nv50/ir: add support for SULDP -> SULDB conversion

[mesa.git] / src / gallium / drivers / nouveau / codegen / nv50_ir_emit_gk110.cpp

diff --git a/src/gallium/drivers/nouveau/codegen/nv50_ir_emit_gk110.cpp b/src/gallium/drivers/nouveau/codegen/nv50_ir_emit_gk110.cpp
index 0c7cd1d81379c950ded7a24ab1a1791a519aff9c..6fbd04df69b9a56ac0ae7df144b0561ba99c1404 100644 (file)
--- a/src/gallium/drivers/nouveau/codegen/nv50_ir_emit_gk110.cpp
+++ b/src/gallium/drivers/nouveau/codegen/nv50_ir_emit_gk110.cpp
@@ -128,6 +128,8 @@ private:
 
    void emitFlow(const Instruction *);
 
+   void emitVOTE(const Instruction *);
+
    inline void defId(const ValueDef&, const int pos);
    inline void srcId(const ValueRef&, const int pos);
    inline void srcId(const ValueRef *, const int pos);
@@ -433,6 +435,10 @@ CodeEmitterGK110::emitForm_21(const Instruction *i, uint32_t opc2,
          srcId(i->src(s), s ? ((s == 2) ? 42 : s1) : 10);
          break;
       default:
+         if (i->op == OP_SELP) {
+            assert(s == 2 && i->src(s).getFile() == FILE_PREDICATE);
+            srcId(i->src(s), 42);
+         }
          // ignore here, can be predicate or flags, but must not be address
          break;
       }
@@ -575,7 +581,7 @@ CodeEmitterGK110::emitIMUL(const Instruction *i)
    assert(!i->src(0).mod.neg() && !i->src(1).mod.neg());
    assert(!i->src(0).mod.abs() && !i->src(1).mod.abs());
 
-   if (isLIMM(i->src(1), TYPE_S32)) {
+   if (i->src(1).getFile() == FILE_IMMEDIATE) {
       emitForm_L(i, 0x280, 2, Modifier(0));
 
       if (i->subOp == NV50_IR_SUBOP_MUL_HIGH)
@@ -746,6 +752,32 @@ CodeEmitterGK110::emitNOT(const Instruction *i)
 void
 CodeEmitterGK110::emitLogicOp(const Instruction *i, uint8_t subOp)
 {
+   if (i->def(0).getFile() == FILE_PREDICATE) {
+      code[0] = 0x00000002 | (subOp << 27);
+      code[1] = 0x84800000;
+
+      emitPredicate(i);
+
+      defId(i->def(0), 5);
+      srcId(i->src(0), 14);
+      if (i->src(0).mod == Modifier(NV50_IR_MOD_NOT)) code[0] |= 1 << 17;
+      srcId(i->src(1), 32);
+      if (i->src(1).mod == Modifier(NV50_IR_MOD_NOT)) code[1] |= 1 << 3;
+
+      if (i->defExists(1)) {
+         defId(i->def(1), 2);
+      } else {
+         code[0] |= 7 << 2;
+      }
+      // (a OP b) OP c
+      if (i->predSrc != 2 && i->srcExists(2)) {
+         code[1] |= subOp << 16;
+         srcId(i->src(2), 42);
+         if (i->src(2).mod == Modifier(NV50_IR_MOD_NOT)) code[1] |= 1 << 13;
+      } else {
+         code[1] |= 7 << 10;
+      }
+   } else
    if (isLIMM(i->src(1), TYPE_S32)) {
       emitForm_L(i, 0x200, 0, i->src(1).mod);
       code[1] |= subOp << 24;
@@ -1045,7 +1077,7 @@ void CodeEmitterGK110::emitSELP(const Instruction *i)
 {
    emitForm_21(i, 0x250, 0x050);
 
-   if ((i->cc == CC_NOT_P) ^ (bool)(i->src(2).mod & Modifier(NV50_IR_MOD_NOT)))
+   if (i->src(2).mod & Modifier(NV50_IR_MOD_NOT))
       code[1] |= 1 << 13;
 }
 
