I was trying to adjust the brightness-levels for the trogdor boards:
https://chromium-review.googlesource.com/c/chromiumos/third_party/kernel/+/2291209
Like on a lot of panels, trogdor's low end needs to be cropped,
and now that we have the interpolation stuff I wanted to make use of it
and bake in even the curve that's customary to have on chromebooks.
I found the current behavior of the pwm_bl driver a little unintuitive
and non-linear. See patch 1 for a suggested fix for this.
A few veyron dts files were relying on this (perhaps weird) behavior.
Those devices also want a minimum brightness like trogdor, so changed
them to use the new way.
Finally, given that trogdor's dts is part of linux-next now, add the
brightness-levels to it, since that's the original reason I was looking at
this.
Changes in v2:
- Fixed type promotion in the driver
- Removed "backlight: pwm_bl: Artificially add 0% during interpolation",
userspace works just fine without it because it already knows how to use
bl_power for turning off the display.
- Added brightness-levels to trogdor as well, now the dts is upstream.
Alexandru Stan (3):
backlight: pwm_bl: Fix interpolation
ARM: dts: rockchip: veyron: Remove 0 point from brightness-levels
arm64: dts: qcom: trogdor: Add brightness-levels
arch/arm/boot/dts/rk3288-veyron-jaq.dts | 2 +-
arch/arm/boot/dts/rk3288-veyron-minnie.dts | 2 +-
arch/arm/boot/dts/rk3288-veyron-tiger.dts | 2 +-
arch/arm64/boot/dts/qcom/sc7180-trogdor.dtsi | 9 +++
drivers/video/backlight/pwm_bl.c | 70 +++++++++-----------
5 files changed, 43 insertions(+), 42 deletions(-)
--
2.28.0
Whenever num-interpolated-steps was larger than the distance
between 2 consecutive brightness levels the table would get really
discontinuous. The slope of the interpolation would stick with
integers only and if it was 0 the whole line segment would get skipped.
Example settings:
brightness-levels = <0 1 2 4 8 16 32 64 128 256>;
num-interpolated-steps = <16>;
The distances between 1 2 4 and 8 would be 1, and only starting with 16
it would start to interpolate properly.
Let's change it so there's always interpolation happening, even if
there's no enough points available (read: values in the table would
appear more than once). This should match the expected behavior much
more closely.
Signed-off-by: Alexandru Stan <[email protected]>
---
drivers/video/backlight/pwm_bl.c | 70 ++++++++++++++------------------
1 file changed, 31 insertions(+), 39 deletions(-)
diff --git a/drivers/video/backlight/pwm_bl.c b/drivers/video/backlight/pwm_bl.c
index dfc760830eb9..3e77f6b73fd9 100644
--- a/drivers/video/backlight/pwm_bl.c
+++ b/drivers/video/backlight/pwm_bl.c
@@ -230,8 +230,7 @@ static int pwm_backlight_parse_dt(struct device *dev,
struct platform_pwm_backlight_data *data)
{
struct device_node *node = dev->of_node;
- unsigned int num_levels = 0;
- unsigned int levels_count;
+ unsigned int num_levels;
unsigned int num_steps = 0;
struct property *prop;
unsigned int *table;
@@ -260,12 +259,11 @@ static int pwm_backlight_parse_dt(struct device *dev,
if (!prop)
return 0;
- data->max_brightness = length / sizeof(u32);
+ num_levels = length / sizeof(u32);
/* read brightness levels from DT property */
- if (data->max_brightness > 0) {
- size_t size = sizeof(*data->levels) * data->max_brightness;
- unsigned int i, j, n = 0;
+ if (num_levels > 0) {
+ size_t size = sizeof(*data->levels) * num_levels;
data->levels = devm_kzalloc(dev, size, GFP_KERNEL);
if (!data->levels)
@@ -273,7 +271,7 @@ static int pwm_backlight_parse_dt(struct device *dev,
ret = of_property_read_u32_array(node, "brightness-levels",
data->levels,
- data->max_brightness);
+ num_levels);
if (ret < 0)
return ret;
@@ -298,7 +296,13 @@ static int pwm_backlight_parse_dt(struct device *dev,
* between two points.
*/
if (num_steps) {
- if (data->max_brightness < 2) {
+ unsigned int num_input_levels = num_levels;
+ unsigned int i;
+ u32 x1, x2, x, dx;
+ u32 y1, y2;
+ s64 dy;
+
+ if (num_input_levels < 2) {
dev_err(dev, "can't interpolate\n");
return -EINVAL;
}
@@ -308,14 +312,7 @@ static int pwm_backlight_parse_dt(struct device *dev,
* taking in consideration the number of interpolated
* steps between two levels.
