IIO: Enable runtime checks on buffers size and related.
Subsystems (6)
Jonathan Cameron
maintainercover IIO: Enable runtime checks on buffers size and related.
From: Jonathan Cameron <[email protected]>
Main aim here is to switch almost all of the light drivers over to iio_push_to_buffers_with_ts() to provide runtime checking that we aren't copying from beyond the provided buffers.
A couple of bugs showed up whilst doing these conversions that should be back ported (patches 1 and 2).
Various other cases benefit from conversion from arrays to structures and putting the temporary storage on the stack. Given these cases are a little more complex, they are done as one or more patch per driver.
The cases that were just a simple change of function call from iio_push_to_buffers_with_timestamp() are grouped together in the final patch as not much to say about them and if we do a patch per driver this and related sets will be unnecessarily long!
If people would prefer I split that up that would be fine.
Jonathan
Jonathan Cameron (16):
iio: light: as73211: Ensure buffer holes are zeroed
iio: light: vcnl4035: Fix endianness vs data placement in buffer
issue.
iio: light: as73211: Use iio_push_to_buffers_with_ts() to allow source
size runtime check
iio: light: vcnl4035: Use a structure to make buffer arrangement
explicit.
iio: light: vcnl4035: Use iio_push_to_buffers_with_ts() to allow
runtime source buffer size check.
iio: light: acpi-als: Use a structure for layout of data to push to
buffer.
iio: light: acpi-als: Use iio_push_to_buffers_with_ts() to allow
runtime source size check
iio: light: adjd_s311: Use iio_push_to_buffers_with_ts() to allow
source size runtime check
iio: light: isl29125: Use iio_push_to_buffers_with_ts() to allow
source size runtime check
iio: light: max44000: Use iio_push_to_buffers_with_ts() to allow
source size runtime check
iio: light: st_uvis25: Use iio_push_to_buffers_with_ts() to allow
source size runtime check
iio: light: tcs3414: Use iio_push_to_buffers_with_ts() to allow source
size runtime check
iio: light: tcs3472: Use iio_push_to_buffers_with_ts() to allow source
size runtime check
iio: light: vcnl4000: Use a structure to make buffer arrangement
explicit.
iio: light: vl6180: Use iio_push_to_buffers_with_ts() to allow source
size runtime check
iio: light: Simple conversions to iio_push_to_buffers_with_ts()
drivers/iio/light/acpi-als.c | 19 ++++++------------- drivers/iio/light/adjd_s311.c | 12 ++++++------ drivers/iio/light/as73211.c | 5 +++-- drivers/iio/light/bh1745.c | 4 ++-- drivers/iio/light/hid-sensor-als.c | 5 +++-- drivers/iio/light/isl29125.c | 14 +++++++------- drivers/iio/light/ltr501.c | 4 ++-- drivers/iio/light/max44000.c | 18 +++++++++--------- drivers/iio/light/opt4060.c | 2 +- drivers/iio/light/rohm-bu27034.c | 3 ++- drivers/iio/light/rpr0521.c | 4 ++-- drivers/iio/light/si1145.c | 5 +++-- drivers/iio/light/st_uvis25.h | 5 ----- drivers/iio/light/st_uvis25_core.c | 12 +++++++++--- drivers/iio/light/tcs3414.c | 15 ++++++++------- drivers/iio/light/tcs3472.c | 14 +++++++------- drivers/iio/light/vcnl4000.c | 11 +++++++---- drivers/iio/light/vcnl4035.c | 13 +++++++++---- drivers/iio/light/veml6030.c | 2 +- drivers/iio/light/vl6180.c | 14 +++++++------- 20 files changed, 94 insertions(+), 87 deletions(-)
Signature · 1 line
2.50.1
Jonathan Cameron
maintainerPATCH 1/16 iio: light: as73211: Ensure buffer holes are zeroed
From: Jonathan Cameron <[email protected]>
Given that the buffer is copied to a kfifo that ultimately user space can read, ensure we zero it.
--- drivers/iio/light/as73211.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/iio/light/as73211.c +1 -1
@@ -639,7 +639,7 @@ static irqreturn_t as73211_trigger_handler(int irq __always_unused, void *p)struct {__le16 chan[4 ];aligned_s64 ts; - } scan; + } scan = { };int data_result, ret;mutex_lock (&data->mutex);
Signature · 1 line
2.50.1
Re: [PATCH 01/16] iio: light: as73211: Ensure buffer holes are zeroed
Quoted 17 lines · Jonathan Cameron
From: Jonathan Cameron <[email protected]>
Given that the buffer is copied to a kfifo that ultimately user space can read, ensure we zero it.
Fixes 403e5586b52e ("iio: light: as73211: New driver")Signed-off-by Jonathan Cameron <[email protected]>Cc Christian Eggers <Christian Eggers <[email protected]>--- drivers/iio/light/as73211.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)drivers/iio/light/as73211.c +0
-0
drivers/iio/light/as73211.c static irqreturn_t as73211_trigger_handler(int irq __always_unused, void *p) · line 645
Quoted 8 lines
drivers/iio/light/as73211.c +1
-1@@ -639,7 +639,7 @@ static irqreturn_t as73211_trigger_handler(int irq __always_unused, void *p)struct {__le16 chan[4 ];aligned_s64 ts; - } scan; + } scan = { };int data_result, ret;mutex_lock (&data->mutex);
Jonathan Cameron
maintainerPATCH 2/16 iio: light: vcnl4035: Fix endianness vs data placement in buffer issue.
In reply to Jonathan CameronFrom: Jonathan Cameron <[email protected]>
The assumption is that the channel ends up in the first 16 bits of the buffer. On a big endian system, the regmap_read() will read a 16 bit value into the 4 byte location, leaving the value in bytes 2 and 3. Fix this by using a a local variable and copying into the current location.
--- drivers/iio/light/vcnl4035.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/iio/light/vcnl4035.c +3 -1
@@ -104,14 +104,16 @@ static irqreturn_t vcnl4035_trigger_consumer_handler(int irq, void *p)struct vcnl4035_data *data =iio_priv (indio_dev);/* Ensure naturally aligned timestamp */ u8 buffer[ALIGN (sizeof (u16),sizeof (s64)) +sizeof (s64)]__aligned(8) = { }; +int val;int ret; - ret =regmap_read (data->regmap,VCNL4035_ALS_DATA , (int *)buffer); + ret =regmap_read (data->regmap,VCNL4035_ALS_DATA , &val);if (ret <0 ) {dev_err (&data->client->dev,"Trigger consumer can't read from sensor.\n" ); goto fail_read; } + *((u16 *)buffer) = val;iio_push_to_buffers_with_timestamp (indio_dev, buffer,iio_get_time_ns (indio_dev));
Signature · 1 line
2.50.1
Re: [PATCH 02/16] iio: light: vcnl4035: Fix endianness vs data placement in buffer issue.
Quoted 6 lines · Jonathan Cameron
From: Jonathan Cameron <[email protected]>
The assumption is that the channel ends up in the first 16 bits of the buffer. On a big endian system, the
regmap_read()will read a 16 bit value into the 4 byte location, leaving the value in bytes
16-bit
4-byte
Quoted 1 line
2 and 3. Fix this by using a a local variable and copying into the
a a --> a
Quoted 1 line
current location.
...
Quoted 1 line
+ int val;
Why signed? regmap API uses an unsigned type for values.
...
Quoted 1 line
+ *((u16 *)buffer) = val;
I don't understand this fix. Does it mean we simply transfer from HW to the user space in whatever endianess HW does this?
Signature · 2 lines
With Best Regards, Andy Shevchenko
Jonathan Cameron
maintainerRe: [PATCH 02/16] iio: light: vcnl4035: Fix endianness vs data placement in buffer issue.
Quoted 15 lines · On Sun, 3 Aug 2025 21:16:27 +0200 Andy Shevchenko <[email protected]>…
Quoted 6 lines · Jonathan Cameron
From: Jonathan Cameron <[email protected]>
The assumption is that the channel ends up in the first 16 bits of the buffer. On a big endian system, the
regmap_read()will read a 16 bit value into the 4 byte location, leaving the value in bytes16-bit
4-byteQuoted 1 line
2 and 3. Fix this by using a a local variable and copying into the
a a --> a
Quoted 8 lines
Quoted 1 line
current location.
...
Quoted 1 line
+ int val;
Why signed? regmap API uses an unsigned type for values.
Good point. I got thrown by the type in the incorrect cast.
Quoted 7 lines
...
Quoted 1 line
+ *((u16 *)buffer) = val;
I don't understand this fix. Does it mean we simply transfer from HW to the user space in whatever endianess HW does this?
I was thinking that was dealt with by the regmap and the iio_chan_spec.
The regmap part is fine as sets REGMAP_ENDIAN_LITTLE but then I'd expect the reads to result in cpu endian from that. The chan spec says little endian though and I think it should CPU endian.
I'll spin a new little series to resolve that as well and fix this one up.
Jonathan
Re: [PATCH 02/16] iio: light: vcnl4035: Fix endianness vs data placement in buffer issue.
