes */
-			ks_read_qmu(ks, (u16 *)skb->data, frame_hdr->len);
-			skb_put(skb, frame_hdr->len - 4);
-			skb->protocol = eth_type_trans(skb, netdev);
-			netif_rx(skb);
-			/* exclude CRC size */
-			netdev->stats.rx_bytes += frame_hdr->len - 4;
-			netdev->stats.rx_packets++;
-		} else {
-			ks_wrreg16(ks, KS_RXQCR, (ks->rc_rxqcr | RXQCR_RRXEF));
-			netdev->stats.rx_dropped++;
-		}
-		frame_hdr++;
-	}
-}
-
-/**
- * ks_update_link_status - link status update.
- * @netdev: The network device being opened.
- * @ks: The chip information
- *
- */
-
-static void ks_update_link_status(struct net_device *netdev, struct ks_net *ks)
-{
-	/* check the status of the link */
-	u32 link_up_status;
-	if (ks_rdreg16(ks, KS_P1SR) & P1SR_LINK_GOOD) {
-		netif_carrier_on(netdev);
-		link_up_status = true;
-	} else {
-		netif_carrier_off(netdev);
-		link_up_status = false;
-	}
-	netif_dbg(ks, link, ks->netdev,
-		  "%s: %s\n", __func__, link_up_status ? "UP" : "DOWN");
-}
-
-/**
- * ks_irq - device interrupt handler
- * @irq: Interrupt number passed from the IRQ handler.
- * @pw: The private word passed to register_irq(), our struct ks_net.
- *
- * This is the handler invoked to find out what happened
- *
- * Read the interrupt status, work out what needs to be done and then clear
- * any of the interrupts that are not needed.
- */
-
-static irqreturn_t ks_irq(int irq, void *pw)
-{
-	struct net_device *netdev = pw;
-	struct ks_net *ks = netdev_priv(netdev);
-	unsigned long flags;
-	u16 status;
-
-	spin_lock_irqsave(&ks->statelock, flags);
-	/*this should be the first in IRQ handler */
-	ks_save_cmd_reg(ks);
-
-	status = ks_rdreg16(ks, KS_ISR);
-	if (unlikely(!status)) {
-		ks_restore_cmd_reg(ks);
-		spin_unlock_irqrestore(&ks->statelock, flags);
-		return IRQ_NONE;
-	}
-
-	ks_wrreg16(ks, KS_ISR, status);
-
-	if (likely(status & IRQ_RXI))
-		ks_rcv(ks, netdev);
-
-	if (unlikely(status & IRQ_LCI))
-		ks_update_link_status(netdev, ks);
-
-	if (unlikely(status & IRQ_TXI))
-		netif_wake_queue(netdev);
-
-	if (unlikely(status & IRQ_LDI)) {
-
-		u16 pmecr = ks_rdreg16(ks, KS_PMECR);
-		pmecr &= ~PMECR_WKEVT_MASK;
-		ks_wrreg16(ks, KS_PMECR, pmecr | PMECR_WKEVT_LINK);
-	}
-
-	if (unlikely(status & IRQ_RXOI))
-		ks->netdev->stats.rx_over_errors++;
-	/* this should be the last in IRQ handler*/
-	ks_restore_cmd_reg(ks);
-	spin_unlock_irqrestore(&ks->statelock, flags);
-	return IRQ_HANDLED;
-}
-
-
-/**
- * ks_net_open - open network device
- * @netdev: The network device being opened.
- *
- * Called when the network device is marked active, such as a user executing
- * 'ifconfig up' on the device.
- */
-static int ks_net_open(struct net_device *netdev)
-{
-	struct ks_net *ks = netdev_priv(netdev);
-	int err;
-
-#define	KS_INT_FLAGS	IRQF_TRIGGER_LOW
-	/* lock the card, even if we may not actually do anything
-	 * else at the moment.
-	 */
-
-	netif_dbg(ks, ifup, ks->netdev, "%s - entry\n", __func__);
-
-	/* reset the HW */
-	err = request_irq(netdev->irq, ks_irq, KS_INT_FLAGS, DRV_NAME, netdev);
-
-	if (err) {
-		pr_err("Failed to request IRQ: %d: %d\n", netdev->irq, err);
-		return err;
-	}
-
-	/* wake up powermode to normal mode */
-	ks_set_powermode(ks, PMECR_PM_NORMAL);
-	mdelay(1);	/* wait for normal mode to take effect */
-
-	ks_wrreg16(ks, KS_ISR, 0xffff);
-	ks_enable_int(ks);
-	ks_enable_qmu(ks);
-	netif_start_queue(ks->netdev);
-
-	netif_dbg(ks, ifup, ks->netdev, "network device up\n");
-
-	return 0;
-}
-
-/**
- * ks_net_stop - close network device
- * @netdev: The device being closed.
- *
- * Called to close down a network device which has been active. Cancell any
- * work, shutdown the RX and TX process and then place the chip into a low
- * power state whilst it is not being used.
- */
-static int ks_net_stop(struct net_device *netdev)
-{
-	struct ks_net *ks = netdev_priv(netdev);
-
-	netif_info(ks, ifdown, netdev, "shutting down\n");
-
-	netif_stop_queue(netdev);
-
-	mutex_lock(&ks->lock);
-
-	/* turn off the IRQs and ack any outstanding */
-	ks_wrreg16(ks, KS_IER, 0x0000);
-	ks_wrreg16(ks, KS_ISR, 0xffff);
-
-	/* shutdown RX/TX QMU */
-	ks_disable_qmu(ks);
-	ks_disable_int(ks);
-
-	/* set powermode to soft power down to save power */
-	ks_set_powermode(ks, PMECR_PM_SOFTDOWN);
-	free_irq(netdev->irq, netdev);
-	mutex_unlock(&ks->lock);
-	return 0;
-}
-
-
-/**
- * ks_write_qmu - write 1 pkt data to the QMU.
- * @ks: The chip information
- * @pdata: buffer address to save 1 pkt
- * @len: Pkt length in byte
- * Here is the sequence to write 1 pkt:
- *	1. set sudo DMA mode
- *	2. write status/length
- *	3. write pkt data
- *	4. reset sudo DMA Mode
- *	5. reset sudo DMA mode
- *	6. Wait until pkt is out
- */
-static void ks_write_qmu(struct ks_net *ks, u8 *pdata, u16 len)
-{
-	/* start header at txb[0] to align txw entries */
-	ks->txh.txw[0] = 0;
-	ks->txh.txw[1] = cpu_to_le16(len);
-
-	/* 1. set sudo-DMA mode */
-	ks_wrreg16(ks, KS_RXQCR, ks->rc_rxqcr | RXQCR_SDA);
-	/* 2. write status/lenth info */
-	ks_outblk(ks, ks->txh.txw, 4);
-	/* 3. write pkt data */
-	ks_outblk(ks, (u16 *)pdata, ALIGN(len, 4));
-	/* 4. reset sudo-DMA mode */
-	ks_wrreg16(ks, KS_RXQCR, ks->rc_rxqcr);
-	/* 5. Enqueue Tx(move the pkt from TX buffer into TXQ) */
-	ks_wrreg16(ks, KS_TXQCR, TXQCR_METFE);
-	/* 6. wait until TXQCR_METFE is auto-cleared */
-	while (ks_rdreg16(ks, KS_TXQCR) & TXQCR_METFE)
-		;
-}
-
-/**
- * ks_start_xmit - transmit packet
- * @skb		: The buffer to transmit
- * @netdev	: The device used to transmit the packet.
- *
- * Called by the network layer to transmit the @skb.
- * spin_lock_irqsave is required because tx and rx should be mutual exclusive.
- * So while tx is in-progress, prevent IRQ interrupt from happenning.
- */
-static netdev_tx_t ks_start_xmit(struct sk_buff *skb, struct net_device *netdev)
-{
-	netdev_tx_t retv = NETDEV_TX_OK;
-	struct ks_net *ks = netdev_priv(netdev);
-	unsigned long flags;
-
-	spin_lock_irqsave(&ks->statelock, flags);
-
-	/* Extra space are required:
-	*  4 byte for alignment, 4 for status/length, 4 for CRC
-	*/
-
-	if (likely(ks_tx_fifo_space(ks) >= skb->len + 12)) {
-		ks_write_qmu(ks, skb->data, skb->len);
-		/* add tx statistics */
-		netdev->stats.tx_bytes += skb->len;
-		netdev->stats.tx_packets++;
-		dev_kfree_skb(skb);
-	} else
-		retv = NETDEV_TX_BUSY;
-	spin_unlock_irqrestore(&ks->statelock, flags);
-	return retv;
-}
-
-/**
- * ks_start_rx - ready to serve pkts
- * @ks		: The chip information
- *
- */
-static void ks_start_rx(struct ks_net *ks)
-{
-	u16 cntl;
-
-	/* Enables QMU Receive (RXCR1). */
-	cntl = ks_rdreg16(ks, KS_RXCR1);
-	cntl |= RXCR1_RXE ;
-	ks_wrreg16(ks, KS_RXCR1, cntl);
-}  /* ks_start_rx */
-
-/**
- * ks_stop_rx - stop to serve pkts
- * @ks		: The chip information
- *
- */
-static void ks_stop_rx(struct ks_net *ks)
-{
-	u16 cntl;
-
-	/* Disables QMU Receive (RXCR1). */
-	cntl = ks_rdreg16(ks, KS_RXCR1);
-	cntl &= ~RXCR1_RXE ;
-	ks_wrreg16(ks, KS_RXCR1, cntl);
-
-}  /* ks_stop_rx */
-
-static unsigned long const ethernet_polynomial = CRC32_POLY_BE;
-
-static unsigned long ether_gen_crc(int length, u8 *data)
-{
-	long crc = -1;
-	while (--length >= 0) {
-		u8 current_octet = *data++;
-		int bit;
-
-		for (bit = 0; bit < 8; bit++, current_octet >>= 1) {
-			crc = (crc << 1) ^
-				((crc < 0) ^ (current_octet & 1) ?
-			ethernet_polynomial : 0);
-		}
-	}
-	return (unsigned long)crc;
-}  /* ether_gen_crc */
-
-/**
-* ks_set_grpaddr - set multicast information
-* @ks : The chip information
-*/
-
