Hi Stephen
Thank you for checking patch
diff --git a/sound/soc/sh/ssi.c b/sound/soc/sh/ssi.c
static int sh4_soc_dai_probe(struct platform_device *pdev) {
- return snd_soc_register_dais(&pdev->dev, sh4_ssi_dai,
ARRAY_SIZE(sh4_ssi_dai));
- return snd_soc_register_component(&pdev->dev, &sh4_ssi_component,
sh4_ssi_dai, ARRAY_SIZE(sh4_ssi_dai));}
ARRAY_SIZE(sh4_ssi_dai) might be 1 or 2 here, based on #ifdef CONFIG_CPU_SUBTYPE_SH7760. If it's 1, then this change switches the code from calling snd_soc_register_dais() to snd_soc_register_dai(). This might be a problem. In practice, perhaps it's won't since the .name fields in the array entries already are "ssi-dai.0" and "ssi-dai.1", so perhaps the device ID is already expected not to be appended (because it's -1 as the device is a singleton)? But, this needs to be checked by someone familiar with the platform...
I'm not sure why, but there was no user for this driver on upstream kernel
Best regards --- Kuninori Morimoto