On Mon, 10 Nov 2014 08:07:50 +0100 Lars-Peter Clausen lars@metafoo.de wrote:
On 11/10/2014 01:45 AM, NeilBrown wrote:
diff --git a/sound/soc/codecs/twl4030.c b/sound/soc/codecs/twl4030.c index b6b0cb399599..613b61cee081 100644 --- a/sound/soc/codecs/twl4030.c +++ b/sound/soc/codecs/twl4030.c @@ -957,6 +957,16 @@ static int snd_soc_put_twl4030_opmode_enum_double(struct snd_kcontrol *kcontrol, { struct snd_soc_codec *codec = snd_soc_kcontrol_codec(kcontrol); struct twl4030_priv *twl4030 = snd_soc_codec_get_drvdata(codec);
- struct snd_ctl_elem_value currentval;
snd_ctl_elem_value is a bit to big to be put onto the kernel stack. Just using twl4030_read() should be fine.
That's a shame, it looked so neat....
Using twl4030_read forces i2c access and misses out on the regmap caching.
What do you think of this?
Thanks, NeilBrown
diff --git a/sound/soc/codecs/twl4030.c b/sound/soc/codecs/twl4030.c index b6b0cb399599..bdb47a045aa5 100644 --- a/sound/soc/codecs/twl4030.c +++ b/sound/soc/codecs/twl4030.c @@ -957,6 +957,18 @@ static int snd_soc_put_twl4030_opmode_enum_double(struct snd_kcontrol *kcontrol, { struct snd_soc_codec *codec = snd_soc_kcontrol_codec(kcontrol); struct twl4030_priv *twl4030 = snd_soc_codec_get_drvdata(codec); + const struct soc_enum *e = &twl4030_op_modes_enum; + unsigned int reg_val; + int ret; + + ret = snd_soc_component_read(snd_kcontrol_chip(kcontrol), + e->reg, ®_val); + if (ret) + return ret; + if (ucontrol->value.enumerated.item[0] == + ((reg_val >> e->shift_l) & e->mask)) + /* no change requested, so do nothing */ + return 0;
if (twl4030->configured) { dev_err(codec->dev,