| Message ID | 20220727064415.940690-1-potin.lai.pt@gmail.com (mailing list archive) |
|---|---|
| State | Changes Requested |
| Headers | show |
| Series | [v2,1/1] iio: humidity: hdc100x: add manufacturer and device ID cehck | expand |
On Wed, Jul 27, 2022 at 8:46 AM Potin Lai <potin.lai.pt@gmail.com> wrote: > > Add manufacturer and device ID checking during probe, and skip the > checking if chip model not supported. > > Supported: > - HDC1000 > - HDC1010 > - HDC1050 > - HDC1080 > > Not supported: > - HDC1008 Thanks for an update, my comments below. ... > + const struct of_device_id *match; Don't you have a compiler warning? Always compile your code with `make W=1` ... > + data = device_get_match_data(&client->dev); > + if (data) { This check is redundant. Too much protective programming. Just oblige that matched ID has to always have an associated data. > + if (!data->support_mfr_check) > + return true; > + } ... > - .probe = hdc100x_probe, > + .probe_new = hdc100x_probe, Make this a separate patch before the presented one.
On 7/27/22 18:00, Andy Shevchenko wrote: > On Wed, Jul 27, 2022 at 8:46 AM Potin Lai <potin.lai.pt@gmail.com> wrote: >> Add manufacturer and device ID checking during probe, and skip the >> checking if chip model not supported. >> >> Supported: >> - HDC1000 >> - HDC1010 >> - HDC1050 >> - HDC1080 >> >> Not supported: >> - HDC1008 > Thanks for an update, my comments below. > > ... > >> + const struct of_device_id *match; > Don't you have a compiler warning? Always compile your code with `make W=1` > > ... You are correct, I will remove this unused variable. >> + data = device_get_match_data(&client->dev); >> + if (data) { > This check is redundant. Too much protective programming. Just oblige > that matched ID has to always have an associated data. Is it guaranteed that device_get_match_data will not return NULL? I find some examples in other drivers, all of them have a check on returned data. Will it be more appropriate if I move device_get_match_data to probe function, and return EINVAL when we get a NULL pointer from device_get_match_data? >> + if (!data->support_mfr_check) >> + return true; >> + } > ... > >> - .probe = hdc100x_probe, >> + .probe_new = hdc100x_probe, > Make this a separate patch before the presented one. > got it, will move this into a separate patch in next version. thanks, Potin
On Wed, Jul 27, 2022 at 12:42 PM Potin Lai <potin.lai.pt@gmail.com> wrote: > On 7/27/22 18:00, Andy Shevchenko wrote: > > On Wed, Jul 27, 2022 at 8:46 AM Potin Lai <potin.lai.pt@gmail.com> wrote: ... > >> + data = device_get_match_data(&client->dev); > >> + if (data) { > > This check is redundant. Too much protective programming. Just oblige > > that matched ID has to always have an associated data. > Is it guaranteed that device_get_match_data will not return NULL? I find some examples in other drivers, all of them have a check on returned data. No, but as I said you may guarantee that by obliging developers not to shoot in their feet. > Will it be more appropriate if I move device_get_match_data to probe function, and return EINVAL when we get a NULL pointer from device_get_match_data? Why is this check needed? We do not like dead code. > >> + if (!data->support_mfr_check) > >> + return true; > >> + }
Andy Shevchenko 於 7/27/2022 7:56 PM 寫道: > On Wed, Jul 27, 2022 at 12:42 PM Potin Lai <potin.lai.pt@gmail.com> wrote: >> On 7/27/22 18:00, Andy Shevchenko wrote: >>> On Wed, Jul 27, 2022 at 8:46 AM Potin Lai <potin.lai.pt@gmail.com> wrote: > ... > >>>> + data = device_get_match_data(&client->dev); >>>> + if (data) { >>> This check is redundant. Too much protective programming. Just oblige >>> that matched ID has to always have an associated data. >> Is it guaranteed that device_get_match_data will not return NULL? I find some examples in other drivers, all of them have a check on returned data. > No, but as I said you may guarantee that by obliging developers not to > shoot in their feet. Thanks for the explanation, I will remove the checking part. Potin >> Will it be more appropriate if I move device_get_match_data to probe function, and return EINVAL when we get a NULL pointer from device_get_match_data? > Why is this check needed? We do not like dead code. > >>>> + if (!data->support_mfr_check) >>>> + return true; >>>> + }
