From 1df3ecc1b87d95130165283154a13ea5b9a498d4 Mon Sep 17 00:00:00 2001 From: Rob Clark Date: Wed, 13 Jan 2016 18:43:14 -0500 Subject: [PATCH] nir: const_index helpers Direct access to intr->const_index[n], where different slots have different meanings, is somewhat confusing. Instead, let's put some extra info in nir_intrinsic_infos[] about which slots map to what, and add some get/set helpers. The helpers validate that the field being accessed (base/writemask/etc) is applicable for the intrinsic opc, for some extra safety. And nir_print can use this to dump out decoded const_index fields. Signed-off-by: Rob Clark Reviewed-by: Jason Ekstrand --- src/compiler/nir/nir.h | 64 ++++++++++- src/compiler/nir/nir_intrinsics.c | 10 +- src/compiler/nir/nir_intrinsics.h | 177 +++++++++++++++--------------- src/compiler/nir/nir_print.c | 40 +++++-- 4 files changed, 191 insertions(+), 100 deletions(-) diff --git a/src/compiler/nir/nir.h b/src/compiler/nir/nir.h index a4dbfde21d0..16203af9cfb 100644 --- a/src/compiler/nir/nir.h +++ b/src/compiler/nir/nir.h @@ -768,7 +768,7 @@ typedef struct { } nir_call_instr; #define INTRINSIC(name, num_srcs, src_components, has_dest, dest_components, \ - num_variables, num_indices, flags) \ + num_variables, num_indices, idx0, idx1, idx2, flags) \ nir_intrinsic_##name, #define LAST_INTRINSIC(name) nir_last_intrinsic = nir_intrinsic_##name, @@ -781,6 +781,8 @@ typedef enum { #undef INTRINSIC #undef LAST_INTRINSIC +#define NIR_INTRINSIC_MAX_CONST_INDEX 3 + /** Represents an intrinsic * * An intrinsic is an instruction type for handling things that are @@ -824,7 +826,7 @@ typedef struct { */ uint8_t num_components; - int const_index[3]; + int const_index[NIR_INTRINSIC_MAX_CONST_INDEX]; nir_deref_var *variables[2]; @@ -853,6 +855,39 @@ typedef enum { NIR_INTRINSIC_CAN_REORDER = (1 << 1), } nir_intrinsic_semantic_flag; +/** + * \name NIR intrinsics const-index flag + * + * Indicates the usage of a const_index slot. + * + * \sa nir_intrinsic_info::index_map + */ +typedef enum { + /** + * Generally instructions that take a offset src argument, can encode + * a constant 'base' value which is added to the offset. + */ + NIR_INTRINSIC_BASE = 1, + + /** + * For store instructions, a writemask for the store. + */ + NIR_INTRINSIC_WRMASK = 2, + + /** + * The stream-id for GS emit_vertex/end_primitive intrinsics. + */ + NIR_INTRINSIC_STREAM_ID = 3, + + /** + * The clip-plane id for load_user_clip_plane intrinsic. + */ + NIR_INTRINSIC_UCP_ID = 4, + + NIR_INTRINSIC_NUM_INDEX_FLAGS, + +} nir_intrinsic_index_flag; + #define NIR_INTRINSIC_MAX_INPUTS 4 typedef struct { @@ -882,12 +917,37 @@ typedef struct { /** the number of constant indices used by the intrinsic */ unsigned num_indices; + /** indicates the usage of