-static void ks_set_grpaddr(struct ks_net *ks)
-{
-	u8	i;
-	u32	index, position, value;
-
-	memset(ks->mcast_bits, 0, sizeof(u8) * HW_MCAST_SIZE);
-
-	for (i = 0; i < ks->mcast_lst_size; i++) {
-		position = (ether_gen_crc(6, ks->mcast_lst[i]) >> 26) & 0x3f;
-		index = position >> 3;
-		value = 1 << (position & 7);
-		ks->mcast_bits[index] |= (u8)value;
-	}
-
-	for (i  = 0; i < HW_MCAST_SIZE; i++) {
-		if (i & 1) {
-			ks_wrreg16(ks, (u16)((KS_MAHTR0 + i) & ~1),
-				(ks->mcast_bits[i] << 8) |
-				ks->mcast_bits[i - 1]);
-		}
-	}
-}  /* ks_set_grpaddr */
-
-/**
-* ks_clear_mcast - clear multicast information
-*
-* @ks : The chip information
-* This routine removes all mcast addresses set in the hardware.
-*/
-
-static void ks_clear_mcast(struct ks_net *ks)
-{
-	u16	i, mcast_size;
-	for (i = 0; i < HW_MCAST_SIZE; i++)
-		ks->mcast_bits[i] = 0;
-
-	mcast_size = HW_MCAST_SIZE >> 2;
-	for (i = 0; i < mcast_size; i++)
-		ks_wrreg16(ks, KS_MAHTR0 + (2*i), 0);
-}
-
-static void ks_set_promis(struct ks_net *ks, u16 promiscuous_mode)
-{
-	u16		cntl;
-	ks->promiscuous = promiscuous_mode;
-	ks_stop_rx(ks);  /* Stop receiving for reconfiguration */
-	cntl = ks_rdreg16(ks, KS_RXCR1);
-
-	cntl &= ~RXCR1_FILTER_MASK;
-	if (promiscuous_mode)
-		/* Enable Promiscuous mode */
-		cntl |= RXCR1_RXAE | RXCR1_RXINVF;
-	else
-		/* Disable Promiscuous mode (default normal mode) */
-		cntl |= RXCR1_RXPAFMA;
-
-	ks_wrreg16(ks, KS_RXCR1, cntl);
-
-	if (ks->enabled)
-		ks_start_rx(ks);
-
-}  /* ks_set_promis */
-
-static void ks_set_mcast(struct ks_net *ks, u16 mcast)
-{
-	u16	cntl;
-
-	ks->all_mcast = mcast;
-	ks_stop_rx(ks);  /* Stop receiving for reconfiguration */
-	cntl = ks_rdreg16(ks, KS_RXCR1);
-	cntl &= ~RXCR1_FILTER_MASK;
-	if (mcast)
-		/* Enable "Perfect with Multicast address passed mode" */
-		cntl |= (RXCR1_RXAE | RXCR1_RXMAFMA | RXCR1_RXPAFMA);
-	else
-		/**
-		 * Disable "Perfect with Multicast address passed
-		 * mode" (normal mode).
-		 */
-		cntl |= RXCR1_RXPAFMA;
-
-	ks_wrreg16(ks, KS_RXCR1, cntl);
-
-	if (ks->enabled)
-		ks_start_rx(ks);
-}  /* ks_set_mcast */
-
-static void ks_set_rx_mode(struct net_device *netdev)
-{
-	struct ks_net *ks = netdev_priv(netdev);
-	struct netdev_hw_addr *ha;
-
-	/* Turn on/off promiscuous mode. */
-	if ((netdev->flags & IFF_PROMISC) == IFF_PROMISC)
-		ks_set_promis(ks,
-			(u16)((netdev->flags & IFF_PROMISC) == IFF_PROMISC));
-	/* Turn on/off all mcast mode. */
-	else if ((netdev->flags & IFF_ALLMULTI) == IFF_ALLMULTI)
-		ks_set_mcast(ks,
-			(u16)((netdev->flags & IFF_ALLMULTI) == IFF_ALLMULTI));
-	else
-		ks_set_promis(ks, false);
-
-	if ((netdev->flags & IFF_MULTICAST) && netdev_mc_count(netdev)) {
-		if (netdev_mc_count(netdev) <= MAX_MCAST_LST) {
-			int i = 0;
-
-			netdev_for_each_mc_addr(ha, netdev) {
-				if (i >= MAX_MCAST_LST)
-					break;
-				memcpy(ks->mcast_lst[i++], ha->addr, ETH_ALEN);
-			}
-			ks->mcast_lst_size = (u8)i;
-			ks_set_grpaddr(ks);
-		} else {
-			/**
-			 * List too big to support so
-			 * turn on all mcast mode.
-			 */
-			ks->mcast_lst_size = MAX_MCAST_LST;
-			ks_set_mcast(ks, true);
-		}
-	} else {
-		ks->mcast_lst_size = 0;
-		ks_clear_mcast(ks);
-	}
-} /* ks_set_rx_mode */
-
-static void ks_set_mac(struct ks_net *ks, u8 *data)
-{
-	u16 *pw = (u16 *)data;
-	u16 w, u;
-
-	ks_stop_rx(ks);  /* Stop receiving for reconfiguration */
-
-	u = *pw++;
-	w = ((u & 0xFF) << 8) | ((u >> 8) & 0xFF);
-	ks_wrreg16(ks, KS_MARH, w);
-
-	u = *pw++;
-	w = ((u & 0xFF) << 8) | ((u >> 8) & 0xFF);
-	ks_wrreg16(ks, KS_MARM, w);
-
-	u = *pw;
-	w = ((u & 0xFF) << 8) | ((u >> 8) & 0xFF);
-	ks_wrreg16(ks, KS_MARL, w);
-
-	memcpy(ks->mac_addr, data, ETH_ALEN);
-
-	if (ks->enabled)
-		ks_start_rx(ks);
-}
-
-static int ks_set_mac_address(struct net_device *netdev, void *paddr)
-{
-	struct ks_net *ks = netdev_priv(netdev);
-	struct sockaddr *addr = paddr;
-	u8 *da;
-
-	memcpy(netdev->dev_addr, addr->sa_data, netdev->addr_len);
-
-	da = (u8 *)netdev->dev_addr;
-
-	ks_set_mac(ks, da);
-	return 0;
-}
-
-static int ks_net_ioctl(struct net_device *netdev, struct ifreq *req, int cmd)
-{
-	struct ks_net *ks = netdev_priv(netdev);
-
-	if (!netif_running(netdev))
-		return -EINVAL;
-
-	return generic_mii_ioctl(&ks->mii, if_mii(req), cmd, NULL);
-}
-
-static const struct net_device_ops ks_netdev_ops = {
-	.ndo_open		= ks_net_open,
-	.ndo_stop		= ks_net_stop,
-	.ndo_do_ioctl		= ks_net_ioctl,
-	.ndo_start_xmit		= ks_start_xmit,
-	.ndo_set_mac_address	= ks_set_mac_address,
-	.ndo_set_rx_mode	= ks_set_rx_mode,
-	.ndo_validate_addr	= eth_validate_addr,
-};
-
-/* ethtool support */
-
-static void ks_get_drvinfo(struct net_device *netdev,
-			       struct ethtool_drvinfo *di)
-{
-	strlcpy(di->driver, DRV_NAME, sizeof(di->driver));
-	strlcpy(di->version, "1.00", sizeof(di->version));
-	strlcpy(di->bus_info, dev_name(netdev->dev.parent),
-		sizeof(di->bus_info));
-}
-
-static u32 ks_get_msglevel(struct net_device *netdev)
-{
-	struct ks_net *ks = netdev_priv(netdev);
-	return ks->msg_enable;
-}
-
-static void ks_set_msglevel(struct net_device *netdev, u32 to)
-{
-	struct ks_net *ks = netdev_priv(netdev);
-	ks->msg_enable = to;
-}
-
-static int ks_get_link_ksettings(struct net_device *netdev,
-				 struct ethtool_link_ksettings *cmd)
-{
-	struct ks_net *ks = netdev_priv(netdev);
-
-	mii_ethtool_get_link_ksettings(&ks->mii, cmd);
-
-	return 0;
-}
-
-static int ks_set_link_ksettings(struct net_device *netdev,
-				 const struct ethtool_link_ksettings *cmd)
-{
-	struct ks_net *ks = netdev_priv(netdev);
-	return mii_ethtool_set_link_ksettings(&ks->mii, cmd);
-}
-
-static u32 ks_get_link(struct net_device *netdev)
-{
-	struct ks_net *ks = netdev_priv(netdev);
-	return mii_link_ok(&ks->mii);
-}
-
-static int ks_nway_reset(struct net_device *netdev)
-{
-	struct ks_net *ks = netdev_priv(netdev);
-	return mii_nway_restart(&ks->mii);
-}
-
-static const struct ethtool_ops ks_ethtool_ops = {
-	.get_drvinfo	= ks_get_drvinfo,
-	.get_msglevel	= ks_get_msglevel,
-	.set_msglevel	= ks_set_msglevel,
-	.get_link	= ks_get_link,
-	.nway_reset	= ks_nway_reset,
-	.get_link_ksettings = ks_get_link_ksettings,
-	.set_link_ksettings = ks_set_link_ksettings,
-};
-
-/* MII interface controls */
-
-/**
- * ks_phy_reg - convert MII register into a KS8851 register
- * @reg: MII register number.
- *
- * Return the KS8851 register number for the corresponding MII PHY register
- * if possible. Return zero if the MII register has no direct mapping to the
- * KS8851 register set.
- */
-static int ks_phy_reg(int reg)
-{
-	switch (reg) {
-	case MII_BMCR:
-		return KS_P1MBCR;
-	case MII_BMSR:
-		return KS_P1MBSR;
-	case MII_PHYSID1:
-		return KS_PHY1ILR;
-	case MII_PHYSID2:
-		return KS_PHY1IHR;
-	case MII_ADVERTISE:
-		return KS_P1ANAR;
-	case MII_LPA:
-		return KS_P1ANLPR;
-	}
-
-	return 0x0;
-}
-
-/**
- * ks_phy_read - MII interface PHY register read.
- * @netdev: The network device the PHY is on.
- * @phy_addr: Address of PHY (ignored as we only have one)
- * @reg: The register to read.
- *
- * This call reads data from the PHY register specified in @reg. Since the
- * device does not support all the MII registers, the non-existent values
- * are always returned as zero.
- *
- * We return zero for unsupported registers as the MII code does not check
- * the value returned for any error status, and simply returns it to the
- * caller. The mii-tool that the driver was tested with takes any -ve error
- * as real PHY capabilities, thus displaying incorrect data to the user.
- */
