radeonsi: remove r600_pipe_common.h
authorMarek Olšák <marek.olsak@amd.com>
Mon, 9 Apr 2018 00:20:39 +0000 (20:20 -0400)
committerMarek Olšák <marek.olsak@amd.com>
Fri, 27 Apr 2018 21:56:04 +0000 (17:56 -0400)
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
14 files changed:
src/gallium/drivers/radeon/r600_pipe_common.h [deleted file]
src/gallium/drivers/radeon/radeon_vce_40_2_2.c
src/gallium/drivers/radeon/radeon_vce_50.c
src/gallium/drivers/radeon/radeon_vcn_enc_1_2.c
src/gallium/drivers/radeonsi/Makefile.sources
src/gallium/drivers/radeonsi/meson.build
src/gallium/drivers/radeonsi/si_buffer.c
src/gallium/drivers/radeonsi/si_pipe.c
src/gallium/drivers/radeonsi/si_pipe.h
src/gallium/drivers/radeonsi/si_query.c
src/gallium/drivers/radeonsi/si_shader.h
src/gallium/drivers/radeonsi/si_state.h
src/gallium/drivers/radeonsi/si_state_streamout.c
src/gallium/drivers/radeonsi/si_texture.c

diff --git a/src/gallium/drivers/radeon/r600_pipe_common.h b/src/gallium/drivers/radeon/r600_pipe_common.h
deleted file mode 100644 (file)
index 66e9a0b..0000000
+++ /dev/null
@@ -1,330 +0,0 @@
-/*
- * Copyright 2013 Advanced Micro Devices, Inc.
- * All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice (including the next
- * paragraph) shall be included in all copies or substantial portions of the
- * Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
- * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
- * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
- * SOFTWARE.
- */
-
-/**
- * This file is going to be removed.
- */
-
-#ifndef R600_PIPE_COMMON_H
-#define R600_PIPE_COMMON_H
-
-#include <stdio.h>
-
-#include "amd/common/ac_binary.h"
-
-#include "radeon/radeon_winsys.h"
-
-#include "util/disk_cache.h"
-#include "util/u_blitter.h"
-#include "util/list.h"
-#include "util/u_range.h"
-#include "util/slab.h"
-#include "util/u_suballoc.h"
-#include "util/u_transfer.h"
-#include "util/u_threaded_context.h"
-
-struct u_log_context;
-struct si_screen;
-struct si_context;
-struct si_perfcounters;
-struct tgsi_shader_info;
-struct si_qbo_state;
-
-/* Only 32-bit buffer allocations are supported, gallium doesn't support more
- * at the moment.
- */
-struct r600_resource {
-       struct threaded_resource        b;
-
-       /* Winsys objects. */
-       struct pb_buffer                *buf;
-       uint64_t                        gpu_address;
-       /* Memory usage if the buffer placement is optimal. */
-       uint64_t                        vram_usage;
-       uint64_t                        gart_usage;
-
-       /* Resource properties. */
-       uint64_t                        bo_size;
-       unsigned                        bo_alignment;
-       enum radeon_bo_domain           domains;
-       enum radeon_bo_flag             flags;
-       unsigned                        bind_history;
-       int                             max_forced_staging_uploads;
-
-       /* The buffer range which is initialized (with a write transfer,
-        * streamout, DMA, or as a random access target). The rest of
-        * the buffer is considered invalid and can be mapped unsynchronized.
-        *
-        * This allows unsychronized mapping of a buffer range which hasn't
-        * been used yet. It's for applications which forget to use
-        * the unsynchronized map flag and expect the driver to figure it out.
-         */
-       struct util_range               valid_buffer_range;
-
-       /* For buffers only. This indicates that a write operation has been
-        * performed by TC L2, but the cache hasn't been flushed.
-        * Any hw block which doesn't use or bypasses TC L2 should check this
-        * flag and flush the cache before using the buffer.
-        *
-        * For example, TC L2 must be flushed if a buffer which has been
-        * modified by a shader store instruction is about to be used as
-        * an index buffer. The reason is that VGT DMA index fetching doesn't
-        * use TC L2.
-        */
-       bool                            TC_L2_dirty;
-
-       /* Whether the resource has been exported via resource_get_handle. */
-       unsigned                        external_usage; /* PIPE_HANDLE_USAGE_* */
-
-       /* Whether this resource is referenced by bindless handles. */
-       bool                            texture_handle_allocated;
-       bool                            image_handle_allocated;
