[AArch64, ARM] Add support for Samsung Exynos M4

Create a separate feature set for Exynos M4 and add test cases.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@334115 91177308-0d34-0410-b5e6-96231b3b80d8

include/llvm/Support/AArch64TargetParser.def

@@ -82,6 +82,8 @@ AARCH64_CPU_NAME("exynos-m2", ARMV8A, FK_CRYPTO_NEON_FP_ARMV8, false,
                 (AArch64::AEK_CRC))
 AARCH64_CPU_NAME("exynos-m3", ARMV8A, FK_CRYPTO_NEON_FP_ARMV8, false,
                 (AArch64::AEK_CRC))
+AARCH64_CPU_NAME("exynos-m4", ARMV8A, FK_CRYPTO_NEON_FP_ARMV8, false,
+                (AArch64::AEK_CRC))
 AARCH64_CPU_NAME("falkor", ARMV8A, FK_CRYPTO_NEON_FP_ARMV8, false,
                 (AArch64::AEK_CRC | AArch64::AEK_RDM))
 AARCH64_CPU_NAME("saphira", ARMV8_3A, FK_CRYPTO_NEON_FP_ARMV8, false,

include/llvm/Support/ARMTargetParser.def

@@ -253,6 +253,7 @@ ARM_CPU_NAME("cyclone", ARMV8A, FK_CRYPTO_NEON_FP_ARMV8, false, ARM::AEK_CRC)
 ARM_CPU_NAME("exynos-m1", ARMV8A, FK_CRYPTO_NEON_FP_ARMV8, false, ARM::AEK_CRC)
 ARM_CPU_NAME("exynos-m2", ARMV8A, FK_CRYPTO_NEON_FP_ARMV8, false, ARM::AEK_CRC)
 ARM_CPU_NAME("exynos-m3", ARMV8A, FK_CRYPTO_NEON_FP_ARMV8, false, ARM::AEK_CRC)
+ARM_CPU_NAME("exynos-m4", ARMV8A, FK_CRYPTO_NEON_FP_ARMV8, false, ARM::AEK_CRC)
 ARM_CPU_NAME("kryo", ARMV8A, FK_CRYPTO_NEON_FP_ARMV8, false, ARM::AEK_CRC)
 // Non-standard Arch names.
 ARM_CPU_NAME("iwmmxt", IWMMXT, FK_NONE, true, ARM::AEK_NONE)

lib/Target/AArch64/AArch64.td

@@ -487,6 +487,7 @@ def : ProcessorModel<"cyclone", CycloneModel, [ProcCyclone]>;
 def : ProcessorModel<"exynos-m1", ExynosM1Model, [ProcExynosM1]>;
 def : ProcessorModel<"exynos-m2", ExynosM1Model, [ProcExynosM2]>;
 def : ProcessorModel<"exynos-m3", ExynosM3Model, [ProcExynosM3]>;
+def : ProcessorModel<"exynos-m4", ExynosM3Model, [ProcExynosM3]>;
 def : ProcessorModel<"falkor", FalkorModel, [ProcFalkor]>;
 def : ProcessorModel<"saphira", FalkorModel, [ProcSaphira]>;
 def : ProcessorModel<"kryo", KryoModel, [ProcKryo]>;

lib/Target/ARM/ARM.td

@@ -992,6 +992,12 @@ def : ProcNoItin<"exynos-m3",                           [ARMv8a, ProcExynosM1,
                                                          FeatureCrypto,
                                                          FeatureCRC]>;
 
+def : ProcNoItin<"exynos-m4",                           [ARMv8a, ProcExynosM1,
+                                                         FeatureHWDivThumb,
+                                                         FeatureHWDivARM,
+                                                         FeatureCrypto,
+                                                         FeatureCRC]>;
+
 def : ProcNoItin<"kryo",                                [ARMv8a, ProcKryo,
                                                          FeatureHWDivThumb,
                                                          FeatureHWDivARM,

test/CodeGen/AArch64/cpus.ll

@@ -12,6 +12,7 @@
 ; RUN: llc < %s -mtriple=arm64-unknown-unknown -mcpu=exynos-m1 2>&1 | FileCheck %s
 ; RUN: llc < %s -mtriple=arm64-unknown-unknown -mcpu=exynos-m2 2>&1 | FileCheck %s
 ; RUN: llc < %s -mtriple=arm64-unknown-unknown -mcpu=exynos-m3 2>&1 | FileCheck %s
+; RUN: llc < %s -mtriple=arm64-unknown-unknown -mcpu=exynos-m4 2>&1 | FileCheck %s
 ; RUN: llc < %s -mtriple=arm64-unknown-unknown -mcpu=falkor 2>&1 | FileCheck %s
 ; RUN: llc < %s -mtriple=arm64-unknown-unknown -mcpu=saphira 2>&1 | FileCheck %s
 ; RUN: llc < %s -mtriple=arm64-unknown-unknown -mcpu=kryo 2>&1 | FileCheck %s

