X-Git-Url: https://git.libre-soc.org/?a=blobdiff_plain;f=src%2Fgallium%2Fdrivers%2Fcell%2Fspu%2Fspu_exec.c;h=0eaae2e451bc12258b213931fb5302cfa3988250;hb=ba002eb1965ccdb09d7c3657d32e22e71ed4096a;hp=cf81bee8fde54b4b76a32c4af40887a7bb6a5194;hpb=2d1f086c12b6d64f5c3fb80474f26775aeb71370;p=mesa.git

diff --git a/src/gallium/drivers/cell/spu/spu_exec.c b/src/gallium/drivers/cell/spu/spu_exec.c
index cf81bee8fde..0eaae2e451b 100644
--- a/src/gallium/drivers/cell/spu/spu_exec.c
+++ b/src/gallium/drivers/cell/spu/spu_exec.c
@@ -50,8 +50,6 @@
  *   Brian Paul
  */
 
-#include <libmisc.h>
-#include <spu_mfcio.h>
 #include <transpose_matrix4x4.h>
 #include <simdmath/ceilf4.h>
 #include <simdmath/cosf4.h>
@@ -65,10 +63,9 @@
 
 #include "pipe/p_compiler.h"
 #include "pipe/p_state.h"
-#include "pipe/p_util.h"
 #include "pipe/p_shader_tokens.h"
-#include "tgsi/util/tgsi_parse.h"
-#include "tgsi/util/tgsi_util.h"
+#include "tgsi/tgsi_parse.h"
+#include "tgsi/tgsi_util.h"
 #include "spu_exec.h"
 #include "spu_main.h"
 #include "spu_vertex_shader.h"
@@ -151,6 +148,7 @@ spu_exec_machine_init(struct spu_exec_machine *mach,
    const qword zero = si_il(0);
    const qword not_zero = si_il(~0);
 
+   (void) numSamplers;
    mach->Samplers = samplers;
    mach->Processor = processor;
    mach->Addrs = &mach->Temps[TGSI_EXEC_NUM_TEMPS];
@@ -384,10 +382,10 @@ fetch_src_file_channel(
          break;
 
       case TGSI_FILE_IMMEDIATE:
-         assert( index->i[0] < (int) mach->ImmLimit );
-         assert( index->i[1] < (int) mach->ImmLimit );
-         assert( index->i[2] < (int) mach->ImmLimit );
-         assert( index->i[3] < (int) mach->ImmLimit );
+         ASSERT( index->i[0] < (int) mach->ImmLimit );
+         ASSERT( index->i[1] < (int) mach->ImmLimit );
+         ASSERT( index->i[2] < (int) mach->ImmLimit );
+         ASSERT( index->i[3] < (int) mach->ImmLimit );
 
          chan->f[0] = mach->Imms[index->i[0]][swizzle];
          chan->f[1] = mach->Imms[index->i[1]][swizzle];
@@ -411,7 +409,7 @@ fetch_src_file_channel(
          break;
 
       default:
-         assert( 0 );
+         ASSERT( 0 );
       }
       break;
 
@@ -424,7 +422,7 @@ fetch_src_file_channel(
       break;
 
    default:
-      assert( 0 );
+      ASSERT( 0 );
    }
 }
 
@@ -473,7 +471,7 @@ fetch_source(
          index.q = si_shli(index.q, 12);
          break;
       default:
-         assert( 0 );
+         ASSERT( 0 );
       }
 
       index.i[0] += reg->SrcRegisterDim.Index;
@@ -560,7 +558,7 @@ store_dest(
       break;
 
    default:
-      assert( 0 );
+      ASSERT( 0 );
       return;
    }
 
@@ -584,11 +582,11 @@ store_dest(
       break;
 
    case TGSI_SAT_MINUS_PLUS_ONE:
-      assert( 0 );
+      ASSERT( 0 );
       break;
 
    default:
-      assert( 0 );
+      ASSERT( 0 );
    }
 }
 
@@ -604,8 +602,8 @@ store_dest(
  * Kill fragment if any of the four values is less than zero.
  */
 static void
-exec_kilp(struct spu_exec_machine *mach,
-          const struct tgsi_full_instruction *inst)
+exec_kil(struct spu_exec_machine *mach,
+         const struct tgsi_full_instruction *inst)
 {
    uint uniquemask;
    uint chan_index;
@@ -641,6 +639,20 @@ exec_kilp(struct spu_exec_machine *mach,
    mach->Temps[TEMP_KILMASK_I].xyzw[TEMP_KILMASK_C].u[0] |= kilmask;
 }
 