-};
-
-struct r600_transfer {
-       struct threaded_transfer        b;
-       struct r600_resource            *staging;
-       unsigned                        offset;
-};
-
-struct r600_fmask_info {
-       uint64_t offset;
-       uint64_t size;
-       unsigned alignment;
-       unsigned pitch_in_pixels;
-       unsigned bank_height;
-       unsigned slice_tile_max;
-       unsigned tile_mode_index;
-       unsigned tile_swizzle;
-};
-
-struct r600_cmask_info {
-       uint64_t offset;
-       uint64_t size;
-       unsigned alignment;
-       unsigned slice_tile_max;
-       uint64_t base_address_reg;
-};
-
-struct r600_texture {
-       struct r600_resource            resource;
-
-       struct radeon_surf              surface;
-       uint64_t                        size;
-       struct r600_texture             *flushed_depth_texture;
-
-       /* Colorbuffer compression and fast clear. */
-       struct r600_fmask_info          fmask;
-       struct r600_cmask_info          cmask;
-       struct r600_resource            *cmask_buffer;
-       uint64_t                        dcc_offset; /* 0 = disabled */
-       unsigned                        cb_color_info; /* fast clear enable bit */
-       unsigned                        color_clear_value[2];
-       unsigned                        last_msaa_resolve_target_micro_mode;
-       unsigned                        num_level0_transfers;
-
-       /* Depth buffer compression and fast clear. */
-       uint64_t                        htile_offset;
-       float                           depth_clear_value;
-       uint16_t                        dirty_level_mask; /* each bit says if that mipmap is compressed */
-       uint16_t                        stencil_dirty_level_mask; /* each bit says if that mipmap is compressed */
-       enum pipe_format                db_render_format:16;
-       uint8_t                         stencil_clear_value;
-       bool                            tc_compatible_htile:1;
-       bool                            depth_cleared:1; /* if it was cleared at least once */
-       bool                            stencil_cleared:1; /* if it was cleared at least once */
-       bool                            upgraded_depth:1; /* upgraded from unorm to Z32_FLOAT */
-       bool                            is_depth:1;
-       bool                            db_compatible:1;
-       bool                            can_sample_z:1;
-       bool                            can_sample_s:1;
-
-       /* We need to track DCC dirtiness, because st/dri usually calls
-        * flush_resource twice per frame (not a bug) and we don't wanna
-        * decompress DCC twice. Also, the dirty tracking must be done even
-        * if DCC isn't used, because it's required by the DCC usage analysis
-        * for a possible future enablement.
-        */
-       bool                            separate_dcc_dirty:1;
-       /* Statistics gathering for the DCC enablement heuristic. */
-       bool                            dcc_gather_statistics:1;
-       /* Counter that should be non-zero if the texture is bound to a
-        * framebuffer.
-        */
-       unsigned                        framebuffers_bound;
-       /* Whether the texture is a displayable back buffer and needs DCC
-        * decompression, which is expensive. Therefore, it's enabled only
-        * if statistics suggest that it will pay off and it's allocated
-        * separately. It can't be bound as a sampler by apps. Limited to
-        * target == 2D and last_level == 0. If enabled, dcc_offset contains
-        * the absolute GPUVM address, not the relative one.
-        */
-       struct r600_resource            *dcc_separate_buffer;
-       /* When DCC is temporarily disabled, the separate buffer is here. */
-       struct r600_resource            *last_dcc_separate_buffer;
-       /* Estimate of how much this color buffer is written to in units of
-        * full-screen draws: ps_invocations / (width * height)
-        * Shader kills, late Z, and blending with trivial discards make it
-        * inaccurate (we need to count CB updates, not PS invocations).
-        */
-       unsigned                        ps_draw_ratio;
-       /* The number of clears since the last DCC usage analysis. */
-       unsigned                        num_slow_clears;
-};
-
-struct r600_surface {
