Re: [PULL] IIO: 3rd set for 6.9 - cleanup.h handling of fwnode_handle_put() related (not ordered wrt to PULL 2)

[Greg KH <gregkh@linuxfoundation.org>]

On Thu, Feb 29, 2024 at 08:23:00PM +0000, Jonathan Cameron wrote:
> The following changes since commit d4551c189d6e6a3fcf7f625bd4b273e770fad35a:
> 
>   Merge tag 'iio-for-6.9a' of http://git.kernel.org/pub/scm/linux/kernel/git/jic23/iio into char-misc-next (2024-02-25 14:11:41 +0100)
> 
> are available in the Git repository at:
> 
>   https://git.kernel.org/pub/scm/linux/kernel/git/jic23/iio.git tags/iio-for-6.9c
> 
> for you to fetch changes up to 64e19caa5564ecc43edaa7fb818d53de650d9b34:
> 
>   iio: adc: rzg2l_adc: Use device_for_each_child_node_scoped() (2024-02-28 19:15:43 +0000)
> 
> ----------------------------------------------------------------
> IIO: 3rd set for 6.9 - cleanup.h related.
> 
> I have separated this set out from the more normal patches as they can
> go separately and that may simplify the merge window.  Greg, up to you
> how you wish to handle this in the char-misc tree.
> 
> Introduces __free() based handling for fwnode_handle_put() to
> allow scope based release of these handles on early exit from functions.

This should be fine, right?  No one complains about these.

> Also introduced device_for_each_child_node_scoped() to provide a
> a convenient way to process child nodes without the need to explicitly
> handle the fwnode_handle_put() needed on early exits from the loop.
> Typically these early exits are a result of error handling or completion
> of a search and have proven very prone to being missed.

This is trickier, there was a load of different versions floating
around, do you have a link to the "last" version of this patch series
that got applied here?

> One instance of such a leaked resource was found during these conversions
> though review of that patch was too late for this pull request.

I don't understand, does this series fix that found problem?  Or is it
coming?

> A number of drivers are also converted over to generic fwnode handling from
> the device tree specific version.

Deleting code is always good, but:


> 
> ----------------------------------------------------------------
> Jonathan Cameron (16):
>       device property: Move fwnode_handle_put() into property.h
>       device property: Add cleanup.h based fwnode_handle_put() scope based cleanup.
>       device property: Introduce device_for_each_child_node_scoped()
>       iio: adc: max11410: Use device_for_each_child_node_scoped()
>       iio: addac: ad74413r: Use device_for_each_child_node_scoped()
>       iio: dac: ltc2688: Use device_for_each_child_node_scoped()
>       iio: adc: fsl-imx25-gcq: Switch from of specific handing to fwnode based.
>       iio: adc: fsl-imx25-gcq: Use devm_* and dev_err_probe() to simplify probe
>       iio: adc: ad7124: Switch from of specific to fwnode based property handling
>       iio: adc: ad7292: Switch from of specific to fwnode property handling
>       iio: adc: ad7192: Convert from of specific to fwnode property handling
>       iio: accel: mma8452: Switch from of specific to fwnode property handling.
>       iio: accel: fxls8962af: Switch from of specific to fwnode based properties.
>       iio: adc: hx711: Switch from of specific to fwnode property handling.
>       iio: temp: ltc2983: Use __free(fwnode_handle) and device_for_each_node_scoped()
>       iio: adc: rzg2l_adc: Use device_for_each_child_node_scoped()

You are mixing the two different handlers in this series, right?  How
about 2 different ones, one for each?  Or do they start to conflict?

thanks,

greg k-h