RE: [EXTERNAL] Re: [PATCH v3 4/5] ALSA: hda/tas2781: Add tas2781 HDA driver

[Ding, Shenghao]

> -----Original Message-----
> From: Takashi Iwai <tiwai@xxxxxxx>
> Sent: Tuesday, May 23, 2023 7:43 PM
> To: Ding, Shenghao <shenghao-ding@xxxxxx>
> Cc: Shenghao Ding <13916275206@xxxxxxx>; broonie@xxxxxxxxxx;
> devicetree@xxxxxxxxxxxxxxx; krzysztof.kozlowski+dt@xxxxxxxxxx;
> robh+dt@xxxxxxxxxx; lgirdwood@xxxxxxxxx; perex@xxxxxxxx; pierre-
> louis.bossart@xxxxxxxxxxxxxxx; Lu, Kevin <kevin-lu@xxxxxx>; alsa-
> devel@xxxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx; Xu, Baojun
> <x1077012@xxxxxx>; Gupta, Peeyush <peeyush@xxxxxx>; Navada Kanyana,
> Mukund <navada@xxxxxx>; gentuser@xxxxxxxxx; Ryan_Chu@xxxxxxxxxxx;
> Sam_Wu@xxxxxxxxxxx
> Subject: Re: [EXTERNAL] Re: [PATCH v3 4/5] ALSA: hda/tas2781: Add tas2781
> HDA driver
> 
> On Tue, 23 May 2023 13:22:03 +0200,
> Ding, Shenghao wrote:
> >
> > > +	[ALC287_FIXUP_TAS2781_I2C_2] = {
> > > +		.type = HDA_FIXUP_FUNC,
> > > +		.v.func = tas2781_fixup_i2c,
> > > +		.chained = true,
> > > +		.chain_id = ALC269_FIXUP_THINKPAD_ACPI,
> > > +	},
> > > +	[ALC287_FIXUP_TAS2781_I2C_4] = {
> > > +		.type = HDA_FIXUP_FUNC,
> > > +		.v.func = tas2781_fixup_i2c,
> > > +		.chained = true,
> > > +		.chain_id = ALC269_FIXUP_THINKPAD_ACPI,
> > > +	},
> >
> > What's a difference between *_2 and *_4?
> > Combine them into ALC287_FIXUP_TAS2781_I2C
> 
> Hm, so there is no difference in stereo and quad speakers?
Yes, our firmware defines the stereo or quad speaker
> 
> > > +static int tas2781_save_calibration(struct tasdevice_priv
> > > +*tas_priv) {
> > > +	efi_guid_t efi_guid = EFI_GUID(0x02f9af02, 0x7734, 0x4233, 0xb4,
> 0x3d,
> > > +		0x93, 0xfe, 0x5a, 0xa3, 0x5d, 0xb3);
> > > +	static efi_char16_t efi_name[] = L"CALI_DATA";
> > > +	struct hda_codec *codec = tas_priv->codec;
> > > +	unsigned int subid = codec->core.subsystem_id & 0xFFFF;
> > > +	struct tm *tm = &tas_priv->tm;
> > > +	unsigned int attr, crc;
> > > +	unsigned int *tmp_val;
> > > +	efi_status_t status;
> > > +	int ret = 0;
> > > +
> > > +	//Lenovo devices
> > > +	if ((subid == 0x387d) || (subid == 0x387e) || (subid == 0x3881)
> > > +		|| (subid == 0x3884) || (subid == 0x3886) || (subid == 0x38a7)
> > > +		|| (subid == 0x38a8) || (subid == 0x38ba) || (subid ==
> 0x38bb)
> > > +		|| (subid == 0x38be) || (subid == 0x38bf) || (subid == 0x38c3)
> > > +		|| (subid == 0x38cb) || (subid == 0x38cd))
> > > +		efi_guid = EFI_GUID(0x1f52d2a1, 0xbb3a, 0x457d, 0xbc, 0x09,
> > > +			0x43, 0xa3, 0xf4, 0x31, 0x0a, 0x92);
> >
> > Here can be a problem: the device ID is embedded here, and it's hard to
> find out.  You'd better to make it some quirk flag that is set in a common
> place and check the flag here instead of checking ID at each place.
> >
> > Do you have example of the solution? I found some quirk flag is static in
> the patch_realtek.c, can't be accessed outside that file.
> 
> You may store some values in struct hda_component, I suppose?
I will try it.
> 
> BTW, please try to fix your mailer to do citation more correctly...
accept
> 
> 
> thanks,
> 
> Takashi