[PATCH 2/2] media: rc: introduce Meson IR blaster driver
Viktor Prutyanov
viktor.prutyanov at phystech.edu
Wed Jul 7 07:40:01 PDT 2021
Hi Sean,
Thank you for the review, I tried to fix issues your found in the 2nd
version, FIFO watermark parameter. Explanation is below.
On Thu, 1 Jul 2021 23:46:46 +0100
Sean Young <sean at mess.org> wrote:
> Hi Viktor,
>
> Thank you for your driver. Is there a datasheet available for this
> hardware?
>
> On Fri, Jul 02, 2021 at 12:51:32AM +0300, Viktor Prutyanov wrote:
> > This patch adds the driver for Amlogic Meson IR blaster.
> >
> > Some Amlogic SoCs such as A311D and T950D4 have IR transmitter
> > (blaster) controller onboard. It is capable of sending IR
> > signals with arbitrary carrier frequency and duty cycle.
> >
> > The driver supports 3 modulation clock sources:
> > - sysclk
> > - xtal3 clock (xtal divided by 3)
> > - 1us clock
> >
> > Signed-off-by: Viktor Prutyanov <viktor.prutyanov at phystech.edu>
> > ---
> > drivers/media/rc/Kconfig | 10 +
> > drivers/media/rc/Makefile | 1 +
> > drivers/media/rc/meson-irblaster.c | 433
> > +++++++++++++++++++++++++++++ 3 files changed, 444 insertions(+)
> > create mode 100644 drivers/media/rc/meson-irblaster.c
> >
> > diff --git a/drivers/media/rc/Kconfig b/drivers/media/rc/Kconfig
> > index d0a8326b75c2..6e60348e1bcf 100644
> > --- a/drivers/media/rc/Kconfig
> > +++ b/drivers/media/rc/Kconfig
> > @@ -246,6 +246,16 @@ config IR_MESON
> > To compile this driver as a module, choose M here: the
> > module will be called meson-ir.
> >
> > +config IR_MESON_IRBLASTER
> > + tristate "Amlogic Meson IR blaster"
> > + depends on ARCH_MESON || COMPILE_TEST
> > + help
> > + Say Y if you want to use the IR blaster available on
> > + Amlogic Meson SoCs.
> > +
> > + To compile this driver as a module, choose M here: the
> > + module will be called meson-irblaster.
> > +
> > config IR_MTK
> > tristate "Mediatek IR remote receiver"
> > depends on ARCH_MEDIATEK || COMPILE_TEST
> > diff --git a/drivers/media/rc/Makefile b/drivers/media/rc/Makefile
> > index 692e9b6b203f..b108f2b0420c 100644
> > --- a/drivers/media/rc/Makefile
> > +++ b/drivers/media/rc/Makefile
> > @@ -28,6 +28,7 @@ obj-$(CONFIG_IR_ITE_CIR) += ite-cir.o
> > obj-$(CONFIG_IR_MCEUSB) += mceusb.o
> > obj-$(CONFIG_IR_FINTEK) += fintek-cir.o
> > obj-$(CONFIG_IR_MESON) += meson-ir.o
> > +obj-$(CONFIG_IR_MESON_IRBLASTER) += meson-irblaster.o
> > obj-$(CONFIG_IR_NUVOTON) += nuvoton-cir.o
> > obj-$(CONFIG_IR_ENE) += ene_ir.o
> > obj-$(CONFIG_IR_REDRAT3) += redrat3.o
> > diff --git a/drivers/media/rc/meson-irblaster.c
> > b/drivers/media/rc/meson-irblaster.c new file mode 100644
> > index 000000000000..ef60c8d3dc3e
> > --- /dev/null
> > +++ b/drivers/media/rc/meson-irblaster.c
> > @@ -0,0 +1,433 @@
> > +// SPDX-License-Identifier: GPL-2.0-only
> > +/**
> > + * meson-irblaster.c - Amlogic Meson IR blaster driver
> > + *
> > + * Copyright (c) 2021, SberDevices. All Rights Reserved.
> > + *
> > + * Author: Viktor Prutyanov <viktor.prutyanov at phystech.edu>
> > + *
>
> No need to include the gpl boilerplate as you already have:
> // SPDX-License-Identifier: GPL-2.0-only
>
> > + * 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; version 2 of the License and no 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, GOOD TITLE
> > or
> > + * NON INFRINGEMENT. See the GNU General Public License for more
> > + * details.
> > + *
> > + * The full GNU General Public License is included in this
> > distribution in
> > + * the file called "COPYING".
