[llvm] [GlobalISel] Support saturated truncate (PR #150219)

[David Green via llvm-commits]

================
@@ -1,6 +1,8 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc < %s -mtriple=aarch64 | FileCheck %s --check-prefixes=CHECK,CHECK-CVT
-; RUN: llc < %s -mtriple=aarch64 -mattr=+fullfp16 | FileCheck %s --check-prefixes=CHECK,CHECK-FP16
+; RUN: llc < %s -mtriple=aarch64 -global-isel=0 | FileCheck %s --check-prefixes=CHECK,CHECK-SD,CHECK-CVT-SD
+; RUN: llc < %s -mtriple=aarch64 -mattr=+fullfp16 -global-isel=0 | FileCheck %s --check-prefixes=CHECK,CHECK-SD,CHECK-FP16-SD
+; RUN: llc < %s -mtriple=aarch64 -global-isel=1 | FileCheck %s --check-prefixes=CHECK,CHECK-GI,CHECK-CVT-GI
+; RUN: llc < %s -mtriple=aarch64 -mattr=+fullfp16 -global-isel=1 | FileCheck %s --check-prefixes=CHECK,CHECK-GI,CHECK-FP16-GI
----------------
davemgreen wrote:

Is it worth doing `--check-prefixes=CHECK,CHECK-CVT,CHECK-CVT-SD` etc, so that the GI and SD "CVT" code can be folded together if they are the same?

https://github.com/llvm/llvm-project/pull/150219