[clang] [Clang][CIR] Add vqshlud and vqshld AArch64 builtins (PR #190728)

[Kartik Ohlan via cfe-commits]

Ko496-glitch wrote:

> > @banach-space i added the test cases to intrinsics.c however there is one concern. The instructions for test file was to use
> > ```
> > // LLVM-LABEL: @test_vduph_lane_bf16
> > // CIR-LABEL: @vduph_lane_bf16(
> > bfloat16_t test_vduph_lane_bf16(bfloat16x4_t v) {
> >   // CIR:
> > 
> >   // LLVM:
> >   return (...)
> > }
> > ```
> > 
> > 
> >     
> >       
> >     
> > 
> >       
> >     
> > 
> >     
> >   
> > but in my case i have the extra _test in CIR-LABEL, any suggestions on how i can fix this or is this fine ?
> > ```
> > 
> > // LLVM-LABEL: @test_vqshlud_n_s64(
> > // CIR-LABEL: @test_vqshlud_n_s64(
> > int64_t test_vqshlud_n_s64(int64_t a) {
> > ```
> 
> Just use this instead:
> 
> ```
> // ALL-LABEL: @test_vqshlud_n_s64
> ```
> 
> Basically, for some intrinsics, there is an indirection (i.e. one intrinsics calls some other intrinsics). and that's usually captured at the CIR level, hence often there is
> 
> * ` // CIR-LABEL: @vduph_lane_bf16(`
> 
> instead of e.g.
> 
> *`// CIR-LABEL: @test_vduph_lane_bf16(`).
> 
> But that is not always the case, as you can see in your example. Just look how the intrinsics that you are lowering is defined in the header file that's included in the test file. Hopefully this is helpful.

thanks! this was really helpfull

https://github.com/llvm/llvm-project/pull/190728