[Intel-gfx] [PATCH 6/6] drm/i915/tgl: Add Clear Color supoort for TGL Render Decompression
Sripada, Radhakrishna
radhakrishna.sripada at intel.com
Thu Sep 19 22:16:34 UTC 2019
Hi Matt,
> -----Original Message-----
> From: Roper, Matthew D
> Sent: Tuesday, September 17, 2019 2:53 PM
> To: Sripada, Radhakrishna <radhakrishna.sripada at intel.com>
> Cc: intel-gfx at lists.freedesktop.org; Pandiyan, Dhinakaran
> <dhinakaran.pandiyan at intel.com>; Syrjala, Ville <ville.syrjala at intel.com>;
> Sharma, Shashank <shashank.sharma at intel.com>; Antognolli, Rafael
> <rafael.antognolli at intel.com>; Chery, Nanley G <nanley.g.chery at intel.com>
> Subject: Re: [PATCH 6/6] drm/i915/tgl: Add Clear Color supoort for TGL
> Render Decompression
>
> On Tue, Sep 17, 2019 at 05:11:55AM -0700, Radhakrishna Sripada wrote:
> > Render Decompression is supported with Y-Tiled main surface. The CCS
> > is linear and has 4 bits of data for each main surface cache line
> > pair, a ratio of 1:256. Additional Clear Color information is passed
> > from the user-space through an offset in the GEM BO. Add a new
> > modifier to identify and parse new Clear Color information and extend
> > Gen12 render decompression functionality to the newly added modifier.
> >
> > Cc: Dhinakaran Pandiyan <dhinakaran.pandiyan at intel.com>
> > Cc: Ville Syrjala <ville.syrjala at intel.com>
> > Cc: Shashank Sharma <shashank.sharma at intel.com>
> > Cc: Rafael Antognolli <rafael.antognolli at intel.com>
> > Cc: Matt Roper <matthew.d.roper at intel.com>
> > Cc: Nanley G Chery <nanley.g.chery at intel.com>
> > Signed-off-by: Radhakrishna Sripada <radhakrishna.sripada at intel.com>
> > ---
> > drivers/gpu/drm/i915/display/intel_display.c | 47 +++++++++++++++++--
> > .../drm/i915/display/intel_display_types.h | 3 ++
> > drivers/gpu/drm/i915/display/intel_sprite.c | 10 +++-
> > drivers/gpu/drm/i915/i915_reg.h | 13 +++++
> > 4 files changed, 69 insertions(+), 4 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/i915/display/intel_display.c
> > b/drivers/gpu/drm/i915/display/intel_display.c
> > index 2da721a6abab..725b9724da49 100644
> > --- a/drivers/gpu/drm/i915/display/intel_display.c
> > +++ b/drivers/gpu/drm/i915/display/intel_display.c
> > @@ -1913,6 +1913,7 @@ intel_tile_width_bytes(const struct
> drm_framebuffer *fb, int color_plane)
> > /* fall through */
> > case I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS:
> > case I915_FORMAT_MOD_Y_TILED_GEN12_MC_CCS:
> > + case I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS_CC:
> > if (color_plane == 1)
> > return cpp;
> > /* fall through */
> > @@ -2051,6 +2052,7 @@ static unsigned int intel_surf_alignment(const
> struct drm_framebuffer *fb,
> > return 0;
> > case I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS:
> > case I915_FORMAT_MOD_Y_TILED_GEN12_MC_CCS:
> > + case I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS_CC:
> > return 4 * 4 * 1024;
> > case I915_FORMAT_MOD_Y_TILED_CCS:
> > case I915_FORMAT_MOD_Yf_TILED_CCS:
> > @@ -2256,6 +2258,7 @@ static bool is_surface_linear(u64 modifier, int
> color_plane)
> > return true;
> > case I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS:
> > case I915_FORMAT_MOD_Y_TILED_GEN12_MC_CCS:
> > + case I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS_CC:
> > return color_plane == 1;
> > default:
> > return false;
> > @@ -2448,6 +2451,7 @@ static unsigned int
> intel_fb_modifier_to_tiling(u64 fb_modifier)
> > case I915_FORMAT_MOD_Y_TILED_CCS:
> > case I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS:
> > case I915_FORMAT_MOD_Y_TILED_GEN12_MC_CCS:
> > + case I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS_CC:
> > return I915_TILING_Y;
> > default:
> > return I915_TILING_NONE;
> > @@ -2497,6 +2501,21 @@ static const struct drm_format_info
> gen12_ccs_formats[] = {
> > .cpp = { 4, 1, }, .hsub = 2, .vsub = 32, .has_alpha = true }, };
> >
> > +/*
> > + * Gen-12 compression uses 4 bits of CCS data for each cache line
> > +pair in the
> > + * main surface. And each 64B CCS cache line represents an area of
> > +4x1 Y-tiles
> > + * in the main surface. With 4 byte pixels and each Y-tile having
> > +dimensions of
> > + * 32x32 pixels, the ratio turns out to 1B in the CCS for every 2 x
> > +32 pixels in
> > + * the main surface. Additional surface is used to pass the Clear
> > +Color
> > + * structure for the driver to program the DE.
