#define SI_BASE_VERTEX_UNKNOWN INT_MIN
#define SI_RESTART_INDEX_UNKNOWN INT_MIN
#define SI_NUM_SMOOTH_AA_SAMPLES 8
-
-#define SI_TRACE_CS 0
-#define SI_TRACE_CS_DWORDS 6
-
-#define SI_MAX_DRAW_CS_DWORDS \
- (/*scratch:*/ 3 + /*derived prim state:*/ 3 + \
- /*draw regs:*/ 16 + /*draw packets:*/ 31)
+#define SI_GS_PER_ES 128
/* Instruction cache. */
#define SI_CONTEXT_INV_ICACHE (R600_CONTEXT_PRIVATE_FLAG << 0)
-/* Cache used by scalar memory (SMEM) instructions. They also use TC
- * as a second level cache, which isn't flushed by this.
- * Other names: constant cache, data cache, DCACHE */
-#define SI_CONTEXT_INV_KCACHE (R600_CONTEXT_PRIVATE_FLAG << 1)
-/* Caches used by vector memory (VMEM) instructions.
- * L1 can optionally be bypassed (GLC=1) and can only be used by shaders.
- * L2 is used by shaders and can be used by other blocks (CP, sDMA). */
-#define SI_CONTEXT_INV_TC_L1 (R600_CONTEXT_PRIVATE_FLAG << 2)
-#define SI_CONTEXT_INV_TC_L2 (R600_CONTEXT_PRIVATE_FLAG << 3)
+/* SMEM L1, other names: KCACHE, constant cache, DCACHE, data cache */
+#define SI_CONTEXT_INV_SMEM_L1 (R600_CONTEXT_PRIVATE_FLAG << 1)
+/* VMEM L1 can optionally be bypassed (GLC=1). Other names: TC L1 */
+#define SI_CONTEXT_INV_VMEM_L1 (R600_CONTEXT_PRIVATE_FLAG << 2)
+/* Used by everything except CB/DB, can be bypassed (SLC=1). Other names: TC L2 */
+#define SI_CONTEXT_INV_GLOBAL_L2 (R600_CONTEXT_PRIVATE_FLAG << 3)
/* Framebuffer caches. */
#define SI_CONTEXT_FLUSH_AND_INV_CB_META (R600_CONTEXT_PRIVATE_FLAG << 4)
#define SI_CONTEXT_FLUSH_AND_INV_DB_META (R600_CONTEXT_PRIVATE_FLAG << 5)
SI_CONTEXT_FLUSH_AND_INV_DB | \
SI_CONTEXT_FLUSH_AND_INV_DB_META)
+#define SI_ENCODE_TRACE_POINT(id) (0xcafe0000 | ((id) & 0xffff))
+#define SI_IS_TRACE_POINT(x) (((x) & 0xcafe0000) == 0xcafe0000)
+#define SI_GET_TRACE_POINT_ID(x) ((x) & 0xffff)
+
+#define SI_MAX_VIEWPORTS 16
+#define SI_MAX_BORDER_COLORS 4096
+
struct si_compute;
+struct hash_table;
struct si_screen {
struct r600_common_screen b;
+ unsigned gs_table_depth;
+
+ /* Whether shaders are monolithic (1-part) or separate (3-part). */
+ bool use_monolithic_shaders;
+
+ pipe_mutex shader_parts_mutex;
+ struct si_shader_part *vs_prologs;
+ struct si_shader_part *vs_epilogs;
+ struct si_shader_part *tcs_epilogs;
+ struct si_shader_part *ps_prologs;
+ struct si_shader_part *ps_epilogs;
+
+ /* Shader cache in memory.
+ *
+ * Design & limitations:
+ * - The shader cache is per screen (= per process), never saved to
+ * disk, and skips redundant shader compilations from TGSI to bytecode.
+ * - It can only be used with one-variant-per-shader support, in which
+ * case only the main (typically middle) part of shaders is cached.
+ * - Only VS, TCS, TES, PS are cached, out of which only the hw VS
+ * variants of VS and TES are cached, so LS and ES aren't.
