[PATCH 1/1] drm/tegra: Add guard to avoid double disable/enable of RGB outputs

[PATCH 1/1] drm/tegra: Add guard to avoid double disable/enable of RGB outputs

[Dmitry Osipenko]

Add guard to check whether RGB output is already enabled in the way it's
done for HDMI output. Fixes possible hang on trying to disable output twice
(first time during driver probe and second on fb registering).

Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
---
 drivers/gpu/drm/tegra/rgb.c | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/drivers/gpu/drm/tegra/rgb.c b/drivers/gpu/drm/tegra/rgb.c
index 338f7f6..0266fb4 100644
--- a/drivers/gpu/drm/tegra/rgb.c
+++ b/drivers/gpu/drm/tegra/rgb.c
@@ -15,6 +15,7 @@
 struct tegra_rgb {
 	struct tegra_output output;
 	struct tegra_dc *dc;
+	bool enabled;
 
 	struct clk *clk_parent;
 	struct clk *clk;
@@ -89,6 +90,9 @@ static int tegra_output_rgb_enable(struct tegra_output *output)
 	struct tegra_rgb *rgb = to_rgb(output);
 	unsigned long value;
 
+	if (rgb->enabled)
+		return 0;
+
 	tegra_dc_write_regs(rgb->dc, rgb_enable, ARRAY_SIZE(rgb_enable));
 
 	value = DE_SELECT_ACTIVE | DE_CONTROL_NORMAL;
@@ -122,6 +126,8 @@ static int tegra_output_rgb_enable(struct tegra_output *output)
 	tegra_dc_writel(rgb->dc, GENERAL_ACT_REQ << 8, DC_CMD_STATE_CONTROL);
 	tegra_dc_writel(rgb->dc, GENERAL_ACT_REQ, DC_CMD_STATE_CONTROL);
 
+	rgb->enabled = true;
+
 	return 0;
 }
 
@@ -130,6 +136,9 @@ static int tegra_output_rgb_disable(struct tegra_output *output)
 	struct tegra_rgb *rgb = to_rgb(output);
 	unsigned long value;
 
+	if (!rgb->enabled)
+		return 0;
+
 	value = tegra_dc_readl(rgb->dc, DC_CMD_DISPLAY_POWER_CONTROL);
 	value &= ~(PW0_ENABLE | PW1_ENABLE | PW2_ENABLE | PW3_ENABLE |
 		   PW4_ENABLE | PM0_ENABLE | PM1_ENABLE);
@@ -144,6 +153,8 @@ static int tegra_output_rgb_disable(struct tegra_output *output)
 
 	tegra_dc_write_regs(rgb->dc, rgb_disable, ARRAY_SIZE(rgb_disable));
 
+	rgb->enabled = false;
+
 	return 0;
 }
 
-- 
1.8.5.3

Re: [PATCH 1/1] drm/tegra: Add guard to avoid double disable/enable of RGB outputs

[Erik Faye-Lund]

On Tue, Feb 11, 2014 at 6:12 PM, Dmitry Osipenko <digetx@gmail.com> wrote:
> Add guard to check whether RGB output is already enabled in the way it's
> done for HDMI output. Fixes possible hang on trying to disable output twice
> (first time during driver probe and second on fb registering).
>
> Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
> ---
>  drivers/gpu/drm/tegra/rgb.c | 11 +++++++++++
>  1 file changed, 11 insertions(+)
>
> diff --git a/drivers/gpu/drm/tegra/rgb.c b/drivers/gpu/drm/tegra/rgb.c
> index 338f7f6..0266fb4 100644
> --- a/drivers/gpu/drm/tegra/rgb.c
> +++ b/drivers/gpu/drm/tegra/rgb.c
> @@ -15,6 +15,7 @@
>  struct tegra_rgb {
>         struct tegra_output output;
>         struct tegra_dc *dc;
> +       bool enabled;
>
>         struct clk *clk_parent;
>         struct clk *clk;
> @@ -89,6 +90,9 @@ static int tegra_output_rgb_enable(struct tegra_output *output)
>         struct tegra_rgb *rgb = to_rgb(output);
>         unsigned long value;
>
> +       if (rgb->enabled)
> +               return 0;
> +
>         tegra_dc_write_regs(rgb->dc, rgb_enable, ARRAY_SIZE(rgb_enable));
>
>         value = DE_SELECT_ACTIVE | DE_CONTROL_NORMAL;
> @@ -122,6 +126,8 @@ static int tegra_output_rgb_enable(struct tegra_output *output)
>         tegra_dc_writel(rgb->dc, GENERAL_ACT_REQ << 8, DC_CMD_STATE_CONTROL);
>         tegra_dc_writel(rgb->dc, GENERAL_ACT_REQ, DC_CMD_STATE_CONTROL);
>
> +       rgb->enabled = true;
> +
>         return 0;
>  }
>
> @@ -130,6 +136,9 @@ static int tegra_output_rgb_disable(struct tegra_output *output)
>         struct tegra_rgb *rgb = to_rgb(output);
>         unsigned long value;
>
> +       if (!rgb->enabled)
> +               return 0;
> +
>         value = tegra_dc_readl(rgb->dc, DC_CMD_DISPLAY_POWER_CONTROL);
>         value &= ~(PW0_ENABLE | PW1_ENABLE | PW2_ENABLE | PW3_ENABLE |
>                    PW4_ENABLE | PM0_ENABLE | PM1_ENABLE);
> @@ -144,6 +153,8 @@ static int tegra_output_rgb_disable(struct tegra_output *output)
>
>         tegra_dc_write_regs(rgb->dc, rgb_disable, ARRAY_SIZE(rgb_disable));
>
> +       rgb->enabled = false;
> +
>         return 0;
>  }
>

