I am not a big fan of the code partition you've selected.
+void sof_rt1015p_set_share_en_spk(void) +{
- /* Two amps share one en pin so there is only one device in acpi
* table*/- rt1015p_quirk |= RT1015P_SHARE_EN_SPK;
+}
This is a function now used in the machine driver, see below [1] This adds a new interface between machine and realtek common code, which we are trying to move to a module with well-defined APIs.
+void sof_rt1015p_dai_link(struct snd_soc_dai_link *link) +{
- link->codecs = rt1015p_dai_link_components;
- if (rt1015p_quirk & RT1015P_SHARE_EN_SPK)
link->num_codecs = 1;- else
link->num_codecs = ARRAY_SIZE(rt1015p_dai_link_components);- link->init = rt1015p_init;
- link->ops = &rt1015p_ops;
+}
+void sof_rt1015p_codec_conf(struct snd_soc_card *card) +{
- if (rt1015p_quirk & RT1015P_SHARE_EN_SPK)
return;- card->codec_conf = rt1015p_codec_confs;
- card->num_configs = ARRAY_SIZE(rt1015p_codec_confs);
+}
could we not handle this quirk inside one of the two dai_link or codec_conf functions above? The machine driver does not care about this RT1015P_SHARE_EN_SPK quirk, it's only used in those two functions so should be set in this scope. There's no need to make it visible to the machine driver, is there?
- /* setup amp quirk if any */
- if (sof_rt5682_quirk & SOF_RT1015P_SPEAKER_AMP_PRESENT) {
/* There may be only one device instance if two amps* sharing one en pin*/if (!acpi_dev_present("RTL1015", "1", -1)) {dev_dbg(&pdev->dev, "Device %s not exist\n",RT1015P_DEV1_NAME);sof_rt1015p_set_share_en_spk();}- }
[1]
I see no problem using the _UID (Unique ID) to check if a second amplifier is present or not. This seems to follow the ACPI spec Section 6.1.12, as long as "the _UID must be unique across all devices with either a common _HID or _CID"
- else if (sof_rt5682_quirk & SOF_RT1015P_SPEAKER_AMP_PRESENT)
sof_rt1015p_codec_conf(&sof_audio_card_rt5682);