> > + */
> > +static const struct drm_format_info gen12_ccs_cc_formats[] = {
> > + { .format = DRM_FORMAT_XRGB8888, .depth = 24, .num_planes = 3,
> .cpp = { 4, 1, 0}, .hsub = 2, .vsub = 32, },
> > + { .format = DRM_FORMAT_XBGR8888, .depth = 24, .num_planes = 3,
> .cpp = { 4, 1, 0}, .hsub = 2, .vsub = 32, },
> > + { .format = DRM_FORMAT_ARGB8888, .depth = 32, .num_planes = 3,
> .cpp = { 4, 1, 0}, .hsub = 2, .vsub = 32, },
> > + { .format = DRM_FORMAT_ABGR8888, .depth = 32, .num_planes = 3,
> .cpp
> > += { 4, 1, 0}, .hsub = 2, .vsub = 32, },
>
> I think these last two were supposed to have .has_alpha = true?
Sure will include in the next rev.
>
> > +};
> > +
> > static const struct drm_format_info * lookup_format_info(const
> > struct drm_format_info formats[],
> > int num_formats, u32 format)
> > @@ -2525,6 +2544,10 @@ intel_get_format_info(const struct
> drm_mode_fb_cmd2 *cmd)
> > return lookup_format_info(gen12_ccs_formats,
> > ARRAY_SIZE(gen12_ccs_formats),
> > cmd->pixel_format);
> > + case I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS_CC:
> > + return lookup_format_info(gen12_ccs_formats,
> > +
> ARRAY_SIZE(gen12_ccs_cc_formats),
> > + cmd->pixel_format);
> > default:
> > return NULL;
> > }
> > @@ -2534,6 +2557,7 @@ bool is_ccs_modifier(u64 modifier) {
> > return modifier == I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS ||
> > modifier == I915_FORMAT_MOD_Y_TILED_GEN12_MC_CCS ||
> > + modifier == I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS_CC
> ||
> > modifier == I915_FORMAT_MOD_Y_TILED_CCS ||
> > modifier == I915_FORMAT_MOD_Yf_TILED_CCS; } @@ -4120,6
> > +4144,9 @@ static u32 skl_plane_ctl_tiling(u64 fb_modifier)
> > case I915_FORMAT_MOD_Y_TILED_CCS:
> > /* fall through */
> > case I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS:
> > + return PLANE_CTL_TILED_Y |
> PLANE_CTL_RENDER_DECOMPRESSION_ENABLE |
> > + ICL_PLANE_CTL_CLEAR_COLOR_DISABLE;
>
> I think this needs to be squashed back into the previous patch that added
> general gen12 CCS.
Sure. I will work with DK to get this changes.
>
>
> > + case I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS_CC:
> > return PLANE_CTL_TILED_Y |
> PLANE_CTL_RENDER_DECOMPRESSION_ENABLE;
> > case I915_FORMAT_MOD_Y_TILED_GEN12_MC_CCS:
> > return PLANE_CTL_TILED_Y |
> PLANE_CTL_MEDIA_DECOMPRESSION_ENABLE;
> > @@ -9897,9 +9924,13 @@ skylake_get_initial_plane_config(struct
> intel_crtc *crtc,
> > case PLANE_CTL_TILED_Y:
> > plane_config->tiling = I915_TILING_Y;
> > if (val & PLANE_CTL_RENDER_DECOMPRESSION_ENABLE)
> > - fb->modifier = INTEL_GEN(dev_priv) >= 12 ?
> > -
> I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS :
> > - I915_FORMAT_MOD_Y_TILED_CCS;
> > + if (INTEL_GEN(dev_priv) >= 12)
> > + fb->modifier = val &
> > +
> ICL_PLANE_CTL_CLEAR_COLOR_DISABLE ?