-       struct pipe_surface             base;
-
-       /* These can vary with block-compressed textures. */
-       uint16_t width0;
-       uint16_t height0;
-
-       bool color_initialized:1;
-       bool depth_initialized:1;
-
-       /* Misc. color flags. */
-       bool color_is_int8:1;
-       bool color_is_int10:1;
-       bool dcc_incompatible:1;
-
-       /* Color registers. */
-       unsigned cb_color_info;
-       unsigned cb_color_view;
-       unsigned cb_color_attrib;
-       unsigned cb_color_attrib2;      /* GFX9 and later */
-       unsigned cb_dcc_control;        /* VI and later */
-       unsigned spi_shader_col_format:8;       /* no blending, no alpha-to-coverage. */
-       unsigned spi_shader_col_format_alpha:8; /* alpha-to-coverage */
-       unsigned spi_shader_col_format_blend:8; /* blending without alpha. */
-       unsigned spi_shader_col_format_blend_alpha:8; /* blending with alpha. */
-
-       /* DB registers. */
-       uint64_t db_depth_base;         /* DB_Z_READ/WRITE_BASE */
-       uint64_t db_stencil_base;
-       uint64_t db_htile_data_base;
-       unsigned db_depth_info;
-       unsigned db_z_info;
-       unsigned db_z_info2;            /* GFX9+ */
-       unsigned db_depth_view;
-       unsigned db_depth_size;
-       unsigned db_depth_slice;
-       unsigned db_stencil_info;
-       unsigned db_stencil_info2;      /* GFX9+ */
-       unsigned db_htile_surface;
-};
-
-struct si_mmio_counter {
-       unsigned busy;
-       unsigned idle;
-};
-
-union si_mmio_counters {
-       struct {
-               /* For global GPU load including SDMA. */
-               struct si_mmio_counter gpu;
-
-               /* GRBM_STATUS */
-               struct si_mmio_counter spi;
-               struct si_mmio_counter gui;
-               struct si_mmio_counter ta;
-               struct si_mmio_counter gds;
-               struct si_mmio_counter vgt;
-               struct si_mmio_counter ia;
-               struct si_mmio_counter sx;
-               struct si_mmio_counter wd;
-               struct si_mmio_counter bci;
-               struct si_mmio_counter sc;
-               struct si_mmio_counter pa;
-               struct si_mmio_counter db;
-               struct si_mmio_counter cp;
-               struct si_mmio_counter cb;
-
-               /* SRBM_STATUS2 */
-               struct si_mmio_counter sdma;
-
-               /* CP_STAT */
-               struct si_mmio_counter pfp;
-               struct si_mmio_counter meq;
-               struct si_mmio_counter me;
-               struct si_mmio_counter surf_sync;
-               struct si_mmio_counter cp_dma;
-               struct si_mmio_counter scratch_ram;
-       } named;
-       unsigned array[0];
-};
-
-struct r600_memory_object {
-       struct pipe_memory_object       b;
-       struct pb_buffer                *buf;
-       uint32_t                        stride;
-       uint32_t                        offset;
-};
-
-/* This encapsulates a state or an operation which can emitted into the GPU
- * command stream. */
-struct r600_atom {
-       void (*emit)(struct si_context *ctx, struct r600_atom *state);
-       unsigned short          id;
-};
-
-/* Saved CS data for debugging features. */
-struct radeon_saved_cs {
-       uint32_t                        *ib;
-       unsigned                        num_dw;
-
-       struct radeon_bo_list_item      *bo_list;
-       unsigned                        bo_count;
-};
-
-/* r600_perfcounters.c */
-void si_perfcounters_destroy(struct si_screen *sscreen);
-
-
-/* Inline helpers. */
-
-static inline struct r600_resource *r600_resource(struct pipe_resource *r)
-{
-       return (struct r600_resource*)r;
-}
-
-static inline void
-r600_resource_reference(struct r600_resource **ptr, struct r600_resource *res)
-{
-       pipe_resource_reference((struct pipe_resource **)ptr,
-                               (struct pipe_resource *)res);
-}
-
-static inline void
-r600_texture_reference(struct r600_texture **ptr, struct r600_texture *res)
-{
-       pipe_resource_reference((struct pipe_resource **)ptr, &res->resource.b.b);
-}
-
-static inline bool
-vi_dcc_enabled(struct r600_texture *tex, unsigned level)
-{
-       return tex->dcc_offset && level < tex->surface.num_dcc_levels;
-}
-
-#endif
index 04e9d7f5e19863ebf58e5dc5bfb005fa91f2301e..66b54dab25a29f0061b660898eed4c4c723417be 100644 (file)
@@ -34,7 +34,7 @@
 