*/
- for (i = 0; i < data->max_brightness - 1; i++) {
- if ((data->levels[i + 1] - data->levels[i]) /
- num_steps)
- num_levels += num_steps;
- else
- num_levels++;
- }
- num_levels++;
+ num_levels = (num_input_levels - 1) * num_steps + 1;
dev_dbg(dev, "new number of brightness levels: %d\n",
num_levels);
@@ -327,24 +324,25 @@ static int pwm_backlight_parse_dt(struct device *dev,
table = devm_kzalloc(dev, size, GFP_KERNEL);
if (!table)
return -ENOMEM;
-
- /* Fill the interpolated table. */
- levels_count = 0;
- for (i = 0; i < data->max_brightness - 1; i++) {
- value = data->levels[i];
- n = (data->levels[i + 1] - value) / num_steps;
- if (n > 0) {
- for (j = 0; j < num_steps; j++) {
- table[levels_count] = value;
- value += n;
- levels_count++;
- }
- } else {
- table[levels_count] = data->levels[i];
- levels_count++;
+ /*
+ * Fill the interpolated table[x] = y
+ * by draw lines between each (x1, y1) to (x2, y2).
+ */
+ dx = num_steps;
+ for (i = 0; i < num_input_levels - 1; i++) {
+ x1 = i * dx;
+ x2 = x1 + dx;
+ y1 = data->levels[i];
+ y2 = data->levels[i + 1];
+ dy = (s64)y2 - y1;
+
+ for (x = x1; x < x2; x++) {
+ table[x] = y1 +
+ div_s64(dy * ((s64)x - x1), dx);
}
}
- table[levels_count] = data->levels[i];
+ /* Fill in the last point, since no line starts here. */
+ table[x2] = y2;
/*
* As we use interpolation lets remove current
@@ -353,15 +351,9 @@ static int pwm_backlight_parse_dt(struct device *dev,
*/
devm_kfree(dev, data->levels);
data->levels = table;
-
- /*
- * Reassign max_brightness value to the new total number
- * of brightness levels.
- */
- data->max_brightness = num_levels;
}
- data->max_brightness--;
+ data->max_brightness = num_levels - 1;
}
return 0;
--
2.28.0
After the "PWM backlight interpolation adjustments" patches, the
backlight interpolation works a little differently. The way these
dts files were working before was relying on a bug (IMHO).
Remove the 0-3 range since otherwise we would have a 252 long
interpolation that would slowly go between 0 and 3, looking really bad
in userspace.
We don't need the 0% point, userspace seems to handle this just fine
because it uses the bl_power property to turn off the display.
Signed-off-by: Alexandru Stan <[email protected]>
---
arch/arm/boot/dts/rk3288-veyron-jaq.dts | 2 +-
arch/arm/boot/dts/rk3288-veyron-minnie.dts | 2 +-
arch/arm/boot/dts/rk3288-veyron-tiger.dts | 2 +-
3 files changed, 3 insertions(+), 3 deletions(-)
diff --git a/arch/arm/boot/dts/rk3288-veyron-jaq.dts b/arch/arm/boot/dts/rk3288-veyron-jaq.dts
index af77ab20586d..4a148cf1defc 100644
--- a/arch/arm/boot/dts/rk3288-veyron-jaq.dts
+++ b/arch/arm/boot/dts/rk3288-veyron-jaq.dts
@@ -20,7 +20,7 @@ / {
&backlight {
/* Jaq panel PWM must be >= 3%, so start non-zero brightness at 8 */
- brightness-levels = <0 8 255>;
+ brightness-levels = <8 255>;
num-interpolated-steps = <247>;
};
diff --git a/arch/arm/boot/dts/rk3288-veyron-minnie.dts b/arch/arm/boot/dts/rk3288-veyron-minnie.dts
index f8b69e0a16a0..82fc6fba9999 100644
--- a/arch/arm/boot/dts/rk3288-veyron-minnie.dts
+++ b/arch/arm/boot/dts/rk3288-veyron-minnie.dts
@@ -39,7 +39,7 @@ volum_up {
&backlight {
/* Minnie panel PWM must be >= 1%, so start non-zero brightness at 3 */
- brightness-levels = <0 3 255>;
+ brightness-levels = <3 255>;
num-interpolated-steps = <252>;
};
diff --git a/arch/arm/boot/dts/rk3288-veyron-tiger.dts b/arch/arm/boot/dts/rk3288-veyron-tiger.dts
index 069f0c2c1fdf..52a84cbe7a90 100644
--- a/arch/arm/boot/dts/rk3288-veyron-tiger.dts
+++ b/arch/arm/boot/dts/rk3288-veyron-tiger.dts
@@ -23,7 +23,7 @@ / {
&backlight {
/* Tiger panel PWM must be >= 1%, so start non-zero brightness at 3 */
- brightness-levels = <0 3 255>;
+ brightness-levels = <3 255>;
num-interpolated-steps = <252>;
};
--
2.28.0
Now that we have better interpolation for the backlight
("backlight: pwm_bl: Fix interpolation"), we can now add the curve to
the trogdor boards, being careful to crop the low end.
