Message-ID: <20150328002844.GA32500@ld-irv-0074>
Date:	Fri, 27 Mar 2015 17:28:44 -0700
From:	Brian Norris <computersforpeace@...il.com>
To:	Kishon Vijay Abraham I <kishon@...com>
Cc:	Tejun Heo <tj@...nel.org>, Rob Herring <robh+dt@...nel.org>,
	Pawel Moll <pawel.moll@....com>,
	Mark Rutland <mark.rutland@....com>,
	Ian Campbell <ijc+devicetree@...lion.org.uk>,
	Kumar Gala <galak@...eaurora.org>,
	Gregory Fong <gregory.0xf0@...il.com>,
	Florian Fainelli <f.fainelli@...il.com>,
	devicetree@...r.kernel.org, linux-arm-kernel@...ts.infradead.org,
	linux-kernel@...r.kernel.org, linux-ide@...r.kernel.org
Subject: Re: [PATCH 4/5] phy: add Broadcom SATA3 PHY driver for Broadcom STB
 SoCs

Hi Kishon,

On Thu, Mar 26, 2015 at 03:29:44AM +0530, Kishon Vijay Abraham I wrote:
> On Thursday 19 March 2015 06:53 AM, Brian Norris wrote:
> > Supports up to two ports which can each be powered on/off and configured
> > independently.
> > 
> > Signed-off-by: Brian Norris <computersforpeace@...il.com>
> > ---
> >  drivers/phy/Kconfig            |   9 ++
> >  drivers/phy/Makefile           |   1 +
> >  drivers/phy/phy-brcmstb-sata.c | 333 +++++++++++++++++++++++++++++++++++++++++
> >  3 files changed, 343 insertions(+)
> >  create mode 100644 drivers/phy/phy-brcmstb-sata.c
> > 
> > diff --git a/drivers/phy/Kconfig b/drivers/phy/Kconfig
> > index 2962de205ba7..c8b22074bcf6 100644
> > --- a/drivers/phy/Kconfig
> > +++ b/drivers/phy/Kconfig
> > @@ -291,4 +291,13 @@ config PHY_QCOM_UFS
> >  	help
> >  	  Support for UFS PHY on QCOM chipsets.
> >  
> > +config PHY_BRCMSTB_SATA
> > +	tristate "Broadcom STB SATA PHY driver"
> > +	depends on ARCH_BRCMSTB
> > +	depends on OF
> > +	select GENERIC_PHY
> > +	help
> > +	  Enable this to support the SATA3 PHY on 28nm Broadcom STB SoCs.
> > +	  Likely useful only with CONFIG_SATA_BRCMSTB enabled.
> > +
> >  endmenu
> > diff --git a/drivers/phy/Makefile b/drivers/phy/Makefile
> > index f080e1bb2a74..28a10804b4f4 100644
> > --- a/drivers/phy/Makefile
> > +++ b/drivers/phy/Makefile
> > @@ -38,3 +38,4 @@ obj-$(CONFIG_PHY_STIH41X_USB)		+= phy-stih41x-usb.o
> >  obj-$(CONFIG_PHY_QCOM_UFS) 	+= phy-qcom-ufs.o
> >  obj-$(CONFIG_PHY_QCOM_UFS) 	+= phy-qcom-ufs-qmp-20nm.o
> >  obj-$(CONFIG_PHY_QCOM_UFS) 	+= phy-qcom-ufs-qmp-14nm.o
> > +obj-$(CONFIG_PHY_BRCMSTB_SATA)		+= phy-brcmstb-sata.o
> > diff --git a/drivers/phy/phy-brcmstb-sata.c b/drivers/phy/phy-brcmstb-sata.c
> > new file mode 100644
> > index 000000000000..413bc94225ac
> > --- /dev/null
> > +++ b/drivers/phy/phy-brcmstb-sata.c
> > @@ -0,0 +1,333 @@
> > +/*
> > + * Broadcom SATA3 AHCI Controller PHY Driver
> > + *
> > + * Copyright © 2009-2015 Broadcom Corporation
> > + *
> > + * This program is free software; you can redistribute it and/or modify
> > + * it under the terms of the GNU General Public License as published by
> > + * the Free Software Foundation; either version 2, or (at your option)
> > + * any later version.
> > + *
> > + * This program is distributed in the hope that it will be useful,
> > + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> > + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> > + * GNU General Public License for more details.
> > + */
> > +
> > +#include <linux/device.h>
> > +#include <linux/init.h>
> > +#include <linux/interrupt.h>
> > +#include <linux/io.h>
> > +#include <linux/kernel.h>
> > +#include <linux/module.h>
> > +#include <linux/of.h>
> > +#include <linux/phy/phy.h>
> > +#include <linux/platform_device.h>
> > +
> > +#define SATA_MDIO_BANK_OFFSET				0x23c
> > +#define SATA_MDIO_REG_OFFSET(ofs)			((ofs) * 4)
> > +#define SATA_MDIO_REG_SPACE_SIZE			0x1000
> > +#define SATA_MDIO_REG_LENGTH				0x1f00
> > +
> > +#define SATA_TOP_CTRL_PHY_CTRL_1			0x0
> > + #define SATA_TOP_CTRL_1_PHY_DEFAULT_POWER_STATE	BIT(14)
> > +
> > +#define SATA_TOP_CTRL_PHY_CTRL_2			0x4
> > + #define SATA_TOP_CTRL_2_SW_RST_MDIOREG			BIT(0)
> > + #define SATA_TOP_CTRL_2_SW_RST_OOB			BIT(1)
> > + #define SATA_TOP_CTRL_2_SW_RST_RX			BIT(2)
> > + #define SATA_TOP_CTRL_2_SW_RST_TX			BIT(3)
> > + #define SATA_TOP_CTRL_2_PHY_GLOBAL_RESET		BIT(14)
> > +
> > +#define MAX_PORTS					2
> > +/* Register offset between PHYs in port-ctrl */
> > +#define SATA_TOP_CTRL_PHY_CTRL_LEN			0x8
> > +/* Register offset between PHYs in PCB space */
> > +#define SATA_MDIO_REG_SPACE_SIZE			0x1000
> > +
> > +struct brcm_sata_port {
> > +	int portnum;
> > +	struct phy *phy;
> > +	struct brcm_sata_phy *phy_priv;
> > +	bool ssc_en;
> > +};
> > +
> > +struct brcm_sata_phy {
> > +	struct device *dev;
> > +	void __iomem *port_ctrl;
> > +	void __iomem *phy_base;
> > +
> > +	struct brcm_sata_port phys[MAX_PORTS];
> 
> Can't we allocate memory for the PHYs dynamically?

