Re: [RFC PATCH 3/4] drm/arm/malidp: Set the AFBC register bits if the framebuffer has AFBC modifier

From: Ayan Halder
Date: Thu Jul 05 2018 - 09:32:01 EST

Next message: Vincent Guittot: "Re: [PATCH v7 00/11] track CPU utilization"
Previous message: Laurentiu Tudor: "[PATCH] docs: kernel-parameters.txt: document xhci-hcd.quirks parameter"
Next in thread: Liviu Dudau: "Re: [RFC PATCH 3/4] drm/arm/malidp: Set the AFBC register bits if the framebuffer has AFBC modifier"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

On Tue, Jun 26, 2018 at 02:17:17PM +0100, Liviu Dudau wrote:
> Hi Ayan,
>
> Thanks for the patch! I have some small comments to make:
>
> On Fri, Jun 15, 2018 at 02:51:33PM +0100, Ayan Kumar Halder wrote:
> > Added the AFBC decoder registers for DP500 , DP550 and DP650.
> > These registers control the processing of AFBC buffers. It controls various
> > features like AFBC decoder enable, lossless transformation and block split
> > as well as setting of the left, right, top and bottom cropping of AFBC buffers
> > (in number of pixels).
> > All the layers (except DE_SMART) support framebuffers with AFBC modifiers.
> > One needs to set the pixel values of the top, left, bottom and right cropping
> > for the AFBC framebuffer.
> > Added the functionality in malidp_de_plane_update() to set the various
> > registers for AFBC decoder, depending on the modifiers.
> >
> > Signed-off-by: Ayan Kumar halder <ayan.halder@xxxxxxx>
> > Reviewed-by: Brian Starkey <brian.starkey@xxxxxxx>
> > ---
> > drivers/gpu/drm/arm/malidp_hw.c | 27 ++++++++-----
> > drivers/gpu/drm/arm/malidp_hw.h | 2 +
> > drivers/gpu/drm/arm/malidp_planes.c | 81 +++++++++++++++++++++++++++++++++----
> > drivers/gpu/drm/arm/malidp_regs.h | 20 +++++++++
> > 4 files changed, 111 insertions(+), 19 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/arm/malidp_hw.c b/drivers/gpu/drm/arm/malidp_hw.c
> > index 4dbf39f..fd6b510 100644
> > --- a/drivers/gpu/drm/arm/malidp_hw.c
> > +++ b/drivers/gpu/drm/arm/malidp_hw.c
> > @@ -76,33 +76,38 @@ static const struct malidp_format_id malidp550_de_formats[] = {
> >
> > static const struct malidp_layer malidp500_layers[] = {
> > { DE_VIDEO1, MALIDP500_DE_LV_BASE, MALIDP500_DE_LV_PTR_BASE,
> > - MALIDP_DE_LV_STRIDE0, MALIDP500_LV_YUV2RGB, ROTATE_ANY },
> > + MALIDP_DE_LV_STRIDE0, MALIDP500_LV_YUV2RGB, ROTATE_ANY,
> > + MALIDP500_DE_LV_AD_CTRL },
> > { DE_GRAPHICS1, MALIDP500_DE_LG1_BASE, MALIDP500_DE_LG1_PTR_BASE,
> > - MALIDP_DE_LG_STRIDE, 0, ROTATE_ANY },
> > + MALIDP_DE_LG_STRIDE, 0, ROTATE_ANY, MALIDP500_DE_LG1_AD_CTRL },