In reply to Jonathan CameronQuoted 26 lines · Jonathan Cameron
From: Jonathan Cameron <[email protected]>
The assumption is that the channel ends up in the first 16 bits of the buffer. On a big endian system, the
regmap_read()will read a 16 bit value into the 4 byte location, leaving the value in bytes 2 and 3. Fix this by using a a local variable and copying into the current location.Fixes 55707294c4eb ("iio: light: Add support for vishay vcnl4035")Signed-off-by Jonathan Cameron <[email protected]>--- drivers/iio/light/vcnl4035.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)drivers/iio/light/vcnl4035.c +1
-1@@ -104,14 +104,16 @@ static irqreturn_t vcnl4035_trigger_consumer_handler(int irq, void *p)struct vcnl4035_data *data =iio_priv (indio_dev);/* Ensure naturally aligned timestamp */ u8 buffer[ALIGN (sizeof (u16),sizeof (s64)) +sizeof (s64)]__aligned(8) = { }; +int val;int ret; - ret =regmap_read (data->regmap,VCNL4035_ALS_DATA , (int *)buffer);
drivers/iio/light/vcnl4035.c static irqreturn_t vcnl4035_trigger_consumer_handler(int irq, void *p) · line 119
Quoted 10 lines
drivers/iio/light/vcnl4035.c +2
-0+ ret =regmap_read (data->regmap,VCNL4035_ALS_DATA , &val);if (ret <0 ) {dev_err (&data->client->dev,"Trigger consumer can't read from sensor.\n" ); goto fail_read; } + *((u16 *)buffer) = val;iio_push_to_buffers_with_timestamp (indio_dev, buffer,iio_get_time_ns (indio_dev));
(Although, I'm not sure if the local buffer could be a packed struct?)
Jonathan Cameron
maintainerPATCH 3/16 iio: light: as73211: Use iio_push_to_buffers_with_ts() to allow source size runtime check
In reply to Jonathan CameronFrom: Jonathan Cameron <[email protected]>
Given the helper routine requires additional aligned space for the timestamp, it is easy to get bugs. The new helper takes a size parameter to sanity check against the space actually used.
--- drivers/iio/light/as73211.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/iio/light/as73211.c +2 -1
@@ -698,7 +698,8 @@ static irqreturn_t as73211_trigger_handler(int irq __always_unused, void *p) } } -iio_push_to_buffers_with_timestamp (indio_dev, &scan,iio_get_time_ns (indio_dev)); +iio_push_to_buffers_with_ts (indio_dev, &scan,sizeof (scan), +iio_get_time_ns (indio_dev)); done:mutex_unlock (&data->mutex);
Signature · 1 line
2.50.1
Jonathan Cameron
maintainerPATCH 4/16 iio: light: vcnl4035: Use a structure to make buffer arrangement explicit.
In reply to Jonathan CameronFrom: Jonathan Cameron <[email protected]>
Using a structure to arrange the data to be passed to iio_push_to_buffers_with_timestamp() makes the placement and padding explicit, removing the need for comments to explain what is going on.
--- drivers/iio/light/vcnl4035.c | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-)
drivers/iio/light/vcnl4035.c +6 -3
@@ -103,7 +103,10 @@ static irqreturn_t vcnl4035_trigger_consumer_handler(int irq, void *p)struct iio_dev *indio_dev = pf->indio_dev;struct vcnl4035_data *data =iio_priv (indio_dev);/* Ensure naturally aligned timestamp */ -u8 buffer[ALIGN (sizeof (u16),sizeof (s64)) +sizeof (s64)]__aligned(8) = { }; +struct { +u16 chan; +aligned_s64 ts; + } scan;int val;int ret;
@@ -113,8 +116,8 @@ static irqreturn_t vcnl4035_trigger_consumer_handler(int irq, void *p) "Trigger consumer can't read from sensor.\n"); goto fail_read; } - *((u16 *)buffer) = val; -iio_push_to_buffers_with_timestamp (indio_dev, buffer, + scan.chan = val; +iio_push_to_buffers_with_timestamp (indio_dev, &scan,iio_get_time_ns (indio_dev)); fail_read:
Signature · 1 line
2.50.1
Re: [PATCH 04/16] iio: light: vcnl4035: Use a structure to make buffer arrangement explicit.
Quoted 28 lines · Jonathan Cameron
From: Jonathan Cameron <[email protected]>
Using a structure to arrange the data to be passed to
iio_push_to_buffers_with_timestamp()makes the placement and padding explicit, removing the need for comments to explain what is going on.Signed-off-by Jonathan Cameron <[email protected]>--- drivers/iio/light/vcnl4035.c | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-)drivers/iio/light/vcnl4035.c +4
-1@@ -103,7 +103,10 @@ static irqreturn_t vcnl4035_trigger_consumer_handler(int irq, void *p)struct iio_dev *indio_dev = pf->indio_dev;struct vcnl4035_data *data =iio_priv (indio_dev);/* Ensure naturally aligned timestamp */ -u8 buffer[ALIGN (sizeof (u16),sizeof (s64)) +sizeof (s64)]__aligned(8) = { }; +struct { +u16 chan; +aligned_s64 ts; + } scan;int val;int ret;
drivers/iio/light/vcnl4035.c static irqreturn_t vcnl4035_trigger_consumer_handler(int irq, void *p) · line 123
Quoted 10 lines
drivers/iio/light/vcnl4035.c +2
-2@@ -113,8 +116,8 @@ static irqreturn_t vcnl4035_trigger_consumer_handler(int irq, void *p) "Trigger consumer can't read from sensor.\n"); goto fail_read; } - *((u16 *)buffer) = val; -iio_push_to_buffers_with_timestamp (indio_dev, buffer, + scan.chan = val; +iio_push_to_buffers_with_timestamp (indio_dev, &scan,iio_get_time_ns (indio_dev)); fail_read:
I suppose I was too hasty commenting the previous patch :) Besides, I misread the buffer definition. I just thought there were room for the u16 and couple of s64s. Hence I suggested a packed struct...
Now, re-reading this and seeing your patch - this looks correct :)
Yours,
-- Matti
Jonathan Cameron
maintainerPATCH 5/16 iio: light: vcnl4035: Use iio_push_to_buffers_with_ts() to allow runtime source buffer size check.
In reply to Jonathan CameronFrom: Jonathan Cameron <[email protected]>
As the sizing of the source data has to include space for an aligned timestamp it is a common source of bugs. Using this new function that takes the size of the provided buffer enables runtime checks for possible bugs.
--- drivers/iio/light/vcnl4035.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/iio/light/vcnl4035.c +2 -2
@@ -117,8 +117,8 @@ static irqreturn_t vcnl4035_trigger_consumer_handler(int irq, void *p) goto fail_read; } scan.chan = val; -iio_push_to_buffers_with_timestamp (indio_dev, &scan, -iio_get_time_ns (indio_dev)); +iio_push_to_buffers_with_ts (indio_dev, &scan,sizeof (scan), +iio_get_time_ns (indio_dev)); fail_read:iio_trigger_notify_done (indio_dev->trig);
Signature · 1 line
2.50.1
Re: [PATCH 05/16] iio: light: vcnl4035: Use iio_push_to_buffers_with_ts() to allow runtime source buffer size check.
Quoted 17 lines · Jonathan Cameron
From: Jonathan Cameron <[email protected]>
As the sizing of the source data has to include space for an aligned timestamp it is a common source of bugs. Using this new function that takes the size of the provided buffer enables runtime checks for possible bugs.
Signed-off-by Jonathan Cameron <[email protected]>--- drivers/iio/light/vcnl4035.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)drivers/iio/light/vcnl4035.c +0
-0
drivers/iio/light/vcnl4035.c static irqreturn_t vcnl4035_trigger_consumer_handler(int irq, void *p) · line 124
Quoted 10 lines
drivers/iio/light/vcnl4035.c +2
-2@@ -117,8 +117,8 @@ static irqreturn_t vcnl4035_trigger_consumer_handler(int irq, void *p) goto fail_read; } scan.chan = val; -iio_push_to_buffers_with_timestamp (indio_dev, &scan, -iio_get_time_ns (indio_dev)); +iio_push_to_buffers_with_ts (indio_dev, &scan,sizeof (scan), +iio_get_time_ns (indio_dev)); fail_read:iio_trigger_notify_done (indio_dev->trig);
(Although, seeing there is 3 patches touching the same spot, I might squash the patch 04/16 with the 03/16. I would leave this one as a standalone so the fix can be backported without this new API, if needed).
Yours,
-- Matti
Jonathan Cameron
maintainerRe: [PATCH 05/16] iio: light: vcnl4035: Use iio_push_to_buffers_with_ts() to allow runtime source buffer size check.
Quoted 26 lines · On Mon, 4 Aug 2025 09:57:46 +0300 Matti Vaittinen <[email protected]>…
Quoted 25 lines · Jonathan Cameron
From: Jonathan Cameron <[email protected]>
As the sizing of the source data has to include space for an aligned timestamp it is a common source of bugs. Using this new function that takes the size of the provided buffer enables runtime checks for possible bugs.