 #include "vl/vl_video_buffer.h"
 
-#include "r600_pipe_common.h"
+#include "si_pipe.h"
 #include "radeon_video.h"
 #include "radeon_vce.h"
 
index b4f9771ecc6a74109e6c45e69c5750f6761fcf42..f4cbc9bb82b53d23623b53a1ad33143c57e3433f 100644 (file)
@@ -34,7 +34,7 @@
 
 #include "vl/vl_video_buffer.h"
 
-#include "r600_pipe_common.h"
+#include "si_pipe.h"
 #include "radeon_video.h"
 #include "radeon_vce.h"
 
index 07493d806eed8606c8ba4cae5d4b72b3a3ebf710..143721211db8417559f3af6c893e2a76f3894c6a 100644 (file)
@@ -34,7 +34,7 @@
 
 #include "vl/vl_video_buffer.h"
 
-#include "r600_pipe_common.h"
+#include "si_pipe.h"
 #include "radeon_video.h"
 #include "radeon_vcn_enc.h"
 
index b20a5497f5ee5eb84d792fcaab5c8eaa12af011e..f760b5b7a699be38a1610c5b2d48e030105e8646 100644 (file)
@@ -47,7 +47,6 @@ C_SOURCES := \
        si_texture.c \
        si_uvd.c \
        ../radeon/r600_perfcounter.c \
-       ../radeon/r600_pipe_common.h \
        ../radeon/radeon_uvd.c \
        ../radeon/radeon_uvd.h \
        ../radeon/radeon_vcn_dec.c \
index f817d591fbed89b4bfb6a85a3f1b8257e7ed1d8d..90498398f3885b1b7cd37330f326d81983fd4c0e 100644 (file)
@@ -63,7 +63,6 @@ files_libradeonsi = files(
   'si_texture.c',
   'si_uvd.c',
   '../radeon/r600_perfcounter.c',
-  '../radeon/r600_pipe_common.h',
   '../radeon/radeon_uvd.c',
   '../radeon/radeon_uvd.h',
   '../radeon/radeon_vcn_enc_1_2.c',
index d17b2c6a831af0bea93c5efda6e78ab7ddf495a9..a0855db571f9970e0cfc69b9d65ca75c02997445 100644 (file)
@@ -25,6 +25,7 @@
 #include "radeonsi/si_pipe.h"
 #include "util/u_memory.h"
 #include "util/u_upload_mgr.h"
+#include "util/u_transfer.h"
 #include <inttypes.h>
 #include <stdio.h>
 
index 3eb89ff0287a81d686dbbd1eb3db454619c37145..971ff19d799646103b33ce5b67e9039975d77f23 100644 (file)
@@ -29,6 +29,7 @@
 #include "sid.h"
 
