aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
Diffstat
-rw-r--r--lib/Target/CellSPU/SPU.h29
-rw-r--r--lib/Target/CellSPU/SPUISelDAGToDAG.cpp15
-rw-r--r--lib/Target/CellSPU/SPUInstrInfo.td2
-rw-r--r--lib/Target/CellSPU/SPUOperands.td10
-rw-r--r--test/CodeGen/CellSPU/and_ops.ll270
5 files changed, 322 insertions, 4 deletions
diff --git a/lib/Target/CellSPU/SPU.h b/lib/Target/CellSPU/SPU.h
index aee87fb..4555fda 100644
--- a/lib/Target/CellSPU/SPU.h
+++ b/lib/Target/CellSPU/SPU.h
@@ -25,7 +25,7 @@ namespace llvm {
FunctionPass *createSPUISelDag(SPUTargetMachine &TM);
FunctionPass *createSPUAsmPrinterPass(std::ostream &o, SPUTargetMachine &tm);
- /* Utility functions/predicates/etc used all over the place: */
+ /*--== Utility functions/predicates/etc used all over the place: --==*/
//! Predicate test for a signed 10-bit value
/*!
\param Value The input value to be tested
@@ -54,6 +54,33 @@ namespace llvm {
inline bool isS10Constant(uint64_t Value) {
return (Value <= ((1 << 9) - 1));
}
+
+ //! Predicate test for an unsigned 10-bit value
+ /*!
+ \param Value The input value to be tested
+
+ This predicate tests for an unsigned 10-bit value, returning the 10-bit value
+ as a short if true.
+ */
+ inline bool isU10Constant(short Value) {
+ return (Value == (Value & 0x3ff));
+ }
+
+ inline bool isU10Constant(int Value) {
+ return (Value == (Value & 0x3ff));
+ }
+
+ inline bool isU10Constant(uint32_t Value) {
+ return (Value == (Value & 0x3ff));
+ }
+
+ inline bool isU10Constant(int64_t Value) {
+ return (Value == (Value & 0x3ff));
+ }
+
+ inline bool isU10Constant(uint64_t Value) {
+ return (Value == (Value & 0x3ff));
+ }
}
// Defines symbolic names for the SPU instructions.
diff --git a/lib/Target/CellSPU/SPUISelDAGToDAG.cpp b/lib/Target/CellSPU/SPUISelDAGToDAG.cpp
index 935064d..ab02a81 100644
--- a/lib/Target/CellSPU/SPUISelDAGToDAG.cpp
+++ b/lib/Target/CellSPU/SPUISelDAGToDAG.cpp
@@ -78,6 +78,21 @@ namespace {
&& isI16IntS10Immediate(cast<ConstantSDNode>(N)));
}
+ //! ConstantSDNode predicate for i16 unsigned 10-bit immediate values
+ bool
+ isI16IntU10Immediate(ConstantSDNode *CN)
+ {
+ return isU10Constant((short) CN->getValue());
+ }
+
+ //! SDNode predicate for i16 sign-extended, 10-bit immediate values
+ bool
+ isI16IntU10Immediate(SDNode *N)
+ {
+ return (N->getOpcode() == ISD::Constant
+ && isI16IntU10Immediate(cast<ConstantSDNode>(N)));
+ }
+
//! ConstantSDNode predicate for signed 16-bit values
/*!
\arg CN The constant SelectionDAG node holding the value
diff --git a/lib/Target/CellSPU/SPUInstrInfo.td b/lib/Target/CellSPU/SPUInstrInfo.td
index faa6a7c..2ec14d2 100644
--- a/lib/Target/CellSPU/SPUInstrInfo.td
+++ b/lib/Target/CellSPU/SPUInstrInfo.td
@@ -1127,7 +1127,7 @@ def ANDHIv8i16:
def ANDHIr16:
RI10Form<0b10101000, (outs R16C:$rT), (ins R16C:$rA, s10imm:$val),
"andhi\t$rT, $rA, $val", IntegerOp,
- [(set R16C:$rT, (and R16C:$rA, i16ImmSExt10:$val))]>;
+ [(set R16C:$rT, (and R16C:$rA, i16ImmU10:$val))]>;
def ANDIv4i32:
RI10Form<0b00101000, (outs VECREG:$rT), (ins VECREG:$rA, s10imm:$val),
diff --git a/lib/Target/CellSPU/SPUOperands.td b/lib/Target/CellSPU/SPUOperands.td
index 05270dd..70103b8 100644
--- a/lib/Target/CellSPU/SPUOperands.td
+++ b/lib/Target/CellSPU/SPUOperands.td
@@ -99,12 +99,18 @@ def i32ImmSExt10 : PatLeaf<(imm), [{
return isI32IntS10Immediate(N);
}]>;
-// i16ImmSExt10 predicate - True if the i32 immediate fits in a 10-bit sign
+// i16ImmSExt10 predicate - True if the i16 immediate fits in a 10-bit sign
// extended field. Used by RI10Form instructions like 'ldq'.