> > +
> I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS :
> > +
> I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS_CC;
> > + else
> > + fb->modifier =
> I915_FORMAT_MOD_Y_TILED_CCS;
> > else if (val &
> PLANE_CTL_RENDER_DECOMPRESSION_ENABLE)
> > fb->modifier =
> I915_FORMAT_MOD_Y_TILED_GEN12_MC_CCS;
> > else
> > @@ -14322,6 +14353,15 @@ static int intel_plane_pin_fb(struct
> > intel_plane_state *plane_state)
> >
> > plane_state->vma = vma;
> >
> > + if (fb->modifier ==
> I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS_CC) {
> > + u32 *ccaddr =
> kmap_atomic(i915_gem_object_get_page(intel_fb_obj(fb),
> > + fb-
> >offsets[2] >> PAGE_SHIFT));
> > +
>
> Will this work on framebuffers in stolen memory? Not that it's likely a regular
> BIOS would actually setup compressed framebuffers to begin with...
It is unlikely that BIOS sets up a compressed fb. However FBC could impact as it works with
Stolen memory. I am not sure about the interaction between fbc and render compression
working at the same time. Apart from that I am unable to think of an issue. Will try to figure
out during testing.
>
> > + plane_state->ccval = ((u64)*(ccaddr +
> CC_VAL_HIGHER_OFFSET) << 32)
> > + | *(ccaddr + CC_VAL_LOWER_OFFSET);
>
> Is there more data in the clear color page besides the color itself? If so,
> maybe we should just use a struct to make the memory layout explicit?
According to BSpec: 43854 the first 3 offsets are for the raw color which acts as an input to
3d engine which fills the info in the later 2 offsets in DE compatible format. The other data
Therefore is redundant and cannot be consumed by the DE hence ignoring.
>
> > + kunmap_atomic(ccaddr);
> > + }
> > +
> > return 0;
> > }
> >
> > @@ -15709,6 +15749,7 @@ static int intel_framebuffer_init(struct
> intel_framebuffer *intel_fb,
> > * tile widths.
> > */
> > if ((fb->modifier ==
> I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS ||
> > + fb->modifier ==
> I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS_CC ||
> > fb->modifier ==
> I915_FORMAT_MOD_Y_TILED_GEN12_MC_CCS) &&
> > i == 0)
> > stride_alignment *= 4;
> > diff --git a/drivers/gpu/drm/i915/display/intel_display_types.h
> > b/drivers/gpu/drm/i915/display/intel_display_types.h
> > index d5cc4b810d9e..a82695a3e0b8 100644
> > --- a/drivers/gpu/drm/i915/display/intel_display_types.h
> > +++ b/drivers/gpu/drm/i915/display/intel_display_types.h
> > @@ -579,6 +579,9 @@ struct intel_plane_state {
> > u32 slave;
> >
> > struct drm_intel_sprite_colorkey ckey;
> > +
> > + /* Clear Color Value */
> > + u64 ccval;
> > };
> >
> > struct intel_initial_plane_config {
> > diff --git a/drivers/gpu/drm/i915/display/intel_sprite.c
> > b/drivers/gpu/drm/i915/display/intel_sprite.c
> > index 1655984955ca..3672e8b9c0bc 100644
> > --- a/drivers/gpu/drm/i915/display/intel_sprite.c
> > +++ b/drivers/gpu/drm/i915/display/intel_sprite.c
> > @@ -549,6 +549,7 @@ skl_program_plane(struct intel_plane *plane,
> > u32 plane_color_ctl = 0;
> > unsigned long irqflags;
> > u32 keymsk, keymax;
> > + u64 ccval = plane_state->ccval;
> >
> > plane_ctl |= skl_plane_ctl_crtc(crtc_state);
> >
> > @@ -609,6 +610,10 @@ skl_program_plane(struct intel_plane *plane,
> > if (fb->format->is_yuv && icl_is_hdr_plane(dev_priv, plane_id))
> > icl_program_input_csc(plane, crtc_state, plane_state);
> >
> > + if (fb->modifier ==
> I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS_CC)
> > + intel_uncore_write64_fw(&dev_priv->uncore,