@@ -1239,7 +1271,7 @@ CodeEmitterGK110::emitQUADOP(const Instruction *i, uint8_t qOp, uint8_t laneMask
 
    defId(i->def(0), 2);
    srcId(i->src(0), 10);
-   srcId(i->srcExists(1) ? i->src(1) : i->src(0), 23);
+   srcId((i->srcExists(1) && i->predSrc != 1) ? i->src(1) : i->src(0), 23);
 
    if (i->op == OP_QUADOP && progType != Program::TYPE_FRAGMENT)
       code[1] |= 1 << 9; // dall
@@ -1267,15 +1299,41 @@ CodeEmitterGK110::emitBAR(const Instruction *i)
    case NV50_IR_SUBOP_BAR_RED_OR:   code[1] |= 0x90; break;
    case NV50_IR_SUBOP_BAR_RED_POPC: code[1] |= 0x10; break;
    default:
-      code[1] |= 0x20;
       assert(i->subOp == NV50_IR_SUBOP_BAR_SYNC);
       break;
    }
 
    emitPredicate(i);
 
-   srcId(i->src(0), 10);
-   srcId(i->src(1), 23);
+   // barrier id
+   if (i->src(0).getFile() == FILE_GPR) {
+      srcId(i->src(0), 10);
+   } else {
+      ImmediateValue *imm = i->getSrc(0)->asImm();
+      assert(imm);
+      code[0] |= imm->reg.data.u32 << 10;
+      code[1] |= 0x8000;
+   }
+
+   // thread count
+   if (i->src(1).getFile() == FILE_GPR) {
+      srcId(i->src(1), 23);
+   } else {
+      ImmediateValue *imm = i->getSrc(0)->asImm();
+      assert(imm);
+      assert(imm->reg.data.u32 <= 0xfff);
+      code[0] |= imm->reg.data.u32 << 23;
+      code[1] |= imm->reg.data.u32 >> 9;
+      code[1] |= 0x4000;
+   }
+
+   if (i->srcExists(2) && (i->predSrc != 2)) {
+      srcId(i->src(2), 32 + 10);
+      if (i->src(2).mod == Modifier(NV50_IR_MOD_NOT))
+         code[1] |= 1 << 13;
+   } else {
+      code[1] |= 7 << 10;
+   }
 }
 
 void CodeEmitterGK110::emitMEMBAR(const Instruction *i)
@@ -1366,6 +1424,24 @@ CodeEmitterGK110::emitFlow(const Instruction *i)
    }
 }
 
+void
+CodeEmitterGK110::emitVOTE(const Instruction *i)
+{
+   assert(i->src(0).getFile() == FILE_PREDICATE &&
+          i->def(1).getFile() == FILE_PREDICATE);
+
+   code[0] = 0x00000002;
+   code[1] = 0x86c00000 | (i->subOp << 19);
+
+   emitPredicate(i);
+
+   defId(i->def(0), 2);
+   defId(i->def(1), 48);
+   if (i->src(0).mod == Modifier(NV50_IR_MOD_NOT))
+      code[1] |= 1 << 13;
+   srcId(i->src(0), 42);
+}
+
 void
 CodeEmitterGK110::emitAFETCH(const Instruction *i)
 {
@@ -1593,16 +1669,20 @@ CodeEmitterGK110::emitSTORE(const Instruction *i)
    switch (i->src(0).getFile()) {
    case FILE_MEMORY_GLOBAL: code[1] = 0xe0000000; code[0] = 0x00000000; break;
    case FILE_MEMORY_LOCAL:  code[1] = 0x7a800000; code[0] = 0x00000002; break;
-   case FILE_MEMORY_SHARED: code[1] = 0x7ac00000; code[0] = 0x00000002; break;
+   case FILE_MEMORY_SHARED:
+      code[0] = 0x00000002;
+      if (i->subOp == NV50_IR_SUBOP_STORE_UNLOCKED)
+         code[1] = 0x78400000;
+      else
+         code[1] = 0x7ac00000;
+      break;
    default:
       assert(!"invalid memory file");
       break;
    }
 