Signed-off-by: Alexandru Stan <[email protected]>
---
arch/arm64/boot/dts/qcom/sc7180-trogdor.dtsi | 9 +++++++++
1 file changed, 9 insertions(+)
diff --git a/arch/arm64/boot/dts/qcom/sc7180-trogdor.dtsi b/arch/arm64/boot/dts/qcom/sc7180-trogdor.dtsi
index bf875589d364..ccdabc6c4994 100644
--- a/arch/arm64/boot/dts/qcom/sc7180-trogdor.dtsi
+++ b/arch/arm64/boot/dts/qcom/sc7180-trogdor.dtsi
@@ -179,6 +179,15 @@ pp3300_fp_tp: pp3300-fp-tp-regulator {
backlight: backlight {
compatible = "pwm-backlight";
+ /* The panels don't seem to like anything below ~ 5% */
+ brightness-levels = <
+ 196 256 324 400 484 576 676 784 900 1024 1156 1296
+ 1444 1600 1764 1936 2116 2304 2500 2704 2916 3136
+ 3364 3600 3844 4096
+ >;
+ num-interpolated-steps = <64>;
+ default-brightness-level = <951>;
+
pwms = <&cros_ec_pwm 1>;
enable-gpios = <&tlmm 12 GPIO_ACTIVE_HIGH>;
power-supply = <&ppvar_sys>;
--
2.28.0
Hi,
On Tue, Oct 13, 2020 at 1:01 AM Alexandru Stan <[email protected]> wrote:
>
> Now that we have better interpolation for the backlight
> ("backlight: pwm_bl: Fix interpolation"), we can now add the curve to
> the trogdor boards, being careful to crop the low end.
Just to make it clear, the patch this depends on hasn't landed yet.
Presumably it will land in the v5.10 timeframe? That means that
without extra coordination this patch can target v5.11.
> Signed-off-by: Alexandru Stan <[email protected]>
> ---
>
> arch/arm64/boot/dts/qcom/sc7180-trogdor.dtsi | 9 +++++++++
> 1 file changed, 9 insertions(+)
>
> diff --git a/arch/arm64/boot/dts/qcom/sc7180-trogdor.dtsi b/arch/arm64/boot/dts/qcom/sc7180-trogdor.dtsi
> index bf875589d364..ccdabc6c4994 100644
> --- a/arch/arm64/boot/dts/qcom/sc7180-trogdor.dtsi
> +++ b/arch/arm64/boot/dts/qcom/sc7180-trogdor.dtsi
> @@ -179,6 +179,15 @@ pp3300_fp_tp: pp3300-fp-tp-regulator {
> backlight: backlight {
> compatible = "pwm-backlight";
>
> + /* The panels don't seem to like anything below ~ 5% */
> + brightness-levels = <
> + 196 256 324 400 484 576 676 784 900 1024 1156 1296
> + 1444 1600 1764 1936 2116 2304 2500 2704 2916 3136
> + 3364 3600 3844 4096
> + >;
> + num-interpolated-steps = <64>;
> + default-brightness-level = <951>;
I haven't done lots of digging here, but this matches what Alexandru
and Matthias agreed upon for the downstream tree and seems sane.
Thus:
Reviewed-by: Douglas Anderson <[email protected]>
On Tue, Oct 13, 2020 at 09:28:38AM -0700, Doug Anderson wrote:
> Hi,
>
> On Tue, Oct 13, 2020 at 1:01 AM Alexandru Stan <[email protected]> wrote:
> >
> > Now that we have better interpolation for the backlight
> > ("backlight: pwm_bl: Fix interpolation"), we can now add the curve to
> > the trogdor boards, being careful to crop the low end.
>
> Just to make it clear, the patch this depends on hasn't landed yet.
> Presumably it will land in the v5.10 timeframe? That means that
> without extra coordination this patch can target v5.11.
You're talking about patch 1 from this set? Despite the title I view
the patch as changing policy (albeit one that does also fix some annoying
quantization errors at the same time) so it's not necessarily a
candidate for merging outside the merge window (I've not checked with
Lee but I think it likely the shutter is already down for features).
Moreover I'm not clear why there a dependency here that would stop the
changes landing in different trees.
Daniel.
> > Signed-off-by: Alexandru Stan <[email protected]>
> > ---
> >
> > arch/arm64/boot/dts/qcom/sc7180-trogdor.dtsi | 9 +++++++++
> > 1 file changed, 9 insertions(+)
> >
> > diff --git a/arch/arm64/boot/dts/qcom/sc7180-trogdor.dtsi b/arch/arm64/boot/dts/qcom/sc7180-trogdor.dtsi
> > index bf875589d364..ccdabc6c4994 100644
> > --- a/arch/arm64/boot/dts/qcom/sc7180-trogdor.dtsi
> > +++ b/arch/arm64/boot/dts/qcom/sc7180-trogdor.dtsi
> > @@ -179,6 +179,15 @@ pp3300_fp_tp: pp3300-fp-tp-regulator {
> > backlight: backlight {
> > compatible = "pwm-backlight";
> >
> > + /* The panels don't seem to like anything below ~ 5% */
> > + brightness-levels = <
> > + 196 256 324 400 484 576 676 784 900 1024 1156 1296
> > + 1444 1600 1764 1936 2116 2304 2500 2704 2916 3136
> > + 3364 3600 3844 4096
> > + >;
> > + num-interpolated-steps = <64>;
> > + default-brightness-level = <951>;
>
> I haven't done lots of digging here, but this matches what Alexandru
> and Matthias agreed upon for the downstream tree and seems sane.
