| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Date: Mon, 29 Jul 2019 19:59:33 +0800
From: Chuhong Yuan <hslester96@...il.com>
To: Brian Masney <masneyb@...tation.org>
Cc: Jonathan Cameron <jic23@...nel.org>,
Hartmut Knaack <knaack.h@....de>,
Lars-Peter Clausen <lars@...afoo.de>,
Peter Meerwald-Stadler <pmeerw@...erw.net>,
linux-iio@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v2] iio: tsl2772: Use device-managed API
Brian Masney <masneyb@...tation.org> 于2019年7月29日周一 下午6:59写道：
>
> Hi Chuhong,
>
> On Mon, Jul 29, 2019 at 06:03:39PM +0800, Chuhong Yuan wrote:
> > Use devm_iio_device_register to simplify
> > the code.
> >
> > Signed-off-by: Chuhong Yuan <hslester96@...il.com>
>
> Thank you for the patch. The patch description doesn't match what all is
> done below. This should also be broken up into multiple patches since a
> patch should only do one thing. The regulator changes should be in their
> own patch, and some of the devm changes may require multiple patches.
> When writing your changelog, if your patch description has the word
> 'and', then that may be a hint that you need to break up your patch a
> little bit. That's not always the case, but something to keep in mind.
>
Thanks for your advice.
I will split it into two patches in next version.
One is to use devm to simpliy the code.
The other is to use regulator_bulk_() to shrink driver size.
> A few minor comments below.
>
> > ---
> > Changes in v2:
> > - Use regulator_bulk_() to shrink driver
> > size.
> > - Utilize more devm functions to simplify
> > the code.
> > - Remove several redundant functions.
> >
> > drivers/iio/light/tsl2772.c | 116 +++++++++++-------------------------
> > 1 file changed, 34 insertions(+), 82 deletions(-)
> >
> > diff --git a/drivers/iio/light/tsl2772.c b/drivers/iio/light/tsl2772.c
> > index 83cece921843..946537c8586f 100644
> > --- a/drivers/iio/light/tsl2772.c
> > +++ b/drivers/iio/light/tsl2772.c
> > @@ -131,7 +131,10 @@ enum {
> > enum {
> > TSL2772_CHIP_UNKNOWN = 0,
> > TSL2772_CHIP_WORKING = 1,
> > - TSL2772_CHIP_SUSPENDED = 2
> > + TSL2772_CHIP_SUSPENDED = 2,
> > + TSL2772_SUPPLY_VDD = 0,
> > + TSL2772_SUPPLY_VDDIO = 1,
> > + TSL2772_NUM_SUPPLIES = 2
> > };
>
> This is a really minor nitpick but can these either use a #define or be
> placed in its own enum block?
>
I refer to drivers/iio/adc/ad7766.c when I use regulator_bulk_().
This file puts them in the enum block.
There are also files using #define, like drivers/iio/dac/ad5064.c and
ad5449.c.
I think both of them are okay.
> >
> > /* Per-device data */
> > @@ -161,8 +164,7 @@ struct tsl2772_chip {
> > struct mutex prox_mutex;
> > struct mutex als_mutex;
> > struct i2c_client *client;
> > - struct regulator *vdd_supply;
> > - struct regulator *vddio_supply;
> > + struct regulator_bulk_data reg[TSL2772_NUM_SUPPLIES];
>
> Since there's other changes, maybe name this 'supplies'? I think of
> 'reg' as an address.
>
Indeed there are files choosing supplies as the name.
But in iio, all regulator_bulk_data use reg (or regs, vref_reg) as the name.
So I also use reg as the name to keep consistency with others.
> > u16 prox_data;
> > struct tsl2772_als_info als_cur_info;
> > struct tsl2772_settings settings;
> > @@ -697,46 +699,7 @@ static void tsl2772_disable_regulators_action(void *_data)
> > {
> > struct tsl2772_chip *chip = _data;
> >
> > - regulator_disable(chip->vdd_supply);
> > - regulator_disable(chip->vddio_supply);
> > -}
> > -
> > -static int tsl2772_enable_regulator(struct tsl2772_chip *chip,
> > - struct regulator *regulator)
> > -{
> > - int ret;
> > -
> > - ret = regulator_enable(regulator);
> > - if (ret < 0) {
> > - dev_err(&chip->client->dev, "Failed to enable regulator: %d\n",
> > - ret);
> > - return ret;
> > - }
> > -
> > - return 0;
> > -}
> > -
> > -static struct regulator *tsl2772_get_regulator(struct tsl2772_chip *chip,
> > - char *name)
> > -{
> > - struct regulator *regulator;
> > - int ret;
> > -
> > - regulator = devm_regulator_get(&chip->client->dev, name);
> > - if (IS_ERR(regulator)) {
> > - if (PTR_ERR(regulator) != -EPROBE_DEFER)
> > - dev_err(&chip->client->dev,
> > - "Failed to get %s regulator %d\n",
> > - name, (int)PTR_ERR(regulator));
> > -
> > - return regulator;
> > - }
> > -
> > - ret = tsl2772_enable_regulator(chip, regulator);
> > - if (ret < 0)
> > - return ERR_PTR(ret);
> > -
> > - return regulator;
> > + regulator_bulk_disable(ARRAY_SIZE(chip->reg), chip->reg);
> > }
> >
> > static int tsl2772_chip_on(struct iio_dev *indio_dev)
> > @@ -860,6 +823,13 @@ static int tsl2772_chip_off(struct iio_dev *indio_dev)
> > return tsl2772_write_control_reg(chip, 0x00);
> > }
> >
> > +static void tsl2772_chip_off_action(void *data)
> > +{
> > + struct iio_dev *indio_dev = data;
> > +
> > + tsl2772_chip_off(indio_dev);
> > +}
> > +
> > /**
> > * tsl2772_invoke_change - power cycle the device to implement the user
> > * parameters
> > @@ -1797,20 +1767,22 @@ static int tsl2772_probe(struct i2c_client *clientp,
> > chip->client = clientp;
> > i2c_set_clientdata(clientp, indio_dev);
> >
> > - chip->vddio_supply = tsl2772_get_regulator(chip, "vddio");
> > - if (IS_ERR(chip->vddio_supply))
> > - return PTR_ERR(chip->vddio_supply);
> > + chip->reg[TSL2772_SUPPLY_VDD].supply = "vdd";
> > + chip->reg[TSL2772_SUPPLY_VDDIO].supply = "vddio";
> >
> > - chip->vdd_supply = tsl2772_get_regulator(chip, "vdd");
> > - if (IS_ERR(chip->vdd_supply)) {
> > - regulator_disable(chip->vddio_supply);
> > - return PTR_ERR(chip->vdd_supply);
> > - }
> > + ret = devm_regulator_bulk_get(&clientp->dev, ARRAY_SIZE(chip->reg),
> > + chip->reg);
> > + if (ret < 0)
> > + return ret;
>
> Add a dev_err like the other error paths in probe function. Users can
> use the tracing subsystem to see why this failed but an error message
> in dmesg is much easier for users to find. Also be sure to check for the
> EPROBE_DEFER case.
>
I will do this in next version.
Regards,
Chuhong
> Brian
Powered by blists - more mailing lists