Yes, but I don't see a lot of point for an IP that supports exactly 2
ports...

> > +};
> > +
> > +enum sata_mdio_phy_regs_28nm {
> > +	PLL_REG_BANK_0				= 0x50,
> > +	PLL_REG_BANK_0_PLLCONTROL_0		= 0x81,
> > +
> > +	TXPMD_REG_BANK				= 0x1a0,
> > +	TXPMD_CONTROL1				= 0x81,
> > +	TXPMD_CONTROL1_TX_SSC_EN_FRC		= BIT(0),
> > +	TXPMD_CONTROL1_TX_SSC_EN_FRC_VAL	= BIT(1),
> > +	TXPMD_TX_FREQ_CTRL_CONTROL1		= 0x82,
> > +	TXPMD_TX_FREQ_CTRL_CONTROL2		= 0x83,
> > +	TXPMD_TX_FREQ_CTRL_CONTROL2_FMIN_MASK	= 0x3ff,
> > +	TXPMD_TX_FREQ_CTRL_CONTROL3		= 0x84,
> > +	TXPMD_TX_FREQ_CTRL_CONTROL3_FMAX_MASK	= 0x3ff,
> > +};
> > +
> > +static inline void __iomem *sata_phy_get_port_ctrl(struct brcm_sata_port *port)
> > +{
> > +	struct brcm_sata_phy *priv = port->phy_priv;
> > +
> > +	return priv->port_ctrl + (port->portnum * SATA_TOP_CTRL_PHY_CTRL_LEN);
> > +}
> > +static inline void __iomem *sata_phy_get_phy_base(struct brcm_sata_port *port)
> 
> please use consistent names for all functions. Prefix all the functions
> with brcmstb.

Will fix.

