On 04/03/2021 1:39 pm, Takashi Iwai wrote: Thank you very much for quick response!

On Wed, 03 Mar 2021 19:29:57 +0100, Vitaly Rodionov wrote:

...

+static const struct hda_verb cs8409_cs42l42_init_verbs[] = {

• { 0x01, AC_VERB_SET_POWER_STATE, 0x0000 }, /* AFG: D0 */

I guess this power state change is superfluous. The AFG node is already powered up when the codec probe or init is called.

Yes, This should be removed

...

• { 0x01, AC_VERB_SET_GPIO_DIRECTION, 0x0020 }, /* GPIO 5 out, 3,4 in */
• { 0x01, AC_VERB_SET_GPIO_DATA, 0x0000 }, /* GPIO data 0 */
• { 0x01, AC_VERB_SET_GPIO_MASK, 0x003f }, /* Enable GPIO */

Those are handled in spec->gpio_dir, gpio->data and gpio->mask fields.

Will handle via spec fields.

...

• { 0x01, AC_VERB_SET_GPIO_WAKE_MASK, 0x0018 }, /* WAKE from GPIO 3,4 */
• { 0x47, AC_VERB_SET_PROC_STATE, 0x0001 }, /* Enable VPW processing */
• { 0x47, AC_VERB_SET_COEF_INDEX, 0x0002 }, /* Configure GPIO 6,7 */
• { 0x47, AC_VERB_SET_PROC_COEF, 0x0080 }, /* I2C mode */
• { 0x47, AC_VERB_SET_COEF_INDEX, 0x005b }, /* Set I2C bus speed */
• { 0x47, AC_VERB_SET_PROC_COEF, 0x0200 }, /* 100kHz I2C_STO = 2 */

Those remaining verbs are good in the init verbs. But I suppose they have to be applied at the resume as well? But...

This was added in the next patch, but you are right we should handle it here.

...

+static int cs8409_cs42l42_fixup(struct hda_codec *codec) +{

• int err = 0;
• struct cs_spec *spec = codec->spec;
• unsigned int pincap = 0;
• /* Basic initial sequence for specific hw configuration */
• snd_hda_sequence_write(codec, cs8409_cs42l42_init_verbs);

... it seems applied only at the fixup call at parsing?

Ditto about cs8409_cs42l42_hw_init(codec).

Yes, needs to be fixed. I will make v2 patch set.

thanks,

Takashi