-   if (i->src(0).getFile() != FILE_MEMORY_GLOBAL)
-      offset &= 0xffffff;
-
    if (code[0] & 0x2) {
+      offset &= 0xffffff;
       emitLoadStoreType(i->dType, 0x33);
       if (i->src(0).getFile() == FILE_MEMORY_LOCAL)
          emitCachingMode(i->cache, 0x2f);
@@ -1613,6 +1693,13 @@ CodeEmitterGK110::emitSTORE(const Instruction *i)
    code[0] |= offset << 23;
    code[1] |= offset >> 9;
 
+   // Unlocked store on shared memory can fail.
+   if (i->src(0).getFile() == FILE_MEMORY_SHARED &&
+       i->subOp == NV50_IR_SUBOP_STORE_UNLOCKED) {
+      assert(i->defExists(0));
+      defId(i->def(0), 32 + 16);
+   }
+
    emitPredicate(i);
 
    srcId(i->src(1), 2);
@@ -1631,7 +1718,13 @@ CodeEmitterGK110::emitLOAD(const Instruction *i)
    switch (i->src(0).getFile()) {
    case FILE_MEMORY_GLOBAL: code[1] = 0xc0000000; code[0] = 0x00000000; break;
    case FILE_MEMORY_LOCAL:  code[1] = 0x7a000000; code[0] = 0x00000002; break;
-   case FILE_MEMORY_SHARED: code[1] = 0x7a400000; code[0] = 0x00000002; break;
+   case FILE_MEMORY_SHARED:
+      code[0] = 0x00000002;
+      if (i->subOp == NV50_IR_SUBOP_LOAD_LOCKED)
+         code[1] = 0x77400000;
+      else
+         code[1] = 0x7a400000;
+      break;
    case FILE_MEMORY_CONST:
       if (!i->src(0).isIndirect(0) && typeSizeof(i->dType) == 4) {
          emitMOV(i);
@@ -1659,6 +1752,13 @@ CodeEmitterGK110::emitLOAD(const Instruction *i)
    code[0] |= offset << 23;
    code[1] |= offset >> 9;
 
+   // Locked store on shared memory can fail.
+   if (i->src(0).getFile() == FILE_MEMORY_SHARED &&
+       i->subOp == NV50_IR_SUBOP_LOAD_LOCKED) {
+      assert(i->defExists(1));
+      defId(i->def(1), 32 + 16);
+   }
+
    emitPredicate(i);
 
    defId(i->def(0), 2);
@@ -1734,6 +1834,9 @@ uses64bitAddress(const Instruction *ldst)
 void
 CodeEmitterGK110::emitATOM(const Instruction *i)
 {
+   const bool hasDst = i->defExists(0);
+   const bool exch = i->subOp == NV50_IR_SUBOP_ATOM_EXCH;
+
    code[0] = 0x00000002;
    if (i->subOp == NV50_IR_SUBOP_ATOM_CAS)
       code[1] = 0x77800000;
@@ -1762,15 +1865,21 @@ CodeEmitterGK110::emitATOM(const Instruction *i)
    /* TODO: cas: flip bits if $r255 is used */
    srcId(i->src(1), 23);
 
-   if (i->defExists(0))
+   if (hasDst) {
       defId(i->def(0), 2);
-   else
+   } else
+   if (!exch) {
       code[0] |= 255 << 2;
+   }
 
-   const int32_t offset = SDATA(i->src(0)).offset;
-   assert(offset < 0x80000 && offset >= -0x80000);
-   code[0] |= (offset & 1) << 31;
-   code[1] |= (offset & 0xffffe) >> 1;
+   if (hasDst || !exch) {
+      const int32_t offset = SDATA(i->src(0)).offset;
+      assert(offset < 0x80000 && offset >= -0x80000);
+      code[0] |= (offset & 1) << 31;
+      code[1] |= (offset & 0xffffe) >> 1;
+   } else {
+      srcAddr32(i->src(0), 31);
+   }
 
    if (i->getIndirect(0, 0)) {
       srcId(i->getIndirect(0, 0), 10);
@@ -2050,6 +2159,9 @@ CodeEmitterGK110::emitInstruction(Instruction *insn)
    case OP_CCTL:
       emitCCTL(insn);
       break;
+   case OP_VOTE:
+      emitVOTE(insn);
+      break;
    case OP_PHI:
    case OP_UNION:
    case OP_CONSTRAINT: