From: "Sharma, Shashank" <shashank.sharma@intel.com>
To: Lionel Landwerlin <lionel.g.landwerlin@intel.com>,
intel-gfx@lists.freedesktop.org, "Bradford,
Robert" <robert.bradford@intel.com>,
"Bish, Jim" <jim.bish@intel.com>
Cc: daniel.vetter@intel.com, kausalmalladi@gmail.com
Subject: Re: [PATCH v8 15/25] drm/i915: CHV: Pipe level degamma correction
Date: Thu, 17 Dec 2015 17:16:23 +0530 [thread overview]
Message-ID: <5672A08F.7030502@intel.com> (raw)
In-Reply-To: <56718C11.1040306@intel.com>
Thanks for the testing and comments Lionel.
My comments inline.
Regards
Shashank
On 12/16/2015 9:36 PM, Lionel Landwerlin wrote:
> I gave a try to this patch on Braswell.
> A couple of comments below.
>
> On 03/12/15 11:36, Shashank Sharma wrote:
>> CHV/BSW supports Degamma color correction, which linearizes all
>> the non-linear color values. This will be applied before Color
>> Transformation.
>>
>> This patch does the following:
>> 1. Attach deGamma property to CRTC
>> 2. Add the core function to program DeGamma correction values for
>> CHV/BSW platform
>> 2. Add DeGamma correction macros/defines
>>
>> Signed-off-by: Shashank Sharma <shashank.sharma@intel.com>
>> Signed-off-by: Kausal Malladi <kausalmalladi@gmail.com>
>> ---
>> drivers/gpu/drm/i915/i915_reg.h | 6 ++
>> drivers/gpu/drm/i915/intel_color_manager.c | 95
>> ++++++++++++++++++++++++++++++
>> drivers/gpu/drm/i915/intel_color_manager.h | 5 ++
>> 3 files changed, 106 insertions(+)
>>
>> diff --git a/drivers/gpu/drm/i915/i915_reg.h
>> b/drivers/gpu/drm/i915/i915_reg.h
>> index 6259537..665e23e 100644
>> --- a/drivers/gpu/drm/i915/i915_reg.h
>> +++ b/drivers/gpu/drm/i915/i915_reg.h
>> @@ -8166,4 +8166,10 @@ enum skl_disp_power_wells {
>> #define _PIPE_GAMMA_BASE(pipe) \
>> (_PIPE3(pipe, PIPEA_CGM_GAMMA, PIPEB_CGM_GAMMA, PIPEC_CGM_GAMMA))
>> +#define PIPEA_CGM_DEGAMMA (VLV_DISPLAY_BASE +
>> 0x66000)
>> +#define PIPEB_CGM_DEGAMMA (VLV_DISPLAY_BASE +
>> 0x68000)
>> +#define PIPEC_CGM_DEGAMMA (VLV_DISPLAY_BASE +
>> 0x6A000)
>> +#define _PIPE_DEGAMMA_BASE(pipe) \
>> + (_PIPE3(pipe, PIPEA_CGM_DEGAMMA, PIPEB_CGM_DEGAMMA,
>> PIPEC_CGM_DEGAMMA))
>> +
>> #endif /* _I915_REG_H_ */
>> diff --git a/drivers/gpu/drm/i915/intel_color_manager.c
>> b/drivers/gpu/drm/i915/intel_color_manager.c
>> index f4334c0..aa89a55 100644
>> --- a/drivers/gpu/drm/i915/intel_color_manager.c
>> +++ b/drivers/gpu/drm/i915/intel_color_manager.c
>> @@ -27,6 +27,95 @@
>> #include "intel_color_manager.h"
>> +static int chv_set_degamma(struct drm_device *dev,
>> + struct drm_property_blob *blob, struct drm_crtc *crtc)
>> +{
>> + u16 red_fract, green_fract, blue_fract;
>> + u32 red, green, blue;
>> + u32 num_samples;
>> + u32 word = 0;
>> + u32 count, cgm_control, cgm_degamma_reg;
>> + i915_reg_t val;
>> + enum pipe pipe;
>> + struct drm_palette *degamma_data;
>> + struct drm_i915_private *dev_priv = dev->dev_private;
>> + struct drm_r32g32b32 *correction_values = NULL;
>> + struct drm_crtc_state *state = crtc->state;
>> +
>> + if (WARN_ON(!blob))
>> + return -EINVAL;
>> +
>> + degamma_data = (struct drm_palette *)blob->data;
>> + pipe = to_intel_crtc(crtc)->pipe;
>> + num_samples = blob->length / sizeof(struct drm_r32g32b32);
>> +
>> + switch (num_samples) {
>> + case GAMMA_DISABLE_VALS:
>> + /* Disable DeGamma functionality on Pipe - CGM Block */
>> + val = _MMIO(_PIPE_CGM_CONTROL(pipe));
>> + cgm_control = I915_READ(val);
>> + cgm_control &= ~CGM_DEGAMMA_EN;
>> + state->palette_before_ctm_blob = NULL;
>> +
>> + I915_WRITE(val, cgm_control);
>> + DRM_DEBUG_DRIVER("DeGamma disabled on Pipe %c\n",
>> + pipe_name(pipe));
>> + break;
>> +
>> + case CHV_DEGAMMA_MAX_VALS:
>> + cgm_degamma_reg = _PIPE_DEGAMMA_BASE(pipe);
>> + count = 0;
>> + correction_values = (struct drm_r32g32b32 *)°amma_data->lut;
>
> The line above is incorrect. This is reading from an invalid memory
> address.
> Just use this :
>
> correction_values = degamma_data->lut;
>
Nothing wrong in the existing way in the code, the pointer is valid. In
fact if I do the way you suggested, we will see a compilation warning
(we tried that previously). We verified the data coming from user space
in kernel, and we are getting valid data.
>
>> + while (count < CHV_DEGAMMA_MAX_VALS) {
>> + blue = correction_values[count].b32;
>> + green = correction_values[count].g32;
>> + red = correction_values[count].r32;
>> +
>> + if (blue > CHV_MAX_GAMMA)
>> + blue = CHV_MAX_GAMMA;
>> +
>> + if (green > CHV_MAX_GAMMA)
>> + green = CHV_MAX_GAMMA;
>> +
>> + if (red > CHV_MAX_GAMMA)
>> + red = CHV_MAX_GAMMA;
>> +
>> + blue_fract = GET_BITS(blue, 8, 14);
>> + green_fract = GET_BITS(green, 8, 14);
>> + red_fract = GET_BITS(red, 8, 14);
> This is reading the wrong bits from the blob here.
> Using your macro, it should be :
>
> blue_fract = GET_BITS(blue, 10, 14);
> green_fract = GET_BITS(green, 10, 14);
> red_fract = GET_BITS(red, 10, 14);
>
Yes, Thanks for pointing this out, I was picking LSB 14 out of 16 bits
I should have taken MSB 14 out of 16 bits, so we could loose precision
of 2 bits.
>> +
>> + /* Green (29:16) and Blue (13:0) in DWORD1 */
>> + SET_BITS(word, green_fract, 16, 14);
>> + SET_BITS(word, blue_fract, 0, 14);
>> + val = _MMIO(cgm_degamma_reg);
>> + I915_WRITE(val, word);
>> + cgm_degamma_reg += 4;
>> +
>> + /* Red (13:0) to be written to DWORD2 */
>> + word = red_fract;
>> + val = _MMIO(cgm_degamma_reg);
>> + I915_WRITE(val, word);
>> + cgm_degamma_reg += 4;
>> + count++;
>> + }
>> +
>> + DRM_DEBUG_DRIVER("DeGamma LUT loaded for Pipe %c\n",
>> + pipe_name(pipe));
>> +
>> + /* Enable DeGamma on Pipe */
>> + val = _MMIO(_PIPE_CGM_CONTROL(pipe));
>> + I915_WRITE(val, I915_READ(val) | CGM_DEGAMMA_EN);
>> + DRM_DEBUG_DRIVER("DeGamma correction enabled on Pipe %c\n",
>> + pipe_name(pipe));
>> + break;
>> +
>> + default:
>> + DRM_ERROR("Invalid number of samples for DeGamma LUT\n");
>> + return -EINVAL;
>> + }
>> + return 0;
>> +}
>> +
>> static int chv_set_gamma(struct drm_device *dev, struct
>> drm_property_blob *blob,
>> struct drm_crtc *crtc)
>> {
>> @@ -158,4 +247,10 @@ void intel_attach_color_properties_to_crtc(struct
>> drm_device *dev,
>> DRM_DEBUG_DRIVER("gamma property attached to CRTC\n");
>> }
>> + /* Degamma correction */
>> + if (config->cm_palette_before_ctm_property) {
>> + drm_object_attach_property(mode_obj,
>> + config->cm_palette_before_ctm_property, 0);
>> + DRM_DEBUG_DRIVER("degamma property attached to CRTC\n");
>> + }
>> }
>> diff --git a/drivers/gpu/drm/i915/intel_color_manager.h
>> b/drivers/gpu/drm/i915/intel_color_manager.h
>> index de706d9..77a2119 100644
>> --- a/drivers/gpu/drm/i915/intel_color_manager.h
>> +++ b/drivers/gpu/drm/i915/intel_color_manager.h
>> @@ -63,5 +63,10 @@
>> #define CHV_GAMMA_SHIFT_GREEN 16
>> #define CHV_MAX_GAMMA ((1 << 24) - 1)
>> +/* Degamma on CHV */
>> +#define CHV_DEGAMMA_MSB_SHIFT 2
>> +#define CHV_DEGAMMA_GREEN_SHIFT 16
>> +
>> /* CHV CGM Block */
>> #define CGM_GAMMA_EN (1 << 2)
>> +#define CGM_DEGAMMA_EN (1 << 0)
>