 #include "radeon/radeon_uvd.h"
+#include "util/disk_cache.h"
 #include "util/hash_table.h"
 #include "util/u_log.h"
 #include "util/u_memory.h"
index 24221d0a55a96a03d437c267e835f3ff45d08794..2ed764bd097ebd447b4e1bcc84e17499a93697ec 100644 (file)
 #define SI_PIPE_H
 
 #include "si_shader.h"
+#include "si_state.h"
 
 #include "util/u_dynarray.h"
 #include "util/u_idalloc.h"
+#include "util/u_range.h"
+#include "util/u_threaded_context.h"
 
 #ifdef PIPE_ARCH_BIG_ENDIAN
 #define SI_BIG_ENDIAN 1
@@ -170,6 +173,245 @@ struct si_compute;
 struct hash_table;
 struct u_suballocator;
 
+/* Only 32-bit buffer allocations are supported, gallium doesn't support more
+ * at the moment.
+ */
+struct r600_resource {
+       struct threaded_resource        b;
+
+       /* Winsys objects. */
+       struct pb_buffer                *buf;
+       uint64_t                        gpu_address;
+       /* Memory usage if the buffer placement is optimal. */
+       uint64_t                        vram_usage;
+       uint64_t                        gart_usage;
+
+       /* Resource properties. */
+       uint64_t                        bo_size;
+       unsigned                        bo_alignment;
+       enum radeon_bo_domain           domains;
+       enum radeon_bo_flag             flags;
+       unsigned                        bind_history;
+       int                             max_forced_staging_uploads;
+
+       /* The buffer range which is initialized (with a write transfer,
+        * streamout, DMA, or as a random access target). The rest of
+        * the buffer is considered invalid and can be mapped unsynchronized.
+        *
+        * This allows unsychronized mapping of a buffer range which hasn't
+        * been used yet. It's for applications which forget to use
+        * the unsynchronized map flag and expect the driver to figure it out.
+         */
+       struct util_range               valid_buffer_range;
+
+       /* For buffers only. This indicates that a write operation has been
+        * performed by TC L2, but the cache hasn't been flushed.
+        * Any hw block which doesn't use or bypasses TC L2 should check this
+        * flag and flush the cache before using the buffer.
+        *
+        * For example, TC L2 must be flushed if a buffer which has been
+        * modified by a shader store instruction is about to be used as
+        * an index buffer. The reason is that VGT DMA index fetching doesn't
+        * use TC L2.
+        */
+       bool                            TC_L2_dirty;
+
+       /* Whether the resource has been exported via resource_get_handle. */
+       unsigned                        external_usage; /* PIPE_HANDLE_USAGE_* */
+
+       /* Whether this resource is referenced by bindless handles. */
+       bool                            texture_handle_allocated;
+       bool                            image_handle_allocated;
+};
+
+struct r600_transfer {
+       struct threaded_transfer        b;
+       struct r600_resource            *staging;
+       unsigned                        offset;
+};
+
+struct r600_fmask_info {
+       uint64_t offset;
+       uint64_t size;
+       unsigned alignment;
+       unsigned pitch_in_pixels;
+       unsigned bank_height;
+       unsigned slice_tile_max;
+       unsigned tile_mode_index;
+       unsigned tile_swizzle;
+};
+
+struct r600_cmask_info {
+       uint64_t offset;
+       uint64_t size;
+       unsigned alignment;
+       unsigned slice_tile_max;
+       uint64_t base_address_reg;
+};
+
+struct r600_texture {
+       struct r600_resource            resource;
+
+       struct radeon_surf              surface;
+       uint64_t                        size;
+       struct r600_texture             *flushed_depth_texture;
+
+       /* Colorbuffer compression and fast clear. */
+       struct r600_fmask_info          fmask;
+       struct r600_cmask_info          cmask;
+       struct r600_resource            *cmask_buffer;
+       uint64_t                        dcc_offset; /* 0 = disabled */
+       unsigned                        cb_color_info; /* fast clear enable bit */
+       unsigned                        color_clear_value[2];
+       unsigned                        last_msaa_resolve_target_micro_mode;
+       unsigned                        num_level0_transfers;
+
+       /* Depth buffer compression and fast clear. */
+       uint64_t                        htile_offset;
+       float                           depth_clear_value;
+       uint16_t                        dirty_level_mask; /* each bit says if that mipmap is compressed */
+       uint16_t                        stencil_dirty_level_mask; /* each bit says if that mipmap is compressed */
+       enum pipe_format                db_render_format:16;
+       uint8_t                         stencil_clear_value;
+       bool                            tc_compatible_htile:1;
+       bool                            depth_cleared:1; /* if it was cleared at least once */
+       bool                            stencil_cleared:1; /* if it was cleared at least once */
+       bool                            upgraded_depth:1; /* upgraded from unorm to Z32_FLOAT */
+       bool                            is_depth:1;
+       bool                            db_compatible:1;
+       bool                            can_sample_z:1;
+       bool                            can_sample_s:1;
