On Tue, Apr 27, 2021 at 11:10:59AM +0300, Pekka Paalanen wrote:
> On Mon, 26 Apr 2021 14:31:28 -0300
> Melissa Wen <melissa.srw@xxxxxxxxx> wrote:
>
> > On 04/26, Daniel Vetter wrote:
> > > On Mon, Apr 26, 2021 at 11:03:15AM +0300, Pekka Paalanen wrote:
> > > > On Sat, 24 Apr 2021 05:25:31 -0300
> > > > Melissa Wen <melissa.srw@xxxxxxxxx> wrote:
> > > >
> > > > > Add support for composing XRGB888 planes in addition to the ARGB8888
> > > > > format. In the case of an XRGB plane at the top, the composition consists
> > > > > of copying the RGB values of a pixel from src to dst and clearing alpha
> > > > > channel, without the need for alpha blending operations for each pixel.
> > > > >
> > > > > Blend equations assume a completely opaque background, i.e., primary plane
> > > > > is not cleared before pixel blending but alpha channel is explicitly
> > > > > opaque (a = 0xff). Also, there is room for performance evaluation in
> > > > > switching pixel blend operation according to the plane format.
> > > > >
> > > > > v4:
> > > > > - clear alpha channel (0xff) after blend color values by pixel
> > > > > - improve comments on blend ops to reflect the current state
> > > > > - describe in the commit message future improvements for plane composition
> > > > >
> > > > > Signed-off-by: Melissa Wen <melissa.srw@xxxxxxxxx>
> > > > > Reviewed-by: Daniel Vetter <daniel.vetter@xxxxxxxx>
> > > > > ---
> > > > > drivers/gpu/drm/vkms/vkms_composer.c | 56 ++++++++++++++++++++++------
> > > > > drivers/gpu/drm/vkms/vkms_plane.c | 7 ++--
> > > > > 2 files changed, 48 insertions(+), 15 deletions(-)
> > > > >
> > > > > diff --git a/drivers/gpu/drm/vkms/vkms_composer.c b/drivers/gpu/drm/vkms/vkms_composer.c
> > > > > index 02642801735d..7e01bc39d2a1 100644
> > > > > --- a/drivers/gpu/drm/vkms/vkms_composer.c
> > > > > +++ b/drivers/gpu/drm/vkms/vkms_composer.c
> > > > > @@ -4,6 +4,7 @@
> > > > >
> > > > > #include <drm/drm_atomic.h>
> > > > > #include <drm/drm_atomic_helper.h>
> > > > > +#include <drm/drm_fourcc.h>
> > > > > #include <drm/drm_gem_framebuffer_helper.h>
> > > > > #include <drm/drm_gem_shmem_helper.h>
> > > > > #include <drm/drm_vblank.h>
> > > > > @@ -64,7 +65,17 @@ static u8 blend_channel(u8 src, u8 dst, u8 alpha)
> > > > > return new_color;
> > > > > }
> > > > >
> > > > > -static void alpha_blending(const u8 *argb_src, u8 *argb_dst)
> > > > > +/**
> > > > > + * alpha_blend - alpha blending equation
> > > > > + * @argb_src: src pixel on premultiplied alpha mode
> > > > > + * @argb_dst: dst pixel completely opaque
> > > > > + *
> > > > > + * blend pixels using premultiplied blend formula. The current DRM assumption
> > > > > + * is that pixel color values have been already pre-multiplied with the alpha
> > > > > + * channel values. See more drm_plane_create_blend_mode_property(). Also, this
> > > > > + * formula assumes a completely opaque background.
> > > > > + */
> > > > > +static void alpha_blend(const u8 *argb_src, u8 *argb_dst)
> > > > > {
> > > > > u8 alpha;
> > > > >
> > > > > @@ -72,8 +83,16 @@ static void alpha_blending(const u8 *argb_src, u8 *argb_dst)
> > > > > argb_dst[0] = blend_channel(argb_src[0], argb_dst[0], alpha);
> > > > > argb_dst[1] = blend_channel(argb_src[1], argb_dst[1], alpha);
> > > > > argb_dst[2] = blend_channel(argb_src[2], argb_dst[2], alpha);
> > > > > - /* Opaque primary */
> > > > > - argb_dst[3] = 0xFF;
> > > > > +}
> > > > > +
> > > > > +/**
> > > > > + * x_blend - blending equation that ignores the pixel alpha
> > > > > + *
> > > > > + * overwrites RGB color value from src pixel to dst pixel.
> > > > > + */
> > > > > +static void x_blend(const u8 *xrgb_src, u8 *xrgb_dst)
> > > > > +{
> > > > > + memcpy(xrgb_dst, xrgb_src, sizeof(u8) * 3);
> > > >
> > > > Hi,
> > > >
> > > > this function very clearly assumes a very specific pixel format on both
> > > > source and destination. I think it would be good if the code comments
> > > > called out exactly which DRM_FORMAT_* they assume. This would be good
> > > > to do on almost every function that makes such assumptions. I believe that
> > > > would help code readability, and also point out explicitly which things
> > > > need to be fixed when you add support for even more pixel formats.
