ASoC: Improve hi6210-i2s DT bindings
Subsystems (3)
PATCH 1/2 ASoC: Improve hi6210-i2s DT bindings
This patch improves the previously submitted hi6210-i2s DT binding, adding extra details to how the multi-dai index value maps to the potential interfaces. (Currently just index 0 -> the S2 interface, as there is only one supported, but in the future other interfaces may be enabled.)
--- Documentation/devicetree/bindings/sound/hisilicon,hi6210-i2s.txt | 8 ++++++++ 1 file changed, 8 insertions(+)
Documentation/devicetree/bindings/sound/hisilicon,hi6210-i2s.txt +8 -0
@@ -17,6 +17,10 @@ Required properties: - dma-names: should be "tx" and "rx" - hisilicon,sysctrl-syscon: phandle to sysctrl syscon - #sound-dai-cells: Should be set to 1 (for multi-dai) + - The dai cell indexes reference the following interfaces: + 0: S2 interface + (Currently that is the only one available, but more may be + supported in the future) Example for the hi6210 i2s controller:
@@ -32,3 +36,7 @@ i2s0: i2s@f7118000{ hisilicon,sysctrl-syscon = <&sys_ctrl>; #sound-dai-cells = <1>; }; + +Then when referencing the i2s controller: + sound-dai = <&i2s0 0>; /* index 0 => S2 interface */ +
Signature · 1 line
2.7.4
PATCH 2/2 ASoC: hisilicon: Minor fixups to hi6210 i2s audio driver
Mark had a few minor fixups he requested to the hi6210 i2s driver, so this patch proves them.
This patch adds a few extra error returns in cases that shouldn't happen, some style nits adding breaks to final default cases in switch statements, and tweaks to use devm_ variant of snd_soc_register_component.
--- sound/soc/hisilicon/hi6210-i2s.c | 14 ++++++-------- 1 file changed, 6 insertions(+), 8 deletions(-)
sound/soc/hisilicon/hi6210-i2s.c +6 -8
@@ -324,6 +324,7 @@ static int hi6210_i2s_hw_params(struct snd_pcm_substream *substream,default : i2s->bits =16 ; dma_data->addr_width =2 ; +break ; } i2s->rate =params_rate (params); i2s->channels =params_channels (params);
@@ -395,6 +396,7 @@ static int hi6210_i2s_hw_params(struct snd_pcm_substream *substream,break ;default :WARN_ONCE (1 ,"Invalid i2s->fmt MASTER_MASK. This shouldn't happen\n" ); +return -EINVAL ; }switch (i2s->format &SND_SOC_DAIFMT_FORMAT_MASK ) {
@@ -409,6 +411,7 @@ static int hi6210_i2s_hw_params(struct snd_pcm_substream *substream,break ;default :WARN_ONCE (1 ,"Invalid i2s->fmt FORMAT_MASK. This shouldn't happen\n" ); +return -EINVAL ; } val =hi6210_read_reg (i2s,HII2S_I2S_CFG );
@@ -440,6 +443,7 @@ static int hi6210_i2s_hw_params(struct snd_pcm_substream *substream, val =hi6210_read_reg (i2s,HII2S_I2S_CFG ); val &= ~HII2S_I2S_CFG__S2_FRAME_MODE ;hi6210_write_reg (i2s,HII2S_I2S_CFG , val); +break ; }/* clear loopback, set signed type and word length */
@@ -587,20 +591,14 @@ static int hi6210_i2s_probe(struct platform_device *pdev)if (ret)return ret; - ret =snd_soc_register_component (&pdev->dev, &hi6210_i2s_i2s_comp, + ret =devm_snd_soc_register_component (&pdev->dev, &hi6210_i2s_i2s_comp, &i2s->dai,1 ); -if (ret) { -dev_err (&pdev->dev,"Failed to register dai\n" ); -return ret; - } - -return 0 ; +return ret; }static int hi6210_i2s_remove (struct platform_device *pdev) {snd_soc_unregister_component (&pdev->dev); -dev_set_drvdata (&pdev->dev,NULL );return 0 ; }
Signature · 1 line
2.7.4
Mark Brown
maintainerRe: [PATCH 2/2] ASoC: hisilicon: Minor fixups to hi6210 i2s audio driver
Quoted 4 lines · John Stultz
This patch adds a few extra error returns in cases that shouldn't happen, some style nits adding breaks to final default cases in switch statements, and tweaks to use
devm_variant ofsnd_soc_register_component.
Please don't make multiple unrelated changes in a single patch, it's bad practice which can slow down the bits of the patch which are OK and makes it harder to review if the changes match up with the changelog.
Quoted 3 lines
static int
hi6210_i2s_remove(struct platform_device *pdev){ snd_soc_unregister_component(&pdev->dev);
Since you converted to devm_ this is now going to be a double free.
Re: [PATCH 2/2] ASoC: hisilicon: Minor fixups to hi6210 i2s audio driver
Quoted 10 lines · Mark Brown
Quoted 4 lines · John Stultz
This patch adds a few extra error returns in cases that shouldn't happen, some style nits adding breaks to final default cases in switch statements, and tweaks to use
devm_variant ofsnd_soc_register_component.Please don't make multiple unrelated changes in a single patch, it's bad practice which can slow down the bits of the patch which are OK and makes it harder to review if the changes match up with the changelog.
Apologies, I'll split it up and resend.
Quoted 5 lines
Quoted 3 lines
static int
hi6210_i2s_remove(struct platform_device *pdev){ snd_soc_unregister_component(&pdev->dev);Since you converted to
devm_this is now going to be a double free.
Ah. So it looks like can yank the whole remove implementation.
thanks
-john