[mips][msa] Added f[cs]af, f[cs]or, f[cs]ueq, f[cs]ul[et], f[cs]une, fsun, ftrunc_[su], hadd_[su], hsub_[su], sr[al]r, sr[al]ri

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@189467 91177308-0d34-0410-b5e6-96231b3b80d8

4 files changed, 976 insertions, 0 deletions

diff --git a/test/CodeGen/Mips/msa/2rf_int_float.ll b/test/CodeGen/Mips/msa/2rf_int_float.ll
index b5eaad2..2e4244c 100644
--- a/test/CodeGen/Mips/msa/2rf_int_float.ll
+++ b/test/CodeGen/Mips/msa/2rf_int_float.ll
@@ -42,6 +42,82 @@ declare <2 x i64> @llvm.mips.fclass.d(<2 x double>) nounwind
 ; CHECK: st.d
 ; CHECK: .size llvm_mips_fclass_d_test
 ;
+@llvm_mips_ftrunc_s_w_ARG1 = global <4 x float> <float 0.000000e+00, float 1.000000e+00, float 2.000000e+00, float 3.000000e+00>, align 16
+@llvm_mips_ftrunc_s_w_RES  = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
+
+define void @llvm_mips_ftrunc_s_w_test() nounwind {
+entry:
+  %0 = load <4 x float>* @llvm_mips_ftrunc_s_w_ARG1
+  %1 = tail call <4 x i32> @llvm.mips.ftrunc.s.w(<4 x float> %0)
+  store <4 x i32> %1, <4 x i32>* @llvm_mips_ftrunc_s_w_RES
+  ret void
+}
+
+declare <4 x i32> @llvm.mips.ftrunc.s.w(<4 x float>) nounwind
+
+; CHECK: llvm_mips_ftrunc_s_w_test:
+; CHECK: ld.w
+; CHECK: ftrunc_s.w
+; CHECK: st.w
+; CHECK: .size llvm_mips_ftrunc_s_w_test
+;
+@llvm_mips_ftrunc_s_d_ARG1 = global <2 x double> <double 0.000000e+00, double 1.000000e+00>, align 16
+@llvm_mips_ftrunc_s_d_RES  = global <2 x i64> <i64 0, i64 0>, align 16
+
+define void @llvm_mips_ftrunc_s_d_test() nounwind {
+entry:
+  %0 = load <2 x double>* @llvm_mips_ftrunc_s_d_ARG1
+  %1 = tail call <2 x i64> @llvm.mips.ftrunc.s.d(<2 x double> %0)
+  store <2 x i64> %1, <2 x i64>* @llvm_mips_ftrunc_s_d_RES
+  ret void
+}
+
+declare <2 x i64> @llvm.mips.ftrunc.s.d(<2 x double>) nounwind
+
+; CHECK: llvm_mips_ftrunc_s_d_test:
+; CHECK: ld.d
+; CHECK: ftrunc_s.d
+; CHECK: st.d
+; CHECK: .size llvm_mips_ftrunc_s_d_test
+;
+@llvm_mips_ftrunc_u_w_ARG1 = global <4 x float> <float 0.000000e+00, float 1.000000e+00, float 2.000000e+00, float 3.000000e+00>, align 16
+@llvm_mips_ftrunc_u_w_RES  = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
+
+define void @llvm_mips_ftrunc_u_w_test() nounwind {
+entry:
+  %0 = load <4 x float>* @llvm_mips_ftrunc_u_w_ARG1
+  %1 = tail call <4 x i32> @llvm.mips.ftrunc.u.w(<4 x float> %0)
+  store <4 x i32> %1, <4 x i32>* @llvm_mips_ftrunc_u_w_RES
+  ret void
+}
+
+declare <4 x i32> @llvm.mips.ftrunc.u.w(<4 x float>) nounwind
+
+; CHECK: llvm_mips_ftrunc_u_w_test:
+; CHECK: ld.w
+; CHECK: ftrunc_u.w
+; CHECK: st.w
+; CHECK: .size llvm_mips_ftrunc_u_w_test
+;
+@llvm_mips_ftrunc_u_d_ARG1 = global <2 x double> <double 0.000000e+00, double 1.000000e+00>, align 16
+@llvm_mips_ftrunc_u_d_RES  = global <2 x i64> <i64 0, i64 0>, align 16
+
+define void @llvm_mips_ftrunc_u_d_test() nounwind {
+entry:
+  %0 = load <2 x double>* @llvm_mips_ftrunc_u_d_ARG1
+  %1 = tail call <2 x i64> @llvm.mips.ftrunc.u.d(<2 x double> %0)
+  store <2 x i64> %1, <2 x i64>* @llvm_mips_ftrunc_u_d_RES
+  ret void
+}
+
+declare <2 x i64> @llvm.mips.ftrunc.u.d(<2 x double>) nounwind
+
+; CHECK: llvm_mips_ftrunc_u_d_test:
+; CHECK: ld.d
+; CHECK: ftrunc_u.d
+; CHECK: st.d
+; CHECK: .size llvm_mips_ftrunc_u_d_test
+;
 @llvm_mips_ftint_s_w_ARG1 = global <4 x float> <float 0.000000e+00, float 1.000000e+00, float 2.000000e+00, float 3.000000e+00>, align 16
 @llvm_mips_ftint_s_w_RES  = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
 
diff --git a/test/CodeGen/Mips/msa/3r-s.ll b/test/CodeGen/Mips/msa/3r-s.ll
index bcc6ba7..9031fd8 100644
--- a/test/CodeGen/Mips/msa/3r-s.ll
+++ b/test/CodeGen/Mips/msa/3r-s.ll
@@ -267,6 +267,94 @@ declare <2 x i64> @llvm.mips.sra.d(<2 x i64>, <2 x i64>) nounwind
 ; CHECK: st.d
 ; CHECK: .size llvm_mips_sra_d_test
 ;
+@llvm_mips_srar_b_ARG1 = global <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 8, i8 9, i8 10, i8 11, i8 12, i8 13, i8 14, i8 15>, align 16
+@llvm_mips_srar_b_ARG2 = global <16 x i8> <i8 16, i8 17, i8 18, i8 19, i8 20, i8 21, i8 22, i8 23, i8 24, i8 25, i8 26, i8 27, i8 28, i8 29, i8 30, i8 31>, align 16
+@llvm_mips_srar_b_RES  = global <16 x i8> <i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0>, align 16
+
+define void @llvm_mips_srar_b_test() nounwind {
+entry:
+  %0 = load <16 x i8>* @llvm_mips_srar_b_ARG1
+  %1 = load <16 x i8>* @llvm_mips_srar_b_ARG2
+  %2 = tail call <16 x i8> @llvm.mips.srar.b(<16 x i8> %0, <16 x i8> %1)
+  store <16 x i8> %2, <16 x i8>* @llvm_mips_srar_b_RES
+  ret void
+}
+
+declare <16 x i8> @llvm.mips.srar.b(<16 x i8>, <16 x i8>) nounwind
+
+; CHECK: llvm_mips_srar_b_test:
+; CHECK: ld.b
+; CHECK: ld.b
+; CHECK: srar.b
+; CHECK: st.b
+; CHECK: .size llvm_mips_srar_b_test
+;
+@llvm_mips_srar_h_ARG1 = global <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>, align 16
+@llvm_mips_srar_h_ARG2 = global <8 x i16> <i16 8, i16 9, i16 10, i16 11, i16 12, i16 13, i16 14, i16 15>, align 16
+@llvm_mips_srar_h_RES  = global <8 x i16> <i16 0, i16 0, i16 0, i16 0, i16 0, i16 0, i16 0, i16 0>, align 16
+
+define void @llvm_mips_srar_h_test() nounwind {
+entry:
+  %0 = load <8 x i16>* @llvm_mips_srar_h_ARG1
+  %1 = load <8 x i16>* @llvm_mips_srar_h_ARG2
