Re: [PATCH v3 8/8] iio: tcs3472: implement wait time and sampling frequency
From: Aldo Conte
Date: Tue May 26 2026 - 11:49:23 EST
On 26/05/26 15:11, Jonathan Cameron wrote:
On Fri, 22 May 2026 14:34:19 +0200
Aldo Conte <aldocontelk@xxxxxxxxx> wrote:
The TCS3472 has a wait state controlled by the WEN bit in the ENABLEHi Aldo,
register and the WAIT register, with an additional WLONG bit in CONFIG
that if set multiplies the wait step by 12. The driver previously
defined TCS3472_WTIME but never used it leaving the TODO comment on
the top of the source file.
Implement control of the wait time through IIO_CHAN_INFO_SAMP_FREQ:
- Reading sampling_frequency returns the chip's current cycle time,
computed as the sum of ATIME, the fixed RGBC initialization time
and the wait time (which depends on WEN and WLONG).
- Writing sampling_frequency programs WTIME so that the resulting
cycle period approximates the requested frequency. If the
requested frequency cannot be reached with any
non-zero wait time, WEN is disabled and the chip runs
back-to-back conversions at the maximum rate allowed by ATIME.
If the requested period exceeds the maximum WTIME range, WLONG
is enabled to extend the wait step from 2.4 ms to 28.8 ms.
- The user's last requested frequency is stored in the driver's
private data so that subsequent changes to integration_time
recompute WTIME and preserve the requested sampling rate as
closely as possible.
Add TCS3472_ENABLE_WEN, TCS3472_ENABLE_RUN and TCS3472_CONFIG_WLONG
bit definitions. TCS3472_ENABLE_RUN bundles the bits
(AEN | PON | WEN) that are simultaneously set when the chip is in
running state and cleared during powerdown, and is used by
tcs3472_probe(), tcs3472_powerdown().
In tcs3472_resume(), only PON and AEN are re-asserted while WEN is
preserved from data->enable.
Remove the "TODO: wait time" comment at the top of the file.
Please look at Sashiko's comments on this particular patch
(the others are about an issue that has nothing to do with your series)
https://sashiko.dev/#/patchset/20260522123420.45495-1-aldocontelk%40gmail.com
The suggestion is that you need to make the timeout in
tcs3472_req_data() longer given with these settings the capture may
take a lot longer than previously.
There is some other general good practice stuff in there. For example:
+ data->enable &= ~TCS3472_ENABLE_WEN;
+ ret = i2c_smbus_write_byte_data(data->client,
+ TCS3472_ENABLE,
+ data->enable);
if you were to use a local variable to do the masking on enable
enable = data->enable & ~TCS3472_ENABLE_WEN;
ret = i2c_smbus_write_byte_data(data->client,..., enable);
if (ret)
return ret;
data->enable = enable;
You'd ensure that 'to best effort' the cached value is not updated
if the write fails. The approx (possibly correct) arguement is that
any failure in i2c_smbus_write_byte_data() is more likely to have
left the original state alone than successfully written the new one.
Or more generally the principle that in ideal world any call that fails
has no side effects.
Anyhow, take a look at all the feedback. If you disagree with any of
it feel free to add a note to the change log of the next version.
Ah. I've just noticed that as part of this you did make the change
I suggested in patch 3. Oops - I should have checked. It is fine
in here.
I didn't crop so that you'd have full context visible. A few specific
cases and comments inline.
Note I've picked up patches 1-7 so it's just this one that needs
revisions for v4.
thanks,
Jonathan
As below.
Signed-off-by: Aldo Conte <aldocontelk@xxxxxxxxx>
---
v3:
- Drop Suggested-by: Andy (inappropriate tag).
- TCS3472_ENABLE_RUN: new style.
- cycle_us: use div64_u64().
- cycle_us: cast val to u64 for 32-bit safety.
- wait_us: keep s64; added comment explaining the range bounds.
- Use 'if (ret)' instead of 'if (ret < 0)' after i2c writes.
- New helper tcs3472_cycle_to_freq() deduplicates val/val2 logic.
- !! -> ? 1 : 0 in probe.
