[Mesa-dev] [PATCH 06/11] r600: add core pieces of image support.
Dave Airlie
airlied at gmail.com
Wed Nov 15 00:11:07 UTC 2017
From: Dave Airlie <airlied at redhat.com>
This adds the atoms and gallium api implementations,
along with support for compress/decompress paths for
shader images.
Signed-off-by: Dave Airlie <airlied at redhat.com>
---
src/gallium/drivers/r600/evergreen_state.c | 294 ++++++++++++++++++++++++++-
src/gallium/drivers/r600/r600_blit.c | 59 ++++++
src/gallium/drivers/r600/r600_hw_context.c | 2 +
src/gallium/drivers/r600/r600_pipe.h | 45 +++-
src/gallium/drivers/r600/r600_shader.h | 1 +
src/gallium/drivers/r600/r600_state_common.c | 30 +++
6 files changed, 428 insertions(+), 3 deletions(-)
diff --git a/src/gallium/drivers/r600/evergreen_state.c b/src/gallium/drivers/r600/evergreen_state.c
index 6920b74..ae7dc76 100644
--- a/src/gallium/drivers/r600/evergreen_state.c
+++ b/src/gallium/drivers/r600/evergreen_state.c
@@ -1673,6 +1673,104 @@ static void evergreen_emit_msaa_state(struct r600_context *rctx, int nr_samples,
}
}
+static void evergreen_emit_image_state(struct r600_context *rctx, struct r600_atom *atom,
+ int immed_id_base, int res_id_base)
+{
+ struct r600_image_state *state = (struct r600_image_state *)atom;
+ struct pipe_framebuffer_state *fb_state = &rctx->framebuffer.state;
+ struct radeon_winsys_cs *cs = rctx->b.gfx.cs;
+ struct r600_texture *rtex;
+ struct r600_resource *resource;
+ uint32_t pkt_flags = 0;
+ int i;
+
+ for (i = 0; i < R600_MAX_IMAGES; i++) {
+ struct r600_image_view *image = &state->views[i];
+ unsigned reloc, immed_reloc;
+ int idx = i;
+
+ idx += fb_state->nr_cbufs + (rctx->dual_src_blend ? 1 : 0);
+ if (!image->base.resource)
+ continue;
+
+ resource = (struct r600_resource *)image->base.resource;
+ if (resource->b.b.target != PIPE_BUFFER)
+ rtex = (struct r600_texture *)image->base.resource;
+ else
+ rtex = NULL;
+
+ reloc = radeon_add_to_buffer_list(&rctx->b,
+ &rctx->b.gfx,
+ resource,
+ RADEON_USAGE_READWRITE,
+ RADEON_PRIO_SHADER_RW_BUFFER);
+
+ immed_reloc = radeon_add_to_buffer_list(&rctx->b,
+ &rctx->b.gfx,
+ resource->immed_buffer,
+ RADEON_USAGE_READWRITE,
+ RADEON_PRIO_SHADER_RW_BUFFER);
+
+ radeon_set_context_reg_seq(cs, R_028C60_CB_COLOR0_BASE + idx * 0x3C, 13);
+
+ radeon_emit(cs, image->cb_color_base); /* R_028C60_CB_COLOR0_BASE */
+ radeon_emit(cs, image->cb_color_pitch); /* R_028C64_CB_COLOR0_PITCH */
+ radeon_emit(cs, image->cb_color_slice); /* R_028C68_CB_COLOR0_SLICE */
+ radeon_emit(cs, image->cb_color_view); /* R_028C6C_CB_COLOR0_VIEW */
+ radeon_emit(cs, image->cb_color_info); /* R_028C70_CB_COLOR0_INFO */
+ radeon_emit(cs, image->cb_color_attrib); /* R_028C74_CB_COLOR0_ATTRIB */
+ radeon_emit(cs, image->cb_color_dim); /* R_028C78_CB_COLOR0_DIM */
+ radeon_emit(cs, rtex ? rtex->cmask.base_address_reg : image->cb_color_base); /* R_028C7C_CB_COLOR0_CMASK */
+ radeon_emit(cs, rtex ? rtex->cmask.slice_tile_max : 0); /* R_028C80_CB_COLOR0_CMASK_SLICE */
