Add missing builtins to altivec.h for ABI compliance (vol. 2)
This patch corresponds to review:
http://reviews.llvm.org/D10875
The bulk of the second round of additions to altivec.h.
The following interfaces were added:
vector double vec_floor(vector double)
vector double vec_madd(vector double, vector double, vector double)
vector float vec_msub(vector float, vector float, vector float)
vector double vec_msub(vector double, vector double, vector double)
vector float vec_mul(vector float, vector float)
vector double vec_mul(vector double, vector double)
vector float vec_nmadd(vector float, vector float, vector float)
vector double vec_nmadd(vector double, vector double, vector double)
vector double vec_nmsub(vector double, vector double, vector double)
vector double vec_nor(vector double, vector double)
vector double vec_or(vector double, vector double)
vector float vec_rint(vector float)
vector double vec_rint(vector double)
vector float vec_nearbyint(vector float)
vector double vec_nearbyint(vector double)
vector float vec_sqrt(vector float)
vector double vec_sqrt(vector double)
vector double vec_rsqrte(vector double)
vector double vec_sel(vector double, vector double, vector unsigned long long)
vector double vec_sel(vector double, vector double, vector unsigned long long)
vector double vec_sub(vector double, vector double)
vector double vec_trunc(vector double)
vector double vec_xor(vector double, vector double)
vector double vec_xor(vector double, vector bool long long)
vector double vec_xor(vector bool long long, vector double)
New VSX paths for the following interfaces:
vector float vec_madd(vector float, vector float, vector float)
vector float vec_nmsub(vector float, vector float, vector float)
vector float vec_rsqrte(vector float)
vector float vec_trunc(vector float)
vector float vec_floor(vector float)
llvm-svn: 241399
diff --git a/clang/lib/Headers/altivec.h b/clang/lib/Headers/altivec.h
index 2c80e24..61ad320 100644
--- a/clang/lib/Headers/altivec.h
+++ b/clang/lib/Headers/altivec.h
@@ -1857,11 +1857,20 @@
/* vec_floor */
-static vector float __attribute__((__always_inline__))
-vec_floor(vector float __a) {
+static vector float __ATTRS_o_ai vec_floor(vector float __a) {
+#ifdef __VSX__
+ return __builtin_vsx_xvrspim(__a);
+#else
return __builtin_altivec_vrfim(__a);
+#endif
}
+#ifdef __VSX__
+static vector double __ATTRS_o_ai vec_floor(vector double __a) {
+ return __builtin_vsx_xvrdpim(__a);
+}
+#endif
+
/* vec_vrfim */
static vector float __attribute__((__always_inline__))
@@ -2532,11 +2541,22 @@
/* vec_madd */
-static vector float __attribute__((__always_inline__))
+static vector float __ATTRS_o_ai
vec_madd(vector float __a, vector float __b, vector float __c) {
+#ifdef __VSX__
+ return __builtin_vsx_xvmaddasp(__a, __b, __c);
+#else
return __builtin_altivec_vmaddfp(__a, __b, __c);
+#endif
}
+#ifdef __VSX__
+static vector double __ATTRS_o_ai
+vec_madd(vector double __a, vector double __b, vector double __c) {
+ return __builtin_vsx_xvmaddadp(__a, __b, __c);
+}
+#endif
+
/* vec_vmaddfp */
static vector float __attribute__((__always_inline__))
@@ -2559,6 +2579,20 @@
return __builtin_altivec_vmhaddshs(__a, __b, __c);
}
+/* vec_msub */
+
+#ifdef __VSX__
+static vector float __ATTRS_o_ai
+vec_msub(vector float __a, vector float __b, vector float __c) {
+ return __builtin_vsx_xvmsubasp(__a, __b, __c);
+}
+
+static vector double __ATTRS_o_ai
+vec_msub(vector double __a, vector double __b, vector double __c) {
+ return __builtin_vsx_xvmsubadp(__a, __b, __c);
+}
+#endif
+
/* vec_max */
static vector signed char __ATTRS_o_ai vec_max(vector signed char __a,
@@ -3677,6 +3711,18 @@
__builtin_altivec_mtvscr((vector int)__a);
}
+/* vec_mul */
+static vector float __ATTRS_o_ai vec_mul(vector float __a, vector float __b) {
+ __a * __b;
+}
+
+#ifdef __VSX__
+static vector double __ATTRS_o_ai
+vec_mul(vector double __a, vector double __b) {
+ __a * __b;
+}
+#endif
+
/* The vmulos* and vmules* instructions have a big endian bias, so
we must reverse the meaning of "even" and "odd" for little endian. */
@@ -3882,13 +3928,38 @@
#endif
}
+/* vec_nmadd */
+
+#ifdef __VSX__
+static vector float __ATTRS_o_ai
+vec_nmadd(vector float __a, vector float __b, vector float __c) {
+ return __builtin_vsx_xvnmaddasp(__a, __b, __c);
+}
+
+static vector double __ATTRS_o_ai
+vec_nmadd(vector double __a, vector double __b, vector double __c) {
+ return __builtin_vsx_xvnmaddadp(__a, __b, __c);
+}
+#endif
+
/* vec_nmsub */
-static vector float __attribute__((__always_inline__))
+static vector float __ATTRS_o_ai
vec_nmsub(vector float __a, vector float __b, vector float __c) {
+#ifdef __VSX__
+ return __builtin_vsx_xvnmsubasp(__a, __b, __c);
+#else
return __builtin_altivec_vnmsubfp(__a, __b, __c);
+#endif
}
+#ifdef __VSX__
+static vector double __ATTRS_o_ai
+vec_nmsub(vector double __a, vector double __b, vector double __c) {
+ return __builtin_vsx_xvnmsubadp(__a, __b, __c);
+}
+#endif
+
/* vec_vnmsubfp */
static vector float __attribute__((__always_inline__))
@@ -3949,6 +4020,15 @@
return (vector float)__res;
}
+#ifdef __VSX__
+static vector double __ATTRS_o_ai
+vec_nor(vector double __a, vector double __b) {
+ vector unsigned long long __res =
+ ~((vector unsigned long long)__a | (vector unsigned long long)__b);
+ return (vector double)__res;
+}
+#endif
+
/* vec_vnor */
static vector signed char __ATTRS_o_ai vec_vnor(vector signed char __a,
@@ -4141,6 +4221,12 @@
}
#ifdef __VSX__
+static vector double __ATTRS_o_ai vec_or(vector double __a, vector double __b) {
+ vector unsigned long long __res =
+ (vector unsigned long long)__a | (vector unsigned long long)__b;
+ return (vector double)__res;
+}
+
static vector signed long long __ATTRS_o_ai
vec_or(vector signed long long __a, vector signed long long __b) {
return __a | __b;
@@ -5237,6 +5323,30 @@
return __builtin_altivec_vrfin(__a);
}
+#ifdef __VSX__
+/* vec_rint */
+
+static vector float __ATTRS_o_ai
+vec_rint(vector float __a) {
+ return __builtin_vsx_xvrspic(__a);
+}
+
+static vector double __ATTRS_o_ai
+vec_rint(vector double __a) {
+ return __builtin_vsx_xvrdpic(__a);
+}
+
+/* vec_nearbyint */
+
+static vector float __ATTRS_o_ai vec_nearbyint(vector float __a) {
+ return __builtin_vsx_xvrspi(__a);
+}
+
+static vector double __ATTRS_o_ai vec_nearbyint(vector double __a) {
+ return __builtin_vsx_xvrdpi(__a);
+}
+#endif
+
/* vec_vrfin */
static vector float __attribute__((__always_inline__))
@@ -5244,13 +5354,35 @@
return __builtin_altivec_vrfin(__a);
}
+/* vec_sqrt */
+
+#ifdef __VSX__
+static vector float __ATTRS_o_ai vec_sqrt(vector float __a) {
+ return __builtin_vsx_xvsqrtsp(__a);
+}
+
+static vector double __ATTRS_o_ai vec_sqrt(vector double __a) {
+ return __builtin_vsx_xvsqrtdp(__a);
+}
+#endif
+
/* vec_rsqrte */
-static __vector float __attribute__((__always_inline__))
+static vector float __ATTRS_o_ai
vec_rsqrte(vector float __a) {
+#ifdef __VSX__
+ return __builtin_vsx_xvrsqrtesp(__a);
+#else
return __builtin_altivec_vrsqrtefp(__a);
+#endif
}
+#ifdef __VSX__
+static vector double __ATTRS_o_ai vec_rsqrte(vector double __a) {
+ return __builtin_vsx_xvrsqrtedp(__a);
+}
+#endif
+
/* vec_vrsqrtefp */
static __vector float __attribute__((__always_inline__))
@@ -5381,6 +5513,22 @@
return (vector float)__res;
}
+#ifdef __VSX__
+static vector double __ATTRS_o_ai vec_sel(vector double __a, vector double __b,
+ vector bool long long __c) {
+ vector long long __res = ((vector long long)__a & ~(vector long long)__c) |
+ ((vector long long)__b & (vector long long)__c);
+ return (vector double)__res;
+}
+
+static vector double __ATTRS_o_ai vec_sel(vector double __a, vector double __b,
+ vector unsigned long long __c) {
+ vector long long __res = ((vector long long)__a & ~(vector long long)__c) |
+ ((vector long long)__b & (vector long long)__c);
+ return (vector double)__res;
+}
+#endif
+
/* vec_vsel */
static vector signed char __ATTRS_o_ai vec_vsel(vector signed char __a,
@@ -7960,6 +8108,13 @@
return __a - __b;
}
+#ifdef __VSX__
+static vector double __ATTRS_o_ai
+vec_sub(vector double __a, vector double __b) {
+ return __a - __b;
+}
+#endif
+
/* vec_vsububm */
#define __builtin_altivec_vsububm vec_vsububm
@@ -8451,11 +8606,21 @@
/* vec_trunc */
-static vector float __attribute__((__always_inline__))
+static vector float __ATTRS_o_ai
vec_trunc(vector float __a) {
+#ifdef __VSX__
+ return __builtin_vsx_xvrspiz(__a);
+#else
return __builtin_altivec_vrfiz(__a);
+#endif
}
+#ifdef __VSX__
+static vector double __ATTRS_o_ai vec_trunc(vector double __a) {
+ return __builtin_vsx_xvrdpiz(__a);
+}
+#endif
+
/* vec_vrfiz */
static vector float __attribute__((__always_inline__))
@@ -8945,6 +9110,24 @@
vector bool long long __b) {
return __a ^ __b;
}
+
+static vector double __ATTRS_o_ai
+vec_xor(vector double __a, vector double __b) {
+ return (vector double)((vector unsigned long long)__a ^
+ (vector unsigned long long)__b);
+}
+
+static vector double __ATTRS_o_ai
+vec_xor(vector double __a, vector bool long long __b) {
+ return (vector double)((vector unsigned long long)__a ^
+ (vector unsigned long long) __b);
+}
+
+static vector double __ATTRS_o_ai
+vec_xor(vector bool long long __a, vector double __b) {
+ return (vector double)((vector unsigned long long)__a ^
+ (vector unsigned long long)__b);
+}
#endif
/* vec_vxor */