test/CodeGen/AArch64/misched-fusion-addr.ll

@@ -1,5 +1,6 @@
 ; RUN: llc %s -o - -mtriple=aarch64-unknown -mattr=fuse-address | FileCheck %s
 ; RUN: llc %s -o - -mtriple=aarch64-unknown -mcpu=exynos-m3     | FileCheck %s
+; RUN: llc %s -o - -mtriple=aarch64-unknown -mcpu=exynos-m4     | FileCheck %s
 
 target triple = "aarch64-unknown"
 

test/CodeGen/AArch64/misched-fusion-aes.ll

@@ -7,6 +7,7 @@
 ; RUN: llc %s -o - -mtriple=aarch64-unknown -mcpu=exynos-m1  | FileCheck %s
 ; RUN: llc %s -o - -mtriple=aarch64-unknown -mcpu=exynos-m2  | FileCheck %s
 ; RUN: llc %s -o - -mtriple=aarch64-unknown -mcpu=exynos-m3  | FileCheck %s
+; RUN: llc %s -o - -mtriple=aarch64-unknown -mcpu=exynos-m4  | FileCheck %s
 
 declare <16 x i8> @llvm.aarch64.crypto.aese(<16 x i8> %d, <16 x i8> %k)
 declare <16 x i8> @llvm.aarch64.crypto.aesmc(<16 x i8> %d)

test/CodeGen/AArch64/misched-fusion-csel.ll

@@ -1,5 +1,6 @@
 ; RUN: llc %s -o - -mtriple=aarch64-unknown -mattr=fuse-csel | FileCheck %s
 ; RUN: llc %s -o - -mtriple=aarch64-unknown -mcpu=exynos-m3  | FileCheck %s
+; RUN: llc %s -o - -mtriple=aarch64-unknown -mcpu=exynos-m4  | FileCheck %s
 
 target triple = "aarch64-unknown"
 

test/CodeGen/AArch64/misched-fusion-lit.ll

@@ -2,6 +2,7 @@
 ; RUN: llc %s -o - -mtriple=aarch64-unknown -mattr=+fuse-literals | FileCheck %s --check-prefix=CHECK --check-prefix=CHECKFUSE
 ; RUN: llc %s -o - -mtriple=aarch64-unknown -mcpu=cortex-a57      | FileCheck %s --check-prefix=CHECK --check-prefix=CHECKFUSE
 ; RUN: llc %s -o - -mtriple=aarch64-unknown -mcpu=exynos-m3       | FileCheck %s --check-prefix=CHECK --check-prefix=CHECKFUSE
+; RUN: llc %s -o - -mtriple=aarch64-unknown -mcpu=exynos-m4       | FileCheck %s --check-prefix=CHECK --check-prefix=CHECKFUSE
 
 @g = common local_unnamed_addr global i8* null, align 8
 

test/CodeGen/AArch64/remat.ll

@@ -8,6 +8,7 @@
 ; RUN: llc -mtriple=aarch64-linux-gnuabi -mcpu=exynos-m1 -o - %s | FileCheck %s
 ; RUN: llc -mtriple=aarch64-linux-gnuabi -mcpu=exynos-m2 -o - %s | FileCheck %s
 ; RUN: llc -mtriple=aarch64-linux-gnuabi -mcpu=exynos-m3 -o - %s | FileCheck %s
+; RUN: llc -mtriple=aarch64-linux-gnuabi -mcpu=exynos-m4 -o - %s | FileCheck %s
 ; RUN: llc -mtriple=aarch64-linux-gnuabi -mcpu=falkor -o - %s | FileCheck %s
 ; RUN: llc -mtriple=aarch64-linux-gnuabi -mcpu=saphira -o - %s | FileCheck %s
 ; RUN: llc -mtriple=aarch64-linux-gnuabi -mcpu=kryo -o - %s | FileCheck %s