+ radeon_emit(cs, image->cb_color_fmask); /* R_028C84_CB_COLOR0_FMASK */
+ radeon_emit(cs, image->cb_color_fmask_slice); /* R_028C88_CB_COLOR0_FMASK_SLICE */
+ radeon_emit(cs, rtex ? rtex->color_clear_value[0] : 0); /* R_028C8C_CB_COLOR0_CLEAR_WORD0 */
+ radeon_emit(cs, rtex ? rtex->color_clear_value[1] : 0); /* R_028C90_CB_COLOR0_CLEAR_WORD1 */
+
+ radeon_emit(cs, PKT3(PKT3_NOP, 0, 0)); /* R_028C60_CB_COLOR0_BASE */
+ radeon_emit(cs, reloc);
+
+ radeon_emit(cs, PKT3(PKT3_NOP, 0, 0)); /* R_028C74_CB_COLOR0_ATTRIB */
+ radeon_emit(cs, reloc);
+
+ radeon_emit(cs, PKT3(PKT3_NOP, 0, 0)); /* R_028C7C_CB_COLOR0_CMASK */
+ radeon_emit(cs, reloc);
+
+ radeon_emit(cs, PKT3(PKT3_NOP, 0, 0)); /* R_028C84_CB_COLOR0_FMASK */
+ radeon_emit(cs, reloc);
+
+ radeon_set_context_reg(cs, R_028B9C_CB_IMMED0_BASE + (idx * 4), resource->immed_buffer->gpu_address >> 8);
+ radeon_emit(cs, PKT3(PKT3_NOP, 0, 0)); /**/
+ radeon_emit(cs, immed_reloc);
+
+ radeon_emit(cs, PKT3(PKT3_SET_RESOURCE, 8, 0) | pkt_flags);
+ radeon_emit(cs, (immed_id_base + i) * 8);
+ radeon_emit_array(cs, image->immed_resource_words, 8);
+
+ radeon_emit(cs, PKT3(PKT3_NOP, 0, 0) | pkt_flags);
+ radeon_emit(cs, immed_reloc);
+
+ radeon_emit(cs, PKT3(PKT3_SET_RESOURCE, 8, 0) | pkt_flags);
+ radeon_emit(cs, (res_id_base + i) * 8);
+ radeon_emit_array(cs, image->resource_words, 8);
+
+ radeon_emit(cs, PKT3(PKT3_NOP, 0, 0) | pkt_flags);
+ radeon_emit(cs, reloc);
+
+ if (!image->skip_mip_address_reloc) {
+ radeon_emit(cs, PKT3(PKT3_NOP, 0, 0) | pkt_flags);
+ radeon_emit(cs, reloc);
+ }
+ }
+}
+
+static void evergreen_emit_fragment_image_state(struct r600_context *rctx, struct r600_atom *atom)
+{
+ evergreen_emit_image_state(rctx, atom,
+ R600_IMAGE_IMMED_RESOURCE_OFFSET,
+ R600_IMAGE_REAL_RESOURCE_OFFSET);
+}
+
static void evergreen_emit_framebuffer_state(struct r600_context *rctx, struct r600_atom *atom)
{
struct radeon_winsys_cs *cs = rctx->b.gfx.cs;
@@ -1748,6 +1846,7 @@ static void evergreen_emit_framebuffer_state(struct r600_context *rctx, struct r
cb->cb_color_info | tex->cb_color_info);
i++;
}
+ i += util_bitcount(rctx->fragment_images.enabled_mask);
for (; i < 8 ; i++)
radeon_set_context_reg(cs, R_028C70_CB_COLOR0_INFO + i * 0x3C, 0);
for (; i < 12; i++)
@@ -1862,9 +1961,9 @@ static void evergreen_emit_cb_misc_state(struct r600_context *rctx, struct r600_
struct r600_cb_misc_state *a = (struct r600_cb_misc_state*)atom;
unsigned fb_colormask = (1ULL << ((unsigned)a->nr_cbufs * 4)) - 1;
unsigned ps_colormask = (1ULL << ((unsigned)a->nr_ps_color_outputs * 4)) - 1;
-
+ unsigned rat_colormask = ((1ULL << ((unsigned)a->nr_image_rats * 4)) - 1) << (a->nr_cbufs * 4);
radeon_set_context_reg_seq(cs, R_028238_CB_TARGET_MASK, 2);
- radeon_emit(cs, a->blend_colormask & fb_colormask); /* R_028238_CB_TARGET_MASK */
+ radeon_emit(cs, (a->blend_colormask & fb_colormask) | rat_colormask); /* R_028238_CB_TARGET_MASK */
/* This must match the used export instructions exactly.