def i16ImmSExt10 : PatLeaf<(imm), [{
return isI16IntS10Immediate(N);
}]>;
+// i16ImmU10 predicate - True if the i16 immediate fits into a 10-bit unsigned
+// value. Used by RI10Form instructions.
+def i16ImmU10 : PatLeaf<(imm), [{
+ return isI16IntU10Immediate(N);
+}]>;
+
def immSExt16 : PatLeaf<(imm), [{
// immSExt16 predicate - True if the immediate fits in a 16-bit sign extended
// field.
@@ -206,7 +212,7 @@ def fpimm18 : PatLeaf<(fpimm), [{
}], FPimm_u18>;
//===----------------------------------------------------------------------===//
-// 64-bit operands:
+// 64-bit operands (TODO):
//===----------------------------------------------------------------------===//
//===----------------------------------------------------------------------===//
diff --git a/test/CodeGen/CellSPU/and_ops.ll b/test/CodeGen/CellSPU/and_ops.ll
new file mode 100644
index 0000000..5c88d7e
--- /dev/null
+++ b/test/CodeGen/CellSPU/and_ops.ll
@@ -0,0 +1,270 @@
+; RUN: llvm-as -o - %s | llc -march=cellspu > %t1.s
+; RUN: grep and %t1.s | count 227
+; RUN: grep andc %t1.s | count 85
+; RUN: grep andi %t1.s | count 36
+; RUN: grep andhi %t1.s | count 31
+; RUN: grep andbi %t1.s | count 1
+
+; AND instruction generation:
+define <4 x i32> @and_v4i32_1(<4 x i32> %arg1, <4 x i32> %arg2) {
+ %A = and <4 x i32> %arg1, %arg2
+ ret <4 x i32> %A
+}
+
+define <4 x i32> @and_v4i32_2(<4 x i32> %arg1, <4 x i32> %arg2) {
+ %A = and <4 x i32> %arg2, %arg1
+ ret <4 x i32> %A
+}
+
+define <8 x i16> @and_v8i16_1(<8 x i16> %arg1, <8 x i16> %arg2) {
+ %A = and <8 x i16> %arg1, %arg2
+ ret <8 x i16> %A
+}
+
+define <8 x i16> @and_v8i16_2(<8 x i16> %arg1, <8 x i16> %arg2) {
+ %A = and <8 x i16> %arg2, %arg1
+ ret <8 x i16> %A
+}
+
+define <16 x i8> @and_v16i8_1(<16 x i8> %arg1, <16 x i8> %arg2) {
+ %A = and <16 x i8> %arg2, %arg1
+ ret <16 x i8> %A
+}
+
+define <16 x i8> @and_v16i8_2(<16 x i8> %arg1, <16 x i8> %arg2) {
+ %A = and <16 x i8> %arg1, %arg2
+ ret <16 x i8> %A
+}
+
+define i32 @and_i32_1(i32 %arg1, i32 %arg2) {
+ %A = and i32 %arg2, %arg1
+ ret i32 %A
+}
+
+define i32 @and_i32_2(i32 %arg1, i32 %arg2) {
+ %A = and i32 %arg1, %arg2
+ ret i32 %A
+}
+
+define i16 @and_i16_1(i16 %arg1, i16 %arg2) {
+ %A = and i16 %arg2, %arg1
+ ret i16 %A
+}
+
+define i16 @and_i16_2(i16 %arg1, i16 %arg2) {
+ %A = and i16 %arg1, %arg2
+ ret i16 %A
+}
+
+define i8 @and_i8_1(i8 %arg1, i8 %arg2) {
+ %A = and i8 %arg2, %arg1
+ ret i8 %A
+}
+
+define i8 @and_i8_2(i8 %arg1, i8 %arg2) {
+ %A = and i8 %arg1, %arg2
+ ret i8 %A
+}
+
+; ANDC instruction generation:
+define <4 x i32> @andc_v4i32_1(<4 x i32> %arg1, <4 x i32> %arg2) {
+ %A = xor <4 x i32> %arg2, < i32 -1, i32 -1, i32 -1, i32 -1 >