+  %2 = tail call <8 x i16> @llvm.mips.srar.h(<8 x i16> %0, <8 x i16> %1)
+  store <8 x i16> %2, <8 x i16>* @llvm_mips_srar_h_RES
+  ret void
+}
+
+declare <8 x i16> @llvm.mips.srar.h(<8 x i16>, <8 x i16>) nounwind
+
+; CHECK: llvm_mips_srar_h_test:
+; CHECK: ld.h
+; CHECK: ld.h
+; CHECK: srar.h
+; CHECK: st.h
+; CHECK: .size llvm_mips_srar_h_test
+;
+@llvm_mips_srar_w_ARG1 = global <4 x i32> <i32 0, i32 1, i32 2, i32 3>, align 16
+@llvm_mips_srar_w_ARG2 = global <4 x i32> <i32 4, i32 5, i32 6, i32 7>, align 16
+@llvm_mips_srar_w_RES  = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
+
+define void @llvm_mips_srar_w_test() nounwind {
+entry:
+  %0 = load <4 x i32>* @llvm_mips_srar_w_ARG1
+  %1 = load <4 x i32>* @llvm_mips_srar_w_ARG2
+  %2 = tail call <4 x i32> @llvm.mips.srar.w(<4 x i32> %0, <4 x i32> %1)
+  store <4 x i32> %2, <4 x i32>* @llvm_mips_srar_w_RES
+  ret void
+}
+
+declare <4 x i32> @llvm.mips.srar.w(<4 x i32>, <4 x i32>) nounwind
+
+; CHECK: llvm_mips_srar_w_test:
+; CHECK: ld.w
+; CHECK: ld.w
+; CHECK: srar.w
+; CHECK: st.w
+; CHECK: .size llvm_mips_srar_w_test
+;
+@llvm_mips_srar_d_ARG1 = global <2 x i64> <i64 0, i64 1>, align 16
+@llvm_mips_srar_d_ARG2 = global <2 x i64> <i64 2, i64 3>, align 16
+@llvm_mips_srar_d_RES  = global <2 x i64> <i64 0, i64 0>, align 16
+
+define void @llvm_mips_srar_d_test() nounwind {
+entry:
+  %0 = load <2 x i64>* @llvm_mips_srar_d_ARG1
+  %1 = load <2 x i64>* @llvm_mips_srar_d_ARG2
+  %2 = tail call <2 x i64> @llvm.mips.srar.d(<2 x i64> %0, <2 x i64> %1)
+  store <2 x i64> %2, <2 x i64>* @llvm_mips_srar_d_RES
+  ret void
+}
+
+declare <2 x i64> @llvm.mips.srar.d(<2 x i64>, <2 x i64>) nounwind
+
+; CHECK: llvm_mips_srar_d_test:
+; CHECK: ld.d
+; CHECK: ld.d
+; CHECK: srar.d
+; CHECK: st.d
+; CHECK: .size llvm_mips_srar_d_test
+;
 @llvm_mips_srl_b_ARG1 = global <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 8, i8 9, i8 10, i8 11, i8 12, i8 13, i8 14, i8 15>, align 16
 @llvm_mips_srl_b_ARG2 = global <16 x i8> <i8 16, i8 17, i8 18, i8 19, i8 20, i8 21, i8 22, i8 23, i8 24, i8 25, i8 26, i8 27, i8 28, i8 29, i8 30, i8 31>, align 16
 @llvm_mips_srl_b_RES  = global <16 x i8> <i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0>, align 16
@@ -355,6 +443,94 @@ declare <2 x i64> @llvm.mips.srl.d(<2 x i64>, <2 x i64>) nounwind
 ; CHECK: st.d
 ; CHECK: .size llvm_mips_srl_d_test
 ;
+@llvm_mips_srlr_b_ARG1 = global <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 8, i8 9, i8 10, i8 11, i8 12, i8 13, i8 14, i8 15>, align 16
+@llvm_mips_srlr_b_ARG2 = global <16 x i8> <i8 16, i8 17, i8 18, i8 19, i8 20, i8 21, i8 22, i8 23, i8 24, i8 25, i8 26, i8 27, i8 28, i8 29, i8 30, i8 31>, align 16
+@llvm_mips_srlr_b_RES  = global <16 x i8> <i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0>, align 16
+
+define void @llvm_mips_srlr_b_test() nounwind {
+entry:
+  %0 = load <16 x i8>* @llvm_mips_srlr_b_ARG1
+  %1 = load <16 x i8>* @llvm_mips_srlr_b_ARG2
+  %2 = tail call <16 x i8> @llvm.mips.srlr.b(<16 x i8> %0, <16 x i8> %1)
+  store <16 x i8> %2, <16 x i8>* @llvm_mips_srlr_b_RES
+  ret void
+}
+
+declare <16 x i8> @llvm.mips.srlr.b(<16 x i8>, <16 x i8>) nounwind
+
+; CHECK: llvm_mips_srlr_b_test:
+; CHECK: ld.b
+; CHECK: ld.b
+; CHECK: srlr.b
+; CHECK: st.b
+; CHECK: .size llvm_mips_srlr_b_test
+;
+@llvm_mips_srlr_h_ARG1 = global <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>, align 16
+@llvm_mips_srlr_h_ARG2 = global <8 x i16> <i16 8, i16 9, i16 10, i16 11, i16 12, i16 13, i16 14, i16 15>, align 16
+@llvm_mips_srlr_h_RES  = global <8 x i16> <i16 0, i16 0, i16 0, i16 0, i16 0, i16 0, i16 0, i16 0>, align 16
+
+define void @llvm_mips_srlr_h_test() nounwind {
+entry:
+  %0 = load <8 x i16>* @llvm_mips_srlr_h_ARG1
+  %1 = load <8 x i16>* @llvm_mips_srlr_h_ARG2
+  %2 = tail call <8 x i16> @llvm.mips.srlr.h(<8 x i16> %0, <8 x i16> %1)
+  store <8 x i16> %2, <8 x i16>* @llvm_mips_srlr_h_RES
+  ret void
+}
+
+declare <8 x i16> @llvm.mips.srlr.h(<8 x i16>, <8 x i16>) nounwind
+
+; CHECK: llvm_mips_srlr_h_test:
+; CHECK: ld.h
+; CHECK: ld.h
+; CHECK: srlr.h
+; CHECK: st.h
+; CHECK: .size llvm_mips_srlr_h_test
+;
+@llvm_mips_srlr_w_ARG1 = global <4 x i32> <i32 0, i32 1, i32 2, i32 3>, align 16
+@llvm_mips_srlr_w_ARG2 = global <4 x i32> <i32 4, i32 5, i32 6, i32 7>, align 16
+@llvm_mips_srlr_w_RES  = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
+
+define void @llvm_mips_srlr_w_test() nounwind {
+entry:
+  %0 = load <4 x i32>* @llvm_mips_srlr_w_ARG1
+  %1 = load <4 x i32>* @llvm_mips_srlr_w_ARG2
+  %2 = tail call <4 x i32> @llvm.mips.srlr.w(<4 x i32> %0, <4 x i32> %1)
+  store <4 x i32> %2, <4 x i32>* @llvm_mips_srlr_w_RES
+  ret void
+}
+
+declare <4 x i32> @llvm.mips.srlr.w(<4 x i32>, <4 x i32>) nounwind
+
+; CHECK: llvm_mips_srlr_w_test:
+; CHECK: ld.w
+; CHECK: ld.w
+; CHECK: srlr.w
+; CHECK: st.w
+; CHECK: .size llvm_mips_srlr_w_test
+;
+@llvm_mips_srlr_d_ARG1 = global <2 x i64> <i64 0, i64 1>, align 16
+@llvm_mips_srlr_d_ARG2 = global <2 x i64> <i64 2, i64 3>, align 16
+@llvm_mips_srlr_d_RES  = global <2 x i64> <i64 0, i64 0>, align 16
+
+define void @llvm_mips_srlr_d_test() nounwind {
+entry:
+  %0 = load <2 x i64>* @llvm_mips_srlr_d_ARG1