> Thus:
>
> Reviewed-by: Douglas Anderson <[email protected]>
On Tue, Oct 13, 2020 at 01:01:02AM -0700, Alexandru Stan wrote:
> After the "PWM backlight interpolation adjustments" patches, the
> backlight interpolation works a little differently. The way these
> dts files were working before was relying on a bug (IMHO).
>
> Remove the 0-3 range since otherwise we would have a 252 long
> interpolation that would slowly go between 0 and 3, looking really bad
> in userspace.
>
> We don't need the 0% point, userspace seems to handle this just fine
> because it uses the bl_power property to turn off the display.
>
> Signed-off-by: Alexandru Stan <[email protected]>
Acked-by: Daniel Thompson <[email protected]>
Note also shouldn't this be patch 1 of the set. AFAICT it makes sense
whether or not the interpolation algorithm is changed.
Daniel.
> ---
>
> arch/arm/boot/dts/rk3288-veyron-jaq.dts | 2 +-
> arch/arm/boot/dts/rk3288-veyron-minnie.dts | 2 +-
> arch/arm/boot/dts/rk3288-veyron-tiger.dts | 2 +-
> 3 files changed, 3 insertions(+), 3 deletions(-)
>
> diff --git a/arch/arm/boot/dts/rk3288-veyron-jaq.dts b/arch/arm/boot/dts/rk3288-veyron-jaq.dts
> index af77ab20586d..4a148cf1defc 100644
> --- a/arch/arm/boot/dts/rk3288-veyron-jaq.dts
> +++ b/arch/arm/boot/dts/rk3288-veyron-jaq.dts
> @@ -20,7 +20,7 @@ / {
>
> &backlight {
> /* Jaq panel PWM must be >= 3%, so start non-zero brightness at 8 */
> - brightness-levels = <0 8 255>;
> + brightness-levels = <8 255>;
> num-interpolated-steps = <247>;
> };
>
> diff --git a/arch/arm/boot/dts/rk3288-veyron-minnie.dts b/arch/arm/boot/dts/rk3288-veyron-minnie.dts
> index f8b69e0a16a0..82fc6fba9999 100644
> --- a/arch/arm/boot/dts/rk3288-veyron-minnie.dts
> +++ b/arch/arm/boot/dts/rk3288-veyron-minnie.dts
> @@ -39,7 +39,7 @@ volum_up {
>
> &backlight {
> /* Minnie panel PWM must be >= 1%, so start non-zero brightness at 3 */
> - brightness-levels = <0 3 255>;
> + brightness-levels = <3 255>;
> num-interpolated-steps = <252>;
> };
>
> diff --git a/arch/arm/boot/dts/rk3288-veyron-tiger.dts b/arch/arm/boot/dts/rk3288-veyron-tiger.dts
> index 069f0c2c1fdf..52a84cbe7a90 100644
> --- a/arch/arm/boot/dts/rk3288-veyron-tiger.dts
> +++ b/arch/arm/boot/dts/rk3288-veyron-tiger.dts
> @@ -23,7 +23,7 @@ / {
>
> &backlight {
> /* Tiger panel PWM must be >= 1%, so start non-zero brightness at 3 */
> - brightness-levels = <0 3 255>;
> + brightness-levels = <3 255>;
> num-interpolated-steps = <252>;
> };
>
> --
> 2.28.0
>
On Tue, Oct 13, 2020 at 01:01:01AM -0700, Alexandru Stan wrote:
> Whenever num-interpolated-steps was larger than the distance
> between 2 consecutive brightness levels the table would get really
> discontinuous. The slope of the interpolation would stick with
> integers only and if it was 0 the whole line segment would get skipped.
>
> Example settings:
> brightness-levels = <0 1 2 4 8 16 32 64 128 256>;
> num-interpolated-steps = <16>;
>
> The distances between 1 2 4 and 8 would be 1, and only starting with 16
> it would start to interpolate properly.
Both comments a perilously close to nitpicking but enough that I wanted
to reply...
I'd suggest that the current behaviour as having two properties.
1. It was designed to generate strictly increasing tables (no repeated
values).
2. It's implementation contains quantization errors when calculating the
step size. This results in both the discards of some interpolated
steps you mentioned (it is possible to insert extra steps between 4
and 8 whilst retaining a strictly increasing table). It also
results in a potentially large undershoot when multiplying a step
size (64 interpolated steps and a gap of 127 is likely to get a visual
jump as we hop through 63 physical steps in one go).
#1 can is a policy that can be changed. #2 is a bug that could be fixed.
To be clear I don't object to generating a monotonically increasing
table but I'd prefer the policy change to be explicitly described in
the description.
> Let's change it so there's always interpolation happening, even if
> there's no enough points available (read: values in the table would
> appear more than once). This should match the expected behavior much
> more closely.