> > > >
> > > > "xrgb" and "argb" are IMO too vague. You might be referring to
> > > > DRM_FORMAT_XRGB* and DRM_FORMAT_ARGB*, or maybe you are referring to any
> > > > pixel format that happens to have or not have an alpha channel in
> > > > addition to the three RGB channels in some order and width.
> > > >
> > > > Being explicit that these refer to specific DRM_FORMAT_* should also
> > > > help understanding how things work on big-endian CPUs. My current
> > > > understanding is that this memcpy is correct also on big-endian, given
> > > > DRM_FORMAT_XRGB8888.
> >
> > This endianess issue seems a little tricky to me. I remember we have
> > already discussed something similar when introducing alpha blend ops. I
> > took little endian as default by a code comment on
> > include/drm/drm_fourcc.h: DRM formats are little endian. But also, I am
> > not sure if I got it well.
>
> DRM format *definitions* are written on a little-endian CPU. When you
> have a big-endian CPU, the byte-to-byte memory contents still remain
> the same. That means if you have a uint32_t pixel in a certain
> DRM_FORMAT_*, you must always access the bits of it like a
> little-endian CPU would.
>
> I think this was the "recently" agreed definition, and drivers who do
> not follow this still exist because fixing them would break userspace?
Legacy AddFb might give you a big endia drm_fourcc on some drivers. AddFb2
will not play such tricks.
Also big-endian is dead, imo if someone cares enough about it they could
make "fix vkms for big-endian" a nice project :-)
-Daniel
> So if you make the assumption that your machine is little-endian, you
> have no worries, but you might want to document that you are making
> this assumption, so that people know it might not be correct on
> big-endian. It is important to document that it is *unknown* if the
> code is correct on big-endian, to make people think rather than blindly
> add a #ifdef big-endian then swap bytes, because the code might be
> correct already - you just don't know yet.
>
> I wouldn't personally bother thinking about big-endian, other than
> acknowledging that I don't think about big-endian when writing code, and
> noticing places where it might make a difference (prime example:
> accessing pixel components via bytes vs. bits-of-uint32).
>
> > > > Hmm, or rather, is this particular function intended to be general in
> > > > the sense that the order of RGB channels does not matter as long as it's
> > > > the same in both source and destination? Which would mean I had a wrong
> > > > assumption from the start.
> > >
> > > Atm all vkms supports is X/ARGB8888, and even there we throw around random
> > > limits. Add support for more pixel formats is definitely on the list, and
> > > then all the blend/compose stuff needs to be quite drastically
> > > rearchitected.
>
> If there are arbitrary limitations, then IMO those are especially
> important to mention.
>
> > yes, currently, we only have on vkms these two formats listed as
> > supported (X/ARGB8888), so, I think it is ok, since we do not expected
> > anything other than these two.
> >
> > >
> > > I think until we're there documenting what's already documented in the
> > > todo list feels like overkill.
>
> I'm literally asking for single-sentence comments added, like:
>
> /* DRM_FORMAT_XRGB8888 */
>
> It makes all the difference to anyone seeing the code for the first
> time. Particularly if people want to review patches into this area,
> because patches are sent via email and therefore completely lack the
> context of the surrounding code at large and knowledge of which kernel
> tree they apply to (I'm not a kernel dev), not to mention the trouble
> of having to apply a patch to be able to look at more context.
>
> Thanks for mentioning https://lkml.org/lkml/2020/8/30/163 in the other
> email!
>
>
> Thanks,
> pq
>
>
> > > -Daniel
> > >
> > > >
> > > > > }
> > > > >
> > > > > /**
> > > > > @@ -82,16 +101,20 @@ static void alpha_blending(const u8 *argb_src, u8 *argb_dst)
> > > > > * @vaddr_src: source address
> > > > > * @dst_composer: destination framebuffer's metadata
> > > > > * @src_composer: source framebuffer's metadata
> > > > > + * @pixel_blend: blending equation based on plane format
> > > > > *
> > > > > - * Blend the vaddr_src value with the vaddr_dst value using the pre-multiplied
> > > > > - * alpha blending equation, since DRM currently assumes that the pixel color
> > > > > - * values have already been pre-multiplied with the alpha channel values. See
> > > > > - * more drm_plane_create_blend_mode_property(). This function uses buffer's
> > > > > - * metadata to locate the new composite values at vaddr_dst.
> > > > > + * Blend the vaddr_src value with the vaddr_dst value using a pixel blend
> > > > > + * equation according to the plane format and clearing alpha channel to an
> > > > > + * completely opaque background. This function uses buffer's metadata to locate
> > > > > + * the new composite values at vaddr_dst.