+  %1 = load <2 x i64>* @llvm_mips_srlr_d_ARG2
+  %2 = tail call <2 x i64> @llvm.mips.srlr.d(<2 x i64> %0, <2 x i64> %1)
+  store <2 x i64> %2, <2 x i64>* @llvm_mips_srlr_d_RES
+  ret void
+}
+
+declare <2 x i64> @llvm.mips.srlr.d(<2 x i64>, <2 x i64>) nounwind
+
+; CHECK: llvm_mips_srlr_d_test:
+; CHECK: ld.d
+; CHECK: ld.d
+; CHECK: srlr.d
+; CHECK: st.d
+; CHECK: .size llvm_mips_srlr_d_test
+;
 @llvm_mips_subs_s_b_ARG1 = global <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 8, i8 9, i8 10, i8 11, i8 12, i8 13, i8 14, i8 15>, align 16
 @llvm_mips_subs_s_b_ARG2 = global <16 x i8> <i8 16, i8 17, i8 18, i8 19, i8 20, i8 21, i8 22, i8 23, i8 24, i8 25, i8 26, i8 27, i8 28, i8 29, i8 30, i8 31>, align 16
 @llvm_mips_subs_s_b_RES  = global <16 x i8> <i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0>, align 16
diff --git a/test/CodeGen/Mips/msa/3rf_int_float.ll b/test/CodeGen/Mips/msa/3rf_int_float.ll
index 3d10ad3..819093b 100644
--- a/test/CodeGen/Mips/msa/3rf_int_float.ll
+++ b/test/CodeGen/Mips/msa/3rf_int_float.ll
@@ -3,6 +3,50 @@
 
 ; RUN: llc -march=mips -mattr=+msa < %s | FileCheck %s
 
+@llvm_mips_fcaf_w_ARG1 = global <4 x float> <float 0.000000e+00, float 1.000000e+00, float 2.000000e+00, float 3.000000e+00>, align 16
+@llvm_mips_fcaf_w_ARG2 = global <4 x float> <float 4.000000e+00, float 5.000000e+00, float 6.000000e+00, float 7.000000e+00>, align 16
+@llvm_mips_fcaf_w_RES  = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
+
+define void @llvm_mips_fcaf_w_test() nounwind {
+entry:
+  %0 = load <4 x float>* @llvm_mips_fcaf_w_ARG1
+  %1 = load <4 x float>* @llvm_mips_fcaf_w_ARG2
+  %2 = tail call <4 x i32> @llvm.mips.fcaf.w(<4 x float> %0, <4 x float> %1)
+  store <4 x i32> %2, <4 x i32>* @llvm_mips_fcaf_w_RES
+  ret void
+}
+
+declare <4 x i32> @llvm.mips.fcaf.w(<4 x float>, <4 x float>) nounwind
+
+; CHECK: llvm_mips_fcaf_w_test:
+; CHECK: ld.w
+; CHECK: ld.w
+; CHECK: fcaf.w
+; CHECK: st.w
+; CHECK: .size llvm_mips_fcaf_w_test
+;
+@llvm_mips_fcaf_d_ARG1 = global <2 x double> <double 0.000000e+00, double 1.000000e+00>, align 16
+@llvm_mips_fcaf_d_ARG2 = global <2 x double> <double 2.000000e+00, double 3.000000e+00>, align 16
+@llvm_mips_fcaf_d_RES  = global <2 x i64> <i64 0, i64 0>, align 16
+
+define void @llvm_mips_fcaf_d_test() nounwind {
+entry:
+  %0 = load <2 x double>* @llvm_mips_fcaf_d_ARG1
+  %1 = load <2 x double>* @llvm_mips_fcaf_d_ARG2
+  %2 = tail call <2 x i64> @llvm.mips.fcaf.d(<2 x double> %0, <2 x double> %1)
+  store <2 x i64> %2, <2 x i64>* @llvm_mips_fcaf_d_RES
+  ret void
+}
+
+declare <2 x i64> @llvm.mips.fcaf.d(<2 x double>, <2 x double>) nounwind
+
+; CHECK: llvm_mips_fcaf_d_test:
+; CHECK: ld.d
+; CHECK: ld.d
+; CHECK: fcaf.d
+; CHECK: st.d
+; CHECK: .size llvm_mips_fcaf_d_test
+;
 @llvm_mips_fceq_w_ARG1 = global <4 x float> <float 0.000000e+00, float 1.000000e+00, float 2.000000e+00, float 3.000000e+00>, align 16
 @llvm_mips_fceq_w_ARG2 = global <4 x float> <float 4.000000e+00, float 5.000000e+00, float 6.000000e+00, float 7.000000e+00>, align 16
 @llvm_mips_fceq_w_RES  = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
@@ -135,6 +179,50 @@ declare <2 x i64> @llvm.mips.fclt.d(<2 x double>, <2 x double>) nounwind
 ; CHECK: st.d
 ; CHECK: .size llvm_mips_fclt_d_test
 ;
+@llvm_mips_fcor_w_ARG1 = global <4 x float> <float 0.000000e+00, float 1.000000e+00, float 2.000000e+00, float 3.000000e+00>, align 16
+@llvm_mips_fcor_w_ARG2 = global <4 x float> <float 4.000000e+00, float 5.000000e+00, float 6.000000e+00, float 7.000000e+00>, align 16
+@llvm_mips_fcor_w_RES  = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
+
+define void @llvm_mips_fcor_w_test() nounwind {
+entry:
+  %0 = load <4 x float>* @llvm_mips_fcor_w_ARG1
+  %1 = load <4 x float>* @llvm_mips_fcor_w_ARG2
+  %2 = tail call <4 x i32> @llvm.mips.fcor.w(<4 x float> %0, <4 x float> %1)
+  store <4 x i32> %2, <4 x i32>* @llvm_mips_fcor_w_RES
+  ret void
+}
+
+declare <4 x i32> @llvm.mips.fcor.w(<4 x float>, <4 x float>) nounwind
+
+; CHECK: llvm_mips_fcor_w_test:
+; CHECK: ld.w
+; CHECK: ld.w
+; CHECK: fcor.w
+; CHECK: st.w
+; CHECK: .size llvm_mips_fcor_w_test
+;
+@llvm_mips_fcor_d_ARG1 = global <2 x double> <double 0.000000e+00, double 1.000000e+00>, align 16
+@llvm_mips_fcor_d_ARG2 = global <2 x double> <double 2.000000e+00, double 3.000000e+00>, align 16
+@llvm_mips_fcor_d_RES  = global <2 x i64> <i64 0, i64 0>, align 16
+
+define void @llvm_mips_fcor_d_test() nounwind {
+entry:
+  %0 = load <2 x double>* @llvm_mips_fcor_d_ARG1
+  %1 = load <2 x double>* @llvm_mips_fcor_d_ARG2
+  %2 = tail call <2 x i64> @llvm.mips.fcor.d(<2 x double> %0, <2 x double> %1)
+  store <2 x i64> %2, <2 x i64>* @llvm_mips_fcor_d_RES
+  ret void
+}
+
+declare <2 x i64> @llvm.mips.fcor.d(<2 x double>, <2 x double>) nounwind
+
+; CHECK: llvm_mips_fcor_d_test:
+; CHECK: ld.d
+; CHECK: ld.d
+; CHECK: fcor.d
+; CHECK: st.d
+; CHECK: .size llvm_mips_fcor_d_test
+;
 @llvm_mips_fcne_w_ARG1 = global <4 x float> <float 0.000000e+00, float 1.000000e+00, float 2.000000e+00, float 3.000000e+00>, align 16
 @llvm_mips_fcne_w_ARG2 = global <4 x float> <float 4.000000e+00, float 5.000000e+00, float 6.000000e+00, float 7.000000e+00>, align 16
