| Message ID | 20240624124941.113010-4-alisa.roman@analog.com (mailing list archive) |
|---|---|
| State | Changes Requested |
| Headers | show |
| Series | iio: adc: ad7192: Improvements | expand |
On Mon, Jun 24, 2024 at 3:50 PM Alisa-Dariana Roman <alisadariana@gmail.com> wrote: > Hello, A few comments inline. > There are actually 4 configuration modes of clock source for AD719X > devices. Either a crystal can be attached externally between MCLK1 and > MCLK2 pins, or an external CMOS-compatible clock can drive the MCLK2 > pin. The other 2 modes make use of the 4.92MHz internal clock. > > Removed properties adi,int-clock-output-enable and adi,clock-xtal were > undocumented. Use cleaner alternative of configuring external clock by > using clock names mclk and xtal. Should we keep the old properties for backwards compatibility? While I like the new approach, the downside is that someone upgrades the kernel and may run into issues with their board, because one of these properties went away. > > Removed functionality of AD7192_CLK_INT_CO restored in complementary > patch. > > Signed-off-by: Alisa-Dariana Roman <alisa.roman@analog.com> > --- > drivers/iio/adc/ad7192.c | 56 ++++++++++++++++++++-------------------- > 1 file changed, 28 insertions(+), 28 deletions(-) > > diff --git a/drivers/iio/adc/ad7192.c b/drivers/iio/adc/ad7192.c > index 334ab90991d4..940517df5429 100644 > --- a/drivers/iio/adc/ad7192.c > +++ b/drivers/iio/adc/ad7192.c > @@ -396,25 +396,37 @@ static inline bool ad7192_valid_external_frequency(u32 freq) > freq <= AD7192_EXT_FREQ_MHZ_MAX); > } > > -static int ad7192_clock_select(struct ad7192_state *st) > +static const char *const ad7192_clock_names[] = { > + "xtal", > + "mclk" Just a thought; no strong opinion. Maybe add a short comment about these being "clock_sel" values AD7192_CLK_EXT_MCLK1_2 & AD7192_CLK_EXT_MCLK2 ? This is mostly due to the "st->clock_sel = AD7192_CLK_EXT_MCLK1_2 + ret;" logic (which is fine) Before, this change, it would > +}; > + > +static int ad7192_clock_setup(struct ad7192_state *st) > { > struct device *dev = &st->sd.spi->dev; > - unsigned int clock_sel; > - > - clock_sel = AD7192_CLK_INT; > + int ret; > > - /* use internal clock */ > - if (!st->mclk) { > - if (device_property_read_bool(dev, "adi,int-clock-output-enable")) > - clock_sel = AD7192_CLK_INT_CO; > + ret = device_property_match_property_string(dev, "clock-names", > + ad7192_clock_names, > + ARRAY_SIZE(ad7192_clock_names)); > + if (ret < 0) { > + st->clock_sel = AD7192_CLK_INT; > + st->fclk = AD7192_INT_FREQ_MHZ; Since this is a new function, an early return can be added here. This would make the else statement redundant, and the function would have less indentation. So, something like: if (ret < 0) { st->clock_sel = AD7192_CLK_INT; st->fclk = AD7192_INT_FREQ_MHZ; return 0; } st->clock_sel = AD7192_CLK_EXT_MCLK1_2 + ret; st->mclk = devm_clk_get_enabled(dev, ad7192_clock_names[ret]); if (IS_ERR(st->mclk)) ................... > } else { > - if (device_property_read_bool(dev, "adi,clock-xtal")) > - clock_sel = AD7192_CLK_EXT_MCLK1_2; > - else > - clock_sel = AD7192_CLK_EXT_MCLK2; > + st->clock_sel = AD7192_CLK_EXT_MCLK1_2 + ret; > + > + st->mclk = devm_clk_get_enabled(dev, ad7192_clock_names[ret]); > + if (IS_ERR(st->mclk)) > + return dev_err_probe(dev, PTR_ERR(st->mclk), > + "Failed to get mclk\n"); > + > + st->fclk = clk_get_rate(st->mclk); > + if (!ad7192_valid_external_frequency(st->fclk)) > + return dev_err_probe(dev, -EINVAL, > + "External clock frequency out of bounds\n"); > } > > - return clock_sel; > + return 0; > } > > static int ad7192_setup(struct iio_dev *indio_dev, struct device *dev) > @@ -1275,21 +1287,9 @@ static int ad7192_probe(struct spi_device *spi) > if (ret) > return ret; > > - st->fclk = AD7192_INT_FREQ_MHZ; > - > - st->mclk = devm_clk_get_optional_enabled(dev, "mclk"); > - if (IS_ERR(st->mclk)) > - return PTR_ERR(st->mclk); > - > - st->clock_sel = ad7192_clock_select(st); > - > - if (st->clock_sel == AD7192_CLK_EXT_MCLK1_2 || > - st->clock_sel == AD7192_CLK_EXT_MCLK2) { > - st->fclk = clk_get_rate(st->mclk); > - if (!ad7192_valid_external_frequency(st->fclk)) > - return dev_err_probe(dev, -EINVAL, > - "External clock frequency out of bounds\n"); > - } > + ret = ad7192_clock_setup(st); > + if (ret) > + return ret; > > ret = ad7192_setup(indio_dev, dev); > if (ret) > -- > 2.34.1 > >
On Tue, 25 Jun 2024 08:30:47 +0300 Alexandru Ardelean <aardelean@baylibre.com> wrote: > On Mon, Jun 24, 2024 at 3:50 PM Alisa-Dariana Roman > <alisadariana@gmail.com> wrote: > > > > Hello, > > A few comments inline. > > > There are actually 4 configuration modes of clock source for AD719X > > devices. Either a crystal can be attached externally between MCLK1 and > > MCLK2 pins, or an external CMOS-compatible clock can drive the MCLK2 > > pin. The other 2 modes make use of the 4.92MHz internal clock. > > > > Removed properties adi,int-clock-output-enable and adi,clock-xtal were > > undocumented. Use cleaner alternative of configuring external clock by > > using clock names mclk and xtal. > > Should we keep the old properties for backwards compatibility? > While I like the new approach, the downside is that someone upgrades > the kernel and may run into issues with their board, because one of > these properties went away. > It's indeed a gamble that no one was actually using them. Whilst the lack of documentation suggests that might be the case, who knows. If it isn't a bit maintenance issue to keep the old property support in the driver it is probably a good idea (with lots of comments to say it's deprecated). > > > > Removed functionality of AD7192_CLK_INT_CO restored in complementary > > patch. > > > > Signed-off-by: Alisa-Dariana Roman <alisa.roman@analog.com> > > --- > > drivers/iio/adc/ad7192.c | 56 ++++++++++++++++++++-------------------- > > 1 file changed, 28 insertions(+), 28 deletions(-) > > > > diff --git a/drivers/iio/adc/ad7192.c b/drivers/iio/adc/ad7192.c > > index 334ab90991d4..940517df5429 100644 > > --- a/drivers/iio/adc/ad7192.c > > +++ b/drivers/iio/adc/ad7192.c > > @@ -396,25 +396,37 @@ static inline bool ad7192_valid_external_frequency(u32 freq) > > freq <= AD7192_EXT_FREQ_MHZ_MAX); > > } > > > > -static int ad7192_clock_select(struct ad7192_state *st) > > +static const char *const ad7192_clock_names[] = { > > + "xtal", > > + "mclk" > > Just a thought; no strong opinion. > Maybe add a short comment about these being "clock_sel" values > AD7192_CLK_EXT_MCLK1_2 & AD7192_CLK_EXT_MCLK2 ? > This is mostly due to the "st->clock_sel = AD7192_CLK_EXT_MCLK1_2 + > ret;" logic (which is fine) > Before, this change, it would Did we lose rest of sentence here? J