> > > > > + *
> > > > > + * TODO: completely clear the primary plane (a = 0xff) before starting to blend
> > > > > + * pixel color values
> > > > > */
> > > > > static void blend(void *vaddr_dst, void *vaddr_src,
> > > > > struct vkms_composer *dst_composer,
> > > > > - struct vkms_composer *src_composer)
> > > > > + struct vkms_composer *src_composer,
> > > > > + void (*pixel_blend)(const u8 *, u8 *))
> > > > > {
> > > > > int i, j, j_dst, i_dst;
> > > > > int offset_src, offset_dst;
> > > > > @@ -119,7 +142,9 @@ static void blend(void *vaddr_dst, void *vaddr_src,
> > > > >
> > > > > pixel_src = (u8 *)(vaddr_src + offset_src);
> > > > > pixel_dst = (u8 *)(vaddr_dst + offset_dst);
> > > > > - alpha_blending(pixel_src, pixel_dst);
> > > > > + pixel_blend(pixel_src, pixel_dst);
> > > > > + /* clearing alpha channel (0xff)*/
> > > > > + memset(vaddr_dst + offset_dst + 3, 0xff, 1);
> > > >
> > > > A one byte memset?
> > > >
> > > > Wouldn't pixel_dst[3] = 0xff; be more clear?
> >
> > yes, I will change it.
> >
> > Thanks for these suggestions,
> >
> > Melissa
> > > >
> > > >
> > > > Thanks,
> > > > pq
> > > >
> > > > > }
> > > > > i_dst++;
> > > > > }
> > > > > @@ -131,6 +156,8 @@ static void compose_plane(struct vkms_composer *primary_composer,
> > > > > {
> > > > > struct drm_gem_object *plane_obj;
> > > > > struct drm_gem_shmem_object *plane_shmem_obj;
> > > > > + struct drm_framebuffer *fb = &plane_composer->fb;
> > > > > + void (*pixel_blend)(const u8 *p_src, u8 *p_dst);
> > > > >
> > > > > plane_obj = drm_gem_fb_get_obj(&plane_composer->fb, 0);
> > > > > plane_shmem_obj = to_drm_gem_shmem_obj(plane_obj);
> > > > > @@ -138,8 +165,13 @@ static void compose_plane(struct vkms_composer *primary_composer,
> > > > > if (WARN_ON(!plane_shmem_obj->vaddr))
> > > > > return;
> > > > >
> > > > > - blend(vaddr_out, plane_shmem_obj->vaddr,
> > > > > - primary_composer, plane_composer);
> > > > > + if (fb->format->format == DRM_FORMAT_ARGB8888)
> > > > > + pixel_blend = &alpha_blend;
> > > > > + else
> > > > > + pixel_blend = &x_blend;
> > > > > +
> > > > > + blend(vaddr_out, plane_shmem_obj->vaddr, primary_composer,
> > > > > + plane_composer, pixel_blend);
> > > > > }
> > > > >
> > > > > static int compose_active_planes(void **vaddr_out,
> > > > > diff --git a/drivers/gpu/drm/vkms/vkms_plane.c b/drivers/gpu/drm/vkms/vkms_plane.c
> > > > > index 135140f8e87a..da4251aff67f 100644
> > > > > --- a/drivers/gpu/drm/vkms/vkms_plane.c
> > > > > +++ b/drivers/gpu/drm/vkms/vkms_plane.c
> > > > > @@ -16,8 +16,9 @@ static const u32 vkms_formats[] = {
> > > > > DRM_FORMAT_XRGB8888,
> > > > > };
> > > > >
> > > > > -static const u32 vkms_cursor_formats[] = {
> > > > > +static const u32 vkms_plane_formats[] = {
> > > > > DRM_FORMAT_ARGB8888,
> > > > > + DRM_FORMAT_XRGB8888
> > > > > };
> > > > >
> > > > > static struct drm_plane_state *
> > > > > @@ -200,8 +201,8 @@ struct vkms_plane *vkms_plane_init(struct vkms_device *vkmsdev,
> > > > > int nformats;
> > > > >
> > > > > if (type == DRM_PLANE_TYPE_CURSOR) {
> > > > > - formats = vkms_cursor_formats;
> > > > > - nformats = ARRAY_SIZE(vkms_cursor_formats);
> > > > > + formats = vkms_plane_formats;
> > > > > + nformats = ARRAY_SIZE(vkms_plane_formats);
> > > > > funcs = &vkms_primary_helper_funcs;
> > > > > } else {
> > > > > formats = vkms_formats;
> > > >
> > >
> > >
> > >
> > > > _______________________________________________
> > > > dri-devel mailing list
> > > > dri-devel@xxxxxxxxxxxxxxxxxxxxx
> > > > https://lists.freedesktop.org/mailman/listinfo/dri-devel
> > >
> > >
> > > --
> > > Daniel Vetter
> > > Software Engineer, Intel Corporation
> > > http://blog.ffwll.ch
>
--
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch
_______________________________________________
dri-devel mailing list
dri-devel@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/dri-devel