Wouldn't it make more sense to make "enabled" and int that counts how
many times tegra_output_rgb_enable has been called? That way you can
have tegra_output_rgb_disable only really disable the display once the
same amount of disables have been performed...

Re: [PATCH 1/1] drm/tegra: Add guard to avoid double disable/enable of RGB outputs

[Dmitry Osipenko]

11.02.2014 23:13, Erik Faye-Lund пишет:
> On Tue, Feb 11, 2014 at 6:12 PM, Dmitry Osipenko <digetx@gmail.com> wrote:
>> Add guard to check whether RGB output is already enabled in the way it's
>> done for HDMI output. Fixes possible hang on trying to disable output twice
>> (first time during driver probe and second on fb registering).
>>
>> Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
>> ---
>>  drivers/gpu/drm/tegra/rgb.c | 11 +++++++++++
>>  1 file changed, 11 insertions(+)
>>
>> diff --git a/drivers/gpu/drm/tegra/rgb.c b/drivers/gpu/drm/tegra/rgb.c
>> index 338f7f6..0266fb4 100644
>> --- a/drivers/gpu/drm/tegra/rgb.c
>> +++ b/drivers/gpu/drm/tegra/rgb.c
>> @@ -15,6 +15,7 @@
>>  struct tegra_rgb {
>>         struct tegra_output output;
>>         struct tegra_dc *dc;
>> +       bool enabled;
>>
>>         struct clk *clk_parent;
>>         struct clk *clk;
>> @@ -89,6 +90,9 @@ static int tegra_output_rgb_enable(struct tegra_output *output)
>>         struct tegra_rgb *rgb = to_rgb(output);
>>         unsigned long value;
>>
>> +       if (rgb->enabled)
>> +               return 0;
>> +
>>         tegra_dc_write_regs(rgb->dc, rgb_enable, ARRAY_SIZE(rgb_enable));
>>
>>         value = DE_SELECT_ACTIVE | DE_CONTROL_NORMAL;
>> @@ -122,6 +126,8 @@ static int tegra_output_rgb_enable(struct tegra_output *output)
>>         tegra_dc_writel(rgb->dc, GENERAL_ACT_REQ << 8, DC_CMD_STATE_CONTROL);
>>         tegra_dc_writel(rgb->dc, GENERAL_ACT_REQ, DC_CMD_STATE_CONTROL);
>>
>> +       rgb->enabled = true;
>> +
>>         return 0;
>>  }
>>
>> @@ -130,6 +136,9 @@ static int tegra_output_rgb_disable(struct tegra_output *output)
>>         struct tegra_rgb *rgb = to_rgb(output);
>>         unsigned long value;
>>
>> +       if (!rgb->enabled)
>> +               return 0;
>> +
>>         value = tegra_dc_readl(rgb->dc, DC_CMD_DISPLAY_POWER_CONTROL);
>>         value &= ~(PW0_ENABLE | PW1_ENABLE | PW2_ENABLE | PW3_ENABLE |
>>                    PW4_ENABLE | PM0_ENABLE | PM1_ENABLE);
>> @@ -144,6 +153,8 @@ static int tegra_output_rgb_disable(struct tegra_output *output)
>>
>>         tegra_dc_write_regs(rgb->dc, rgb_disable, ARRAY_SIZE(rgb_disable));
>>
>> +       rgb->enabled = false;
>> +
>>         return 0;
>>  }
>>
> 
> Wouldn't it make more sense to make "enabled" and int that counts how
> many times tegra_output_rgb_enable has been called? That way you can
> have tegra_output_rgb_disable only really disable the display once the
> same amount of disables have been performed...
> 
RGB is a part of DC, i.e. one per DC. tegra_output_rgb_enable() would be called
for each individual DC, so it doesn't make any sense. Besides, I doubt tegra
device with two panels ever existed or will be :)