>
> Signed-off-by: Alexandru Stan <[email protected]>
> ---
>
> drivers/video/backlight/pwm_bl.c | 70 ++++++++++++++------------------
> 1 file changed, 31 insertions(+), 39 deletions(-)
>
> diff --git a/drivers/video/backlight/pwm_bl.c b/drivers/video/backlight/pwm_bl.c
> index dfc760830eb9..3e77f6b73fd9 100644
> --- a/drivers/video/backlight/pwm_bl.c
> +++ b/drivers/video/backlight/pwm_bl.c
> @@ -230,8 +230,7 @@ static int pwm_backlight_parse_dt(struct device *dev,
> struct platform_pwm_backlight_data *data)
> {
> struct device_node *node = dev->of_node;
> - unsigned int num_levels = 0;
> - unsigned int levels_count;
> + unsigned int num_levels;
> unsigned int num_steps = 0;
> struct property *prop;
> unsigned int *table;
> @@ -260,12 +259,11 @@ static int pwm_backlight_parse_dt(struct device *dev,
> if (!prop)
> return 0;
>
> - data->max_brightness = length / sizeof(u32);
> + num_levels = length / sizeof(u32);
>
> /* read brightness levels from DT property */
> - if (data->max_brightness > 0) {
> - size_t size = sizeof(*data->levels) * data->max_brightness;
> - unsigned int i, j, n = 0;
> + if (num_levels > 0) {
> + size_t size = sizeof(*data->levels) * num_levels;
>
> data->levels = devm_kzalloc(dev, size, GFP_KERNEL);
> if (!data->levels)
> @@ -273,7 +271,7 @@ static int pwm_backlight_parse_dt(struct device *dev,
>
> ret = of_property_read_u32_array(node, "brightness-levels",
> data->levels,
> - data->max_brightness);
> + num_levels);
> if (ret < 0)
> return ret;
>
> @@ -298,7 +296,13 @@ static int pwm_backlight_parse_dt(struct device *dev,
> * between two points.
> */
> if (num_steps) {
> - if (data->max_brightness < 2) {
> + unsigned int num_input_levels = num_levels;
> + unsigned int i;
> + u32 x1, x2, x, dx;
> + u32 y1, y2;
> + s64 dy;
> +
> + if (num_input_levels < 2) {
> dev_err(dev, "can't interpolate\n");
> return -EINVAL;
> }
> @@ -308,14 +312,7 @@ static int pwm_backlight_parse_dt(struct device *dev,
> * taking in consideration the number of interpolated
> * steps between two levels.
> */
> - for (i = 0; i < data->max_brightness - 1; i++) {
> - if ((data->levels[i + 1] - data->levels[i]) /
> - num_steps)
> - num_levels += num_steps;
> - else
> - num_levels++;
> - }
> - num_levels++;
> + num_levels = (num_input_levels - 1) * num_steps + 1;
> dev_dbg(dev, "new number of brightness levels: %d\n",
> num_levels);
>
> @@ -327,24 +324,25 @@ static int pwm_backlight_parse_dt(struct device *dev,
> table = devm_kzalloc(dev, size, GFP_KERNEL);
> if (!table)
> return -ENOMEM;
> -
> - /* Fill the interpolated table. */
> - levels_count = 0;
> - for (i = 0; i < data->max_brightness - 1; i++) {
> - value = data->levels[i];
> - n = (data->levels[i + 1] - value) / num_steps;
> - if (n > 0) {
> - for (j = 0; j < num_steps; j++) {
> - table[levels_count] = value;
> - value += n;
> - levels_count++;
> - }
> - } else {
> - table[levels_count] = data->levels[i];
> - levels_count++;
> + /*
> + * Fill the interpolated table[x] = y
> + * by draw lines between each (x1, y1) to (x2, y2).
> + */
> + dx = num_steps;
> + for (i = 0; i < num_input_levels - 1; i++) {
> + x1 = i * dx;
> + x2 = x1 + dx;
> + y1 = data->levels[i];
> + y2 = data->levels[i + 1];
> + dy = (s64)y2 - y1;
> +
> + for (x = x1; x < x2; x++) {
> + table[x] = y1 +
> + div_s64(dy * ((s64)x - x1), dx);
I don't think it is possible for x - x1 to be negative (e.g. what is the
s64 for). Obviously it makes little functional difference whether the
cast is there or not but I don't like fixed point code that has been
written with "just in case" casts.
Daniel.
> }
> }
> - table[levels_count] = data->levels[i];
> + /* Fill in the last point, since no line starts here. */
> + table[x2] = y2;
>
> /*
> * As we use interpolation lets remove current
> @@ -353,15 +351,9 @@ static int pwm_backlight_parse_dt(struct device *dev,
> */
> devm_kfree(dev, data->levels);
> data->levels = table;
> -
> - /*
> - * Reassign max_brightness value to the new total number
> - * of brightness levels.