Signed-off-by Jonathan Cameron <[email protected]>--- drivers/iio/light/vcnl4035.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)drivers/iio/light/vcnl4035.c +2
-2@@ -117,8 +117,8 @@ static irqreturn_t vcnl4035_trigger_consumer_handler(int irq, void *p) goto fail_read; } scan.chan = val; -iio_push_to_buffers_with_timestamp (indio_dev, &scan, -iio_get_time_ns (indio_dev)); +iio_push_to_buffers_with_ts (indio_dev, &scan,sizeof (scan), +iio_get_time_ns (indio_dev));
Quoted 8 lines
drivers/iio/light/vcnl4035.c static irqreturn_t vcnl4035_trigger_consumer_handler(int irq, void *p) · line 124
Quoted 2 lines
drivers/iio/light/vcnl4035.c +0
-0fail_read:iio_trigger_notify_done (indio_dev->trig);Reviewed-by Matti Vaittinen <[email protected]>(Although, seeing there is 3 patches touching the same spot, I might squash the patch 04/16 with the 03/16. I would leave this one as a standalone so the fix can be backported without this new API, if needed).
Whilst all the same spot I'm keen to keep them separate as they are all unrelated changes.
I'll be respinning this anyway with an additional fix for the buffer type.
Jonathan
Quoted 4 lines
Yours,
-- Matti
Jonathan Cameron
maintainerPATCH 6/16 iio: light: acpi-als: Use a structure for layout of data to push to buffer.
In reply to Jonathan CameronFrom: Jonathan Cameron <[email protected]>
Using a structure makes the padding and alignment rules explicit, removing the need for a comment.
Also move the storage to the stack as it is only 16 bytes.
--- drivers/iio/light/acpi-als.c | 19 ++++++------------- 1 file changed, 6 insertions(+), 13 deletions(-)
drivers/iio/light/acpi-als.c +6 -13
@@ -49,20 +49,10 @@ static const struct iio_chan_spec acpi_als_channels[] = { IIO_CHAN_SOFT_TIMESTAMP(1), }; -/* -* The event buffer contains timestamp and all the data from -* the ACPI0008 block. There are multiple, but so far we only -* support _ALI (illuminance): One channel, padding and timestamp. -*/ -#define ACPI_ALS_EVT_BUFFER_SIZE \ - (sizeof(s32) + sizeof(s32) + sizeof(s64)) -struct acpi_als {struct acpi_device *device;struct mutex lock;struct iio_trigger *trig; - -s32 evt_buffer[ACPI_ALS_EVT_BUFFER_SIZE /sizeof (s32)]__aligned(8) ; };/*
@@ -152,7 +142,10 @@ static irqreturn_t acpi_als_trigger_handler(int irq, void *p)struct iio_poll_func *pf = p;struct iio_dev *indio_dev = pf->indio_dev;struct acpi_als *als =iio_priv (indio_dev); -s32 *buffer = als->evt_buffer; +struct { +s32 light; +aligned_s64 ts; + } scan = { };s32 val;int ret;
@@ -161,7 +154,7 @@ static irqreturn_t acpi_als_trigger_handler(int irq, void *p) ret =acpi_als_read_value (als,ACPI_ALS_ILLUMINANCE , &val);if (ret <0 ) goto out; - *buffer = val; + scan.light = val;/* * When coming from own trigger via polls, set polling function
@@ -174,7 +167,7 @@ static irqreturn_t acpi_als_trigger_handler(int irq, void *p)if (!pf->timestamp) pf->timestamp =iio_get_time_ns (indio_dev); -iio_push_to_buffers_with_timestamp (indio_dev, buffer, pf->timestamp); +iio_push_to_buffers_with_timestamp (indio_dev, &scan, pf->timestamp); out:mutex_unlock (&als->lock);iio_trigger_notify_done (indio_dev->trig);
Signature · 1 line
2.50.1
Re: [PATCH 06/16] iio: light: acpi-als: Use a structure for layout of data to push to buffer.
Quoted 5 lines · Jonathan Cameron
Using a structure makes the padding and alignment rules explicit, removing the need for a comment.
Also move the storage to the stack as it is only 16 bytes.
...
drivers/iio/light/acpi-als.c static const struct iio_chan_spec acpi_als_channels[] = { · line 58
Quoted 7 lines
drivers/iio/light/acpi-als.c +0
-7-/* -* The event buffer contains timestamp and all the data from -* the ACPI0008 block. There are multiple, but so far we only -* support _ALI (illuminance): One channel, padding and timestamp. -*/ -#define ACPI_ALS_EVT_BUFFER_SIZE \ - (sizeof(s32) + sizeof(s32) + sizeof(s64))
IIUC this definition is named in a bit ambiguous way. This size is related to the driver for ACPI light sensor, but it doesn't mean that size is derived from any of the ACPI specification, killing that is a right thing to do, hence I like this patch very much.
Signature · 2 lines
With Best Regards, Andy Shevchenko
Re: [PATCH 06/16] iio: light: acpi-als: Use a structure for layout of data to push to buffer.
Quoted 9 lines · On Sun, 3 Aug 2025 21:24:13 +0200 Andy Shevchenko <[email protected]>…
Quoted 5 lines · Jonathan Cameron
Using a structure makes the padding and alignment rules explicit, removing the need for a comment.
Also move the storage to the stack as it is only 16 bytes.
...
Quoted 13 lines
drivers/iio/light/acpi-als.c static const struct iio_chan_spec acpi_als_channels[] = { · line 58
Quoted 7 lines
drivers/iio/light/acpi-als.c +0
-7-/* -* The event buffer contains timestamp and all the data from -* the ACPI0008 block. There are multiple, but so far we only -* support _ALI (illuminance): One channel, padding and timestamp. -*/ -#define ACPI_ALS_EVT_BUFFER_SIZE \ - (sizeof(s32) + sizeof(s32) + sizeof(s64))IIUC this definition is named in a bit ambiguous way. This size is related to the driver for ACPI light sensor, but it doesn't mean that size is derived from any of the ACPI specification, killing that is a right thing to do, hence I like this patch very much.
Applied.
J
Jonathan Cameron
maintainerPATCH 7/16 iio: light: acpi-als: Use iio_push_to_buffers_with_ts() to allow runtime source size check
In reply to Jonathan CameronFrom: Jonathan Cameron <[email protected]>
This function allows for runtime detection of undersized storage which can be non obvious due to the injection of a timestamp within the helper.
--- drivers/iio/light/acpi-als.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/iio/light/acpi-als.c +1 -1
@@ -167,7 +167,7 @@ static irqreturn_t acpi_als_trigger_handler(int irq, void *p)if (!pf->timestamp) pf->timestamp =iio_get_time_ns (indio_dev); -iio_push_to_buffers_with_timestamp (indio_dev, &scan, pf->timestamp); +iio_push_to_buffers_with_ts (indio_dev, &scan,sizeof (scan), pf->timestamp); out:mutex_unlock (&als->lock);iio_trigger_notify_done (indio_dev->trig);
Signature · 1 line
2.50.1
Re: [PATCH 07/16] iio: light: acpi-als: Use iio_push_to_buffers_with_ts() to allow runtime source size check
Quoted 7 lines · On Sat, 2 Aug 2025 17:44:27 +0100 Jonathan Cameron <[email protected]> wrote:
From: Jonathan Cameron <[email protected]>
This function allows for runtime detection of undersized storage which can be non obvious due to the injection of a timestamp within the helper.
Signed-off-by Jonathan Cameron <[email protected]>Cc Gwendal Grignou <[email protected]>
Applied
Quoted 17 lines
--- drivers/iio/light/acpi-als.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)drivers/iio/light/acpi-als.c +1
-1@@ -167,7 +167,7 @@ static irqreturn_t acpi_als_trigger_handler(int irq, void *p)if (!pf->timestamp) pf->timestamp =iio_get_time_ns (indio_dev); -iio_push_to_buffers_with_timestamp (indio_dev, &scan, pf->timestamp); +iio_push_to_buffers_with_ts (indio_dev, &scan,sizeof (scan), pf->timestamp); out:mutex_unlock (&als->lock);iio_trigger_notify_done (indio_dev->trig);
Jonathan Cameron
maintainerPATCH 8/16 iio: light: adjd_s311: Use iio_push_to_buffers_with_ts() to allow source size runtime check
In reply to Jonathan CameronFrom: Jonathan Cameron <[email protected]>
Also move the structure used as the source to the stack as it is only 16 bytes and not the target of an DMA or similar.