-static int ks_phy_read(struct net_device *netdev, int phy_addr, int reg)
-{
-	struct ks_net *ks = netdev_priv(netdev);
-	int ksreg;
-	int result;
-
-	ksreg = ks_phy_reg(reg);
-	if (!ksreg)
-		return 0x0;	/* no error return allowed, so use zero */
-
-	mutex_lock(&ks->lock);
-	result = ks_rdreg16(ks, ksreg);
-	mutex_unlock(&ks->lock);
-
-	return result;
-}
-
-static void ks_phy_write(struct net_device *netdev,
-			     int phy, int reg, int value)
-{
-	struct ks_net *ks = netdev_priv(netdev);
-	int ksreg;
-
-	ksreg = ks_phy_reg(reg);
-	if (ksreg) {
-		mutex_lock(&ks->lock);
-		ks_wrreg16(ks, ksreg, value);
-		mutex_unlock(&ks->lock);
-	}
-}
-
-/**
- * ks_read_selftest - read the selftest memory info.
- * @ks: The device state
- *
- * Read and check the TX/RX memory selftest information.
- */
-static int ks_read_selftest(struct ks_net *ks)
-{
-	unsigned both_done = MBIR_TXMBF | MBIR_RXMBF;
-	int ret = 0;
-	unsigned rd;
-
-	rd = ks_rdreg16(ks, KS_MBIR);
-
-	if ((rd & both_done) != both_done) {
-		netdev_warn(ks->netdev, "Memory selftest not finished\n");
-		return 0;
-	}
-
-	if (rd & MBIR_TXMBFA) {
-		netdev_err(ks->netdev, "TX memory selftest fails\n");
-		ret |= 1;
-	}
-
-	if (rd & MBIR_RXMBFA) {
-		netdev_err(ks->netdev, "RX memory selftest fails\n");
-		ret |= 2;
-	}
-
-	netdev_info(ks->netdev, "the selftest passes\n");
-	return ret;
-}
-
-static void ks_setup(struct ks_net *ks)
-{
-	u16	w;
-
-	/**
-	 * Configure QMU Transmit
-	 */
-
-	/* Setup Transmit Frame Data Pointer Auto-Increment (TXFDPR) */
-	ks_wrreg16(ks, KS_TXFDPR, TXFDPR_TXFPAI);
-
-	/* Setup Receive Frame Data Pointer Auto-Increment */
-	ks_wrreg16(ks, KS_RXFDPR, RXFDPR_RXFPAI);
-
-	/* Setup Receive Frame Threshold - 1 frame (RXFCTFC) */
-	ks_wrreg16(ks, KS_RXFCTR, 1 & RXFCTR_RXFCT_MASK);
-
-	/* Setup RxQ Command Control (RXQCR) */
-	ks->rc_rxqcr = RXQCR_CMD_CNTL;
-	ks_wrreg16(ks, KS_RXQCR, ks->rc_rxqcr);
-
-	/**
-	 * set the force mode to half duplex, default is full duplex
-	 *  because if the auto-negotiation fails, most switch uses
-	 *  half-duplex.
-	 */
-
-	w = ks_rdreg16(ks, KS_P1MBCR);
-	w &= ~BMCR_FULLDPLX;
-	ks_wrreg16(ks, KS_P1MBCR, w);
-
-	w = TXCR_TXFCE | TXCR_TXPE | TXCR_TXCRC | TXCR_TCGIP;
-	ks_wrreg16(ks, KS_TXCR, w);
-
-	w = RXCR1_RXFCE | RXCR1_RXBE | RXCR1_RXUE | RXCR1_RXME | RXCR1_RXIPFCC;
-
-	if (ks->promiscuous)         /* bPromiscuous */
-		w |= (RXCR1_RXAE | RXCR1_RXINVF);
-	else if (ks->all_mcast) /* Multicast address passed mode */
-		w |= (RXCR1_RXAE | RXCR1_RXMAFMA | RXCR1_RXPAFMA);
-	else                                   /* Normal mode */
-		w |= RXCR1_RXPAFMA;
-
-	ks_wrreg16(ks, KS_RXCR1, w);
-}  /*ks_setup */
-
-
-static void ks_setup_int(struct ks_net *ks)
-{
-	ks->rc_ier = 0x00;
-	/* Clear the interrupts status of the hardware. */
-	ks_wrreg16(ks, KS_ISR, 0xffff);
-
-	/* Enables the interrupts of the hardware. */
-	ks->rc_ier = (IRQ_LCI | IRQ_TXI | IRQ_RXI);
-}  /* ks_setup_int */
-
-static int ks_hw_init(struct ks_net *ks)
-{
-#define	MHEADER_SIZE	(sizeof(struct type_frame_head) * MAX_RECV_FRAMES)
-	ks->promiscuous = 0;
-	ks->all_mcast = 0;
-	ks->mcast_lst_size = 0;
-
-	ks->frame_head_info = devm_kmalloc(&ks->pdev->dev, MHEADER_SIZE,
-					   GFP_KERNEL);
-	if (!ks->frame_head_info)
-		return false;
-
-	ks_set_mac(ks, KS_DEFAULT_MAC_ADDRESS);
-	return true;
-}
-
-#if defined(CONFIG_OF)
-static const struct of_device_id ks8851_ml_dt_ids[] = {
-	{ .compatible = "micrel,ks8851-mll" },
-	{ /* sentinel */ }
-};
-MODULE_DEVICE_TABLE(of, ks8851_ml_dt_ids);
-#endif
-
-static int ks8851_probe(struct platform_device *pdev)
-{
-	int err;
-	struct net_device *netdev;
-	struct ks_net *ks;
-	u16 id, data;
-	const char *mac;
-
-	netdev = alloc_etherdev(sizeof(struct ks_net));
-	if (!netdev)
-		return -ENOMEM;
-
-	SET_NETDEV_DEV(netdev, &pdev->dev);
-
-	ks = netdev_priv(netdev);
-	ks->netdev = netdev;
-
-	ks->hw_addr = devm_platform_ioremap_resource(pdev, 0);
-	if (IS_ERR(ks->hw_addr)) {
-		err = PTR_ERR(ks->hw_addr);
-		goto err_free;
-	}
-
-	ks->hw_addr_cmd = devm_platform_ioremap_resource(pdev, 1);
-	if (IS_ERR(ks->hw_addr_cmd)) {
-		err = PTR_ERR(ks->hw_addr_cmd);
-		goto err_free;
-	}
-
-	err = ks_check_endian(ks);
-	if (err)
-		goto err_free;
-
-	netdev->irq = platform_get_irq(pdev, 0);
-
-	if ((int)netdev->irq < 0) {
-		err = netdev->irq;
-		goto err_free;
-	}
-
-	ks->pdev = pdev;
-
-	mutex_init(&ks->lock);
-	spin_lock_init(&ks->statelock);
-
-	netdev->netdev_ops = &ks_netdev_ops;
-	netdev->ethtool_ops = &ks_ethtool_ops;
-
-	/* setup mii state */
-	ks->mii.dev             = netdev;
-	ks->mii.phy_id          = 1,
-	ks->mii.phy_id_mask     = 1;
-	ks->mii.reg_num_mask    = 0xf;
-	ks->mii.mdio_read       = ks_phy_read;
-	ks->mii.mdio_write      = ks_phy_write;
-
-	netdev_info(netdev, "message enable is %d\n", msg_enable);
-	/* set the default message enable */
-	ks->msg_enable = netif_msg_init(msg_enable, (NETIF_MSG_DRV |
-						     NETIF_MSG_PROBE |
-						     NETIF_MSG_LINK));
-	ks_read_config(ks);
-
-	/* simple check for a valid chip being connected to the bus */
-	if ((ks_rdreg16(ks, KS_CIDER) & ~CIDER_REV_MASK) != CIDER_ID) {
-		netdev_err(netdev, "failed to read device ID\n");
-		err = -ENODEV;
-		goto err_free;
-	}
-
-	if (ks_read_selftest(ks)) {
-		netdev_err(netdev, "failed to read device ID\n");
-		err = -ENODEV;
-		goto err_free;
-	}
-
-	err = register_netdev(netdev);
-	if (err)
-		goto err_free;
-
-	platform_set_drvdata(pdev, netdev);
-
-	ks_soft_reset(ks, GRR_GSR);
-	ks_hw_init(ks);
-	ks_disable_qmu(ks);
-	ks_setup(ks);
-	ks_setup_int(ks);
-
-	data = ks_rdreg16(ks, KS_OBCR);
-	ks_wrreg16(ks, KS_OBCR, data | OBCR_ODS_16mA);
-
-	/* overwriting the default MAC address */
-	if (pdev->dev.of_node) {
-		mac = of_get_mac_address(pdev->dev.of_node);
-		if (!IS_ERR(mac))
-			ether_addr_copy(ks->mac_addr, mac);
-	} else {
-		struct ks8851_mll_platform_data *pdata;
-
-		pdata = dev_get_platdata(&pdev->dev);
-		if (!pdata) {
-			netdev_err(netdev, "No platform data\n");
-			err = -ENODEV;
-			goto err_pdata;
-		}
-		memcpy(ks->mac_addr, pdata->mac_addr, ETH_ALEN);
-	}
-	if (!is_valid_ether_addr(ks->mac_addr)) {
-		/* Use random MAC address if none passed */
-		eth_random_addr(ks->mac_addr);
-		netdev_info(netdev, "Using random mac address\n");
-	}
-	netdev_info(netdev, "Mac address is: %pM\n", ks->mac_addr);
-
-	memcpy(netdev->dev_addr, ks->mac_addr, ETH_ALEN);
-
-	ks_set_mac(ks, netdev->dev_addr);
-
-	id = ks_rdreg16(ks, KS_CIDER);
-
-	netdev_info(netdev, "Found chip, family: 0x%x, id: 0x%x, rev: 0x%x\n",
-		    (id >> 8) & 0xff, (id >> 4) & 0xf, (id >> 1) & 0x7);
-	return 0;
-
-err_pdata:
-	unregister_netdev(netdev);
-err_free:
-	free_netdev(netdev);
-	return err;
-}
-
-static int ks8851_remove(struct platform_device *pdev)
-{
-	struct net_device *netdev = platform_get_drvdata(pdev);
-
-	unregister_netdev(netdev);
-	free_netdev(netdev);
-	return 0;
-
-}
-
-static struct platform_driver ks8851_platform_driver = {
-	.driver = {
-		.name = DRV_NAME,
-		.of_match_table	= of_match_ptr(ks8851_ml_dt_ids),
-	},
-	.probe = ks8851_probe,
-	.remove = ks8851_remove,
-};
-
-module_platform_driver(ks8851_platform_driver);
-
-MODULE_DESCRIPTION("KS8851 MLL Network driver");
-MODULE_AUTHOR("David Choi <[email protected]>");
-MODULE_LICENSE("GPL");
-module_param_named(message, msg_enable, int, 0);
-MODULE_PARM_DESC(message, "Message verbosity level (0=none, 31=all)");
-
Signature · 1 line
2.25.1
#21 lore raw
In reply to Marek Vasut
Quoted 12 lines · Marek Vasut

