Re: [PATCH v7 2/4] iio: adc: Add Xilinx AMS driver
From: Andy Shevchenko
Date: Mon Oct 25 2021 - 06:11:52 EST
On Tue, Oct 19, 2021 at 6:22 PM Anand Ashok Dumbre
<anand.ashok.dumbre@xxxxxxxxxx> wrote:
>
> The AMS includes an ADC as well as on-chip sensors that can be used to
> sample external voltages and monitor on-die operating conditions, such
> as temperature and supply voltage levels. The AMS has two SYSMON blocks.
> PL-SYSMON block is capable of monitoring off chip voltage and
> temperature.
> PL-SYSMON block has DRP, JTAG and I2C interface to enable monitoring
> from external master. Out of these interface currently only DRP is
from an external
interfaces
> supported.
> Other block PS-SYSMON is memory mapped to PS.
> The AMS can use internal channels to monitor voltage and temperature as
> well as one primary and up to 16 auxiliary channels for measuring
> external voltages.
> The voltage and temperature monitoring channels also have event
> capability which allows to generate an interrupt when their value falls
> below or raises above a set threshold.
>
> Signed-off-by: Manish Narani <manish.narani@xxxxxxxxxx>
What does this SoB mean here? Have you read Submitting Patches?
> Signed-off-by: Anand Ashok Dumbre <anand.ashok.dumbre@xxxxxxxxxx>
...
> +config XILINX_AMS
> + tristate "Xilinx AMS driver"
> + depends on ARCH_ZYNQMP || COMPILE_TEST
> + depends on HAS_IOMEM
> + help
> + Say yes here to have support for the Xilinx AMS.
It's not important for most of the users. Please, strat help with more
useful information like below.
> + The driver supports Voltage and Temperature monitoring on Xilinx Ultrascale
> + devices.
> +
> + The driver can also be built as a module. If so, the module will be called
> + xilinx-ams.
...
> + * Manish Narani <mnarani@xxxxxxxxxx>
A-ha! You probably forgot the Co-developed-by tag above.
> + * Rajnikant Bhojani <rajnikant.bhojani@xxxxxxxxxx>
...
Missed headers, like bits.h.
> +#include <linux/clk.h>
> +#include <linux/delay.h>
> +#include <linux/interrupt.h>
> +#include <linux/io.h>
> +#include <linux/iopoll.h>
> +#include <linux/kernel.h>
> +#include <linux/module.h>
> +#include <linux/of_address.h>
Do you need this? Maybe mod_devicetable.h?
> +#include <linux/platform_device.h>
> +#include <linux/slab.h>
...
> +static const unsigned int AMS_UNMASK_TIMEOUT_MS = 500;
Why not define (esp. taking into account another similar define below?
...
> +#define AMS_REGCFG1_ALARM_MASK 0xF0F
> +#define AMS_REGCFG3_ALARM_MASK 0x3F
> +#define AMS_PL_ALARM_MASK 0xFFFF0000U
> +#define AMS_ISR0_ALARM_MASK 0xFFFFFFFFU
> +#define AMS_ISR1_ALARM_MASK 0xE000001FU
> +#define AMS_ISR1_EOC_MASK 0x00000008U
What is so special about these that they are not using combinations of
GENMASK() / BIT()?
...
> +enum ams_alarm_bit {
> + AMS_ALARM_BIT_TEMP,
> + AMS_ALARM_BIT_SUPPLY1,
> + AMS_ALARM_BIT_SUPPLY2,
> + AMS_ALARM_BIT_SUPPLY3,
> + AMS_ALARM_BIT_SUPPLY4,
> + AMS_ALARM_BIT_SUPPLY5,
> + AMS_ALARM_BIT_SUPPLY6,
> + AMS_ALARM_BIT_RESERVED,
> + AMS_ALARM_BIT_SUPPLY7,
> + AMS_ALARM_BIT_SUPPLY8,
> + AMS_ALARM_BIT_SUPPLY9,
> + AMS_ALARM_BIT_SUPPLY10,
> + AMS_ALARM_BIT_VCCAMS,
> + AMS_ALARM_BIT_TEMP_REMOTE
Is it terminator line? Doesn't sound like it to me. So, please add a
comma. Same for the rest.
