Adding front-end support to several intrinsics (bit scanning, conversion and state...

Adding LLVM front-end support to two intrinsics dealing with bit scan: _bit_scan_forward and _bit_scan_reverse.
 Their functionality is as described in Intel intrinsics guide:
 https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_bit_scan_forward&expand=371,370
 https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_bit_scan_reverse&expand=371,370

 Furthermore, adding clang front-end support to these conversion intrinsics: _mm256_cvtsd_f64, _mm256_cvtsi256_si32 and _mm256_cvtss_f32.

 Finally, adding tests to all of the above, as well as to the state reading intrinsics _rdpmc and _rdtsc.
  Their functionality is also specified in the Intel intrinsics guide.

Commit on behalf of Omer Paparo Bivas

git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@271387 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/include/clang/Basic/BuiltinsX86.def b/include/clang/Basic/BuiltinsX86.def
index a5580612ff95c95b723ad6aeb9aa45e9e3ed13ef..b4c44437d907cfb34b6a323383f02400f5fbf23f 100644 (file)
--- a/include/clang/Basic/BuiltinsX86.def
+++ b/include/clang/Basic/BuiltinsX86.def
@@ -35,6 +35,10 @@ BUILTIN(__builtin_ms_va_start, "vc*&.", "nt")
 BUILTIN(__builtin_ms_va_end, "vc*&", "n")
 BUILTIN(__builtin_ms_va_copy, "vc*&c*&", "n")
 
+// Bit scan
+TARGET_BUILTIN(__builtin_ia32_bit_scan_forward, "ii", "", "")
+TARGET_BUILTIN(__builtin_ia32_bit_scan_reverse, "ii", "", "")
+
 // Undefined Values
 //
 TARGET_BUILTIN(__builtin_ia32_undef128, "V2d", "nc", "")

diff --git a/lib/Headers/avxintrin.h b/lib/Headers/avxintrin.h
index 8bb1b55e1e4b5143dba652700c379dd8ec4b075c..6a7a4b903a6f87d4a4e201c88becfc82cdc82fc1 100644 (file)
--- a/lib/Headers/avxintrin.h
+++ b/lib/Headers/avxintrin.h
@@ -2123,6 +2123,25 @@ _mm256_cvttps_epi32(__m256 __a)
   return (__m256i)__builtin_ia32_cvttps2dq256((__v8sf) __a);
 }
 
+static __inline double __DEFAULT_FN_ATTRS
+_mm256_cvtsd_f64(__m256d __a)
+{
+ return __a[0];
+}
+
+static __inline int __DEFAULT_FN_ATTRS
+_mm256_cvtsi256_si32(__m256i __a)
+{
+ __v8si __b = (__v8si)__a;
+ return __b[0];
+}
+
+static __inline float __DEFAULT_FN_ATTRS
+_mm256_cvtss_f32(__m256 __a)
+{
+ return __a[0];
+}
+
 /* Vector replicate */
 static __inline __m256 __DEFAULT_FN_ATTRS
 _mm256_movehdup_ps(__m256 __a)

diff --git a/lib/Headers/ia32intrin.h b/lib/Headers/ia32intrin.h
index b2f82bb59e38a3c95c0bd868595679a6b89aade7..397f3fd13e010d3cd869af09a67a0c10b08fbe2c 100644 (file)
--- a/lib/Headers/ia32intrin.h
+++ b/lib/Headers/ia32intrin.h
@@ -74,4 +74,6 @@ __rdtscp(unsigned int *__A) {
 
 #define _rdtsc() __rdtsc()
 
+#define _rdpmc(A) __rdpmc(A)
+
 #endif /* __IA32INTRIN_H */

diff --git a/lib/Headers/immintrin.h b/lib/Headers/immintrin.h
index d6c6ae5e4c57ffe379f5850e30cc1732735941ba..7ed2035761db0f7c38e3c39e4d2af1a8a16ed1c0 100644 (file)
--- a/lib/Headers/immintrin.h
+++ b/lib/Headers/immintrin.h
@@ -169,6 +169,18 @@ _rdrand32_step(unsigned int *__p)
   return __builtin_ia32_rdrand32_step(__p);
 }
 