> > { DE_GRAPHICS2, MALIDP500_DE_LG2_BASE, MALIDP500_DE_LG2_PTR_BASE,
> > - MALIDP_DE_LG_STRIDE, 0, ROTATE_ANY },
> > + MALIDP_DE_LG_STRIDE, 0, ROTATE_ANY, MALIDP500_DE_LG2_AD_CTRL },
> > };
> >
> > static const struct malidp_layer malidp550_layers[] = {
> > { DE_VIDEO1, MALIDP550_DE_LV1_BASE, MALIDP550_DE_LV1_PTR_BASE,
> > - MALIDP_DE_LV_STRIDE0, MALIDP550_LV_YUV2RGB, ROTATE_ANY },
> > + MALIDP_DE_LV_STRIDE0, MALIDP550_LV_YUV2RGB, ROTATE_ANY,
> > + MALIDP550_DE_LV1_AD_CTRL },
> > { DE_GRAPHICS1, MALIDP550_DE_LG_BASE, MALIDP550_DE_LG_PTR_BASE,
> > - MALIDP_DE_LG_STRIDE, 0, ROTATE_ANY },
> > + MALIDP_DE_LG_STRIDE, 0, ROTATE_ANY, MALIDP550_DE_LG_AD_CTRL },
> > { DE_VIDEO2, MALIDP550_DE_LV2_BASE, MALIDP550_DE_LV2_PTR_BASE,
> > - MALIDP_DE_LV_STRIDE0, MALIDP550_LV_YUV2RGB, ROTATE_ANY },
> > + MALIDP_DE_LV_STRIDE0, MALIDP550_LV_YUV2RGB, ROTATE_ANY,
> > + MALIDP550_DE_LV2_AD_CTRL },
> > { DE_SMART, MALIDP550_DE_LS_BASE, MALIDP550_DE_LS_PTR_BASE,
> > - MALIDP550_DE_LS_R1_STRIDE, 0, ROTATE_NONE },
> > + MALIDP550_DE_LS_R1_STRIDE, 0, ROTATE_NONE, 0 },
> > };
> >
> > static const struct malidp_layer malidp650_layers[] = {
> > { DE_VIDEO1, MALIDP550_DE_LV1_BASE, MALIDP550_DE_LV1_PTR_BASE,
> > - MALIDP_DE_LV_STRIDE0, MALIDP550_LV_YUV2RGB, ROTATE_ANY },
> > + MALIDP_DE_LV_STRIDE0, MALIDP550_LV_YUV2RGB, ROTATE_ANY,
> > + MALIDP550_DE_LV1_AD_CTRL },
> > { DE_GRAPHICS1, MALIDP550_DE_LG_BASE, MALIDP550_DE_LG_PTR_BASE,
> > - MALIDP_DE_LG_STRIDE, 0, ROTATE_COMPRESSED },
> > + MALIDP_DE_LG_STRIDE, 0, ROTATE_COMPRESSED, MALIDP550_DE_LG_AD_CTRL },
> > { DE_VIDEO2, MALIDP550_DE_LV2_BASE, MALIDP550_DE_LV2_PTR_BASE,
> > - MALIDP_DE_LV_STRIDE0, MALIDP550_LV_YUV2RGB, ROTATE_ANY },
> > + MALIDP_DE_LV_STRIDE0, MALIDP550_LV_YUV2RGB, ROTATE_ANY,
> > + MALIDP550_DE_LV2_AD_CTRL },
> > { DE_SMART, MALIDP550_DE_LS_BASE, MALIDP550_DE_LS_PTR_BASE,
> > - MALIDP550_DE_LS_R1_STRIDE, 0, ROTATE_NONE },
> > + MALIDP550_DE_LS_R1_STRIDE, 0, ROTATE_NONE, 0 },
> > };
> >
> > #define SE_N_SCALING_COEFFS 96
> > diff --git a/drivers/gpu/drm/arm/malidp_hw.h b/drivers/gpu/drm/arm/malidp_hw.h
> > index 4390243..bbe6883 100644
> > --- a/drivers/gpu/drm/arm/malidp_hw.h
> > +++ b/drivers/gpu/drm/arm/malidp_hw.h
> > @@ -67,6 +67,8 @@ struct malidp_layer {
> > u16 stride_offset; /* offset to the first stride register. */
> > s16 yuv2rgb_offset; /* offset to the YUV->RGB matrix entries */
> > enum rotation_features rot; /* type of rotation supported */
> > + /* address offset for the AFBC decoder registers */
> > + u16 afbc_decoder_offset;
> > };
> >
> > enum malidp_scaling_coeff_set {
