tgsi: add support for image operations to tgsi_exec. (v2.1)

This adds support for load/store/atomic operations on images
along with image tracking support.

v2: add RESQ support. (Ilia)
v2.1: constify interface (Brian)
split get_image_coord_dim (Brian)

Reviewed-by: Brian Paul <[email protected]>
Signed-off-by: Dave Airlie <[email protected]>

2 files changed, 315 insertions, 2 deletions

diff --git a/src/gallium/auxiliary/tgsi/tgsi_exec.c b/src/gallium/auxiliary/tgsi/tgsi_exec.c
index fa1c9161bcc..53d5937b2df 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_exec.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_exec.c
@@ -853,7 +853,8 @@ void
 tgsi_exec_machine_bind_shader(
    struct tgsi_exec_machine *mach,
    const struct tgsi_token *tokens,
-   struct tgsi_sampler *sampler)
+   struct tgsi_sampler *sampler,
+   struct tgsi_image *image)
 {
    uint k;
    struct tgsi_parse_context parse;
@@ -871,6 +872,7 @@ tgsi_exec_machine_bind_shader(
 
    mach->Tokens = tokens;
    mach->Sampler = sampler;
+   mach->Image = image;
 
    if (!tokens) {
       /* unbind and free all */
@@ -3706,6 +3708,247 @@ exec_dfracexp(struct tgsi_exec_machine *mach,
    }
 }
 
+static int
+get_image_coord_dim(unsigned tgsi_tex)
+{
+   int dim;
+   switch (tgsi_tex) {
+   case TGSI_TEXTURE_BUFFER:
+   case TGSI_TEXTURE_1D:
+      dim = 1;
+      break;
+   case TGSI_TEXTURE_2D:
+   case TGSI_TEXTURE_RECT:
+   case TGSI_TEXTURE_1D_ARRAY:
+   case TGSI_TEXTURE_2D_MSAA:
+      dim = 2;
+      break;
+   case TGSI_TEXTURE_3D:
+   case TGSI_TEXTURE_CUBE:
+   case TGSI_TEXTURE_2D_ARRAY:
+   case TGSI_TEXTURE_2D_ARRAY_MSAA:
+   case TGSI_TEXTURE_CUBE_ARRAY:
+      dim = 3;
+      break;
+   default:
+      assert(!"unknown texture target");
+      dim = 0;
+      break;
+   }
+
+   return dim;
+}
+
+static int
+get_image_coord_sample(unsigned tgsi_tex)
+{
+   int sample = 0;
+   switch (tgsi_tex) {
+   case TGSI_TEXTURE_2D_MSAA:
+      sample = 3;
+      break;
+   case TGSI_TEXTURE_2D_ARRAY_MSAA:
+      sample = 4;
+      break;
+   default:
+      break;
+   }
+   return sample;
+}
+
+static void
+exec_load(struct tgsi_exec_machine *mach,
+          const struct tgsi_full_instruction *inst)
+{
+   union tgsi_exec_channel r[4], sample_r;
+   uint unit;
+   int sample;
+   int i, j;
+   int dim;
+   uint chan;
+   float rgba[TGSI_NUM_CHANNELS][TGSI_QUAD_SIZE];
+   struct tgsi_image_params params;
+   int kilmask = mach->Temps[TEMP_KILMASK_I].xyzw[TEMP_KILMASK_C].u[0];
+
+   unit = fetch_sampler_unit(mach, inst, 0);
+   dim = get_image_coord_dim(inst->Memory.Texture);
+   sample = get_image_coord_sample(inst->Memory.Texture);
+   assert(dim <= 3);
+
+   params.execmask = mach->ExecMask & mach->NonHelperMask & ~kilmask;
+   params.unit = unit;
+   params.tgsi_tex_instr = inst->Memory.Texture;
+   params.format = inst->Memory.Format;
+
+   for (i = 0; i < dim; i++) {
+      IFETCH(&r[i], 1, TGSI_CHAN_X + i);
+   }
+
+   if (sample)
+      IFETCH(&sample_r, 1, TGSI_CHAN_X + sample);
+
+   mach->Image->load(mach->Image, &params,
+                     r[0].i, r[1].i, r[2].i, sample_r.i,
+                     rgba);
+   for (j = 0; j < TGSI_QUAD_SIZE; j++) {
+      r[0].f[j] = rgba[0][j];
+      r[1].f[j] = rgba[1][j];
+      r[2].f[j] = rgba[2][j];
+      r[3].f[j] = rgba[3][j];
+   }
+   for (chan = 0; chan < TGSI_NUM_CHANNELS; chan++) {
+      if (inst->Dst[0].Register.WriteMask & (1 << chan)) {
+         store_dest(mach, &r[chan], &inst->Dst[0], inst, chan, TGSI_EXEC_DATA_FLOAT);
+      }
+   }
+}
+
+static void
+exec_store(struct tgsi_exec_machine *mach,
+           const struct tgsi_full_instruction *inst)
+{
+   union tgsi_exec_channel r[3], sample_r;
+   union tgsi_exec_channel value[4];
+   float rgba[TGSI_NUM_CHANNELS][TGSI_QUAD_SIZE];
+   struct tgsi_image_params params;
+   int dim;
+   int sample;
+   int i, j;
+   uint unit;
+   int kilmask = mach->Temps[TEMP_KILMASK_I].xyzw[TEMP_KILMASK_C].u[0];
+   unit = inst->Dst[0].Register.Index;
+   dim = get_image_coord_dim(inst->Memory.Texture);
+   sample = get_image_coord_sample(inst->Memory.Texture);
+   assert(dim <= 3);
+
+   params.execmask = mach->ExecMask & mach->NonHelperMask & ~kilmask;
+   params.unit = unit;
+   params.tgsi_tex_instr = inst->Memory.Texture;
+   params.format = inst->Memory.Format;
+
+   for (i = 0; i < dim; i++) {
+      IFETCH(&r[i], 0, TGSI_CHAN_X + i);
+   }
+
+   for (i = 0; i < 4; i++) {
+      FETCH(&value[i], 1, TGSI_CHAN_X + i);
+   }
+   if (sample)
+      IFETCH(&sample_r, 0, TGSI_CHAN_X + sample);
+
+   for (j = 0; j < TGSI_QUAD_SIZE; j++) {
+      rgba[0][j] = value[0].f[j];
+      rgba[1][j] = value[1].f[j];
+      rgba[2][j] = value[2].f[j];
+      rgba[3][j] = value[3].f[j];
+   }
+
+   mach->Image->store(mach->Image, &params,
+                      r[0].i, r[1].i, r[2].i, sample_r.i,
+                      rgba);
+}
+
+static void
+exec_atomop(struct tgsi_exec_machine *mach,
+            const struct tgsi_full_instruction *inst)
+{
+   union tgsi_exec_channel r[3], sample_r;
+   union tgsi_exec_channel value[4], value2[4];
+   float rgba[TGSI_NUM_CHANNELS][TGSI_QUAD_SIZE];
+   float rgba2[TGSI_NUM_CHANNELS][TGSI_QUAD_SIZE];
+   struct tgsi_image_params params;
+   int dim;
+   int sample;
+   int i, j;
+   uint unit, chan;
+   int kilmask = mach->Temps[TEMP_KILMASK_I].xyzw[TEMP_KILMASK_C].u[0];
+   unit = fetch_sampler_unit(mach, inst, 0);
+   dim = get_image_coord_dim(inst->Memory.Texture);
+   sample = get_image_coord_sample(inst->Memory.Texture);
+   assert(dim <= 3);
+
+   params.execmask = mach->ExecMask & mach->NonHelperMask & ~kilmask;
+   params.unit = unit;
+   params.tgsi_tex_instr = inst->Memory.Texture;
+   params.format = inst->Memory.Format;
+
+   for (i = 0; i < dim; i++) {
+      IFETCH(&r[i], 1, TGSI_CHAN_X + i);
+   }
+
+   for (i = 0; i < 4; i++) {
+      FETCH(&value[i], 2, TGSI_CHAN_X + i);
+      if (inst->Instruction.Opcode == TGSI_OPCODE_ATOMCAS)
+         FETCH(&value2[i], 3, TGSI_CHAN_X + i);
+   }
+   if (sample)
+      IFETCH(&sample_r, 1, TGSI_CHAN_X + sample);
+
+   for (j = 0; j < TGSI_QUAD_SIZE; j++) {
+      rgba[0][j] = value[0].f[j];
+      rgba[1][j] = value[1].f[j];
+      rgba[2][j] = value[2].f[j];
+      rgba[3][j] = value[3].f[j];
+   }
+   if (inst->Instruction.Opcode == TGSI_OPCODE_ATOMCAS) {
+      for (j = 0; j < TGSI_QUAD_SIZE; j++) {
+         rgba2[0][j] = value2[0].f[j];
+         rgba2[1][j] = value2[1].f[j];
+         rgba2[2][j] = value2[2].f[j];
+         rgba2[3][j] = value2[3].f[j];
+      }
+   }
+
+   mach->Image->op(mach->Image, &params, inst->Instruction.Opcode,
+                   r[0].i, r[1].i, r[2].i, sample_r.i,
+                   rgba, rgba2);
+
+   for (j = 0; j < TGSI_QUAD_SIZE; j++) {
+      r[0].f[j] = rgba[0][j];
+      r[1].f[j] = rgba[1][j];
+      r[2].f[j] = rgba[2][j];
+      r[3].f[j] = rgba[3][j];
+   }
+   for (chan = 0; chan < TGSI_NUM_CHANNELS; chan++) {
+      if (inst->Dst[0].Register.WriteMask & (1 << chan)) {
+         store_dest(mach, &r[chan], &inst->Dst[0], inst, chan, TGSI_EXEC_DATA_FLOAT);
+      }
+   }
+}
+
+static void
+exec_resq(struct tgsi_exec_machine *mach,
+          const struct tgsi_full_instruction *inst)
+{
+   int result[4];
+   union tgsi_exec_channel r[4];
+   uint unit;
+   int i, chan, j;
+   struct tgsi_image_params params;
+   int kilmask = mach->Temps[TEMP_KILMASK_I].xyzw[TEMP_KILMASK_C].u[0];
+
+   unit = fetch_sampler_unit(mach, inst, 0);
+
+   params.execmask = mach->ExecMask & mach->NonHelperMask & ~kilmask;
+   params.unit = unit;
+   params.tgsi_tex_instr = inst->Memory.Texture;
+   params.format = inst->Memory.Format;
+
+   mach->Image->get_dims(mach->Image, &params, result);
+
+   for (i = 0; i < TGSI_QUAD_SIZE; i++) {
+      for (j = 0; j < 4; j++) {
+         r[j].i[i] = result[j];
+      }
+   }
+
+   for (chan = 0; chan < TGSI_NUM_CHANNELS; chan++) {
+      if (inst->Dst[0].Register.WriteMask & (1 << chan)) {
+         store_dest(mach, &r[chan], &inst->Dst[0], inst, chan,
+                    TGSI_EXEC_DATA_INT);
+      }
+   }
+}
 