> > +{
> > +	struct brcm_sata_phy *priv = port->phy_priv;
> > +
> > +	return priv->phy_base + (port->portnum * SATA_MDIO_REG_SPACE_SIZE);
> > +}
> > +
> > +static void brcm_sata_mdio_wr(void __iomem *addr, u32 bank, u32 ofs,
> > +			      u32 msk, u32 value)
> > +{
> > +	u32 tmp;
> > +
> > +	writel(bank, addr + SATA_MDIO_BANK_OFFSET);
> > +	tmp = readl(addr + SATA_MDIO_REG_OFFSET(ofs));
> > +	tmp = (tmp & msk) | value;
> > +	writel(tmp, addr + SATA_MDIO_REG_OFFSET(ofs));
> > +}
> > +
> > +/* These defaults were characterized by H/W group */
> > +#define FMIN_VAL_DEFAULT	0x3df
> > +#define FMAX_VAL_DEFAULT	0x3df
> > +#define FMAX_VAL_SSC		0x83
> > +
> > +static void cfg_ssc_28nm(struct brcm_sata_port *port)
> > +{
> > +	void __iomem *base = sata_phy_get_phy_base(port);
> > +	struct brcm_sata_phy *priv = port->phy_priv;
> > +	u32 tmp;
> > +
> > +	/* override the TX spread spectrum setting */
> > +	tmp = TXPMD_CONTROL1_TX_SSC_EN_FRC_VAL | TXPMD_CONTROL1_TX_SSC_EN_FRC;
> > +	brcm_sata_mdio_wr(base, TXPMD_REG_BANK, TXPMD_CONTROL1, ~tmp, tmp);
> > +
> > +	/* set fixed min freq */
> > +	brcm_sata_mdio_wr(base, TXPMD_REG_BANK, TXPMD_TX_FREQ_CTRL_CONTROL2,
> > +			  ~TXPMD_TX_FREQ_CTRL_CONTROL2_FMIN_MASK,
> > +			  FMIN_VAL_DEFAULT);
> > +
> > +	/* set fixed max freq depending on SSC config */
> > +	if (port->ssc_en) {
> > +		dev_info(priv->dev, "enabling SSC on port %d\n", port->portnum);
> > +		tmp = FMAX_VAL_SSC;
> > +	} else {
> > +		tmp = FMAX_VAL_DEFAULT;
> > +	}
> > +
> > +	brcm_sata_mdio_wr(base, TXPMD_REG_BANK, TXPMD_TX_FREQ_CTRL_CONTROL3,
> > +			  ~TXPMD_TX_FREQ_CTRL_CONTROL3_FMAX_MASK, tmp);
> > +}
> > +
> > +static void brcm_sata_phy_enable(struct brcm_sata_port *port)
> > +{
> > +	void __iomem *port_ctrl = sata_phy_get_port_ctrl(port);
> > +	void __iomem *p;
> > +	u32 reg;
> > +
> > +	/* clear PHY_DEFAULT_POWER_STATE */
> > +	p = port_ctrl + SATA_TOP_CTRL_PHY_CTRL_1;
> > +	reg = readl(p);
> > +	reg &= ~SATA_TOP_CTRL_1_PHY_DEFAULT_POWER_STATE;
> > +	writel(reg, p);
> > +
> > +	/* reset the PHY digital logic */
> > +	p = port_ctrl + SATA_TOP_CTRL_PHY_CTRL_2;
> > +	reg = readl(p);
> > +	reg &= ~(SATA_TOP_CTRL_2_SW_RST_MDIOREG | SATA_TOP_CTRL_2_SW_RST_OOB |
> > +		 SATA_TOP_CTRL_2_SW_RST_RX);
> > +	reg |= SATA_TOP_CTRL_2_SW_RST_TX;
> > +	writel(reg, p);
> > +	reg = readl(p);
> > +	reg |= SATA_TOP_CTRL_2_PHY_GLOBAL_RESET;
> > +	writel(reg, p);
> > +	reg = readl(p);
> > +	reg &= ~SATA_TOP_CTRL_2_PHY_GLOBAL_RESET;
> > +	writel(reg, p);
> > +	(void)readl(p);
> > +}
> > +
> > +static void brcm_sata_phy_disable(struct brcm_sata_port *port)
> > +{
> > +	void __iomem *port_ctrl = sata_phy_get_port_ctrl(port);
> > +	void __iomem *p;
> > +	u32 reg;
> > +
> > +	/* power-off the PHY digital logic */
> > +	p = port_ctrl + SATA_TOP_CTRL_PHY_CTRL_2;
> > +	reg = readl(p);
> > +	reg |= (SATA_TOP_CTRL_2_SW_RST_MDIOREG | SATA_TOP_CTRL_2_SW_RST_OOB |
> > +		SATA_TOP_CTRL_2_SW_RST_RX | SATA_TOP_CTRL_2_SW_RST_TX |
> > +		SATA_TOP_CTRL_2_PHY_GLOBAL_RESET);
> > +	writel(reg, p);
> > +
> > +	/* set PHY_DEFAULT_POWER_STATE */
> > +	p = port_ctrl + SATA_TOP_CTRL_PHY_CTRL_1;
> > +	reg = readl(p);
> > +	reg |= SATA_TOP_CTRL_1_PHY_DEFAULT_POWER_STATE;
> > +	writel(reg, p);
> > +}
> > +
> > +static int brcmstb_sata_phy_power_on(struct phy *phy)
> > +{
> > +	struct brcm_sata_port *port = phy_get_drvdata(phy);
> > +
> > +	dev_info(port->phy_priv->dev, "powering on port %d\n", port->portnum);
> 
> dev_dbg?