+ * - GS and CS aren't cached, but it's certainly possible to cache
+ * those as well.
+ */
+ pipe_mutex shader_cache_mutex;
+ struct hash_table *shader_cache;
+};
+
+struct si_blend_color {
+ struct r600_atom atom;
+ struct pipe_blend_color state;
};
struct si_sampler_view {
struct pipe_sampler_view base;
struct list_head list;
struct r600_resource *resource;
+ struct r600_resource *dcc_buffer;
/* [0..7] = image descriptor
* [4..7] = buffer descriptor */
uint32_t state[8];
uint32_t fmask_state[8];
+ bool is_stencil_sampler;
};
struct si_sampler_state {
uint32_t val[4];
- uint32_t border_color[4];
};
struct si_cs_shader_state {
struct si_textures_info {
struct si_sampler_views views;
- struct si_sampler_states states;
uint32_t depth_texture_mask; /* which textures are depth */
uint32_t compressed_colortex_mask;
};
unsigned log_samples;
unsigned cb0_is_integer;
unsigned compressed_cb_mask;
- unsigned export_16bpc;
+ unsigned spi_shader_col_format;
+ unsigned spi_shader_col_format_alpha;
+ unsigned spi_shader_col_format_blend;
+ unsigned spi_shader_col_format_blend_alpha;
+ unsigned color_is_int8; /* bitmask */
+ unsigned dirty_cbufs;
+ bool dirty_zsbuf;
+};
+
+struct si_clip_state {
+ struct r600_atom atom;
+ struct pipe_clip_state state;
+};
+
+struct si_sample_mask {
+ struct r600_atom atom;
+ uint16_t sample_mask;
+};
+
+struct si_scissors {
+ struct r600_atom atom;
+ unsigned dirty_mask;
+ struct pipe_scissor_state states[SI_MAX_VIEWPORTS];
};
-#define SI_NUM_ATOMS(sctx) (sizeof((sctx)->atoms)/sizeof((sctx)->atoms.array[0]))
+struct si_viewports {
+ struct r600_atom atom;
+ unsigned dirty_mask;
+ struct pipe_viewport_state states[SI_MAX_VIEWPORTS];
+};
+
+/* A shader state consists of the shader selector, which is a constant state
+ * object shared by multiple contexts and shouldn't be modified, and
+ * the current shader variant selected for this context.
+ */
+struct si_shader_ctx_state {
+ struct si_shader_selector *cso;
+ struct si_shader *current;
+};
struct si_context {
struct r600_common_context b;
void *custom_blend_fastclear;
void *pstipple_sampler_state;
struct si_screen *screen;
- struct si_pm4_state *init_config;
struct pipe_fence_handle *last_gfx_fence;
- struct si_shader_selector *fixed_func_tcs_shader;
-
- union {
- struct {
- /* The order matters. */
- struct r600_atom *const_buffers[SI_NUM_SHADERS];
- struct r600_atom *rw_buffers[SI_NUM_SHADERS];
- struct r600_atom *sampler_views[SI_NUM_SHADERS];
- struct r600_atom *sampler_states[SI_NUM_SHADERS];
- /* Caches must be flushed after resource descriptors are
- * updated in memory. */
- struct r600_atom *cache_flush;
- struct r600_atom *streamout_begin;
- struct r600_atom *streamout_enable; /* must be after streamout_begin */
- struct r600_atom *framebuffer;
- struct r600_atom *msaa_sample_locs;
- struct r600_atom *db_render_state;
- struct r600_atom *msaa_config;
- struct r600_atom *clip_regs;
- struct r600_atom *shader_userdata;
- } s;
- struct r600_atom *array[0];
- } atoms;
