Re: [PATCH] i2c: tegra: Add i2c support

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Sun, Feb 20, 2011 at 11:49:56AM -0800, Colin Cross wrote:
> On Tue, Feb 15, 2011 at 3:48 PM, Ben Dooks <ben-i2c@xxxxxxxxx> wrote:
> > On 08/02/11 12:44, Mark Brown wrote:
> >> From: Colin Cross <ccross@xxxxxxxxxxx>
> >
> > Some sort of explanation here would have been nice.
> Will fix
> 
> >> Signed-off-by: Colin Cross <ccross@xxxxxxxxxxx>
> >> Signed-off-by: Mark Brown <broonie@xxxxxxxxxxxxxxxxxxxxxxxxxxx>
> >> ---
> >>  drivers/i2c/busses/Kconfig     |    7 +
> >>  drivers/i2c/busses/Makefile    |    1 +
> >>  drivers/i2c/busses/i2c-tegra.c |  665 ++++++++++++++++++++++++++++++++++++++++
> >>  include/linux/i2c-tegra.h      |   25 ++
> >>  4 files changed, 698 insertions(+), 0 deletions(-)
> >>  create mode 100644 drivers/i2c/busses/i2c-tegra.c
> >>  create mode 100644 include/linux/i2c-tegra.h
> >>
> >> diff --git a/drivers/i2c/busses/Kconfig b/drivers/i2c/busses/Kconfig
> >> index 30f8dbd..b76a2a4 100644
> >> --- a/drivers/i2c/busses/Kconfig
> >> +++ b/drivers/i2c/busses/Kconfig
> >
> > Fine.
> >
> >> diff --git a/drivers/i2c/busses/Makefile b/drivers/i2c/busses/Makefile
> >> index 3c630b7..f505253 100644
> >> --- a/drivers/i2c/busses/Makefile
> >> +++ b/drivers/i2c/busses/Makefile
> >
> > Fine.
> >
> >> diff --git a/drivers/i2c/busses/i2c-tegra.c b/drivers/i2c/busses/i2c-tegra.c
> >> new file mode 100644
> >> index 0000000..cfa0084
> >> --- /dev/null
> >> +++ b/drivers/i2c/busses/i2c-tegra.c
> >> @@ -0,0 +1,665 @@
> >> +/*
> >> + * drivers/i2c/busses/i2c-tegra.c
> >> + *
> >> + * Copyright (C) 2010 Google, Inc.
> >> + * Author: Colin Cross <ccross@xxxxxxxxxxx>
> >> + *
> >> + * This software is licensed under the terms of the GNU General Public
> >> + * License version 2, as published by the Free Software Foundation, and
> >> + * may be copied, distributed, and modified under those terms.
> >> + *
> >> + * 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/kernel.h>
> >> +#include <linux/init.h>
> >> +#include <linux/platform_device.h>
> >> +#include <linux/clk.h>
> >> +#include <linux/i2c.h>
> >> +#include <linux/io.h>
> >> +#include <linux/interrupt.h>
> >> +#include <linux/delay.h>
> >> +#include <linux/slab.h>
> >> +#include <linux/i2c-tegra.h>
> >> +
> >> +#include <asm/unaligned.h>
> >> +
> >> +#include <mach/clk.h>
> >
> > I don't think you should be including
> mach/clk.h provides the definitions of tegra_assert_periph_reset.  The
> Tegra resets are part of the clock controller, and the I2C bus
> requires resets at various times, including after a NACK.  It does not
> export any of the private clock implementation.
> 
> >> +
> >> +struct tegra_i2c_dev {
> >> +     struct device *dev;
> >> +     struct i2c_adapter adapter;
> >> +     struct clk *clk;
> >> +     struct clk *i2c_clk;
> >> +     struct resource *iomem;
> >> +     void __iomem *base;
> >> +     int cont_id;
> >> +     int irq;
> >> +     int is_dvc;
> >> +     struct completion msg_complete;
> >> +     int msg_err;
> >> +     u8 *msg_buf;
> >> +     size_t msg_buf_remaining;
> >> +     int msg_read;
> >> +     int msg_transfer_complete;
> >> +     unsigned long bus_clk_rate;
> >> +     bool is_suspended;
> >> +};
> >
> > would have been nice to have some kerneldoc for this.
> Will do
> 
> >> +/*
> >> + * i2c_writel and i2c_readl will offset the register if necessary to talk
> >> + * to the I2C block inside the DVC block
> >> + */
> >> +static void i2c_writel(struct tegra_i2c_dev *i2c_dev, u32 val, unsigned long reg)
> >> +{
> >> +     if (i2c_dev->is_dvc)
> >> +             reg += (reg >= I2C_TX_FIFO) ? 0x10 : 0x40;
> >> +     writel(val, i2c_dev->base + reg);
> >> +}
> >> +
> >> +static u32 i2c_readl(struct tegra_i2c_dev *i2c_dev, unsigned long reg)
> >> +{
> >> +     if (i2c_dev->is_dvc)
> >> +             reg += (reg >= I2C_TX_FIFO) ? 0x10 : 0x40;
> >> +     return readl(i2c_dev->base + reg);
> >> +}
> >
> > as a note, would have been better to wrap up the address changes into
> > one place to ensure that they're consistent.
> Will fix
> 
> > [snip]
> >
> >> +static void tegra_i2c_set_clk(struct tegra_i2c_dev *i2c_dev, unsigned int freq)
> >> +{
> >> +     clk_set_rate(i2c_dev->clk, freq * 8);
> >> +}
> >
> > hmm, really need a separate function?
> Will fix
> 
> >> +static int tegra_i2c_flush_fifos(struct tegra_i2c_dev *i2c_dev)
> >> +{
> >> +     unsigned long timeout = jiffies + HZ;
> >> +     u32 val = i2c_readl(i2c_dev, I2C_FIFO_CONTROL);
> >> +     val |= I2C_FIFO_CONTROL_TX_FLUSH | I2C_FIFO_CONTROL_RX_FLUSH;
> >> +     i2c_writel(i2c_dev, val, I2C_FIFO_CONTROL);
> >> +
> >> +     while (i2c_readl(i2c_dev, I2C_FIFO_CONTROL) &
> >> +             (I2C_FIFO_CONTROL_TX_FLUSH | I2C_FIFO_CONTROL_RX_FLUSH)) {
> >> +             if (time_after(jiffies, timeout)) {
> >> +                     dev_warn(i2c_dev->dev, "timeout waiting for fifo flush\n");
> >> +                     return -ETIMEDOUT;
> >> +             }
> >> +             msleep(1);
> >> +     }
> >> +     return 0;
> >> +}
> >> +
> >> +static int tegra_i2c_empty_rx_fifo(struct tegra_i2c_dev *i2c_dev)
> >> +{
> >> +     u32 val;
> >> +     int rx_fifo_avail;
> >> +     int word;
> >> +     u8 *buf = i2c_dev->msg_buf;
> >> +     size_t buf_remaining = i2c_dev->msg_buf_remaining;
> >> +     int words_to_transfer;
> >> +
> >> +     val = i2c_readl(i2c_dev, I2C_FIFO_STATUS);
> >> +     rx_fifo_avail = (val & I2C_FIFO_STATUS_RX_MASK) >>
> >> +             I2C_FIFO_STATUS_RX_SHIFT;
> >> +
> >> +     words_to_transfer = buf_remaining / BYTES_PER_FIFO_WORD;
> >> +     if (words_to_transfer > rx_fifo_avail)
> >> +             words_to_transfer = rx_fifo_avail;
> >> +
> >> +     for (word = 0; word < words_to_transfer; word++) {
> >> +             val = i2c_readl(i2c_dev, I2C_RX_FIFO);
> >> +             put_unaligned_le32(val, buf);
> >> +             buf += BYTES_PER_FIFO_WORD;
> >> +             buf_remaining -= BYTES_PER_FIFO_WORD;
> >> +             rx_fifo_avail--;
> >> +     }
> >
> > you know, there's a readsl() function that does this.
> I think readsl can't handle the possibly unaligned buf pointer.