--- drivers/iio/light/adjd_s311.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-)
drivers/iio/light/adjd_s311.c +6 -6
@@ -54,10 +54,6 @@struct adjd_s311_data {struct i2c_client *client; -struct { -s16 chans[4 ]; -aligned_s64 ts; - } scan; };enum adjd_s311_channel_idx {
@@ -120,6 +116,10 @@ static irqreturn_t adjd_s311_trigger_handler(int irq, void *p)struct adjd_s311_data *data =iio_priv (indio_dev);s64 time_ns =iio_get_time_ns (indio_dev);int i, j =0 ; +struct { +s16 chans[4 ]; +aligned_s64 ts; + } scan = { };int ret =adjd_s311_req_data (indio_dev);if (ret <0 )
@@ -131,10 +131,10 @@ static irqreturn_t adjd_s311_trigger_handler(int irq, void *p)if (ret <0 ) goto done; - data->scan.chans[j++] = ret &ADJD_S311_DATA_MASK ; + scan.chans[j++] = ret &ADJD_S311_DATA_MASK ; } -iio_push_to_buffers_with_timestamp (indio_dev, &data->scan, time_ns); +iio_push_to_buffers_with_ts (indio_dev, &scan,sizeof (scan), time_ns); done:iio_trigger_notify_done (indio_dev->trig);
Signature · 1 line
2.50.1
Re: [PATCH 08/16] iio: light: adjd_s311: Use iio_push_to_buffers_with_ts() to allow source size runtime check
Quoted 39 lines · Jonathan Cameron
From: Jonathan Cameron <[email protected]>
Also move the structure used as the source to the stack as it is only 16 bytes and not the target of an DMA or similar.
Signed-off-by Jonathan Cameron <[email protected]>--- drivers/iio/light/adjd_s311.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-)drivers/iio/light/adjd_s311.c +4
-4@@ -54,10 +54,6 @@struct adjd_s311_data {struct i2c_client *client; -struct { -s16 chans[4 ]; -aligned_s64 ts; - } scan; };enum adjd_s311_channel_idx {@@ -120,6 +116,10 @@ static irqreturn_t adjd_s311_trigger_handler(int irq, void *p)struct adjd_s311_data *data =iio_priv (indio_dev);s64 time_ns =iio_get_time_ns (indio_dev);int i, j =0 ; +struct { +s16 chans[4 ]; +aligned_s64 ts; + } scan = { };int ret =adjd_s311_req_data (indio_dev);if (ret <0 )@@ -131,10 +131,10 @@ static irqreturn_t adjd_s311_trigger_handler(int irq, void *p)if (ret <0 ) goto done;
drivers/iio/light/adjd_s311.c static irqreturn_t adjd_s311_trigger_handler(int irq, void *p) · line 140
Quoted 10 lines
drivers/iio/light/adjd_s311.c +2
-2- data->scan.chans[j++] = ret &ADJD_S311_DATA_MASK ; + scan.chans[j++] = ret &ADJD_S311_DATA_MASK ; } -iio_push_to_buffers_with_timestamp (indio_dev, &data->scan, time_ns); +iio_push_to_buffers_with_ts (indio_dev, &scan,sizeof (scan), time_ns); done:iio_trigger_notify_done (indio_dev->trig);
Re: [PATCH 08/16] iio: light: adjd_s311: Use iio_push_to_buffers_with_ts() to allow source size runtime check
Quoted 7 lines · On Mon, 4 Aug 2025 10:04:40 +0300 Matti Vaittinen <[email protected]>…
Quoted 6 lines · Jonathan Cameron
From: Jonathan Cameron <[email protected]>
Also move the structure used as the source to the stack as it is only 16 bytes and not the target of an DMA or similar.
Signed-off-by Jonathan Cameron <[email protected]>
Applied.
Jonathan Cameron
maintainerPATCH 9/16 iio: light: isl29125: Use iio_push_to_buffers_with_ts() to allow source size runtime check
In reply to Jonathan CameronFrom: Jonathan Cameron <[email protected]>
Also move the structure used as the source to the stack as it is only 16 bytes and not the target of an DMA or similar.
--- drivers/iio/light/isl29125.c | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-)
drivers/iio/light/isl29125.c +7 -7
@@ -51,11 +51,6 @@struct isl29125_data {struct i2c_client *client;u8 conf1; -/* Ensure timestamp is naturally aligned */ -struct { -u16 chans[3 ]; -aligned_s64 timestamp; - } scan; };#define ISL29125_CHANNEL (_color, _si) { \
@@ -179,6 +174,11 @@ static irqreturn_t isl29125_trigger_handler(int irq, void *p)struct iio_dev *indio_dev = pf->indio_dev;struct isl29125_data *data =iio_priv (indio_dev);int i, j =0 ; +/* Ensure timestamp is naturally aligned */ +struct { +u16 chans[3 ]; +aligned_s64 timestamp; + } scan = { };iio_for_each_active_channel (indio_dev, i) { int ret = i2c_smbus_read_word_data(data->client,
@@ -186,10 +186,10 @@ static irqreturn_t isl29125_trigger_handler(int irq, void *p)if (ret <0 ) goto done; - data->scan.chans[j++] = ret; + scan.chans[j++] = ret; } -iio_push_to_buffers_with_timestamp (indio_dev, &data->scan, +iio_push_to_buffers_with_ts (indio_dev, &scan,sizeof (scan),iio_get_time_ns (indio_dev)); done:
Signature · 1 line
2.50.1
Re: [PATCH 09/16] iio: light: isl29125: Use iio_push_to_buffers_with_ts() to allow source size runtime check
Quoted 41 lines · Jonathan Cameron
From: Jonathan Cameron <[email protected]>
Also move the structure used as the source to the stack as it is only 16 bytes and not the target of an DMA or similar.
Signed-off-by Jonathan Cameron <[email protected]>--- drivers/iio/light/isl29125.c | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-)drivers/iio/light/isl29125.c +5
-5@@ -51,11 +51,6 @@struct isl29125_data {struct i2c_client *client;u8 conf1; -/* Ensure timestamp is naturally aligned */ -struct { -u16 chans[3 ]; -aligned_s64 timestamp; - } scan; };#define ISL29125_CHANNEL (_color, _si) { \@@ -179,6 +174,11 @@ static irqreturn_t isl29125_trigger_handler(int irq, void *p)struct iio_dev *indio_dev = pf->indio_dev;struct isl29125_data *data =iio_priv (indio_dev);int i, j =0 ; +/* Ensure timestamp is naturally aligned */ +struct { +u16 chans[3 ]; +aligned_s64 timestamp; + } scan = { };iio_for_each_active_channel (indio_dev, i) { int ret = i2c_smbus_read_word_data(data->client,@@ -186,10 +186,10 @@ static irqreturn_t isl29125_trigger_handler(int irq, void *p)if (ret <0 ) goto done;
drivers/iio/light/isl29125.c static irqreturn_t isl29125_trigger_handler(int irq, void *p) · line 195
Quoted 10 lines
drivers/iio/light/isl29125.c +2
-2- data->scan.chans[j++] = ret; + scan.chans[j++] = ret; } -iio_push_to_buffers_with_timestamp (indio_dev, &data->scan, +iio_push_to_buffers_with_ts (indio_dev, &scan,sizeof (scan),iio_get_time_ns (indio_dev)); done:
Re: [PATCH 09/16] iio: light: isl29125: Use iio_push_to_buffers_with_ts() to allow source size runtime check
Quoted 7 lines · On Mon, 4 Aug 2025 10:06:30 +0300 Matti Vaittinen <[email protected]>…
Quoted 6 lines · Jonathan Cameron
From: Jonathan Cameron <[email protected]>
Also move the structure used as the source to the stack as it is only 16 bytes and not the target of an DMA or similar.
Signed-off-by Jonathan Cameron <[email protected]>
Applied.
Quoted 47 lines
Quoted 35 lines
--- drivers/iio/light/isl29125.c | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-)drivers/iio/light/isl29125.c +5
-5@@ -51,11 +51,6 @@struct isl29125_data {struct i2c_client *client;u8 conf1; -/* Ensure timestamp is naturally aligned */ -struct { -u16 chans[3 ]; -aligned_s64 timestamp; - } scan; };#define ISL29125_CHANNEL (_color, _si) { \@@ -179,6 +174,11 @@ static irqreturn_t isl29125_trigger_handler(int irq, void *p)struct iio_dev *indio_dev = pf->indio_dev;struct isl29125_data *data =iio_priv (indio_dev);int i, j =0 ; +/* Ensure timestamp is naturally aligned */ +struct { +u16 chans[3 ]; +aligned_s64 timestamp; + } scan = { };iio_for_each_active_channel (indio_dev, i) { int ret = i2c_smbus_read_word_data(data->client,@@ -186,10 +186,10 @@ static irqreturn_t isl29125_trigger_handler(int irq, void *p)if (ret <0 ) goto done;drivers/iio/light/isl29125.c static irqreturn_t isl29125_trigger_handler(int irq, void *p) · line 195
Quoted 10 lines
drivers/iio/light/isl29125.c +2
-2- data->scan.chans[j++] = ret; + scan.chans[j++] = ret; } -iio_push_to_buffers_with_timestamp (indio_dev, &data->scan, +iio_push_to_buffers_with_ts (indio_dev, &scan,sizeof (scan),iio_get_time_ns (indio_dev)); done:Reviewed-by Matti Vaittinen <[email protected]
Jonathan Cameron
maintainerPATCH 10/16 iio: light: max44000: Use iio_push_to_buffers_with_ts() to allow source size runtime check
In reply to Jonathan CameronFrom: Jonathan Cameron <[email protected]>
Also move the structure used as the source to the stack as it is only 16 bytes and not the target of an DMA or similar.