The KS8851SNL/SNLI and KS8851-16MLL/MLLI/MLLU are very much the same pieces of silicon, except the former has an SPI interface, while the later has a parallel bus interface. Thus far, Linux has two separate drivers for each and they are diverging considerably.

This series unifies them into a single driver with small SPI and parallel bus specific parts. The approach here is to first separate out the SPI specific parts into a separate file, then add parallel bus accessors in another separate file and then finally remove the old parallel bus driver. The reason for replacing the old parallel bus driver is because the SPI bus driver is much higher quality.

Quoted 4 lines

NOTE: This series depends on "net: ks8851-ml: Fix IO operations, again"

NOTE: The performance regression on KS8851-16MLL is now fixes, the TX
throughput is back to ~75 Mbit/s , RX is still 50 Mbit/s .

I also managed to implement RX NAPI (see attached demo patch if you want to measure something on the SPI variant), but the RX performance didn't improve dramatically. It's been 52 Mbit/s, with the RX NAPI it is 56 Mbit/s, on the parallel variant.

[...]

From 9ef3977d7b3f73246ff0600022d23faaa32dfb75 Mon Sep 17 00:00:00 2001
From: Marek Vasut <[email protected]>
Date: Sun, 29 Mar 2020 18:09:34 +0200
Subject: [PATCH] RX-NAPI

---
 drivers/net/ethernet/micrel/ks8851.h        |   3 +
 drivers/net/ethernet/micrel/ks8851_common.c | 147 +++++++++++++++-----
 2 files changed, 115 insertions(+), 35 deletions(-)
drivers/net/ethernet/micrel/ks8851.h +3 -0
index 649436032405..8f4f6d14e28a 100644
@@ -394,6 +394,9 @@ struct ks8851_net {
 	struct regulator	*vdd_reg;
 	struct regulator	*vdd_io;
 	int			gpio;
+	struct napi_struct	napi;
+	bool			acked;
+	u16			rxfc;
 };
 
 int ks8851_probe_common(struct net_device *netdev, struct device *dev);
drivers/net/ethernet/micrel/ks8851_common.c +112 -35

220 lines not on this page · raw message

Signature · 1 line
2.25.1
#22 lore raw
In reply to Marek Vasut
Quoted 11 lines · Marek Vasut

The KS8851SNL/SNLI and KS8851-16MLL/MLLI/MLLU are very much the same pieces of silicon, except the former has an SPI interface, while the later has a parallel bus interface. Thus far, Linux has two separate drivers for each and they are diverging considerably.

This series unifies them into a single driver with small SPI and parallel bus specific parts. The approach here is to first separate out the SPI specific parts into a separate file, then add parallel bus accessors in another separate file and then finally remove the old parallel bus driver. The reason for replacing the old parallel bus driver is because the SPI bus driver is much higher quality.

Sorry for the delay Marek.

ks8851.ko (SPI variant) no longer compiles with this series. The attached 0001 patch fixes it.

Both drivers can only be compiled as modules with this series:
If only one of them is built-in, there's a linker error because of the module_param_named() for msg_enable.
If both are built-in, the symbol collisions you've mentioned occur.

It seems Kbuild can't support building a .o file with a different name than the corresponding .c file because of the implicit rules used everywhere. However, ks8851_common.c can be renamed to be a header file (a library of sorts) which is included by the two .c files. I've renamed ks8851_spi.c back to the original ks8851.c and ks8851_par.c back to ks8851_mll.c. The result is the attached 0002 patch. Compiles without any errors regardless if one or both drivers are built-in or modules.

I'll be back at the office this week and will conduct performance measurements with this version.