it does. see arch/arm/lib/io-writesl.S for proof.

> > this whole function gives me the creeps, is there any reason why
> > we can't use the readsl or similar functions for this?
> Same here - readsl can't handle the alignment requirements, readsb
> can't  handle the required 32 bit register read, and the bytes in the
> same word but after the end buf may not be part of buf, so byte writes
> to buf are required.

You'll find it can, too. arch/arm/lib/io-readsl.S.

[snip]
 
> >> +static int tegra_i2c_init(struct tegra_i2c_dev *i2c_dev)
> >> +{
> >> +     u32 val;
> >> +     int err = 0;
> >> +
> >> +     clk_enable(i2c_dev->clk);
> >> +
> >> +     tegra_periph_reset_assert(i2c_dev->clk);
> >> +     udelay(2);
> >> +     tegra_periph_reset_deassert(i2c_dev->clk);
> >> +
> >> +     if (i2c_dev->is_dvc)
> >> +             tegra_dvc_init(i2c_dev);
> >> +
> >> +     val = I2C_CNFG_NEW_MASTER_FSM | I2C_CNFG_PACKET_MODE_EN;
> >> +     i2c_writel(i2c_dev, val, I2C_CNFG);
> >> +     i2c_writel(i2c_dev, 0, I2C_INT_MASK);
> >> +     tegra_i2c_set_clk(i2c_dev, i2c_dev->bus_clk_rate);
> >> +
> >> +     val = 7 << I2C_FIFO_CONTROL_TX_TRIG_SHIFT |
> >> +             0 << I2C_FIFO_CONTROL_RX_TRIG_SHIFT;
> >> +     i2c_writel(i2c_dev, val, I2C_FIFO_CONTROL);
> >> +
> >> +     if (tegra_i2c_flush_fifos(i2c_dev))
> >> +             err = -ETIMEDOUT;
> >> +
> >> +     clk_disable(i2c_dev->clk);
> >> +     return 0;
> >> +}
> >
> > err never gets returned. please remove or return.
> Should be returned, will fix.

good. unused return codes make maintainers cry.
 