--- drivers/iio/light/max44000.c | 18 +++++++++--------- 1 file changed, 9 insertions(+), 9 deletions(-)
drivers/iio/light/max44000.c +9 -9
@@ -75,11 +75,6 @@struct max44000_data {struct mutex lock;struct regmap *regmap; -/* Ensure naturally aligned timestamp */ -struct { -u16 channels[2 ]; -aligned_s64 ts; - } scan; };/* Default scale is set to the minimum of 0.03125 or 1 / (1 << 5) lux */
@@ -496,24 +491,29 @@ static irqreturn_t max44000_trigger_handler(int irq, void *p)int index =0 ;unsigned int regval;int ret; +struct { +u16 channels[2 ]; +aligned_s64 ts; + } scan = { }; +mutex_lock (&data->lock);if (test_bit (MAX44000_SCAN_INDEX_ALS , indio_dev->active_scan_mask)) { ret =max44000_read_alsval (data);if (ret <0 ) goto out_unlock; - data->scan.channels[index++] = ret; + scan.channels[index++] = ret; }if (test_bit (MAX44000_SCAN_INDEX_PRX , indio_dev->active_scan_mask)) { ret =regmap_read (data->regmap,MAX44000_REG_PRX_DATA , ®val);if (ret <0 ) goto out_unlock; - data->scan.channels[index] = regval; + scan.channels[index] = regval; }mutex_unlock (&data->lock); -iio_push_to_buffers_with_timestamp (indio_dev, &data->scan, -iio_get_time_ns (indio_dev)); +iio_push_to_buffers_with_ts (indio_dev, &scan,sizeof (scan), +iio_get_time_ns (indio_dev));iio_trigger_notify_done (indio_dev->trig);return IRQ_HANDLED ;
Signature · 1 line
2.50.1
Re: [PATCH 10/16] iio: light: max44000: Use iio_push_to_buffers_with_ts() to allow source size runtime check
On Sat, Aug 2, 2025 at 6:46 PM Jonathan Cameron <[email protected]> wrote:
The below (here and in the other patches) reads incomplete. Yes, I understand that the first sentence is in the Subject, but to me repeating it here makes things clearer.
Quoted 2 lines
Also move the structure used as the source to the stack as it is only 16 bytes and not the target of an DMA or similar.
Signature · 2 lines
With Best Regards, Andy Shevchenko
Jonathan Cameron
maintainerRe: [PATCH 10/16] iio: light: max44000: Use iio_push_to_buffers_with_ts() to allow source size runtime check
Quoted 6 lines · On Sun, 3 Aug 2025 21:26:57 +0200 Andy Shevchenko <[email protected]>…
On Sat, Aug 2, 2025 at 6:46 PM Jonathan Cameron <[email protected]> wrote:
The below (here and in the other patches) reads incomplete. Yes, I understand that the first sentence is in the Subject, but to me repeating it here makes things clearer.
Tweaked whilst applying. Thanks,
Quoted 4 lines
Quoted 3 lines
Also move the structure used as the source to the stack as it is only 16 bytes and not the target of an DMA or similar.
Re: [PATCH 10/16] iio: light: max44000: Use iio_push_to_buffers_with_ts() to allow source size runtime check
In reply to Jonathan CameronQuoted 50 lines · Jonathan Cameron
From: Jonathan Cameron <[email protected]>
Also move the structure used as the source to the stack as it is only 16 bytes and not the target of an DMA or similar.
Signed-off-by Jonathan Cameron <[email protected]>--- drivers/iio/light/max44000.c | 18 +++++++++--------- 1 file changed, 9 insertions(+), 9 deletions(-)drivers/iio/light/max44000.c +7
-7@@ -75,11 +75,6 @@struct max44000_data {struct mutex lock;struct regmap *regmap; -/* Ensure naturally aligned timestamp */ -struct { -u16 channels[2 ]; -aligned_s64 ts; - } scan; };/* Default scale is set to the minimum of 0.03125 or 1 / (1 << 5) lux */ @@ -496,24 +491,29 @@ static irqreturn_t max44000_trigger_handler(int irq, void *p)int index =0 ;unsigned int regval;int ret; +struct { +u16 channels[2 ]; +aligned_s64 ts; + } scan = { }; +mutex_lock (&data->lock);if (test_bit (MAX44000_SCAN_INDEX_ALS , indio_dev->active_scan_mask)) { ret =max44000_read_alsval (data);if (ret <0 ) goto out_unlock; - data->scan.channels[index++] = ret; + scan.channels[index++] = ret; }if (test_bit (MAX44000_SCAN_INDEX_PRX , indio_dev->active_scan_mask)) { ret =regmap_read (data->regmap,MAX44000_REG_PRX_DATA , ®val);if (ret <0 ) goto out_unlock; - data->scan.channels[index] = regval; + scan.channels[index] = regval;
drivers/iio/light/max44000.c static irqreturn_t max44000_trigger_handler(int irq, void *p) · line 519
Quoted 10 lines
drivers/iio/light/max44000.c +2
-2}mutex_unlock (&data->lock); -iio_push_to_buffers_with_timestamp (indio_dev, &data->scan, -iio_get_time_ns (indio_dev)); +iio_push_to_buffers_with_ts (indio_dev, &scan,sizeof (scan), +iio_get_time_ns (indio_dev));iio_trigger_notify_done (indio_dev->trig);return IRQ_HANDLED ;
Jonathan Cameron
maintainerPATCH 11/16 iio: light: st_uvis25: Use iio_push_to_buffers_with_ts() to allow source size runtime check
In reply to Jonathan CameronFrom: Jonathan Cameron <[email protected]>
Also move the structure used as the source to the stack as it is only 16 bytes and not the target of an DMA or similar.
--- drivers/iio/light/st_uvis25.h | 5 ----- drivers/iio/light/st_uvis25_core.c | 12 +++++++++--- 2 files changed, 9 insertions(+), 8 deletions(-)
drivers/iio/light/st_uvis25.h +0 -5
@@ -27,11 +27,6 @@ struct st_uvis25_hw {struct iio_trigger *trig;bool enabled;int irq; -/* Ensure timestamp is naturally aligned */ -struct { -u8 chan; -aligned_s64 ts; - } scan; };extern const struct dev_pm_ops st_uvis25_pm_ops;
drivers/iio/light/st_uvis25_core.c +9 -3
@@ -234,15 +234,21 @@ static irqreturn_t st_uvis25_buffer_handler_thread(int irq, void *p)struct st_uvis25_hw *hw =iio_priv (iio_dev);unsigned int val;int err; +/* Ensure timestamp is naturally aligned */ +struct { +u8 chan; +aligned_s64 ts; + } scan = { }; + err =regmap_read (hw->regmap,ST_UVIS25_REG_OUT_ADDR , &val);if (err <0 ) goto out; - hw->scan.chan = val; + scan.chan = val; -iio_push_to_buffers_with_timestamp (iio_dev, &hw->scan, -iio_get_time_ns (iio_dev)); +iio_push_to_buffers_with_ts (iio_dev, &scan,sizeof (scan), +iio_get_time_ns (iio_dev)); out:iio_trigger_notify_done (hw->trig);
Signature · 1 line
2.50.1
Re: [PATCH 11/16] iio: light: st_uvis25: Use iio_push_to_buffers_with_ts() to allow source size runtime check
Quoted 6 lines · Jonathan Cameron
From: Jonathan Cameron <[email protected]>
Also move the structure used as the source to the stack as it is only 16 bytes and not the target of an DMA or similar.
Signed-off-by Jonathan Cameron <[email protected]>
Quoted 4 lines
--- drivers/iio/light/st_uvis25.h | 5 ----- drivers/iio/light/st_uvis25_core.c | 12 +++++++++--- 2 files changed, 9 insertions(+), 8 deletions(-)
Jonathan Cameron
maintainerRe: [PATCH 11/16] iio: light: st_uvis25: Use iio_push_to_buffers_with_ts() to allow source size runtime check
Quoted 9 lines · On Mon, 4 Aug 2025 10:09:56 +0300 Matti Vaittinen <[email protected]>…
Quoted 6 lines · Jonathan Cameron
From: Jonathan Cameron <[email protected]>
Also move the structure used as the source to the stack as it is only 16 bytes and not the target of an DMA or similar.
Signed-off-by Jonathan Cameron <[email protected]>Reviewed-by Matti Vaittinen <[email protected]>
Tweaked description inline with Andy's comment on earlier patch and applied.
Quoted 7 lines
Quoted 5 lines
--- drivers/iio/light/st_uvis25.h | 5 ----- drivers/iio/light/st_uvis25_core.c | 12 +++++++++--- 2 files changed, 9 insertions(+), 8 deletions(-)
Jonathan Cameron
maintainerPATCH 12/16 iio: light: tcs3414: Use iio_push_to_buffers_with_ts() to allow source size runtime check
In reply to Jonathan CameronFrom: Jonathan Cameron <[email protected]>
Also move the structure used as the source to the stack as it is only 16 bytes and not the target of an DMA or similar.