Thanks,

Lukas

From 5d6d0fc40c5e37bf1bea95bf11c598c61c2a9bc2 Mon Sep 17 00:00:00 2001
From: Lukas Wunner <[email protected]>
Date: Sun, 5 Apr 2020 13:08:50 +0200
Subject: [PATCH 1/2] net: ks8851: Unbreak the build of CONFIG_KS8851

drivers/net/ethernet/micrel/ks8851_spi.c: In function 'ks8851_start_xmit':
drivers/net/ethernet/micrel/ks8851_spi.c:360:20: error: implicit declaration of function 'calc_txlen' [-Werror=implicit-function-declaration]

  unsigned needed = calc_txlen(skb->len);
                    ^~~~~~~~~~
drivers/net/ethernet/micrel/ks8851_spi.c:377:19: error: 'struct ks8851_net' has no member named 'tx_work'
  schedule_work(&ks->tx_work);
                   ^~
---
 drivers/net/ethernet/micrel/ks8851_common.c | 12 ------------
 drivers/net/ethernet/micrel/ks8851_spi.c    | 15 ++++++++++++++-
 2 files changed, 14 insertions(+), 13 deletions(-)
drivers/net/ethernet/micrel/ks8851_common.c +0 -12
index b708c01f9f33..9878369efda2 100644
@@ -363,18 +363,6 @@ static irqreturn_t ks8851_irq(int irq, void *_ks)
 	return IRQ_HANDLED;
 }
 
-/**
- * calc_txlen - calculate size of message to send packet
- * @len: Length of data
- *
- * Returns the size of the TXFIFO message needed to send
- * this packet.
- */
-static inline unsigned calc_txlen(unsigned len)
-{
-	return ALIGN(len + 4, 4);
-}
-
 /**
  * ks8851_net_open - open network device
  * @dev: The network device being opened.
drivers/net/ethernet/micrel/ks8851_spi.c +14 -1
index e245ad0b82f4..d08a51c55236 100644
@@ -236,6 +236,18 @@ void ks8851_rdfifo(struct ks8851_net *ks, u8 *buff, unsigned int len)
 		netdev_err(ks->netdev, "%s: spi_sync() failed\n", __func__);
 }
 
+/**
+ * calc_txlen - calculate size of message to send packet
+ * @len: Length of data
+ *
+ * Returns the size of the TXFIFO message needed to send
+ * this packet.
+ */
+static inline unsigned calc_txlen(unsigned len)
+{
+	return ALIGN(len + 4, 4);
+}
+
 /**
  * ks8851_wrfifo - write packet to TX FIFO via SPI
  * @ks: The device state.
@@ -357,6 +369,7 @@ void ks8851_flush_tx_work(struct ks8851_net *ks)
 netdev_tx_t ks8851_start_xmit(struct sk_buff *skb, struct net_device *dev)
 {
 	struct ks8851_net *ks = netdev_priv(dev);
+	struct ks8851_net_spi *kss = to_ks8851_spi(ks);
 	unsigned needed = calc_txlen(skb->len);
 	netdev_tx_t ret = NETDEV_TX_OK;
 
@@ -374,7 +387,7 @@ netdev_tx_t ks8851_start_xmit(struct sk_buff *skb, struct net_device *dev)
 	}
 
 	spin_unlock(&ks->statelock);
-	schedule_work(&ks->tx_work);
+	schedule_work(&kss->tx_work);
 
 	return ret;
 }
Signature · 1 line
2.25.0

From 9b914fcf37d622f330a5997e0bc660c422bcc2f1 Mon Sep 17 00:00:00 2001
From: Lukas Wunner <[email protected]>
Date: Mon, 6 Apr 2020 04:54:14 +0200
Subject: [PATCH 2/2] net: ks8851: Unbreak the build of CONFIG_KS8851=y ||
CONFIG_KS8851_MLL=y

If one of ks8851 or ks8851_mll is built-in:

arm-linux-gnueabihf-ld: drivers/net/ethernet/micrel/ks8851_common.o:(__param+0x4): undefined reference to `__this_module'

If both are built-in:

arm-linux-gnueabihf-ld: drivers/net/ethernet/micrel/ks8851_par.o: in function `ks8851_lock':
drivers/net/ethernet/micrel/ks8851_par.c:68: multiple definition of `ks8851_lock'; drivers/net/ethernet/micrel/ks8851_spi.o:drivers/net/ethernet/micrel/ks8851_spi.c:80: first defined here
arm-linux-gnueabihf-ld: drivers/net/ethernet/micrel/ks8851_par.o: in function `ks8851_unlock':
drivers/net/ethernet/micrel/ks8851_par.c:82: multiple definition of `ks8851_unlock'; drivers/net/ethernet/micrel/ks8851_spi.o:drivers/net/ethernet/micrel/ks8851_spi.c:94: first defined here
arm-linux-gnueabihf-ld: drivers/net/ethernet/micrel/ks8851_par.o: in function `ks8851_wrreg16':
drivers/net/ethernet/micrel/ks8851_par.c:141: multiple definition of `ks8851_wrreg16'; drivers/net/ethernet/micrel/ks8851_spi.o:drivers/net/ethernet/micrel/ks8851_spi.c:116: first defined here
arm-linux-gnueabihf-ld: drivers/net/ethernet/micrel/ks8851_par.o: in function `ks8851_rdreg16':
drivers/net/ethernet/micrel/ks8851_par.c:157: multiple definition of `ks8851_rdreg16'; drivers/net/ethernet/micrel/ks8851_spi.o:drivers/net/ethernet/micrel/ks8851_spi.c:195: first defined here
arm-linux-gnueabihf-ld: drivers/net/ethernet/micrel/ks8851_par.o: in function `ks8851_rdfifo':
drivers/net/ethernet/micrel/ks8851_par.c:175: multiple definition of `ks8851_rdfifo'; drivers/net/ethernet/micrel/ks8851_spi.o:drivers/net/ethernet/micrel/ks8851_spi.c:212: first defined here
arm-linux-gnueabihf-ld: drivers/net/ethernet/micrel/ks8851_par.o: in function `ks8851_wrfifo':
drivers/net/ethernet/micrel/ks8851_par.c:196: multiple definition of `ks8851_wrfifo'; drivers/net/ethernet/micrel/ks8851_spi.o:drivers/net/ethernet/micrel/ks8851_spi.c:263: first defined here
arm-linux-gnueabihf-ld: drivers/net/ethernet/micrel/ks8851_par.o: in function `ks8851_rx_skb':
drivers/net/ethernet/micrel/ks8851_par.c:221: multiple definition of `ks8851_rx_skb'; drivers/net/ethernet/micrel/ks8851_spi.o:drivers/net/ethernet/micrel/ks8851_spi.c:303: first defined here
arm-linux-gnueabihf-ld: drivers/net/ethernet/micrel/ks8851_par.o: in function `ks8851_flush_tx_work':
drivers/net/ethernet/micrel/ks8851_par.c:230: multiple definition of `ks8851_flush_tx_work'; drivers/net/ethernet/micrel/ks8851_spi.o:drivers/net/ethernet/micrel/ks8851_spi.c:350: first defined here
arm-linux-gnueabihf-ld: drivers/net/ethernet/micrel/ks8851_par.o: in function `ks8851_start_xmit':
drivers/net/ethernet/micrel/ks8851_par.c:247: multiple definition of `ks8851_start_xmit'; drivers/net/ethernet/micrel/ks8851_spi.o:drivers/net/ethernet/micrel/ks8851_spi.c:370: first defined here

---
 drivers/net/ethernet/micrel/Makefile          |  2 -
 .../micrel/{ks8851_spi.c => ks8851.c}         | 42 ++++++-------------
 drivers/net/ethernet/micrel/ks8851.h          | 27 ++++--------
 .../{ks8851_common.c => ks8851_common.h}      | 19 ++-------
 .../micrel/{ks8851_par.c => ks8851_mll.c}     | 41 ++++++------------
 5 files changed, 38 insertions(+), 93 deletions(-)
 rename drivers/net/ethernet/micrel/{ks8851_spi.c => ks8851.c} (91%)
 rename drivers/net/ethernet/micrel/{ks8851_common.c => ks8851_common.h} (98%)
 rename drivers/net/ethernet/micrel/{ks8851_par.c => ks8851_mll.c} (89%)
drivers/net/ethernet/micrel/Makefile +0 -2
index 5cc00d22c708..6d8ac5527aef 100644
@@ -5,7 +5,5 @@
 
 obj-$(CONFIG_KS8842) += ks8842.o
 obj-$(CONFIG_KS8851) += ks8851.o
-ks8851-objs = ks8851_common.o ks8851_spi.o
 obj-$(CONFIG_KS8851_MLL) += ks8851_mll.o
-ks8851_mll-objs = ks8851_common.o ks8851_par.o
 obj-$(CONFIG_KSZ884X_PCI) += ksz884x.o
drivers/net/ethernet/micrel/ks8851.c renamed from drivers/net/ethernet/micrel/ks8851_spi.c +13 -29

113 lines not on this page · raw message

drivers/net/ethernet/micrel/ks8851.h +9 -18
index 7932996c6ec4..a8fd2ece2345 100644
@@ -304,6 +304,15 @@
 #define TXFR_TXFID_MASK				(0x3f << 0)
 #define TXFR_TXFID_SHIFT			(0)
 
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
+#define DEBUG
+
+#include <linux/kernel.h>
+#include <linux/eeprom_93cx6.h>
+#include <linux/mii.h>
+#include <linux/netdevice.h>
+
 /**
  * struct ks8851_rxctrl - KS8851 driver rx control
  * @mchash: Multicast hash-table data.
@@ -403,24 +412,6 @@ struct ks8851_net {
 					  struct sk_buff *txp, bool irq);
 };
 