Re: [PATCH 1/1] drm/tegra: Add guard to avoid double disable/enable of RGB outputs

[Thierry Reding]

[-- Attachment #1: Type: text/plain, Size: 3165 bytes --]

On Tue, Feb 11, 2014 at 08:13:14PM +0100, Erik Faye-Lund wrote:
> On Tue, Feb 11, 2014 at 6:12 PM, Dmitry Osipenko <digetx@gmail.com> wrote:
> > Add guard to check whether RGB output is already enabled in the way it's
> > done for HDMI output. Fixes possible hang on trying to disable output twice
> > (first time during driver probe and second on fb registering).
> >
> > Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
> > ---
> >  drivers/gpu/drm/tegra/rgb.c | 11 +++++++++++
> >  1 file changed, 11 insertions(+)
> >
> > diff --git a/drivers/gpu/drm/tegra/rgb.c b/drivers/gpu/drm/tegra/rgb.c
> > index 338f7f6..0266fb4 100644
> > --- a/drivers/gpu/drm/tegra/rgb.c
> > +++ b/drivers/gpu/drm/tegra/rgb.c
> > @@ -15,6 +15,7 @@
> >  struct tegra_rgb {
> >         struct tegra_output output;
> >         struct tegra_dc *dc;
> > +       bool enabled;
> >
> >         struct clk *clk_parent;
> >         struct clk *clk;
> > @@ -89,6 +90,9 @@ static int tegra_output_rgb_enable(struct tegra_output *output)
> >         struct tegra_rgb *rgb = to_rgb(output);
> >         unsigned long value;
> >
> > +       if (rgb->enabled)
> > +               return 0;
> > +
> >         tegra_dc_write_regs(rgb->dc, rgb_enable, ARRAY_SIZE(rgb_enable));
> >
> >         value = DE_SELECT_ACTIVE | DE_CONTROL_NORMAL;
> > @@ -122,6 +126,8 @@ static int tegra_output_rgb_enable(struct tegra_output *output)
> >         tegra_dc_writel(rgb->dc, GENERAL_ACT_REQ << 8, DC_CMD_STATE_CONTROL);
> >         tegra_dc_writel(rgb->dc, GENERAL_ACT_REQ, DC_CMD_STATE_CONTROL);
> >
> > +       rgb->enabled = true;
> > +
> >         return 0;
> >  }
> >
> > @@ -130,6 +136,9 @@ static int tegra_output_rgb_disable(struct tegra_output *output)
> >         struct tegra_rgb *rgb = to_rgb(output);
> >         unsigned long value;
> >
> > +       if (!rgb->enabled)
> > +               return 0;
> > +
> >         value = tegra_dc_readl(rgb->dc, DC_CMD_DISPLAY_POWER_CONTROL);
> >         value &= ~(PW0_ENABLE | PW1_ENABLE | PW2_ENABLE | PW3_ENABLE |
> >                    PW4_ENABLE | PM0_ENABLE | PM1_ENABLE);
> > @@ -144,6 +153,8 @@ static int tegra_output_rgb_disable(struct tegra_output *output)
> >
> >         tegra_dc_write_regs(rgb->dc, rgb_disable, ARRAY_SIZE(rgb_disable));
> >
> > +       rgb->enabled = false;
> > +
> >         return 0;
> >  }
> >
> 
> Wouldn't it make more sense to make "enabled" and int that counts how
> many times tegra_output_rgb_enable has been called? That way you can
> have tegra_output_rgb_disable only really disable the display once the
> same amount of disables have been performed...

I don't think that will work. Calls to tegra_output_rgb_enable/disable()
and the .enable/disable for any other outputs aren't guaranteed to be
balanced. Therefore reference counting the enables and disables will
break things.

Thinking about it, maybe the current interface to outputs isn't the best
solution. It's simple and with the guards in place it works pretty well.
Perhaps if I can find some spare time I'll investigate whether something
a little cleaner can be done.

Thierry

[-- Attachment #2: Type: application/pgp-signature, Size: 836 bytes --]

Re: [PATCH 1/1] drm/tegra: Add guard to avoid double disable/enable of RGB outputs

[Thierry Reding]

[-- Attachment #1: Type: text/plain, Size: 468 bytes --]

On Tue, Feb 11, 2014 at 09:12:27PM +0400, Dmitry Osipenko wrote:
> Add guard to check whether RGB output is already enabled in the way it's
> done for HDMI output. Fixes possible hang on trying to disable output twice
> (first time during driver probe and second on fb registering).
> 
> Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
> ---
>  drivers/gpu/drm/tegra/rgb.c | 11 +++++++++++
>  1 file changed, 11 insertions(+)

Applied, thanks!

Thierry

[-- Attachment #2: Type: application/pgp-signature, Size: 836 bytes --]