On Wed, Aug 11, 2021 at 8:46 AM Mugilraj Dhavachelvan <dmugil2000@xxxxxxxxx> wrote: > On Tue, Aug 10, 2021 at 03:49:52PM +0300, Andy Shevchenko wrote: > > On Mon, Aug 9, 2021 at 10:59 AM Mugilraj Dhavachelvan > > <dmugil2000@xxxxxxxxx> wrote: ... > > > + data->tol = data->cfg->kohms * (val & GENMASK(6, 0)) * 10 / 8; > > > + if (!(val & BIT(7))) > > > + data->tol *= -1; > > > > Shouldn't you simple use corresponding sign_extend*()? > > > I'm not able see any sign_extend for 16 bit. Is there any other way? So, then add it in bitops.h the same way it's done for s32 and s64. ... > > > + if (!data->cfg) > > > + data->cfg = &ad5110_cfg[i2c_match_id(ad5110_id, client)->driver_data]; > > > > Not sure this is not a dead code since you are using ->probe_new(). > > > Even I'm suspecting that and also removing id_table. But I'm not sure of > it so just left as it is. I²C ID table is good to have without direct use, but ->probe_new() is called if and only if there is a compatible string or ACPI ID match. In such case data->cfg may be NULL if and only if the corresponding table missed it, but this will be a bug anyway, so the above code will rather hide the bug. Hence, please remove it. -- With Best Regards, Andy Shevchenko