+
+       /* We need to track DCC dirtiness, because st/dri usually calls
+        * flush_resource twice per frame (not a bug) and we don't wanna
+        * decompress DCC twice. Also, the dirty tracking must be done even
+        * if DCC isn't used, because it's required by the DCC usage analysis
+        * for a possible future enablement.
+        */
+       bool                            separate_dcc_dirty:1;
+       /* Statistics gathering for the DCC enablement heuristic. */
+       bool                            dcc_gather_statistics:1;
+       /* Counter that should be non-zero if the texture is bound to a
+        * framebuffer.
+        */
+       unsigned                        framebuffers_bound;
+       /* Whether the texture is a displayable back buffer and needs DCC
+        * decompression, which is expensive. Therefore, it's enabled only
+        * if statistics suggest that it will pay off and it's allocated
+        * separately. It can't be bound as a sampler by apps. Limited to
+        * target == 2D and last_level == 0. If enabled, dcc_offset contains
+        * the absolute GPUVM address, not the relative one.
+        */
+       struct r600_resource            *dcc_separate_buffer;
+       /* When DCC is temporarily disabled, the separate buffer is here. */
+       struct r600_resource            *last_dcc_separate_buffer;
+       /* Estimate of how much this color buffer is written to in units of
+        * full-screen draws: ps_invocations / (width * height)
+        * Shader kills, late Z, and blending with trivial discards make it
+        * inaccurate (we need to count CB updates, not PS invocations).
+        */
+       unsigned                        ps_draw_ratio;
+       /* The number of clears since the last DCC usage analysis. */
+       unsigned                        num_slow_clears;
+};
+
+struct r600_surface {
+       struct pipe_surface             base;
+
+       /* These can vary with block-compressed textures. */
+       uint16_t width0;
+       uint16_t height0;
+
+       bool color_initialized:1;
+       bool depth_initialized:1;
+
+       /* Misc. color flags. */
+       bool color_is_int8:1;
+       bool color_is_int10:1;
+       bool dcc_incompatible:1;
+
+       /* Color registers. */
+       unsigned cb_color_info;
+       unsigned cb_color_view;
+       unsigned cb_color_attrib;
+       unsigned cb_color_attrib2;      /* GFX9 and later */
+       unsigned cb_dcc_control;        /* VI and later */
+       unsigned spi_shader_col_format:8;       /* no blending, no alpha-to-coverage. */
+       unsigned spi_shader_col_format_alpha:8; /* alpha-to-coverage */
+       unsigned spi_shader_col_format_blend:8; /* blending without alpha. */
+       unsigned spi_shader_col_format_blend_alpha:8; /* blending with alpha. */
+
+       /* DB registers. */
+       uint64_t db_depth_base;         /* DB_Z_READ/WRITE_BASE */
+       uint64_t db_stencil_base;
+       uint64_t db_htile_data_base;
+       unsigned db_depth_info;
+       unsigned db_z_info;
+       unsigned db_z_info2;            /* GFX9+ */
+       unsigned db_depth_view;
+       unsigned db_depth_size;
+       unsigned db_depth_slice;
+       unsigned db_stencil_info;
+       unsigned db_stencil_info2;      /* GFX9+ */
+       unsigned db_htile_surface;
+};
+
+struct si_mmio_counter {
+       unsigned busy;
+       unsigned idle;
+};
+
+union si_mmio_counters {
+       struct {
+               /* For global GPU load including SDMA. */
+               struct si_mmio_counter gpu;
+
+               /* GRBM_STATUS */
+               struct si_mmio_counter spi;
+               struct si_mmio_counter gui;
+               struct si_mmio_counter ta;
+               struct si_mmio_counter gds;
+               struct si_mmio_counter vgt;
+               struct si_mmio_counter ia;
+               struct si_mmio_counter sx;
+               struct si_mmio_counter wd;
+               struct si_mmio_counter bci;
+               struct si_mmio_counter sc;
+               struct si_mmio_counter pa;
+               struct si_mmio_counter db;
+               struct si_mmio_counter cp;
+               struct si_mmio_counter cb;
+
+               /* SRBM_STATUS2 */
+               struct si_mmio_counter sdma;
+
+               /* CP_STAT */
+               struct si_mmio_counter pfp;
+               struct si_mmio_counter meq;
+               struct si_mmio_counter me;
+               struct si_mmio_counter surf_sync;
+               struct si_mmio_counter cp_dma;
+               struct si_mmio_counter scratch_ram;
+       } named;
+       unsigned array[0];
+};
+
+struct r600_memory_object {
+       struct pipe_memory_object       b;
+       struct pb_buffer                *buf;
+       uint32_t                        stride;
+       uint32_t                        offset;
+};
+
+/* Saved CS data for debugging features. */
+struct radeon_saved_cs {
+       uint32_t                        *ib;
+       unsigned                        num_dw;
+
+       struct radeon_bo_list_item      *bo_list;
+       unsigned                        bo_count;
+};
+
 struct si_screen {
        struct pipe_screen              b;
        struct radeon_winsys            *ws;
@@ -975,6 +1217,9 @@ unsigned si_end_counter(struct si_screen *sscreen, unsigned type,
 /* si_compute.c */
 void si_init_compute_functions(struct si_context *sctx);
 