+/* __bit_scan_forward */
+static __inline__ int __attribute__((__always_inline__, __nodebug__))
+_bit_scan_forward(int __A) {
+ return __builtin_ia32_bit_scan_forward(__A);
+}
+
+/* __bit_scan_reverse */
+static __inline__ int __attribute__((__always_inline__, __nodebug__))
+_bit_scan_reverse(int __A) {
+ return __builtin_ia32_bit_scan_reverse(__A);
+}
+
 #ifdef __x86_64__
 static __inline__ int __attribute__((__always_inline__, __nodebug__, __target__("rdrnd")))
 _rdrand64_step(unsigned long long *__p)
@@ -227,6 +239,7 @@ _writegsbase_u64(unsigned long long __V)
 {
   return __builtin_ia32_wrgsbase64(__V);
 }
+
 #endif
 #endif /* __FSGSBASE__ */
 

diff --git a/test/CodeGen/avx-builtins.c b/test/CodeGen/avx-builtins.c
index e15e22a2ed6f3c043b638dabae8c101e909f2a4d..c1b8cf2c919aa11074e10cf74274abff319479f9 100644 (file)
--- a/test/CodeGen/avx-builtins.c
+++ b/test/CodeGen/avx-builtins.c
@@ -1385,3 +1385,24 @@ void test_mm256_zeroupper() {
   // CHECK: call void @llvm.x86.avx.vzeroupper()
   return _mm256_zeroupper();
 }
+
+double test_mm256_cvtsd_f64(__m256d __a)
+{
+ // CHECK-LABEL: @test_mm256_cvtsd_f64
+ // CHECK: extractelement <4 x double> %{{.*}}, i32 0
+ return _mm256_cvtsd_f64(__a);
+}
+
+int test_mm256_cvtsi256_si32(__m256i __a)
+{
+ // CHECK-LABEL: @test_mm256_cvtsi256_si32
+ // CHECK: extractelement <8 x i32> %{{.*}}, i32 0
+ return _mm256_cvtsi256_si32(__a);
+}
+
+float test_mm256_cvtss_f32(__m256 __a)
+{
+ // CHECK-LABEL: @test_mm256_cvtss_f32
+ // CHECK: extractelement <8 x float> %{{.*}}, i32 0
+ return _mm256_cvtss_f32(__a);
+}

diff --git a/test/CodeGen/bitscan-builtins.c b/test/CodeGen/bitscan-builtins.c
new file mode 100644 (file)
index 0000000..85a7542
--- /dev/null
+++ b/test/CodeGen/bitscan-builtins.c
@@ -0,0 +1,17 @@
+// RUN: %clang_cc1 -triple x86_64-unknown-unknown -emit-llvm -o - %s | FileCheck %s
+
+// Don't include mm_malloc.h, it's system specific.
+#define __MM_MALLOC_H
+#include <immintrin.h>
+
+int test_bit_scan_forward(int a) {
+  return _bit_scan_forward(a);
+// CHECK: @test_bit_scan_forward
+// CHECK: call i32 @llvm.x86.bit.scan.forward
+}
+
+int test_bit_scan_reverse(int a) {
+  return _bit_scan_reverse(a);
+// CHECK: @test_bit_scan_reverse
+// CHECK: call i32 @llvm.x86.bit.scan.reverse
+}

diff --git a/test/CodeGen/rd-builtins.c b/test/CodeGen/rd-builtins.c
new file mode 100644 (file)
index 0000000..5cad903
--- /dev/null
+++ b/test/CodeGen/rd-builtins.c
@@ -0,0 +1,18 @@
+// RUN: %clang_cc1 -triple x86_64-unknown-unknown -emit-llvm -o - %s | FileCheck %s
+
+// Don't include mm_malloc.h, it's system specific.
+#define __MM_MALLOC_H
+
+#include <x86intrin.h>
+
+unsigned long long test_rdpmc(int a) {
+  return _rdpmc(a);
+// CHECK: @test_rdpmc
+// CHECK: call i64 @llvm.x86.rdpmc
+}
+
+int test_rdtsc() {
+  return _rdtsc();
+// CHECK: @test_rdtsc
+// CHECK: call i64 @llvm.x86.rdtsc
+}