+ struct si_shader_ctx_state fixed_func_tcs_shader;
+ LLVMTargetMachineRef tm;
+ bool gfx_flush_in_progress;
+
+ /* Atoms (direct states). */
+ union si_state_atoms atoms;
+ unsigned dirty_atoms; /* mask */
+ /* PM4 states (precomputed immutable states) */
+ union si_state queued;
+ union si_state emitted;
+ /* Atom declarations. */
+ struct r600_atom cache_flush;
struct si_framebuffer framebuffer;
- struct si_vertex_element *vertex_elements;
- /* for saving when using blitter */
- struct pipe_stencil_ref stencil_ref;
+ struct r600_atom msaa_sample_locs;
+ struct r600_atom db_render_state;
+ struct r600_atom msaa_config;
+ struct si_sample_mask sample_mask;
+ struct r600_atom cb_render_state;
+ struct si_blend_color blend_color;
+ struct r600_atom clip_regs;
+ struct si_clip_state clip_state;
+ struct si_shader_data shader_userdata;
+ struct si_scissors scissors;
+ struct si_viewports viewports;
+ struct si_stencil_ref stencil_ref;
+ struct r600_atom spi_map;
+
+ /* Precomputed states. */
+ struct si_pm4_state *init_config;
+ struct si_pm4_state *init_config_gs_rings;
+ bool init_config_has_vgt_flush;
+ struct si_pm4_state *vgt_shader_config[4];
+
/* shaders */
- struct si_shader_selector *ps_shader;
- struct si_shader_selector *gs_shader;
- struct si_shader_selector *vs_shader;
- struct si_shader_selector *tcs_shader;
- struct si_shader_selector *tes_shader;
+ struct si_shader_ctx_state ps_shader;
+ struct si_shader_ctx_state gs_shader;
+ struct si_shader_ctx_state vs_shader;
+ struct si_shader_ctx_state tcs_shader;
+ struct si_shader_ctx_state tes_shader;
struct si_cs_shader_state cs_shader_state;
- struct si_shader_data shader_userdata;
+
/* shader information */
+ struct si_vertex_element *vertex_elements;
unsigned sprite_coord_enable;
bool flatshade;
+
+ /* shader descriptors */
struct si_descriptors vertex_buffers;
struct si_buffer_resources const_buffers[SI_NUM_SHADERS];
struct si_buffer_resources rw_buffers[SI_NUM_SHADERS];
struct si_textures_info samplers[SI_NUM_SHADERS];
- struct r600_resource *scratch_buffer;
- struct r600_resource *border_color_table;
- unsigned border_color_offset;
- struct r600_atom clip_regs;
- struct r600_atom msaa_sample_locs;
- struct r600_atom msaa_config;
- int ps_iter_samples;
- bool smoothing_enabled;
+ /* other shader resources */
+ struct pipe_constant_buffer null_const_buf; /* used for set_constant_buffer(NULL) on CIK */
+ struct pipe_resource *esgs_ring;
+ struct pipe_resource *gsvs_ring;
+ struct pipe_resource *tf_ring;
+ union pipe_color_union *border_color_table; /* in CPU memory, any endian */
+ struct r600_resource *border_color_buffer;
+ union pipe_color_union *border_color_map; /* in VRAM (slow access), little endian */
+ unsigned border_color_count;
/* Vertex and index buffers. */
- bool vertex_buffers_dirty;
- struct pipe_index_buffer index_buffer;
- struct pipe_vertex_buffer vertex_buffer[SI_NUM_VERTEX_BUFFERS];
-
- /* With rasterizer discard, there doesn't have to be a pixel shader.