See comments below.

> > +
> > +	brcm_sata_phy_enable(port);
> > +	cfg_ssc_28nm(port);
> > +
> > +	return 0;
> > +}
> > +
> > +static int brcmstb_sata_phy_power_off(struct phy *phy)
> > +{
> > +	struct brcm_sata_port *port = phy_get_drvdata(phy);
> > +
> > +	dev_info(port->phy_priv->dev, "powering off port %d\n", port->portnum);
> 
> same here..
> > +
> > +	brcm_sata_phy_disable(port);
> > +
> > +	return 0;
> > +}
> > +
> > +static struct phy_ops phy_ops_28nm = {
> > +	.power_on	= brcmstb_sata_phy_power_on,
> > +	.power_off	= brcmstb_sata_phy_power_off,
> > +	.owner		= THIS_MODULE,
> > +};
> > +
> > +static struct phy *brcm_sata_phy_xlate(struct device *dev,
> > +				       struct of_phandle_args *args)
> > +{
> > +	struct brcm_sata_phy *priv = dev_get_drvdata(dev);
> > +	int i = args->args[0];
> > +
> > +	if (i >= MAX_PORTS || !priv->phys[i].phy) {
> > +		dev_err(dev, "invalid phy: %d\n", i);
> > +		return ERR_PTR(-ENODEV);
> > +	}
> > +
> > +	return priv->phys[i].phy;
> > +}
> 
> this xlate is not required at all if the controller device tree node has
> phandle to the phy node (sub node) instead of the phy provider device tree
> node.

That doesn't match any convention I see in existing SATA phy bindings,
nor do I see how the existing of_phy_simple_xlate() would support this,
unless I instantiate a device for each port's PHY. If I adjust the
device tree as you suggest, and use of_phy_simple_xlate() instead of
this, of_phy_get() can't find the PHY provider, because the provider is
registered to the parent, not the subnode.

Can you elaborate on your suggestion?

> > +
> > +static const struct of_device_id brcmstb_sata_phy_of_match[] = {
> > +	{ .compatible	= "brcm,bcm7445-sata-phy" },
> > +	{},
> > +};
> > +MODULE_DEVICE_TABLE(of, brcmstb_sata_phy_of_match);
> > +
> > +static int brcmstb_sata_phy_probe(struct platform_device *pdev)
> > +{
> > +	struct device *dev = &pdev->dev;
> > +	struct device_node *dn = dev->of_node, *child;
> > +	struct brcm_sata_phy *priv;
> > +	struct resource *res;
> > +	struct phy_provider *provider;
> > +	int count = 0;
> > +
> > +	if (of_get_child_count(dn) == 0)
> > +		return -ENODEV;
> > +
> > +	priv = devm_kzalloc(dev, sizeof(*priv), GFP_KERNEL);
> > +	if (!priv)
> > +		return -ENOMEM;
> > +	dev_set_drvdata(dev, priv);
> > +	priv->dev = dev;
> > +
> > +	res = platform_get_resource_byname(pdev, IORESOURCE_MEM, "port-ctrl");
> > +	if (!res) {
> > +		dev_err(dev, "couldn't get port-ctrl resource\n");
> > +		return -EINVAL;
> > +	}
> > +	/*
> > +	 * Don't request region, since it may be within a region owned by the
> > +	 * SATA driver
> 
> It should be in the SATA driver then. Why is it here?