+/**
+ * Execute NVIDIA-style KIL which is predicated by a condition code.
+ * Kill fragment if the condition code is TRUE.
+ */
+static void
+exec_kilp(struct tgsi_exec_machine *mach,
+          const struct tgsi_full_instruction *inst)
+{
+   uint kilmask = 0; /* bit 0 = pixel 0, bit 1 = pixel 1, etc */
+
+   /* TODO: build kilmask from CC mask */
+
+   mach->Temps[TEMP_KILMASK_I].xyzw[TEMP_KILMASK_C].u[0] |= kilmask;
+}
 
 /*
  * Fetch a texel using STR texture coordinates.
@@ -659,9 +671,10 @@ fetch_texel( struct spu_sampler *sampler,
    qword rgba[4];
    qword out[4];
 
-   sampler->get_samples(sampler, s->f, t->f, p->f, lodbias, (float *) rgba);
+   sampler->get_samples(sampler, s->f, t->f, p->f, lodbias, 
+			(float (*)[4]) rgba);
 
-   _transpose_matrix4x4(out, rgba);
+   _transpose_matrix4x4((vec_float4 *) out, (vec_float4 *) rgba);
    r->q = out[0];
    g->q = out[1];
    b->q = out[2];
@@ -672,7 +685,7 @@ fetch_texel( struct spu_sampler *sampler,
 static void
 exec_tex(struct spu_exec_machine *mach,
          const struct tgsi_full_instruction *inst,
-         boolean biasLod)
+         boolean biasLod, boolean projected)
 {
    const uint unit = inst->FullSrcRegisters[1].SrcRegister.Index;
    union spu_exec_channel r[8];
@@ -686,17 +699,9 @@ exec_tex(struct spu_exec_machine *mach,
 
       FETCH(&r[0], 0, CHAN_X);
 
-      switch (inst->FullSrcRegisters[0].SrcRegisterExtSwz.ExtDivide) {
-      case TGSI_EXTSWIZZLE_W:
+      if (projected) {
          FETCH(&r[1], 0, CHAN_W);
          r[0].q = micro_div(r[0].q, r[1].q);
-         break;
-
-      case TGSI_EXTSWIZZLE_ONE:
-         break;
-
-      default:
-         assert (0);
       }
 
       if (biasLod) {
@@ -718,19 +723,11 @@ exec_tex(struct spu_exec_machine *mach,
       FETCH(&r[1], 0, CHAN_Y);
       FETCH(&r[2], 0, CHAN_Z);
 
-      switch (inst->FullSrcRegisters[0].SrcRegisterExtSwz.ExtDivide) {
-      case TGSI_EXTSWIZZLE_W:
+      if (projected) {
          FETCH(&r[3], 0, CHAN_W);
          r[0].q = micro_div(r[0].q, r[3].q);
          r[1].q = micro_div(r[1].q, r[3].q);
          r[2].q = micro_div(r[2].q, r[3].q);
-         break;
-
-      case TGSI_EXTSWIZZLE_ONE:
-         break;
-
-      default:
-         assert (0);
       }
 
       if (biasLod) {
@@ -752,19 +749,11 @@ exec_tex(struct spu_exec_machine *mach,
       FETCH(&r[1], 0, CHAN_Y);
       FETCH(&r[2], 0, CHAN_Z);
 