* Other values may lead to undefined behavior and hangs.
*/
@@ -3761,6 +3860,195 @@ static void evergreen_set_hw_atomic_buffers(struct pipe_context *ctx,
}
}
+static void evergreen_set_shader_images(struct pipe_context *ctx,
+ enum pipe_shader_type shader, unsigned start_slot,
+ unsigned count,
+ const struct pipe_image_view *images)
+{
+ struct r600_context *rctx = (struct r600_context *)ctx;
+ struct r600_screen *rscreen = (struct r600_screen *)ctx->screen;
+ int i;
+ struct r600_image_view *rview;
+ struct pipe_resource *image;
+ struct r600_resource *resource;
+ struct r600_tex_color_info color;
+ struct eg_buf_res_params buf_params;
+ struct eg_tex_res_params tex_params;
+ unsigned old_mask;
+ bool skip_reloc = false;
+ struct r600_image_state *istate = NULL;
+ int idx;
+ if (shader != PIPE_SHADER_FRAGMENT && count == 0)
+ return;
+
+ istate = &rctx->fragment_images;
+
+ assert (shader == PIPE_SHADER_FRAGMENT);
+ old_mask = istate->enabled_mask;
+ for (i = start_slot, idx = 0; i < start_slot + count; i++, idx++) {
+ unsigned res_type;
+ const struct pipe_image_view *iview;
+ rview = &istate->views[i];
+
+ if (!images || !images[idx].resource) {
+ pipe_resource_reference((struct pipe_resource **)&rview->base.resource, NULL);
+ istate->enabled_mask &= ~(1 << i);
+ continue;
+ }
+
+ iview = &images[idx];
+ image = iview->resource;
+ resource = (struct r600_resource *)image;
+
+ r600_context_add_resource_size(ctx, image);
+
+ rview->base = *iview;
+ rview->base.resource = NULL;
+ pipe_resource_reference((struct pipe_resource **)&rview->base.resource, image);
+
+ if (!resource->immed_buffer) {
+ int immed_size = (rscreen->b.info.max_se * 256 * 64) * util_format_get_blocksize(iview->format);
+
+ eg_resource_alloc_immed(&rscreen->b, resource, immed_size);
+ }
+
+ bool is_buffer = image->target == PIPE_BUFFER;
+ struct r600_texture *rtex = (struct r600_texture *)image;
+ if (!is_buffer & rtex->db_compatible)
+ istate->compressed_depthtex_mask |= 1 << i;
+ else
+ istate->compressed_depthtex_mask &= ~(1 << i);
+
+ if (!is_buffer && rtex->cmask.size)
+ istate->compressed_colortex_mask |= 1 << i;
+ else
+ istate->compressed_colortex_mask &= ~(1 << i);
+ if (!is_buffer) {
+
+ evergreen_set_color_surface_common(rctx, rtex,
+ iview->u.tex.level,
+ iview->u.tex.first_layer,
+ iview->u.tex.last_layer,
+ iview->format,
+ &color);
+ color.dim = S_028C78_WIDTH_MAX(u_minify(image->width0, iview->u.tex.level) - 1) |
+ S_028C78_HEIGHT_MAX(u_minify(image->height0, iview->u.tex.level) - 1);
+ } else {
+ color.offset = 0;
+ color.view = 0;
+ evergreen_set_color_surface_buffer(rctx, resource,
+ iview->format,
+ iview->u.buf.offset,
+ iview->u.buf.size,
+ &color);
+ }
+
+ switch (image->target) {
+ case PIPE_BUFFER:
+ res_type = V_028C70_BUFFER;
+ break;
+ case PIPE_TEXTURE_1D:
+ res_type = V_028C70_TEXTURE1D;
+ break;
+ case PIPE_TEXTURE_1D_ARRAY:
+ res_type = V_028C70_TEXTURE1DARRAY;
+ break;
+ case PIPE_TEXTURE_2D:
+ case PIPE_TEXTURE_RECT:
+ res_type = V_028C70_TEXTURE2D;
+ break;
+ case PIPE_TEXTURE_3D:
+ res_type = V_028C70_TEXTURE3D;
+ break;