> - */
> - data->max_brightness = num_levels;
> }
>
> - data->max_brightness--;
> + data->max_brightness = num_levels - 1;
> }
>
> return 0;
> --
> 2.28.0
>
Hi,
On Wed, Oct 14, 2020 at 4:33 AM Daniel Thompson
<[email protected]> wrote:
>
> On Tue, Oct 13, 2020 at 09:28:38AM -0700, Doug Anderson wrote:
> > Hi,
> >
> > On Tue, Oct 13, 2020 at 1:01 AM Alexandru Stan <[email protected]> wrote:
> > >
> > > Now that we have better interpolation for the backlight
> > > ("backlight: pwm_bl: Fix interpolation"), we can now add the curve to
> > > the trogdor boards, being careful to crop the low end.
> >
> > Just to make it clear, the patch this depends on hasn't landed yet.
> > Presumably it will land in the v5.10 timeframe? That means that
> > without extra coordination this patch can target v5.11.
>
> You're talking about patch 1 from this set? Despite the title I view
> the patch as changing policy (albeit one that does also fix some annoying
> quantization errors at the same time) so it's not necessarily a
> candidate for merging outside the merge window (I've not checked with
> Lee but I think it likely the shutter is already down for features).
Ugh, I'm off by one. :( Right. New features prepared for v5.10
should already have been baking in linuxnext and merge requests have
already started flowing towards Linus. After v5.10-rc1 then it'd just
fixes and this doesn't really qualify. So the timing dictates that
patch #1 will land in v5.11, not v5.10.
> Moreover I'm not clear why there a dependency here that would stop the
> changes landing in different trees.
I haven't tried Alexandru's device tree patch without the associated
code changes, but I guess I just assumed that it would make a really
ugly (non)ideal backlight curve and we'd be better off with what we
currently have (AKA no curve specified at all).
Hrm, thinking about it, I guess the worst case is a slightly non-ideal
backlight curve and it would be all good in the final v5.11 which
would have the device tree and code changes, so you're right that both
the code and device tree could target v5.11 without anything too
bad...
> Daniel.
>
>
> > > Signed-off-by: Alexandru Stan <[email protected]>
> > > ---
> > >
> > > arch/arm64/boot/dts/qcom/sc7180-trogdor.dtsi | 9 +++++++++
> > > 1 file changed, 9 insertions(+)
> > >
> > > diff --git a/arch/arm64/boot/dts/qcom/sc7180-trogdor.dtsi b/arch/arm64/boot/dts/qcom/sc7180-trogdor.dtsi
> > > index bf875589d364..ccdabc6c4994 100644
> > > --- a/arch/arm64/boot/dts/qcom/sc7180-trogdor.dtsi
> > > +++ b/arch/arm64/boot/dts/qcom/sc7180-trogdor.dtsi
> > > @@ -179,6 +179,15 @@ pp3300_fp_tp: pp3300-fp-tp-regulator {
> > > backlight: backlight {
> > > compatible = "pwm-backlight";
> > >
> > > + /* The panels don't seem to like anything below ~ 5% */
> > > + brightness-levels = <
> > > + 196 256 324 400 484 576 676 784 900 1024 1156 1296
> > > + 1444 1600 1764 1936 2116 2304 2500 2704 2916 3136
> > > + 3364 3600 3844 4096
> > > + >;
> > > + num-interpolated-steps = <64>;
> > > + default-brightness-level = <951>;
> >
> > I haven't done lots of digging here, but this matches what Alexandru
> > and Matthias agreed upon for the downstream tree and seems sane.
> > Thus:
> >
> > Reviewed-by: Douglas Anderson <[email protected]>
On Wed, Oct 14, 2020 at 06:51:19AM -0700, Doug Anderson wrote:
> Hi,
>
> On Wed, Oct 14, 2020 at 4:33 AM Daniel Thompson
> <[email protected]> wrote:
> >
> > On Tue, Oct 13, 2020 at 09:28:38AM -0700, Doug Anderson wrote:
> > > Hi,
> > >
> > > On Tue, Oct 13, 2020 at 1:01 AM Alexandru Stan <[email protected]> wrote:
> > > >
> > > > Now that we have better interpolation for the backlight
> > > > ("backlight: pwm_bl: Fix interpolation"), we can now add the curve to
> > > > the trogdor boards, being careful to crop the low end.
> > >
> > > Just to make it clear, the patch this depends on hasn't landed yet.
> > > Presumably it will land in the v5.10 timeframe? That means that
> > > without extra coordination this patch can target v5.11.
> >
> > You're talking about patch 1 from this set? Despite the title I view
> > the patch as changing policy (albeit one that does also fix some annoying
> > quantization errors at the same time) so it's not necessarily a
> > candidate for merging outside the merge window (I've not checked with
> > Lee but I think it likely the shutter is already down for features).
>
> Ugh, I'm off by one. :( Right. New features prepared for v5.10
> should already have been baking in linuxnext and merge requests have
> already started flowing towards Linus. After v5.10-rc1 then it'd just
> fixes and this doesn't really qualify. So the timing dictates that
> patch #1 will land in v5.11, not v5.10.
>
>
> > Moreover I'm not clear why there a dependency here that would stop the
> > changes landing in different trees.
>
> I haven't tried Alexandru's device tree patch without the associated
> code changes, but I guess I just assumed that it would make a really
> ugly (non)ideal backlight curve and we'd be better off with what we
> currently have (AKA no curve specified at all).
