[llvm] [InstCombine] Fix Failure to convert vector fp comparisons that can be represented as integers #82241 (PR #83274)

[Yingwei Zheng via llvm-commits]

================
@@ -509,3 +509,61 @@ define i1 @i128_cast_cmp_oeq_int_inf_uitofp(i128 %i) {
   %cmp = fcmp oeq float %f, 0x7FF0000000000000
   ret i1 %cmp
 }
+
+define <2 x i1> @i32_vec_cast_cmp_oeq_vec_int_0_sitofp(<2 x i32> %i) {
----------------
dtcxzyw wrote:

Please pre-commit these tests.


https://github.com/llvm/llvm-project/pull/83274