> > + */
> > +
> > +#include <linux/device.h>
> > +#include <linux/module.h>
> > +#include <linux/sched.h>
> > +#include <linux/platform_device.h>
> > +#include <linux/of.h>
> > +#include <linux/interrupt.h>
> > +#include <linux/spinlock.h>
> > +#include <linux/of_irq.h>
> > +#include <linux/clk.h>
> > +#include <media/rc-core.h>
> > +
> > +#define DRIVER_NAME "meson-irblaster"
> > +
> > +#define dprintk(x...) { if (debug) pr_info(DRIVER_NAME ": "
> > x); }
>
> Please use dev_dbg().
>
> > +#define IRB_DEFAULT_CARRIER 38000
> > +#define IRB_DEFAULT_DUTY_CYCLE 50
> > +
> > +#define IRB_FIFO_LEN 128
> > +#define IRB_DEFAULT_MAX_FIFO_LEVEL 96
> > +
> > +#define IRB_ADDR0 0x0
> > +#define IRB_ADDR1 0x4
> > +#define IRB_ADDR2 0x8
> > +#define IRB_ADDR3 0xc
> > +
> > +#define IRB_MAX_DELAY (1 << 10)
> > +#define IRB_DELAY_MASK (IRB_MAX_DELAY - 1)
> > +
> > +/* IRCTRL_IR_BLASTER_ADDR0 */
> > +#define IRB_MOD_CLK(x) ((x) << 12)
> > +#define IRB_MOD_SYS_CLK 0
> > +#define IRB_MOD_XTAL3_CLK 1
> > +#define IRB_MOD_1US_CLK 2
> > +#define IRB_MOD_10US_CLK 3
> > +#define IRB_INIT_HIGH BIT(2)
> > +#define IRB_ENABLE BIT(0)
> > +
> > +/* IRCTRL_IR_BLASTER_ADDR2 */
> > +#define IRB_MOD_COUNT(lo, hi) ((((lo) - 1) << 16) | ((hi) -
> > 1)) +
> > +/* IRCTRL_IR_BLASTER_ADDR2 */
> > +#define IRB_WRITE_FIFO BIT(16)
> > +#define IRB_MOD_ENABLE BIT(12)
> > +#define IRB_TB_1US (0x0 << 10)
> > +#define IRB_TB_10US (0x1 << 10)
> > +#define IRB_TB_100US (0x2 << 10)
> > +#define IRB_TB_MOD_CLK (0x3 << 10)
> > +
> > +/* IRCTRL_IR_BLASTER_ADDR3 */
> > +#define IRB_FIFO_THD_PENDING BIT(16)
> > +#define IRB_FIFO_IRQ_ENABLE BIT(8)
> > +
> > +static bool debug;
> > +module_param(debug, bool, 0644);
> > +MODULE_PARM_DESC(debug, "Enable debug messages");
>
> With dynamic debug, you don't need this module option.
>
> > +static unsigned int max_fifo_level = IRB_DEFAULT_MAX_FIFO_LEVEL;
> > +module_param(max_fifo_level, uint, 0444);
> > +MODULE_PARM_DESC(max_fifo_level, "Max blaster FIFO filling
> > level");
>
> Why would you want to lower the fifo size? Is this module parameter
> ever needed?
The idea is following. FIFO size is 128 entries. Interrupt appears when
FIFO IRQ threshold is passed. If we set the threshold to 0, IRQ appears
right after the FIFO becomes empty. It means that IR blaster do nothing
while we pushing next entries. But for example if we set threshold to 24
= 128 - 96, IRQ appears when 24 entries are about to transmit and we
have enough time to push new entries.
Of course, it has more sense in previous version with threaded IRQ, but
I think it still OK to have time reserve to push entries with a large
number.
