Re: [PATCH 10/23] iio: accel: kxcjk-1013: Convert to i2c's .probe_new

["Uwe Kleine-König" <u.kleine-koenig@pengutronix.de>]

[-- Attachment #1: Type: text/plain, Size: 2282 bytes --]

On Sun, Oct 23, 2022 at 10:06:11PM +0300, Andy Shevchenko wrote:
> On Sun, Oct 23, 2022 at 03:22:49PM +0200, Uwe Kleine-König wrote:
> > .probe_new() doesn't get the i2c_device_id * parameter, so determine
> > that explicitly in .probe(). The device_id array has to move up for that
> > to work.
> 
> ...
> 
> > +static const struct i2c_device_id kxcjk1013_id[] = {
> > +	{"kxcjk1013", KXCJK1013},
> > +	{"kxcj91008", KXCJ91008},
> > +	{"kxtj21009", KXTJ21009},
> > +	{"kxtf9",     KXTF9},
> > +	{"kx023-1025", KX0231025},
> > +	{"SMO8500",   KXCJ91008},
> > +	{}
> > +};
> > +MODULE_DEVICE_TABLE(i2c, kxcjk1013_id);
> 
> I don't like this part. Can we, please, find a way how to dereference this
> table via struct i2c_client, please?

It would be possible to do (on top of my patch here as PoC):

diff --git a/drivers/iio/accel/kxcjk-1013.c b/drivers/iio/accel/kxcjk-1013.c
index e043dd698747..00269b25af99 100644
--- a/drivers/iio/accel/kxcjk-1013.c
+++ b/drivers/iio/accel/kxcjk-1013.c
@@ -1445,7 +1445,7 @@ MODULE_DEVICE_TABLE(i2c, kxcjk1013_id);
 
 static int kxcjk1013_probe(struct i2c_client *client)
 {
-	const struct i2c_device_id *id = i2c_match_id(kxcjk1013_id, client);
+	const struct i2c_device_id *id = i2c_match_id(to_i2c_driver(client->dev.driver)->id_table, client);
 	struct kxcjk1013_data *data;
 	struct iio_dev *indio_dev;
 	struct kxcjk_1013_platform_data *pdata;

(only compile tested), you could even create a function or macro to make
this a bit prettier on the source level. For the compiler loading the
address from a local symbol instead of from two pointer dereferences is
(I guess) a bit more effective and IMHO more natural.

*shrug*, I don't care much, but I don't like to have to rework this
series just because you don't like this part. You even didn't give a
rationale, I can imagine several different ones:

 [ ] it makes the patch bigger
 [ ] it results in an unnatural order of symbols in the driver
 [ ] it's some kind of duplication
 [ ] something else
     please elaborate: ________________________________

Best regards
Uwe

-- 
Pengutronix e.K.                           | Uwe Kleine-König            |
Industrial Linux Solutions                 | https://www.pengutronix.de/ |

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]