--- drivers/iio/light/tcs3414.c | 15 ++++++++------- 1 file changed, 8 insertions(+), 7 deletions(-)
drivers/iio/light/tcs3414.c +8 -7
@@ -53,11 +53,6 @@ struct tcs3414_data {u8 control;u8 gain;u8 timing; -/* Ensure timestamp is naturally aligned */ -struct { -u16 chans[4 ]; -aligned_s64 timestamp; - } scan; };#define TCS3414_CHANNEL (_color, _si, _addr) { \
@@ -204,6 +199,12 @@ static irqreturn_t tcs3414_trigger_handler(int irq, void *p)struct iio_dev *indio_dev = pf->indio_dev;struct tcs3414_data *data =iio_priv (indio_dev);int i, j =0 ; +/* Ensure timestamp is naturally aligned */ +struct { +u16 chans[4 ]; +aligned_s64 timestamp; + } scan = { }; +iio_for_each_active_channel (indio_dev, i) { int ret = i2c_smbus_read_word_data(data->client,
@@ -211,10 +212,10 @@ static irqreturn_t tcs3414_trigger_handler(int irq, void *p)if (ret <0 ) goto done; - data->scan.chans[j++] = ret; + scan.chans[j++] = ret; } -iio_push_to_buffers_with_timestamp (indio_dev, &data->scan, +iio_push_to_buffers_with_ts (indio_dev, &scan,sizeof (scan),iio_get_time_ns (indio_dev)); done:
Signature · 1 line
2.50.1
Jonathan Cameron
maintainerRe: [PATCH 12/16] iio: light: tcs3414: Use iio_push_to_buffers_with_ts() to allow source size runtime check
Quoted 6 lines · On Sat, 2 Aug 2025 17:44:32 +0100 Jonathan Cameron <[email protected]> wrote:
From: Jonathan Cameron <[email protected]>
Also move the structure used as the source to the stack as it is only 16 bytes and not the target of an DMA or similar.
Signed-off-by Jonathan Cameron <[email protected]>
Applied with tweaked description as per feedback on earlier patch.
Thanks,
Jonathan
Jonathan Cameron
maintainerPATCH 13/16 iio: light: tcs3472: Use iio_push_to_buffers_with_ts() to allow source size runtime check
In reply to Jonathan CameronFrom: Jonathan Cameron <[email protected]>
Also move the structure used as the source to the stack as it is only 16 bytes and not the target of an DMA or similar.
--- drivers/iio/light/tcs3472.c | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-)
drivers/iio/light/tcs3472.c +7 -7
@@ -64,11 +64,6 @@ struct tcs3472_data {u8 control;u8 atime;u8 apers; -/* Ensure timestamp is naturally aligned */ -struct { -u16 chans[4 ]; -aligned_s64 timestamp; - } scan; };static const struct iio_event_spec tcs3472_events[] = {
@@ -377,6 +372,11 @@ static irqreturn_t tcs3472_trigger_handler(int irq, void *p)struct iio_dev *indio_dev = pf->indio_dev;struct tcs3472_data *data =iio_priv (indio_dev);int i, j =0 ; +/* Ensure timestamp is naturally aligned */ +struct { +u16 chans[4 ]; +aligned_s64 timestamp; + } scan = { };int ret =tcs3472_req_data (data);if (ret <0 )
@@ -388,10 +388,10 @@ static irqreturn_t tcs3472_trigger_handler(int irq, void *p)if (ret <0 ) goto done; - data->scan.chans[j++] = ret; + scan.chans[j++] = ret; } -iio_push_to_buffers_with_timestamp (indio_dev, &data->scan, +iio_push_to_buffers_with_ts (indio_dev, &scan,sizeof (scan),iio_get_time_ns (indio_dev)); done:
Signature · 1 line
2.50.1
Re: [PATCH 13/16] iio: light: tcs3472: Use iio_push_to_buffers_with_ts() to allow source size runtime check
Quoted 4 lines · On Sat, 2 Aug 2025 17:44:33 +0100 Jonathan Cameron <[email protected]> wrote:
From: Jonathan Cameron <[email protected]>
Also move the structure used as the source to the stack as it is only 16 bytes and not the target of an DMA or similar.
Applied. Description tweaked similar to earlier patches.
Jonathan Cameron
maintainerPATCH 14/16 iio: light: vcnl4000: Use a structure to make buffer arrangement explicit.
In reply to Jonathan CameronFrom: Jonathan Cameron <[email protected]>
Removes the need for comments to describe the buffer passed to iio_push_to_buffers_with_timestamp().
--- drivers/iio/light/vcnl4000.c | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-)
drivers/iio/light/vcnl4000.c +6 -3
@@ -1662,7 +1662,10 @@ static irqreturn_t vcnl4010_trigger_handler(int irq, void *p)struct iio_dev *indio_dev = pf->indio_dev;struct vcnl4000_data *data =iio_priv (indio_dev);const unsigned long *active_scan_mask = indio_dev->active_scan_mask; -u16 buffer[8 ]__aligned(8) = {0 };/* 1x16-bit + naturally aligned ts */ +struct { +u16 chan; +aligned_s64 ts; + } scan = { };bool data_read = false;unsigned long isr;int val =0 ;
@@ -1682,7 +1685,7 @@ static irqreturn_t vcnl4010_trigger_handler(int irq, void *p)if (ret <0 ) goto end; - buffer[0 ] = val; + scan.chan = val; data_read = true; } }
@@ -1695,7 +1698,7 @@ static irqreturn_t vcnl4010_trigger_handler(int irq, void *p)if (!data_read) goto end; -iio_push_to_buffers_with_timestamp (indio_dev, buffer, +iio_push_to_buffers_with_timestamp (indio_dev, &scan,iio_get_time_ns (indio_dev)); end:
Signature · 1 line
2.50.1
Re: [PATCH 14/16] iio: light: vcnl4000: Use a structure to make buffer arrangement explicit.
Kind regards
Mårten
Quoted 46 lines · Jonathan Cameron
From: Jonathan Cameron <[email protected]>
Removes the need for comments to describe the buffer passed to
iio_push_to_buffers_with_timestamp().Signed-off-by Jonathan Cameron <[email protected]>Cc Astrid Rost <[email protected]>Cc Mårten Lindahl <[email protected]>--- drivers/iio/light/vcnl4000.c | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-)drivers/iio/light/vcnl4000.c +6
-3@@ -1662,7 +1662,10 @@ static irqreturn_t vcnl4010_trigger_handler(int irq, void *p)struct iio_dev *indio_dev = pf->indio_dev;struct vcnl4000_data *data =iio_priv (indio_dev);const unsigned long *active_scan_mask = indio_dev->active_scan_mask; -u16 buffer[8 ]__aligned(8) = {0 };/* 1x16-bit + naturally aligned ts */ +struct { +u16 chan; +aligned_s64 ts; + } scan = { };bool data_read = false;unsigned long isr;int val =0 ;@@ -1682,7 +1685,7 @@ static irqreturn_t vcnl4010_trigger_handler(int irq, void *p)if (ret <0 ) goto end; - buffer[0 ] = val; + scan.chan = val; data_read = true; } }@@ -1695,7 +1698,7 @@ static irqreturn_t vcnl4010_trigger_handler(int irq, void *p)if (!data_read) goto end; -iio_push_to_buffers_with_timestamp (indio_dev, buffer, +iio_push_to_buffers_with_timestamp (indio_dev, &scan,iio_get_time_ns (indio_dev)); end:
Re: [PATCH 14/16] iio: light: vcnl4000: Use a structure to make buffer arrangement explicit.
Quoted 1 line · On Mon, 4 Aug 2025 09:00:17 +0200 Mårten Lindahl <[email protected]> wrote:
Reviewed-by Mårten Lindahl <[email protected]>
Applied. thanks
Quoted 53 lines
Kind regards
Mårten
Quoted 46 lines · Jonathan Cameron
From: Jonathan Cameron <[email protected]>
Removes the need for comments to describe the buffer passed to
iio_push_to_buffers_with_timestamp().Signed-off-by Jonathan Cameron <[email protected]>Cc Astrid Rost <[email protected]>Cc Mårten Lindahl <[email protected]>--- drivers/iio/light/vcnl4000.c | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-)drivers/iio/light/vcnl4000.c +6
-3@@ -1662,7 +1662,10 @@ static irqreturn_t vcnl4010_trigger_handler(int irq, void *p)struct iio_dev *indio_dev = pf->indio_dev;struct vcnl4000_data *data =iio_priv (indio_dev);const unsigned long *active_scan_mask = indio_dev->active_scan_mask; -u16 buffer[8 ]__aligned(8) = {0 };/* 1x16-bit + naturally aligned ts */ +struct { +u16 chan; +aligned_s64 ts; + } scan = { };bool data_read = false;unsigned long isr;int val =0 ;@@ -1682,7 +1685,7 @@ static irqreturn_t vcnl4010_trigger_handler(int irq, void *p)if (ret <0 ) goto end; - buffer[0 ] = val; + scan.chan = val; data_read = true; } }@@ -1695,7 +1698,7 @@ static irqreturn_t vcnl4010_trigger_handler(int irq, void *p)if (!data_read) goto end; -iio_push_to_buffers_with_timestamp (indio_dev, buffer, +iio_push_to_buffers_with_timestamp (indio_dev, &scan,iio_get_time_ns (indio_dev)); end:
Jonathan Cameron
maintainerPATCH 15/16 iio: light: vl6180: Use iio_push_to_buffers_with_ts() to allow source size runtime check
In reply to Jonathan CameronFrom: Jonathan Cameron <[email protected]>
Also move the structure used as the source to the stack as it is only 16 bytes and not the target of an DMA or similar.