>
> > +
> > +struct irblaster_dev {
> > + unsigned int irq;
> > + void __iomem *reg_base;
> > + unsigned int *buf;
> > + unsigned int buf_len;
> > + unsigned int buf_head;
> > + unsigned int carrier;
> > + unsigned int duty_cycle;
> > + spinlock_t lock;
> > + struct completion completion;
> > + unsigned int max_fifo_level;
> > + unsigned int clk_nr;
> > + unsigned long clk_rate;
> > +};
> > +
> > +static void irb_set_mod(struct irblaster_dev *irb)
> > +{
> > + unsigned int cnt = irb->clk_rate / irb->carrier;
> > + unsigned int pulse_cnt = cnt * irb->duty_cycle / 100;
> > + unsigned int space_cnt = cnt - pulse_cnt;
> > +
> > + dprintk("F_mod = %uHz, T_mod = %luns, duty_cycle = %u%%\n",
> > + irb->carrier, NSEC_PER_SEC / irb->clk_rate * cnt,
> > + 100 * pulse_cnt / cnt);
>
> dev_dbg()
>
> > +
> > + writel(IRB_MOD_COUNT(pulse_cnt, space_cnt),
> > + irb->reg_base + IRB_ADDR1);
> > +}
> > +
> > +static void irb_setup(struct irblaster_dev *irb)
> > +{
> > + unsigned int fifo_irq_threshold = IRB_FIFO_LEN -
> > irb->max_fifo_level; +
> > + /*
> > + * Disable the blaster, set modulator clock tick and set
> > initialize
> > + * output to be high. Set up carrier frequency and duty
> > cycle. Then
> > + * unset initialize output. Enable FIFO interrupt, set
> > FIFO interrupt
> > + * threshold. Finally, enable the blaster back.
> > + */
> > + writel(~IRB_ENABLE & (IRB_MOD_CLK(irb->clk_nr) |
> > IRB_INIT_HIGH),
> > + irb->reg_base + IRB_ADDR0);
> > + irb_set_mod(irb);
> > + writel(readl(irb->reg_base + IRB_ADDR0) & ~IRB_INIT_HIGH,
> > + irb->reg_base + IRB_ADDR0);
> > + writel(IRB_FIFO_IRQ_ENABLE | fifo_irq_threshold,
> > + irb->reg_base + IRB_ADDR3);
> > + writel(readl(irb->reg_base + IRB_ADDR0) | IRB_ENABLE,
> > + irb->reg_base + IRB_ADDR0);
> > +}
> > +
> > +static void irb_fifo_push_pulse(struct irblaster_dev *irb,
> > unsigned int time) +{
> > + unsigned int delay;
> > + unsigned int tb = IRB_TB_MOD_CLK;
> > + unsigned int tb_us = USEC_PER_SEC / irb->carrier;
> > +
> > + delay = (DIV_ROUND_CLOSEST_ULL(time, tb_us) - 1) &
> > IRB_DELAY_MASK;
> > + writel((IRB_WRITE_FIFO | IRB_MOD_ENABLE) | tb | delay,
> > + irb->reg_base + IRB_ADDR2);
> > +}
> > +
> > +static void irb_fifo_push_space(struct irblaster_dev *irb,
> > unsigned int time) +{
> > + unsigned int delay;
> > + unsigned int tb = IRB_TB_100US;
> > + unsigned int tb_us = 100;
> > +
> > + if (time <= IRB_MAX_DELAY) {
> > + tb = IRB_TB_1US;
> > + tb_us = 1;
> > + } else if (time <= 10 * IRB_MAX_DELAY) {
> > + tb = IRB_TB_10US;
> > + tb_us = 10;
> > + } else if (time <= 100 * IRB_MAX_DELAY) {
> > + tb = IRB_TB_100US;
> > + tb_us = 100;
> > + }
> > +
> > + delay = (DIV_ROUND_CLOSEST_ULL(time, tb_us) - 1) &
> > IRB_DELAY_MASK;
> > + writel((IRB_WRITE_FIFO & ~IRB_MOD_ENABLE) | tb | delay,
> > + irb->reg_base + IRB_ADDR2);
> > +}
> > +
> > +static void irb_send_buffer(struct irblaster_dev *irb)
> > +{
> > + unsigned long flags;
> > + unsigned int nr = 0;
> > +
> > + spin_lock_irqsave(&irb->lock, flags);
> > + while (irb->buf_head < irb->buf_len && nr <
> > irb->max_fifo_level) {
> > + if (irb->buf_head % 2 == 0)
> > + irb_fifo_push_pulse(irb,
> > irb->buf[irb->buf_head]);
> > + else
> > + irb_fifo_push_space(irb,
> > irb->buf[irb->buf_head]); +
> > + irb->buf_head++;
> > + nr++;
> > + }
> > + spin_unlock_irqrestore(&irb->lock, flags);
> > +}
> > +
> > +static bool irb_check_buf(struct irblaster_dev *irb,
> > + unsigned int *buf, unsigned int len)
> > +{
> > + unsigned int i;
> > +
> > + for (i = 0; i < len; i++) {
> > + unsigned int max_tb_us;
> > + /*
> > + * Max space timebase is 100 us.
> > + * Pulse timebase equals to carrier period.
