Re: [PATCH 1/7] iio: pressure: bmp280: Apply iwyu principles to includes.

[Andy Shevchenko <andriy.shevchenko@intel.com>]

On Thu, Jul 10, 2025 at 06:11:01PM +0100, Jonathan Cameron wrote:
> From: Jonathan Cameron <Jonathan.Cameron@huawei.com>
> 
> The recent introduction of a warning on missing include of
> linux/export.h when W=1 motivated revisiting the includes in affected
> drivers.  In general IWYU principles avoid complex include paths that
> make it hard to refactor headers.
> 
> - Move linux/unaligned.h entry to appropriate place.
> - Drop comment on linux/irq.h as we don't generally keep a record
>   in code of why includes are there.
> 
> Remove
> - linux/device.h from bmp280-regmap.c as struct device forwards definition
>   is enough.
> - linux/module.h from bmp280-regmap.c as the module stuff is all in the
>   other files.
> Add all of:
> - linux/array_size.h for ARRAY_SIZE()
> - linux/device.h in the i2c and spi drivers
> - linux/err.h for PTR_ERR() etc and also assume includes errno.h
> - linux/export.h for EXPORT_SYMBOL*()
> - linux/jiffies.h for msecs_to_jiffies
> - linux/log2.h for ilog2()
> - linux/math64.h for div64_s64
> - linux/minmax.h for clamp_val()
> - linux/mod_devicetable.h for spi_device_id etc
> - linux/mutex.h for various mutex calls
> - linux/pm.h for pm_ptr()
> - linux/stddef.h for false / true
> - linux/string.h for memcpy
> - linux/string_choices.h for str_enable_disable
> - linux/time.h for USEC_PER_MSEC
> - linux/types.h for local bool definition.
> 
> Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
> ---
> I'm not sure on whether we should bother with stddef.h or types.h
> Looking for feedback on any others.

types.h is superior.

-- 
With Best Regards,
Andy Shevchenko