[PATCH v6 2/2] soc/imx: Add GPCv2 power gating driver
Dong Aisheng
dongas86 at gmail.com
Thu Mar 23 20:58:18 PDT 2017
On Mon, Mar 20, 2017 at 01:34:19PM -0700, Andrey Smirnov wrote:
> On Mon, Mar 20, 2017 at 12:03 AM, Dong Aisheng <dongas86 at gmail.com> wrote:
> > On Thu, Mar 16, 2017 at 06:30:50AM -0700, Andrey Smirnov wrote:
> >> Add code allowing for control of various power domains managed by GPCv2
> >> IP block found in i.MX7 series of SoCs. Power domains covered by this
> >> patch are:
> >>
> >> - PCIE PHY
> >> - MIPI PHY
> >> - USB HSIC PHY
> >> - USB OTG1/2 PHY
> >>
> >> Support for any other power domain controlled by GPC is not present, and
> >> can be added at some later point.
> >>
> >> Testing of this code was done against a PCIe driver.
> >>
> >> Cc: yurovsky at gmail.com
> >> Cc: Lucas Stach <l.stach at pengutronix.de>
> >> Cc: Fabio Estevam <fabio.estevam at nxp.com>
> >> Cc: linux-arm-kernel at lists.infradead.org
> >> Cc: linux-kernel at vger.kernel.org
> >> Signed-off-by: Andrey Smirnov <andrew.smirnov at gmail.com>
> >> ---
> >> drivers/soc/Kconfig | 1 +
> >> drivers/soc/imx/Kconfig | 10 ++
> >> drivers/soc/imx/Makefile | 1 +
> >> drivers/soc/imx/gpcv2.c | 385 +++++++++++++++++++++++++++++++++++++++++++++++
> >
> > Personally i'd like more naming gpc-imx7.c since gpcv2 is too generic,
> > and actually you also mix using imx7d name below.
> >
>
> This IP block is already referenced as GPCv2 in
> drivers/irqchip/irq-imx-gpcv2.c, so IMHO re-naming it to gpc-imx7
> would be confusing.
>
Okay, i see.
I got a NXP internal GPC block guide which also names it as GPCv2.
So internally we call it GPCv2.
And we can still grep two GPCv2 in RM, although most has been
changed to GPC.
Due to above reasons and to stop the bothering, i think we can
use the name as it is now to make things keep going.
> > And AFAIK only MX7 will use this new driver.
> >
> > Probably should also change the exist gpc to gpc-imx6.
> >
> >> 4 files changed, 397 insertions(+)
> >> create mode 100644 drivers/soc/imx/Kconfig
> >> create mode 100644 drivers/soc/imx/gpcv2.c
> >>
> >> diff --git a/drivers/soc/Kconfig b/drivers/soc/Kconfig
> >> index f31bceb..55a4eb8 100644
> >> --- a/drivers/soc/Kconfig
> >> +++ b/drivers/soc/Kconfig
> >> @@ -2,6 +2,7 @@ menu "SOC (System On Chip) specific Drivers"
> >>
> >> source "drivers/soc/bcm/Kconfig"
> >> source "drivers/soc/fsl/Kconfig"
> >> +source "drivers/soc/imx/Kconfig"
> >> source "drivers/soc/mediatek/Kconfig"
> >> source "drivers/soc/qcom/Kconfig"
> >> source "drivers/soc/rockchip/Kconfig"
> >> diff --git a/drivers/soc/imx/Kconfig b/drivers/soc/imx/Kconfig
> >> new file mode 100644
> >> index 0000000..bc7f0ee0
> >> --- /dev/null
> >> +++ b/drivers/soc/imx/Kconfig
> >> @@ -0,0 +1,10 @@
> >> +menu "i.MX SoC drivers"
> >> +
> >> +config IMX7_PM_DOMAINS
> >> + bool "i.MX7 PM domains"
> >> + select PM_GENERIC_DOMAINS
> >> + depends on SOC_IMX7D || (COMPILE_TEST && OF)
> >> + default y if SOC_IMX7D
> >> +
> >> +endmenu
> >> +
> >> diff --git a/drivers/soc/imx/Makefile b/drivers/soc/imx/Makefile
> >> index 35861f5..5b6e396 100644
> >> --- a/drivers/soc/imx/Makefile
> >> +++ b/drivers/soc/imx/Makefile
> >> @@ -1 +1,2 @@
> >> obj-y += gpc.o
> >> +obj-$(CONFIG_IMX7_PM_DOMAINS) += gpcv2.o
> >> diff --git a/drivers/soc/imx/gpcv2.c b/drivers/soc/imx/gpcv2.c
> >> new file mode 100644
> >> index 0000000..73c58ca
> >> --- /dev/null
> >> +++ b/drivers/soc/imx/gpcv2.c
> >> @@ -0,0 +1,385 @@
> >> +/*
> >> + * Copyright 2017 Impinj, Inc
> >> + * Author: Andrey Smirnov <andrew.smirnov at gmail.com>
> >> + *
> >> + * Based on the code of analogus driver:
> >> + *
> >> + * Copyright 2015-2017 Pengutronix, Lucas Stach <kernel at pengutronix.de>
> >> + *
> >> + * The code contained herein is licensed under the GNU General Public
> >> + * License. You may obtain a copy of the GNU General Public License
> >> + * Version 2 or later at the following locations:
> >> + *
> >> + * http://www.opensource.org/licenses/gpl-license.html
> >> + * http://www.gnu.org/copyleft/gpl.html
> >> + */
> >> +
> >> +#include <linux/clk.h>
> >
> > Where did you use it?