+ %B = and <4 x i32> %arg1, %A
+ ret <4 x i32> %B
+}
+
+define <4 x i32> @andc_v4i32_2(<4 x i32> %arg1, <4 x i32> %arg2) {
+ %A = xor <4 x i32> %arg1, < i32 -1, i32 -1, i32 -1, i32 -1 >
+ %B = and <4 x i32> %arg2, %A
+ ret <4 x i32> %B
+}
+
+define <4 x i32> @andc_v4i32_3(<4 x i32> %arg1, <4 x i32> %arg2) {
+ %A = xor <4 x i32> %arg1, < i32 -1, i32 -1, i32 -1, i32 -1 >
+ %B = and <4 x i32> %A, %arg2
+ ret <4 x i32> %B
+}
+
+define <8 x i16> @andc_v8i16_1(<8 x i16> %arg1, <8 x i16> %arg2) {
+ %A = xor <8 x i16> %arg2, < i16 -1, i16 -1, i16 -1, i16 -1,
+ i16 -1, i16 -1, i16 -1, i16 -1 >
+ %B = and <8 x i16> %arg1, %A
+ ret <8 x i16> %B
+}
+
+define <8 x i16> @andc_v8i16_2(<8 x i16> %arg1, <8 x i16> %arg2) {
+ %A = xor <8 x i16> %arg1, < i16 -1, i16 -1, i16 -1, i16 -1,
+ i16 -1, i16 -1, i16 -1, i16 -1 >
+ %B = and <8 x i16> %arg2, %A
+ ret <8 x i16> %B
+}
+
+define <16 x i8> @andc_v16i8_1(<16 x i8> %arg1, <16 x i8> %arg2) {
+ %A = xor <16 x i8> %arg1, < i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1,
+ i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1,
+ i8 -1, i8 -1, i8 -1, i8 -1 >
+ %B = and <16 x i8> %arg2, %A
+ ret <16 x i8> %B
+}
+
+define <16 x i8> @andc_v16i8_2(<16 x i8> %arg1, <16 x i8> %arg2) {
+ %A = xor <16 x i8> %arg2, < i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1,
+ i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1,
+ i8 -1, i8 -1, i8 -1, i8 -1 >
+ %B = and <16 x i8> %arg1, %A
+ ret <16 x i8> %B
+}
+
+define <16 x i8> @andc_v16i8_3(<16 x i8> %arg1, <16 x i8> %arg2) {
+ %A = xor <16 x i8> %arg2, < i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1,
+ i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1,
+ i8 -1, i8 -1, i8 -1, i8 -1 >
+ %B = and <16 x i8> %A, %arg1
+ ret <16 x i8> %B
+}
+
+define i32 @andc_i32_1(i32 %arg1, i32 %arg2) {
+ %A = xor i32 %arg2, -1
+ %B = and i32 %A, %arg1
+ ret i32 %B
+}
+
+define i32 @andc_i32_2(i32 %arg1, i32 %arg2) {
+ %A = xor i32 %arg1, -1
+ %B = and i32 %A, %arg2
+ ret i32 %B
+}
+
+define i32 @andc_i32_3(i32 %arg1, i32 %arg2) {
+ %A = xor i32 %arg2, -1
+ %B = and i32 %arg1, %A
+ ret i32 %B
+}
+
+define i16 @andc_i16_1(i16 %arg1, i16 %arg2) {
+ %A = xor i16 %arg2, -1
+ %B = and i16 %A, %arg1
+ ret i16 %B
+}
+
+define i16 @andc_i16_2(i16 %arg1, i16 %arg2) {
+ %A = xor i16 %arg1, -1
+ %B = and i16 %A, %arg2
+ ret i16 %B
+}
+
+define i16 @andc_i16_3(i16 %arg1, i16 %arg2) {
+ %A = xor i16 %arg2, -1
+ %B = and i16 %arg1, %A
+ ret i16 %B
+}
+
+define i8 @andc_i8_1(i8 %arg1, i8 %arg2) {
+ %A = xor i8 %arg2, -1
+ %B = and i8 %A, %arg1
+ ret i8 %B
+}
+
+define i8 @andc_i8_2(i8 %arg1, i8 %arg2) {
+ %A = xor i8 %arg1, -1
+ %B = and i8 %A, %arg2