 @llvm_mips_fcne_w_RES  = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
@@ -179,6 +267,138 @@ declare <2 x i64> @llvm.mips.fcne.d(<2 x double>, <2 x double>) nounwind
 ; CHECK: st.d
 ; CHECK: .size llvm_mips_fcne_d_test
 ;
+@llvm_mips_fcueq_w_ARG1 = global <4 x float> <float 0.000000e+00, float 1.000000e+00, float 2.000000e+00, float 3.000000e+00>, align 16
+@llvm_mips_fcueq_w_ARG2 = global <4 x float> <float 4.000000e+00, float 5.000000e+00, float 6.000000e+00, float 7.000000e+00>, align 16
+@llvm_mips_fcueq_w_RES  = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
+
+define void @llvm_mips_fcueq_w_test() nounwind {
+entry:
+  %0 = load <4 x float>* @llvm_mips_fcueq_w_ARG1
+  %1 = load <4 x float>* @llvm_mips_fcueq_w_ARG2
+  %2 = tail call <4 x i32> @llvm.mips.fcueq.w(<4 x float> %0, <4 x float> %1)
+  store <4 x i32> %2, <4 x i32>* @llvm_mips_fcueq_w_RES
+  ret void
+}
+
+declare <4 x i32> @llvm.mips.fcueq.w(<4 x float>, <4 x float>) nounwind
+
+; CHECK: llvm_mips_fcueq_w_test:
+; CHECK: ld.w
+; CHECK: ld.w
+; CHECK: fcueq.w
+; CHECK: st.w
+; CHECK: .size llvm_mips_fcueq_w_test
+;
+@llvm_mips_fcueq_d_ARG1 = global <2 x double> <double 0.000000e+00, double 1.000000e+00>, align 16
+@llvm_mips_fcueq_d_ARG2 = global <2 x double> <double 2.000000e+00, double 3.000000e+00>, align 16
+@llvm_mips_fcueq_d_RES  = global <2 x i64> <i64 0, i64 0>, align 16
+
+define void @llvm_mips_fcueq_d_test() nounwind {
+entry:
+  %0 = load <2 x double>* @llvm_mips_fcueq_d_ARG1
+  %1 = load <2 x double>* @llvm_mips_fcueq_d_ARG2
+  %2 = tail call <2 x i64> @llvm.mips.fcueq.d(<2 x double> %0, <2 x double> %1)
+  store <2 x i64> %2, <2 x i64>* @llvm_mips_fcueq_d_RES
+  ret void
+}
+
+declare <2 x i64> @llvm.mips.fcueq.d(<2 x double>, <2 x double>) nounwind
+
+; CHECK: llvm_mips_fcueq_d_test:
+; CHECK: ld.d
+; CHECK: ld.d
+; CHECK: fcueq.d
+; CHECK: st.d
+; CHECK: .size llvm_mips_fcueq_d_test
+;
+@llvm_mips_fcult_w_ARG1 = global <4 x float> <float 0.000000e+00, float 1.000000e+00, float 2.000000e+00, float 3.000000e+00>, align 16
+@llvm_mips_fcult_w_ARG2 = global <4 x float> <float 4.000000e+00, float 5.000000e+00, float 6.000000e+00, float 7.000000e+00>, align 16
+@llvm_mips_fcult_w_RES  = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
+
+define void @llvm_mips_fcult_w_test() nounwind {
+entry:
+  %0 = load <4 x float>* @llvm_mips_fcult_w_ARG1
+  %1 = load <4 x float>* @llvm_mips_fcult_w_ARG2
+  %2 = tail call <4 x i32> @llvm.mips.fcult.w(<4 x float> %0, <4 x float> %1)
+  store <4 x i32> %2, <4 x i32>* @llvm_mips_fcult_w_RES
+  ret void
+}
+
+declare <4 x i32> @llvm.mips.fcult.w(<4 x float>, <4 x float>) nounwind
+
+; CHECK: llvm_mips_fcult_w_test:
+; CHECK: ld.w
+; CHECK: ld.w
+; CHECK: fcult.w
+; CHECK: st.w
+; CHECK: .size llvm_mips_fcult_w_test
+;
+@llvm_mips_fcult_d_ARG1 = global <2 x double> <double 0.000000e+00, double 1.000000e+00>, align 16
+@llvm_mips_fcult_d_ARG2 = global <2 x double> <double 2.000000e+00, double 3.000000e+00>, align 16
+@llvm_mips_fcult_d_RES  = global <2 x i64> <i64 0, i64 0>, align 16
+
+define void @llvm_mips_fcult_d_test() nounwind {
+entry:
+  %0 = load <2 x double>* @llvm_mips_fcult_d_ARG1
+  %1 = load <2 x double>* @llvm_mips_fcult_d_ARG2
+  %2 = tail call <2 x i64> @llvm.mips.fcult.d(<2 x double> %0, <2 x double> %1)
+  store <2 x i64> %2, <2 x i64>* @llvm_mips_fcult_d_RES
+  ret void
+}
+
+declare <2 x i64> @llvm.mips.fcult.d(<2 x double>, <2 x double>) nounwind
+
+; CHECK: llvm_mips_fcult_d_test:
+; CHECK: ld.d
+; CHECK: ld.d
+; CHECK: fcult.d
+; CHECK: st.d
+; CHECK: .size llvm_mips_fcult_d_test
+;
+@llvm_mips_fcule_w_ARG1 = global <4 x float> <float 0.000000e+00, float 1.000000e+00, float 2.000000e+00, float 3.000000e+00>, align 16
+@llvm_mips_fcule_w_ARG2 = global <4 x float> <float 4.000000e+00, float 5.000000e+00, float 6.000000e+00, float 7.000000e+00>, align 16
+@llvm_mips_fcule_w_RES  = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
+
+define void @llvm_mips_fcule_w_test() nounwind {
+entry:
+  %0 = load <4 x float>* @llvm_mips_fcule_w_ARG1
+  %1 = load <4 x float>* @llvm_mips_fcule_w_ARG2
+  %2 = tail call <4 x i32> @llvm.mips.fcule.w(<4 x float> %0, <4 x float> %1)
+  store <4 x i32> %2, <4 x i32>* @llvm_mips_fcule_w_RES
+  ret void
+}
+
+declare <4 x i32> @llvm.mips.fcule.w(<4 x float>, <4 x float>) nounwind
+
+; CHECK: llvm_mips_fcule_w_test:
+; CHECK: ld.w
+; CHECK: ld.w
+; CHECK: fcule.w
+; CHECK: st.w
+; CHECK: .size llvm_mips_fcule_w_test
+;
+@llvm_mips_fcule_d_ARG1 = global <2 x double> <double 0.000000e+00, double 1.000000e+00>, align 16
+@llvm_mips_fcule_d_ARG2 = global <2 x double> <double 2.000000e+00, double 3.000000e+00>, align 16
+@llvm_mips_fcule_d_RES  = global <2 x i64> <i64 0, i64 0>, align 16
+
+define void @llvm_mips_fcule_d_test() nounwind {
+entry:
+  %0 = load <2 x double>* @llvm_mips_fcule_d_ARG1
+  %1 = load <2 x double>* @llvm_mips_fcule_d_ARG2
+  %2 = tail call <2 x i64> @llvm.mips.fcule.d(<2 x double> %0, <2 x double> %1)
+  store <2 x i64> %2, <2 x i64>* @llvm_mips_fcule_d_RES
+  ret void
+}
+
+declare <2 x i64> @llvm.mips.fcule.d(<2 x double>, <2 x double>) nounwind
+
+; CHECK: llvm_mips_fcule_d_test:
