Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753488AbcDYFhI (ORCPT ); Mon, 25 Apr 2016 01:37:08 -0400 Received: from mail-io0-f179.google.com ([209.85.223.179]:36322 "EHLO mail-io0-f179.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752099AbcDYFhF (ORCPT ); Mon, 25 Apr 2016 01:37:05 -0400 MIME-Version: 1.0 In-Reply-To: <1461319754-12040-4-git-send-email-ldewangan@nvidia.com> References: <1461319754-12040-1-git-send-email-ldewangan@nvidia.com> <1461319754-12040-4-git-send-email-ldewangan@nvidia.com> From: Alexandre Courbot Date: Mon, 25 Apr 2016 14:36:45 +0900 Message-ID: Subject: Re: [PATCH V4 4/4] gpio: tegra: Add support for gpio debounce To: Laxman Dewangan Cc: Stephen Warren , Linus Walleij , Thierry Reding , "linux-gpio@vger.kernel.org" , "linux-tegra@vger.kernel.org" , Linux Kernel Mailing List Content-Type: text/plain; charset=UTF-8 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 7471 Lines: 181 Sorry, just realized I commented on v3... On Fri, Apr 22, 2016 at 7:09 PM, Laxman Dewangan wrote: > NVIDIA's Tegra210 support the HW debounce in the GPIO controller > for all its GPIO pins. > > Add support for setting debounce timing by implementing the > set_debounce callback of gpiochip. > > Signed-off-by: Laxman Dewangan > > --- > Changes from V1: > - Write debounce count before enable. > - Make sure the debounce count do not have any boot residuals. > > Changes from V2: > - Only access register for debounce when SoC support debounce. > > Changes from V3: > - Add locking mechanism in debounce count register update. > - Move DBC register from prev patch to here. > > drivers/gpio/gpio-tegra.c | 69 ++++++++++++++++++++++++++++++++++++++++++++++- > 1 file changed, 68 insertions(+), 1 deletion(-) > > diff --git a/drivers/gpio/gpio-tegra.c b/drivers/gpio/gpio-tegra.c > index 6af5eb2..45d80ec 100644 > --- a/drivers/gpio/gpio-tegra.c > +++ b/drivers/gpio/gpio-tegra.c > @@ -46,10 +46,13 @@ > #define GPIO_INT_ENB(t, x) (GPIO_REG(t, x) + 0x50) > #define GPIO_INT_LVL(t, x) (GPIO_REG(t, x) + 0x60) > #define GPIO_INT_CLR(t, x) (GPIO_REG(t, x) + 0x70) > +#define GPIO_DBC_CNT(t, x) (GPIO_REG(t, x) + 0xF0) > + > > #define GPIO_MSK_CNF(t, x) (GPIO_REG(t, x) + t->soc->upper_offset + 0x00) > #define GPIO_MSK_OE(t, x) (GPIO_REG(t, x) + t->soc->upper_offset + 0x10) > #define GPIO_MSK_OUT(t, x) (GPIO_REG(t, x) + t->soc->upper_offset + 0X20) > +#define GPIO_MSK_DBC_EN(t, x) (GPIO_REG(t, x) + t->soc->upper_offset + 0x30) > #define GPIO_MSK_INT_STA(t, x) (GPIO_REG(t, x) + t->soc->upper_offset + 0x40) > #define GPIO_MSK_INT_ENB(t, x) (GPIO_REG(t, x) + t->soc->upper_offset + 0x50) > #define GPIO_MSK_INT_LVL(t, x) (GPIO_REG(t, x) + t->soc->upper_offset + 0x60) > @@ -67,6 +70,7 @@ struct tegra_gpio_bank { > int bank; > int irq; > spinlock_t lvl_lock[4]; > + spinlock_t dbc_lock[4]; /* Lock for updating debounce count register */ I'm nit'ing here, but maybe one spinlock shared by all ports would be enough? (the same would apply to lvl_lock, so feel free to do this as a separate patch) I don't think we expect *that* many concurrent accesses, do we? > #ifdef CONFIG_PM_SLEEP > u32 cnf[4]; > u32 out[4]; > @@ -74,11 +78,14 @@ struct tegra_gpio_bank { > u32 int_enb[4]; > u32 int_lvl[4]; > u32 wake_enb[4]; > + u32 dbc_enb[4]; > #endif > + u32 dbc_cnt[4]; > struct tegra_gpio_info *tgi; > }; > > struct tegra_gpio_soc_config { > + bool debounce_supported; > u32 bank_stride; > u32 upper_offset; > }; > @@ -182,6 +189,38 @@ static int tegra_gpio_direction_output(struct gpio_chip *chip, unsigned offset, > return 0; > } > > +static int tegra_gpio_set_debounce(struct gpio_chip *chip, unsigned int offset, > + unsigned int debounce) > +{ > + struct tegra_gpio_info *tgi = gpiochip_get_data(chip); > + unsigned int debounce_ms = DIV_ROUND_UP(debounce, 1000); > + unsigned long flags; > + int port = GPIO_PORT(offset); > + int bank = GPIO_BANK(offset); Maybe declare "bank" as follows: struct tegra_gpio_bank *bank = &tgi->bank_info[GPIO_BANK(offset)]; This will allow you to simplify the code that follows: > + > + if (!debounce_ms) { > + tegra_gpio_mask_write(tgi, GPIO_MSK_DBC_EN(tgi, offset), > + offset, 0); > + return 0; > + } > + > + debounce_ms = min(debounce_ms, 255U); > + > + /* There is only one debounce count register per port and hence > + * set the maximum of current and requested debounce time. > + */ > + spin_lock_irqsave(&tgi->bank_info[bank].dbc_lock[port], flags); > + if (tgi->bank_info[bank].dbc_cnt[port] < debounce_ms) { > + tegra_gpio_writel(tgi, debounce_ms, GPIO_DBC_CNT(tgi, offset)); > + tgi->bank_info[bank].dbc_cnt[port] = debounce_ms; > + } > + spin_unlock_irqrestore(&tgi->bank_info[bank].dbc_lock[port], flags); Becomes: spin_lock_irqsave(bank->dbc_lock[port], flags); if (bank->dbc_cnt[port] < debounce_ms) { tegra_gpio_writel(tgi, debounce_ms, GPIO_DBC_CNT(tgi, offset)); bank->dbc_cnt[port] = debounce_ms; } spin_unlock_irqrestore(&bank->dbc_lock[port], flags); Which is nicer to the eyes. Extra points if you initialize port and bank after we ensure that debounce_ms is not zero and that their value will actually be used. > + > + tegra_gpio_mask_write(tgi, GPIO_MSK_DBC_EN(tgi, offset), offset, 1); > + > + return 0; > +} > + > static int tegra_gpio_to_irq(struct gpio_chip *chip, unsigned offset) > { > struct tegra_gpio_info *tgi = gpiochip_get_data(chip); > @@ -197,6 +236,7 @@ static struct gpio_chip tegra_gpio_chip = { > .get = tegra_gpio_get, > .direction_output = tegra_gpio_direction_output, > .set = tegra_gpio_set, > + .set_debounce = tegra_gpio_set_debounce, > .to_irq = tegra_gpio_to_irq, > .base = 0, > }; > @@ -360,6 +400,14 @@ static int tegra_gpio_resume(struct device *dev) > unsigned int gpio = (b<<5) | (p<<3); > tegra_gpio_writel(tgi, bank->cnf[p], > GPIO_CNF(tgi, gpio)); > + > + if (tgi->soc->debounce_supported) { > + tegra_gpio_writel(tgi, bank->dbc_cnt[p], > + GPIO_DBC_CNT(tgi, gpio)); > + tegra_gpio_writel(tgi, bank->dbc_enb[p], > + GPIO_MSK_DBC_EN(tgi, gpio)); > + } > + > tegra_gpio_writel(tgi, bank->out[p], > GPIO_OUT(tgi, gpio)); > tegra_gpio_writel(tgi, bank->oe[p], > @@ -395,6 +443,13 @@ static int tegra_gpio_suspend(struct device *dev) > GPIO_OUT(tgi, gpio)); > bank->oe[p] = tegra_gpio_readl(tgi, > GPIO_OE(tgi, gpio)); > + if (tgi->soc->debounce_supported) { > + bank->dbc_enb[p] = tegra_gpio_readl(tgi, > + GPIO_MSK_DBC_EN(tgi, gpio)); > + bank->dbc_enb[p] = (bank->dbc_enb[p] << 8) | > + bank->dbc_enb[p]; > + } > + > bank->int_enb[p] = tegra_gpio_readl(tgi, > GPIO_INT_ENB(tgi, gpio)); > bank->int_lvl[p] = tegra_gpio_readl(tgi, > @@ -547,6 +602,9 @@ static int tegra_gpio_probe(struct platform_device *pdev) > > platform_set_drvdata(pdev, tgi); > > + if (!config->debounce_supported) > + tgi->gc->set_debounce = NULL; Yep, we really want one gpio_chip instance in the tegra_gpio_info struct, otherwise this kind of limits the purpose of getting rid of these global variables...