+/* r600_perfcounters.c */
+void si_perfcounters_destroy(struct si_screen *sscreen);
+
 /* si_perfcounters.c */
 void si_init_perfcounters(struct si_screen *screen);
 
@@ -1060,6 +1305,39 @@ void si_init_context_texture_functions(struct si_context *sctx);
  * common helpers
  */
 
+static inline struct r600_resource *r600_resource(struct pipe_resource *r)
+{
+       return (struct r600_resource*)r;
+}
+
+static inline void
+r600_resource_reference(struct r600_resource **ptr, struct r600_resource *res)
+{
+       pipe_resource_reference((struct pipe_resource **)ptr,
+                               (struct pipe_resource *)res);
+}
+
+static inline void
+r600_texture_reference(struct r600_texture **ptr, struct r600_texture *res)
+{
+       pipe_resource_reference((struct pipe_resource **)ptr, &res->resource.b.b);
+}
+
+static inline bool
+vi_dcc_enabled(struct r600_texture *tex, unsigned level)
+{
+       return tex->dcc_offset && level < tex->surface.num_dcc_levels;
+}
+
+static inline unsigned
+si_tile_mode_index(struct r600_texture *rtex, unsigned level, bool stencil)
+{
+       if (stencil)
+               return rtex->surface.u.legacy.stencil_tiling_index[level];
+       else
+               return rtex->surface.u.legacy.tiling_index[level];
+}
+
 static inline void
 si_context_add_resource_size(struct si_context *sctx, struct pipe_resource *r)
 {
index 3de47c056b2e1e537e602494d7a0a69e146c3e05..c859cd081ca8e1764bd35e96673cc7d32767dcb8 100644 (file)
@@ -29,6 +29,7 @@
 #include "util/u_memory.h"
 #include "util/u_upload_mgr.h"
 #include "util/os_time.h"
+#include "util/u_suballoc.h"
 #include "tgsi/tgsi_text.h"
 #include "amd/common/sid.h"
 