+; CHECK: ld.d
+; CHECK: ld.d
+; CHECK: fcule.d
+; CHECK: st.d
+; CHECK: .size llvm_mips_fcule_d_test
+;
 @llvm_mips_fcun_w_ARG1 = global <4 x float> <float 0.000000e+00, float 1.000000e+00, float 2.000000e+00, float 3.000000e+00>, align 16
 @llvm_mips_fcun_w_ARG2 = global <4 x float> <float 4.000000e+00, float 5.000000e+00, float 6.000000e+00, float 7.000000e+00>, align 16
 @llvm_mips_fcun_w_RES  = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
@@ -223,6 +443,94 @@ declare <2 x i64> @llvm.mips.fcun.d(<2 x double>, <2 x double>) nounwind
 ; CHECK: st.d
 ; CHECK: .size llvm_mips_fcun_d_test
 ;
+@llvm_mips_fcune_w_ARG1 = global <4 x float> <float 0.000000e+00, float 1.000000e+00, float 2.000000e+00, float 3.000000e+00>, align 16
+@llvm_mips_fcune_w_ARG2 = global <4 x float> <float 4.000000e+00, float 5.000000e+00, float 6.000000e+00, float 7.000000e+00>, align 16
+@llvm_mips_fcune_w_RES  = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
+
+define void @llvm_mips_fcune_w_test() nounwind {
+entry:
+  %0 = load <4 x float>* @llvm_mips_fcune_w_ARG1
+  %1 = load <4 x float>* @llvm_mips_fcune_w_ARG2
+  %2 = tail call <4 x i32> @llvm.mips.fcune.w(<4 x float> %0, <4 x float> %1)
+  store <4 x i32> %2, <4 x i32>* @llvm_mips_fcune_w_RES
+  ret void
+}
+
+declare <4 x i32> @llvm.mips.fcune.w(<4 x float>, <4 x float>) nounwind
+
+; CHECK: llvm_mips_fcune_w_test:
+; CHECK: ld.w
+; CHECK: ld.w
+; CHECK: fcune.w
+; CHECK: st.w
+; CHECK: .size llvm_mips_fcune_w_test
+;
+@llvm_mips_fcune_d_ARG1 = global <2 x double> <double 0.000000e+00, double 1.000000e+00>, align 16
+@llvm_mips_fcune_d_ARG2 = global <2 x double> <double 2.000000e+00, double 3.000000e+00>, align 16
+@llvm_mips_fcune_d_RES  = global <2 x i64> <i64 0, i64 0>, align 16
+
+define void @llvm_mips_fcune_d_test() nounwind {
+entry:
+  %0 = load <2 x double>* @llvm_mips_fcune_d_ARG1
+  %1 = load <2 x double>* @llvm_mips_fcune_d_ARG2
+  %2 = tail call <2 x i64> @llvm.mips.fcune.d(<2 x double> %0, <2 x double> %1)
+  store <2 x i64> %2, <2 x i64>* @llvm_mips_fcune_d_RES
+  ret void
+}
+
+declare <2 x i64> @llvm.mips.fcune.d(<2 x double>, <2 x double>) nounwind
+
+; CHECK: llvm_mips_fcune_d_test:
+; CHECK: ld.d
+; CHECK: ld.d
+; CHECK: fcune.d
+; CHECK: st.d
+; CHECK: .size llvm_mips_fcune_d_test
+;
+@llvm_mips_fsaf_w_ARG1 = global <4 x float> <float 0.000000e+00, float 1.000000e+00, float 2.000000e+00, float 3.000000e+00>, align 16
+@llvm_mips_fsaf_w_ARG2 = global <4 x float> <float 4.000000e+00, float 5.000000e+00, float 6.000000e+00, float 7.000000e+00>, align 16
+@llvm_mips_fsaf_w_RES  = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
+
+define void @llvm_mips_fsaf_w_test() nounwind {
+entry:
+  %0 = load <4 x float>* @llvm_mips_fsaf_w_ARG1
+  %1 = load <4 x float>* @llvm_mips_fsaf_w_ARG2
+  %2 = tail call <4 x i32> @llvm.mips.fsaf.w(<4 x float> %0, <4 x float> %1)
+  store <4 x i32> %2, <4 x i32>* @llvm_mips_fsaf_w_RES
+  ret void
+}
+
+declare <4 x i32> @llvm.mips.fsaf.w(<4 x float>, <4 x float>) nounwind
+
+; CHECK: llvm_mips_fsaf_w_test:
+; CHECK: ld.w
+; CHECK: ld.w
+; CHECK: fsaf.w
+; CHECK: st.w
+; CHECK: .size llvm_mips_fsaf_w_test
+;
+@llvm_mips_fsaf_d_ARG1 = global <2 x double> <double 0.000000e+00, double 1.000000e+00>, align 16
+@llvm_mips_fsaf_d_ARG2 = global <2 x double> <double 2.000000e+00, double 3.000000e+00>, align 16
+@llvm_mips_fsaf_d_RES  = global <2 x i64> <i64 0, i64 0>, align 16
+
+define void @llvm_mips_fsaf_d_test() nounwind {
+entry:
+  %0 = load <2 x double>* @llvm_mips_fsaf_d_ARG1
+  %1 = load <2 x double>* @llvm_mips_fsaf_d_ARG2
+  %2 = tail call <2 x i64> @llvm.mips.fsaf.d(<2 x double> %0, <2 x double> %1)
+  store <2 x i64> %2, <2 x i64>* @llvm_mips_fsaf_d_RES
+  ret void
+}
+
+declare <2 x i64> @llvm.mips.fsaf.d(<2 x double>, <2 x double>) nounwind
+
+; CHECK: llvm_mips_fsaf_d_test:
+; CHECK: ld.d
+; CHECK: ld.d
+; CHECK: fsaf.d
+; CHECK: st.d
+; CHECK: .size llvm_mips_fsaf_d_test
+;
 @llvm_mips_fseq_w_ARG1 = global <4 x float> <float 0.000000e+00, float 1.000000e+00, float 2.000000e+00, float 3.000000e+00>, align 16
 @llvm_mips_fseq_w_ARG2 = global <4 x float> <float 4.000000e+00, float 5.000000e+00, float 6.000000e+00, float 7.000000e+00>, align 16
 @llvm_mips_fseq_w_RES  = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
@@ -355,6 +663,50 @@ declare <2 x i64> @llvm.mips.fslt.d(<2 x double>, <2 x double>) nounwind
 ; CHECK: st.d
 ; CHECK: .size llvm_mips_fslt_d_test
 ;
+@llvm_mips_fsor_w_ARG1 = global <4 x float> <float 0.000000e+00, float 1.000000e+00, float 2.000000e+00, float 3.000000e+00>, align 16
+@llvm_mips_fsor_w_ARG2 = global <4 x float> <float 4.000000e+00, float 5.000000e+00, float 6.000000e+00, float 7.000000e+00>, align 16
+@llvm_mips_fsor_w_RES  = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
+
+define void @llvm_mips_fsor_w_test() nounwind {
+entry:
+  %0 = load <4 x float>* @llvm_mips_fsor_w_ARG1
+  %1 = load <4 x float>* @llvm_mips_fsor_w_ARG2
+  %2 = tail call <4 x i32> @llvm.mips.fsor.w(<4 x float> %0, <4 x float> %1)
+  store <4 x i32> %2, <4 x i32>* @llvm_mips_fsor_w_RES
+  ret void
+}
+
+declare <4 x i32> @llvm.mips.fsor.w(<4 x float>, <4 x float>) nounwind
+