-      switch (inst->FullSrcRegisters[0].SrcRegisterExtSwz.ExtDivide) {
-      case TGSI_EXTSWIZZLE_W:
+      if (projected) {
          FETCH(&r[3], 0, CHAN_W);
          r[0].q = micro_div(r[0].q, r[3].q);
          r[1].q = micro_div(r[1].q, r[3].q);
          r[2].q = micro_div(r[2].q, r[3].q);
-         break;
-
-      case TGSI_EXTSWIZZLE_ONE:
-         break;
-
-      default:
-         assert (0);
       }
 
       if (biasLod) {
@@ -780,7 +769,7 @@ exec_tex(struct spu_exec_machine *mach,
       break;
 
    default:
-      assert (0);
+      ASSERT (0);
    }
 
    FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
@@ -854,13 +843,11 @@ exec_declaration(struct spu_exec_machine *mach,
          unsigned first, last, mask;
          interpolation_func interp;
 
-         assert( decl->Declaration.Declare == TGSI_DECLARE_RANGE );
-
-         first = decl->u.DeclarationRange.First;
-         last = decl->u.DeclarationRange.Last;
+         first = decl->DeclarationRange.First;
+         last = decl->DeclarationRange.Last;
          mask = decl->Declaration.UsageMask;
 
-         switch( decl->Interpolation.Interpolate ) {
+         switch( decl->Declaration.Interpolate ) {
          case TGSI_INTERPOLATE_CONSTANT:
             interp = constant_interpolation;
             break;
@@ -874,7 +861,7 @@ exec_declaration(struct spu_exec_machine *mach,
             break;
 
          default:
-            assert( 0 );
+            ASSERT( 0 );
          }
 
          if( mask == TGSI_WRITEMASK_XYZW ) {
@@ -922,7 +909,7 @@ exec_instruction(
       break;
 
    case TGSI_OPCODE_MOV:
-   /* TGSI_OPCODE_SWZ */
+   case TGSI_OPCODE_SWZ:
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( &r[0], 0, chan_index );
          STORE( &r[0], 0, chan_index );
@@ -965,7 +952,6 @@ exec_instruction(
       break;
 
    case TGSI_OPCODE_RCP:
-   /* TGSI_OPCODE_RECIP */
       FETCH( &r[0], 0, CHAN_X );
       r[0].q = micro_div(mach->Temps[TEMP_1_I].xyzw[TEMP_1_C].q, r[0].q);
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
@@ -974,7 +960,6 @@ exec_instruction(
       break;
 
    case TGSI_OPCODE_RSQ:
-   /* TGSI_OPCODE_RECIPSQRT */
       FETCH( &r[0], 0, CHAN_X );
       r[0].q = micro_sqrt(r[0].q);
       r[0].q = micro_div(mach->Temps[TEMP_1_I].xyzw[TEMP_1_C].q, r[0].q);
@@ -984,11 +969,11 @@ exec_instruction(
       break;
 
    case TGSI_OPCODE_EXP:
-      assert (0);
+      ASSERT (0);
       break;
 
    case TGSI_OPCODE_LOG:
-      assert (0);
+      ASSERT (0);
       break;
 
    case TGSI_OPCODE_MUL:
@@ -1128,7 +1113,6 @@ exec_instruction(
       break;
 
    case TGSI_OPCODE_MAD:
-   /* TGSI_OPCODE_MADD */
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( &r[0], 0, chan_index );
          FETCH( &r[1], 1, chan_index );
@@ -1149,8 +1133,7 @@ exec_instruction(
       }
       break;
 
-   case TGSI_OPCODE_LERP:
-   /* TGSI_OPCODE_LRP */
+   case TGSI_OPCODE_LRP:
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH(&r[0], 0, chan_index);
          FETCH(&r[1], 1, chan_index);
@@ -1164,28 +1147,14 @@ exec_instruction(
       break;
 
    case TGSI_OPCODE_CND:
-      assert (0);
-      break;
-
-   case TGSI_OPCODE_CND0:
-      assert (0);
-      break;
-
-   case TGSI_OPCODE_DOT2ADD:
-      /* TGSI_OPCODE_DP2A */
-      assert (0);
-      break;
-
-   case TGSI_OPCODE_INDEX:
-      assert (0);
+      ASSERT (0);
       break;
 
