sgtl5000 fails after suspend/resume
diff mbox

Message ID CAOMZO5CyE5T-H=K2Q9A=tYoY9zt0CO4YdFWK5PMS=NR2XbqPzQ@mail.gmail.com
State New
Headers show

Commit Message

Fabio Estevam Feb. 15, 2018, 4:36 p.m. UTC
On Thu, Feb 15, 2018 at 11:35 AM, Fabio Estevam <festevam@gmail.com> wrote:
> Hi,
>
> I noticed that sgtl5000 does not work after a suspend/resume cycle:
>
> # aplay /media/a2002011001-e02.wav
> Playing WAVE '/media/a2002011001-e02.wav' : Signed 16 bit Little
> Endian, Rate 44100 Hz, Stereo
> aplay: pcm_write:2051: write error: Input/output error
>
> Tested several 4.16-rc1 (and also previous kernels like 4.4) and all
> of them failed.
>
> I think it worked in some point in time.
>
> Tested on a imx53qsb and imx6 wandboard.
>
> Does anyone have any suggestions?

Reverting the commit below on top of 4.16-rc1 makes sgtl5000
suspend/resume to work again:

commit 8419caa7270291e26f8b34b12b29680586c85d30
Author: Eric Nelson <eric@nelint.com>
Date:   Tue Jun 7 01:14:52 2016 +0200

    ASoC: sgtl5000: Do not disable regulators in SND_SOC_BIAS_OFF

    Disabling the SGTL5000 through regulators would certainly save more
    power than simply disabling the reference voltages as described in the
    data sheet, but won't properly restore things on resume.

    This driver does not support active regulators. So we simply disable the
    reference bias currents.

    Signed-off-by: Eric Nelson <eric@nelint.com>
    Signed-off-by: Clemens Gruber <clemens.gruber@pqgruber.com>
    Reviewed-by: Fabio Estevam <fabio.estevam@nxp.com>
    Signed-off-by: Mark Brown <broonie@kernel.org>

Or if I do the change below against linux-next it also works:

        }

Any suggestions for a proper fix? Eric?

Thanks

Comments

Fabio Estevam Feb. 15, 2018, 6:11 p.m. UTC | #1
On Thu, Feb 15, 2018 at 2:36 PM, Fabio Estevam <festevam@gmail.com> wrote:

> Or if I do the change below against linux-next it also works:
>
> --- a/sound/soc/codecs/sgtl5000.c
> +++ b/sound/soc/codecs/sgtl5000.c
> @@ -880,8 +880,6 @@ static int sgtl5000_set_bias_level(struct
> snd_soc_component *component,
>                                     SGTL5000_REFTOP_POWERUP);
>                 break;
>         case SND_SOC_BIAS_OFF:
> -               snd_soc_component_update_bits(component,
> SGTL5000_CHIP_ANA_POWER,
> -                                   SGTL5000_REFTOP_POWERUP, 0);
>                 break;
>         }
>
> Any suggestions for a proper fix? Eric?

I sent a patch that that removes sgtl5000_set_bias_level() completely
and fixes the sgtl5000 suspend/resume issue.
Mark Brown Feb. 15, 2018, 6:16 p.m. UTC | #2
On Thu, Feb 15, 2018 at 02:36:00PM -0200, Fabio Estevam wrote:

> Reverting the commit below on top of 4.16-rc1 makes sgtl5000
> suspend/resume to work again:

> @@ -880,8 +880,6 @@ static int sgtl5000_set_bias_level(struct
> snd_soc_component *component,
>                                     SGTL5000_REFTOP_POWERUP);
>                 break;
>         case SND_SOC_BIAS_OFF:
> -               snd_soc_component_update_bits(component,
> SGTL5000_CHIP_ANA_POWER,
> -                                   SGTL5000_REFTOP_POWERUP, 0);
>                 break;
>         }

> Any suggestions for a proper fix? Eric?

Does it need a register map resync when powering on again?
Max Krummenacher Feb. 15, 2018, 6:25 p.m. UTC | #3
Hi Fabio

On Thu, 2018-02-15 at 14:36 -0200, Fabio Estevam wrote:
> On Thu, Feb 15, 2018 at 11:35 AM, Fabio Estevam <festevam@gmail.com> wrote:
> > 
> > Hi,
> > 
> > I noticed that sgtl5000 does not work after a suspend/resume cycle:
> > 
> > # aplay /media/a2002011001-e02.wav
> > Playing WAVE '/media/a2002011001-e02.wav' : Signed 16 bit Little
> > Endian, Rate 44100 Hz, Stereo
> > aplay: pcm_write:2051: write error: Input/output error
> > 
> > Tested several 4.16-rc1 (and also previous kernels like 4.4) and all
> > of them failed.
> > 
> > I think it worked in some point in time.
> > 
> > Tested on a imx53qsb and imx6 wandboard.
> > 
> > Does anyone have any suggestions?
> 
> Reverting the commit below on top of 4.16-rc1 makes sgtl5000
> suspend/resume to work again:
> 
> commit 8419caa7270291e26f8b34b12b29680586c85d30
> Author: Eric Nelson <eric@nelint.com>
> Date:   Tue Jun 7 01:14:52 2016 +0200
> 
>     ASoC: sgtl5000: Do not disable regulators in SND_SOC_BIAS_OFF
> 
>     Disabling the SGTL5000 through regulators would certainly save more
>     power than simply disabling the reference voltages as described in the
>     data sheet, but won't properly restore things on resume.
> 
>     This driver does not support active regulators. So we simply disable the
>     reference bias currents.
> 
>     Signed-off-by: Eric Nelson <eric@nelint.com>
>     Signed-off-by: Clemens Gruber <clemens.gruber@pqgruber.com>
>     Reviewed-by: Fabio Estevam <fabio.estevam@nxp.com>
>     Signed-off-by: Mark Brown <broonie@kernel.org>
> 
> Or if I do the change below against linux-next it also works:
> 
> --- a/sound/soc/codecs/sgtl5000.c
> +++ b/sound/soc/codecs/sgtl5000.c
> @@ -880,8 +880,6 @@ static int sgtl5000_set_bias_level(struct
> snd_soc_component *component,
>                                     SGTL5000_REFTOP_POWERUP);
>                 break;
>         case SND_SOC_BIAS_OFF:
> -               snd_soc_component_update_bits(component,
> SGTL5000_CHIP_ANA_POWER,
> -                                   SGTL5000_REFTOP_POWERUP, 0);
>                 break;
>         }
> 
> Any suggestions for a proper fix? Eric?
> 

I tested this on a Colibri iMX6DL with the same results.
Vanilla 4.16-rc1 does resume but audio is silent after resume.
The interrupted aplay process is in the process list and never finishes.

Both your fixes make aplay continue playing sound again until the end
of the played sound file.

Strangely if I use our downstream kernel (based on NXP/linux-fslc
4.9-1.0.x-imx which has the 8419caa7270291e26f8b34b12b29680586c85d30
commit) suspend/resume works.

Max

Patch
diff mbox

--- a/sound/soc/codecs/sgtl5000.c
+++ b/sound/soc/codecs/sgtl5000.c
@@ -880,8 +880,6 @@  static int sgtl5000_set_bias_level(struct
snd_soc_component *component,
                                    SGTL5000_REFTOP_POWERUP);
                break;
        case SND_SOC_BIAS_OFF:
-               snd_soc_component_update_bits(component,
SGTL5000_CHIP_ANA_POWER,
-                                   SGTL5000_REFTOP_POWERUP, 0);
                break;