[llvm] [LoongArch] Fix fp_to_uint/fp_to_sint conversion errors for lasx (PR #137129)

[via llvm-commits]

================
@@ -31,9 +31,9 @@ define void @fptosi_v4f64_v4i32(ptr %res, ptr %in){
 ; CHECK-LABEL: fptosi_v4f64_v4i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    xvld $xr0, $a1, 0
+; CHECK-NEXT:    xvftintrz.l.d $xr0, $xr0
 ; CHECK-NEXT:    xvpermi.d $xr1, $xr0, 238
-; CHECK-NEXT:    xvfcvt.s.d $xr0, $xr1, $xr0
-; CHECK-NEXT:    xvftintrz.w.s $xr0, $xr0
+; CHECK-NEXT:    xvpickev.w $xr0, $xr1, $xr0
 ; CHECK-NEXT:    vst $vr0, $a0, 0
----------------
heiher wrote:

Just to correct my earlier comment - it was inaccurate and has been deleted. :P

According to the LLVM IR [`fptosi`](https://llvm.org/docs/LangRef.html#id275) spec:

> If the value cannot fit in ty2, the result is a poison value.

So unless overflow/invalid recording is required, using `xvftintrz.l.d + xvpickev.w` is also valid, and this sequence may offer better throughput.

https://github.com/llvm/llvm-project/pull/137129