>
> Not sure, I'll double check and drop it if it's unused.
>
> >
> >> +#include <linux/delay.h>
> >> +#include <linux/io.h>
> >> +#include <linux/of_device.h>
> >> +#include <linux/platform_device.h>
> >> +#include <linux/pm_domain.h>
> >> +#include <linux/regmap.h>
> >> +#include <linux/regulator/consumer.h>
> >> +#include <dt-bindings/power/imx7-power.h>
> >> +
> >> +#define GPC_PGC_CPU_MAPPING 0xec
> >> +#define USB_HSIC_PHY_A7_DOMAIN BIT(6)
> >> +#define USB_OTG2_PHY_A7_DOMAIN BIT(5)
> >> +#define USB_OTG1_PHY_A7_DOMAIN BIT(4)
> >> +#define PCIE_PHY_A7_DOMAIN BIT(3)
> >> +#define MIPI_PHY_A7_DOMAIN BIT(2)
> >> +
> >> +#define GPC_PU_PGC_SW_PUP_REQ 0xf8
> >> +#define GPC_PU_PGC_SW_PDN_REQ 0x104
> >> +#define USB_HSIC_PHY_SW_Pxx_REQ BIT(4)
> >> +#define USB_OTG2_PHY_SW_Pxx_REQ BIT(3)
> >> +#define USB_OTG1_PHY_SW_Pxx_REQ BIT(2)
> >> +#define PCIE_PHY_SW_Pxx_REQ BIT(1)
> >> +#define MIPI_PHY_SW_Pxx_REQ BIT(0)
> >> +
> >> +#define GPC_MAX_REGISTER 0x1000
> >> +
> >> +#define GPC_PGC_nCTRL_PCR BIT(0)
> >> +
> >> +struct imx7_pgc_domain {
> >> + struct generic_pm_domain genpd;
> >> + struct regmap *regmap;
> >> + struct regulator *regulator;
> >> +
> >> + unsigned int pgc_nctrl;
> >> +
> >> + const struct {
> >> + u32 pxx;
> >> + u32 map;
> >> + } bits;
> >> +
> >> + const int voltage;
> >> + struct device *dev;
> >> +};
> >> +
> >> +static int imx7_gpc_pu_pgc_sw_pxx_req(struct generic_pm_domain *genpd,
> >> + bool on)
> >> +{
> >> + struct imx7_pgc_domain *domain = container_of(genpd,
> >> + struct imx7_pgc_domain,
> >> + genpd);
> >> + unsigned int offset = on ?