> > diff --git a/drivers/gpu/drm/arm/malidp_planes.c b/drivers/gpu/drm/arm/malidp_planes.c
> > index 533cdde..3950504 100644
> > --- a/drivers/gpu/drm/arm/malidp_planes.c
> > +++ b/drivers/gpu/drm/arm/malidp_planes.c
> > @@ -330,6 +330,71 @@ static void malidp_de_set_color_encoding(struct malidp_plane *plane,
> > }
> > }
> >
> > +static void malidp_set_plane_base_addr(struct drm_framebuffer *fb,
> > + struct malidp_plane *mp,
> > + int plane_index)
> > +{
> > + dma_addr_t paddr;
> > + u16 ptr;
> > + struct drm_plane *plane = &mp->base;
> > + bool afbc = fb->modifier ? true : false;
>
> The decision to set afbc based on wether the fb has a modifier or not
> seems a bit weak. Should we also (at least) check that the modifier is
> an ARM one?

In my next patch (of this series) ie
https://patchwork.kernel.org/patch/10466537/, we have checked for the
validity of modifier and format in malidp_format_mod_supported(). When
drm_atomic_commit() is invoked, it first calls (indirectly via some
intermediate functions)
malidp_format_mod_supported() and then invokes
malidp_set_plane_base_addr(). So there is no way that an invalid
modifier could be passed to malidp_set_plane_base_addr(). If
malidp_format_mod_supported() returns error then drm_atomic_commit()
would return error too.
>
> > +
> > + ptr = mp->layer->ptr + (plane_index << 4);
> > +
> > + /*
> > + * For AFBC buffers, cropping is handled by AFBC decoder rather than
> > + * pointer manipulation.
> > + */
> > + if (!afbc) {
> > + paddr = drm_fb_cma_get_gem_addr(fb, plane->state,
> > + plane_index);
> > + } else {
> > + struct drm_gem_cma_object *obj;
> > +
> > + obj = drm_fb_cma_get_gem_obj(fb, plane_index);
> > +
> > + if (WARN_ON(!obj))
> > + return;
> > + paddr = obj->paddr;
> > + }
> > +
> > + malidp_hw_write(mp->hwdev, lower_32_bits(paddr), ptr);
> > + malidp_hw_write(mp->hwdev, upper_32_bits(paddr), ptr + 4);
> > +}
> > +
> > +static void malidp_de_set_plane_afbc(struct drm_plane *plane)
> > +{
> > + struct malidp_plane *mp;
> > + u32 src_w, src_h, val = 0, src_x, src_y;
> > + struct drm_framebuffer *fb = plane->state->fb;
> > +
> > + mp = to_malidp_plane(plane);
> > +
> > + /* convert src values from Q16 fixed point to integer */
> > + src_w = plane->state->src_w >> 16;
> > + src_h = plane->state->src_h >> 16;
> > + src_x = plane->state->src_x >> 16;
> > + src_y = plane->state->src_y >> 16;
> > +
> > + val = ((fb->width - (src_x + src_w)) << MALIDP_AD_CROP_RIGHT_OFFSET) |
> > + src_x;
> > + malidp_hw_write(mp->hwdev, val,
> > + mp->layer->afbc_decoder_offset + MALIDP_AD_CROP_H);
> > +
> > + val = ((fb->height - (src_y + src_h)) << MALIDP_AD_CROP_BOTTOM_OFFSET) |