>
> Hrm, thinking about it, I guess the worst case is a slightly non-ideal
> backlight curve and it would be all good in the final v5.11 which
> would have the device tree and code changes, so you're right that both
> the code and device tree could target v5.11 without anything too
> bad...
Excellent. I'll try to remember this for v3 so I can get my Acked-by
versus Reviewed-by tags correct ;-) .
Daniel.
>
>
> > Daniel.
> >
> >
> > > > Signed-off-by: Alexandru Stan <[email protected]>
> > > > ---
> > > >
> > > > arch/arm64/boot/dts/qcom/sc7180-trogdor.dtsi | 9 +++++++++
> > > > 1 file changed, 9 insertions(+)
> > > >
> > > > diff --git a/arch/arm64/boot/dts/qcom/sc7180-trogdor.dtsi b/arch/arm64/boot/dts/qcom/sc7180-trogdor.dtsi
> > > > index bf875589d364..ccdabc6c4994 100644
> > > > --- a/arch/arm64/boot/dts/qcom/sc7180-trogdor.dtsi
> > > > +++ b/arch/arm64/boot/dts/qcom/sc7180-trogdor.dtsi
> > > > @@ -179,6 +179,15 @@ pp3300_fp_tp: pp3300-fp-tp-regulator {
> > > > backlight: backlight {
> > > > compatible = "pwm-backlight";
> > > >
> > > > + /* The panels don't seem to like anything below ~ 5% */
> > > > + brightness-levels = <
> > > > + 196 256 324 400 484 576 676 784 900 1024 1156 1296
> > > > + 1444 1600 1764 1936 2116 2304 2500 2704 2916 3136
> > > > + 3364 3600 3844 4096
> > > > + >;
> > > > + num-interpolated-steps = <64>;
> > > > + default-brightness-level = <951>;
> > >
> > > I haven't done lots of digging here, but this matches what Alexandru
> > > and Matthias agreed upon for the downstream tree and seems sane.
> > > Thus:
> > >
> > > Reviewed-by: Douglas Anderson <[email protected]>
Hi Alexandru,
On Tue, Oct 13, 2020 at 1:57 PM Alexandru Stan <[email protected]> wrote:
> Whenever num-interpolated-steps was larger than the distance
> between 2 consecutive brightness levels the table would get really
> discontinuous. The slope of the interpolation would stick with
> integers only and if it was 0 the whole line segment would get skipped.
>
> Example settings:
> brightness-levels = <0 1 2 4 8 16 32 64 128 256>;
> num-interpolated-steps = <16>;
>
> The distances between 1 2 4 and 8 would be 1, and only starting with 16
> it would start to interpolate properly.
>
> Let's change it so there's always interpolation happening, even if
> there's no enough points available (read: values in the table would
> appear more than once). This should match the expected behavior much
> more closely.
>
> Signed-off-by: Alexandru Stan <[email protected]>
Thanks for your patch!
> --- a/drivers/video/backlight/pwm_bl.c
> +++ b/drivers/video/backlight/pwm_bl.c
> @@ -327,24 +324,25 @@ static int pwm_backlight_parse_dt(struct device *dev,
> table = devm_kzalloc(dev, size, GFP_KERNEL);
> if (!table)
> return -ENOMEM;
> -
> - /* Fill the interpolated table. */
> - levels_count = 0;
> - for (i = 0; i < data->max_brightness - 1; i++) {
> - value = data->levels[i];
> - n = (data->levels[i + 1] - value) / num_steps;
> - if (n > 0) {
> - for (j = 0; j < num_steps; j++) {
> - table[levels_count] = value;
> - value += n;
> - levels_count++;
> - }
> - } else {
> - table[levels_count] = data->levels[i];
> - levels_count++;
> + /*
> + * Fill the interpolated table[x] = y
> + * by draw lines between each (x1, y1) to (x2, y2).
> + */
> + dx = num_steps;
> + for (i = 0; i < num_input_levels - 1; i++) {
> + x1 = i * dx;
> + x2 = x1 + dx;
> + y1 = data->levels[i];
> + y2 = data->levels[i + 1];
> + dy = (s64)y2 - y1;
> +
> + for (x = x1; x < x2; x++) {
> + table[x] = y1 +
> + div_s64(dy * ((s64)x - x1), dx);
Yummy, 64-by-32 divisions.
Shouldn't this use a rounded division?
Nevertheless, I think it would be worthwhile to implement this using
a (modified) Bresenham algorithm, avoiding multiplications and
divisions, and possibly increasing accuracy as well.
https://en.wikipedia.org/wiki/Bresenham%27s_line_algorithm
> }
> }
> - table[levels_count] = data->levels[i];
> + /* Fill in the last point, since no line starts here. */
> + table[x2] = y2;
>
> /*
> * As we use interpolation lets remove current
Gr{oetje,eeting}s,
Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds
On Wed, Oct 14, 2020 at 7:19 AM Daniel Thompson
<[email protected]> wrote:
>
> On Tue, Oct 13, 2020 at 01:01:02AM -0700, Alexandru Stan wrote:
> > After the "PWM backlight interpolation adjustments" patches, the
> > backlight interpolation works a little differently. The way these
> > dts files were working before was relying on a bug (IMHO).