+; CHECK: llvm_mips_fsor_w_test:
+; CHECK: ld.w
+; CHECK: ld.w
+; CHECK: fsor.w
+; CHECK: st.w
+; CHECK: .size llvm_mips_fsor_w_test
+;
+@llvm_mips_fsor_d_ARG1 = global <2 x double> <double 0.000000e+00, double 1.000000e+00>, align 16
+@llvm_mips_fsor_d_ARG2 = global <2 x double> <double 2.000000e+00, double 3.000000e+00>, align 16
+@llvm_mips_fsor_d_RES  = global <2 x i64> <i64 0, i64 0>, align 16
+
+define void @llvm_mips_fsor_d_test() nounwind {
+entry:
+  %0 = load <2 x double>* @llvm_mips_fsor_d_ARG1
+  %1 = load <2 x double>* @llvm_mips_fsor_d_ARG2
+  %2 = tail call <2 x i64> @llvm.mips.fsor.d(<2 x double> %0, <2 x double> %1)
+  store <2 x i64> %2, <2 x i64>* @llvm_mips_fsor_d_RES
+  ret void
+}
+
+declare <2 x i64> @llvm.mips.fsor.d(<2 x double>, <2 x double>) nounwind
+
+; CHECK: llvm_mips_fsor_d_test:
+; CHECK: ld.d
+; CHECK: ld.d
+; CHECK: fsor.d
+; CHECK: st.d
+; CHECK: .size llvm_mips_fsor_d_test
+;
 @llvm_mips_fsne_w_ARG1 = global <4 x float> <float 0.000000e+00, float 1.000000e+00, float 2.000000e+00, float 3.000000e+00>, align 16
 @llvm_mips_fsne_w_ARG2 = global <4 x float> <float 4.000000e+00, float 5.000000e+00, float 6.000000e+00, float 7.000000e+00>, align 16
 @llvm_mips_fsne_w_RES  = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
@@ -399,3 +751,223 @@ declare <2 x i64> @llvm.mips.fsne.d(<2 x double>, <2 x double>) nounwind
 ; CHECK: st.d
 ; CHECK: .size llvm_mips_fsne_d_test
 ;
+@llvm_mips_fsueq_w_ARG1 = global <4 x float> <float 0.000000e+00, float 1.000000e+00, float 2.000000e+00, float 3.000000e+00>, align 16
+@llvm_mips_fsueq_w_ARG2 = global <4 x float> <float 4.000000e+00, float 5.000000e+00, float 6.000000e+00, float 7.000000e+00>, align 16
+@llvm_mips_fsueq_w_RES  = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
+
+define void @llvm_mips_fsueq_w_test() nounwind {
+entry:
+  %0 = load <4 x float>* @llvm_mips_fsueq_w_ARG1
+  %1 = load <4 x float>* @llvm_mips_fsueq_w_ARG2
+  %2 = tail call <4 x i32> @llvm.mips.fsueq.w(<4 x float> %0, <4 x float> %1)
+  store <4 x i32> %2, <4 x i32>* @llvm_mips_fsueq_w_RES
+  ret void
+}
+
+declare <4 x i32> @llvm.mips.fsueq.w(<4 x float>, <4 x float>) nounwind
+
+; CHECK: llvm_mips_fsueq_w_test:
+; CHECK: ld.w
+; CHECK: ld.w
+; CHECK: fsueq.w
+; CHECK: st.w
+; CHECK: .size llvm_mips_fsueq_w_test
+;
+@llvm_mips_fsueq_d_ARG1 = global <2 x double> <double 0.000000e+00, double 1.000000e+00>, align 16
+@llvm_mips_fsueq_d_ARG2 = global <2 x double> <double 2.000000e+00, double 3.000000e+00>, align 16
+@llvm_mips_fsueq_d_RES  = global <2 x i64> <i64 0, i64 0>, align 16
+
+define void @llvm_mips_fsueq_d_test() nounwind {
+entry:
+  %0 = load <2 x double>* @llvm_mips_fsueq_d_ARG1
+  %1 = load <2 x double>* @llvm_mips_fsueq_d_ARG2
+  %2 = tail call <2 x i64> @llvm.mips.fsueq.d(<2 x double> %0, <2 x double> %1)
+  store <2 x i64> %2, <2 x i64>* @llvm_mips_fsueq_d_RES
+  ret void
+}
+
+declare <2 x i64> @llvm.mips.fsueq.d(<2 x double>, <2 x double>) nounwind
+
+; CHECK: llvm_mips_fsueq_d_test:
+; CHECK: ld.d
+; CHECK: ld.d
+; CHECK: fsueq.d
+; CHECK: st.d
+; CHECK: .size llvm_mips_fsueq_d_test
+;
+@llvm_mips_fsult_w_ARG1 = global <4 x float> <float 0.000000e+00, float 1.000000e+00, float 2.000000e+00, float 3.000000e+00>, align 16
+@llvm_mips_fsult_w_ARG2 = global <4 x float> <float 4.000000e+00, float 5.000000e+00, float 6.000000e+00, float 7.000000e+00>, align 16
+@llvm_mips_fsult_w_RES  = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
+
+define void @llvm_mips_fsult_w_test() nounwind {
+entry:
+  %0 = load <4 x float>* @llvm_mips_fsult_w_ARG1
+  %1 = load <4 x float>* @llvm_mips_fsult_w_ARG2
+  %2 = tail call <4 x i32> @llvm.mips.fsult.w(<4 x float> %0, <4 x float> %1)
+  store <4 x i32> %2, <4 x i32>* @llvm_mips_fsult_w_RES
+  ret void
+}
+
+declare <4 x i32> @llvm.mips.fsult.w(<4 x float>, <4 x float>) nounwind
+
+; CHECK: llvm_mips_fsult_w_test:
+; CHECK: ld.w
+; CHECK: ld.w
+; CHECK: fsult.w
+; CHECK: st.w
+; CHECK: .size llvm_mips_fsult_w_test
+;
+@llvm_mips_fsult_d_ARG1 = global <2 x double> <double 0.000000e+00, double 1.000000e+00>, align 16
+@llvm_mips_fsult_d_ARG2 = global <2 x double> <double 2.000000e+00, double 3.000000e+00>, align 16
+@llvm_mips_fsult_d_RES  = global <2 x i64> <i64 0, i64 0>, align 16
+
+define void @llvm_mips_fsult_d_test() nounwind {
+entry:
+  %0 = load <2 x double>* @llvm_mips_fsult_d_ARG1
+  %1 = load <2 x double>* @llvm_mips_fsult_d_ARG2
+  %2 = tail call <2 x i64> @llvm.mips.fsult.d(<2 x double> %0, <2 x double> %1)
+  store <2 x i64> %2, <2 x i64>* @llvm_mips_fsult_d_RES
+  ret void
+}
+
+declare <2 x i64> @llvm.mips.fsult.d(<2 x double>, <2 x double>) nounwind
+
+; CHECK: llvm_mips_fsult_d_test:
+; CHECK: ld.d
+; CHECK: ld.d
+; CHECK: fsult.d
+; CHECK: st.d
+; CHECK: .size llvm_mips_fsult_d_test
+;
+@llvm_mips_fsule_w_ARG1 = global <4 x float> <float 0.000000e+00, float 1.000000e+00, float 2.000000e+00, float 3.000000e+00>, align 16
+@llvm_mips_fsule_w_ARG2 = global <4 x float> <float 4.000000e+00, float 5.000000e+00, float 6.000000e+00, float 7.000000e+00>, align 16
+@llvm_mips_fsule_w_RES  = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
+
+define void @llvm_mips_fsule_w_test() nounwind {
+entry:
+  %0 = load <4 x float>* @llvm_mips_fsule_w_ARG1
+  %1 = load <4 x float>* @llvm_mips_fsule_w_ARG2
+  %2 = tail call <4 x i32> @llvm.mips.fsule.w(<4 x float> %0, <4 x float> %1)
+  store <4 x i32> %2, <4 x i32>* @llvm_mips_fsule_w_RES
+  ret void
+}
+
+declare <4 x i32> @llvm.mips.fsule.w(<4 x float>, <4 x float>) nounwind
+
+; CHECK: llvm_mips_fsule_w_test:
+; CHECK: ld.w
+; CHECK: ld.w
+; CHECK: fsule.w
+; CHECK: st.w
+; CHECK: .size llvm_mips_fsule_w_test
+;
+@llvm_mips_fsule_d_ARG1 = global <2 x double> <double 0.000000e+00, double 1.000000e+00>, align 16
+@llvm_mips_fsule_d_ARG2 = global <2 x double> <double 2.000000e+00, double 3.000000e+00>, align 16
+@llvm_mips_fsule_d_RES  = global <2 x i64> <i64 0, i64 0>, align 16
+
+define void @llvm_mips_fsule_d_test() nounwind {
+entry:
+  %0 = load <2 x double>* @llvm_mips_fsule_d_ARG1
+  %1 = load <2 x double>* @llvm_mips_fsule_d_ARG2
+  %2 = tail call <2 x i64> @llvm.mips.fsule.d(<2 x double> %0, <2 x double> %1)
+  store <2 x i64> %2, <2 x i64>* @llvm_mips_fsule_d_RES
+  ret void
+}
+
+declare <2 x i64> @llvm.mips.fsule.d(<2 x double>, <2 x double>) nounwind
+
+; CHECK: llvm_mips_fsule_d_test:
+; CHECK: ld.d
+; CHECK: ld.d
+; CHECK: fsule.d
+; CHECK: st.d
+; CHECK: .size llvm_mips_fsule_d_test
+;
+@llvm_mips_fsun_w_ARG1 = global <4 x float> <float 0.000000e+00, float 1.000000e+00, float 2.000000e+00, float 3.000000e+00>, align 16
+@llvm_mips_fsun_w_ARG2 = global <4 x float> <float 4.000000e+00, float 5.000000e+00, float 6.000000e+00, float 7.000000e+00>, align 16
+@llvm_mips_fsun_w_RES  = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
+
+define void @llvm_mips_fsun_w_test() nounwind {
+entry:
+  %0 = load <4 x float>* @llvm_mips_fsun_w_ARG1
+  %1 = load <4 x float>* @llvm_mips_fsun_w_ARG2
+  %2 = tail call <4 x i32> @llvm.mips.fsun.w(<4 x float> %0, <4 x float> %1)
+  store <4 x i32> %2, <4 x i32>* @llvm_mips_fsun_w_RES
+  ret void
+}
+
+declare <4 x i32> @llvm.mips.fsun.w(<4 x float>, <4 x float>) nounwind
+
+; CHECK: llvm_mips_fsun_w_test:
+; CHECK: ld.w
+; CHECK: ld.w
+; CHECK: fsun.w
+; CHECK: st.w
+; CHECK: .size llvm_mips_fsun_w_test
+;
+@llvm_mips_fsun_d_ARG1 = global <2 x double> <double 0.000000e+00, double 1.000000e+00>, align 16
+@llvm_mips_fsun_d_ARG2 = global <2 x double> <double 2.000000e+00, double 3.000000e+00>, align 16
+@llvm_mips_fsun_d_RES  = global <2 x i64> <i64 0, i64 0>, align 16
+
+define void @llvm_mips_fsun_d_test() nounwind {
+entry:
+  %0 = load <2 x double>* @llvm_mips_fsun_d_ARG1
+  %1 = load <2 x double>* @llvm_mips_fsun_d_ARG2
+  %2 = tail call <2 x i64> @llvm.mips.fsun.d(<2 x double> %0, <2 x double> %1)
+  store <2 x i64> %2, <2 x i64>* @llvm_mips_fsun_d_RES
+  ret void
+}
+
+declare <2 x i64> @llvm.mips.fsun.d(<2 x double>, <2 x double>) nounwind
+
+; CHECK: llvm_mips_fsun_d_test:
+; CHECK: ld.d
+; CHECK: ld.d
+; CHECK: fsun.d
+; CHECK: st.d
+; CHECK: .size llvm_mips_fsun_d_test
+;
+@llvm_mips_fsune_w_ARG1 = global <4 x float> <float 0.000000e+00, float 1.000000e+00, float 2.000000e+00, float 3.000000e+00>, align 16
+@llvm_mips_fsune_w_ARG2 = global <4 x float> <float 4.000000e+00, float 5.000000e+00, float 6.000000e+00, float 7.000000e+00>, align 16
+@llvm_mips_fsune_w_RES  = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
+
+define void @llvm_mips_fsune_w_test() nounwind {
+entry:
+  %0 = load <4 x float>* @llvm_mips_fsune_w_ARG1
+  %1 = load <4 x float>* @llvm_mips_fsune_w_ARG2
+  %2 = tail call <4 x i32> @llvm.mips.fsune.w(<4 x float> %0, <4 x float> %1)
+  store <4 x i32> %2, <4 x i32>* @llvm_mips_fsune_w_RES
+  ret void
+}
+
+declare <4 x i32> @llvm.mips.fsune.w(<4 x float>, <4 x float>) nounwind
+
+; CHECK: llvm_mips_fsune_w_test:
+; CHECK: ld.w
+; CHECK: ld.w
+; CHECK: fsune.w
+; CHECK: st.w
+; CHECK: .size llvm_mips_fsune_w_test
+;
+@llvm_mips_fsune_d_ARG1 = global <2 x double> <double 0.000000e+00, double 1.000000e+00>, align 16
+@llvm_mips_fsune_d_ARG2 = global <2 x double> <double 2.000000e+00, double 3.000000e+00>, align 16
+@llvm_mips_fsune_d_RES  = global <2 x i64> <i64 0, i64 0>, align 16
+
+define void @llvm_mips_fsune_d_test() nounwind {
+entry:
+  %0 = load <2 x double>* @llvm_mips_fsune_d_ARG1
+  %1 = load <2 x double>* @llvm_mips_fsune_d_ARG2
+  %2 = tail call <2 x i64> @llvm.mips.fsune.d(<2 x double> %0, <2 x double> %1)
+  store <2 x i64> %2, <2 x i64>* @llvm_mips_fsune_d_RES
+  ret void
+}
+
+declare <2 x i64> @llvm.mips.fsune.d(<2 x double>, <2 x double>) nounwind
+
+; CHECK: llvm_mips_fsune_d_test:
+; CHECK: ld.d
+; CHECK: ld.d
+; CHECK: fsune.d
+; CHECK: st.d
+; CHECK: .size llvm_mips_fsune_d_test
+;
diff --git a/test/CodeGen/Mips/msa/bit.ll b/test/CodeGen/Mips/msa/bit.ll
index 28bf0e7..f39f2c7 100644
--- a/test/CodeGen/Mips/msa/bit.ll
+++ b/test/CodeGen/Mips/msa/bit.ll
@@ -306,6 +306,82 @@ declare <2 x i64> @llvm.mips.srai.d(<2 x i64>, i32) nounwind
 ; CHECK: st.d
 ; CHECK: .size llvm_mips_srai_d_test
 ;
+@llvm_mips_srari_b_ARG1 = global <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 8, i8 9, i8 10, i8 11, i8 12, i8 13, i8 14, i8 15>, align 16
+@llvm_mips_srari_b_RES  = global <16 x i8> <i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0>, align 16
+