> > + src_y;
> > + malidp_hw_write(mp->hwdev, val,
> > + mp->layer->afbc_decoder_offset + MALIDP_AD_CROP_V);
> > +
> > + val = MALIDP_AD_EN;
> > + if (fb->modifier & AFBC_FORMAT_MOD_SPLIT)
> > + val |= MALIDP_AD_BS;
> > + if (fb->modifier & AFBC_FORMAT_MOD_YTR)
> > + val |= MALIDP_AD_YTR;
> > +
> > + malidp_hw_write(mp->hwdev, val, mp->layer->afbc_decoder_offset);
> > +}
> > +
> > static void malidp_de_plane_update(struct drm_plane *plane,
> > struct drm_plane_state *old_state)
> > {
> > @@ -338,6 +403,7 @@ static void malidp_de_plane_update(struct drm_plane *plane,
> > u32 src_w, src_h, dest_w, dest_h, val;
> > int i;
> > bool format_has_alpha = plane->state->fb->format->has_alpha;
> > + struct drm_framebuffer *fb = plane->state->fb;
> >
> > mp = to_malidp_plane(plane);
> >
> > @@ -349,15 +415,9 @@ static void malidp_de_plane_update(struct drm_plane *plane,
> >
> > malidp_hw_write(mp->hwdev, ms->format, mp->layer->base);
> >
> > - for (i = 0; i < ms->n_planes; i++) {
> > - /* calculate the offset for the layer's plane registers */
> > - u16 ptr = mp->layer->ptr + (i << 4);
> > - dma_addr_t fb_addr = drm_fb_cma_get_gem_addr(plane->state->fb,
> > - plane->state, i);
> > + for (i = 0; i < ms->n_planes; i++)
> > + malidp_set_plane_base_addr(fb, mp, i);
> >
> > - malidp_hw_write(mp->hwdev, lower_32_bits(fb_addr), ptr);
> > - malidp_hw_write(mp->hwdev, upper_32_bits(fb_addr), ptr + 4);
> > - }
> > malidp_de_set_plane_pitches(mp, ms->n_planes,
> > plane->state->fb->pitches);
> >
> > @@ -381,6 +441,11 @@ static void malidp_de_plane_update(struct drm_plane *plane,
> > LAYER_H_VAL(src_w) | LAYER_V_VAL(src_h),
> > mp->layer->base + MALIDP550_LS_R1_IN_SIZE);
> >
> > + if (fb->modifier)
> > + malidp_de_set_plane_afbc(plane);
> > + else
> > + malidp_hw_write(mp->hwdev, 0, mp->layer->afbc_decoder_offset);
>
> Given that you are testing for the non-zero value of fb->modifier inside
> malidp_de_set_plane_afbc() function before setting the afbc_decoder
> register value, I feel that you could take the 'else' branch from here
> and merge it into malidp_de_set_plane_afbc() function (possibly also
> rename it to malidp_de_set_plane).

I am not exactly sure what you meant here. We are testing for the non
zero value of fb->modifier in malidp_de_plane_update(). If
fb->modifier is not zero, we set the afbc specific registers in
malidp_de_set_plane_afbc(). Thus, malidp_de_plane_update() will
invoke malidp_de_set_plane_afbc() for setting afbc stuff and will
continue with the other register settings. I do not see a clear reason
to refactor this code.

