[Intel-gfx] [PATCH v3 11/11] drm/i915/tgl: Add Clear Color supoort for TGL Render Decompression
Dhinakaran Pandiyan
dhinakaran.pandiyan at intel.com
Sat Oct 5 00:17:07 UTC 2019
On Fri, 2019-10-04 at 16:52 -0700, Matt Roper wrote:
> On Fri, Sep 27, 2019 at 03:28:37PM -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.
> >
> > v2: Fix has_alpha flag for modifiers, omit CC modifier during initial
> > plane config(Matt). Fix Lookup error.
> > v3: Fix the panic while running kms_cube
> >
> > 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 | 52 +++++++++++++++++++
> > .../drm/i915/display/intel_display_types.h | 3 ++
> > drivers/gpu/drm/i915/display/intel_sprite.c | 11 +++-
> > drivers/gpu/drm/i915/i915_reg.h | 12 +++++
> > 4 files changed, 77 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/gpu/drm/i915/display/intel_display.c
> > b/drivers/gpu/drm/i915/display/intel_display.c
> > index 4971c296f951..822237e98f00 100644
> > --- a/drivers/gpu/drm/i915/display/intel_display.c
> > +++ b/drivers/gpu/drm/i915/display/intel_display.c
> > @@ -1919,6 +1919,10 @@ intel_tile_width_bytes(const struct drm_framebuffer *fb, int color_plane)
> > if (color_plane == 1)
> > return 64;
> > /* fall through */
> > + case I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS_CC:
> > + if (color_plane == 1 || color_plane == 2)
> > + return 64;
> > + /* fall through */
> > case I915_FORMAT_MOD_Y_TILED:
> > if (IS_GEN(dev_priv, 2) || HAS_128_BYTE_Y_TILING(dev_priv))
> > return 128;
> > @@ -2060,6 +2064,7 @@ static unsigned int intel_surf_alignment(const struct drm_framebuffer *fb,
> > return 256 * 1024;
> > return 0;
> > case I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS:
> > + case I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS_CC:
> > return 16 * 1024;
> > case I915_FORMAT_MOD_Y_TILED_CCS:
> > case I915_FORMAT_MOD_Yf_TILED_CCS:
> > @@ -2265,6 +2270,8 @@ static bool is_surface_linear(u64 modifier, int color_plane)
> > return true;
> > case I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS:
> > return color_plane == 1;
> > + case I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS_CC:
> > + return color_plane == 1 || color_plane == 2;
> > case I915_FORMAT_MOD_Y_TILED_GEN12_MC_CCS:
> > return color_plane == 1 || color_plane == 3;
> > default:
> > @@ -2458,6 +2465,7 @@ 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;
> > @@ -2511,6 +2519,25 @@ static const struct drm_format_info gen12_ccs_formats[] = {
> > .cpp = { 1, 1, 2, 1}, .hsub = 2, .vsub = 2, .is_yuv = 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.
> > + */
>
> Rather than duplicating the previous comment's text I'd just say
>
> "Same as gen12_ccs_formats[] above, but with an additional surface used
> to pass..."
>
> > +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, .has_alpha = true },
> > + { .format = DRM_FORMAT_ABGR8888, .depth = 32, .num_planes = 3,
> > + .cpp = { 4, 1, 0, }, .hsub = 2, .vsub = 32, .has_alpha = true },
> > +};
> > +
> > static const struct drm_format_info *
> > lookup_format_info(const struct drm_format_info formats[],
> > int num_formats, u32 format)
> > @@ -2538,6 +2565,10 @@ 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_cc_formats,
> > + ARRAY_SIZE(gen12_ccs_cc_formats),
> > + cmd->pixel_format);
> > default:
> > return NULL;
> > }
> > @@ -2547,6 +2578,7 @@ 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;
> > }
> > @@ -2734,6 +2766,8 @@ static bool is_ccs_plane(u64 modifier, int color_plane)
> > return false;
> > else if (modifier == I915_FORMAT_MOD_Y_TILED_GEN12_MC_CCS)
> > return color_plane == 3 || color_plane == 1;
> > + else if (modifier == I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS_CC)
> > + return color_plane == 1 || color_plane == 2;
> > else
> > return color_plane == 1;
> > }
> > @@ -2805,6 +2839,14 @@ intel_fill_fb_info(struct drm_i915_private *dev_priv,
> > cpp = fb->format->cpp[i];
> > intel_fb_plane_dims(&width, &height, fb, i);
> >
> > + /*
> > + * Plane 2 of Render Compression with Clear Color fb modifier is consumed
> > + * by the driver and not passed to DE. Skip the arithmetic related to
> > + * alignment and offset calculation.
> > + */
> > + if (i == 2 && fb->modifier == I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS_CC)
> > + continue;
> > +
>
> We might as well do this before the cpp and width/height assignments too?
>
> Should we also be checking IS_ALIGNED(offsets[2], PAGE_SIZE)? We seem
> to rely on that after kmap'ing below.
I am wondering if it needs to be page aligned. Since Clear Color data follows a linear CCS, would a
64B alignment suffice?
-DK
>
>
> Matt
>
> > ret = intel_fb_offset_to_xy(&x, &y, fb, i);
> > if (ret) {
> > DRM_DEBUG_KMS("bad fb plane %d offset: 0x%x\n",
> > @@ -4274,6 +4316,7 @@ static u32 skl_plane_ctl_tiling(u64 fb_modifier)
> > case I915_FORMAT_MOD_Y_TILED:
> > return PLANE_CTL_TILED_Y;
> > case I915_FORMAT_MOD_Y_TILED_CCS:
> > + 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_RC_CCS:
> > return PLANE_CTL_TILED_Y |
> > @@ -14536,6 +14579,15 @@ 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));
> > +
> > + plane_state->ccval = ((u64)*(ccaddr + CC_VAL_HIGHER_OFFSET) << 32)
> > + | *(ccaddr + CC_VAL_LOWER_OFFSET);
> > + kunmap_atomic(ccaddr);
> > + }
> > +
> > return 0;
> > }
> >
> > diff --git a/drivers/gpu/drm/i915/display/intel_display_types.h
> > b/drivers/gpu/drm/i915/display/intel_display_types.h
> > index 5998b959225c..77d3f6b634b0 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 planar_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 ce79373154fc..24ae8bf9744d 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_mc_ccs[] = {
> > I915_FORMAT_MOD_Y_TILED_GEN12_MC_CCS,
> > I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS,
> > + I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS_CC,
> > I915_FORMAT_MOD_Y_TILED,
> > I915_FORMAT_MOD_X_TILED,
> > DRM_FORMAT_MOD_LINEAR,
> > @@ -2161,6 +2168,7 @@ static const u64 gen12_plane_format_modifiers_mc_ccs[] = {
> >
> > static const u64 gen12_plane_format_modifiers_rc_ccs[] = {
> > I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS,
> > + I915_FORMAT_MOD_Y_TILED_GEN12_RC_CCS_CC,
> > I915_FORMAT_MOD_Y_TILED,
> > I915_FORMAT_MOD_X_TILED,
> > DRM_FORMAT_MOD_LINEAR,
> > @@ -2334,6 +2342,7 @@ 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 0abf9a3917ed..01023fc880c8 100644
> > --- a/drivers/gpu/drm/i915/i915_reg.h
> > +++ b/drivers/gpu/drm/i915/i915_reg.h
> > @@ -6734,6 +6734,8 @@ 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
> > @@ -6773,6 +6775,16 @@ 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
> >
>
>
More information about the Intel-gfx
mailing list