+ case PIPE_TEXTURE_2D_ARRAY:
+ case PIPE_TEXTURE_CUBE:
+ case PIPE_TEXTURE_CUBE_ARRAY:
+ res_type = V_028C70_TEXTURE2DARRAY;
+ break;
+ default:
+ assert(0);
+ res_type = 0;
+ break;
+ }
+
+ rview->cb_color_base = color.offset;
+ rview->cb_color_dim = color.dim;
+ rview->cb_color_info = color.info |
+ S_028C70_RAT(1) |
+ S_028C70_RESOURCE_TYPE(res_type);
+ rview->cb_color_pitch = color.pitch;
+ rview->cb_color_slice = color.slice;
+ rview->cb_color_view = color.view;
+ rview->cb_color_attrib = color.attrib;
+ rview->cb_color_fmask = color.fmask;
+ rview->cb_color_fmask_slice = color.fmask_slice;
+
+ memset(&buf_params, 0, sizeof(buf_params));
+ buf_params.pipe_format = iview->format;
+ buf_params.size = resource->immed_buffer->b.b.width0;
+ buf_params.swizzle[0] = PIPE_SWIZZLE_X;
+ buf_params.swizzle[1] = PIPE_SWIZZLE_Y;
+ buf_params.swizzle[2] = PIPE_SWIZZLE_Z;
+ buf_params.swizzle[3] = PIPE_SWIZZLE_W;
+ buf_params.uncached = 1;
+ evergreen_fill_buffer_resource_words(rctx, &resource->immed_buffer->b.b,
+ &buf_params, &skip_reloc,
+ rview->immed_resource_words);
+
+
+ if (image->target != PIPE_BUFFER) {
+ memset(&tex_params, 0, sizeof(tex_params));
+ tex_params.pipe_format = iview->format;
+ tex_params.force_level = 0;
+ tex_params.width0 = image->width0;
+ tex_params.height0 = image->height0;
+ tex_params.first_level = iview->u.tex.level;
+ tex_params.last_level = iview->u.tex.level;
+ tex_params.first_layer = iview->u.tex.first_layer;
+ tex_params.last_layer = iview->u.tex.last_layer;
+ tex_params.target = image->target;
+ tex_params.swizzle[0] = PIPE_SWIZZLE_X;
+ tex_params.swizzle[1] = PIPE_SWIZZLE_Y;
+ tex_params.swizzle[2] = PIPE_SWIZZLE_Z;
+ tex_params.swizzle[3] = PIPE_SWIZZLE_W;
+ evergreen_fill_tex_resource_words(rctx, &resource->b.b, &tex_params,
+ &rview->skip_mip_address_reloc,
+ rview->resource_words);
+
+ } else {
+ memset(&buf_params, 0, sizeof(buf_params));
+ buf_params.pipe_format = iview->format;
+ buf_params.size = iview->u.buf.size;
+ buf_params.offset = iview->u.buf.offset;
+ buf_params.swizzle[0] = PIPE_SWIZZLE_X;
+ buf_params.swizzle[1] = PIPE_SWIZZLE_Y;
+ buf_params.swizzle[2] = PIPE_SWIZZLE_Z;
+ buf_params.swizzle[3] = PIPE_SWIZZLE_W;
+ evergreen_fill_buffer_resource_words(rctx, &resource->b.b,
+ &buf_params,
+ &rview->skip_mip_address_reloc,
+ rview->resource_words);
+ }
+ istate->enabled_mask |= (1 << i);
+ }
+
+ istate->atom.num_dw = util_bitcount(istate->enabled_mask) * 46;
+ istate->dirty_buffer_constants = TRUE;
+ rctx->b.flags |= R600_CONTEXT_WAIT_3D_IDLE | R600_CONTEXT_FLUSH_AND_INV;
+ rctx->b.flags |= R600_CONTEXT_FLUSH_AND_INV_CB |
+ R600_CONTEXT_FLUSH_AND_INV_CB_META;
+
+ if (old_mask != istate->enabled_mask)
+ r600_mark_atom_dirty(rctx, &rctx->framebuffer.atom);
+
+ if (rctx->cb_misc_state.nr_image_rats != util_bitcount(istate->enabled_mask)) {
+ rctx->cb_misc_state.nr_image_rats = util_bitcount(istate->enabled_mask);
+ r600_mark_atom_dirty(rctx, &rctx->cb_misc_state.atom);
+ }
+
+ r600_mark_atom_dirty(rctx, &istate->atom);
+}
+
void evergreen_init_state_functions(struct r600_context *rctx)
{
unsigned id = 1;
