From d144da63d0f5fb309ac868bb9e13b4d4ae1e882b Mon Sep 17 00:00:00 2001
From: Alex Maclean <amaclean@nvidia.com>
Date: Thu, 17 Apr 2025 17:37:39 +0000
Subject: [PATCH] [DAGCombiner] Fold and/or of NaN SETCC - tests follow up

---
 llvm/test/CodeGen/NVPTX/and-or-setcc.ll |   4 +-
 llvm/test/CodeGen/X86/and-or-setcc.ll   | 118 ++++++++++++++++++++++++
 2 files changed, 120 insertions(+), 2 deletions(-)

diff --git a/llvm/test/CodeGen/NVPTX/and-or-setcc.ll b/llvm/test/CodeGen/NVPTX/and-or-setcc.ll
index 21be9df94d553..6c3514c1ad946 100644
--- a/llvm/test/CodeGen/NVPTX/and-or-setcc.ll
+++ b/llvm/test/CodeGen/NVPTX/and-or-setcc.ll
@@ -1,6 +1,6 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
-; RUN: llc < %s -mtriple=nvptx64 | FileCheck %s
-; RUN: %if ptxas %{ llc < %s -mtriple=nvptx64 | %ptxas-verify %}
+; RUN: llc < %s | FileCheck %s
+; RUN: %if ptxas %{ llc < %s | %ptxas-verify %}
 
 target triple = "nvptx64-nvidia-cuda"
 