Did you read the discussion branching here?

http://article.gmane.org/gmane.linux.drivers.devicetree/114637

I've seen the exact opposite suggestion already (move it to the PHY
driver), and I'm not sure either suggestion is correct. The same
register block has registers for both the PHY and the SATA controller.

> > +	 */
> > +	priv->port_ctrl = devm_ioremap(dev, res->start, resource_size(res));
> > +	if (!priv->port_ctrl) {
> > +		dev_err(dev, "couldn't remap: %pR\n", res);
> > +		return -ENOMEM;
> > +	}
> > +
> > +	res = platform_get_resource_byname(pdev, IORESOURCE_MEM, "phy");
> > +	priv->phy_base = devm_ioremap_resource(dev, res);
> > +	if (IS_ERR(priv->phy_base))
> > +		return PTR_ERR(priv->phy_base);
> > +
> > +	for_each_available_child_of_node(dn, child) {
> > +		unsigned int id;
> > +		struct brcm_sata_port *port;
> > +
> > +		if (of_property_read_u32(child, "reg", &id)) {
> > +			dev_err(dev, "missing reg property in node %s\n",
> > +					child->name);
> > +			return -EINVAL;
> > +		}
> > +
> > +		if (id >= MAX_PORTS) {
> > +			dev_err(dev, "invalid reg: %u\n", id);
> > +			return -EINVAL;
> > +		}
> > +		if (priv->phys[id].phy) {
> > +			dev_err(dev, "already registered port %u\n", id);
> > +			return -EINVAL;
> > +		}
> > +
> > +		port = &priv->phys[id];
> > +		port->portnum = id;
> > +		port->phy_priv = priv;
> > +		port->phy = devm_phy_create(dev, NULL, &phy_ops_28nm);
> > +		port->ssc_en = of_property_read_bool(child, "brcm,enable-ssc");
> > +		if (IS_ERR(port->phy)) {
> > +			dev_err(dev, "failed to create PHY\n");
> > +			return PTR_ERR(port->phy);
> > +		}
> > +
> > +		phy_set_drvdata(port->phy, port);
> > +		count++;
> > +	}
> > +
> > +	provider = devm_of_phy_provider_register(dev, brcm_sata_phy_xlate);
> > +	if (IS_ERR(provider)) {
> > +		dev_err(dev, "could not register PHY provider\n");
> > +		return PTR_ERR(provider);
> > +	}
> > +
> > +	dev_info(dev, "registered %d ports\n", count);
> 
> lets not make the boot noisy. Make it dev_vdbg if it is required.

I think it's important to have at least some of the three informational
prints that you're suggesting turn into dbg. It's pretty important to
see that we're powering on one or more PHY ports, for both
power/correctness concerns (trying to power on a port that is
OTP-disabled, for instance) and debugging concerns (the suggestions you
made about the device tree yielded a dead SATA, and it was obvious,
because the "powering on" prints were missing).

I'd kinda like to see the previous power on/off prints above stay as
dev_info(), though the "registered" print might be superfluous, as the
registration info should show up in sysfs.

Related: I don't see any API for monitoring the PHY status from user
space. e.g., there's nothing useful in /sys/class/phy/*/.

> > +
> > +	return 0;
> > +}
> > +
> > +static int brcmstb_sata_phy_remove(struct platform_device *pdev)
> > +{
> > +	return 0;
> > +}
> 
> Remove this function if it doesn't have anything to do.

I was confused; I thought that that the driver model would not allow a
device to be detached if there was no remove function. Will remove now.

Thanks for the review,
Brian
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Powered by blists - more mailing lists