- Fix tcs3472_resume(): preserve WEN from data->enable instead of
forcing TCS3472_ENABLE_RUN.
drivers/iio/light/tcs3472.c | 245 +++++++++++++++++++++++++++++++++---
1 file changed, 227 insertions(+), 18 deletions(-)
diff --git a/drivers/iio/light/tcs3472.c b/drivers/iio/light/tcs3472.c
index 1f597ca93697..81549d7eba35 100644
--- a/drivers/iio/light/tcs3472.c
+++ b/drivers/iio/light/tcs3472.c
@@ -9,8 +9,6 @@
* TCS34727)
*
* Datasheet: http://ams.com/eng/content/download/319364/1117183/file/TCS3472_Datasheet_EN_v2.pdf
- *
- * TODO: wait time
*/
#include <linux/cleanup.h>
@@ -53,19 +51,27 @@
#define TCS3472_STATUS_AINT BIT(4)
#define TCS3472_STATUS_AVALID BIT(0)
#define TCS3472_ENABLE_AIEN BIT(4)
+#define TCS3472_ENABLE_WEN BIT(3)
#define TCS3472_ENABLE_AEN BIT(1)
#define TCS3472_ENABLE_PON BIT(0)
+#define TCS3472_ENABLE_RUN \
+ (TCS3472_ENABLE_AEN | TCS3472_ENABLE_PON | TCS3472_ENABLE_WEN)
#define TCS3472_CONTROL_AGAIN_MASK (BIT(0) | BIT(1))
+#define TCS3472_CONFIG_WLONG BIT(1)
struct tcs3472_data {
struct i2c_client *client;
struct mutex lock;
+ int target_freq_hz;
+ int target_freq_uhz;
u16 low_thresh;
u16 high_thresh;
u8 enable;
u8 control;
u8 atime;
u8 apers;
+ u8 wtime;
+ bool wlong;
};
static const struct iio_event_spec tcs3472_events[] = {
@@ -91,6 +97,7 @@ static const struct iio_event_spec tcs3472_events[] = {
.info_mask_separate = BIT(IIO_CHAN_INFO_RAW), \
.info_mask_shared_by_type = BIT(IIO_CHAN_INFO_CALIBSCALE) | \
BIT(IIO_CHAN_INFO_INT_TIME), \
+ .info_mask_shared_by_all = BIT(IIO_CHAN_INFO_SAMP_FREQ), \
.channel2 = IIO_MOD_LIGHT_##_color, \
.address = _addr, \
.scan_index = _si, \
@@ -114,6 +121,47 @@ static const struct iio_chan_spec tcs3472_channels[] = {
IIO_CHAN_SOFT_TIMESTAMP(4),
};
+/*
+ * The chip's cycle time is the sum of three components:
+ * - ATIME: the programmable RGBC integration time.
+ * - The fixed RGBC initialization time (2400 us).
+ * - WTIME: the wait time, used only if WEN is set. If WLONG is active,
+ * the wait step is multiplied by 12 (2400 us -> 28800 us).
+ */
+static unsigned int tcs3472_cycle_time_us(struct tcs3472_data *data)
+{
+ unsigned int atime_us = (256 - data->atime) * 2400;
+ unsigned int init_us = 2400;
+ unsigned int wtime_us;
+
+ if (!(data->enable & TCS3472_ENABLE_WEN))
+ wtime_us = 0;
+ else if (data->wlong)
+ wtime_us = (256 - data->wtime) * 28800;
+ else
+ wtime_us = (256 - data->wtime) * 2400;
+
+ return atime_us + init_us + wtime_us;
+}
+
+/*
+ * Convert a cycle time in microseconds to a frequency in Hz and microhertz.