diff --git a/llvm/test/CodeGen/X86/and-or-setcc.ll b/llvm/test/CodeGen/X86/and-or-setcc.ll
index cb8ecca9348e6..a6a9362908811 100644
--- a/llvm/test/CodeGen/X86/and-or-setcc.ll
+++ b/llvm/test/CodeGen/X86/and-or-setcc.ll
@@ -55,3 +55,121 @@ define i1 @or_uno(float %a, float %b) {
   %e = or i1 %c, %d
   ret i1 %e
 }
+
+define <4 x i1> @and_ord_vec(<4 x float> %a, <4 x float> %b) {
+; X86-LABEL: and_ord_vec:
+; X86:       # %bb.0:
+; X86-NEXT:    pushl %eax
+; X86-NEXT:    .cfi_def_cfa_offset 8
+; X86-NEXT:    flds {{[0-9]+}}(%esp)
+; X86-NEXT:    fstps (%esp) # 4-byte Folded Spill
+; X86-NEXT:    flds {{[0-9]+}}(%esp)
+; X86-NEXT:    flds {{[0-9]+}}(%esp)
+; X86-NEXT:    flds {{[0-9]+}}(%esp)
+; X86-NEXT:    flds {{[0-9]+}}(%esp)
+; X86-NEXT:    flds {{[0-9]+}}(%esp)
+; X86-NEXT:    flds {{[0-9]+}}(%esp)
+; X86-NEXT:    flds {{[0-9]+}}(%esp)
+; X86-NEXT:    fucompp
+; X86-NEXT:    fnstsw %ax
+; X86-NEXT:    fucompp
+; X86-NEXT:    # kill: def $ah killed $ah killed $ax
+; X86-NEXT:    sahf
+; X86-NEXT:    fnstsw %ax
+; X86-NEXT:    fucompp
+; X86-NEXT:    setnp %dh
+; X86-NEXT:    shlb $2, %dh
+; X86-NEXT:    # kill: def $ah killed $ah killed $ax
+; X86-NEXT:    sahf
+; X86-NEXT:    fnstsw %ax
+; X86-NEXT:    flds (%esp) # 4-byte Folded Reload
+; X86-NEXT:    fxch %st(1)
+; X86-NEXT:    fucompp
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
+; X86-NEXT:    setnp %dl
+; X86-NEXT:    shlb $3, %dl
+; X86-NEXT:    orb %dh, %dl
+; X86-NEXT:    # kill: def $ah killed $ah killed $ax
+; X86-NEXT:    sahf
+; X86-NEXT:    setnp %dh
+; X86-NEXT:    fnstsw %ax
+; X86-NEXT:    # kill: def $ah killed $ah killed $ax
+; X86-NEXT:    sahf
+; X86-NEXT:    setnp %al
+; X86-NEXT:    addb %al, %al
+; X86-NEXT:    orb %dh, %al
+; X86-NEXT:    orb %dl, %al
+; X86-NEXT:    movb %al, (%ecx)
+; X86-NEXT:    movl %ecx, %eax
+; X86-NEXT:    popl %ecx
+; X86-NEXT:    .cfi_def_cfa_offset 4
+; X86-NEXT:    retl $4
+;
+; X64-LABEL: and_ord_vec:
+; X64:       # %bb.0:
+; X64-NEXT:    cmpordps %xmm1, %xmm0
+; X64-NEXT:    retq
+  %c = fcmp ord <4 x float> %a, %a
+  %d = fcmp ord <4 x float> %b, %b
+  %e = and <4 x i1> %c, %d
+  ret <4 x i1> %e
+}
+
+define <4 x i1> @or_uno_vec(<4 x float> %a, <4 x float> %b) {
+; X86-LABEL: or_uno_vec:
+; X86:       # %bb.0:
+; X86-NEXT:    pushl %eax
+; X86-NEXT:    .cfi_def_cfa_offset 8
+; X86-NEXT:    flds {{[0-9]+}}(%esp)
+; X86-NEXT:    fstps (%esp) # 4-byte Folded Spill
+; X86-NEXT:    flds {{[0-9]+}}(%esp)
+; X86-NEXT:    flds {{[0-9]+}}(%esp)
+; X86-NEXT:    flds {{[0-9]+}}(%esp)
+; X86-NEXT:    flds {{[0-9]+}}(%esp)
+; X86-NEXT:    flds {{[0-9]+}}(%esp)
+; X86-NEXT:    flds {{[0-9]+}}(%esp)
+; X86-NEXT:    flds {{[0-9]+}}(%esp)
+; X86-NEXT:    fucompp
+; X86-NEXT:    fnstsw %ax
+; X86-NEXT:    fucompp
+; X86-NEXT:    # kill: def $ah killed $ah killed $ax
+; X86-NEXT:    sahf
+; X86-NEXT:    fnstsw %ax
+; X86-NEXT:    fucompp
+; X86-NEXT:    setp %dh
+; X86-NEXT:    shlb $2, %dh
+; X86-NEXT:    # kill: def $ah killed $ah killed $ax
+; X86-NEXT:    sahf
+; X86-NEXT:    fnstsw %ax
+; X86-NEXT:    flds (%esp) # 4-byte Folded Reload
+; X86-NEXT:    fxch %st(1)
+; X86-NEXT:    fucompp
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
+; X86-NEXT:    setp %dl
+; X86-NEXT:    shlb $3, %dl
+; X86-NEXT:    orb %dh, %dl
+; X86-NEXT:    # kill: def $ah killed $ah killed $ax
+; X86-NEXT:    sahf
+; X86-NEXT:    setp %dh
+; X86-NEXT:    fnstsw %ax
+; X86-NEXT:    # kill: def $ah killed $ah killed $ax
+; X86-NEXT:    sahf
+; X86-NEXT:    setp %al
+; X86-NEXT:    addb %al, %al
+; X86-NEXT:    orb %dh, %al
+; X86-NEXT:    orb %dl, %al
+; X86-NEXT:    movb %al, (%ecx)
+; X86-NEXT:    movl %ecx, %eax
+; X86-NEXT:    popl %ecx
+; X86-NEXT:    .cfi_def_cfa_offset 4
+; X86-NEXT:    retl $4
+;
+; X64-LABEL: or_uno_vec:
+; X64:       # %bb.0:
+; X64-NEXT:    cmpunordps %xmm1, %xmm0
+; X64-NEXT:    retq
+  %c = fcmp uno <4 x float> %a, %a
+  %d = fcmp uno <4 x float> %b, %b
+  %e = or <4 x i1> %c, %d
+  ret <4 x i1> %e
+}