> > + */
> > + if (i % 2 == 0)
> > + max_tb_us = USEC_PER_SEC / irb->carrier;
> > + else
> > + max_tb_us = 100;
> > +
> > + if (buf[i] >= max_tb_us * IRB_MAX_DELAY)
> > + return false;
> > + }
> > +
> > + return true;
> > +}
> > +
> > +static void irb_send(struct irblaster_dev *irb,
> > + unsigned int *buf, unsigned int len)
> > +{
> > + reinit_completion(&irb->completion);
> > +
> > + irb->buf = buf;
> > + irb->buf_len = len;
> > + irb->buf_head = 0;
> > +
> > + dprintk("tx started, buffer length = %u\n", len);
> > + irb_send_buffer(irb);
> > + wait_for_completion_interruptible(&irb->completion);
> > + dprintk("tx completed\n");
> > +}
> > +
> > +static irqreturn_t irb_irqhandler(int irq, void *data)
> > +{
> > + struct irblaster_dev *irb = data;
> > +
> > + writel(readl(irb->reg_base + IRB_ADDR3) &
> > ~IRB_FIFO_THD_PENDING,
> > + irb->reg_base + IRB_ADDR3);
> > +
> > + if (irb->buf_head < irb->buf_len)
> > + return IRQ_WAKE_THREAD;
> > +
> > + complete(&irb->completion);
> > +
> > + return IRQ_HANDLED;
> > +}
> > +
> > +static irqreturn_t irb_thread_irqhandler(int irq, void *data)
> > +{
> > + struct irblaster_dev *irb = data;
> > +
> > + irb_send_buffer(irb);
> > +
> > + return IRQ_HANDLED;
> > +}
> > +
> > +static int irb_set_tx_carrier(struct rc_dev *rc, u32 carrier)
> > +{
> > + struct irblaster_dev *irb = rc->priv;
> > +
> > + irb->carrier = carrier;
>
> carrier might be 0 for unmodulated IR. This will make irb_set_mod()
> do a division by zero.
>
> Please check appropriate range for carrier and support unmodulated IR
> (carrier = 0) if possible.
>
> > + irb_set_mod(irb);
> > +
> > + return 0;
> > +}
> > +
> > +static int irb_set_tx_duty_cycle(struct rc_dev *rc, u32 duty_cycle)
> > +{
> > + struct irblaster_dev *irb = rc->priv;
> > +
> > + irb->duty_cycle = duty_cycle;
> > + irb_set_mod(irb);
> > +
> > + return 0;
> > +}
> > +
> > +static int irb_tx_ir(struct rc_dev *rc, unsigned int *buf,
> > unsigned int len) +{
> > + struct irblaster_dev *irb = rc->priv;
> > +
> > + if (!irb_check_buf(irb, buf, len))
> > + return -EINVAL;
> > +
> > + irb_send(irb, buf, len);
> > +
> > + return len;
> > +}
> > +
> > +static int irb_mod_clock_probe(struct irblaster_dev *irb, struct
> > device *dev) +{
> > + struct device_node *np = dev->of_node;
> > + struct clk *clock;
> > + const char *clock_name;
> > +
> > + if (!np)
> > + return -ENODEV;
> > +
> > + if (!of_property_read_string(np, "mod-clock",
> > &clock_name)) {
> > + if (!strcmp(clock_name, "sysclk"))
> > + irb->clk_nr = IRB_MOD_SYS_CLK;
> > + else if (!strcmp(clock_name, "xtal"))
> > + irb->clk_nr = IRB_MOD_XTAL3_CLK;
> > + else
> > + return -EINVAL;
> > +
> > + clock = devm_clk_get(dev, clock_name);
> > + if (IS_ERR(clock) || clk_prepare_enable(clock))
> > + return -ENODEV;
> > + } else {
> > + irb->clk_nr = IRB_MOD_1US_CLK;
> > + }
> > +
> > + switch (irb->clk_nr) {
> > + case IRB_MOD_SYS_CLK:
> > + irb->clk_rate = clk_get_rate(clock);
> > + break;
> > + case IRB_MOD_XTAL3_CLK:
> > + irb->clk_rate = clk_get_rate(clock) / 3;
> > + break;
> > + case IRB_MOD_1US_CLK:
> > + irb->clk_rate = 1000000;
> > + break;
> > + }
> > +
> > + dprintk("F_clk = %luHz\n", irb->clk_rate);
> > +
> > + return 0;
> > +}
> > +
> > +static int __init irblaster_probe(struct platform_device *pdev)
> > +{
> > + struct device *dev = &pdev->dev;
> > + struct irblaster_dev *irb;
> > + struct rc_dev *rc;
> > + struct resource *range;
> > + int ret;