test/CodeGen/ARM/build-attributes.ll

@@ -143,6 +143,9 @@
 ; RUN: llc < %s -mtriple=armv8-linux-gnueabi -mcpu=exynos-m3 | FileCheck %s --check-prefix=EXYNOS-M3
 ; RUN: llc < %s -mtriple=armv8-linux-gnueabi -mcpu=exynos-m3  -enable-unsafe-fp-math -disable-fp-elim -enable-no-infs-fp-math -enable-no-nans-fp-math -fp-contract=fast | FileCheck %s --check-prefix=EXYNOS-M1-FAST
 ; RUN: llc < %s -mtriple=armv8-linux-gnueabi -mcpu=exynos-m3 -enable-sign-dependent-rounding-fp-math | FileCheck %s --check-prefix=DYN-ROUNDING
+; RUN: llc < %s -mtriple=armv8-linux-gnueabi -mcpu=exynos-m4 | FileCheck %s --check-prefix=EXYNOS-M4
+; RUN: llc < %s -mtriple=armv8-linux-gnueabi -mcpu=exynos-m4  -enable-unsafe-fp-math -disable-fp-elim -enable-no-infs-fp-math -enable-no-nans-fp-math -fp-contract=fast | FileCheck %s --check-prefix=EXYNOS-M1-FAST
+; RUN: llc < %s -mtriple=armv8-linux-gnueabi -mcpu=exynos-m4 -enable-sign-dependent-rounding-fp-math | FileCheck %s --check-prefix=DYN-ROUNDING
 ; RUN: llc < %s -mtriple=armv8.1a-linux-gnueabi  -enable-unsafe-fp-math -disable-fp-elim -enable-no-infs-fp-math -enable-no-nans-fp-math -fp-contract=fast | FileCheck %s --check-prefix=GENERIC-ARMV8_1-A-FAST
 ; RUN: llc < %s -mtriple=armv8.1a-linux-gnueabi -enable-sign-dependent-rounding-fp-math | FileCheck %s --check-prefix=DYN-ROUNDING
 ; RUN: llc < %s -mtriple=armv7-none-linux-gnueabi -mcpu=cortex-a7 | FileCheck %s  --check-prefix=CORTEX-A7-CHECK
@@ -182,6 +185,8 @@
 ; RUN: llc < %s -mtriple=armv8-none-linux-gnueabi -mcpu=exynos-m2 -mattr=+strict-align | FileCheck %s --check-prefix=STRICT-ALIGN
 ; RUN: llc < %s -mtriple=armv8-none-linux-gnueabi -mcpu=exynos-m3 | FileCheck %s --check-prefix=NO-STRICT-ALIGN
 ; RUN: llc < %s -mtriple=armv8-none-linux-gnueabi -mcpu=exynos-m3 -mattr=+strict-align | FileCheck %s --check-prefix=STRICT-ALIGN
+; RUN: llc < %s -mtriple=armv8-none-linux-gnueabi -mcpu=exynos-m4 | FileCheck %s --check-prefix=NO-STRICT-ALIGN
+; RUN: llc < %s -mtriple=armv8-none-linux-gnueabi -mcpu=exynos-m4 -mattr=+strict-align | FileCheck %s --check-prefix=STRICT-ALIGN
 
 ; ARMv7a
 ; RUN: llc < %s -mtriple=armv7-none-linux-gnueabi -mcpu=cortex-a7 | FileCheck %s --check-prefix=NO-STRICT-ALIGN
@@ -1613,6 +1618,29 @@
 ; EXYNOS-M3-NOT:  .eabi_attribute 28
 ; EXYNOS-M3:  .eabi_attribute 38, 1
 
+; EXYNOS-M4:  .cpu exynos-m4
+; EXYNOS-M4:  .eabi_attribute 6, 14
+; EXYNOS-M4:  .eabi_attribute 7, 65
+; EXYNOS-M4:  .eabi_attribute 8, 1
+; EXYNOS-M4:  .eabi_attribute 9, 2
+; EXYNOS-M4:  .fpu crypto-neon-fp-armv8
+; EXYNOS-M4:  .eabi_attribute 12, 3
+; EXYNOS-M4-NOT:  .eabi_attribute 27
+; EXYNOS-M4:  .eabi_attribute 36, 1
+; EXYNOS-M4:  .eabi_attribute 42, 1
+; EXYNOS-M4-NOT:  .eabi_attribute 44
+; EXYNOS-M4:  .eabi_attribute 68, 3
+; EXYNOS-M4-NOT:   .eabi_attribute 19
+;; We default to IEEE 754 compliance
+; EXYNOS-M4:  .eabi_attribute 20, 1
+; EXYNOS-M4:  .eabi_attribute 21, 1
+; EXYNOS-M4-NOT:  .eabi_attribute 22
+; EXYNOS-M4:  .eabi_attribute 23, 3
+; EXYNOS-M4:  .eabi_attribute 24, 1
+; EXYNOS-M4:  .eabi_attribute 25, 1
+; EXYNOS-M4-NOT:  .eabi_attribute 28
+; EXYNOS-M4:  .eabi_attribute 38, 1
+
 ; GENERIC-FPU-VFPV3-FP16: .fpu vfpv3-fp16
 ; GENERIC-FPU-VFPV3-D16-FP16: .fpu vfpv3-d16-fp16
 ; GENERIC-FPU-VFPV3XD: .fpu vfpv3xd