intr->const_index[n] */ + unsigned index_map[NIR_INTRINSIC_NUM_INDEX_FLAGS]; + /** semantic flags for calls to this intrinsic */ nir_intrinsic_semantic_flag flags; } nir_intrinsic_info; extern const nir_intrinsic_info nir_intrinsic_infos[nir_num_intrinsics]; + +#define INTRINSIC_IDX_ACCESSORS(name, flag, type) \ +static inline type \ +nir_intrinsic_##name(nir_intrinsic_instr *instr) \ +{ \ + const nir_intrinsic_info *info = &nir_intrinsic_infos[instr->intrinsic]; \ + assert(info->index_map[NIR_INTRINSIC_##flag] > 0); \ + return instr->const_index[info->index_map[NIR_INTRINSIC_##flag] - 1]; \ +} \ +static inline void \ +nir_intrinsic_set_##name(nir_intrinsic_instr *instr, type val) \ +{ \ + const nir_intrinsic_info *info = &nir_intrinsic_infos[instr->intrinsic]; \ + assert(info->index_map[NIR_INTRINSIC_##flag] > 0); \ + instr->const_index[info->index_map[NIR_INTRINSIC_##flag] - 1] = val; \ +} + +INTRINSIC_IDX_ACCESSORS(write_mask, WRMASK, unsigned) +INTRINSIC_IDX_ACCESSORS(base, BASE, int) +INTRINSIC_IDX_ACCESSORS(stream_id, STREAM_ID, unsigned) +INTRINSIC_IDX_ACCESSORS(ucp_id, UCP_ID, unsigned) + /** * \group texture information * diff --git a/src/compiler/nir/nir_intrinsics.c b/src/compiler/nir/nir_intrinsics.c index a7c868c39af..0257b19b348 100644 --- a/src/compiler/nir/nir_intrinsics.c +++ b/src/compiler/nir/nir_intrinsics.c @@ -30,7 +30,8 @@ #define OPCODE(name) nir_intrinsic_##name #define INTRINSIC(_name, _num_srcs, _src_components, _has_dest, \ - _dest_components, _num_variables, _num_indices, _flags) \ + _dest_components, _num_variables, _num_indices, \ + idx0, idx1, idx2, _flags) \ { \ .name = #_name, \ .num_srcs = _num_srcs, \ @@ -39,9 +40,16 @@ .dest_components = _dest_components, \ .num_variables = _num_variables, \ .num_indices = _num_indices, \ + .index_map = { \ + [NIR_INTRINSIC_ ## idx0] = 1, \ + [NIR_INTRINSIC_ ## idx1] = 2, \ + [NIR_INTRINSIC_ ## idx2] = 3, \ + }, \ .flags = _flags \ }, +#define NIR_INTRINSIC_xx 0 + #define LAST_INTRINSIC(name) const nir_intrinsic_info nir_intrinsic_infos[nir_num_intrinsics] = { diff --git a/src/compiler/nir/nir_intrinsics.h b/src/compiler/nir/nir_intrinsics.h index 62eead4878a..00725562874 100644 --- a/src/compiler/nir/nir_intrinsics.h +++ b/src/compiler/nir/nir_intrinsics.h @@ -30,7 +30,7 @@ * expands to a list of macros of the form: * * INTRINSIC(name, num_srcs, src_components, has_dest, dest_components, - * num_variables, num_indices, flags) + * num_variables, num_indices, idx0, idx1, idx2, flags) * * Which should correspond one-to-one with the nir_intrinsic_info structure. It * is included in both ir.h to create the nir_intrinsic enum (with members of @@ -42,9 +42,9 @@ #define ARR(...) { __VA_ARGS__ } -INTRINSIC(load_var, 0, ARR(), true, 0, 1, 