diff --git a/drivers/iio/adc/ad7192.c b/drivers/iio/adc/ad7192.c index 334ab90991d4..940517df5429 100644 --- a/drivers/iio/adc/ad7192.c +++ b/drivers/iio/adc/ad7192.c @@ -396,25 +396,37 @@ static inline bool ad7192_valid_external_frequency(u32 freq) freq <= AD7192_EXT_FREQ_MHZ_MAX); } -static int ad7192_clock_select(struct ad7192_state *st) +static const char *const ad7192_clock_names[] = { + "xtal", + "mclk" +}; + +static int ad7192_clock_setup(struct ad7192_state *st) { struct device *dev = &st->sd.spi->dev; - unsigned int clock_sel; - - clock_sel = AD7192_CLK_INT; + int ret; - /* use internal clock */ - if (!st->mclk) { - if (device_property_read_bool(dev, "adi,int-clock-output-enable")) - clock_sel = AD7192_CLK_INT_CO; + ret = device_property_match_property_string(dev, "clock-names", + ad7192_clock_names, + ARRAY_SIZE(ad7192_clock_names)); + if (ret < 0) { + st->clock_sel = AD7192_CLK_INT; + st->fclk = AD7192_INT_FREQ_MHZ; } else { - if (device_property_read_bool(dev, "adi,clock-xtal")) - clock_sel = AD7192_CLK_EXT_MCLK1_2; - else - clock_sel = AD7192_CLK_EXT_MCLK2; + st->clock_sel = AD7192_CLK_EXT_MCLK1_2 + ret; + + st->mclk = devm_clk_get_enabled(dev, ad7192_clock_names[ret]); + if (IS_ERR(st->mclk)) + return dev_err_probe(dev, PTR_ERR(st->mclk), + "Failed to get mclk\n"); + + st->fclk = clk_get_rate(st->mclk); + if (!ad7192_valid_external_frequency(st->fclk)) + return dev_err_probe(dev, -EINVAL, + "External clock frequency out of bounds\n"); } - return clock_sel; + return 0; } static int ad7192_setup(struct iio_dev *indio_dev, struct device *dev) @@ -1275,21 +1287,9 @@ static int ad7192_probe(struct spi_device *spi) if (ret) return ret; - st->fclk = AD7192_INT_FREQ_MHZ; - - st->mclk = devm_clk_get_optional_enabled(dev, "mclk"); - if (IS_ERR(st->mclk)) - return PTR_ERR(st->mclk); - - st->clock_sel = ad7192_clock_select(st); - - if (st->clock_sel == AD7192_CLK_EXT_MCLK1_2 || - st->clock_sel == AD7192_CLK_EXT_MCLK2) { - st->fclk = clk_get_rate(st->mclk); - if (!ad7192_valid_external_frequency(st->fclk)) - return dev_err_probe(dev, -EINVAL, - "External clock frequency out of bounds\n"); - } + ret = ad7192_clock_setup(st); + if (ret) + return ret; ret = ad7192_setup(indio_dev, dev); if (ret)
There are actually 4 configuration modes of clock source for AD719X devices. Either a crystal can be attached externally between MCLK1 and MCLK2 pins, or an external CMOS-compatible clock can drive the MCLK2 pin. The other 2 modes make use of the 4.92MHz internal clock. Removed properties adi,int-clock-output-enable and adi,clock-xtal were undocumented. Use cleaner alternative of configuring external clock by using clock names mclk and xtal. Removed functionality of AD7192_CLK_INT_CO restored in complementary patch. Signed-off-by: Alisa-Dariana Roman <alisa.roman@analog.com> --- drivers/iio/adc/ad7192.c | 56 ++++++++++++++++++++-------------------- 1 file changed, 28 insertions(+), 28 deletions(-)