- * In that case, we bind this one: */
- void *dummy_pixel_shader;
- struct r600_atom cache_flush;
- struct pipe_constant_buffer null_const_buf; /* used for set_constant_buffer(NULL) on CIK */
-
- /* VGT states. */
- struct si_pm4_state *vgt_shader_config[4];
- struct si_pm4_state *gs_rings;
- struct pipe_resource *esgs_ring;
- struct pipe_resource *gsvs_ring;
- struct si_pm4_state *tf_state;
- struct pipe_resource *tf_ring;
-
- LLVMTargetMachineRef tm;
+ bool vertex_buffers_dirty;
+ struct pipe_index_buffer index_buffer;
+ struct pipe_vertex_buffer vertex_buffer[SI_NUM_VERTEX_BUFFERS];
- /* SI state handling */
- union si_state queued;
- union si_state emitted;
+ /* MSAA config state. */
+ int ps_iter_samples;
+ bool smoothing_enabled;
/* DB render state. */
- struct r600_atom db_render_state;
bool dbcb_depth_copy_enabled;
bool dbcb_stencil_copy_enabled;
unsigned dbcb_copy_sample;
- bool db_inplace_flush_enabled;
+ bool db_flush_depth_inplace;
+ bool db_flush_stencil_inplace;
bool db_depth_clear;
bool db_depth_disable_expclear;
+ bool db_stencil_clear;
+ bool db_stencil_disable_expclear;
unsigned ps_db_shader_control;
- /* Draw state. */
+ /* Emitted draw state. */
int last_base_vertex;
int last_start_instance;
int last_sh_base_reg;
int last_gs_out_prim;
int last_prim;
int last_multi_vgt_param;
+ int last_ls_hs_config;
int last_rast_prim;
unsigned last_sc_line_stipple;
int current_rast_prim; /* primitive type after TES, GS */
+ unsigned last_gsvs_itemsize;
/* Scratch buffer */
+ struct r600_resource *scratch_buffer;
boolean emit_scratch_reloc;
unsigned scratch_waves;
unsigned spi_tmpring_size;
+
+ /* Emitted derived tessellation state. */
+ struct si_shader *last_ls; /* local shader (VS) */
+ struct si_shader_selector *last_tcs;
+ int last_num_tcs_input_cp;
+ int last_tes_sh_base;
+
+ /* Debug state. */
+ bool is_debug;
+ uint32_t *last_ib;
+ unsigned last_ib_dw_size;
+ struct r600_resource *last_trace_buf;
+ struct r600_resource *trace_buf;
+ unsigned trace_id;
+ uint64_t dmesg_timestamp;
+ unsigned last_bo_count;
+ struct radeon_bo_list_item *last_bo_list;
};
/* cik_sdma.c */
unsigned src_level,
const struct pipe_box *src_box);
+/* si_cp_dma.c */
+void si_copy_buffer(struct si_context *sctx,
+ struct pipe_resource *dst, struct pipe_resource *src,
+ uint64_t dst_offset, uint64_t src_offset, unsigned size,
+ bool is_framebuffer);
+void si_init_cp_dma_functions(struct si_context *sctx);
+
+/* si_debug.c */
+void si_init_debug_functions(struct si_context *sctx);
+void si_check_vm_faults(struct si_context *sctx);
+bool si_replace_shader(unsigned num, struct radeon_shader_binary *binary);
+
/* si_dma.c */
void si_dma_copy(struct pipe_context *ctx,
struct pipe_resource *dst,
void si_context_gfx_flush(void *context, unsigned flags,
struct pipe_fence_handle **fence);
void si_begin_new_cs(struct si_context *ctx);
-void si_need_cs_space(struct si_context *ctx, unsigned num_dw, boolean count_draw_in);
-
-#if SI_TRACE_CS
-void si_trace_emit(struct si_context *sctx);
-#endif
+void si_need_cs_space(struct si_context *ctx);
/* si_compute.c */
void si_init_compute_functions(struct si_context *sctx);
+/* si_perfcounters.c */
+void si_init_perfcounters(struct si_screen *screen);
+
/* si_uvd.c */
struct pipe_video_codec *si_uvd_create_decoder(struct pipe_context *context,
const struct pipe_video_codec *templ);
sctx->last_sh_base_reg = -1; /* reset to an unknown value */
}
+static inline void
+si_set_atom_dirty(struct si_context *sctx,
+ struct r600_atom *atom, bool dirty)
+{
+ unsigned bit = 1 << (atom->id - 1);
+
+ if (dirty)
+ sctx->dirty_atoms |= bit;
+ else
+ sctx->dirty_atoms &= ~bit;
+}
+
+static inline void
+si_mark_atom_dirty(struct si_context *sctx,
+ struct r600_atom *atom)
+{
+ si_set_atom_dirty(sctx, atom, true);
+}
+
#endif