--- drivers/iio/light/vl6180.c | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-)
drivers/iio/light/vl6180.c +7 -7
@@ -96,11 +96,6 @@ struct vl6180_data {unsigned int als_it_ms;unsigned int als_meas_rate;unsigned int range_meas_rate; - -struct { -u16 chan[2 ]; -aligned_s64 timestamp; - } scan; };enum {VL6180_ALS ,VL6180_RANGE ,VL6180_PROX };
@@ -545,6 +540,11 @@ static irqreturn_t vl6180_trigger_handler(int irq, void *priv)struct vl6180_data *data =iio_priv (indio_dev);s64 time_ns =iio_get_time_ns (indio_dev);int ret, bit, i =0 ; +struct { +u16 chan[2 ]; +aligned_s64 timestamp; + } scan = { }; +iio_for_each_active_channel (indio_dev, bit) {if (vl6180_chan_regs_table[bit].word)
@@ -560,10 +560,10 @@ static irqreturn_t vl6180_trigger_handler(int irq, void *priv)return IRQ_HANDLED ; } - data->scan.chan[i++] = ret; + scan.chan[i++] = ret; } -iio_push_to_buffers_with_timestamp (indio_dev, &data->scan, time_ns); +iio_push_to_buffers_with_ts (indio_dev, &scan,sizeof (scan), time_ns);iio_trigger_notify_done (indio_dev->trig);/* Clear the interrupt flag after data read */
Signature · 1 line
2.50.1
Re: [PATCH 15/16] iio: light: vl6180: Use iio_push_to_buffers_with_ts() to allow source size runtime check
Quoted 7 lines · On Sat, 2 Aug 2025 17:44:35 +0100 Jonathan Cameron <[email protected]> wrote:
From: Jonathan Cameron <[email protected]>
Also move the structure used as the source to the stack as it is only 16 bytes and not the target of an DMA or similar.
Signed-off-by Jonathan Cameron <[email protected]>Cc Abhash Jha <[email protected]>
Applied. Description tweaked as per earlier feedback.
Jonathan Cameron
maintainerPATCH 16/16 iio: light: Simple conversions to iio_push_to_buffers_with_ts()
In reply to Jonathan CameronFrom: Jonathan Cameron <[email protected]>
This new function allows for runtime checks on the provided source buffer being of sufficient size to accommodate the enabled channel data layout and the naturally aligned s64 timestamp (which is non obvious and a frequent source of bugs in the past).
This patch includes the remaining simple cases for light sensor drivers.
--- drivers/iio/light/bh1745.c | 4 ++-- drivers/iio/light/hid-sensor-als.c | 5 +++-- drivers/iio/light/ltr501.c | 4 ++-- drivers/iio/light/opt4060.c | 2 +- drivers/iio/light/rohm-bu27034.c | 3 ++- drivers/iio/light/rpr0521.c | 4 ++-- drivers/iio/light/si1145.c | 5 +++-- drivers/iio/light/vcnl4000.c | 4 ++-- drivers/iio/light/veml6030.c | 2 +- 9 files changed, 18 insertions(+), 15 deletions(-)
drivers/iio/light/bh1745.c +2 -2
@@ -755,8 +755,8 @@ static irqreturn_t bh1745_trigger_handler(int interrupt, void *p) scan.chans[j++] = value; } -iio_push_to_buffers_with_timestamp (indio_dev, &scan, -iio_get_time_ns (indio_dev)); +iio_push_to_buffers_with_ts (indio_dev, &scan,sizeof (scan), +iio_get_time_ns (indio_dev)); err:iio_trigger_notify_done (indio_dev->trig);
drivers/iio/light/hid-sensor-als.c +3 -2
@@ -262,8 +262,9 @@ static int als_proc_event(struct hid_sensor_hub_device *hsdev,if (!als_state->timestamp) als_state->timestamp =iio_get_time_ns (indio_dev); -iio_push_to_buffers_with_timestamp (indio_dev, &als_state->scan, - als_state->timestamp); +iio_push_to_buffers_with_ts (indio_dev, &als_state->scan, +sizeof (als_state->scan), + als_state->timestamp); als_state->timestamp =0 ; }
drivers/iio/light/ltr501.c +2 -2
@@ -1315,8 +1315,8 @@ static irqreturn_t ltr501_trigger_handler(int irq, void *p) scan.channels[j++] = psdata <R501_PS_DATA_MASK ; } -iio_push_to_buffers_with_timestamp (indio_dev, &scan, -iio_get_time_ns (indio_dev)); +iio_push_to_buffers_with_ts (indio_dev, &scan,sizeof (scan), +iio_get_time_ns (indio_dev)); done:iio_trigger_notify_done (indio_dev->trig);
drivers/iio/light/opt4060.c +1 -1
@@ -1104,7 +1104,7 @@ static irqreturn_t opt4060_trigger_handler(int irq, void *p) } } -iio_push_to_buffers_with_timestamp (idev, &raw, pf->timestamp); +iio_push_to_buffers_with_ts (idev, &raw,sizeof (raw), pf->timestamp); err_read:iio_trigger_notify_done (idev->trig);return IRQ_HANDLED ;
drivers/iio/light/rohm-bu27034.c +2 -1
@@ -1193,7 +1193,8 @@ static int bu27034_buffer_thread(void *arg)*/ data->scan.mlux = (u32 )mlux; } -iio_push_to_buffers_with_timestamp (idev, &data->scan, tstamp); +iio_push_to_buffers_with_ts (idev, &data->scan, +sizeof (data->scan), tstamp); }return 0 ;
drivers/iio/light/rpr0521.c +2 -2
@@ -457,8 +457,8 @@ static irqreturn_t rpr0521_trigger_consumer_handler(int irq, void *p) data->scan.channels, (3 * 2) + 1); /* 3 * 16-bit + (discarded) int clear reg. */if (!err) -iio_push_to_buffers_with_timestamp (indio_dev, - &data->scan, pf->timestamp); +iio_push_to_buffers_with_ts (indio_dev, &data->scan, +sizeof (data->scan), pf->timestamp);else dev_err (&data->client->dev,"Trigger consumer can't read from sensor.\n" );
drivers/iio/light/si1145.c +3 -2
@@ -494,8 +494,9 @@ static irqreturn_t si1145_trigger_handler(int irq, void *private) goto done; } -iio_push_to_buffers_with_timestamp (indio_dev, data->buffer, -iio_get_time_ns (indio_dev)); +iio_push_to_buffers_with_ts (indio_dev, data->buffer, +sizeof (data->buffer), +iio_get_time_ns (indio_dev)); done:iio_trigger_notify_done (indio_dev->trig);
drivers/iio/light/vcnl4000.c +2 -2
@@ -1698,8 +1698,8 @@ static irqreturn_t vcnl4010_trigger_handler(int irq, void *p)if (!data_read) goto end; -iio_push_to_buffers_with_timestamp (indio_dev, &scan, -iio_get_time_ns (indio_dev)); +iio_push_to_buffers_with_ts (indio_dev, &scan,sizeof (scan), +iio_get_time_ns (indio_dev)); end:iio_trigger_notify_done (indio_dev->trig);
drivers/iio/light/veml6030.c +1 -1
@@ -903,7 +903,7 @@ static irqreturn_t veml6030_trigger_handler(int irq, void *p) scan.chans[i++] = reg; } -iio_push_to_buffers_with_timestamp (iio, &scan, pf->timestamp); +iio_push_to_buffers_with_ts (iio, &scan,sizeof (scan), pf->timestamp); done:iio_trigger_notify_done (iio->trig);
Signature · 1 line
2.50.1
Re: [PATCH 16/16] iio: light: Simple conversions to iio_push_to_buffers_with_ts()
Quoted 10 lines · Jonathan Cameron
From: Jonathan Cameron <[email protected]>
This new function allows for runtime checks on the provided source buffer being of sufficient size to accommodate the enabled channel data layout and the naturally aligned s64 timestamp (which is non obvious and a frequent source of bugs in the past).
This patch includes the remaining simple cases for light sensor drivers.