diff --git a/drivers/iio/humidity/hdc100x.c b/drivers/iio/humidity/hdc100x.c index 9e0fce917ce4c..8c69b4a610c4f 100644 --- a/drivers/iio/humidity/hdc100x.c +++ b/drivers/iio/humidity/hdc100x.c @@ -34,6 +34,23 @@ #define HDC100X_REG_CONFIG_ACQ_MODE BIT(12) #define HDC100X_REG_CONFIG_HEATER_EN BIT(13) +#define HDC100X_REG_MFR_ID 0xFE +#define HDC100X_REG_DEV_ID 0xFF + +#define HDC100X_MFR_ID 0x5449 + +struct hdc100x_chip_data { + bool support_mfr_check; +}; + +static const struct hdc100x_chip_data hdc100x_chip_data = { + .support_mfr_check = true, +}; + +static const struct hdc100x_chip_data hdc1008_chip_data = { + .support_mfr_check = false, +}; + struct hdc100x_data { struct i2c_client *client; struct mutex lock; @@ -351,8 +368,38 @@ static const struct iio_info hdc100x_info = { .attrs = &hdc100x_attribute_group, }; -static int hdc100x_probe(struct i2c_client *client, - const struct i2c_device_id *id) +static int hdc100x_read_mfr_id(struct i2c_client *client) +{ + return i2c_smbus_read_word_swapped(client, HDC100X_REG_MFR_ID); +} + +static int hdc100x_read_dev_id(struct i2c_client *client) +{ + return i2c_smbus_read_word_swapped(client, HDC100X_REG_DEV_ID); +} + +static bool is_valid_hdc100x(struct i2c_client *client) +{ + const struct of_device_id *match; + struct hdc100x_chip_data *data; + int mfr_id, dev_id; + + data = device_get_match_data(&client->dev); + if (data) { + if (!data->support_mfr_check) + return true; + } + + mfr_id = hdc100x_read_mfr_id(client); + dev_id = hdc100x_read_dev_id(client); + if (mfr_id == HDC100X_MFR_ID && + (dev_id == 0x1000 || dev_id == 0x1050)) + return true; + + return false; +} + +static int hdc100x_probe(struct i2c_client *client) { struct iio_dev *indio_dev; struct hdc100x_data *data; @@ -362,6 +409,9 @@ static int hdc100x_probe(struct i2c_client *client, I2C_FUNC_SMBUS_BYTE | I2C_FUNC_I2C)) return -EOPNOTSUPP; + if (!is_valid_hdc100x(client)) + return -EOPNOTSUPP; + indio_dev = devm_iio_device_alloc(&client->dev, sizeof(*data)); if (!indio_dev) return -ENOMEM; @@ -397,22 +447,22 @@ static int hdc100x_probe(struct i2c_client *client, } static const struct i2c_device_id hdc100x_id[] = { - { "hdc100x", 0 }, - { "hdc1000", 0 }, - { "hdc1008", 0 }, - { "hdc1010", 0 }, - { "hdc1050", 0 }, - { "hdc1080", 0 }, + { "hdc100X", (kernel_ulong_t)&hdc100x_chip_data }, + { "hdc1000", (kernel_ulong_t)&hdc100x_chip_data }, + { "hdc1008", (kernel_ulong_t)&hdc1008_chip_data }, + { "hdc1010", (kernel_ulong_t)&hdc100x_chip_data }, + { "hdc1050", (kernel_ulong_t)&hdc100x_chip_data }, + { "hdc1080", (kernel_ulong_t)&hdc100x_chip_data }, { } }; MODULE_DEVICE_TABLE(i2c, hdc100x_id); static const struct of_device_id hdc100x_dt_ids[] = { - { .compatible = "ti,hdc1000" }, - { .compatible = "ti,hdc1008" }, - { .compatible = "ti,hdc1010" }, - { .compatible = "ti,hdc1050" }, - { .compatible = "ti,hdc1080" }, + { .compatible = "ti,hdc1000", .data = &hdc100x_chip_data }, + { .compatible = "ti,hdc1008", .data = &hdc1008_chip_data }, + { .compatible = "ti,hdc1010", .data = &hdc100x_chip_data }, + { .compatible = "ti,hdc1050", .data = &hdc100x_chip_data }, + { .compatible = "ti,hdc1080", .data = &hdc100x_chip_data }, { } }; MODULE_DEVICE_TABLE(of, hdc100x_dt_ids); @@ -422,7 +472,7 @@ static struct i2c_driver hdc100x_driver = { .name = "hdc100x", .of_match_table = hdc100x_dt_ids, }, - .probe = hdc100x_probe, + .probe_new = hdc100x_probe, .id_table = hdc100x_id, }; module_i2c_driver(hdc100x_driver);
Add manufacturer and device ID checking during probe, and skip the checking if chip model not supported. Supported: - HDC1000 - HDC1010 - HDC1050 - HDC1080 Not supported: - HDC1008 Signed-off-by: Potin Lai <potin.lai.pt@gmail.com> --- LINK: [v1] https://lore.kernel.org/all/20220722172035.44977-1-potin.lai.pt@gmail.com/ changes v1 --> v2: - fix typo in commit message - change to use probe_new - use device_get_match_data instead of i2c_of_match_device --- drivers/iio/humidity/hdc100x.c | 78 ++++++++++++++++++++++++++++------ 1 file changed, 64 insertions(+), 14 deletions(-)