+ *
+ * Given cycle_us = T (the cycle period in microseconds), the corresponding
+ * frequency is:
+ * f = 1e6 / T [Hz]
+ *
+ * The result is split into the IIO_VAL_INT_PLUS_MICRO format:
+ * val = floor(1e6 / T) [Hz]
+ * val2 = (1e6 mod T) * 1e6 / T [microhertz]
+ */
+static void tcs3472_cycle_to_freq(unsigned int cycle_us, int *val, int *val2)
+{
+ *val = USEC_PER_SEC / cycle_us;
+ *val2 = div_u64((u64)(USEC_PER_SEC % cycle_us) * USEC_PER_SEC,
+ cycle_us);
+}
+
static int tcs3472_req_data(struct tcs3472_data *data)
{
int tries = 50;
@@ -166,16 +214,131 @@ static int tcs3472_read_raw(struct iio_dev *indio_dev,
*val = 0;
*val2 = (256 - data->atime) * 2400;
return IIO_VAL_INT_PLUS_MICRO;
+ case IIO_CHAN_INFO_SAMP_FREQ: {
+ unsigned int cycle_us = tcs3472_cycle_time_us(data);
+
+ tcs3472_cycle_to_freq(cycle_us, val, val2);
+ return IIO_VAL_INT_PLUS_MICRO;
+ }
default:
return -EINVAL;
}
}
+static int tcs3472_set_sampling_freq(struct tcs3472_data *data,
+ int val, int val2)
+{
+ unsigned int atime_us;
+ unsigned int init_us = 2400;
+ u64 cycle_us;
+ s64 wait_us;
+ int wtime;
+ bool wlong = false;
+ u8 config;
+ int ret;
+
+ if (val < 0 || val2 < 0 || (val == 0 && val2 == 0))
+ return -EINVAL;
+
+ guard(mutex)(&data->lock);
+
+ atime_us = (256 - data->atime) * 2400;
+ cycle_us = div64_u64(PSEC_PER_SEC,
+ (u64)val * USEC_PER_SEC + val2);
+
+ /*
+ * wait_us can be negative when the requested frequency is too high
+ * to be reached, or very large when the requested frequency is
+ * close to zero. Use s64 to cover the full range:
+ *
+ * cycle_us = PSEC_PER_SEC / (val * USEC_PER_SEC + val2)
+ *
+ * The divisor of the formula above reaches its maximum when
+ * val = val2 = INT_MAX:
+ * INT_MAX * USEC_PER_SEC + INT_MAX = ~2.15e18
+ * so cycle_us_min = floor(1e12 / 2.15e18) = 0.
+ *
+ * The divisor reaches its minimum (1) when val = 0 and val2 = 1,
+ * so cycle_us_max = 1e12 / 1 = 1e12.
+ *
+ * Therefore:
+ * wait_us_min = 0 - 2400 - 612000 = -616800
+ * wait_us_max = 1e12 - 2400 - 2400 = 999999995200
+ *
+ * Both fit comfortably in s64.
+ */
+ wait_us = (s64)cycle_us - init_us - atime_us;
+
+ if (wait_us < 2400) {
+ if (data->enable & TCS3472_ENABLE_WEN) {
+ data->enable &= ~TCS3472_ENABLE_WEN;
+ ret = i2c_smbus_write_byte_data(data->client,
+ TCS3472_ENABLE,
+ data->enable);
+ if (ret)
+ return ret;
+ }
+
+ data->target_freq_hz = val;
+ data->target_freq_uhz = val2;
+ return 0;
+ }
+
+ /*
+ * Wait state is needed: make sure WEN is active before programming
+ * WTIME (and possibly WLONG).
+ */
+ if (!(data->enable & TCS3472_ENABLE_WEN)) {
+ data->enable |= TCS3472_ENABLE_WEN;
Similar to below. Can we only update data->enable if the write succeeds.
+ ret = i2c_smbus_write_byte_data(data->client, TCS3472_ENABLE,This would be more consistent and easier to read if you have
+ data->enable);
+ if (ret)
+ return ret;
+ }
+
+ wtime = 256 - DIV_ROUND_CLOSEST_ULL(wait_us, 2400);
+ if (wtime < 0) {
+ wlong = true;
+ wtime = 256 - DIV_ROUND_CLOSEST_ULL(wait_us, 28800);
+ }
+ wtime = clamp(wtime, 0, 255);
+
+ if (wlong != data->wlong) {
+ ret = i2c_smbus_read_byte_data(data->client, TCS3472_CONFIG);
+ if (ret < 0)
+ return ret;
+
+ config = ret;
+ if (wlong)
+ config |= TCS3472_CONFIG_WLONG;
+ else
+ config &= ~TCS3472_CONFIG_WLONG;
+
+ ret = i2c_smbus_write_byte_data(data->client, TCS3472_CONFIG,
+ config);
+ if (ret)
+ return ret;
+
+ data->wlong = wlong;
+ }
+
+ ret = i2c_smbus_write_byte_data(data->client, TCS3472_WTIME, wtime);
+ if (ret)
+ return ret;
+
+ data->wtime = wtime;
+ data->target_freq_hz = val;
+ data->target_freq_uhz = val2;
+
+ return 0;
+}
+
static int tcs3472_write_raw(struct iio_dev *indio_dev,
struct iio_chan_spec const *chan,
int val, int val2, long mask)
{
struct tcs3472_data *data = iio_priv(indio_dev);
+ int ret;
int i;
switch (mask) {
@@ -196,15 +359,29 @@ static int tcs3472_write_raw(struct iio_dev *indio_dev,
if (val != 0)
return -EINVAL;
for (i = 0; i < 256; i++) {
- if (val2 == (256 - i) * 2400) {
- data->atime = i;
- return i2c_smbus_write_byte_data(
- data->client, TCS3472_ATIME,
- data->atime);
- }
-
+ if (val2 != (256 - i) * 2400)
+ continue;
+
+ data->atime = i;
+ ret = i2c_smbus_write_byte_data(data->client,
+ TCS3472_ATIME,
+ data->atime);
+ if (ret)
+ return ret;
+
+ /*
+ * ATIME just changed, so the cycle time changed too.