-int ks8851_probe_common(struct net_device *netdev, struct device *dev);
-int ks8851_remove_common(struct device *dev);
-int ks8851_suspend(struct device *dev);
-int ks8851_resume(struct device *dev);
-
-void ks8851_lock(struct ks8851_net *ks, unsigned long *flags);
-void ks8851_unlock(struct ks8851_net *ks, unsigned long *flags);
-
-unsigned int ks8851_rdreg16(struct ks8851_net *ks, unsigned int reg);
-void ks8851_wrreg16(struct ks8851_net *ks, unsigned int reg, unsigned int val);
-void ks8851_rdfifo(struct ks8851_net *ks, u8 *buff, unsigned int len);
-void ks8851_wrfifo(struct ks8851_net *ks, struct sk_buff *txp, bool irq);
-
-void ks8851_rx_skb(struct sk_buff *skb);
-void ks8851_flush_tx_work(struct ks8851_net *ks);
-
-netdev_tx_t ks8851_start_xmit(struct sk_buff *skb, struct net_device *dev);
-
 static SIMPLE_DEV_PM_OPS(ks8851_pm_ops, ks8851_suspend, ks8851_resume);
 
 /**
drivers/net/ethernet/micrel/ks8851_common.h renamed from drivers/net/ethernet/micrel/ks8851_common.c +4 -15
similarity index 98%
rename from drivers/net/ethernet/micrel/ks8851_common.c
rename to drivers/net/ethernet/micrel/ks8851_common.h
index 9878369efda2..5f96230e08cd 100644
@@ -6,28 +6,18 @@
  *	Ben Dooks <[email protected]>
  */
 
-#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
-
-#define DEBUG
-
 #include <linux/interrupt.h>
 #include <linux/module.h>
-#include <linux/kernel.h>
-#include <linux/netdevice.h>
 #include <linux/etherdevice.h>
 #include <linux/ethtool.h>
 #include <linux/cache.h>
 #include <linux/crc32.h>
-#include <linux/mii.h>
-#include <linux/eeprom_93cx6.h>
 #include <linux/regulator/consumer.h>
 
 #include <linux/gpio.h>
 #include <linux/of_gpio.h>
 #include <linux/of_net.h>
 
-#include "ks8851.h"
-
 static int msg_enable;
 
 /**
@@ -930,7 +920,7 @@ static int ks8851_read_selftest(struct ks8851_net *ks)
 
 #ifdef CONFIG_PM_SLEEP
 
-int ks8851_suspend(struct device *dev)
+static int ks8851_suspend(struct device *dev)
 {
 	struct ks8851_net *ks = dev_get_drvdata(dev);
 	struct net_device *netdev = ks->netdev;
@@ -943,7 +933,7 @@ int ks8851_suspend(struct device *dev)
 	return 0;
 }
 
-int ks8851_resume(struct device *dev)
+static int ks8851_resume(struct device *dev)
 {
 	struct ks8851_net *ks = dev_get_drvdata(dev);
 	struct net_device *netdev = ks->netdev;
@@ -957,7 +947,7 @@ int ks8851_resume(struct device *dev)
 }
 #endif
 
-int ks8851_probe_common(struct net_device *netdev, struct device *dev)
+static int ks8851_probe_common(struct net_device *netdev, struct device *dev)
 {
 	struct ks8851_net *ks = netdev_priv(netdev);
 	unsigned cider;
@@ -1086,7 +1076,7 @@ int ks8851_probe_common(struct net_device *netdev, struct device *dev)
 	return ret;
 }
 
-int ks8851_remove_common(struct device *dev)
+static int ks8851_remove_common(struct device *dev)
 {
 	struct ks8851_net *priv = dev_get_drvdata(dev);
 
@@ -1104,4 +1094,3 @@ int ks8851_remove_common(struct device *dev)
 
 module_param_named(message, msg_enable, int, 0);
 MODULE_PARM_DESC(message, "Message verbosity level (0=none, 31=all)");
-MODULE_ALIAS("spi:ks8851");
drivers/net/ethernet/micrel/ks8851_mll.c renamed from drivers/net/ethernet/micrel/ks8851_par.c +12 -29

109 lines not on this page · raw message

Signature · 1 line
2.25.0
#23 lore raw
Quoted 32 lines · Lukas Wunner
Quoted 11 lines · Marek Vasut

The KS8851SNL/SNLI and KS8851-16MLL/MLLI/MLLU are very much the same pieces of silicon, except the former has an SPI interface, while the later has a parallel bus interface. Thus far, Linux has two separate drivers for each and they are diverging considerably.

This series unifies them into a single driver with small SPI and parallel bus specific parts. The approach here is to first separate out the SPI specific parts into a separate file, then add parallel bus accessors in another separate file and then finally remove the old parallel bus driver. The reason for replacing the old parallel bus driver is because the SPI bus driver is much higher quality.

Sorry for the delay Marek.

ks8851.ko (SPI variant) no longer compiles with this series. The attached 0001 patch fixes it.

Both drivers can only be compiled as modules with this series:
If only one of them is built-in, there's a linker error because of the module_param_named() for msg_enable.
If both are built-in, the symbol collisions you've mentioned occur.

It seems Kbuild can't support building a .o file with a different name than the corresponding .c file because of the implicit rules used everywhere. However, ks8851_common.c can be renamed to be a header file (a library of sorts) which is included by the two .c files. I've renamed ks8851_spi.c back to the original ks8851.c and ks8851_par.c back to ks8851_mll.c. The result is the attached 0002 patch. Compiles without any errors regardless if one or both drivers are built-in or modules.

Quoted 2 lines

I'll be back at the office this week and will conduct performance measurements with this version.

This looks like a hack, I'm more inclined to go back to using callbacks for the various functions, since I don't see any performance problems there. We're still talking about 25 MHz SPI bus here and the SPI subsystem is full of such indirection anyway, so I'm not even sure what you're hoping to gain here ; it seems to me like a premature optimization which only causes trouble.

#24 lore raw
Quoted 36 lines · Marek Vasut
Quoted 32 lines · Lukas Wunner
Quoted 11 lines · Marek Vasut

The KS8851SNL/SNLI and KS8851-16MLL/MLLI/MLLU are very much the same pieces of silicon, except the former has an SPI interface, while the later has a parallel bus interface. Thus far, Linux has two separate drivers for each and they are diverging considerably.

This series unifies them into a single driver with small SPI and parallel bus specific parts. The approach here is to first separate out the SPI specific parts into a separate file, then add parallel bus accessors in another separate file and then finally remove the old parallel bus driver. The reason for replacing the old parallel bus driver is because the SPI bus driver is much higher quality.

Sorry for the delay Marek.

ks8851.ko (SPI variant) no longer compiles with this series. The attached 0001 patch fixes it.

Both drivers can only be compiled as modules with this series:
If only one of them is built-in, there's a linker error because of the module_param_named() for msg_enable.
If both are built-in, the symbol collisions you've mentioned occur.

It seems Kbuild can't support building a .o file with a different name than the corresponding .c file because of the implicit rules used everywhere. However, ks8851_common.c can be renamed to be a header file (a library of sorts) which is included by the two .c files. I've renamed ks8851_spi.c back to the original ks8851.c and ks8851_par.c back to ks8851_mll.c. The result is the attached 0002 patch. Compiles without any errors regardless if one or both drivers are built-in or modules.

Quoted 2 lines

I'll be back at the office this week and will conduct performance measurements with this version.

Quoted 6 lines

This looks like a hack, I'm more inclined to go back to using callbacks for the various functions, since I don't see any performance problems there. We're still talking about 25 MHz SPI bus here and the SPI subsystem is full of such indirection anyway, so I'm not even sure what you're hoping to gain here ; it seems to me like a premature optimization which only causes trouble.

So I got the KS8851SNL development kit to test the SPI option. The current driver in linux-next gives me SPI transfer timeouts at 1 MHz (I also tried 6 MHz, 10 MHz, same problem, I also checked the signal quality which is OK) both on iMX6Q and on STM32MP1 with ~2 cm long wiring between the SoM and the KS8851SNL devkit, so that's where my testing ends, sadly. Unless you have some idea what the problem might be ?

That said, I at least did test of the impact of the pointer indirection on the KS8851-16MLL with [1] (direct access test is with the top 7 patches removed, indirect access test is with the branch as-is). The KS8851-16MLL also has higher bandwidth, so it calls the IO accessors more often than the SPI variant, and so it's likely more relevant when testing whether their overhead matters.

- SoC is stm32mp157c at 650 MHz
- link partner
Intel Corporation 82579LM Gigabit Network Connection (Lewisville) (rev 06)
- Test commands (repeated thrice, averaged,
after three discarded runs to prime the system):

  $ iperf -sei 1               # RX
  $ iperf -c 192.168.1.1 -ei 1 # TX
             | accessor |     RX       |     TX
KS8851-16MLL | indirect | 52.86 Mbit/s | 74.03 Mbit/s
KS8851-16MLL |  direct  | 53.80 Mbit/s | 73.90 Mbit/s

There is ~1 Mbit/s throughput increase on the RX average, however the throughput fluctuates by a few Mbit/s during the iperf RX tests between 50.7 Mbit/s and 54.1 Mbit/s. There is also some decrease on TX, but it is marginal. Interestingly enough, the TX test throughput fluctuates less than RX, only by small hundreds of kbit/s.

