From: Alex Hung <alex.hung@amd.com>
To: <dri-devel@lists.freedesktop.org>, <amd-gfx@lists.freedesktop.org>
Cc: <wayland-devel@lists.freedesktop.org>, <harry.wentland@amd.com>,
<alex.hung@amd.com>, <leo.liu@amd.com>,
<ville.syrjala@linux.intel.com>, <pekka.paalanen@collabora.com>,
<contact@emersion.fr>, <mwen@igalia.com>, <jadahl@redhat.com>,
<sebastian.wick@redhat.com>, <shashank.sharma@amd.com>,
<agoins@nvidia.com>, <joshua@froggi.es>, <mdaenzer@redhat.com>,
<aleixpol@kde.org>, <xaver.hugl@gmail.com>,
<victoria@system76.com>, <daniel@ffwll.ch>,
<uma.shankar@intel.com>, <quic_naseer@quicinc.com>,
<quic_cbraga@quicinc.com>, <quic_abhinavk@quicinc.com>,
<marcan@marcan.st>, <Liviu.Dudau@arm.com>,
<sashamcintosh@google.com>, <chaitanya.kumar.borah@intel.com>,
<louis.chauvet@bootlin.com>
Subject: [PATCH V8 16/43] drm/vkms: Use s32 for internal color pipeline precision
Date: Wed, 26 Mar 2025 17:46:57 -0600 [thread overview]
Message-ID: <20250326234748.2982010-17-alex.hung@amd.com> (raw)
In-Reply-To: <20250326234748.2982010-1-alex.hung@amd.com>
From: Harry Wentland <harry.wentland@amd.com>
Certain operations require us to preserve values below 0.0 and
above 1.0 (0x0 and 0xffff respectively in 16 bpc unorm). One
such operation is a BT709 encoding operation followed by its
decoding operation, or the reverse.
We'll use s32 values as intermediate in and outputs of our
color operations, for the operations where it matters.
For now this won't apply to LUT operations. We might want to
update those to work on s32 as well, but it's unclear how
that should work for unorm LUT definitions. We'll revisit
that once we add LUT + CTM tests.
In order to allow for this we'll also invert the nesting of our
colorop processing loops. We now use the pixel iteration loop
on the outside and the colorop iteration on the inside.
Reviewed-by: Louis Chauvet <louis.chauvet@bootlin.com>
Signed-off-by: Alex Hung <alex.hung@amd.com>
Signed-off-by: Harry Wentland <harry.wentland@amd.com>
---
v7:
- Fix checkpatch warnings
- Add a commit messages
- Fix code styles by adding and removing spaces (new lines, tabs and so on)
v6:
- use clamp_val instead of manual clamping (Louis Chauvet)
v4:
- Clarify that we're packing 16-bit UNORM into s32, not
converting values to a different representation (Pekka)
v3:
- Use new colorop->next pointer
drivers/gpu/drm/vkms/vkms_composer.c | 27 +++++++++++++++++++++++++--
drivers/gpu/drm/vkms/vkms_drv.h | 4 ++++
2 files changed, 29 insertions(+), 2 deletions(-)
diff --git a/drivers/gpu/drm/vkms/vkms_composer.c b/drivers/gpu/drm/vkms/vkms_composer.c
index 72890ffea13a..36f0caabb67b 100644
--- a/drivers/gpu/drm/vkms/vkms_composer.c
+++ b/drivers/gpu/drm/vkms/vkms_composer.c
@@ -128,7 +128,7 @@ static void apply_lut(const struct vkms_crtc_state *crtc_state, struct line_buff
}
}
-static void apply_colorop(struct pixel_argb_u16 *pixel, struct drm_colorop *colorop)
+static void apply_colorop(struct pixel_argb_s32 *pixel, struct drm_colorop *colorop)
{
struct drm_colorop_state *colorop_state = colorop->state;
struct drm_device *dev = colorop->dev;
@@ -157,9 +157,26 @@ static void apply_colorop(struct pixel_argb_u16 *pixel, struct drm_colorop *colo
static void pre_blend_color_transform(const struct vkms_plane_state *plane_state,
struct line_buffer *output_buffer)
{
+ struct pixel_argb_s32 pixel;
+
for (size_t x = 0; x < output_buffer->n_pixels; x++) {
struct drm_colorop *colorop = plane_state->base.base.color_pipeline;
+ /*
+ * Some operations, such as applying a BT709 encoding matrix,
+ * followed by a decoding matrix, require that we preserve
+ * values above 1.0 and below 0.0 until the end of the pipeline.
+ *
+ * Pack the 16-bit UNORM values into s32 to give us head-room to
+ * avoid clipping until we're at the end of the pipeline. Clip
+ * intentionally at the end of the pipeline before packing
+ * UNORM values back into u16.
+ */
+ pixel.a = output_buffer->pixels[x].a;
+ pixel.r = output_buffer->pixels[x].r;
+ pixel.g = output_buffer->pixels[x].g;
+ pixel.b = output_buffer->pixels[x].b;
+
while (colorop) {
struct drm_colorop_state *colorop_state;
@@ -169,10 +186,16 @@ static void pre_blend_color_transform(const struct vkms_plane_state *plane_state
return;
if (!colorop_state->bypass)
- apply_colorop(&output_buffer->pixels[x], colorop);
+ apply_colorop(&pixel, colorop);
colorop = colorop->next;
}
+
+ /* clamp values */
+ output_buffer->pixels[x].a = clamp_val(pixel.a, 0, 0xffff);
+ output_buffer->pixels[x].r = clamp_val(pixel.r, 0, 0xffff);
+ output_buffer->pixels[x].g = clamp_val(pixel.g, 0, 0xffff);
+ output_buffer->pixels[x].b = clamp_val(pixel.b, 0, 0xffff);
}
}
diff --git a/drivers/gpu/drm/vkms/vkms_drv.h b/drivers/gpu/drm/vkms/vkms_drv.h
index 999625de0050..d9196b9db025 100644
--- a/drivers/gpu/drm/vkms/vkms_drv.h
+++ b/drivers/gpu/drm/vkms/vkms_drv.h
@@ -43,6 +43,10 @@ struct vkms_frame_info {
unsigned int rotation;
};
+struct pixel_argb_s32 {
+ s32 a, r, g, b;
+};
+
struct pixel_argb_u16 {
u16 a, r, g, b;
};
--
2.43.0
next prev parent reply other threads:[~2025-03-26 23:52 UTC|newest]
Thread overview: 128+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-03-26 23:46 [PATCH V8 00/43] Color Pipeline API w/ VKMS Alex Hung
2025-03-26 23:46 ` [PATCH V8 01/43] drm: Add helper for conversion from signed-magnitude Alex Hung
2025-03-26 23:46 ` [PATCH V8 02/43] drm/vkms: Add kunit tests for VKMS LUT handling Alex Hung
2025-03-26 23:46 ` [PATCH V8 03/43] drm/doc/rfc: Describe why prescriptive color pipeline is needed Alex Hung
2025-03-29 15:26 ` Simon Ser
2025-04-01 0:10 ` Alex Hung
2025-04-01 7:26 ` Simon Ser
2025-03-31 16:24 ` Shengyu Qu
2025-03-31 16:41 ` Alex Hung
2025-03-31 16:54 ` Shengyu Qu
2025-03-26 23:46 ` [PATCH V8 04/43] drm/colorop: Introduce new drm_colorop mode object Alex Hung
2025-03-26 23:46 ` [PATCH V8 05/43] drm/colorop: Add TYPE property Alex Hung
2025-03-26 23:46 ` [PATCH V8 06/43] drm/colorop: Add 1D Curve subtype Alex Hung
2025-04-01 15:14 ` Daniel Stone
2025-04-01 19:53 ` Simon Ser
2025-04-01 21:02 ` Harry Wentland
2025-04-08 16:40 ` Daniel Stone
2025-04-08 17:30 ` Harry Wentland
2025-04-08 18:28 ` Daniel Stone
2025-04-10 7:53 ` Pipeline vs. no pipeline (Re: [PATCH V8 06/43] drm/colorop: Add 1D Curve subtype) Pekka Paalanen
2025-04-15 15:29 ` Harry Wentland
2025-04-16 14:39 ` Xaver Hugl
2025-04-17 8:28 ` Pekka Paalanen
2025-04-10 10:05 ` [PATCH V8 06/43] drm/colorop: Add 1D Curve subtype Simon Ser
2025-04-15 11:12 ` Borah, Chaitanya Kumar
2025-04-17 15:13 ` Simon Ser
2025-03-26 23:46 ` [PATCH V8 07/43] drm/colorop: Add BYPASS property Alex Hung
2025-03-26 23:46 ` [PATCH V8 08/43] drm/colorop: Add NEXT property Alex Hung
2025-03-27 23:26 ` Simon Ser
2025-03-26 23:46 ` [PATCH V8 09/43] drm/colorop: Add atomic state print for drm_colorop Alex Hung