unittests/Support/TargetParserTest.cpp

@@ -253,18 +253,23 @@ TEST(TargetParserTest, testARMCPU) {
                          "8-A"));
   EXPECT_TRUE(testARMCPU("exynos-m1", "armv8-a", "crypto-neon-fp-armv8",
                          ARM::AEK_CRC | ARM::AEK_SEC | ARM::AEK_MP |
-                             ARM::AEK_VIRT | ARM::AEK_HWDIVARM |
-                             ARM::AEK_HWDIVTHUMB | ARM::AEK_DSP,
+                         ARM::AEK_VIRT | ARM::AEK_HWDIVARM |
+                         ARM::AEK_HWDIVTHUMB | ARM::AEK_DSP,
                          "8-A"));
   EXPECT_TRUE(testARMCPU("exynos-m2", "armv8-a", "crypto-neon-fp-armv8",
                          ARM::AEK_CRC | ARM::AEK_SEC | ARM::AEK_MP |
-                             ARM::AEK_VIRT | ARM::AEK_HWDIVARM |
-                             ARM::AEK_HWDIVTHUMB | ARM::AEK_DSP,
+                         ARM::AEK_VIRT | ARM::AEK_HWDIVARM |
+                         ARM::AEK_HWDIVTHUMB | ARM::AEK_DSP,
                          "8-A"));
   EXPECT_TRUE(testARMCPU("exynos-m3", "armv8-a", "crypto-neon-fp-armv8",
                          ARM::AEK_CRC | ARM::AEK_SEC | ARM::AEK_MP |
-                             ARM::AEK_VIRT | ARM::AEK_HWDIVARM |
-                             ARM::AEK_HWDIVTHUMB | ARM::AEK_DSP,
+                         ARM::AEK_VIRT | ARM::AEK_HWDIVARM |
+                         ARM::AEK_HWDIVTHUMB | ARM::AEK_DSP,
+                         "8-A"));
+  EXPECT_TRUE(testARMCPU("exynos-m4", "armv8-a", "crypto-neon-fp-armv8",
+                         ARM::AEK_CRC | ARM::AEK_SEC | ARM::AEK_MP |
+                         ARM::AEK_VIRT | ARM::AEK_HWDIVARM |
+                         ARM::AEK_HWDIVTHUMB | ARM::AEK_DSP,
                          "8-A"));
   EXPECT_TRUE(testARMCPU("cortex-m23", "armv8-m.base", "none",
                          ARM::AEK_HWDIVTHUMB, "8-M.Baseline"));
@@ -279,7 +284,7 @@ TEST(TargetParserTest, testARMCPU) {
                          "7-S"));
 }
 
-static constexpr unsigned NumARMCPUArchs = 82;
+static constexpr unsigned NumARMCPUArchs = 83;
 
 TEST(TargetParserTest, testARMCPUArchList) {
   SmallVector<StringRef, NumARMCPUArchs> List;
@@ -727,6 +732,10 @@ TEST(TargetParserTest, testAArch64CPU) {
       "exynos-m3", "armv8-a", "crypto-neon-fp-armv8",
       AArch64::AEK_CRC | AArch64::AEK_CRYPTO | AArch64::AEK_FP |
       AArch64::AEK_SIMD, "8-A"));
+  EXPECT_TRUE(testAArch64CPU(
+      "exynos-m4", "armv8-a", "crypto-neon-fp-armv8",
+      AArch64::AEK_CRC | AArch64::AEK_CRYPTO | AArch64::AEK_FP |
+      AArch64::AEK_SIMD, "8-A"));
   EXPECT_TRUE(testAArch64CPU(
       "falkor", "armv8-a", "crypto-neon-fp-armv8",
       AArch64::AEK_CRC | AArch64::AEK_CRYPTO | AArch64::AEK_FP |
@@ -761,7 +770,7 @@ TEST(TargetParserTest, testAArch64CPU) {
       "8-A"));
 }
 
-static constexpr unsigned NumAArch64CPUArchs = 19;
+static constexpr unsigned NumAArch64CPUArchs = 20;
 
 TEST(TargetParserTest, testAArch64CPUArchList) {
   SmallVector<StringRef, NumAArch64CPUArchs> List;
@@ -824,6 +833,8 @@ TEST(TargetParserTest, testAArch64Extension) {
                                     AArch64::ArchKind::INVALID, "ras"));
   EXPECT_FALSE(testAArch64Extension("exynos-m3",
                                     AArch64::ArchKind::INVALID, "ras"));
+  EXPECT_FALSE(testAArch64Extension("exynos-m4",
+                                    AArch64::ArchKind::INVALID, "ras"));
   EXPECT_TRUE(testAArch64Extension("falkor",
                                    AArch64::ArchKind::INVALID, "rdm"));
   EXPECT_FALSE(testAArch64Extension("kryo",