> > + PLANE_CC_VAL(pipe, plane_id),
> ccval);
> > +
> > skl_write_plane_wm(plane, crtc_state);
> >
> > I915_WRITE_FW(PLANE_KEYVAL(pipe, plane_id), key->min_value);
> @@
> > -1738,7 +1743,8 @@ static int skl_plane_check_fb(const struct
> intel_crtc_state *crtc_state,
> > fb->modifier == I915_FORMAT_MOD_Y_TILED_CCS ||
> > fb->modifier == I915_FORMAT_MOD_Yf_TILED_CCS ||
> > fb->modifier == I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS ||
> > - fb->modifier == I915_FORMAT_MOD_Y_TILED_GEN12_MC_CCS))
> {
> > + fb->modifier == I915_FORMAT_MOD_Y_TILED_GEN12_MC_CCS
> ||
> > + fb->modifier ==
> I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS_CC)) {
> > DRM_DEBUG_KMS("Y/Yf tiling not supported in IF-ID
> mode\n");
> > return -EINVAL;
> > }
> > @@ -2153,6 +2159,7 @@ static const u64
> > skl_plane_format_modifiers_ccs[] = { static const u64
> gen12_plane_format_modifiers_ccs[] = {
> > I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS,
> > I915_FORMAT_MOD_Y_TILED_GEN12_MC_CCS,
> > + I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS_CC,
> > I915_FORMAT_MOD_Y_TILED,
> > I915_FORMAT_MOD_X_TILED,
> > DRM_FORMAT_MOD_LINEAR,
> > @@ -2321,6 +2328,7 @@ static bool
> gen12_plane_format_mod_supported(struct drm_plane *_plane,
> > case I915_FORMAT_MOD_X_TILED:
> > case I915_FORMAT_MOD_Y_TILED:
> > case I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS:
> > + case I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS_CC:
> > break;
> > default:
> > return false;
> > diff --git a/drivers/gpu/drm/i915/i915_reg.h
> > b/drivers/gpu/drm/i915/i915_reg.h index 096be95ec7f9..00c59ed8b114
> > 100644
> > --- a/drivers/gpu/drm/i915/i915_reg.h
> > +++ b/drivers/gpu/drm/i915/i915_reg.h
> > @@ -6681,6 +6681,7 @@ enum {
> > #define PLANE_CTL_YUV422_VYUY (3 << 16)
> > #define PLANE_CTL_RENDER_DECOMPRESSION_ENABLE (1 << 15)
> > #define PLANE_CTL_TRICKLE_FEED_DISABLE (1 << 14)
> > +#define ICL_PLANE_CTL_CLEAR_COLOR_DISABLE (1 << 13)
>
> Looks like the whitespace is off here.
Will fix it next rev.
Thanks,
RK
>
>
> Matt
>
> > #define PLANE_CTL_PLANE_GAMMA_DISABLE (1 << 13) /*
> Pre-GLK */
> > #define PLANE_CTL_TILED_MASK (0x7 << 10)
> > #define PLANE_CTL_TILED_LINEAR (0 << 10)
> > @@ -6721,6 +6722,8 @@ enum {
> > #define _PLANE_KEYMAX_1_A 0x701a0
> > #define _PLANE_KEYMAX_2_A 0x702a0
> > #define PLANE_KEYMAX_ALPHA(a) ((a) << 24)
> > +#define _PLANE_CC_VAL_1_A 0x701b4
> > +#define _PLANE_CC_VAL_2_A 0x702b4
> > #define _PLANE_AUX_DIST_1_A 0x701c0
> > #define _PLANE_AUX_DIST_2_A 0x702c0
> > #define _PLANE_AUX_OFFSET_1_A 0x701c4
> > @@ -6760,6 +6763,16 @@ enum {
> > #define _PLANE_NV12_BUF_CFG_1_A 0x70278
> > #define _PLANE_NV12_BUF_CFG_2_A 0x70378
> >
> > +#define _PLANE_CC_VAL_1_B 0x711b4
> > +#define _PLANE_CC_VAL_2_B 0x712b4
> > +#define _PLANE_CC_VAL_1(pipe) _PIPE(pipe, _PLANE_CC_VAL_1_A,
> _PLANE_CC_VAL_1_B)
> > +#define _PLANE_CC_VAL_2(pipe) _PIPE(pipe, _PLANE_CC_VAL_2_A,
> _PLANE_CC_VAL_2_B)
> > +#define PLANE_CC_VAL(pipe, plane) \
> > + _MMIO_PLANE(plane, _PLANE_CC_VAL_1(pipe),
> _PLANE_CC_VAL_2(pipe))
> > +
> > +#define CC_VAL_LOWER_OFFSET 4
> > +#define CC_VAL_HIGHER_OFFSET 5
> > +
> > /* Input CSC Register Definitions */
> > #define _PLANE_INPUT_CSC_RY_GY_1_A 0x701E0
> > #define _PLANE_INPUT_CSC_RY_GY_2_A 0x702E0
> > --
> > 2.20.1
> >
>
> --
> Matt Roper
> Graphics Software Engineer
> VTT-OSGC Platform Enablement
> Intel Corporation
> (916) 356-2795
More information about the Intel-gfx
mailing list