> +};
...
> + AMS_SEQ_MAX
This seems correct, no comma is needed :-)
...
> +struct ams {
> + void __iomem *base;
> + void __iomem *ps_base;
> + void __iomem *pl_base;
> + struct clk *clk;
> + struct device *dev;
> + /* check kernel doc above */
Useless
> + struct mutex lock;
> + /* check kernel doc above */
Ditto.
> + spinlock_t intr_lock;
> + unsigned int alarm_mask;
> + unsigned int masked_alarm;
> + u64 intr_mask;
> + int irq;
> + struct delayed_work ams_unmask_work;
> +};
...
> + writel((val & ~mask) | (data & mask), ams->ps_base + offset);
Split to assignment and simple writel() call. Same to the rest.
...
> + ams->intr_mask &= ~mask;
> + ams->intr_mask |= (val & mask);
This may be combined to one line as it's a standard pattern.
...
> + if (ams->ps_base) {
> + /* Configuring PS alarm enable */
> + cfg = ~((alarm_mask & AMS_ISR0_ALARM_2_TO_0_MASK) <<
> + AMS_CONF1_ALARM_2_TO_0_SHIFT);
> + cfg &= ~((alarm_mask & AMS_ISR0_ALARM_6_TO_3_MASK) <<
> + AMS_CONF1_ALARM_6_TO_3_SHIFT);
> + ams_ps_update_reg(ams, AMS_REG_CONFIG1, AMS_REGCFG1_ALARM_MASK,
> + cfg);
> +
> + cfg = ~((alarm_mask >> AMS_CONF3_ALARM_12_TO_7_SHIFT) &
> + AMS_ISR0_ALARM_12_TO_7_MASK);
> + ams_ps_update_reg(ams, AMS_REG_CONFIG3, AMS_REGCFG3_ALARM_MASK,
> + cfg);
> + }
By factoring out the body of the conditional to a helper function you may:
- decrease indentation
- make code better to read
- reduce LOCs
> + if (ams->pl_base) {
> + pl_alarm_mask = (alarm_mask >> AMS_PL_ALARM_START);
> + pl_alarm_mask = FIELD_GET(AMS_PL_ALARM_MASK, alarm_mask);
> + /* Configuring PL alarm enable */
> + cfg = ~((pl_alarm_mask & AMS_ISR0_ALARM_2_TO_0_MASK) <<
> + AMS_CONF1_ALARM_2_TO_0_SHIFT);
> + cfg &= ~((pl_alarm_mask & AMS_ISR0_ALARM_6_TO_3_MASK) <<
> + AMS_CONF1_ALARM_6_TO_3_SHIFT);
> + ams_pl_update_reg(ams, AMS_REG_CONFIG1,
> + AMS_REGCFG1_ALARM_MASK, cfg);
> +
> + cfg = ~((pl_alarm_mask >> AMS_CONF3_ALARM_12_TO_7_SHIFT) &
> + AMS_ISR0_ALARM_12_TO_7_MASK);
> + ams_pl_update_reg(ams, AMS_REG_CONFIG3,
> + AMS_REGCFG3_ALARM_MASK, cfg);
> + }
Ditto. And the same applies to all the rest where it gains something
from the above list of improvements.
...
> + int i;
> + unsigned long long scan_mask;
> + struct ams *ams = iio_priv(indio_dev);
Reversed xmas tree order, please.
Same for the rest.
...
> + /* Run calibration of PS & PL as part of the sequence */
> + scan_mask = 0x1 | BIT(AMS_PS_SEQ_MAX);
BIT(0) ?
...
> + ams_update_intrmask(ams, ~0, ~0);
Replace ~0 to proper GENMASK()./BIT() combination which takes into
account real bits used by hardware.
...
> + case IIO_CHAN_INFO_RAW:
> + mutex_lock(&ams->lock);
> + if (chan->scan_index >= (AMS_PS_SEQ_MAX * 3)) {
Too many parens.