-   case TGSI_OPCODE_NEGATE:
-      assert (0);
+   case TGSI_OPCODE_DP2A:
+      ASSERT (0);
       break;
 
-   case TGSI_OPCODE_FRAC:
-   /* TGSI_OPCODE_FRC */
+   case TGSI_OPCODE_FRC:
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( &r[0], 0, chan_index );
          r[0].q = micro_frc(r[0].q);
@@ -1194,11 +1163,10 @@ exec_instruction(
       break;
 
    case TGSI_OPCODE_CLAMP:
-      assert (0);
+      ASSERT (0);
       break;
 
-   case TGSI_OPCODE_FLOOR:
-   /* TGSI_OPCODE_FLR */
+   case TGSI_OPCODE_FLR:
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
          FETCH( &r[0], 0, chan_index );
          r[0].q = micro_flr(r[0].q);
@@ -1214,8 +1182,7 @@ exec_instruction(
       }
       break;
 
-   case TGSI_OPCODE_EXPBASE2:
-    /* TGSI_OPCODE_EX2 */
+   case TGSI_OPCODE_EX2:
       FETCH(&r[0], 0, CHAN_X);
 
       r[0].q = micro_pow(mach->Temps[TEMP_2_I].xyzw[TEMP_2_C].q, r[0].q);
@@ -1225,8 +1192,7 @@ exec_instruction(
       }
       break;
 
-   case TGSI_OPCODE_LOGBASE2:
-   /* TGSI_OPCODE_LG2 */
+   case TGSI_OPCODE_LG2:
       FETCH( &r[0], 0, CHAN_X );
       r[0].q = micro_lg2(r[0].q);
       FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
@@ -1234,8 +1200,7 @@ exec_instruction(
       }
       break;
 
-   case TGSI_OPCODE_POWER:
-      /* TGSI_OPCODE_POW */
+   case TGSI_OPCODE_POW:
       FETCH(&r[0], 0, CHAN_X);
       FETCH(&r[1], 1, CHAN_X);
 
@@ -1246,7 +1211,7 @@ exec_instruction(
       }
       break;
 
-   case TGSI_OPCODE_CROSSPRODUCT:
+   case TGSI_OPCODE_XPD:
       /* TGSI_OPCODE_XPD */
       FETCH(&r[0], 0, CHAN_Y);
       FETCH(&r[1], 1, CHAN_Z);
@@ -1288,10 +1253,6 @@ exec_instruction(
       }
       break;
 
-    case TGSI_OPCODE_MULTIPLYMATRIX:
-       assert (0);
-       break;
-
     case TGSI_OPCODE_ABS:
        FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
           FETCH(&r[0], 0, chan_index);
@@ -1303,7 +1264,7 @@ exec_instruction(
        break;
 
    case TGSI_OPCODE_RCC:
-      assert (0);
+      ASSERT (0);
       break;
 
    case TGSI_OPCODE_DPH:
@@ -1362,28 +1323,27 @@ exec_instruction(
       break;
 
    case TGSI_OPCODE_KIL:
-      /* for enabled ExecMask bits, set the killed bit */
-      mach->Temps[TEMP_KILMASK_I].xyzw[TEMP_KILMASK_C].u[0] |= mach->ExecMask;
+      exec_kil (mach, inst);
       break;
 
    case TGSI_OPCODE_PK2H:
-      assert (0);
+      ASSERT (0);
       break;
 
    case TGSI_OPCODE_PK2US:
-      assert (0);
+      ASSERT (0);
       break;
 
    case TGSI_OPCODE_PK4B:
-      assert (0);
+      ASSERT (0);
       break;
 
    case TGSI_OPCODE_PK4UB:
-      assert (0);
+      ASSERT (0);
       break;
 
    case TGSI_OPCODE_RFL:
-      assert (0);
+      ASSERT (0);
       break;
 
    case TGSI_OPCODE_SEQ:
@@ -1398,7 +1358,7 @@ exec_instruction(
       break;
 
    case TGSI_OPCODE_SFL:
-      assert (0);
+      ASSERT (0);
       break;
 