index 7734bfd2f5d3d8ea2d598e23ec700b82ef521424..c26ccafdd69636511b7751ec073a5684ac21b76e 100644 (file)
 #include <llvm-c/Core.h> /* LLVMModuleRef */
 #include <llvm-c/TargetMachine.h>
 #include "tgsi/tgsi_scan.h"
+#include "util/u_inlines.h"
 #include "util/u_queue.h"
 
 #include "ac_binary.h"
 #include "ac_llvm_build.h"
-#include "si_state.h"
+
+#include <stdio.h>
 
 struct nir_shader;
+struct si_shader;
+struct si_context;
 
+#define SI_MAX_ATTRIBS         16
 #define SI_MAX_VS_OUTPUTS      40
 
 /* Shader IO unique indices are supported for TGSI_SEMANTIC_GENERIC with an
index e9849a992f0cd688b1da3c222b5c158e25c82a7e..628cce87ef9b3554289dc65c76ca47afea3e44aa 100644 (file)
 #define SI_STATE_H
 
 #include "si_pm4.h"
-#include "radeon/r600_pipe_common.h"
 
 #include "pipebuffer/pb_slab.h"
+#include "util/u_blitter.h"
 
 #define SI_NUM_GRAPHICS_SHADERS (PIPE_SHADER_TESS_EVAL+1)
 #define SI_NUM_SHADERS (PIPE_SHADER_COMPUTE+1)
 
-#define SI_MAX_ATTRIBS                 16
 #define SI_NUM_VERTEX_BUFFERS          SI_MAX_ATTRIBS
 #define SI_NUM_SAMPLERS                        32 /* OpenGL textures units per shader */
 #define SI_NUM_CONST_BUFFERS           16
 struct si_screen;
 struct si_shader;
 struct si_shader_selector;
+struct r600_texture;
+struct si_qbo_state;
+
+/* This encapsulates a state or an operation which can emitted into the GPU
+ * command stream. */
+struct r600_atom {
+       void (*emit)(struct si_context *ctx, struct r600_atom *state);
+       unsigned short          id;
+};
 
 struct si_state_blend {
        struct si_pm4_state     pm4;
@@ -448,15 +456,6 @@ void si_update_prims_generated_query_state(struct si_context *sctx,
 void si_init_streamout_functions(struct si_context *sctx);
 
 
-static inline unsigned
-si_tile_mode_index(struct r600_texture *rtex, unsigned level, bool stencil)
-{
-       if (stencil)
-               return rtex->surface.u.legacy.stencil_tiling_index[level];
-       else
-               return rtex->surface.u.legacy.tiling_index[level];
-}
-
 static inline unsigned si_get_constbuf_slot(unsigned slot)
 {
        /* Constant buffers are in slots [16..31], ascending */
index e77eafe838819d06203717dde4e30b53347661d1..4cd5d2fc6b35ba3fda278fef6990cd0f555048a5 100644 (file)
@@ -25,6 +25,7 @@
 #include "si_build_pm4.h"
 
 #include "util/u_memory.h"
+#include "util/u_suballoc.h"
 
 static void si_set_streamout_enable(struct si_context *sctx, bool enable);
 
index ca8073ae84ddadbb8d18639e700c83559c1f05ad..c07a580cd4730bdccf3f49fb891b1759f592bcde 100644 (file)
@@ -31,6 +31,7 @@
 #include "util/u_pack_color.h"
 #include "util/u_resource.h"
 #include "util/u_surface.h"
+#include "util/u_transfer.h"
 #include "util/os_time.h"
 #include <errno.h>
 #include <inttypes.h>