Re: [PATCH v4 06/13] iio: dac: ds4424: use device match data for chip info

[Jonathan Cameron]

On Tue, 3 Feb 2026 16:56:00 +0200
Andy Shevchenko <andriy.shevchenko@xxxxxxxxx> wrote:

> On Tue, Feb 03, 2026 at 01:00:02PM +0100, Oleksij Rempel wrote:
> > On Tue, Feb 03, 2026 at 01:51:23PM +0200, Andy Shevchenko wrote:  
> > > On Tue, Feb 03, 2026 at 11:17:42AM +0100, Oleksij Rempel wrote:  
> > > > On Tue, Feb 03, 2026 at 12:03:23PM +0200, Andy Shevchenko wrote:  
> > > > > On Tue, Feb 03, 2026 at 10:34:26AM +0100, Oleksij Rempel wrote:  
> 
> ...
> 
> > > > > > -	indio_dev->name = id->name;  
> > > > >   
> > > > > > +	indio_dev->name = client->name;  
> > > > > 
> > > > > Isn't this an ABI breakage?  
> > > > 
> > > > I can't confirm it.
> > > > 
> > > > before all patches:
> > > > root@DistroKit:~ cat /sys/bus/iio/devices/iio:device3/name 
> > > > ds4424
> > > > 
> > > > after:
> > > > root@DistroKit:~ cat /sys/bus/iio/devices/iio:device3/name 
> > > > ds4424  
> > > 
> > > In ACPI case it might look differently, but I have no means to test this.
> > > 
> > > id->name comes strictly from an i2c table, while client->name is constructed
> > > using specifics of the firmware enumeration. In DT due to some (historical?)
> > > reasons the client->name has no vendor substring and hence matches 1:1 to
> > > id->name. In ACPI, IIRC, the client->name is ACPI device instance name,
> > > something like ABCD0123:00.  
> > 
> > Ok, I see. Should I revert this line?  
> 
> Just do not introduce that change (change of the ->name field) in the original
> patch, in that case no revert churn would be needed.
> 
I think this got dealt with in discussion of next version but
safest route is just have an extra copy of the name in the
chip_info structure.  Then we know it's stable against different
firmware types etc.

The few places we have client->name are all ancient bugs that
are really hard to fix years later without risk :(

Jonathan