> With those small changes:
>
> Reviewed-by: Liviu Dudau <liviu.dudau@xxxxxxx>
>
> Best regards,
> Liviu
>
>
> > +
> > /* first clear the rotation bits */
> > val = malidp_hw_read(mp->hwdev, mp->layer->base + MALIDP_LAYER_CONTROL);
> > val &= ~LAYER_ROT_MASK;
> > diff --git a/drivers/gpu/drm/arm/malidp_regs.h b/drivers/gpu/drm/arm/malidp_regs.h
> > index 149024f..54f4ec5 100644
> > --- a/drivers/gpu/drm/arm/malidp_regs.h
> > +++ b/drivers/gpu/drm/arm/malidp_regs.h
> > @@ -180,10 +180,13 @@
> > #define MALIDP500_LV_YUV2RGB ((s16)(-0xB8))
> > #define MALIDP500_DE_LV_BASE 0x00100
> > #define MALIDP500_DE_LV_PTR_BASE 0x00124
> > +#define MALIDP500_DE_LV_AD_CTRL 0x00400
> > #define MALIDP500_DE_LG1_BASE 0x00200
> > #define MALIDP500_DE_LG1_PTR_BASE 0x0021c
> > +#define MALIDP500_DE_LG1_AD_CTRL 0x0040c
> > #define MALIDP500_DE_LG2_BASE 0x00300
> > #define MALIDP500_DE_LG2_PTR_BASE 0x0031c
> > +#define MALIDP500_DE_LG2_AD_CTRL 0x00418
> > #define MALIDP500_SE_BASE 0x00c00
> > #define MALIDP500_SE_CONTROL 0x00c0c
> > #define MALIDP500_SE_PTR_BASE 0x00e0c
> > @@ -208,10 +211,13 @@
> > #define MALIDP550_LV_YUV2RGB 0x00084
> > #define MALIDP550_DE_LV1_BASE 0x00100
> > #define MALIDP550_DE_LV1_PTR_BASE 0x00124
> > +#define MALIDP550_DE_LV1_AD_CTRL 0x001B8
> > #define MALIDP550_DE_LV2_BASE 0x00200
> > #define MALIDP550_DE_LV2_PTR_BASE 0x00224
> > +#define MALIDP550_DE_LV2_AD_CTRL 0x002B8
> > #define MALIDP550_DE_LG_BASE 0x00300
> > #define MALIDP550_DE_LG_PTR_BASE 0x0031c
> > +#define MALIDP550_DE_LG_AD_CTRL 0x00330
> > #define MALIDP550_DE_LS_BASE 0x00400
> > #define MALIDP550_DE_LS_PTR_BASE 0x0042c
> > #define MALIDP550_DE_PERF_BASE 0x00500
> > @@ -223,6 +229,20 @@
> > #define MALIDP550_CONFIG_VALID 0x0c014
> > #define MALIDP550_CONFIG_ID 0x0ffd4
> >
> > +/* AFBC register offsets relative to MALIDPXXX_DE_LX_AD_CTRL */
> > +/* The following register offsets are common for DP500, DP550 and DP650 */
> > +#define MALIDP_AD_CROP_H 0x4
> > +#define MALIDP_AD_CROP_V 0x8
> > +#define MALIDP_AD_END_PTR_LOW 0xc
> > +#define MALIDP_AD_END_PTR_HIGH 0x10
> > +
> > +/* AFBC decoder Registers */
> > +#define MALIDP_AD_EN BIT(0)
> > +#define MALIDP_AD_YTR BIT(4)
> > +#define MALIDP_AD_BS BIT(8)
> > +#define MALIDP_AD_CROP_RIGHT_OFFSET 16
> > +#define MALIDP_AD_CROP_BOTTOM_OFFSET 16
> > +
> > /*
> > * Starting with DP550 the register map blocks has been standardised to the
> > * following layout:
> > --
> > 2.7.4
> >
>
> --
> ====================
> | I would like to |
> | fix the world, |
> | but they're not |
> | giving me the |
> \ source code! /
> ---------------
> ??\_(???)_/??
> _______________________________________________
> dri-devel mailing list
> dri-devel@xxxxxxxxxxxxxxxxxxxxx
> https://lists.freedesktop.org/mailman/listinfo/dri-devel

Next message: Vincent Guittot: "Re: [PATCH v7 00/11] track CPU utilization"
Previous message: Laurentiu Tudor: "[PATCH] docs: kernel-parameters.txt: document xhci-hcd.quirks parameter"
Next in thread: Liviu Dudau: "Re: [RFC PATCH 3/4] drm/arm/malidp: Set the AFBC register bits if the framebuffer has AFBC modifier"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]