> >> +static irqreturn_t tegra_i2c_isr(int irq, void *dev_id)
> >> +{
> >> +     u32 status;
> >> +     const u32 status_err = I2C_INT_NO_ACK | I2C_INT_ARBITRATION_LOST;
> >> +     struct tegra_i2c_dev *i2c_dev = dev_id;
> >> +
> >> +     status = i2c_readl(i2c_dev, I2C_INT_STATUS);
> >> +
> >> +     if (status == 0) {
> >> +             dev_warn(i2c_dev->dev, "irq status 0 %08x\n",
> >> +                     i2c_readl(i2c_dev, I2C_PACKET_TRANSFER_STATUS));
> >> +             return IRQ_HANDLED;
> >
> > maybe return IRQ_UNHANDED?
> Sure - there are some poorly understood cases where the I2C controller
> can assert its interrupt with no status bits, but the handling to
> detect it and reset the controller is not in this patch, so dev_warn
> and IRQ_UNHANDLED is better.

I'm not too concerned if you have `handled` it, but please note what
is happening.

> >> +     if (WARN_ON(ret == 0)) {
> >
> > suppose a WARN_ON() is OK here.
> >
> >
> >> +             dev_err(i2c_dev->dev, "i2c transfer timed out\n");
> >> +
> >> +             tegra_i2c_init(i2c_dev);
> >> +             return -ETIMEDOUT;
> >> +     }
> >> +
> >> +     pr_debug("transfer complete: %d %d %d\n", ret, completion_done(&i2c_dev->msg_complete), i2c_dev->msg_err);
> >
> > not dev_dbg()?
> Will fix
 
> >> +     if (likely(i2c_dev->msg_err == I2C_ERR_NONE))
> >> +             return 0;
> >> +
> >> +     tegra_i2c_init(i2c_dev);
> >> +     if (i2c_dev->msg_err == I2C_ERR_NO_ACK) {
> >> +             if (msg->flags & I2C_M_IGNORE_NAK)
> >> +                     return 0;
> >> +             return -EREMOTEIO;
> >> +     }
> >> +
> >> +     return -EIO;
> >> +}
> >
> > think that's fine.
> >
> > [snip]
> >
> >> +static int tegra_i2c_probe(struct platform_device *pdev)
> >> +{
> >> +     struct tegra_i2c_dev *i2c_dev;
> >> +     struct tegra_i2c_platform_data *pdata = pdev->dev.platform_data;
> >> +     struct resource *res;
> >> +     struct resource *iomem;
> >> +     struct clk *clk;
> >> +     struct clk *i2c_clk;
> >> +     void *base;
> >> +     int irq;
> >> +     int ret = 0;
> >> +
> >> +     res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
> >> +     if (!res) {
> >> +             dev_err(&pdev->dev, "no mem resource?\n");
> >> +             return -ENODEV;
> >> +     }
> >
> > -ENODEV gets ignored by the upper layer, maybe find something else to
> > return.
> I'll return -EINVAL

I really should get around to finish writing a good guide to this.
 