+ * Re-run the sampling frequency logic to recompute
+ * WTIME and preserve the user's last requested
+ * frequency.
+ */
+ return tcs3472_set_sampling_freq(data,
+ data->target_freq_hz,
+ data->target_freq_uhz);
}
return -EINVAL;
+ case IIO_CHAN_INFO_SAMP_FREQ:
+ return tcs3472_set_sampling_freq(data, val, val2);
default:
return -EINVAL;
}
@@ -434,16 +611,15 @@ static const struct iio_info tcs3472_info = {
static int tcs3472_powerdown(struct tcs3472_data *data)
{
int ret;
- u8 enable_mask = TCS3472_ENABLE_AEN | TCS3472_ENABLE_PON;
guard(mutex)(&data->lock);
ret = i2c_smbus_write_byte_data(data->client, TCS3472_ENABLE,
- data->enable & ~enable_mask);
+ data->enable & ~TCS3472_ENABLE_RUN);
if (ret)
return ret;
- data->enable &= ~enable_mask;
+ data->enable &= ~TCS3472_ENABLE_RUN;
u8 value;
value = data->enable & ~TCS3472_ENABLE_AEN | TCS3472_ENABLE_PON
value &= ~TCS3472_ENABLE_RUN;
ret = i2c_...
if (ret)
return ret;
data->enable = value;
return 0;
Or something along those lines. No need for separate enable_mask local
variable.
>
return 0;
}
@@ -458,6 +634,7 @@ static int tcs3472_probe(struct i2c_client *client)
struct device *dev = &client->dev;
struct tcs3472_data *data;
struct iio_dev *indio_dev;
+ unsigned int cycle_us;
int ret;
indio_dev = devm_iio_device_alloc(dev, sizeof(*data));
@@ -498,6 +675,16 @@ static int tcs3472_probe(struct i2c_client *client)
return ret;
data->atime = ret;
+ ret = i2c_smbus_read_byte_data(data->client, TCS3472_WTIME);
+ if (ret < 0)
+ return ret;
+ data->wtime = ret;
+
+ ret = i2c_smbus_read_byte_data(data->client, TCS3472_CONFIG);
+ if (ret < 0)
+ return ret;
+ data->wlong = (ret & TCS3472_CONFIG_WLONG) ? 1 : 0;
+
ret = i2c_smbus_read_word_data(data->client, TCS3472_AILT);
if (ret < 0)
return ret;
@@ -518,14 +705,29 @@ static int tcs3472_probe(struct i2c_client *client)
if (ret < 0)
return ret;
- /* enable device */
- data->enable = ret | TCS3472_ENABLE_PON | TCS3472_ENABLE_AEN;
+ /*
+ * Enable the chip in its full running state, including WEN. The
+ * actual wait time is controlled by the WTIME and WLONG registers,
+ * which retain their power-on defaults until userspace writes to
+ * sampling_frequency.
+ */
+ data->enable = ret | TCS3472_ENABLE_RUN;
data->enable &= ~TCS3472_ENABLE_AIEN;
Here is a case that should look more like the one that follows.
Use a local variable to build what is written and update the cache
only if the write succeeds.
ret = i2c_smbus_write_byte_data(data->client, TCS3472_ENABLE,
data->enable);
if (ret < 0)
return ret;
+ /*
+ * Initialize target frequency from the chip's current state so that
+ * subsequent integration_time changes via IIO_CHAN_INFO_INT_TIME can
+ * preserve a meaningful sampling rate, even before userspace writes
+ * sampling_frequency for the first time.
