[Beignet] [PATCH 1/2] Fix vector allocation and 64-bit reading

[Homer Hsing]

Vector allocation gave each register DWORD * SIMD_WIDTH.
When the register is QWORD, allocation is wrong.
This patch fixes vector allocation.

This patch also fixes 64-bit reading.
When reading, temporary registers is next to destination register.

Signed-off-by: Homer Hsing <homer.xing at intel.com>
---
 backend/src/backend/gen_encoder.cpp        |  4 ++--
 backend/src/backend/gen_reg_allocation.cpp | 18 ++++++++++++++++--
 2 files changed, 18 insertions(+), 4 deletions(-)

diff --git a/backend/src/backend/gen_encoder.cpp b/backend/src/backend/gen_encoder.cpp
index f84c6dd..0a2b673 100644
--- a/backend/src/backend/gen_encoder.cpp
+++ b/backend/src/backend/gen_encoder.cpp
@@ -367,10 +367,10 @@ namespace gbe
 
   void GenEncoder::READ_FLOAT64(GenRegister dst, GenRegister src, uint32_t bti, uint32_t elemNum) {
     int w = curr.execWidth;
-    dst = GenRegister::h2(dst);
     dst.type = GEN_TYPE_UD;
     src.type = GEN_TYPE_UD;
-    GenRegister r = GenRegister::retype(GenRegister::suboffset(src, w*2), GEN_TYPE_UD);
+    GenRegister r = GenRegister::retype(GenRegister::suboffset(dst, w*2), GEN_TYPE_UD);
+    dst = GenRegister::h2(dst);
     GenRegister imm4 = GenRegister::immud(4);
     GenInstruction *insn;
     insn = next(GEN_OPCODE_SEND);
diff --git a/backend/src/backend/gen_reg_allocation.cpp b/backend/src/backend/gen_reg_allocation.cpp
index e7c96ac..56d9261 100644
--- a/backend/src/backend/gen_reg_allocation.cpp
+++ b/backend/src/backend/gen_reg_allocation.cpp
@@ -475,16 +475,30 @@ namespace gbe
         const SelectionVector *vector = it->second.first;
         const uint32_t simdWidth = ctx.getSimdWidth();
         const uint32_t alignment = simdWidth * sizeof(uint32_t);
-        const uint32_t size = vector->regNum * alignment;
+        int num = 0;
+        for (uint32_t regID = 0; regID < vector->regNum; ++regID) {
+          const ir::Register reg = vector->reg[regID].reg();
+          const ir::RegisterFamily f = selection.getRegisterFamily(reg);
+          if (f == ir::FAMILY_QWORD)
+            num += 2;
+          else
+            num ++;
+        }
+        const uint32_t size = num * alignment;
         uint32_t grfOffset;
         while ((grfOffset = ctx.allocate(size, alignment)) == 0) {
           const bool success = this->expireGRF(interval);
           if (success == false) return false;
         }
-        for (uint32_t regID = 0; regID < vector->regNum; ++regID, grfOffset += alignment) {
+        for (uint32_t regID = 0; regID < vector->regNum; ++regID) {
           const ir::Register reg = vector->reg[regID].reg();
           GBE_ASSERT(RA.contains(reg) == false);
           RA.insert(std::make_pair(reg, grfOffset));
+          const ir::RegisterFamily f = selection.getRegisterFamily(reg);
+          if (f == ir::FAMILY_QWORD)
+            grfOffset += 2 * alignment;
+          else
+            grfOffset += alignment;
         }
       }
       // Case 2: This is a regular scalar register, allocate it alone
-- 
1.8.1.2