[TargetLibraryInfo] add mappings from LLVM sin/cos intrinsics to SVML calls

These weren't included in D19544 - probably just an oversight.
D40044 made it more likely that we'll have LLVM math intrinsics rather 
than libcalls, so this bug was more easily exposed.
As the tests/code show, we already have the complete mappings for pow/exp/log.

I don't have any experience with SVML, so I don't know if anything else is 
missing. It's also not clear to me that we should be doing this transform in 
IR rather than DAG/isel, but that's a separate issue.

Differential Revision: https://reviews.llvm.org/D47610

llvm-svn: 334211
diff --git a/llvm/lib/Analysis/TargetLibraryInfo.cpp b/llvm/lib/Analysis/TargetLibraryInfo.cpp
index b223662..7fff22e 100644
--- a/llvm/lib/Analysis/TargetLibraryInfo.cpp
+++ b/llvm/lib/Analysis/TargetLibraryInfo.cpp
@@ -1461,6 +1461,14 @@
         {"sinf", "__svml_sinf8", 8},
         {"sinf", "__svml_sinf16", 16},
 
+        {"llvm.sin.f64", "__svml_sin2", 2},
+        {"llvm.sin.f64", "__svml_sin4", 4},
+        {"llvm.sin.f64", "__svml_sin8", 8},
+
+        {"llvm.sin.f32", "__svml_sinf4", 4},
+        {"llvm.sin.f32", "__svml_sinf8", 8},
+        {"llvm.sin.f32", "__svml_sinf16", 16},
+
         {"cos", "__svml_cos2", 2},
         {"cos", "__svml_cos4", 4},
         {"cos", "__svml_cos8", 8},
@@ -1469,6 +1477,14 @@
         {"cosf", "__svml_cosf8", 8},
         {"cosf", "__svml_cosf16", 16},
 
+        {"llvm.cos.f64", "__svml_cos2", 2},
+        {"llvm.cos.f64", "__svml_cos4", 4},
+        {"llvm.cos.f64", "__svml_cos8", 8},
+
+        {"llvm.cos.f32", "__svml_cosf4", 4},
+        {"llvm.cos.f32", "__svml_cosf8", 8},
+        {"llvm.cos.f32", "__svml_cosf16", 16},
+
         {"pow", "__svml_pow2", 2},
         {"pow", "__svml_pow4", 4},
         {"pow", "__svml_pow8", 8},