    case TGSI_OPCODE_SGT:
@@ -1443,21 +1403,21 @@ exec_instruction(
       break;
 
    case TGSI_OPCODE_STR:
-      assert (0);
+      ASSERT (0);
       break;
 
    case TGSI_OPCODE_TEX:
       /* simple texture lookup */
       /* src[0] = texcoord */
       /* src[1] = sampler unit */
-      exec_tex(mach, inst, FALSE);
+      exec_tex(mach, inst, FALSE, FALSE);
       break;
 
    case TGSI_OPCODE_TXB:
       /* Texture lookup with lod bias */
       /* src[0] = texcoord (src[0].w = load bias) */
       /* src[1] = sampler unit */
-      exec_tex(mach, inst, TRUE);
+      exec_tex(mach, inst, TRUE, FALSE);
       break;
 
    case TGSI_OPCODE_TXD:
@@ -1466,46 +1426,53 @@ exec_instruction(
       /* src[1] = d[strq]/dx */
       /* src[2] = d[strq]/dy */
       /* src[3] = sampler unit */
-      assert (0);
+      ASSERT (0);
       break;
 
    case TGSI_OPCODE_TXL:
       /* Texture lookup with explit LOD */
       /* src[0] = texcoord (src[0].w = load bias) */
       /* src[1] = sampler unit */
-      exec_tex(mach, inst, TRUE);
+      exec_tex(mach, inst, TRUE, FALSE);
+      break;
+
+   case TGSI_OPCODE_TXP:
+      /* Texture lookup with projection */
+      /* src[0] = texcoord (src[0].w = projection) */
+      /* src[1] = sampler unit */
+      exec_tex(mach, inst, TRUE, TRUE);
       break;
 
    case TGSI_OPCODE_UP2H:
-      assert (0);
+      ASSERT (0);
       break;
 
    case TGSI_OPCODE_UP2US:
-      assert (0);
+      ASSERT (0);
       break;
 
    case TGSI_OPCODE_UP4B:
-      assert (0);
+      ASSERT (0);
       break;
 
    case TGSI_OPCODE_UP4UB:
-      assert (0);
+      ASSERT (0);
       break;
 
    case TGSI_OPCODE_X2D:
-      assert (0);
+      ASSERT (0);
       break;
 
    case TGSI_OPCODE_ARA:
-      assert (0);
+      ASSERT (0);
       break;
 
    case TGSI_OPCODE_ARR:
-      assert (0);
+      ASSERT (0);
       break;
 
    case TGSI_OPCODE_BRA:
-      assert (0);
+      ASSERT (0);
       break;
 
    case TGSI_OPCODE_CAL:
@@ -1514,14 +1481,14 @@ exec_instruction(
          /* do the call */
 
          /* push the Cond, Loop, Cont stacks */
-         assert(mach->CondStackTop < TGSI_EXEC_MAX_COND_NESTING);
+         ASSERT(mach->CondStackTop < TGSI_EXEC_MAX_COND_NESTING);
          mach->CondStack[mach->CondStackTop++] = mach->CondMask;
-         assert(mach->LoopStackTop < TGSI_EXEC_MAX_LOOP_NESTING);
+         ASSERT(mach->LoopStackTop < TGSI_EXEC_MAX_LOOP_NESTING);
          mach->LoopStack[mach->LoopStackTop++] = mach->LoopMask;
-         assert(mach->ContStackTop < TGSI_EXEC_MAX_LOOP_NESTING);
+         ASSERT(mach->ContStackTop < TGSI_EXEC_MAX_LOOP_NESTING);
          mach->ContStack[mach->ContStackTop++] = mach->ContMask;
 
-         assert(mach->FuncStackTop < TGSI_EXEC_MAX_CALL_NESTING);
+         ASSERT(mach->FuncStackTop < TGSI_EXEC_MAX_CALL_NESTING);
          mach->FuncStack[mach->FuncStackTop++] = mach->FuncMask;
 
          /* note that PC was already incremented above */
@@ -1545,13 +1512,13 @@ exec_instruction(
          *pc = mach->CallStack[--mach->CallStackTop];
 