> >> + GPC_PU_PGC_SW_PUP_REQ : GPC_PU_PGC_SW_PDN_REQ;
> >> + const bool enable_power_control = domain->pgc_nctrl && !on;
> >> + const bool has_regulator = !IS_ERR(domain->regulator);
> >> + unsigned long deadline;
> >> + int ret = 0;
> >> +
> >> + regmap_update_bits(domain->regmap, GPC_PGC_CPU_MAPPING,
> >> + domain->bits.map, domain->bits.map);
> >> +
> >> + if (has_regulator && on) {
> >> + ret = regulator_enable(domain->regulator);
> >> + if (ret) {
> >> + dev_err(domain->dev, "failed to enable regulator\n");
> >> + goto unmap;
> >> + }
> >> + }
> >> +
> >> + if (enable_power_control)
> >> + regmap_update_bits(domain->regmap, domain->pgc_nctrl,
> >> + GPC_PGC_nCTRL_PCR, GPC_PGC_nCTRL_PCR);
> >> +
> >> + regmap_update_bits(domain->regmap, offset,
> >> + domain->bits.pxx, domain->bits.pxx);
> >> +
> >> + /*
> >> + * As per "5.5.9.4 Example Code 4" in IMX7DRM.pdf wait
> >> + * for PUP_REQ/PDN_REQ bit to be cleared
> >> + */
> >> + deadline = jiffies + msecs_to_jiffies(1);
> >> + while (true) {
> >> + u32 pxx_req;
> >> +
> >> + regmap_read(domain->regmap, offset, &pxx_req);
> >> +
> >> + if (!(pxx_req & domain->bits.pxx))
> >> + break;
> >> +
> >> + if (time_after(jiffies, deadline)) {
> >> + dev_err(domain->dev, "falied to command PGC\n");
> >> + ret = -ETIMEDOUT;
> >> + /*
> >> + * If we were in a process of enabling a
> >> + * domain and failed we might as well disable
> >> + * the regulator we just enabled. And if it
> >> + * was the opposite situation and we failed to
> >> + * power down -- keep the regulator on
> >> + */
> >> + on = !on;
> >> + break;
> >> + }
> >> +
> >> + cpu_relax();
> >> + }
> >> +
> >> + if (enable_power_control)
> >> + regmap_update_bits(domain->regmap, domain->pgc_nctrl,
> >> + GPC_PGC_nCTRL_PCR, 0);
> >> +
> >> + if (has_regulator && !on) {
> >> + int err;
> >> +
> >> + err = regulator_disable(domain->regulator);
> >> + if (err)
> >> + dev_err(domain->dev,
> >> + "failed to disable regulator: %d\n", ret);
> >> + /* Preserve earlier error code */
> >> + ret = ret ?: err;
> >> + }
> >> +unmap:
> >> + regmap_update_bits(domain->regmap, GPC_PGC_CPU_MAPPING,
> >> + domain->bits.map, 0);
> >> + return ret;
> >> +}
> >> +
> >> +static int imx7_gpc_pu_pgc_sw_pup_req(struct generic_pm_domain *genpd)
> >> +{
> >> + return imx7_gpc_pu_pgc_sw_pxx_req(genpd, true);
> >> +}
> >> +
> >> +static int imx7_gpc_pu_pgc_sw_pdn_req(struct generic_pm_domain *genpd)
> >> +{
> >> + return imx7_gpc_pu_pgc_sw_pxx_req(genpd, false);
> >> +}
> >> +
> >> +static struct imx7_pgc_domain imx7_pgc_domains[] = {
> >> + [IMX7_POWER_DOMAIN_USB_HSIC_PHY] = {
> >> + .genpd = {
> >> + .name = "usb-hsic-phy",
> >> + },
> >> + .bits = {
> >> + .pxx = USB_HSIC_PHY_SW_Pxx_REQ,
> >> + .map = USB_HSIC_PHY_A7_DOMAIN,
> >> + },
> >> + .voltage = 1200000,
> >> + .pgc_nctrl = 0x0d00,
> >> + },
> >> +
> >> + [IMX7_POWER_DOMAIN_USB_OTG2_PHY] = {
> >> + .genpd = {
> >> + .name = "usb-otg2-phy",
> >> + },
> >> + .bits = {
> >> + .pxx = USB_OTG2_PHY_SW_Pxx_REQ,
> >> + .map = USB_OTG2_PHY_A7_DOMAIN,
> >> + },
> >> + },
> >> +
> >> + [IMX7_POWER_DOMAIN_USB_OTG1_PHY] = {
> >> + .genpd = {
> >> + .name = "usb-otg1-phy",
> >> + },
> >> + .bits = {
> >> + .pxx = USB_OTG1_PHY_SW_Pxx_REQ,
> >> + .map = USB_OTG1_PHY_A7_DOMAIN,
> >> + },
> >> + },
> >> +
> >> + [IMX7_POWER_DOMAIN_PCIE_PHY] = {
> >> + .genpd = {
> >> + .name = "pcie-phy",
> >> + },
> >> + .bits = {
> >> + .pxx = PCIE_PHY_SW_Pxx_REQ,
> >> + .map = PCIE_PHY_A7_DOMAIN,
> >> + },
> >> + .voltage = 1000000,
> >> + .pgc_nctrl = 0x0c40,
> >> + },
> >> +
> >> + [IMX7_POWER_DOMAIN_MIPI_PHY] = {
> >> + .genpd = {
> >> + .name = "mipi-phy",