0, NIR_INTRINSIC_CAN_ELIMINATE) -INTRINSIC(store_var, 1, ARR(0), false, 0, 1, 1, 0) -INTRINSIC(copy_var, 0, ARR(), false, 0, 2, 0, 0) +INTRINSIC(load_var, 0, ARR(), true, 0, 1, 0, xx, xx, xx, NIR_INTRINSIC_CAN_ELIMINATE) +INTRINSIC(store_var, 1, ARR(0), false, 0, 1, 1, WRMASK, xx, xx, 0) +INTRINSIC(copy_var, 0, ARR(), false, 0, 2, 0, xx, xx, xx, 0) /* * Interpolation of input. The interp_var_at* intrinsics are similar to the @@ -54,25 +54,25 @@ INTRINSIC(copy_var, 0, ARR(), false, 0, 2, 0, 0) * respectively. */ -INTRINSIC(interp_var_at_centroid, 0, ARR(0), true, 0, 1, 0, +INTRINSIC(interp_var_at_centroid, 0, ARR(0), true, 0, 1, 0, xx, xx, xx, NIR_INTRINSIC_CAN_ELIMINATE | NIR_INTRINSIC_CAN_REORDER) -INTRINSIC(interp_var_at_sample, 1, ARR(1), true, 0, 1, 0, +INTRINSIC(interp_var_at_sample, 1, ARR(1), true, 0, 1, 0, xx, xx, xx, NIR_INTRINSIC_CAN_ELIMINATE | NIR_INTRINSIC_CAN_REORDER) -INTRINSIC(interp_var_at_offset, 1, ARR(2), true, 0, 1, 0, +INTRINSIC(interp_var_at_offset, 1, ARR(2), true, 0, 1, 0, xx, xx, xx, NIR_INTRINSIC_CAN_ELIMINATE | NIR_INTRINSIC_CAN_REORDER) /* * Ask the driver for the size of a given buffer. It takes the buffer index * as source. */ -INTRINSIC(get_buffer_size, 1, ARR(1), true, 1, 0, 0, +INTRINSIC(get_buffer_size, 1, ARR(1), true, 1, 0, 0, xx, xx, xx, NIR_INTRINSIC_CAN_ELIMINATE | NIR_INTRINSIC_CAN_REORDER) /* * a barrier is an intrinsic with no inputs/outputs but which can't be moved * around/optimized in general */ -#define BARRIER(name) INTRINSIC(name, 0, ARR(), false, 0, 0, 0, 0) +#define BARRIER(name) INTRINSIC(name, 0, ARR(), false, 0, 0, 0, xx, xx, xx, 0) BARRIER(barrier) BARRIER(discard) @@ -89,7 +89,7 @@ BARRIER(memory_barrier) * The latter can be used as code motion barrier, which is currently not * feasible with NIR. */ -INTRINSIC(shader_clock, 0, ARR(), true, 1, 0, 0, NIR_INTRINSIC_CAN_ELIMINATE) +INTRINSIC(shader_clock, 0, ARR(), true, 1, 0, 0, xx, xx, xx, NIR_INTRINSIC_CAN_ELIMINATE) /* * Memory barrier with semantics analogous to the compute shader @@ -103,7 +103,7 @@ BARRIER(memory_barrier_image) BARRIER(memory_barrier_shared) /** A conditional discard, with a single boolean source. */ -INTRINSIC(discard_if, 1, ARR(1), false, 0, 0, 0, 0) +INTRINSIC(discard_if, 1, ARR(1), false, 0, 0, 0, xx, xx, xx, 0) /** * Basic Geometry Shader intrinsics. @@ -113,8 +113,8 @@ INTRINSIC(discard_if, 1, ARR(1), false, 0, 0, 0, 0) * * end_primitive implements GLSL's EndPrimitive() built-in. */ -INTRINSIC(emit_vertex, 0, ARR(), false, 0, 0, 1, 0) -INTRINSIC(end_primitive, 0, ARR(), false, 0, 0, 1, 0) +INTRINSIC(emit_vertex, 0, ARR(), false, 0, 0, 1, STREAM_ID, xx, xx, 0) +INTRINSIC(end_primitive, 0, ARR(), false, 0, 0, 1, STREAM_ID, xx, xx, 0) /** * Geometry Shader intrinsics with a vertex count. @@ -125,9 +125,9 @@ INTRINSIC(end_primitive, 0, ARR(), false, 0, 0, 1, 0) * These maintain a count of the number of vertices emitted, as an additional * unsigned integer source. */ -INTRINSIC(emit_vertex_with_counter, 1, ARR(1), false, 0, 0, 1, 0) -INTRINSIC(end_primitive_with_counter, 1, ARR(1), false, 0, 0, 1, 0) -INTRINSIC(set_vertex_count, 1, ARR(1), false, 0, 0, 0, 0) +INTRINSIC(emit_vertex_with_counter, 1, ARR(1), false, 0, 0, 1, STREAM_ID, xx, xx, 0) +INTRINSIC(end_primitive_with_counter, 1, ARR(1), false, 0, 0, 1, STREAM_ID, xx, xx, 0) +INTRINSIC(set_vertex_count, 1, ARR(1), false, 0, 0, 0, xx, xx, xx, 0) /* * Atomic counters @@ -137,8 +137,8 @@ INTRINSIC(set_vertex_count, 1, ARR(1), false, 0, 0, 0, 0) */ #define ATOMIC(name, flags) \ - INTRINSIC(atomic_counter_##name##_var, 0, ARR(), true, 1, 1, 0, flags) \ - INTRINSIC(atomic_counter_##name, 1, ARR(1), true, 1, 0, 1, flags) + INTRINSIC(atomic_counter_##name##_var, 0, ARR(), true, 1, 1, 0, xx, xx, xx, flags) \ + INTRINSIC(atomic_counter_##name, 1, ARR(1), true, 1, 0, 1, BASE, xx, xx, flags) ATOMIC(inc, 0) ATOMIC(dec, 0) @@ -159,20 +159,20 @@ ATOMIC(read, NIR_INTRINSIC_CAN_ELIMINATE) * either one or two additional scalar arguments with the same meaning as in * the ARB_shader_image_load_store specification. */ -INTRINSIC(image_load, 2, ARR(4, 1), true, 4, 1, 0, +INTRINSIC(image_load, 2, ARR(4, 1), true, 4, 1, 0, xx, xx, xx, NIR_INTRINSIC_CAN_ELIMINATE) -INTRINSIC(image_store, 3, ARR(4, 1, 4), false, 0, 1, 0, 0) -INTRINSIC(image_atomic_add, 3, ARR(4, 1, 1), true, 1, 1, 0, 0) -INTRINSIC(image_atomic_min, 3, ARR(4, 1, 1), true, 1, 1, 0, 0) -INTRINSIC(image_atomic_max, 3, ARR(4, 1, 1), true, 1, 1, 0, 0) -INTRINSIC(image_atomic_and, 3, ARR(4, 1, 1), true, 1, 1, 0, 0) -INTRINSIC(image_atomic_or, 3, ARR(4, 1, 1), true, 1, 1, 0, 0) -INTRINSIC(image_atomic_xor, 3, ARR(4, 1, 1), true, 1, 1, 0, 0) -INTRINSIC(image_atomic_exchange, 3, ARR(4, 1, 1), true, 1, 1, 0, 0) -INTRINSIC(image_atomic_comp_swap, 4, ARR(4, 1, 1, 1), true, 1, 1, 0, 0) -INTRINSIC(image_size, 0, ARR(), true, 4, 1, 0, +INTRINSIC(image_store, 3, ARR(4, 1, 4), false, 0, 1, 0, xx, xx, xx, 0) +INTRINSIC(image_atomic_add, 3, ARR(4, 1, 1), true, 1, 1, 0, xx, xx, xx, 0) +INTRINSIC(image_atomic_min, 3, ARR(4, 1, 1), true, 1, 1, 0, xx, xx, xx, 0) +INTRINSIC(image_atomic_max, 3, ARR(4, 1, 1), true, 1, 1, 0, xx, xx, xx, 0) +INTRINSIC(image_atomic_and, 3, ARR(4, 1, 1), true, 1, 1, 0, xx, xx, xx, 0) +INTRINSIC(image_atomic_or, 3, ARR(4, 1, 1), true, 1, 1, 0, xx, xx, xx, 0) +INTRINSIC(image_atomic_xor, 3, ARR(4, 1, 1), true, 1, 1, 0, xx, xx, xx, 0) +INTRINSIC(image_atomic_exchange, 3, ARR(4, 1, 1), true, 1, 1, 0, xx, xx, xx, 0) +INTRINSIC(image_atomic_comp_swap, 4, ARR(4, 1, 1, 1), true, 1, 1, 0, xx, xx, xx, 0) +INTRINSIC(image_size, 0, ARR(), true, 