> + ret = ams_read_vcc_reg(ams, chan->address, val);
> + if (ret) {
> + mutex_unlock(&ams->lock);
> + return -EINVAL;
> + }
> + ams_enable_channel_sequence(indio_dev);
> + } else if (chan->scan_index >= AMS_PS_SEQ_MAX)
> + *val = readl(ams->pl_base + chan->address);
> + else
> + *val = readl(ams->ps_base + chan->address);
> + mutex_unlock(&ams->lock);
> +
> + return IIO_VAL_INT;
...
> + return -EINVAL;
Use corresponding defaul cases in each of the switches.
...
> + int offset = 0;
Make the assignment as an else branch, so all offset assignments will
be grouped together.
> + if (dir == IIO_EV_DIR_FALLING) {
> + if (scan_index < AMS_SEQ_SUPPLY7)
> + offset = AMS_ALARM_THRESHOLD_OFF_10;
> + else
> + offset = AMS_ALARM_THRESHOLD_OFF_20;
> + }
...
> + return 0;
default case.
> +}
...
> +static const struct iio_chan_spec
> +*ams_event_to_channel(struct iio_dev *indio_dev, u32 event)
Unusual indentation.
...
> + case AMS_ALARM_BIT_TEMP_REMOTE:
> + scan_index += AMS_SEQ_TEMP_REMOTE;
> + break;
default?
Same for the rest of the cases like this.
...
> + return (ams->alarm_mask & ams_get_alarm_mask(chan->scan_index)) ? 1 : 0;
!! would work as well.
...
> + /*
> + * The temperature channel only supports over-temperature
> + * events
Missed period.
> + */
...
> + /* only process alarms that are not masked */
Inconsistent style (here capitalization is missed). Make all comments
in the code consistent.
> + isr0 &= ~((ams->intr_mask & AMS_ISR0_ALARM_MASK) | ams->masked_alarm);
> +
Redundant blank line.
> + if (!isr0)
How did you test this branch? (Hint: something very important should
be done here)
> + return IRQ_NONE;
...
> + for_each_child_of_node(chan_node, child) {
> + ret = of_property_read_u32(child, "reg", ®);
> + if (ret || reg > (AMS_PL_MAX_EXT_CHANNEL + 30))
> + continue;
> +
> + memcpy(&channels[num_channels], &ams_pl_channels[reg +
> + AMS_PL_MAX_FIXED_CHANNEL - 30], sizeof(*channels));
> +
> + if (of_property_read_bool(child, "xlnx,bipolar"))
> + channels[num_channels].scan_type.sign = 's';
> +
> + num_channels++;
> + }
Use device property API here instead of *of_*() calls.
...
> + /* Initialize buffer for channel specification */
> + ams_channels = kzalloc(sizeof(ams_ps_channels) +
> + sizeof(ams_pl_channels) +
> + sizeof(ams_ctrl_channels), GFP_KERNEL);
Use the corresponding macro from overflow.h.
> + if (!ams_channels)
> + return -ENOMEM;
...
> + if (of_device_is_available(np)) {
fwnode_device_is_available()
> + ret = ams_init_module(indio_dev, np, ams_channels);
> + if (ret < 0)
> + goto err;
> +
> + num_channels += ret;
> + }
...
> + for_each_child_of_node(np, child_node) {
> + if (of_device_is_available(child_node)) {
> + ret = ams_init_module(indio_dev, child_node,
> + ams_channels + num_channels);
> + if (ret < 0)
> + goto err;
> +
> + num_channels += ret;
> + }
> + }
As per above.
...
> + if (!pdev->dev.of_node)
> + return -ENODEV;
Drop this, please. It will allow reuse of the driver in ACPI environments.
...
> + ams->irq = platform_get_irq(pdev, 0);
> + if (ams->irq == -EPROBE_DEFER) {
Is IRQ optional or not?
> + ret = -EPROBE_DEFER;
> + return ret;
> + }
...
> + ret = devm_iio_device_register(&pdev->dev, indio_dev);
> +
> + return ret;
return devm_...
...
> + clk_prepare_enable(ams->clk);
It might fail.
--
With Best Regards,
Andy Shevchenko