Correct the double bug in HSW.

Should set the nomask in mov_df_imm and need handle exec_width=4 case in setHeader.

Signed-off-by: Yang Rong <rong.r.yang@intel.com>
Reviewed-by: Junyan He <junyan.he@inbox.com>

diff --git a/backend/src/backend/gen75_encoder.cpp b/backend/src/backend/gen75_encoder.cpp
index 7fc2d1d..9e758d6 100644 (file)
--- a/backend/src/backend/gen75_encoder.cpp
+++ b/backend/src/backend/gen75_encoder.cpp
@@ -183,6 +183,7 @@ namespace gbe
     GenRegister r = GenRegister::retype(tmp, GEN_TYPE_UD);
     push();
     curr.predicate = GEN_PREDICATE_NONE;
+    curr.noMask = 1;
     curr.execWidth = 1;
     MOV(r, GenRegister::immud(u.u[0]));
     MOV(GenRegister::suboffset(r, 1), GenRegister::immud(u.u[1]));

diff --git a/backend/src/backend/gen75_encoder.hpp b/backend/src/backend/gen75_encoder.hpp
index a9042b3..60dea7b 100644 (file)
--- a/backend/src/backend/gen75_encoder.hpp
+++ b/backend/src/backend/gen75_encoder.hpp
@@ -31,11 +31,15 @@ namespace gbe
   class Gen75Encoder : public GenEncoder
   {
   public:
+    /*! exec width of the double data type */    
+    const int GEN75_DOUBLE_EXEC_WIDTH = 4;
     virtual ~Gen75Encoder(void) { }
 
     Gen75Encoder(uint32_t simdWidth, uint32_t gen, uint32_t deviceID)
          : GenEncoder(simdWidth, gen, deviceID, 8) { }
 
+    /*! Get double/long exec width */
+    virtual int getDoubleExecWidth(void) { return GEN75_DOUBLE_EXEC_WIDTH; }
     virtual void MOV_DF(GenRegister dest, GenRegister src0, GenRegister tmp = GenRegister::null());
     virtual void LOAD_DF_IMM(GenRegister dest, GenRegister tmp, double value);
     virtual void ATOMIC(GenRegister dst, uint32_t function, GenRegister src, uint32_t bti, uint32_t srcNum);

diff --git a/backend/src/backend/gen_encoder.cpp b/backend/src/backend/gen_encoder.cpp
index aa30b05..7078dcb 100644 (file)
--- a/backend/src/backend/gen_encoder.cpp
+++ b/backend/src/backend/gen_encoder.cpp
@@ -240,6 +240,8 @@ namespace gbe
       insn->header.execution_size = GEN_WIDTH_8;
     else if (this->curr.execWidth == 16)
       insn->header.execution_size = GEN_WIDTH_16;
+    else if (this->curr.execWidth == 4)
+      insn->header.execution_size = GEN_WIDTH_4;
     else if (this->curr.execWidth == 1)
       insn->header.execution_size = GEN_WIDTH_1;
     else
@@ -624,6 +626,7 @@ namespace gbe
                             GenRegister src0, GenRegister src1 = GenRegister::null()) {
        int w = p->curr.execWidth;
        p->push();
+       p->curr.execWidth = p->getDoubleExecWidth();
        p->curr.nibControl = 0;
        GenNativeInstruction *insn = p->next(opcode);
        p->setHeader(insn);

diff --git a/backend/src/backend/gen_encoder.hpp b/backend/src/backend/gen_encoder.hpp
index 9eaa1c2..e0bb4cc 100644 (file)
--- a/backend/src/backend/gen_encoder.hpp
+++ b/backend/src/backend/gen_encoder.hpp
@@ -70,6 +70,8 @@ namespace gbe
     virtual ~GenEncoder(void) { }
     /*! Size of the stack (should be large enough) */
     enum { MAX_STATE_NUM = 16 };
+    /*! gen7 exec width of the double data type */
+    const int GEN7_DOUBLE_EXEC_WIDTH = 8;
     /*! Push the current instruction state */
     void push(void);
     /*! Pop the latest pushed state */
@@ -135,6 +137,8 @@ namespace gbe
 #undef ALU2
 #undef ALU2_MOD
 #undef ALU3
+    /*! Get double/long exec width */
+    virtual int getDoubleExecWidth(void) { return GEN7_DOUBLE_EXEC_WIDTH; }
     virtual void MOV_DF(GenRegister dest, GenRegister src0, GenRegister tmp = GenRegister::null());
     virtual void LOAD_DF_IMM(GenRegister dest, GenRegister tmp, double value);
     void LOAD_INT64_IMM(GenRegister dest, int64_t value);