[llvm] [InstCombine] Call InstSimplify for cast instructions (PR #162849)

[Nikita Popov via llvm-commits]

================
@@ -11,10 +11,8 @@ define i16 @test5(i16 %A) !dbg !34 {
   call void @llvm.dbg.value(metadata i32 %C, metadata !37, metadata !DIExpression()), !dbg !41
 
   ; Preserve the dbg.value for the DCE'd 32-bit 'and'.
-  ;
-  ; The high 16 bits of the original 'and' require sign-extending the new 16-bit and:
   ; CHECK-NEXT: #dbg_value(i16 [[and]], [[C:![0-9]+]],
-  ; CHECK-SAME:    !DIExpression(DW_OP_LLVM_convert, 16, DW_ATE_signed, DW_OP_LLVM_convert, 32, DW_ATE_signed, DW_OP_stack_value)
+  ; CHECK-SAME:    !DIExpression(DW_OP_LLVM_convert, 16, DW_ATE_unsigned, DW_OP_LLVM_convert, 32, DW_ATE_unsigned, DW_OP_stack_value)
----------------
nikic wrote:

My reading of this DIExpression is that we're first extending to 32-bit and then truncation to 16-bit, in which case the signedness does not matter. But I'm not confident on that reading...

The difference in behavior is that we're directly removing a cast pair instead of converting it into a bitcast first, which has some special handling here, though I have no idea what it's doing: https://github.com/nikic/llvm-project/blob/087ad0bea2e991d6c6b546170994cb2d172a6261/llvm/lib/Transforms/InstCombine/InstCombineCasts.cpp#L169-L171

https://github.com/llvm/llvm-project/pull/162849