+ */
+ cycle_us = tcs3472_cycle_time_us(data);
+ tcs3472_cycle_to_freq(cycle_us, &data->target_freq_hz,
+ &data->target_freq_uhz);
+
ret = devm_add_action_or_reset(dev, tcs3472_powerdown_action, data);
if (ret)
return ret;
@@ -561,16 +763,23 @@ static int tcs3472_resume(struct device *dev)
struct tcs3472_data *data = iio_priv(i2c_get_clientdata(
to_i2c_client(dev)));
int ret;
- u8 enable_mask = TCS3472_ENABLE_AEN | TCS3472_ENABLE_PON;
+ u8 value;
guard(mutex)(&data->lock);
- ret = i2c_smbus_write_byte_data(data->client, TCS3472_ENABLE,
- data->enable | enable_mask);
+ /*
+ * On resume, only PON and AEN need to be re-asserted. WEN must not
+ * be forced on: its desired state was set by the user via
+ * sampling_frequency and is already stored in data->enable, so we
+ * read it from there to preserve the user's choice.
+ */
+ value = data->enable | TCS3472_ENABLE_PON | TCS3472_ENABLE_AEN;
+
+ ret = i2c_smbus_write_byte_data(data->client, TCS3472_ENABLE, value);
if (ret)
return ret;
- data->enable |= enable_mask;
+ data->enable = value;
return 0;
}
Hi Jonathan,
Thanks for the review. I'd like to ask a design
question about Sashiko's finding on the cached data->enable.
Sashiko pointed out:
> @@ -434,16 +611,15 @@ static const struct iio_info tcs3472_info = {
> static int tcs3472_powerdown(struct tcs3472_data *data)
> {
> int ret;
> - u8 enable_mask = TCS3472_ENABLE_AEN | TCS3472_ENABLE_PON;
>
> guard(mutex)(&data->lock);
>
> ret = i2c_smbus_write_byte_data(data->client, TCS3472_ENABLE,
> - data->enable & ~enable_mask);
> + data->enable & ~TCS3472_ENABLE_RUN);
> if (ret)
> return ret;
>
> - data->enable &= ~enable_mask;
> + data->enable &= ~TCS3472_ENABLE_RUN;
Does this permanently lose the WEN configuration after a suspend/resume cycle?
Because TCS3472_ENABLE_WEN is part of the TCS3472_ENABLE_RUN mask, this clears
the WEN bit in the cached data->enable. Upon resume, tcs3472_resume() attempts
to restore the configuration by reading this cache:
value = data->enable | TCS3472_ENABLE_PON | TCS3472_ENABLE_AEN;
But since the WEN bit was permanently cleared in the cache during suspend, the
device always resumes with WEN disabled, discarding the user's sampling
frequency configuration.
I see two possible fixes:
1. In tcs3472_powerdown(), clear only PON and AEN from data->enable
(keeping WEN as the "user's last desired state"). This is minimal
but introduces a divergence between data->enable and the actual
hardware ENABLE register, which breaks the simple "cache of the
HW register" semantic.
2. Add a separate field (e.g. bool target_wen) that holds the user's
desired WEN state, set in tcs3472_set_sampling_freq() and read
back in tcs3472_resume(). data->enable then strictly remains a
pure cache of the HW register.
With solution2, the functions would behave as follows:
- tcs3472_set_sampling_freq(): when activating wait state, sets
target_wen = true; when disabling it (frequency too high), sets
target_wen = false. data->enable is then updated to match what
is written to the chip.
- tcs3472_cycle_time_us(): reads target_wen instead of
(data->enable & WEN) to decide whether to include wtime_us in
the cycle. This avoids reporting a stale wait time during a
suspended state.
- tcs3472_powerdown(): unchanged in semantics clears PON, AEN
and WEN from both the chip and data->enable. target_wen is left
untouched, since it represents the user's wish, not the HW
state.
- tcs3472_resume(): rebuilds the ENABLE value from
PON | AEN | (target_wen ? WEN : 0), writes it to the chip, then
updates data->enable to match.
- tcs3472_probe(): initializes target_wen from the chip's current
WEN bit (read at probe time).
Solution 2 feels cleaner to me, but it adds a field and a bit of
synchronization between target_wen and (data->enable & WEN). Do you
have another approach in mind?
Thanks,
Aldo