> > +
> > + irb = devm_kzalloc(dev, sizeof(*irb), GFP_KERNEL);
> > + if (!irb)
> > + return -ENOMEM;
> > +
> > + range = platform_get_resource(pdev, IORESOURCE_MEM, 0);
> > + if (!range) {
> > + dev_err(dev, "no memory resource found\n");
> > + return -ENODEV;
> > + }
> > +
> > + irb->reg_base = devm_ioremap_resource(dev, range);
> > + if (IS_ERR(irb->reg_base)) {
> > + dev_err(dev, "ioremap failed\n");
> > + return PTR_ERR(irb->reg_base);
> > + }
> > +
> > + irb->irq = platform_get_irq(pdev, 0);
> > + if (irb->irq < 0) {
> > + dev_err(dev, "no irq resource found\n");
> > + return -ENODEV;
> > + }
> > +
> > + if (max_fifo_level <= IRB_FIFO_LEN)
> > + irb->max_fifo_level = max_fifo_level;
> > + else {
> > + irb->max_fifo_level = IRB_FIFO_LEN;
> > + dev_warn(dev, "max FIFO level param truncated to
> > %u",
> > + IRB_FIFO_LEN);
> > + }
> > +
> > + irb->carrier = IRB_DEFAULT_CARRIER;
> > + irb->duty_cycle = IRB_DEFAULT_DUTY_CYCLE;
> > + init_completion(&irb->completion);
> > + spin_lock_init(&irb->lock);
> > +
> > + ret = irb_mod_clock_probe(irb, dev);
> > + if (ret) {
> > + dev_err(dev, "modulator clock setup failed\n");
> > + return ret;
> > + }
> > + irb_setup(irb);
> > +
> > + ret = devm_request_threaded_irq(dev, irb->irq,
> > + irb_irqhandler,
> > + irb_thread_irqhandler,
> > + IRQF_TRIGGER_RISING,
> > + DRIVER_NAME, irb);
>
> Rather than using threaded irqs, would it make more sense to convert
> the tx data to the right format before starting tx, thus avoiding
> doing expensive(ish) conversions during interrupt handling.
>
> Then the interrupt handler would just need to feed the fifo from a
> buffer, which can be done without a threaded irq. Threaded irq might
> be an issue if the thread handler does not get called in time and the
> hardware runs out of tx data.
>
> > + if (ret) {
> > + dev_err(dev, "irq request failed\n");
> > + return ret;
> > + }
> > +
> > + rc = rc_allocate_device(RC_DRIVER_IR_RAW_TX);
> > + if (!rc)
> > + return -ENOMEM;
> > +
> > + rc->driver_name = DRIVER_NAME;
>
> Please set rc->device_name as well.
>
> > + rc->priv = irb;
> > +
> > + rc->tx_ir = irb_tx_ir;
> > + rc->s_tx_carrier = irb_set_tx_carrier;
> > + rc->s_tx_duty_cycle = irb_set_tx_duty_cycle;
> > +
> > + ret = rc_register_device(rc);
> > + if (ret < 0) {
> > + dev_err(dev, "rc_dev registration failed\n");
> > + rc_free_device(rc);
> > + return ret;
> > + }
> > +
> > + platform_set_drvdata(pdev, rc);
> > +
> > + return 0;
> > +}
> > +
> > +static int irblaster_remove(struct platform_device *pdev)
> > +{
> > + struct rc_dev *rc = platform_get_drvdata(pdev);
> > +
> > + rc_unregister_device(rc);
> > +
> > + return 0;
> > +}
> > +
> > +static const struct of_device_id irblaster_dt_match[] = {
> > + {
> > + .compatible = "amlogic,meson-irblaster",
> > + },
> > + {},
> > +};
> > +MODULE_DEVICE_TABLE(of, irblaster_dt_match);
> > +
> > +static struct platform_driver irblaster_pd = {
> > + .remove = irblaster_remove,
> > + .driver = {
> > + .name = DRIVER_NAME,
> > + .owner = THIS_MODULE,
> > + .of_match_table = irblaster_dt_match,
> > + },
> > +};
> > +
> > +module_platform_driver_probe(irblaster_pd, irblaster_probe);
> > +
> > +MODULE_DESCRIPTION("Meson IR blaster driver");
> > +MODULE_AUTHOR("Viktor Prutyanov <viktor.prutyanov at phystech.edu>");
> > +MODULE_LICENSE("GPL");
> > --
> > 2.21.0
>
> Thanks,
>
> Sean
Best regards,
Viktor Prutyanov
More information about the linux-amlogic
mailing list