> >
> > Remove the 0-3 range since otherwise we would have a 252 long
> > interpolation that would slowly go between 0 and 3, looking really bad
> > in userspace.
> >
> > We don't need the 0% point, userspace seems to handle this just fine
> > because it uses the bl_power property to turn off the display.
> >
> > Signed-off-by: Alexandru Stan <[email protected]>
>
> Acked-by: Daniel Thompson <[email protected]>
Thank you!
>
> Note also shouldn't this be patch 1 of the set. AFAICT it makes sense
> whether or not the interpolation algorithm is changed.
Yeah, I guess it could be. Sorry I didn't think of it that way before,
I'm used to landing things in a group.
In particular on veyron I assume it will almost be a noop without
having my driver patch (especially with the findings of 0% not being
that important).
Feel free to land this independently.
>
>
> Daniel.
Alexandru Stan (amstan)
On Wed, 14 Oct 2020 at 23:55, Geert Uytterhoeven <[email protected]> wrote:
>
> Hi Alexandru,
>
> On Tue, Oct 13, 2020 at 1:57 PM Alexandru Stan <[email protected]> wrote:
> > Whenever num-interpolated-steps was larger than the distance
> > between 2 consecutive brightness levels the table would get really
> > discontinuous. The slope of the interpolation would stick with
> > integers only and if it was 0 the whole line segment would get skipped.
> >
> > Example settings:
> > brightness-levels = <0 1 2 4 8 16 32 64 128 256>;
> > num-interpolated-steps = <16>;
> >
> > The distances between 1 2 4 and 8 would be 1, and only starting with 16
> > it would start to interpolate properly.
> >
> > Let's change it so there's always interpolation happening, even if
> > there's no enough points available (read: values in the table would
> > appear more than once). This should match the expected behavior much
> > more closely.
> >
> > Signed-off-by: Alexandru Stan <[email protected]>
>
> Thanks for your patch!
Thanks for your reply!
I'm sorry I haven't replied earlier. Looks like your reply was marked as spam.
Rest be assured my spam filter has been disciplined! :D
>
> > --- a/drivers/video/backlight/pwm_bl.c
> > +++ b/drivers/video/backlight/pwm_bl.c
> > @@ -327,24 +324,25 @@ static int pwm_backlight_parse_dt(struct device *dev,
> > table = devm_kzalloc(dev, size, GFP_KERNEL);
> > if (!table)
> > return -ENOMEM;
> > -
> > - /* Fill the interpolated table. */
> > - levels_count = 0;
> > - for (i = 0; i < data->max_brightness - 1; i++) {
> > - value = data->levels[i];
> > - n = (data->levels[i + 1] - value) / num_steps;
> > - if (n > 0) {
> > - for (j = 0; j < num_steps; j++) {
> > - table[levels_count] = value;
> > - value += n;
> > - levels_count++;
> > - }
> > - } else {
> > - table[levels_count] = data->levels[i];
> > - levels_count++;
> > + /*
> > + * Fill the interpolated table[x] = y
> > + * by draw lines between each (x1, y1) to (x2, y2).
> > + */
> > + dx = num_steps;
> > + for (i = 0; i < num_input_levels - 1; i++) {
> > + x1 = i * dx;
> > + x2 = x1 + dx;
> > + y1 = data->levels[i];
> > + y2 = data->levels[i + 1];
> > + dy = (s64)y2 - y1;
> > +
> > + for (x = x1; x < x2; x++) {
> > + table[x] = y1 +
> > + div_s64(dy * ((s64)x - x1), dx);
>
> Yummy, 64-by-32 divisions.
> Shouldn't this use a rounded division?
It won't hurt. But it really doesn't make much of a difference either way.
>
> Nevertheless, I think it would be worthwhile to implement this using
> a (modified) Bresenham algorithm, avoiding multiplications and
> divisions, and possibly increasing accuracy as well.
>
> https://en.wikipedia.org/wiki/Bresenham%27s_line_algorithm
Sure, it might be a little faster to use Bresenham's line algorithm.
Looks like to implement it I would have to deal with some fixed point
math and still have to do divisions occasionally.
I don't think performance is critical here, the values get calculated
only once when the driver loads, and the algorithm's accuracy
improvements might be at most 1 LSB.
Meanwhile the formula I already implemented is almost the same as the
formulas found at
https://en.wikipedia.org/wiki/Linear_interpolation#:~:text=gives
I would like to keep it as is, as straightforward as possible.
>
> > }
> > }
> > - table[levels_count] = data->levels[i];
> > + /* Fill in the last point, since no line starts here. */
> > + table[x2] = y2;
> >
> > /*
> > * As we use interpolation lets remove current
>
> Gr{oetje,eeting}s,
>
> Geert
>
> --
> Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]
>
> In personal conversations with technical people, I call myself a hacker. But
> when I'm talking to journalists I just say "programmer" or something like that.
> -- Linus Torvalds
Alexandru Stan (amstan)