[PATCH 1/3] Added support for On Chip OTP in i.MX23/28
Maxime Ripard
maxime.ripard at free-electrons.com
Fri Jul 5 04:03:51 EDT 2013
Hi Christoph,
On Wed, Jul 03, 2013 at 02:39:12PM +0200, Christoph G. Baumann wrote:
> From: "Christoph G. Baumann" <cb at sgoc.de>
>
> Signed-off-by: Christoph G. Baumann <cb at sgoc.de>
> ---
> drivers/misc/fsl_otp.c | 249 ++++++++++++++++++++++++++++++++++++++++++++++++
> drivers/misc/fsl_otp.h | 201 ++++++++++++++++++++++++++++++++++++++
> 2 files changed, 450 insertions(+)
> create mode 100644 drivers/misc/fsl_otp.c
> create mode 100644 drivers/misc/fsl_otp.h
>
> diff --git a/drivers/misc/fsl_otp.c b/drivers/misc/fsl_otp.c
> new file mode 100644
> index 0000000..fd1941a
> --- /dev/null
> +++ b/drivers/misc/fsl_otp.c
> @@ -0,0 +1,249 @@
> +/*
> + * Freescale On-Chip OTP driver
> + *
> + * Copyright (C) 2010 Freescale Semiconductor, Inc. All Rights Reserved.
> + *
> + * Ported to 3.7 by Christoph G. Baumann <cgb at debian.org>
Maybe it's not worth mentionning the version here, since it's wrong
anyway.
> + * 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 of the License, 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.
> + *
> + * You should have received a copy of the GNU General Public License
> + * along with this program; if not, write to the Free Software
> + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
> + */
> +#include <linux/kobject.h>
> +#include <linux/string.h>
> +#include <linux/sysfs.h>
> +#include <linux/module.h>
> +#include <linux/init.h>
> +#include <linux/delay.h>
> +#include <linux/fcntl.h>
> +#include <linux/mutex.h>
> +#include <linux/clk.h>
> +#include <linux/err.h>
> +#include <linux/io.h>
> +#include <linux/slab.h>
> +#include <linux/platform_device.h>
> +
> +#include "fsl_otp.h"
> +
> +static DEFINE_MUTEX(otp_mutex);
> +static struct kobject *otp_kobj;
> +static struct attribute **attrs;
> +static struct kobj_attribute *kattr;
> +static struct attribute_group attr_group;
> +
> +static inline unsigned int get_reg_index(struct kobj_attribute *tmp)
> +{
> + return tmp - kattr;
> +}
> +
> +static int otp_wait_busy(u32 flags)
> +{
> + int count;
> + u32 c;
> +
> + for (count = 10000; count >= 0; count--) {
> + c = __raw_readl(REGS_OCOTP_BASE + HW_OCOTP_CTRL);
> + if (!(c & (BM_OCOTP_CTRL_BUSY | BM_OCOTP_CTRL_ERROR | flags)))
> + break;
> + cpu_relax();
> + }
> +
> + if (count < 0)
> + return -ETIMEDOUT;
> + return 0;
> +}
> +
> +static ssize_t otp_show(struct kobject *kobj, struct kobj_attribute *attr,
> + char *buf)
> +{
> + unsigned int index = get_reg_index(attr);
> + u32 value = 0;
> +
> + /* sanity check */
> + if (index >= otp_data.fuse_num)
> + return 0;
> +
> + mutex_lock(&otp_mutex);
> +
> + if (otp_read_prepare()) {
> + mutex_unlock(&otp_mutex);
> + return 0;
> + }
> + value = __raw_readl(REGS_OCOTP_BASE + HW_OCOTP_CUST_N(index));
> + otp_read_post();
> +
> + mutex_unlock(&otp_mutex);
> + return sprintf(buf, "0x%x\n", value);
> +}
> +
> +static int otp_write_bits(int addr, u32 data, u32 magic)
> +{
> + u32 c; /* for control register */
> +
> + /* init the control register */
> + c = __raw_readl(REGS_OCOTP_BASE + HW_OCOTP_CTRL);
> + c &= ~BM_OCOTP_CTRL_ADDR;
> + c |= BF(addr, OCOTP_CTRL_ADDR);
> + c |= BF(magic, OCOTP_CTRL_WR_UNLOCK);
> + __raw_writel(c, REGS_OCOTP_BASE + HW_OCOTP_CTRL);
> +
> + /* init the data register */
> + __raw_writel(data, REGS_OCOTP_BASE + HW_OCOTP_DATA);
You shouldn't use __raw_readl/__raw_writel directly here, but
readl/writel. The __raw_* functions lack the endianness conversions and
the memory barrier readl has.
> + otp_wait_busy(0);
> +
> + mdelay(2); /* Write Postamble */
Why do you need the mdelay after your call to otp_wait_busy ?