+ ret i8 %B
+}
+
+define i8 @andc_i8_3(i8 %arg1, i8 %arg2) {
+ %A = xor i8 %arg2, -1
+ %B = and i8 %arg1, %A
+ ret i8 %B
+}
+
+; ANDI instruction generation (i32 data type):
+define <4 x i32> @andi_v4i32_1(<4 x i32> %in) {
+ %tmp2 = and <4 x i32> %in, < i32 511, i32 511, i32 511, i32 511 >
+ ret <4 x i32> %tmp2
+}
+
+define <4 x i32> @andi_v4i32_2(<4 x i32> %in) {
+ %tmp2 = and <4 x i32> %in, < i32 510, i32 510, i32 510, i32 510 >
+ ret <4 x i32> %tmp2
+}
+
+define <4 x i32> @andi_v4i32_3(<4 x i32> %in) {
+ %tmp2 = and <4 x i32> %in, < i32 -1, i32 -1, i32 -1, i32 -1 >
+ ret <4 x i32> %tmp2
+}
+
+define <4 x i32> @andi_v4i32_4(<4 x i32> %in) {
+ %tmp2 = and <4 x i32> %in, < i32 -512, i32 -512, i32 -512, i32 -512 >
+ ret <4 x i32> %tmp2
+}
+
+define i32 @andi_u32(i32 zeroext %in) zeroext {
+ %tmp37 = and i32 %in, 37
+ ret i32 %tmp37
+}
+
+define i32 @andi_i32(i32 signext %in) signext {
+ %tmp38 = and i32 %in, 37
+ ret i32 %tmp38
+}
+
+define i32 @andi_i32_1(i32 %in) {
+ %tmp37 = and i32 %in, 37
+ ret i32 %tmp37
+}
+
+; ANDHI instruction generation (i16 data type):
+define <8 x i16> @andhi_v8i16_1(<8 x i16> %in) {
+ %tmp2 = and <8 x i16> %in, < i16 511, i16 511, i16 511, i16 511,
+ i16 511, i16 511, i16 511, i16 511 >
+ ret <8 x i16> %tmp2
+}
+
+define <8 x i16> @andhi_v8i16_2(<8 x i16> %in) {
+ %tmp2 = and <8 x i16> %in, < i16 510, i16 510, i16 510, i16 510,
+ i16 510, i16 510, i16 510, i16 510 >
+ ret <8 x i16> %tmp2
+}
+
+define <8 x i16> @andhi_v8i16_3(<8 x i16> %in) {
+ %tmp2 = and <8 x i16> %in, < i16 -1, i16 -1, i16 -1, i16 -1, i16 -1,
+ i16 -1, i16 -1, i16 -1 >
+ ret <8 x i16> %tmp2
+}
+
+define <8 x i16> @andhi_v8i16_4(<8 x i16> %in) {
+ %tmp2 = and <8 x i16> %in, < i16 -512, i16 -512, i16 -512, i16 -512,
+ i16 -512, i16 -512, i16 -512, i16 -512 >
+ ret <8 x i16> %tmp2
+}
+
+define i16 @andhi_u16(i16 zeroext %in) zeroext {
+ %tmp37 = and i16 %in, 37 ; <i16> [#uses=1]
+ ret i16 %tmp37
+}
+
+define i16 @andhi_i16(i16 signext %in) signext {
+ %tmp38 = and i16 %in, 37 ; <i16> [#uses=1]
+ ret i16 %tmp38
+}
+
+; i8 data type (s/b ANDBI if 8-bit registers were supported):
+define <16 x i8> @and_v16i8(<16 x i8> %in) {
+ ; ANDBI generated for vector types
+ %tmp2 = and <16 x i8> %in, < i8 42, i8 42, i8 42, i8 42, i8 42, i8 42,
+ i8 42, i8 42, i8 42, i8 42, i8 42, i8 42,
+ i8 42, i8 42, i8 42, i8 42 >
+ ret <16 x i8> %tmp2
+}
+
+define i8 @and_u8(i8 zeroext %in) zeroext {
+ ; ANDI generated:
+ %tmp37 = and i8 %in, 37 ; <i8> [#uses=1]
+ ret i8 %tmp37
+}
+
+define i8 @and_i8(i8 signext %in) signext {
+ ; ANDHI generated
+ %tmp38 = and i8 %in, 37 ; <i8> [#uses=1]
+ ret i8 %tmp38
+}
generated by cgit v1.1 at 2025-01-13 18:52:32 +0000