          /* pop the Cond, Loop, Cont stacks */
-         assert(mach->CondStackTop > 0);
+         ASSERT(mach->CondStackTop > 0);
          mach->CondMask = mach->CondStack[--mach->CondStackTop];
-         assert(mach->LoopStackTop > 0);
+         ASSERT(mach->LoopStackTop > 0);
          mach->LoopMask = mach->LoopStack[--mach->LoopStackTop];
-         assert(mach->ContStackTop > 0);
+         ASSERT(mach->ContStackTop > 0);
          mach->ContMask = mach->ContStack[--mach->ContStackTop];
-         assert(mach->FuncStackTop > 0);
+         ASSERT(mach->FuncStackTop > 0);
          mach->FuncMask = mach->FuncStack[--mach->FuncStackTop];
 
          UPDATE_EXEC_MASK(mach);
@@ -1559,7 +1526,7 @@ exec_instruction(
       break;
 
    case TGSI_OPCODE_SSG:
-      assert (0);
+      ASSERT (0);
       break;
 
    case TGSI_OPCODE_CMP:
@@ -1599,11 +1566,11 @@ exec_instruction(
       break;
 
    case TGSI_OPCODE_NRM:
-      assert (0);
+      ASSERT (0);
       break;
 
    case TGSI_OPCODE_DIV:
-      assert( 0 );
+      ASSERT( 0 );
       break;
 
    case TGSI_OPCODE_DP2:
@@ -1622,7 +1589,7 @@ exec_instruction(
 
    case TGSI_OPCODE_IF:
       /* push CondMask */
-      assert(mach->CondStackTop < TGSI_EXEC_MAX_COND_NESTING);
+      ASSERT(mach->CondStackTop < TGSI_EXEC_MAX_COND_NESTING);
       mach->CondStack[mach->CondStackTop++] = mach->CondMask;
       FETCH( &r[0], 0, CHAN_X );
       /* update CondMask */
@@ -1646,7 +1613,7 @@ exec_instruction(
       /* invert CondMask wrt previous mask */
       {
          uint prevMask;
-         assert(mach->CondStackTop > 0);
+         ASSERT(mach->CondStackTop > 0);
          prevMask = mach->CondStack[mach->CondStackTop - 1];
          mach->CondMask = ~mach->CondMask & prevMask;
          UPDATE_EXEC_MASK(mach);
@@ -1656,7 +1623,7 @@ exec_instruction(
 
    case TGSI_OPCODE_ENDIF:
       /* pop CondMask */
-      assert(mach->CondStackTop > 0);
+      ASSERT(mach->CondStackTop > 0);
       mach->CondMask = mach->CondStack[--mach->CondStackTop];
       UPDATE_EXEC_MASK(mach);
       break;
@@ -1667,19 +1634,19 @@ exec_instruction(
       break;
 
    case TGSI_OPCODE_REP:
-      assert (0);
+      ASSERT (0);
       break;
 
    case TGSI_OPCODE_ENDREP:
-       assert (0);
+       ASSERT (0);
        break;
 
    case TGSI_OPCODE_PUSHA:
-      assert (0);
+      ASSERT (0);
       break;
 
    case TGSI_OPCODE_POPA:
-      assert (0);
+      ASSERT (0);
       break;
 
    case TGSI_OPCODE_CEIL:
@@ -1753,7 +1720,7 @@ exec_instruction(
       break;
 
    case TGSI_OPCODE_MOD:
-      assert (0);
+      ASSERT (0);
       break;
 
    case TGSI_OPCODE_XOR:
@@ -1766,15 +1733,15 @@ exec_instruction(
       break;
 
    case TGSI_OPCODE_SAD:
-      assert (0);
+      ASSERT (0);
       break;
 
    case TGSI_OPCODE_TXF:
-      assert (0);
+      ASSERT (0);
       break;
 
    case TGSI_OPCODE_TXQ:
-      assert (0);
+      ASSERT (0);
       break;
 
    case TGSI_OPCODE_EMIT:
@@ -1787,21 +1754,21 @@ exec_instruction(
       mach->Primitives[mach->Temps[TEMP_PRIMITIVE_I].xyzw[TEMP_PRIMITIVE_C].u[0]] = 0;
       break;
 
-   case TGSI_OPCODE_LOOP:
+   case TGSI_OPCODE_BGNFOR:
       /* fall-through (for now) */
-   case TGSI_OPCODE_BGNLOOP2:
+   case TGSI_OPCODE_BGNLOOP:
       /* push LoopMask and ContMasks */
-      assert(mach->LoopStackTop < TGSI_EXEC_MAX_LOOP_NESTING);
+      ASSERT(mach->LoopStackTop < TGSI_EXEC_MAX_LOOP_NESTING);
       mach->LoopStack[mach->LoopStackTop++] = mach->LoopMask;
-      assert(mach->ContStackTop < TGSI_EXEC_MAX_LOOP_NESTING);
+      ASSERT(mach->ContStackTop < TGSI_EXEC_MAX_LOOP_NESTING);
       mach->ContStack[mach->ContStackTop++] = mach->ContMask;
       break;
 
-   case TGSI_OPCODE_ENDLOOP:
+   case TGSI_OPCODE_ENDFOR:
       /* fall-through (for now at least) */
-   case TGSI_OPCODE_ENDLOOP2:
+   case TGSI_OPCODE_ENDLOOP:
       /* Restore ContMask, but don't pop */
-      assert(mach->ContStackTop > 0);
+      ASSERT(mach->ContStackTop > 0);
       mach->ContMask = mach->ContStack[mach->ContStackTop - 1];
       if (mach->LoopMask) {
          /* repeat loop: jump to instruction just past BGNLOOP */
@@ -1809,10 +1776,10 @@ exec_instruction(
       }
       else {
          /* exit loop: pop LoopMask */
-         assert(mach->LoopStackTop > 0);
+         ASSERT(mach->LoopStackTop > 0);
          mach->LoopMask = mach->LoopStack[--mach->LoopStackTop];
          /* pop ContMask */
-         assert(mach->ContStackTop > 0);
+         ASSERT(mach->ContStackTop > 0);
          mach->ContMask = mach->ContStack[--mach->ContStackTop];
       }
       UPDATE_EXEC_MASK(mach);
@@ -1841,26 +1808,26 @@ exec_instruction(
       break;
 
    case TGSI_OPCODE_NOISE1:
-      assert( 0 );
+      ASSERT( 0 );
       break;
 
    case TGSI_OPCODE_NOISE2:
-      assert( 0 );
+      ASSERT( 0 );
       break;
 
    case TGSI_OPCODE_NOISE3:
-      assert( 0 );
+      ASSERT( 0 );
       break;
 
    case TGSI_OPCODE_NOISE4:
-      assert( 0 );
+      ASSERT( 0 );
       break;
 
    case TGSI_OPCODE_NOP:
       break;
 
    default:
-      assert( 0 );
+      ASSERT( 0 );
    }
 }
 
@@ -1881,11 +1848,11 @@ spu_exec_machine_run( struct spu_exec_machine *mach )
    mach->FuncMask = 0xf;
    mach->ExecMask = 0xf;
 
-   mach->CondStackTop = 0; /* temporarily subvert this assertion */
-   assert(mach->CondStackTop == 0);
-   assert(mach->LoopStackTop == 0);
-   assert(mach->ContStackTop == 0);
-   assert(mach->CallStackTop == 0);
+   mach->CondStackTop = 0; /* temporarily subvert this ASSERTion */
+   ASSERT(mach->CondStackTop == 0);
+   ASSERT(mach->LoopStackTop == 0);
+   ASSERT(mach->ContStackTop == 0);
+   ASSERT(mach->CallStackTop == 0);
 
    mach->Temps[TEMP_KILMASK_I].xyzw[TEMP_KILMASK_C].u[0] = 0;
    mach->Temps[TEMP_OUTPUT_I].xyzw[TEMP_OUTPUT_C].u[0] = 0;