I'd say that the pointer indirection doesn't play any role here and it is better to keep the driver simple and go with it, unless the SPI option is affected, which I doubt it is.

[1]
https://git.kernel.org/pub/scm/linux/kernel/git/marex/linux-2.6.git/log/?h=ks8851-v4

#25 lore raw
Quoted 4 lines · Marek Vasut
Quoted 3 lines · Marek Vasut
Quoted 2 lines · Lukas Wunner

I'll be back at the office this week and will conduct performance measurements with this version.

Latency without this series (ping -A -c 100000):

100000 packets transmitted, 100000 received, 0% packet loss, time 205ms rtt min/avg/max/mdev = 0.790/1.695/2.624/0.046 ms, ipg/ewma 2.000/1.693 ms

                             ^^^^^

Latency with this series:

100000 packets transmitted, 100000 received, 0% packet loss, time 220ms rtt min/avg/max/mdev = 0.880/1.717/2.428/0.059 ms, ipg/ewma 2.000/1.710 ms

Latency with this series + my patch to inline accessors:

100000 packets transmitted, 100000 received, 0% packet loss, time 219ms rtt min/avg/max/mdev = 0.874/1.716/3.296/0.058 ms, ipg/ewma 2.000/1.719 ms

RX throughput without this series (iperf3 -f k -i 0 -c):

[ 5] 0.00-10.00 sec 19.0 MBytes 15958 Kbits/sec receiver

RX throughput with this series:

[ 5] 0.00-10.00 sec 18.4 MBytes 15452 Kbits/sec receiver

RX throughput with this series + my patch to inline accessors:

[ 5] 0.00-10.00 sec 18.5 MBytes 15506 Kbits/sec receiver

TX throughput without this series (iperf3 -R -f k -i 0 -c):

[ 5] 0.00-10.00 sec 18.6 MBytes 15604 Kbits/sec receiver

TX throughput with this series:

[ 5] 0.00-10.00 sec 18.3 MBytes 15314 Kbits/sec receiver

TX throughput with this series + my patch to inline accessors:

[ 5] 0.00-10.00 sec 18.3 MBytes 15349 Kbits/sec receiver

The commands were invoked from a machine with a Broadcom tg3 Gigabit Ethernet controller.

Conclusion: The series does incur a measurable performance penalty which should be fixed before it gets merged. Inlining the accessors only yields a very small improvement. I'm wondering where the performance penalty is originating from: Perhaps because of the 16-bit read of RXFC in ks8851_rx_pkts()?

Quoted 6 lines

So I got the KS8851SNL development kit to test the SPI option. The current driver in linux-next gives me SPI transfer timeouts at 1 MHz (I also tried 6 MHz, 10 MHz, same problem, I also checked the signal quality which is OK) both on iMX6Q and on STM32MP1 with ~2 cm long wiring between the SoM and the KS8851SNL devkit, so that's where my testing ends, sadly. Unless you have some idea what the problem might be ?

Check the signal quality with an oscilloscope.
Crank up drive strength of the SPI pins if supported by the pin controller. Check if the driver for the SPI controller is buggy or somehow limits the speed.

We use the KSZ8851SNL both with STM32-based products (at 20 MHz SPI clock, but without an operating system) and with Raspberry Pi based products (at up to 25 MHz SPI clock, with Linux). It is only the latter that I'm really familiar with. We've invested considerable resources to fix bugs and speed up the Raspberry Pi's SPI driver as much as possible. By now it works pretty well with the ks8851.

A Raspberry Pi 3 is readily available for just a few Euros, so you may want to pick up one of those. My most recent speed improvements for the Raspberry Pi's SPI and DMA drivers went into v5.4, so be sure to use at least that. Should signal quality be a problem, it's possible to increase GPIO drive strength by putting a dt-blob.bin in the /boot partition:

https://www.raspberrypi.org/documentation/configuration/pin-configuration.md

I forgot to test whether reading the MAC address from the external EEPROM works with v3 of your series, but I'll be back in the office on Tuesday to take another look.

Thanks,

Lukas

#26 lore raw
Quoted 43 lines · Lukas Wunner
Quoted 4 lines · Marek Vasut
Quoted 3 lines · Marek Vasut
Quoted 2 lines · Lukas Wunner

I'll be back at the office this week and will conduct performance measurements with this version.

Latency without this series (ping -A -c 100000):

100000 packets transmitted, 100000 received, 0% packet loss, time 205ms rtt min/avg/max/mdev = 0.790/1.695/2.624/0.046 ms, ipg/ewma 2.000/1.693 ms

                             ^^^^^

Latency with this series:

100000 packets transmitted, 100000 received, 0% packet loss, time 220ms rtt min/avg/max/mdev = 0.880/1.717/2.428/0.059 ms, ipg/ewma 2.000/1.710 ms

Latency with this series + my patch to inline accessors:

100000 packets transmitted, 100000 received, 0% packet loss, time 219ms rtt min/avg/max/mdev = 0.874/1.716/3.296/0.058 ms, ipg/ewma 2.000/1.719 ms

RX throughput without this series (iperf3 -f k -i 0 -c):

[ 5] 0.00-10.00 sec 19.0 MBytes 15958 Kbits/sec receiver

RX throughput with this series:

[ 5] 0.00-10.00 sec 18.4 MBytes 15452 Kbits/sec receiver

RX throughput with this series + my patch to inline accessors:

[ 5] 0.00-10.00 sec 18.5 MBytes 15506 Kbits/sec receiver

TX throughput without this series (iperf3 -R -f k -i 0 -c):

[ 5] 0.00-10.00 sec 18.6 MBytes 15604 Kbits/sec receiver

TX throughput with this series:

[ 5] 0.00-10.00 sec 18.3 MBytes 15314 Kbits/sec receiver

Quoted 7 lines

TX throughput with this series + my patch to inline accessors:

[ 5] 0.00-10.00 sec 18.3 MBytes 15349 Kbits/sec receiver

The commands were invoked from a machine with a Broadcom tg3 Gigabit Ethernet controller.

It would be helpful if we could at least run the same test, so the results are comparable.

Quoted 3 lines

Conclusion: The series does incur a measurable performance penalty which should be fixed before it gets merged. Inlining the accessors only yields a very small improvement.

OK, so we finally agree on this, thanks.

Quoted 3 lines

I'm wondering where the
performance penalty is originating from: Perhaps because of the 16-bit read of RXFC in ks8851_rx_pkts()?

Can you patch that part away to see whether that's the case ?

Quoted 9 lines
Quoted 6 lines

So I got the KS8851SNL development kit to test the SPI option. The current driver in linux-next gives me SPI transfer timeouts at 1 MHz (I also tried 6 MHz, 10 MHz, same problem, I also checked the signal quality which is OK) both on iMX6Q and on STM32MP1 with ~2 cm long wiring between the SoM and the KS8851SNL devkit, so that's where my testing ends, sadly. Unless you have some idea what the problem might be ?

Check the signal quality with an oscilloscope.
Crank up drive strength of the SPI pins if supported by the pin controller.

I already tried both, got nothing useful.

Quoted 2 lines

Check if the driver for the SPI controller is buggy or somehow limits the speed.

I used two different drivers -- the iMX SPI and the STM32 SPI -- I would say that if both show the same behavior, it's unlikely to be the driver.

Quoted 11 lines

We use the KSZ8851SNL both with STM32-based products (at 20 MHz SPI clock, but without an operating system) and with Raspberry Pi based products (at up to 25 MHz SPI clock, with Linux). It is only the latter that I'm really familiar with. We've invested considerable resources to fix bugs and speed up the Raspberry Pi's SPI driver as much as possible. By now it works pretty well with the ks8851.

A Raspberry Pi 3 is readily available for just a few Euros, so you may want to pick up one of those. My most recent speed improvements for the Raspberry Pi's SPI and DMA drivers went into v5.4, so be sure to use at least that.

This is based off linux-next, so that shouldn't be a problem.

I'll try to set up the RPi3 and see if that's any better.

Quoted 9 lines

Should signal quality be a problem, it's possible to increase GPIO drive strength by putting a dt-blob.bin in the /boot partition:

https://www.raspberrypi.org/documentation/configuration/pin-configuration.md

I forgot to test whether reading the MAC address from the external EEPROM works with v3 of your series, but I'll be back in the office on Tuesday to take another look.

While you're at it, can you take a look at the latency ? I think you should be able to git-bisect it rather easily. Thanks

#27 lore raw
Quoted 4 lines · Marek Vasut

NOTE: The V4 is now tested on RPi3B with KSZ8851SNL DEMO Board at 25 MHz.
The "ping -c 1000 -i 0.01" latency test is fluctuating around
rtt min/avg/max/mdev = 1.448/1.540/1.699/0.030 ms
either way, with or without this series.

Compiling this series fails if CONFIG_KS8851=m and CONFIG_KS8851_MLL=y:

arm-linux-gnueabihf-ld: drivers/net/ethernet/micrel/ks8851_common.o:(__param+0x4): undefined reference to `__this_module'

I already reported this in my e-mail of April 6th, it's caused by the module_param_named() for msg_enable.