> + return 0;
> +}
> +
> +static ssize_t otp_store(struct kobject *kobj, struct kobj_attribute *attr,
> + const char *buf, size_t count)
> +{
> + unsigned int index = get_reg_index(attr);
> + int value;
> +
> + /* sanity check */
> + if (index >= otp_data.fuse_num)
> + return 0;
> +
> + sscanf(buf, "0x%x", &value);
> +
> + mutex_lock(&otp_mutex);
> + if (otp_write_prepare()) {
> + mutex_unlock(&otp_mutex);
> + return 0;
> + }
> + otp_write_bits(index, value, 0x3e77);
What's that magic value?
> + otp_write_post();
> + mutex_unlock(&otp_mutex);
> +
> + return count;
> +}
> +
> +static void free_otp_attr(void)
> +{
> + kfree(attrs);
> + attrs = NULL;
> +
> + kfree(kattr);
> + kattr = NULL;
> +}
> +
> +static int alloc_otp_attr(void)
> +{
> + int i;
> +
> + /* The last one is NULL, which is used to detect the end */
> + attrs = kzalloc((otp_data.fuse_num + 1) * sizeof(attrs[0]),
> + GFP_KERNEL);
> + kattr = kzalloc(otp_data.fuse_num * sizeof(struct kobj_attribute),
> + GFP_KERNEL);
> +
> + if (!attrs || !kattr)
> + goto error_out;
> +
> + for (i = 0; i < otp_data.fuse_num; i++) {
> + kattr[i].attr.name = otp_data.fuse_name[i];
> + kattr[i].attr.mode = 0600;
> + kattr[i].show = otp_show;
> + kattr[i].store = otp_store;
> + attrs[i] = &kattr[i].attr;
> + sysfs_attr_init(attrs[i]);
> + }
> + memset(&attr_group, 0, sizeof(attr_group));
> + attr_group.attrs = attrs;
> + return 0;
> +
> +error_out:
> + free_otp_attr();
> + return -ENOMEM;
> +}
> +
> +static int fsl_otp_probe(struct platform_device *pdev)
Your function names could use some common prefix, it's a bit messy for
now. You have:
- free_otp_*
- otp_*
- fsl_otp_*
You could use the prefix mxs_otp for all your functions, for example.
> +{
> + int retval;
> + struct fsl_otp_data *pdata;
> +
> +
> + pdata = &otp_data;
> + if (pdev->dev.platform_data == NULL)
> + pdev->dev.platform_data = &otp_data;
Where is that otp_data coming from?
> +
> + retval = alloc_otp_attr();
> + if (retval)
> + return retval;
> +
> + retval = map_ocotp_addr(pdev);
> + if (retval)
> + goto error;
> +
> + otp_kobj = kobject_create_and_add("fsl_otp", NULL);
> + if (!otp_kobj) {
> + retval = -ENOMEM;
> + goto error;
> + }
> +
> + retval = sysfs_create_group(otp_kobj, &attr_group);
> + if (retval)
> + goto error;
> +
> + mutex_init(&otp_mutex);
> + return 0;
> +error:
> + kobject_put(otp_kobj);
> + otp_kobj = NULL;
> + free_otp_attr();
> + unmap_ocotp_addr();
Where is that unmap_ocotp_addr function declared?
> + return retval;
> +}
> +
> +static int otp_remove(struct platform_device *pdev)
> +{
> + kobject_put(otp_kobj);
> + otp_kobj = NULL;
> +
> + free_otp_attr();
> + unmap_ocotp_addr();
Ditto.
> + return 0;
> +}
> +
> +static const struct of_device_id mxs_otp_dt_ids[] = {
> + { .compatible = "fsl,ocotp", .data = NULL, },
> + { /* sentinel */ }
> +};
> +MODULE_DEVICE_TABLE(of, mxs_otp_dt_ids);
> +
> +static struct platform_driver ocotp_driver = {
> + .probe = fsl_otp_probe,
> + .remove = otp_remove,
> + .driver = {
> + .name = "ocotp",
> + .owner = THIS_MODULE,
> + .of_match_table = mxs_otp_dt_ids,
> + },
> +};
> +
> +static int __init fsl_otp_init(void)
> +{
> + return platform_driver_register(&ocotp_driver);
> +}
> +
> +static void __exit fsl_otp_exit(void)
> +{
> + platform_driver_unregister(&ocotp_driver);
> +}
> +module_init(fsl_otp_init);
> +module_exit(fsl_otp_exit);
You can use module_platform_driver here to remove all this boilerplate
code.
> +MODULE_LICENSE("GPL");
> +MODULE_AUTHOR("Huang Shijie <b32955 at freescale.com>");
> +MODULE_DESCRIPTION("Common driver for OTP controller");
> diff --git a/drivers/misc/fsl_otp.h b/drivers/misc/fsl_otp.h
> new file mode 100644
> index 0000000..b429479
> --- /dev/null
> +++ b/drivers/misc/fsl_otp.h
> @@ -0,0 +1,201 @@
> +/*
> + * Copyright (C) 2010 Freescale Semiconductor, Inc. All Rights Reserved.
> + *
> + * Ported to 3.7 by Christoph G. Baumann <cgb at debian.org>
> + *
> + * 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 of the License, 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.
> + *
> + * You should have received a copy of the GNU General Public License along
> + * with this program; if not, write to the Free Software Foundation, Inc.,
> + * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
> + */
> +#ifndef __FREESCALE_OTP__
> +#define __FREESCALE_OTP__
> +
> +
> +static int otp_wait_busy(u32 flags);
> +
> +/* IMX23 and IMX28 share most of the defination ========================= */
> +#if (defined(CONFIG_SOC_IMX23) || defined(CONFIG_SOC_IMX28))
> +
> +#if defined(CONFIG_SOC_IMX28) && defined(CONFIG_SOC_IMX23)
> +#undef CONFIG_SOC_IMX23
> +#endif
> +
> +#include <linux/regulator/consumer.h>
> +#include <mach/hardware.h>
> +#include <mach/mxs.h>
> +
> +#if defined(CONFIG_SOC_IMX23)
> +#include <mach/mx23.h>
> +#include <mach/mx23-ocotp.h>
> +#else
> +#include <mach/mx28.h>
> +#include <mach/mx28-ocotp.h>
> +#endif
Are you even sure it compiles? These headers don't exist.
Also, now that the mxs platform enabled multi-platform support, you have
to keep in mind that this driver can be used on both an imx23 *and* an
imx28 at run time, so all these compile-time options should be removed.
> +#define OCOTP_PHYS_ADDR MXS_OCOTP_BASE_ADDR
> +#define REGS_OCOTP_BASE (MXS_IO_ADDRESS(OCOTP_PHYS_ADDR))
> +#define BF(value, field) (((value) << BP_##field) & BM_##field)
> +
> +#if defined(CONFIG_SOC_IMX28)
> +/* Building up eight registers's names of a bank */
> +#define MXS_OTP_BANK(a, b, c, d, e, f, g, h) \
> + {\
> + ("HW_OCOTP_"#a), ("HW_OCOTP_"#b), ("HW_OCOTP_"#c), ("HW_OCOTP_"#d), \
> + ("HW_OCOTP_"#e), ("HW_OCOTP_"#f), ("HW_OCOTP_"#g), ("HW_OCOTP_"#h) \
> + }
> +
> +#define MXS_OTP_BANKS (5)
> +#define MXS_OTP_BANK_ITEMS (8)
> +static const char *bank_reg_desc[MXS_OTP_BANKS][MXS_OTP_BANK_ITEMS] = {
> +MXS_OTP_BANK(CUST0, CUST1, CUST2, CUST3, CRYPTO0, CRYPTO1, CRYPTO2, CRYPTO3),
> +MXS_OTP_BANK(HWCAP0, HWCAP1, HWCAP2, HWCAP3, HWCAP4, HWCAP5, SWCAP, CUSTCAP),
> +MXS_OTP_BANK(LOCK, OPS0, OPS1, OPS2, OPS3, UN0, UN1, UN2),
> +MXS_OTP_BANK(ROM0, ROM1, ROM2, ROM3, ROM4, ROM5, ROM6, ROM7),
> +MXS_OTP_BANK(SRK0, SRK1, SRK2, SRK3, SRK4, SRK5, SRK6, SRK7),
> +};
> +#endif /* if defined(CONFIG_SOC_IMX28) */
> +
> +
> +#if defined(CONFIG_SOC_IMX23)
> +/* Building up eight registers's names of a bank */
> +#define MXS_OTP_BANK(a, b, c, d, e, f, g, h) \
> + {\
> + ("HW_OCOTP_"#a), ("HW_OCOTP_"#b), ("HW_OCOTP_"#c), ("HW_OCOTP_"#d), \
> + ("HW_OCOTP_"#e), ("HW_OCOTP_"#f), ("HW_OCOTP_"#g), ("HW_OCOTP_"#h) \
> + }
> +
> +#define MXS_OTP_BANKS (4)
> +#define MXS_OTP_BANK_ITEMS (8)
> +static const char *bank_reg_desc[MXS_OTP_BANKS][MXS_OTP_BANK_ITEMS] = {
> +MXS_OTP_BANK(CUST0, CUST1, CUST2, CUST3, CRYPTO0, CRYPTO1, CRYPTO2, CRYPTO3),
> +MXS_OTP_BANK(HWCAP0, HWCAP1, HWCAP2, HWCAP3, HWCAP4, HWCAP5, SWCAP, CUSTCAP),
