[RFC,v1] tlv320aic32x4: delay i2c access by 1 ms after hardware reset
diff mbox series

Message ID 20190210154519.2506-1-ps.report@gmx.net
State New
Headers show
Series
  • [RFC,v1] tlv320aic32x4: delay i2c access by 1 ms after hardware reset
Related show

Commit Message

Peter Seiderer Feb. 10, 2019, 3:45 p.m. UTC
As stated in 'TLV320AIC3254 Application Reference Guide' ([1]):

  3.2 Device Startup Lockout Times
    After the TLV320AIC3254 initializes through hardware reset at power-up
    or software reset, the internal registers initialize to default values.
    This initialization takes place within 1ms after pulling the RESET
    signal high. During this initialization phase, no register-read or
    register-write operation should be performed on ADC or DAC coefficient
    buffers. Also, no block within the codec should be powered up during
    the initialization phase.

[1] http://www.ti.com/lit/an/slaa408a/slaa408a.pdf

Signed-off-by: Peter Seiderer <ps.report@gmx.net>
---
Note: This came up after the electronic departement took a deeper look at
the i2c-bus/vcc/reset pins with a logic analyzer and the documentation
because of a vcc/reset timing problem (now fixed on custom board).
Not 100% sure if the 1 ms is needed (as restricted to ADC/DAC coefficent
and codec block power up).
---
 sound/soc/codecs/tlv320aic32x4.c | 2 ++
 1 file changed, 2 insertions(+)

Comments

Mark Brown Feb. 11, 2019, 3:04 p.m. UTC | #1
On Sun, Feb 10, 2019 at 04:45:19PM +0100, Peter Seiderer wrote:

> @@ -972,6 +972,8 @@ static int aic32x4_component_probe(struct snd_soc_component *component)
>  		gpio_set_value(aic32x4->rstn_gpio, 1);
>  	}
>  
> +	mdelay(1);
> +

Perhaps only do this if we toggled the GPIO?  If the device wasn't in
reset then there's no need to wait.
Peter Seiderer Feb. 11, 2019, 9:02 p.m. UTC | #2
Hello Mark,

On Mon, 11 Feb 2019 15:04:25 +0000, Mark Brown <broonie@kernel.org> wrote:

> On Sun, Feb 10, 2019 at 04:45:19PM +0100, Peter Seiderer wrote:
> 
> > @@ -972,6 +972,8 @@ static int aic32x4_component_probe(struct snd_soc_component *component)
> >  		gpio_set_value(aic32x4->rstn_gpio, 1);
> >  	}
> >  
> > +	mdelay(1);
> > +  
> 
> Perhaps only do this if we toggled the GPIO?  If the device wasn't in
> reset then there's no need to wait.

You are absolutely right with this, will re-spin the patch with
the suggested change, thanks for review...

Regards,
Peter

Patch
diff mbox series

diff --git a/sound/soc/codecs/tlv320aic32x4.c b/sound/soc/codecs/tlv320aic32x4.c
index e1bfba62fc08..b3beb0ca87c3 100644
--- a/sound/soc/codecs/tlv320aic32x4.c
+++ b/sound/soc/codecs/tlv320aic32x4.c
@@ -972,6 +972,8 @@  static int aic32x4_component_probe(struct snd_soc_component *component)
 		gpio_set_value(aic32x4->rstn_gpio, 1);
 	}
 
+	mdelay(1);
+
 	snd_soc_component_write(component, AIC32X4_RESET, 0x01);
 
 	if (aic32x4->setup)