4, 1, 0, xx, xx, xx, NIR_INTRINSIC_CAN_ELIMINATE | NIR_INTRINSIC_CAN_REORDER) -INTRINSIC(image_samples, 0, ARR(), true, 1, 1, 0, +INTRINSIC(image_samples, 0, ARR(), true, 1, 1, 0, xx, xx, xx, NIR_INTRINSIC_CAN_ELIMINATE | NIR_INTRINSIC_CAN_REORDER) /* @@ -192,16 +192,16 @@ INTRINSIC(image_samples, 0, ARR(), true, 1, 1, 0, * in ssbo_atomic_add, etc). * 3: For CompSwap only: the second data parameter. */ -INTRINSIC(ssbo_atomic_add, 3, ARR(1, 1, 1), true, 1, 0, 0, 0) -INTRINSIC(ssbo_atomic_imin, 3, ARR(1, 1, 1), true, 1, 0, 0, 0) -INTRINSIC(ssbo_atomic_umin, 3, ARR(1, 1, 1), true, 1, 0, 0, 0) -INTRINSIC(ssbo_atomic_imax, 3, ARR(1, 1, 1), true, 1, 0, 0, 0) -INTRINSIC(ssbo_atomic_umax, 3, ARR(1, 1, 1), true, 1, 0, 0, 0) -INTRINSIC(ssbo_atomic_and, 3, ARR(1, 1, 1), true, 1, 0, 0, 0) -INTRINSIC(ssbo_atomic_or, 3, ARR(1, 1, 1), true, 1, 0, 0, 0) -INTRINSIC(ssbo_atomic_xor, 3, ARR(1, 1, 1), true, 1, 0, 0, 0) -INTRINSIC(ssbo_atomic_exchange, 3, ARR(1, 1, 1), true, 1, 0, 0, 0) -INTRINSIC(ssbo_atomic_comp_swap, 4, ARR(1, 1, 1, 1), true, 1, 0, 0, 0) +INTRINSIC(ssbo_atomic_add, 3, ARR(1, 1, 1), true, 1, 0, 0, xx, xx, xx, 0) +INTRINSIC(ssbo_atomic_imin, 3, ARR(1, 1, 1), true, 1, 0, 0, xx, xx, xx, 0) +INTRINSIC(ssbo_atomic_umin, 3, ARR(1, 1, 1), true, 1, 0, 0, xx, xx, xx, 0) +INTRINSIC(ssbo_atomic_imax, 3, ARR(1, 1, 1), true, 1, 0, 0, xx, xx, xx, 0) +INTRINSIC(ssbo_atomic_umax, 3, ARR(1, 1, 1), true, 1, 0, 0, xx, xx, xx, 0) +INTRINSIC(ssbo_atomic_and, 3, ARR(1, 1, 1), true, 1, 0, 0, xx, xx, xx, 0) +INTRINSIC(ssbo_atomic_or, 3, ARR(1, 1, 1), true, 1, 0, 0, xx, xx, xx, 0) +INTRINSIC(ssbo_atomic_xor, 3, ARR(1, 1, 1), true, 1, 0, 0, xx, xx, xx, 0) +INTRINSIC(ssbo_atomic_exchange, 3, ARR(1, 1, 1), true, 1, 0, 0, xx, xx, xx, 0) +INTRINSIC(ssbo_atomic_comp_swap, 4, ARR(1, 1, 1, 1), true, 1, 0, 0, xx, xx, xx, 0) /* * CS shared variable atomic intrinsics @@ -219,42 +219,43 @@ INTRINSIC(ssbo_atomic_comp_swap, 4, ARR(1, 1, 1, 1), true, 1, 0, 0, 0) * in shared_atomic_add, etc). * 2: For CompSwap only: the second data parameter. */ -INTRINSIC(shared_atomic_add, 2, ARR(1, 1), true, 1, 0, 0, 0) -INTRINSIC(shared_atomic_imin, 2, ARR(1, 1), true, 1, 0, 0, 0) -INTRINSIC(shared_atomic_umin, 2, ARR(1, 1), true, 1, 0, 0, 0) -INTRINSIC(shared_atomic_imax, 2, ARR(1, 1), true, 1, 0, 0, 0) -INTRINSIC(shared_atomic_umax, 2, ARR(1, 1), true, 1, 0, 0, 0) -INTRINSIC(shared_atomic_and, 2, ARR(1, 1), true, 1, 0, 0, 0) -INTRINSIC(shared_atomic_or, 2, ARR(1, 1), true, 1, 0, 0, 0) -INTRINSIC(shared_atomic_xor, 2, ARR(1, 1), true, 1, 0, 0, 0) -INTRINSIC(shared_atomic_exchange, 2, ARR(1, 1), true, 1, 0, 0, 0) -INTRINSIC(shared_atomic_comp_swap, 3, ARR(1, 1, 1), true, 1, 0, 0, 0) +INTRINSIC(shared_atomic_add, 2, ARR(1, 1), true, 1, 0, 0, xx, xx, xx, 