Re: [PATCH] i2c: Allow recovery of the initial IRQ by an I2C client device.

[Charles Keepax <ckeepax@opensource.cirrus.com>]

On Fri, Feb 22, 2019 at 11:15:59AM +0100, Benjamin Tissoires wrote:
> On Fri, Feb 22, 2019 at 12:26 AM Wolfram Sang <wsa@the-dreams.de> wrote:
> > On Tue, Feb 19, 2019 at 11:30:27AM -0800, Jim Broadus wrote:
> > > A previous change allowed I2C client devices to discover new IRQs upon
> > > reprobe by clearing the IRQ in i2c_device_remove. However, if an IRQ was
> > > assigned in i2c_new_device, that information is lost.
> > >
> > > For example, the touchscreen and trackpad devices on a Dell Inspiron laptop
> > > are I2C devices whose IRQs are defined by ACPI extended IRQ types. The
> > > client device structures are initialized during an ACPI walk. After
> > > removing the i2c_hid device, modprobe fails.
> > >
> > > This change caches the initial IRQ value in i2c_new_device and then resets
> > > the client device IRQ to the initial value in i2c_device_remove.
> > >
> > > Fixes: 6f108dd70d30 ("i2c: Clear client->irq in i2c_device_remove")
> > > Signed-off-by: Jim Broadus <jbroadus@gmail.com>
> >
> > Adding Benjamin to CC
> 
> Sorry, I should have answered earlier.
> 
> I am a little bit hesitant regarding this patch. The effect is
> correct, and I indeed realized a few weeks ago that something were
> wrong as we couldn't rmmod/modprobe i2c-hid.
> 
> But I still have the feeling that the problem is not solved at the
> right place. In i2c_new_device() we are storing parts of the fields of
> struct i2c_board_info, and when resetting the irq we are losing
> information. This patch solves that, but I wonder if the IRQ should
> not be 'simply' set in i2c_device_probe(). This means we also need to
> store the .resources of info, but I have a feeling this will be less
> error prone in the future.
> 

I would be somewhat inclined to agree here, it does seem odd that
on some paths we are allocating the IRQ on the new_device side
and on some on the probe side.

> But this is just my guts telling me something is not right. I would
> perfectly understand if we want to get this merged ASAP.
> 
> So given that the code is correct, this is my:
> Reviewed-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
> 

This would be my thinking as well we should merge this to avoid
the regression.

Thanks,
Charles