Signed-off-by Jonathan Cameron <[email protected]>
Quoted 5 lines
Cc Mudit Sharma <[email protected]>Cc Jiri Kosina <[email protected]>Cc Srinivas Pandruvada <[email protected]>Cc Javier Carrasco <[email protected]>Cc Matti Vaittinen <[email protected]>
Quoted 11 lines
--- drivers/iio/light/bh1745.c | 4 ++-- drivers/iio/light/hid-sensor-als.c | 5 +++-- drivers/iio/light/ltr501.c | 4 ++-- drivers/iio/light/opt4060.c | 2 +- drivers/iio/light/rohm-bu27034.c | 3 ++- drivers/iio/light/rpr0521.c | 4 ++-- drivers/iio/light/si1145.c | 5 +++-- drivers/iio/light/vcnl4000.c | 4 ++-- drivers/iio/light/veml6030.c | 2 +- 9 files changed, 18 insertions(+), 15 deletions(-)
Yours,
-- Matti
Re: [PATCH 16/16] iio: light: Simple conversions to iio_push_to_buffers_with_ts()
In reply to Jonathan CameronQuoted 13 lines · Jonathan Cameron
From: Jonathan Cameron <[email protected]>
This new function allows for runtime checks on the provided source buffer
being of sufficient size to accommodate the enabled channel data layout and
the naturally aligned s64 timestamp (which is non obvious and a frequent
source of bugs in the past).This patch includes the remaining simple cases for light sensor drivers.
Quoted 6 lines
Signed-off-by Jonathan Cameron <[email protected]>Cc Mudit Sharma <[email protected]>Cc Jiri Kosina <[email protected]>Cc Srinivas Pandruvada <[email protected]>Cc Javier Carrasco <[email protected]>Cc Matti Vaittinen <[email protected]>
Quoted 175 lines
---
drivers/iio/light/bh1745.c | 4 ++--
drivers/iio/light/hid-sensor-als.c | 5 +++--
drivers/iio/light/ltr501.c | 4 ++--
drivers/iio/light/opt4060.c | 2 +-
drivers/iio/light/rohm-bu27034.c | 3 ++-
drivers/iio/light/rpr0521.c | 4 ++--
drivers/iio/light/si1145.c | 5 +++--
drivers/iio/light/vcnl4000.c | 4 ++--
drivers/iio/light/veml6030.c | 2 +-9 files changed, 18 insertions(+), 15 deletions(-)drivers/iio/light/bh1745.c +0
-0@@ -755,8 +755,8 @@ static irqreturn_t bh1745_trigger_handler(int interrupt, void *p) scan.chans[j++] = value; }drivers/iio/light/bh1745.c +0
-2- iio_push_to_buffers_with_timestamp(indio_dev, &scan, -
iio_get_time_ns(indio_dev));drivers/iio/light/bh1745.c +2
-0+iio_push_to_buffers_with_ts (indio_dev, &scan,sizeof (scan), + iio_get_time_ns(indio_dev));
err:
iio_trigger_notify_done(indio_dev->trig);drivers/iio/light/hid-sensor-als.c +0
-0@@ -262,8 +262,9 @@ static int als_proc_event(struct hid_sensor_hub_device *hsdev, if (!als_state->timestamp)
als_state->timestamp=iio_get_time_ns(indio_dev);drivers/iio/light/hid-sensor-als.c +0
-1- iio_push_to_buffers_with_timestamp(indio_dev,
&als_state->scan,drivers/iio/light/hid-sensor-als.c +0
-1- als_state-Quoted 1 line
timestamp);
drivers/iio/light/hid-sensor-als.c +1
-0+ iio_push_to_buffers_with_ts(indio_dev, &als_state-Quoted 1 line
scan,
drivers/iio/light/hid-sensor-als.c +2
-0+ sizeof(als_state->scan), + als_state->timestamp);
als_state->timestamp= 0;
}drivers/iio/light/ltr501.c +0
-0@@ -1315,8 +1315,8 @@ static irqreturn_t ltr501_trigger_handler(int irq, void *p) scan.channels[j++] = psdata & LTR501_PS_DATA_MASK; }drivers/iio/light/ltr501.c +0
-2- iio_push_to_buffers_with_timestamp(indio_dev, &scan, -
iio_get_time_ns(indio_dev));drivers/iio/light/ltr501.c +2
-0+iio_push_to_buffers_with_ts (indio_dev, &scan,sizeof (scan), + iio_get_time_ns(indio_dev));
done:
iio_trigger_notify_done(indio_dev->trig);drivers/iio/light/opt4060.c +0
-0@@ -1104,7 +1104,7 @@ static irqreturn_t opt4060_trigger_handler(int irq, void *p) } }drivers/iio/light/opt4060.c +0
-1- iio_push_to_buffers_with_timestamp(idev, &raw, pf-Quoted 1 line
timestamp);
drivers/iio/light/opt4060.c +1
-0+ iio_push_to_buffers_with_ts(idev, &raw, sizeof(raw), pf-Quoted 1 line
timestamp);
err_read:
iio_trigger_notify_done(idev->trig);
returnIRQ_HANDLED;drivers/iio/light/rohm-bu27034.c +2
-1@@ -1193,7 +1193,8 @@ static int bu27034_buffer_thread(void *arg)*/ data->scan.mlux = (u32)mlux; } -iio_push_to_buffers_with_timestamp (idev, &data- >scan, tstamp); +iio_push_to_buffers_with_ts (idev, &data->scan, + sizeof(data->scan), tstamp); }
return 0;drivers/iio/light/rpr0521.c +0
-0@@ -457,8 +457,8 @@ static irqreturn_t rpr0521_trigger_consumer_handler(int irq, void *p) data->scan.channels,(3 * 2) + 1); /* 3 * 16-bit + (discarded) int
clear reg. */
if (!err)drivers/iio/light/rpr0521.c +0
-2- iio_push_to_buffers_with_timestamp(indio_dev, - &data->scan, pf-Quoted 1 line
timestamp);
drivers/iio/light/rpr0521.c +2
-0+ iio_push_to_buffers_with_ts(indio_dev, &data->scan, + sizeof(data->scan), pf-Quoted 1 line
timestamp);
else
dev_err(&data->client->dev, "Trigger consumer can't read from sensor.\n");drivers/iio/light/si1145.c +0
-0@@ -494,8 +494,9 @@ static irqreturn_t si1145_trigger_handler(int irq, void *private) goto done; }drivers/iio/light/si1145.c +3
-2-iio_push_to_buffers_with_timestamp (indio_dev, data->buffer, -iio_get_time_ns (indio_dev)); +iio_push_to_buffers_with_ts (indio_dev, data->buffer, + sizeof(data->buffer), + iio_get_time_ns(indio_dev));
done:
iio_trigger_notify_done(indio_dev->trig);drivers/iio/light/vcnl4000.c +0
-0@@ -1698,8 +1698,8 @@ static irqreturn_t vcnl4010_trigger_handler(int irq, void *p) if (!data_read) goto end;drivers/iio/light/vcnl4000.c +0
-2- iio_push_to_buffers_with_timestamp(indio_dev, &scan, -
iio_get_time_ns(indio_dev));drivers/iio/light/vcnl4000.c +2
-0+iio_push_to_buffers_with_ts (indio_dev, &scan,sizeof (scan), + iio_get_time_ns(indio_dev));
end:
iio_trigger_notify_done(indio_dev->trig);drivers/iio/light/veml6030.c +0
-0@@ -903,7 +903,7 @@ static irqreturn_t veml6030_trigger_handler(int irq, void *p) scan.chans[i++] = reg; }drivers/iio/light/veml6030.c +0
-1- iio_push_to_buffers_with_timestamp(iio, &scan, pf-Quoted 1 line
timestamp);
drivers/iio/light/veml6030.c +1
-0+ iio_push_to_buffers_with_ts(iio, &scan, sizeof(scan), pf-Quoted 1 line
timestamp);
done:
iio_trigger_notify_done(iio->trig);
Re: [PATCH 16/16] iio: light: Simple conversions to iio_push_to_buffers_with_ts()
Quoted 14 lines · On Mon, 04 Aug 2025 06:29:10 -0700 srinivas pandruvada…
Quoted 13 lines · Jonathan Cameron
From: Jonathan Cameron <[email protected]>
This new function allows for runtime checks on the provided source buffer
being of sufficient size to accommodate the enabled channel data layout and
the naturally aligned s64 timestamp (which is non obvious and a frequent
source of bugs in the past).This patch includes the remaining simple cases for light sensor drivers.
Quoted 8 lines
Quoted 6 lines
Signed-off-by Jonathan Cameron <[email protected]>Cc Mudit Sharma <[email protected]>Cc Jiri Kosina <[email protected]>Cc Srinivas Pandruvada <[email protected]>Cc Javier Carrasco <[email protected]>Cc Matti Vaittinen <[email protected]>
Applied.
Thanks,
Jonathan
Quoted 7 lines · Jonathan Cameron
From: Jonathan Cameron <[email protected]>
Main aim here is to switch almost all of the light drivers over to
iio_push_to_buffers_with_ts()to provide runtime checking that we aren't copying from beyond the provided buffers.
Quoted 6 lines
A couple of bugs showed up whilst doing these conversions that should be back ported (patches 1 and 2).
Various other cases benefit from conversion from arrays to structures and putting the temporary storage on the stack. Given these cases are a little more complex, they are done as one or more patch per driver.
patches
Quoted 6 lines
The cases that were just a simple change of function call from
iio_push_to_buffers_with_timestamp()are grouped together in the final patch as not much to say about them and if we do a patch per driver this and related sets will be unnecessarily long!If people would prefer I split that up that would be fine.
with a few nit-picks here and there...
Signature · 2 lines
With Best Regards, Andy Shevchenko