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
next prev parent reply other threads:[~2015-12-17 11:46 UTC|newest]
Thread overview: 29+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-12-03 11:36 [PATCH v8 00/25] Color Management for DRM framework Shashank Sharma
2015-12-03 11:36 ` [PATCH v8 01/25] drm: Create Color Management DRM properties Shashank Sharma
2015-12-03 11:36 ` [PATCH v8 02/25] drm: Create Color Management query properties Shashank Sharma
2015-12-03 11:36 ` [PATCH v8 03/25] drm: Add color correction blobs in CRTC state Shashank Sharma
2015-12-03 11:36 ` [PATCH v8 04/25] drm: Add set property support for color manager Shashank Sharma
2015-12-03 11:36 ` [PATCH v8 05/25] drm: Add get " Shashank Sharma
2015-12-03 11:36 ` [PATCH v8 06/25] drm: Add drm structures for palette color property Shashank Sharma
2015-12-03 11:36 ` [PATCH v8 07/25] drm: Add structure for CTM " Shashank Sharma
2015-12-03 11:36 ` [PATCH v8 08/25] drm: Add color correction state flag Shashank Sharma
2015-12-03 11:36 ` [PATCH v8 09/25] drm/i915: Add set property interface for CRTC Shashank Sharma
2015-12-03 11:36 ` [PATCH v8 10/25] drm/i915: Create color management files Shashank Sharma
2015-12-03 11:36 ` [PATCH v8 11/25] drm/i915: Register color correction capabilities Shashank Sharma
2015-12-03 11:36 ` [PATCH v8 12/25] drm/i915: CHV: Load gamma color correction values Shashank Sharma
2015-12-03 11:36 ` [PATCH v8 13/25] drm/i915: CHV: Load degamma " Shashank Sharma
2015-12-03 11:36 ` [PATCH v8 14/25] drm/i915: CHV: Pipe level Gamma correction Shashank Sharma
2015-12-03 11:36 ` [PATCH v8 15/25] drm/i915: CHV: Pipe level degamma correction Shashank Sharma
2015-12-16 16:06 ` Lionel Landwerlin
2015-12-17 11:46 ` Sharma, Shashank [this message]
2015-12-03 11:36 ` [PATCH v8 16/25] drm/i915: CHV: Pipe level CSC correction Shashank Sharma
2015-12-03 11:36 ` [PATCH v8 17/25] drm/i915: Commit color correction to CRTC Shashank Sharma
2015-12-03 11:36 ` [PATCH v8 18/25] drm/i915: Attach color properties " Shashank Sharma
2015-12-03 11:36 ` [PATCH v8 19/25] drm/i915: BDW: Load gamma correction values Shashank Sharma
2015-12-03 11:36 ` [PATCH v8 20/25] drm/i915: BDW: Pipe level Gamma correction Shashank Sharma
2015-12-09 15:15 ` [PATCH v9 " Lionel Landwerlin
2015-12-03 11:36 ` [PATCH v8 21/25] drm/i915: BDW: Load degamma correction values Shashank Sharma
2015-12-03 11:36 ` [PATCH v8 22/25] drm/i915: BDW: Pipe level degamma correction Shashank Sharma
2015-12-03 11:36 ` [PATCH v8 23/25] drm/i915: BDW: Pipe level CSC correction Shashank Sharma
2015-12-03 11:37 ` [PATCH v8 24/25] drm/i915: disable plane gamma Shashank Sharma
2015-12-03 11:37 ` [PATCH v8 25/25] drm/i915: Commit color correction only when needed Shashank Sharma
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=5672A08F.7030502@intel.com \
--to=shashank.sharma@intel.com \
--cc=daniel.vetter@intel.com \
--cc=intel-gfx@lists.freedesktop.org \
--cc=jim.bish@intel.com \
--cc=kausalmalladi@gmail.com \
--cc=lionel.g.landwerlin@intel.com \
--cc=robert.bradford@intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox