Re: [PATCH] ALSA: hda/conexant: Add quirk for Acer TravelLite TL05-41M

[Takashi Iwai <tiwai@suse.de>]

On Fri, 06 Nov 2026 11:23:45 +0100,
Devajith Jijush wrote:
> 
> The Acer TravelLite TL05-41M (subsystem 0x1558a600) uses a Conexant
> CX11880 codec whose analog output amplifier is enabled by codec GPIO
> bit 1.  The driver has no quirk for this machine, so the line is never
> raised and both the internal speakers and the 3.5 mm combo jack stay
> silent, while the microphone on the same jack, USB and Bluetooth audio
> work normally.
> 
> Add a fixup that sets GPIO bit 1 on this subsystem ID.  It is applied
> from the HDA_FIXUP_ACT_INIT hook, since the codec loses the GPIO state
> on a power transition and the line would otherwise stay off after the
> first runtime power save.
> 
> Note that the existing CXT_FIXUP_GPIO1 fixup raises GPIO bit 0, which
> has no effect on this board.
> 
> Link: https://bugzilla.kernel.org/show_bug.cgi?id=222238
> Signed-off-by: Devajith Jijush <devajuice@zohomail.in>
> ---
> Note for reviewers: the GPIO behaviour was verified on the affected
> machine with hda-verb; I have not built a kernel with this patch, so the
> code path itself is untested.  Setting the same three GPIO bits from
> userspace makes both the internal speakers and the headphone jack work.
> 
>  sound/hda/codecs/conexant.c | 19 +++++++++++++++++++
>  1 file changed, 19 insertions(+)
> 
> diff --git a/sound/hda/codecs/conexant.c b/sound/hda/codecs/conexant.c
> index a141c94..623e466 100644
> --- a/sound/hda/codecs/conexant.c
> +++ b/sound/hda/codecs/conexant.c
> @@ -289,6 +289,7 @@ enum {
>  	CXT_FIXUP_HEADPHONE_MIC_PIN,
>  	CXT_FIXUP_HEADPHONE_MIC,
>  	CXT_FIXUP_GPIO1,
> +	CXT_FIXUP_AMP_GPIO1,
>  	CXT_FIXUP_ASPIRE_DMIC,
>  	CXT_FIXUP_THINKPAD_ACPI,
>  	CXT_FIXUP_LENOVO_XPAD_ACPI,
> @@ -826,6 +827,17 @@ static void cxt_fixup_hp_a_u(struct hda_codec *codec,
>  		cxt_setup_gpio_unmute(codec, 0x2);
>  }
>  
> +/* Acer TravelLite TL05-41M: the analog output amplifier is enabled by
> + * codec GPIO bit 1.  The codec loses the GPIO state on a power
> + * transition, so this must be applied on every init, not only at probe.
> + */
> +static void cxt_fixup_amp_gpio1(struct hda_codec *codec,
> +				const struct hda_fixup *fix, int action)
> +{
> +	if (action == HDA_FIXUP_ACT_INIT)
> +		snd_hda_codec_set_gpio(codec, 0x02, 0x02, 0x02, 0);
> +}
> +
>  /* Huawei MateBook 14 (SN6140): the speaker output follows the headphone
>   * pin's connection select in hardware.  Both pins must be routed to the
>   * shared DAC 0x10, which carries the playback stream; the generic
> @@ -1001,6 +1013,10 @@ static const struct hda_fixup cxt_fixups[] = {
>  			{ }
>  		},
>  	},
> +	[CXT_FIXUP_AMP_GPIO1] = {
> +		.type = HDA_FIXUP_FUNC,
> +		.v.func = cxt_fixup_amp_gpio1,
> +	},

There is already a GPIO bit 0 setup (called gpio1)

	[CXT_FIXUP_GPIO1] = {
		.type = HDA_FIXUP_VERBS,
		.v.verbs = (const struct hda_verb[]) {
			{ 0x01, AC_VERB_SET_GPIO_MASK, 0x01 },
			{ 0x01, AC_VERB_SET_GPIO_DIRECTION, 0x01 },
			{ 0x01, AC_VERB_SET_GPIO_DATA, 0x01 },
			{ }
		},
	},

so you can implement similarly,

	[CXT_FIXUP_GPIO2] = {
		.type = HDA_FIXUP_VERBS,
		.v.verbs = (const struct hda_verb[]) {
			{ 0x01, AC_VERB_SET_GPIO_MASK, 0x02 },
			{ 0x01, AC_VERB_SET_GPIO_DIRECTION, 0x02 },
			{ 0x01, AC_VERB_SET_GPIO_DATA, 0x02 },
			{ }
		},
	},

and use this in the quirk, instead.  Then you don't need to create a
function.

> @@ -1255,6 +1273,7 @@ static const struct hda_quirk cxt5066_fixups[] = {
>  static const struct hda_model_fixup cxt5066_fixup_models[] = {
>  	{ .id = CXT_FIXUP_STEREO_DMIC, .name = "stereo-dmic" },
>  	{ .id = CXT_FIXUP_GPIO1, .name = "gpio1" },
> +	{ .id = CXT_FIXUP_AMP_GPIO1, .name = "amp-gpio1" },
>  	{ .id = CXT_FIXUP_HEADPHONE_MIC_PIN, .name = "headphone-mic-pin" },
>  	{ .id = CXT_PINCFG_LENOVO_TP410, .name = "tp410" },
>  	{ .id = CXT_FIXUP_THINKPAD_ACPI, .name = "thinkpad" },

Please avoid adding a new model string unless needed inevitably.
We can apply the existing quirk with the alias (xxxx:yyyy), too.


thanks,

Takashi