> >> +     iomem = request_mem_region(res->start, resource_size(res), pdev->name);
> >> +     if (!iomem) {
> >> +             dev_err(&pdev->dev, "I2C region already claimed\n");
> >> +             return -EBUSY;
> >> +     }
> >> +
> >> +     base = ioremap(iomem->start, resource_size(iomem));
> >> +     if (!base) {
> >> +             dev_err(&pdev->dev, "Can't ioremap I2C region\n");
> >
> > would prefer Cannot
> Will fix
> 
> >> +             return -ENOMEM;
> >> +     }
> >> +
> >> +     res = platform_get_resource(pdev, IORESOURCE_IRQ, 0);
> >> +     if (!res) {
> >> +             dev_err(&pdev->dev, "no irq resource?\n");
> >
> > No need for ? after this.
> Will fix
> 
> >> +             ret = -ENODEV;
> >
> > and see again.
> Will fix
> 
> >> +             goto err_iounmap;
> >> +     }
> >> +     irq = res->start;
> >> +
> >> +     clk = clk_get(&pdev->dev, NULL);
> >> +     if (!clk) {
> >
> > you've departed from no dev_err() here.
> Will fix
> 
> >> +             ret = -ENOMEM;
> >> +             goto err_release_region;
> >> +     }
> >> +
> >
> >
> >> +     i2c_clk = clk_get(&pdev->dev, "i2c");
> >> +     if (!i2c_clk) {
> >
> > see again.
> Will fix
> 
> >> +             ret = -ENOMEM;
> >> +             goto err_clk_put;
> >> +     }
> >> +
> >> +     i2c_dev = kzalloc(sizeof(struct tegra_i2c_dev), GFP_KERNEL);
> >> +     if (!i2c_dev) {
> >> +             ret = -ENOMEM;
> >> +             goto err_i2c_clk_put;
> >> +     }
> >> +
> >> +     i2c_dev->base = base;
> >> +     i2c_dev->clk = clk;
> >> +     i2c_dev->i2c_clk = i2c_clk;
> >> +     i2c_dev->iomem = iomem;
> >> +     i2c_dev->adapter.algo = &tegra_i2c_algo;
> >> +     i2c_dev->irq = irq;
> >> +     i2c_dev->cont_id = pdev->id;
> >> +     i2c_dev->dev = &pdev->dev;
> >> +     i2c_dev->bus_clk_rate = pdata ? pdata->bus_clk_rate : 100000;
> >> +
> >> +     if (pdev->id == 3)
> >> +             i2c_dev->is_dvc = 1;
> >> +     init_completion(&i2c_dev->msg_complete);
> >> +
> >> +     platform_set_drvdata(pdev, i2c_dev);
> >> +
> >> +     ret = tegra_i2c_init(i2c_dev);
> >> +     if (ret)
> >> +             goto err_free;
> >
> > does tegra_i2c_init() print an appropriate error?
> Will fix
> 
> >> +
> >> +     ret = request_irq(i2c_dev->irq, tegra_i2c_isr, IRQF_DISABLED,
> >> +             pdev->name, i2c_dev);
> >
> > hmm, do you really want to be running IRQF_DISABLED?
> No, I'll drop it.
> 
> >> +     if (ret) {
> >> +             dev_err(&pdev->dev, "Failed to request irq %i\n", i2c_dev->irq);
> >> +             goto err_free;
> >> +     }
> >> +
> >> +     clk_enable(i2c_dev->i2c_clk);
> >> +
> >> +     i2c_set_adapdata(&i2c_dev->adapter, i2c_dev);
> >> +     i2c_dev->adapter.owner = THIS_MODULE;
> >> +     i2c_dev->adapter.class = I2C_CLASS_HWMON;
> >> +     strlcpy(i2c_dev->adapter.name, "Tegra I2C adapter",
> >> +             sizeof(i2c_dev->adapter.name));
> >> +     i2c_dev->adapter.algo = &tegra_i2c_algo;
> >> +     i2c_dev->adapter.dev.parent = &pdev->dev;
> >> +     i2c_dev->adapter.nr = pdev->id;
> >> +
> >> +     ret = i2c_add_numbered_adapter(&i2c_dev->adapter);
> >> +     if (ret) {
> >> +             dev_err(&pdev->dev, "Failed to add I2C adapter\n");
> >> +             goto err_free_irq;
> >> +     }
> >> +
> >> +     return 0;
> >> +err_free_irq:
> >> +     free_irq(i2c_dev->irq, i2c_dev);
> >> +err_free:
> >> +     kfree(i2c_dev);
> >> +err_i2c_clk_put:
> >> +     clk_put(i2c_clk);
> >> +err_clk_put:
> >> +     clk_put(clk);
> >> +err_release_region:
> >> +     release_mem_region(iomem->start, resource_size(iomem));
> >> +err_iounmap:
> >> +     iounmap(base);
> >> +     return ret;
> >> +}
> >
> > [snip]
> >
> > how about MODULE_* decelerations at the end of this, with license, etc?
> Will fix
> >
> >> diff --git a/include/linux/i2c-tegra.h b/include/linux/i2c-tegra.h
> >> new file mode 100644
> >> index 0000000..9c85da4
> >> --- /dev/null
> >
> > looks ok.
> >
> > could do with a few tidies before inclusion.
> >
> > --
> > Ben
> >
> >

-- 
Ben Dooks, ben@xxxxxxxxx, http://www.fluff.org/ben/

Large Hadron Colada: A large Pina Colada that makes the universe disappear.

--
To unsubscribe from this list: send the line "unsubscribe linux-i2c" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html


[Index of Archives]     [Linux GPIO]     [Linux SPI]     [Linux Hardward Monitoring]     [LM Sensors]     [Linux USB Devel]     [Linux Media]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]

  Powered by Linux