Reading the MAC address from an external EEPROM works with this series.

I still get worse performance with this series than without it (perhaps unsurprisingly so because not much has changed from v3 to v4).

We're using CONFIG_PREEMPT_RT_FULL=y. I'm sorry for not mentioning this earlier, I didn't assume it would make such a big difference but apparently it does.

This is the branch I've tested today:
https://github.com/l1k/linux/commits/revpi-4.19-marek-v4

Latency without this series (ping -A -c 100000):
rtt min/avg/max/mdev = 0.793/1.694/2.321/0.044 ms, ipg/ewma 2.000/1.691 ms

Latency with this series:
rtt min/avg/max/mdev = 0.957/1.715/2.652/0.043 ms, ipg/ewma 2.000/1.716 ms

RX throughput without this series (iperf3 -f k -i 0 -c):
[ 5] 0.00-10.00 sec 19.0 MBytes 15960 Kbits/sec receiver

RX throughput with this series:
[ 5] 0.00-10.00 sec 18.5 MBytes 15498 Kbits/sec receiver

TX throughput without this series (iperf3 -R -f k -i 0 -c):
[ 5] 0.00-10.00 sec 18.6 MBytes 15614 Kbits/sec receiver

TX throughput with this series:
[ 5] 0.00-10.00 sec 18.3 MBytes 15371 Kbits/sec receiver

So this is pretty much the same performance degredation as before.

Quoted 5 lines
Quoted 3 lines

I'm wondering where the
performance penalty is originating from: Perhaps because of the 16-bit read of RXFC in ks8851_rx_pkts()?

Can you patch that part away to see whether that's the case ?

Will do.

Quoted 5 lines
Quoted 2 lines

Check if the driver for the SPI controller is buggy or somehow limits the speed.

I used two different drivers -- the iMX SPI and the STM32 SPI -- I would say that if both show the same behavior, it's unlikely to be the driver.

Hm, so why did it work with the RasPi but not with the others?

Thanks,

Lukas

#28 lore raw
Quoted 14 lines · Lukas Wunner
Quoted 4 lines · Marek Vasut

NOTE: The V4 is now tested on RPi3B with KSZ8851SNL DEMO Board at 25 MHz.
The "ping -c 1000 -i 0.01" latency test is fluctuating around
rtt min/avg/max/mdev = 1.448/1.540/1.699/0.030 ms
either way, with or without this series.

Compiling this series fails if CONFIG_KS8851=m and CONFIG_KS8851_MLL=y:

arm-linux-gnueabihf-ld: drivers/net/ethernet/micrel/ks8851_common.o:(__param+0x4): undefined reference to `__this_module'

I already reported this in my e-mail of April 6th, it's caused by the module_param_named() for msg_enable.

Quoted 5 lines

Reading the MAC address from an external EEPROM works with this series.

I still get worse performance with this series than without it (perhaps unsurprisingly so because not much has changed from v3 to v4).

I reinstated the indirect access, so things did change. Besides, there performance for the parallel option is back where it was with the old driver, which is important for me.

Quoted 3 lines

We're using CONFIG_PREEMPT_RT_FULL=y. I'm sorry for not mentioning this earlier, I didn't assume it would make such a big difference but apparently it does.

Do you also have the RT patch applied ?

Quoted 2 lines

This is the branch I've tested today:
https://github.com/l1k/linux/commits/revpi-4.19-marek-v4

You seem to have quite a few more patches in that repository than just this series, some of them even touching the RPi SPI driver and it's DMA implementation.

Could it be that's why I don't see your performance problems too ?

Quoted 23 lines

Latency without this series (ping -A -c 100000):
rtt min/avg/max/mdev = 0.793/1.694/2.321/0.044 ms, ipg/ewma 2.000/1.691 ms

Latency with this series:
rtt min/avg/max/mdev = 0.957/1.715/2.652/0.043 ms, ipg/ewma 2.000/1.716 ms

RX throughput without this series (iperf3 -f k -i 0 -c):
[ 5] 0.00-10.00 sec 19.0 MBytes 15960 Kbits/sec receiver

RX throughput with this series:
[ 5] 0.00-10.00 sec 18.5 MBytes 15498 Kbits/sec receiver

TX throughput without this series (iperf3 -R -f k -i 0 -c):
[ 5] 0.00-10.00 sec 18.6 MBytes 15614 Kbits/sec receiver

TX throughput with this series:
[ 5] 0.00-10.00 sec 18.3 MBytes 15371 Kbits/sec receiver

So this is pretty much the same performance degredation as before.

Quoted 7 lines
Quoted 5 lines
Quoted 3 lines

I'm wondering where the
performance penalty is originating from: Perhaps because of the 16-bit read of RXFC in ks8851_rx_pkts()?

Can you patch that part away to see whether that's the case ?

Will do.

Thanks

Quoted 7 lines
Quoted 5 lines
Quoted 2 lines

Check if the driver for the SPI controller is buggy or somehow limits the speed.

I used two different drivers -- the iMX SPI and the STM32 SPI -- I would say that if both show the same behavior, it's unlikely to be the driver.

Hm, so why did it work with the RasPi but not with the others?

I didn't have a chance to debug this yet.

#29 lore raw
Quoted 4 lines · Marek Vasut

On 4/15/20 4:39 PM, Lukas Wunner wrote:
I reinstated the indirect access, so things did change. Besides, there performance for the parallel option is back where it was with the old driver, which is important for me.

Okay, I see.

Quoted 5 lines
Quoted 3 lines

We're using CONFIG_PREEMPT_RT_FULL=y. I'm sorry for not mentioning this earlier, I didn't assume it would make such a big difference but apparently it does.

Do you also have the RT patch applied ?

Yes, the branch I linked to also contains the RT patches.

Quoted 6 lines
Quoted 2 lines

This is the branch I've tested today:
https://github.com/l1k/linux/commits/revpi-4.19-marek-v4

You seem to have quite a few more patches in that repository than just this series, some of them even touching the RPi SPI driver and it's DMA implementation.

Those are just the patches I mainlined, but backported to v4.19. This branch is based on the Raspberry Pi Foundation's downstream repository, they still ship v4.19. Their repo has more performant drivers for USB, SD/MMC etc. which haven't been mainlined yet. But the SPI portion is the same as in mainline because I always submit to mainline and let the patches percolate down to their repo.

Quoted 6 lines
Quoted 4 lines
Quoted 2 lines

I used two different drivers -- the iMX SPI and the STM32 SPI -- I would say that if both show the same behavior, it's unlikely to be the driver.

Hm, so why did it work with the RasPi but not with the others?

I didn't have a chance to debug this yet.

I was just curious if those drivers don't work as well as the one for the RasPi. :-) That would be funny because the RasPi is generally considered a toy in the embedded space and the platforms you've mentioned are taken more seriously I think.

Thanks,

Lukas

#30 lore raw
Quoted 9 lines · Lukas Wunner
Quoted 4 lines · Marek Vasut

On 4/15/20 4:39 PM, Lukas Wunner wrote:
I reinstated the indirect access, so things did change. Besides, there performance for the parallel option is back where it was with the old driver, which is important for me.

Okay, I see.

Quoted 7 lines
Quoted 5 lines
Quoted 3 lines

We're using CONFIG_PREEMPT_RT_FULL=y. I'm sorry for not mentioning this earlier, I didn't assume it would make such a big difference but apparently it does.

Do you also have the RT patch applied ?

Yes, the branch I linked to also contains the RT patches.

OK, I see. I am only testing against latest linux-next with no extra patches. Can you do that as well ?

Here is all you need:
https://git.kernel.org/pub/scm/linux/kernel/git/marex/linux-2.6.git/log/?h=ks8851-v5

Quoted 7 lines
Quoted 6 lines
Quoted 2 lines

This is the branch I've tested today:
https://github.com/l1k/linux/commits/revpi-4.19-marek-v4

You seem to have quite a few more patches in that repository than just this series, some of them even touching the RPi SPI driver and it's DMA implementation.

Quoted 6 lines

Those are just the patches I mainlined, but backported to v4.19. This branch is based on the Raspberry Pi Foundation's downstream repository, they still ship v4.19. Their repo has more performant drivers for USB, SD/MMC etc. which haven't been mainlined yet. But the SPI portion is the same as in mainline because I always submit to mainline and let the patches percolate down to their repo.

Above I linked current linux-next with KS8851 patches applied (plus fixes for the next version of this series, including the RPi3 DT patch and config I used to test the SPI version), so please use that.

Quoted 11 lines
Quoted 6 lines
Quoted 4 lines
Quoted 2 lines

I used two different drivers -- the iMX SPI and the STM32 SPI -- I would say that if both show the same behavior, it's unlikely to be the driver.

Hm, so why did it work with the RasPi but not with the others?

I didn't have a chance to debug this yet.

I was just curious if those drivers don't work as well as the one for the RasPi. :-) That would be funny because the RasPi is generally considered a toy in the embedded space and the platforms you've mentioned are taken more seriously I think.

My guess would be that the SPI in the RPi is just ignoring various non-critical transmission errors to cater for all the "toy" hardware, but that's just pure speculation.

[...]