Re: [PATCH v3 3/5] ASoC: qcom: Add support for lpass hdmi driver

[Stephen Boyd]

Quoting Srinivasa Rao Mandadapu (2020-08-30 23:39:22)
> diff --git a/sound/soc/qcom/lpass-hdmi.c b/sound/soc/qcom/lpass-hdmi.c
> new file mode 100644
> index 0000000..7e18113
> --- /dev/null
> +++ b/sound/soc/qcom/lpass-hdmi.c
> @@ -0,0 +1,684 @@
> +// SPDX-License-Identifier: GPL-2.0-only
[...]
> +
> +static int lpass_hdmi_daiops_hw_params(struct snd_pcm_substream *substream,
> +               struct snd_pcm_hw_params *params, struct snd_soc_dai *dai)
> +{
[...]
> +       data_format = LPASS_DATA_FORMAT_LINEAR;
> +       ch_sts_buf0 = (((data_format << LPASS_DATA_FORMAT_SHIFT) & LPASS_DATA_FORMAT_MASK)
> +                               | ((sampling_freq << LPASS_FREQ_BIT_SHIFT) & LPASS_FREQ_BIT_MASK));
> +       ch_sts_buf1 = (word_length) & LPASS_WORDLENGTH_MASK;
> +
> +       ret = regmap_field_write(drvdata->tx_ctl->soft_reset, LPASS_TX_CTL_RESET);
> +       if (ret) {
> +               dev_err(dai->dev, "%s error writing to softreset enable : %d\n",

All of these strings bloat the kernel image. Can we just return ret
instead and if something goes wrong we can use a debug patch to figure
out which register write failed? Would a register write even fail to
begin with?

> +                                       __func__, ret);
> +               return ret;
> +       }
> +
> +       ret = regmap_field_write(drvdata->tx_ctl->soft_reset, LPASS_TX_CTL_CLEAR);
> +       if (ret) {
> +               dev_err(dai->dev, "%s error writing to softreset disable : %d\n",
> +                                       __func__, ret);
> +               return ret;
> +       }
> +
> +       ret = regmap_field_write(drvdata->legacy->legacy_en,
> +                               LPASS_HDMITX_LEGACY_DISABLE);
> +       if (ret) {
> +               dev_err(dai->dev, "%s error writing to legacy_en field : %d\n",
> +                                       __func__, ret);
> +               return ret;
> +       }
> +
[...]
>  #define LPAIF_DMACTL_REG(v, chan, dir) __LPAIF_DMA_REG(v, chan, dir, CTL)
>  #define LPAIF_DMABASE_REG(v, chan, dir) __LPAIF_DMA_REG(v, chan, dir, BASE)
> diff --git a/sound/soc/qcom/lpass-platform.c b/sound/soc/qcom/lpass-platform.c
> index df692ed..607f4c4 100644
> --- a/sound/soc/qcom/lpass-platform.c
> +++ b/sound/soc/qcom/lpass-platform.c
> @@ -553,7 +702,21 @@ static irqreturn_t lpass_platform_lpaif_irq(int irq, void *data)
>  
>         /* Handle per channel interrupts */
>         for (chan = 0; chan < LPASS_MAX_DMA_CHANNELS; chan++) {
> -               if (irqs & LPAIF_IRQ_ALL(chan) && drvdata->substream[chan]) {
> +               switch (v->id) {
> +               case HDMI_INTERFACE:
> +                       val = LPAIF_IRQ_HDMI_REQ_ON_PRELOAD(chan) |
> +                               LPAIF_IRQ_HDMI_METADONE |
> +                               LPAIF_IRQ_HDMI_SDEEP_AUD_DIS(chan);
> +                       break;
> +               case I2S_INTERFACE:
> +                       val = 0;
> +                       break;
> +               default:
> +                       pr_err("%s: invalid  %d interface\n", __func__, v->id);

Any reason we can't use dev_err() here?

> +                       return -EINVAL;
> +               }
> +               if (irqs & (LPAIF_IRQ_ALL(chan) | val)
> +                       && drvdata->substream[chan]) {
>                         rv = lpass_dma_interrupt_handler(
>                                                 drvdata->substream[chan],
>                                                 drvdata, chan, irqs);
> @@ -644,15 +807,15 @@ int asoc_qcom_lpass_platform_register(struct platform_device *pdev)
>  
>         /* ensure audio hardware is disabled */
>         ret = regmap_write(drvdata->lpaif_map,
> -                       LPAIF_IRQEN_REG(v, LPAIF_IRQ_PORT_HOST), 0);
> +                       IRQ_EN(v, LPAIF_IRQ_PORT_HOST), 0);
>         if (ret) {
>                 dev_err(&pdev->dev, "error writing to irqen reg: %d\n", ret);
>                 return ret;
>         }
>  
>         ret = devm_request_irq(&pdev->dev, drvdata->lpaif_irq,
> -                       lpass_platform_lpaif_irq, IRQF_TRIGGER_RISING,
> -                       "lpass-irq-lpaif", drvdata);
> +               lpass_platform_lpaif_irq, IRQF_TRIGGER_RISING,

Can we use the irq flags from the firmware, i.e. whatever the DT or ACPI
tables say?

> +               pdev->name, drvdata);
>         if (ret) {
>                 dev_err(&pdev->dev, "irq request failed: %d\n", ret);
>                 return ret;