 static void
 micro_i2f(union tgsi_exec_channel *dst,
@@ -5172,6 +5415,34 @@ exec_instruction(
    case TGSI_OPCODE_D2U:
       exec_d2u(mach, inst);
       break;
+
+   case TGSI_OPCODE_LOAD:
+      exec_load(mach, inst);
+      break;
+
+   case TGSI_OPCODE_STORE:
+      exec_store(mach, inst);
+      break;
+
+   case TGSI_OPCODE_ATOMUADD:
+   case TGSI_OPCODE_ATOMXCHG:
+   case TGSI_OPCODE_ATOMCAS:
+   case TGSI_OPCODE_ATOMAND:
+   case TGSI_OPCODE_ATOMOR:
+   case TGSI_OPCODE_ATOMXOR:
+   case TGSI_OPCODE_ATOMUMIN:
+   case TGSI_OPCODE_ATOMUMAX:
+   case TGSI_OPCODE_ATOMIMIN:
+   case TGSI_OPCODE_ATOMIMAX:
+      exec_atomop(mach, inst);
+      break;
+
+   case TGSI_OPCODE_RESQ:
+      exec_resq(mach, inst);
+      break;
+   case TGSI_OPCODE_BARRIER:
+   case TGSI_OPCODE_MEMBAR:
+      break;
    default:
       assert( 0 );
    }
diff --git a/src/gallium/auxiliary/tgsi/tgsi_exec.h b/src/gallium/auxiliary/tgsi/tgsi_exec.h
index 6bcf45b4dae..2c81d5e827d 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_exec.h
+++ b/src/gallium/auxiliary/tgsi/tgsi_exec.h
@@ -98,6 +98,46 @@ enum tgsi_sampler_control
    TGSI_SAMPLER_GATHER,
 };
 