> +MXS_OTP_BANK(LOCK, OPS0, OPS1, OPS2, OPS3, UN0, UN1, UN2),
> +MXS_OTP_BANK(ROM0, ROM1, ROM2, ROM3, ROM4, ROM5, ROM6, ROM7),
> +};
> +
> +#endif/* if defined(CONFIG_SOC_IMX23) */
> +
> +static unsigned long otp_hclk_saved;
> +static u32 otp_voltage_saved;
> +struct regulator *regu;
> +
> +struct fsl_otp_data {
> + char **fuse_name;
> + char *regulator_name;
> + unsigned int fuse_num;
> +};
> +
> +static struct fsl_otp_data otp_data = {
> + .fuse_name = (char **)bank_reg_desc,
> + .regulator_name = "vddio-sd0",
> + .fuse_num = MXS_OTP_BANKS * MXS_OTP_BANK_ITEMS,
> +};
> +
> +/* open the bank for the imx23/imx28 */
> +static int otp_read_prepare(void)
> +{
> + int r;
> +
> + /* [1] set the HCLK */
> + /* [2] check BUSY and ERROR bit */
> + r = otp_wait_busy(0);
> + if (r < 0)
> + goto error;
> +
> + /* [3] open bank */
> + __raw_writel(BM_OCOTP_CTRL_RD_BANK_OPEN,
> + REGS_OCOTP_BASE + HW_OCOTP_CTRL_SET);
> + udelay(10);
> +
> + /* [4] wait for BUSY */
> + r = otp_wait_busy(0);
> + return 0;
> +error:
> + return r;
> +}
> +
> +static int otp_read_post(void)
> +{
> + /* [5] close bank */
> + __raw_writel(BM_OCOTP_CTRL_RD_BANK_OPEN,
> + REGS_OCOTP_BASE + HW_OCOTP_CTRL_CLR);
> + return 0;
> +}
> +
> +static int otp_write_prepare(void)
> +{
> + struct clk *hclk;
> + int err = 0;
> +
> + /* [1] HCLK to 24MHz. */
> + hclk = clk_get_sys("hbus", NULL);
> + if (IS_ERR(hclk)) {
> + err = PTR_ERR(hclk);
> + goto out;
> + }
> + /*
> + WARNING ACHTUNG UWAGA
> +
> + the code below changes HCLK clock rate to 24M. This is
> + required to write OTP bits (7.2.2 in STMP378x Target
> + Specification), and might affect LCD operation, for example.
> + Moreover, this hacky code changes VDDIO to 2.8V; and resto-
> + res it only on otp_close(). This may affect... anything.
> +
> + You are warned now.
> + */
> + otp_hclk_saved = clk_get_rate(hclk);
> + clk_set_rate(hclk, 24000000);
> +
> + /* [2] The voltage is set to 2.8V */
> + regu = regulator_get(NULL, otp_data.regulator_name);
> + otp_voltage_saved = regulator_get_voltage(regu);
> + regulator_set_voltage(regu, 2800000, 2800000);
> +
> + /* [3] wait for BUSY and ERROR */
> + err = otp_wait_busy(BM_OCOTP_CTRL_RD_BANK_OPEN);
> +out:
> + return err;
> +}
> +
> +static int otp_write_post(void)
> +{
> + struct clk *hclk;
> +
> + hclk = clk_get_sys("hbus", NULL);
> + if (IS_ERR(hclk))
> + return PTR_ERR(hclk);
> +
> + /* restore the clock and voltage */
> + clk_set_rate(hclk, otp_hclk_saved);
> + regulator_set_voltage(regu, otp_voltage_saved, otp_voltage_saved);
> + otp_wait_busy(0);
> +
> + /* reset */
> + __raw_writel(BM_OCOTP_CTRL_RELOAD_SHADOWS,
> + REGS_OCOTP_BASE + HW_OCOTP_CTRL_SET);
> + otp_wait_busy(BM_OCOTP_CTRL_RELOAD_SHADOWS);
> + return 0;
> +}
> +
> +static int __init map_ocotp_addr(struct platform_device *pdev)
> +{
> + return 0;
> +}
> +static void unmap_ocotp_addr(void)
> +{
> +}
Ah, so here are your functions and structures. Putting !inline functions
in a header is considered a bad practice, so you should move them in the
driver directly.
Thanks!
Maxime
--
Maxime Ripard, Free Electrons
Embedded Linux, Kernel and Android engineering
http://free-electrons.com
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: Digital signature
URL: <http://lists.infradead.org/pipermail/linux-arm-kernel/attachments/20130705/4bca9dbb/attachment.sig>
More information about the linux-arm-kernel
mailing list