@@ -3779,6 +4067,7 @@ void evergreen_init_state_functions(struct r600_context *rctx)
rctx->config_state.dyn_gpr_enabled = true;
}
r600_init_atom(rctx, &rctx->framebuffer.atom, id++, evergreen_emit_framebuffer_state, 0);
+ r600_init_atom(rctx, &rctx->fragment_images.atom, id++, evergreen_emit_fragment_image_state, 0);
/* shader const */
r600_init_atom(rctx, &rctx->constbuf_state[PIPE_SHADER_VERTEX].atom, id++, evergreen_emit_vs_constant_buffers, 0);
r600_init_atom(rctx, &rctx->constbuf_state[PIPE_SHADER_GEOMETRY].atom, id++, evergreen_emit_gs_constant_buffers, 0);
@@ -3847,6 +4136,7 @@ void evergreen_init_state_functions(struct r600_context *rctx)
rctx->b.b.set_min_samples = evergreen_set_min_samples;
rctx->b.b.set_tess_state = evergreen_set_tess_state;
rctx->b.b.set_hw_atomic_buffers = evergreen_set_hw_atomic_buffers;
+ rctx->b.b.set_shader_images = evergreen_set_shader_images;
if (rctx->b.chip_class == EVERGREEN)
rctx->b.b.get_sample_position = evergreen_get_sample_position;
else
diff --git a/src/gallium/drivers/r600/r600_blit.c b/src/gallium/drivers/r600/r600_blit.c
index 64f06e8..05d04f5 100644
--- a/src/gallium/drivers/r600/r600_blit.c
+++ b/src/gallium/drivers/r600/r600_blit.c
@@ -291,6 +291,40 @@ void r600_decompress_depth_textures(struct r600_context *rctx,
}
}
+void r600_decompress_depth_images(struct r600_context *rctx,
+ struct r600_image_state *images)
+{
+ unsigned i;
+ unsigned depth_texture_mask = images->compressed_depthtex_mask;
+
+ while (depth_texture_mask) {
+ struct r600_image_view *view;
+ struct r600_texture *tex;
+
+ i = u_bit_scan(&depth_texture_mask);
+
+ view = &images->views[i];
+ assert(view);
+
+ tex = (struct r600_texture *)view->base.resource;
+ assert(tex->db_compatible);
+
+ if (r600_can_sample_zs(tex, false)) {
+ r600_blit_decompress_depth_in_place(rctx, tex,
+ false,
+ view->base.u.tex.level,
+ view->base.u.tex.level,
+ 0, util_max_layer(&tex->resource.b.b, view->base.u.tex.level));
+ } else {
+ r600_blit_decompress_depth(&rctx->b.b, tex, NULL,
+ view->base.u.tex.level,
+ view->base.u.tex.level,
+ 0, util_max_layer(&tex->resource.b.b, view->base.u.tex.level),
+ 0, u_max_sample(&tex->resource.b.b));
+ }
+ }
+}
+
static void r600_blit_decompress_color(struct pipe_context *ctx,
struct r600_texture *rtex,
unsigned first_level, unsigned last_level,
@@ -360,6 +394,31 @@ void r600_decompress_color_textures(struct r600_context *rctx,
}
}
+void r600_decompress_color_images(struct r600_context *rctx,
+ struct r600_image_state *images)
+{
+ unsigned i;
+ unsigned mask = images->compressed_colortex_mask;
+
+ while (mask) {
+ struct r600_image_view *view;
+ struct r600_texture *tex;
+
+ i = u_bit_scan(&mask);
+
+ view = &images->views[i];
+ assert(view);
+
+ tex = (struct r600_texture *)view->base.resource;
+ assert(tex->cmask.size);
+
+ r600_blit_decompress_color(&rctx->b.b, tex,
+ view->base.u.tex.level, view->base.u.tex.level,
+ view->base.u.tex.first_layer,
+ view->base.u.tex.last_layer);
+ }
+}
+
/* Helper for decompressing a portion of a color or depth resource before
* blitting if any decompression is needed.
