The _get_maxmin_fp_NaN() function is used in instructions xvmaxsp, xvminsp, xvmindp, xvmaxdp. The case where srcA is QNaN and src is SNaN returns SNaN. However, in this case the HW sets the result to QNaN.
Created attachment 118974 [details] fix the _get_maxmin_fp_NAN function The attached patch was committed to fix the issue. commit e9986500957ae53ef6374e86938188bca4d1c6df Author: Carl Love <carll@us.ibm.com> Date: Fri Mar 22 12:32:29 2019 -0500
No regressions found by automated regression testing. Closing.