0) +INTRINSIC(shared_atomic_imin, 2, ARR(1, 1), true, 1, 0, 0, xx, xx, xx, 0) +INTRINSIC(shared_atomic_umin, 2, ARR(1, 1), true, 1, 0, 0, xx, xx, xx, 0) +INTRINSIC(shared_atomic_imax, 2, ARR(1, 1), true, 1, 0, 0, xx, xx, xx, 0) +INTRINSIC(shared_atomic_umax, 2, ARR(1, 1), true, 1, 0, 0, xx, xx, xx, 0) +INTRINSIC(shared_atomic_and, 2, ARR(1, 1), true, 1, 0, 0, xx, xx, xx, 0) +INTRINSIC(shared_atomic_or, 2, ARR(1, 1), true, 1, 0, 0, xx, xx, xx, 0) +INTRINSIC(shared_atomic_xor, 2, ARR(1, 1), true, 1, 0, 0, xx, xx, xx, 0) +INTRINSIC(shared_atomic_exchange, 2, ARR(1, 1), true, 1, 0, 0, xx, xx, xx, 0) +INTRINSIC(shared_atomic_comp_swap, 3, ARR(1, 1, 1), true, 1, 0, 0, xx, xx, xx, 0) -#define SYSTEM_VALUE(name, components, num_indices) \ +#define SYSTEM_VALUE(name, components, num_indices, idx0, idx1, idx2) \ INTRINSIC(load_##name, 0, ARR(), true, components, 0, num_indices, \ + idx0, idx1, idx2, \ NIR_INTRINSIC_CAN_ELIMINATE | NIR_INTRINSIC_CAN_REORDER) -SYSTEM_VALUE(front_face, 1, 0) -SYSTEM_VALUE(vertex_id, 1, 0) -SYSTEM_VALUE(vertex_id_zero_base, 1, 0) -SYSTEM_VALUE(base_vertex, 1, 0) -SYSTEM_VALUE(instance_id, 1, 0) -SYSTEM_VALUE(base_instance, 1, 0) -SYSTEM_VALUE(draw_id, 1, 0) -SYSTEM_VALUE(sample_id, 1, 0) -SYSTEM_VALUE(sample_pos, 2, 0) -SYSTEM_VALUE(sample_mask_in, 1, 0) -SYSTEM_VALUE(primitive_id, 1, 0) -SYSTEM_VALUE(invocation_id, 1, 0) -SYSTEM_VALUE(tess_coord, 3, 0) -SYSTEM_VALUE(tess_level_outer, 4, 0) -SYSTEM_VALUE(tess_level_inner, 2, 0) -SYSTEM_VALUE(patch_vertices_in, 1, 0) -SYSTEM_VALUE(local_invocation_id, 3, 0) -SYSTEM_VALUE(work_group_id, 3, 0) -SYSTEM_VALUE(user_clip_plane, 4, 1) /* const_index[0] is user_clip_plane[idx] */ -SYSTEM_VALUE(num_work_groups, 3, 0) -SYSTEM_VALUE(helper_invocation, 1, 0) +SYSTEM_VALUE(front_face, 1, 0, xx, xx, xx) +SYSTEM_VALUE(vertex_id, 1, 0, xx, xx, xx) +SYSTEM_VALUE(vertex_id_zero_base, 1, 0, xx, xx, xx) +SYSTEM_VALUE(base_vertex, 1, 0, xx, xx, xx) +SYSTEM_VALUE(instance_id, 1, 0, xx, xx, xx) +SYSTEM_VALUE(base_instance, 1, 0, xx, xx, xx) +SYSTEM_VALUE(draw_id, 1, 0, xx, xx, xx) +SYSTEM_VALUE(sample_id, 1, 0, xx, xx, xx) +SYSTEM_VALUE(sample_pos, 2, 0, xx, xx, xx) +SYSTEM_VALUE(sample_mask_in, 1, 0, xx, xx, xx) +SYSTEM_VALUE(primitive_id, 1, 0, xx, xx, xx) +SYSTEM_VALUE(invocation_id, 1, 0, xx, xx, xx) +SYSTEM_VALUE(tess_coord, 3, 0, xx, xx, xx) +SYSTEM_VALUE(tess_level_outer, 4, 0, xx, xx, xx) +SYSTEM_VALUE(tess_level_inner, 2, 0, xx, xx, xx) +SYSTEM_VALUE(patch_vertices_in, 1, 0, xx, xx, xx) +SYSTEM_VALUE(local_invocation_id, 3, 0, xx, xx, xx) +SYSTEM_VALUE(work_group_id, 3, 0, xx, xx, xx) +SYSTEM_VALUE(user_clip_plane, 4, 1, UCP_ID, xx, xx) +SYSTEM_VALUE(num_work_groups, 3, 0, xx, xx, xx) +SYSTEM_VALUE(helper_invocation, 1, 0, xx, xx, xx) /* * Load operations pull data from some piece of GPU memory. All load @@ -274,25 +275,25 @@ SYSTEM_VALUE(helper_invocation, 1, 0) * offsets are always in bytes. */ -#define LOAD(name, srcs, indices, flags) \ - INTRINSIC(load_##name, srcs, ARR(1, 1, 1, 1), true, 0, 0, indices, flags) +#define LOAD(name, srcs, num_indices, idx0, idx1, idx2, flags) \ + INTRINSIC(load_##name, srcs, ARR(1, 1, 1, 1), true, 0, 0, num_indices, idx0, idx1, idx2, flags) /* src[] = { offset }. const_index[] = { base } */ -LOAD(uniform, 1, 1, NIR_INTRINSIC_CAN_ELIMINATE | NIR_INTRINSIC_CAN_REORDER) +LOAD(uniform, 1, 1, BASE, xx, xx, NIR_INTRINSIC_CAN_ELIMINATE | NIR_INTRINSIC_CAN_REORDER) /* src[] = { buffer_index, offset }. No const_index */ -LOAD(ubo, 2, 0, NIR_INTRINSIC_CAN_ELIMINATE | NIR_INTRINSIC_CAN_REORDER) +LOAD(ubo, 2, 0, xx, xx, xx, NIR_INTRINSIC_CAN_ELIMINATE | NIR_INTRINSIC_CAN_REORDER) /* src[] = { offset }. const_index[] = { base } */ -LOAD(input, 1, 1, NIR_INTRINSIC_CAN_ELIMINATE | NIR_INTRINSIC_CAN_REORDER) +LOAD(input, 1, 1, BASE, xx, xx, NIR_INTRINSIC_CAN_ELIMINATE | NIR_INTRINSIC_CAN_REORDER) /* src[] = { vertex, offset }. const_index[] = { base } */ -LOAD(per_vertex_input, 2, 1, NIR_INTRINSIC_CAN_ELIMINATE | NIR_INTRINSIC_CAN_REORDER) +LOAD(per_vertex_input, 2, 1, BASE, xx, xx, NIR_INTRINSIC_CAN_ELIMINATE | NIR_INTRINSIC_CAN_REORDER) /* src[] = { buffer_index, offset }. No const_index */ -LOAD(ssbo, 2, 0, NIR_INTRINSIC_CAN_ELIMINATE) +LOAD(ssbo, 2, 0, xx, xx, xx, NIR_INTRINSIC_CAN_ELIMINATE) /* src[] = { offset }. const_index[] = { base } */ -LOAD(output, 1, 1, NIR_INTRINSIC_CAN_ELIMINATE) +LOAD(output, 1, 1, BASE, xx, xx, NIR_INTRINSIC_CAN_ELIMINATE) /* src[] = { vertex, offset }. const_index[] = { base } */ -LOAD(per_vertex_output, 2, 1, NIR_INTRINSIC_CAN_ELIMINATE) +LOAD(per_vertex_output, 2, 1, BASE, xx, xx, NIR_INTRINSIC_CAN_ELIMINATE) /* src[] = { offset }. const_index[] = { base } */ -LOAD(shared, 1, 1, NIR_INTRINSIC_CAN_ELIMINATE) +LOAD(shared, 1, 1, BASE, xx, xx, NIR_INTRINSIC_CAN_ELIMINATE) /* * Stores work the same way as loads, except now the first source is the value @@ -301,16 +302,16 @@ LOAD(shared, 1, 1, NIR_INTRINSIC_CAN_ELIMINATE) * const_index[0]. */ -#define STORE(name, srcs, indices, flags) \ - INTRINSIC(store_##name, srcs, ARR(0, 1, 1, 1), false, 0, 0, indices, flags) +#define STORE(name, srcs, num_indices, idx0, idx1, idx2, flags) \ + INTRINSIC(store_##name, srcs, ARR(0, 1, 1, 1), false, 0, 0, num_indices, idx0, idx1, idx2, flags) /* src[] = { value, offset }. const_index[] = { base, write_mask } */ -STORE(output, 2, 2, 0) +STORE(output, 2, 2, BASE, WRMASK, xx, 0) /* src[] = { value, vertex, offset }. const_index[] = { base, write_mask } */ -STORE(per_vertex_output, 3, 2, 0) +STORE(per_vertex_output, 3, 2, BASE, WRMASK, xx, 0) /* src[] = { value, block_index, offset }. const_index[] = { write_mask } */ -STORE(ssbo, 3, 1, 0) +STORE(ssbo, 3, 1, WRMASK, xx, xx, 0) /* src[] = { value, offset }. const_index[] = { base, write_mask } */ -STORE(shared, 2, 2, 0) +STORE(shared, 2, 2, BASE, WRMASK, xx, 0) LAST_INTRINSIC(store_shared) diff --git a/src/compiler/nir/nir_print.c b/src/compiler/nir/nir_print.c index 48ecb48a620..101688eb00f 100644 --- a/src/compiler/nir/nir_print.c +++ b/src/compiler/nir/nir_print.c @@ -444,15 +444,16 @@ print_deref(nir_deref_var *deref, print_state *state) static void print_intrinsic_instr(nir_intrinsic_instr *instr, print_state *state) { - unsigned num_srcs = nir_intrinsic_infos[instr->intrinsic].num_srcs; + const nir_intrinsic_info *info = &nir_intrinsic_infos[instr->intrinsic]; + unsigned num_srcs = info->num_srcs; FILE *fp = state->fp; - if (nir_intrinsic_infos[instr->intrinsic].has_dest) { + if (info->has_dest) { print_dest(&instr->dest, state); fprintf(fp, " = "); } - fprintf(fp, "intrinsic %s (", nir_intrinsic_infos[instr->intrinsic].name); + fprintf(fp, "intrinsic %s (", info->name); for (unsigned i = 0; i < num_srcs; i++) { if (i != 0) @@ -463,9 +464,7 @@ print_intrinsic_instr(nir_intrinsic_instr *instr, print_state *state) fprintf(fp, ") ("); - unsigned num_vars = nir_intrinsic_infos[instr->intrinsic].num_variables; - - for (unsigned i = 0; i < num_vars; i++) { + for (unsigned i = 0; i < info->num_variables; i++) { if (i != 0) fprintf(fp, ", "); @@ -474,9 +473,7 @@ print_intrinsic_instr(nir_intrinsic_instr *instr, print_state *state) fprintf(fp, ") ("); - unsigned num_indices = nir_intrinsic_infos[instr->intrinsic].num_indices; - - for (unsigned i = 0; i < num_indices; i++) { + for (unsigned i = 0; i < info->num_indices; i++) { if (i != 0) fprintf(fp, ", "); @@ -485,6 +482,31 @@ print_intrinsic_instr(nir_intrinsic_instr *instr, print_state *state) fprintf(fp, ")"); + static const char *index_name[NIR_INTRINSIC_NUM_INDEX_FLAGS] = { + [NIR_INTRINSIC_BASE] = "base", + [NIR_INTRINSIC_WRMASK] = "wrmask", + [NIR_INTRINSIC_STREAM_ID] = "stream-id", + [NIR_INTRINSIC_UCP_ID] = "ucp-id", + }; + for (unsigned idx = 1; idx < NIR_INTRINSIC_NUM_INDEX_FLAGS; idx++) { + if (!info->index_map[idx]) + continue; + fprintf(fp, " /*"); + if (idx == NIR_INTRINSIC_WRMASK) { + /* special case wrmask to show it as a writemask.. */ + unsigned wrmask = nir_intrinsic_write_mask(instr); + fprintf(fp, " wrmask="); + for (unsigned i = 0; i < 4; i++) + if ((wrmask >> i) & 1) + fprintf(fp, "%c", "xyzw"[i]); + } else { + unsigned off = info->index_map[idx] - 1; + assert(index_name[idx]); /* forgot to update index_name table? */ + fprintf(fp, " %s=%d", index_name[idx], instr->const_index[off]); + } + fprintf(fp, " */"); + } + if (!state->shader) return; -- 2.30.2