* The driver doesn't decompress resources automatically while u_blitter is
diff --git a/src/gallium/drivers/r600/r600_hw_context.c b/src/gallium/drivers/r600/r600_hw_context.c
index a821c35..727ad9c 100644
--- a/src/gallium/drivers/r600/r600_hw_context.c
+++ b/src/gallium/drivers/r600/r600_hw_context.c
@@ -348,6 +348,8 @@ void r600_begin_new_cs(struct r600_context *ctx)
r600_mark_atom_dirty(ctx, &ctx->db_misc_state.atom);
r600_mark_atom_dirty(ctx, &ctx->db_state.atom);
r600_mark_atom_dirty(ctx, &ctx->framebuffer.atom);
+ if (ctx->b.chip_class >= EVERGREEN)
+ r600_mark_atom_dirty(ctx, &ctx->fragment_images.atom);
r600_mark_atom_dirty(ctx, &ctx->hw_shader_stages[R600_HW_STAGE_PS].atom);
r600_mark_atom_dirty(ctx, &ctx->poly_offset_state.atom);
r600_mark_atom_dirty(ctx, &ctx->vgt_state.atom);
diff --git a/src/gallium/drivers/r600/r600_pipe.h b/src/gallium/drivers/r600/r600_pipe.h
index 4fbc82b..ad7e499 100644
--- a/src/gallium/drivers/r600/r600_pipe.h
+++ b/src/gallium/drivers/r600/r600_pipe.h
@@ -38,7 +38,16 @@
#include "tgsi/tgsi_scan.h"
-#define R600_NUM_ATOMS 52
+#define R600_NUM_ATOMS 53
+
+#define R600_MAX_IMAGES 8
+/*
+ * ranges reserved for images on evergreen
+ * first set for the immediate buffers,
+ * second for the actual resources for RESQ.
+ */
+#define R600_IMAGE_IMMED_RESOURCE_OFFSET 160
+#define R600_IMAGE_REAL_RESOURCE_OFFSET 168
/* read caches */
#define R600_CONTEXT_INV_VERTEX_CACHE (R600_CONTEXT_PRIVATE_FLAG << 0)
@@ -135,6 +144,7 @@ struct r600_cb_misc_state {
unsigned blend_colormask; /* 8*4 bits for 8 RGBA colorbuffers */
unsigned nr_cbufs;
unsigned nr_ps_color_outputs;
+ unsigned nr_image_rats;
bool multiwrite;
bool dual_src_blend;
};
@@ -425,6 +435,33 @@ struct r600_atomic_buffer_state {
struct pipe_shader_buffer buffer[EG_MAX_ATOMIC_BUFFERS];
};
+struct r600_image_view {
+ struct pipe_image_view base;
+ uint32_t cb_color_base;
+ uint32_t cb_color_pitch;
+ uint32_t cb_color_slice;
+ uint32_t cb_color_view;
+ uint32_t cb_color_info;
+ uint32_t cb_color_attrib;
+ uint32_t cb_color_dim;
+ uint32_t cb_color_fmask;
+ uint32_t cb_color_fmask_slice;
+ uint32_t immed_resource_words[8];
+ uint32_t resource_words[8];
+ bool skip_mip_address_reloc;
+ uint32_t buf_size;
+};
+
+struct r600_image_state {
+ struct r600_atom atom;
+ uint32_t enabled_mask;
+ uint32_t dirty_mask;
+ uint32_t compressed_depthtex_mask;
+ uint32_t compressed_colortex_mask;
+ boolean dirty_buffer_constants;
+ struct r600_image_view views[R600_MAX_IMAGES];
+};
+
struct r600_context {
struct r600_common_context b;
struct r600_screen *screen;
@@ -480,6 +517,8 @@ struct r600_context {
struct r600_stencil_ref_state stencil_ref;
struct r600_vgt_state vgt_state;
struct r600_atomic_buffer_state atomic_buffer_state;
+ /* only have images on fragment shader */
+ struct r600_image_state fragment_images;
/* Shaders and shader resources. */
struct r600_cso_state vertex_fetch_shader;