+struct tgsi_image_params {
+   unsigned unit;
+   unsigned tgsi_tex_instr;
+   enum pipe_format format;
+   unsigned execmask;
+};
+
+struct tgsi_image {
+   /* image interfaces */
+   void (*load)(const struct tgsi_image *image,
+                const struct tgsi_image_params *params,
+                const int s[TGSI_QUAD_SIZE],
+                const int t[TGSI_QUAD_SIZE],
+                const int r[TGSI_QUAD_SIZE],
+                const int sample[TGSI_QUAD_SIZE],
+                float rgba[TGSI_NUM_CHANNELS][TGSI_QUAD_SIZE]);
+
+   void (*store)(const struct tgsi_image *image,
+                 const struct tgsi_image_params *params,
+                 const int s[TGSI_QUAD_SIZE],
+                 const int t[TGSI_QUAD_SIZE],
+                 const int r[TGSI_QUAD_SIZE],
+                 const int sample[TGSI_QUAD_SIZE],
+                 float rgba[TGSI_NUM_CHANNELS][TGSI_QUAD_SIZE]);
+
+   void (*op)(const struct tgsi_image *image,
+              const struct tgsi_image_params *params,
+              unsigned opcode,
+              const int s[TGSI_QUAD_SIZE],
+              const int t[TGSI_QUAD_SIZE],
+              const int r[TGSI_QUAD_SIZE],
+              const int sample[TGSI_QUAD_SIZE],
+              float rgba[TGSI_NUM_CHANNELS][TGSI_QUAD_SIZE],
+              float rgba2[TGSI_NUM_CHANNELS][TGSI_QUAD_SIZE]);
+
+   void (*get_dims)(const struct tgsi_image *image,
+                    const struct tgsi_image_params *params,
+                    int dims[4]);
+};
+
 /**
  * Information for sampling textures, which must be implemented
  * by code outside the TGSI executor.
@@ -293,6 +333,7 @@ struct tgsi_exec_machine
 
    struct tgsi_sampler           *Sampler;
 
+   struct tgsi_image             *Image;
    unsigned                      ImmLimit;
 
    const void *Consts[PIPE_MAX_CONSTANT_BUFFERS];
@@ -382,7 +423,8 @@ void
 tgsi_exec_machine_bind_shader(
    struct tgsi_exec_machine *mach,
    const struct tgsi_token *tokens,
-   struct tgsi_sampler *sampler);
+   struct tgsi_sampler *sampler,
+   struct tgsi_image *image);
 
 uint
 tgsi_exec_machine_run(