+define void @llvm_mips_srari_b_test() nounwind {
+entry:
+  %0 = load <16 x i8>* @llvm_mips_srari_b_ARG1
+  %1 = tail call <16 x i8> @llvm.mips.srari.b(<16 x i8> %0, i32 7)
+  store <16 x i8> %1, <16 x i8>* @llvm_mips_srari_b_RES
+  ret void
+}
+
+declare <16 x i8> @llvm.mips.srari.b(<16 x i8>, i32) nounwind
+
+; CHECK: llvm_mips_srari_b_test:
+; CHECK: ld.b
+; CHECK: srari.b
+; CHECK: st.b
+; CHECK: .size llvm_mips_srari_b_test
+;
+@llvm_mips_srari_h_ARG1 = global <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>, align 16
+@llvm_mips_srari_h_RES  = global <8 x i16> <i16 0, i16 0, i16 0, i16 0, i16 0, i16 0, i16 0, i16 0>, align 16
+
+define void @llvm_mips_srari_h_test() nounwind {
+entry:
+  %0 = load <8 x i16>* @llvm_mips_srari_h_ARG1
+  %1 = tail call <8 x i16> @llvm.mips.srari.h(<8 x i16> %0, i32 7)
+  store <8 x i16> %1, <8 x i16>* @llvm_mips_srari_h_RES
+  ret void
+}
+
+declare <8 x i16> @llvm.mips.srari.h(<8 x i16>, i32) nounwind
+
+; CHECK: llvm_mips_srari_h_test:
+; CHECK: ld.h
+; CHECK: srari.h
+; CHECK: st.h
+; CHECK: .size llvm_mips_srari_h_test
+;
+@llvm_mips_srari_w_ARG1 = global <4 x i32> <i32 0, i32 1, i32 2, i32 3>, align 16
+@llvm_mips_srari_w_RES  = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
+
+define void @llvm_mips_srari_w_test() nounwind {
+entry:
+  %0 = load <4 x i32>* @llvm_mips_srari_w_ARG1
+  %1 = tail call <4 x i32> @llvm.mips.srari.w(<4 x i32> %0, i32 7)
+  store <4 x i32> %1, <4 x i32>* @llvm_mips_srari_w_RES
+  ret void
+}
+
+declare <4 x i32> @llvm.mips.srari.w(<4 x i32>, i32) nounwind
+
+; CHECK: llvm_mips_srari_w_test:
+; CHECK: ld.w
+; CHECK: srari.w
+; CHECK: st.w
+; CHECK: .size llvm_mips_srari_w_test
+;
+@llvm_mips_srari_d_ARG1 = global <2 x i64> <i64 0, i64 1>, align 16
+@llvm_mips_srari_d_RES  = global <2 x i64> <i64 0, i64 0>, align 16
+
+define void @llvm_mips_srari_d_test() nounwind {
+entry:
+  %0 = load <2 x i64>* @llvm_mips_srari_d_ARG1
+  %1 = tail call <2 x i64> @llvm.mips.srari.d(<2 x i64> %0, i32 7)
+  store <2 x i64> %1, <2 x i64>* @llvm_mips_srari_d_RES
+  ret void
+}
+
+declare <2 x i64> @llvm.mips.srari.d(<2 x i64>, i32) nounwind
+
+; CHECK: llvm_mips_srari_d_test:
+; CHECK: ld.d
+; CHECK: srari.d
+; CHECK: st.d
+; CHECK: .size llvm_mips_srari_d_test
+;
 @llvm_mips_srli_b_ARG1 = global <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 8, i8 9, i8 10, i8 11, i8 12, i8 13, i8 14, i8 15>, align 16
 @llvm_mips_srli_b_RES  = global <16 x i8> <i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0>, align 16
 
@@ -382,3 +458,79 @@ declare <2 x i64> @llvm.mips.srli.d(<2 x i64>, i32) nounwind
 ; CHECK: st.d
 ; CHECK: .size llvm_mips_srli_d_test
 ;
+@llvm_mips_srlri_b_ARG1 = global <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 8, i8 9, i8 10, i8 11, i8 12, i8 13, i8 14, i8 15>, align 16
+@llvm_mips_srlri_b_RES  = global <16 x i8> <i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0>, align 16
+
+define void @llvm_mips_srlri_b_test() nounwind {
+entry:
+  %0 = load <16 x i8>* @llvm_mips_srlri_b_ARG1
+  %1 = tail call <16 x i8> @llvm.mips.srlri.b(<16 x i8> %0, i32 7)
+  store <16 x i8> %1, <16 x i8>* @llvm_mips_srlri_b_RES
+  ret void
+}
+
+declare <16 x i8> @llvm.mips.srlri.b(<16 x i8>, i32) nounwind
+
+; CHECK: llvm_mips_srlri_b_test:
+; CHECK: ld.b
+; CHECK: srlri.b
+; CHECK: st.b
+; CHECK: .size llvm_mips_srlri_b_test
+;
+@llvm_mips_srlri_h_ARG1 = global <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>, align 16
+@llvm_mips_srlri_h_RES  = global <8 x i16> <i16 0, i16 0, i16 0, i16 0, i16 0, i16 0, i16 0, i16 0>, align 16
+
+define void @llvm_mips_srlri_h_test() nounwind {
+entry:
+  %0 = load <8 x i16>* @llvm_mips_srlri_h_ARG1
+  %1 = tail call <8 x i16> @llvm.mips.srlri.h(<8 x i16> %0, i32 7)
+  store <8 x i16> %1, <8 x i16>* @llvm_mips_srlri_h_RES
+  ret void
+}
+
+declare <8 x i16> @llvm.mips.srlri.h(<8 x i16>, i32) nounwind
+
+; CHECK: llvm_mips_srlri_h_test:
+; CHECK: ld.h
+; CHECK: srlri.h
+; CHECK: st.h
+; CHECK: .size llvm_mips_srlri_h_test
+;
+@llvm_mips_srlri_w_ARG1 = global <4 x i32> <i32 0, i32 1, i32 2, i32 3>, align 16
+@llvm_mips_srlri_w_RES  = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
+
+define void @llvm_mips_srlri_w_test() nounwind {
+entry:
+  %0 = load <4 x i32>* @llvm_mips_srlri_w_ARG1
+  %1 = tail call <4 x i32> @llvm.mips.srlri.w(<4 x i32> %0, i32 7)
+  store <4 x i32> %1, <4 x i32>* @llvm_mips_srlri_w_RES
+  ret void
+}
+
+declare <4 x i32> @llvm.mips.srlri.w(<4 x i32>, i32) nounwind
+
+; CHECK: llvm_mips_srlri_w_test:
+; CHECK: ld.w
+; CHECK: srlri.w
+; CHECK: st.w
+; CHECK: .size llvm_mips_srlri_w_test
+;
+@llvm_mips_srlri_d_ARG1 = global <2 x i64> <i64 0, i64 1>, align 16
+@llvm_mips_srlri_d_RES  = global <2 x i64> <i64 0, i64 0>, align 16
+
+define void @llvm_mips_srlri_d_test() nounwind {
+entry:
+  %0 = load <2 x i64>* @llvm_mips_srlri_d_ARG1
+  %1 = tail call <2 x i64> @llvm.mips.srlri.d(<2 x i64> %0, i32 7)
+  store <2 x i64> %1, <2 x i64>* @llvm_mips_srlri_d_RES
+  ret void
+}
+
+declare <2 x i64> @llvm.mips.srlri.d(<2 x i64>, i32) nounwind
+
+; CHECK: llvm_mips_srlri_d_test:
+; CHECK: ld.d
+; CHECK: srlri.d
+; CHECK: st.d
+; CHECK: .size llvm_mips_srlri_d_test
+;