i965/fs: Add support for non-shadow textureGrad (TXD) on gen4.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>

diff --git a/src/mesa/drivers/dri/i965/brw_fs_emit.cpp b/src/mesa/drivers/dri/i965/brw_fs_emit.cpp
index 844b89b7dc322904ecc105ccbe36c1f8d2fda534..1c522cbad02420a23ecb8ac0eb28e0e69a63dfa1 100644 (file)
--- a/src/mesa/drivers/dri/i965/brw_fs_emit.cpp
+++ b/src/mesa/drivers/dri/i965/brw_fs_emit.cpp
@@ -312,7 +312,9 @@ fs_visitor::generate_tex(fs_inst *inst, struct brw_reg dst, struct brw_reg src)
         }
         break;
       case FS_OPCODE_TXD:
-        assert(!"TXD isn't supported on gen4 yet.");
+        assert(!inst->shadow_compare); // not supported yet
+        assert(inst->mlen == 7 || inst->mlen == 10);
+        msg_type = BRW_SAMPLER_MESSAGE_SIMD8_SAMPLE_GRADIENTS;
         break;
       }
    }

diff --git a/src/mesa/drivers/dri/i965/brw_fs_visitor.cpp b/src/mesa/drivers/dri/i965/brw_fs_visitor.cpp
index 638eaa416c8285379a4f5f8365cf28e28b4a0fb0..a5479e1178b39d54e739c9b03997b41a18567f97 100644 (file)
--- a/src/mesa/drivers/dri/i965/brw_fs_visitor.cpp
+++ b/src/mesa/drivers/dri/i965/brw_fs_visitor.cpp
@@ -595,7 +595,42 @@ fs_visitor::emit_texture_gen4(ir_texture *ir, fs_reg dst, fs_reg coordinate,
       /* gen4's SIMD8 sampler always has the slots for u,v,r present. */
       mlen += 3;
    } else if (ir->op == ir_txd) {
-      assert(!"TXD isn't supported on gen4 yet.");
+      ir->lod_info.grad.dPdx->accept(this);
+      fs_reg dPdx = this->result;
+
+      ir->lod_info.grad.dPdy->accept(this);
+      fs_reg dPdy = this->result;
+
+      for (int i = 0; i < ir->coordinate->type->vector_elements; i++) {
+        emit(BRW_OPCODE_MOV, fs_reg(MRF, base_mrf + mlen + i), coordinate);
+        coordinate.reg_offset++;
+      }
+      /* the slots for u and v are always present, but r is optional */
+      mlen += MAX2(ir->coordinate->type->vector_elements, 2);
+
+      /*  P   = u, v, r
+       * dPdx = dudx, dvdx, drdx
+       * dPdy = dudy, dvdy, drdy
+       *
+       * 2-arg: dudx   dvdx   dudy   dvdy
+       *        dPdx.x dPdx.y dPdy.x dPdy.y
+       *        m4     m5     m6     m7
+       *
+       * 3-arg: dudx   dvdx   drdx   dudy   dvdy   drdy
+       *        dPdx.x dPdx.y dPdx.z dPdy.x dPdy.y dPdy.z
+       *        m5     m6     m7     m8     m9     m10
+       */
+      for (int i = 0; i < ir->lod_info.grad.dPdx->type->vector_elements; i++) {
+        emit(BRW_OPCODE_MOV, fs_reg(MRF, base_mrf + mlen), dPdx);
+        dPdx.reg_offset++;
+        mlen++;
+      }
+
+      for (int i = 0; i < ir->lod_info.grad.dPdy->type->vector_elements; i++) {
+        emit(BRW_OPCODE_MOV, fs_reg(MRF, base_mrf + mlen), dPdy);
+        dPdy.reg_offset++;
+        mlen++;
+      }
    } else {
       /* Oh joy.  gen4 doesn't have SIMD8 non-shadow-compare bias/lod
        * instructions.  We'll need to do SIMD16 here.