2025-03-27 23:29 ` Simon Ser
2025-03-26 23:46 ` [PATCH V8 10/43] drm/plane: Add COLOR PIPELINE property Alex Hung
2025-03-29 14:33 ` Simon Ser
2025-03-26 23:46 ` [PATCH V8 11/43] drm/colorop: Introduce DRM_CLIENT_CAP_PLANE_COLOR_PIPELINE Alex Hung
2025-03-29 14:37 ` Simon Ser
2025-04-01 1:42 ` Shengyu Qu
2025-03-26 23:46 ` [PATCH V8 12/43] Documentation/gpu: document drm_colorop Alex Hung
2025-03-29 14:40 ` Simon Ser
2025-03-26 23:46 ` [PATCH V8 13/43] drm/vkms: Add enumerated 1D curve colorop Alex Hung
2025-03-26 23:46 ` [PATCH V8 14/43] drm/vkms: Add kunit tests for linear and sRGB LUTs Alex Hung
2025-03-26 23:46 ` [PATCH V8 15/43] drm/colorop: Add 3x4 CTM type Alex Hung
2025-03-26 23:46 ` Alex Hung [this message]
2025-03-26 23:46 ` [PATCH V8 17/43] drm/vkms: add 3x4 matrix in color pipeline Alex Hung
2025-03-26 23:46 ` [PATCH V8 18/43] drm/tests: Add a few tests around drm_fixed.h Alex Hung
2025-03-26 23:47 ` [PATCH V8 19/43] drm/vkms: Add tests for CTM handling Alex Hung
2025-03-26 23:47 ` [PATCH V8 20/43] drm/colorop: pass plane_color_pipeline client cap to atomic check Alex Hung
2025-03-29 15:32 ` Simon Ser
2025-03-26 23:47 ` [PATCH V8 21/43] drm/colorop: define a new macro for_each_new_colorop_in_state Alex Hung
2025-03-26 23:47 ` [PATCH V8 22/43] drm/amd/display: Ignore deprecated props when plane_color_pipeline set Alex Hung
2025-03-26 23:47 ` [PATCH V8 23/43] drm/amd/display: Add bypass COLOR PIPELINE Alex Hung
2025-03-26 23:47 ` [PATCH V8 24/43] drm/amd/display: Skip color pipeline initialization for cursor plane Alex Hung
2025-03-30 9:48 ` Shengyu Qu
2025-03-30 12:59 ` Shengyu Qu
2025-03-31 14:28 ` Alex Hung
2025-03-31 15:43 ` Shengyu Qu
2025-03-31 16:06 ` Alex Hung
2025-03-31 16:12 ` Shengyu Qu
2025-03-31 16:26 ` Alex Hung
2025-03-31 16:31 ` Shengyu Qu
2025-03-31 16:34 ` Alex Hung
2025-03-31 16:50 ` Shengyu Qu
2025-03-31 17:04 ` Shengyu Qu
2025-03-31 17:42 ` Alex Hung
2025-03-31 18:53 ` Xaver Hugl
2025-04-01 0:28 ` Alex Hung
2025-04-01 15:04 ` Xaver Hugl
2025-04-01 15:45 ` Melissa Wen
2025-04-01 19:39 ` Harry Wentland
2025-04-01 1:04 ` Shengyu Qu
2025-04-01 1:24 ` Alex Hung
2025-04-01 9:56 ` Michel Dänzer
2025-04-01 12:32 ` Shengyu Qu
2025-04-01 14:11 ` Michel Dänzer
2025-04-01 15:45 ` Shengyu Qu
2025-04-01 19:45 ` Harry Wentland
2025-04-02 3:47 ` Qu Shengyu
2025-04-01 16:24 ` Shengyu Qu
2025-03-26 23:47 ` [PATCH V8 25/43] drm/amd/display: Add support for sRGB EOTF in DEGAM block Alex Hung
2025-03-26 23:47 ` [PATCH V8 26/43] drm/amd/display: Add support for sRGB Inverse EOTF in SHAPER block Alex Hung
2025-03-26 23:47 ` [PATCH V8 27/43] drm/amd/display: Add support for sRGB EOTF in BLND block Alex Hung
2025-03-26 23:47 ` [PATCH V8 28/43] drm/colorop: Add PQ 125 EOTF and its inverse Alex Hung
2025-03-29 14:48 ` Simon Ser
2025-03-26 23:47 ` [PATCH V8 29/43] drm/amd/display: Enable support for PQ 125 EOTF and Inverse Alex Hung
2025-03-26 23:47 ` [PATCH V8 30/43] drm/colorop: add BT2020/BT709 OETF and Inverse OETF Alex Hung
2025-03-29 14:53 ` Simon Ser
2025-03-26 23:47 ` [PATCH V8 31/43] drm/amd/display: Add support for BT.709 and BT.2020 TFs Alex Hung
2025-03-26 23:47 ` [PATCH V8 32/43] drm/colorop: Add 1D Curve Custom LUT type Alex Hung
2025-03-29 14:55 ` Simon Ser
2025-04-15 6:09 ` Shankar, Uma
2025-04-15 6:16 ` Simon Ser
2025-04-15 6:40 ` Shankar, Uma
2025-04-15 15:05 ` Harry Wentland
2025-04-15 16:25 ` Simon Ser
2025-05-22 11:33 ` Shankar, Uma
2025-05-30 13:58 ` Pekka Paalanen
2025-06-03 8:30 ` Shankar, Uma
2025-06-03 10:51 ` Pekka Paalanen
2025-06-03 20:26 ` Harry Wentland
2025-06-04 18:59 ` Shankar, Uma
2025-06-05 7:30 ` Pekka Paalanen
2025-03-26 23:47 ` [PATCH V8 33/43] drm/amd/display: add shaper and blend colorops for 1D Curve Custom LUT Alex Hung
2025-03-26 23:47 ` [PATCH V8 34/43] drm/amd/display: add 3x4 matrix colorop Alex Hung
2025-03-26 23:47 ` [PATCH V8 35/43] drm/colorop: Add mutliplier type Alex Hung
2025-03-26 23:47 ` [PATCH V8 36/43] drm/amd/display: add multiplier colorop Alex Hung
2025-03-26 23:47 ` [PATCH V8 37/43] drm/amd/display: Swap matrix and multiplier Alex Hung
2025-03-26 23:47 ` [PATCH V8 38/43] drm/colorop: Define LUT_1D interpolation Alex Hung
2025-03-26 23:47 ` [PATCH V8 39/43] drm/colorop: allow non-bypass colorops Alex Hung
2025-03-29 15:41 ` Simon Ser
2025-03-26 23:47 ` [PATCH V8 40/43] drm/colorop: Add 3D LUT support to color pipeline Alex Hung
2025-03-29 14:57 ` Simon Ser
2025-04-25 13:50 ` Leandro Ribeiro
2025-05-13 3:39 ` Alex Hung
2025-05-17 1:22 ` Xaver Hugl
2025-05-17 11:53 ` Simon Ser
2025-05-17 22:32 ` Xaver Hugl
2025-05-19 23:43 ` Simon Ser
2025-05-20 20:13 ` Harry Wentland
2025-05-21 19:18 ` Harry Wentland
2025-05-22 10:14 ` Simon Ser
2025-05-22 11:46 ` Shankar, Uma
2025-05-17 17:36 ` Autumn Ashton
2025-03-26 23:47 ` [PATCH V8 41/43] drm/amd/display: add 3D LUT colorop Alex Hung
2025-03-26 23:47 ` [PATCH V8 42/43] drm/amd/display: Add AMD color pipeline doc Alex Hung
2025-03-26 23:47 ` [PATCH V8 43/43] drm/colorop: Add destroy functions for color pipeline Alex Hung
2025-03-29 15:48 ` Simon Ser
2025-04-01 2:42 ` Alex Hung
2025-04-10 16:18 ` Simon Ser
2025-03-29 15:51 ` [PATCH V8 00/43] Color Pipeline API w/ VKMS Simon Ser
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=20250326234748.2982010-17-alex.hung@amd.com \
--to=alex.hung@amd.com \
--cc=Liviu.Dudau@arm.com \
--cc=agoins@nvidia.com \
--cc=aleixpol@kde.org \
--cc=amd-gfx@lists.freedesktop.org \
--cc=chaitanya.kumar.borah@intel.com \
--cc=contact@emersion.fr \
--cc=daniel@ffwll.ch \
--cc=dri-devel@lists.freedesktop.org \
--cc=harry.wentland@amd.com \
--cc=jadahl@redhat.com \
--cc=joshua@froggi.es \
--cc=leo.liu@amd.com \
--cc=louis.chauvet@bootlin.com \
--cc=marcan@marcan.st \
--cc=mdaenzer@redhat.com \
--cc=mwen@igalia.com \
--cc=pekka.paalanen@collabora.com \
--cc=quic_abhinavk@quicinc.com \
--cc=quic_cbraga@quicinc.com \
--cc=quic_naseer@quicinc.com \
--cc=sashamcintosh@google.com \
--cc=sebastian.wick@redhat.com \
--cc=shashank.sharma@amd.com \
--cc=uma.shankar@intel.com \
--cc=victoria@system76.com \
--cc=ville.syrjala@linux.intel.com \
--cc=wayland-devel@lists.freedesktop.org \
--cc=xaver.hugl@gmail.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.