> >> + },
> >> + .bits = {
> >> + .pxx = MIPI_PHY_SW_Pxx_REQ,
> >> + .map = MIPI_PHY_A7_DOMAIN,
> >> + },
> >> + .voltage = 1000000,
> >> + .pgc_nctrl = 0x0c00,
> >> + },
> >> +};
> >> +
> >> +static int imx7_pgc_domain_probe(struct platform_device *pdev)
> >> +{
> >> + struct imx7_pgc_domain *domain = pdev->dev.platform_data;
> >> + int ret;
> >> +
> >> + domain->dev = &pdev->dev;
> >> +
> >> + ret = pm_genpd_init(&domain->genpd, NULL, true);
> >> + if (ret) {
> >> + dev_err(domain->dev, "Failed to init power domain\n");
> >> + return ret;
> >> + }
> >> +
> >> + domain->regulator = devm_regulator_get_optional(domain->dev, "power");
> >> + if (IS_ERR(domain->regulator) &&
> >> + PTR_ERR(domain->regulator) != -ENODEV) {
> >> + dev_err(domain->dev, "Failed to get domain's regulator\n");
> >> + return PTR_ERR(domain->regulator);
> >> + }
> >> +
> >> + if (!IS_ERR(domain->regulator)) {
> >> + if (!domain->voltage) {
> >> + WARN(1, "No voltage configured for domain's regulator");
> >> + return -EINVAL;
> >> + }
> >
> > This is somehow a bit mess.
> > Not sure if it's regulator API issue or our using issue...
> >
>
> You are going to have to give me something to work with. If you think
> this code can be improved please provide concrete suggestions.
>
I will look into it later.
> >> +
> >> + regulator_set_voltage(domain->regulator,
> >> + domain->voltage, domain->voltage);
> >> + }
> >> +
> >> + ret = of_genpd_add_provider_simple(domain->dev->of_node,
> >> + &domain->genpd);
> >> + if (ret) {
> >> + dev_err(domain->dev, "Failed to add genpd provider\n");
> >> + pm_genpd_remove(&domain->genpd);
> >> + }
> >> +
> >> + return ret;
> >> +}
> >> +
> >> +static int imx7_pgc_domain_remove(struct platform_device *pdev)
> >
> > here
> >
> >> +{
> >> + struct imx7_pgc_domain *domain = pdev->dev.platform_data;
> >> +
> >> + of_genpd_del_provider(domain->dev->of_node);
> >> + pm_genpd_remove(&domain->genpd);
> >> +
> >> + return 0;
> >> +}
> >> +
> >> +static const struct platform_device_id imx7_pgc_domain_id[] = {
> >> + { "imx7-pgc-domain", },
> >
> > here
> >
> >> + { },
> >> +};
> >> +
> >> +static struct platform_driver imx7_pgc_domain_driver = {
> >> + .driver = {
> >> + .name = "imx7-pgc",
> >
> > here
> >
> > All you're using imx7, not gpcv2..
> >
> > That why i prefer gpc-imx7.c
> >
>
> I am fine with this inconsistency and I am going to let Shawn decide.
>
> Shawn, please let me know if you want me to do any renaming.
>
>
> >> + },
> >> + .probe = imx7_pgc_domain_probe,
> >> + .remove = imx7_pgc_domain_remove,
> >> + .id_table = imx7_pgc_domain_id,
> >> +};
> >> +builtin_platform_driver(imx7_pgc_domain_driver)
> >
> > I did not understand too much on why we choose this approach to
> > add a power domain. AFAIK per current power domain design, a provider
> > can support multiple domains. But with this approach, actually we register
> > each domain as a provider which is actually a bit violate the real HW.
> >
> > Lucas, can you please help provide some information on this new approach
> > decision history?
> >
> >> +
> >> +static bool imx_gpcv2_readable_reg(struct device *dev, unsigned int reg)
> >> +{
> >> + return reg % 4 == 0 &&
> >> + reg <= GPC_MAX_REGISTER;
> >> +}
> >> +
> >> +static bool imx_gpcv2_volatile_reg(struct device *dev, unsigned int reg)
> >> +{
> >> + return reg == GPC_PU_PGC_SW_PUP_REQ ||
> >> + reg == GPC_PU_PGC_SW_PDN_REQ;
> >> +}
> >> +
> >> +static int imx_gpcv2_probe(struct platform_device *pdev)
> >> +{
> >> + static const struct regmap_config regmap_config = {
> >
> > If any special reason to put the structure in this function?
> >
> >> + .cache_type = REGCACHE_FLAT,
> >> + .reg_bits = 32,
> >> + .val_bits = 32,
> >> + .reg_stride = 4,
> >> +
> >> + .readable_reg = imx_gpcv2_readable_reg,
> >
> > No really need this.
> >
>
> OK, I'll drop it.
>
> >> + .volatile_reg = imx_gpcv2_volatile_reg,
> >
> > Can you check if you have the same cache issue as GPC which
> > i just sent out a series to fix?
> >
>
> It probably does. I'll adopt your fix from GPC driver.
>
> >> +
> >> + .max_register = GPC_MAX_REGISTER,
> >> + };
> >> + struct device *dev = &pdev->dev;
> >> + struct device_node *pgc_np, *np;
> >> + struct regmap *regmap;
> >> + struct resource *res;
> >> + void __iomem *base;
> >> + int ret;
> >> +
> >> + pgc_np = of_get_child_by_name(dev->of_node, "pgc");
> >> + if (!pgc_np) {
> >> + dev_err(dev, "No power domains specified in DT\n");
> >> + return -EINVAL;
> >> + }
> >> +
> >> + res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
> >> + base = devm_ioremap_resource(dev, res);
> >> + if (IS_ERR(base))
> >> + return PTR_ERR(base);
> >> +
> >> + regmap = devm_regmap_init_mmio(dev, base, ®map_config);
> >> + if (IS_ERR(regmap)) {
> >> + ret = PTR_ERR(regmap);
> >> + dev_err(dev, "failed to init regmap (%d)\n", ret);
> >> + return ret;
> >> + }
> >> +
> >> + for_each_child_of_node(pgc_np, np) {
> >> + struct platform_device *pd_pdev;
> >> + struct imx7_pgc_domain *domain;
> >> + u32 domain_index;
> >> +
> >> + ret = of_property_read_u32(np, "reg", &domain_index);
> >> + if (ret) {
> >> + dev_err(dev, "Failed to read 'reg' property\n");
> >> + of_node_put(np);
> >> + return ret;
> >> + }
> >> +
> >> + if (domain_index >= ARRAY_SIZE(imx7_pgc_domains)) {
> >> + dev_warn(dev,
> >> + "Domain index %d is out of bounds\n",
> >> + domain_index);
> >> + continue;
> >> + }
> >> +
> >> + domain = &imx7_pgc_domains[domain_index];
> >> + domain->regmap = regmap;
> >> + domain->genpd.power_on = imx7_gpc_pu_pgc_sw_pup_req;
> >> + domain->genpd.power_off = imx7_gpc_pu_pgc_sw_pdn_req;
> >
> > Is imx7_gpc_pu_pgc_sw_pup_req used by all power domains?
> > The name looks like a bit specific to PU.
> >
>
> All power domains currently supported by the driver, which is limited
> to domains controlled by GPC_PU_PGC_SW_PUP_REQ register, hence the
> name.
>
I did check the RM carefully, the power domain this patch supports
are all PU type domain. If we want to support others in the future,
a new named API may be introduced.
So, i'm okay with this name currently.
> >> +
> >> + pd_pdev = platform_device_alloc("imx7-pgc-domain",
> >> + domain_index);
> >> + if (!pd_pdev) {
> >> + dev_err(dev, "Failed to allocate platform device\n");
> >> + of_node_put(np);
> >> + return -ENOMEM;
> >> + }
> >> +
> >> + pd_pdev->dev.platform_data = domain;
> >> + pd_pdev->dev.parent = dev;
> >> + pd_pdev->dev.of_node = np;
> >> +
> >> + ret = platform_device_add(pd_pdev);
> >> + if (ret) {
> >> + platform_device_put(pd_pdev);
> >> + of_node_put(np);
> >> + return ret;
> >> + }
> >> + }
> >> +
> >> + return 0;
> >> +}
> >> +
> >> +static const struct of_device_id imx_gpcv2_dt_ids[] = {
> >> + { .compatible = "fsl,imx7d-gpc" },
> >> + { }
> >> +};
> >> +
> >> +static struct platform_driver imx_gpc_driver = {
> >> + .driver = {
> >> + .name = "imx-gpcv2",
> >> + .of_match_table = imx_gpcv2_dt_ids,
> >> + },
> >> + .probe = imx_gpcv2_probe,
> >> +};
> >> +builtin_platform_driver(imx_gpc_driver)
> >> --
> >> 2.9.3
Regards
Dong Aisheng
> >>
More information about the linux-arm-kernel
mailing list