struct r600_shader_state hw_shader_stages[EG_NUM_HW_STAGES];
@@ -648,8 +687,12 @@ bool evergreen_adjust_gprs(struct r600_context *rctx);
void r600_init_blit_functions(struct r600_context *rctx);
void r600_decompress_depth_textures(struct r600_context *rctx,
struct r600_samplerview_state *textures);
+void r600_decompress_depth_images(struct r600_context *rctx,
+ struct r600_image_state *images);
void r600_decompress_color_textures(struct r600_context *rctx,
struct r600_samplerview_state *textures);
+void r600_decompress_color_images(struct r600_context *rctx,
+ struct r600_image_state *images);
void r600_resource_copy_region(struct pipe_context *ctx,
struct pipe_resource *dst,
unsigned dst_level,
diff --git a/src/gallium/drivers/r600/r600_shader.h b/src/gallium/drivers/r600/r600_shader.h
index c6efbb9..ccbcd67 100644
--- a/src/gallium/drivers/r600/r600_shader.h
+++ b/src/gallium/drivers/r600/r600_shader.h
@@ -118,6 +118,7 @@ struct r600_shader {
boolean uses_atomics;
boolean uses_images;
uint8_t atomic_base;
+ uint8_t rat_base;
};
union r600_shader_key {
diff --git a/src/gallium/drivers/r600/r600_state_common.c b/src/gallium/drivers/r600/r600_state_common.c
index 4ae18fc..1bd0a76 100644
--- a/src/gallium/drivers/r600/r600_state_common.c
+++ b/src/gallium/drivers/r600/r600_state_common.c
@@ -753,6 +753,26 @@ static int r600_get_hw_atomic_count(const struct pipe_context *ctx,
return value;
}
+static void r600_update_compressed_colortex_mask_images(struct r600_image_state *images)
+{
+ uint32_t mask = images->enabled_mask;
+
+ while (mask) {
+ unsigned i = u_bit_scan(&mask);
+ struct pipe_resource *res = images->views[i].base.resource;
+
+ if (res && res->target != PIPE_BUFFER) {
+ struct r600_texture *rtex = (struct r600_texture *)res;
+
+ if (rtex->cmask.size) {
+ images->compressed_colortex_mask |= 1 << i;
+ } else {
+ images->compressed_colortex_mask &= ~(1 << i);
+ }
+ }
+ }
+}
+
/* Compute the key for the hw shader variant */
static inline void r600_shader_selector_key(const struct pipe_context *ctx,
const struct r600_pipe_shader_selector *sel,
@@ -1473,6 +1493,7 @@ static void r600_update_compressed_resource_state(struct r600_context *rctx)
for (i = 0; i < PIPE_SHADER_TYPES; ++i) {
r600_update_compressed_colortex_mask(&rctx->samplers[i].views);
}
+ r600_update_compressed_colortex_mask_images(&rctx->fragment_images);
}
/* Decompress textures if needed. */
@@ -1485,6 +1506,15 @@ static void r600_update_compressed_resource_state(struct r600_context *rctx)
r600_decompress_color_textures(rctx, views);
}
}
+
+ {
+ struct r600_image_state *istate;
+ istate = &rctx->fragment_images;
+ if (istate->compressed_depthtex_mask)
+ r600_decompress_depth_images(rctx, istate);
+ if (istate->compressed_colortex_mask)
+ r600_decompress_color_images(rctx, istate);
+ }
}
#define SELECT_SHADER_OR_FAIL(x) do { \
--
2.9.5
More information about the mesa-dev
mailing list