[Patch] Support for Nuxi CloudABI

[Jonathan Roelofs]

On 3/8/15 5:32 AM, Ed Schouten wrote:
> 2015-03-07 17:46 GMT+01:00 Eric Christopher <echristo at gmail.com>:
>> That's fine. Just reply with the patch with testcase for the record. :)
>
> Sure!
>
> Before Eric pointed me to the unit tests for Triple, I was under the
> impression that we only tested triples by having accidental testing
> coverage for them through the IR dumps we have in test/. My idea was
> therefore to at least add some coverage for CloudABI by adjusting one
> of the existing tests to use the CloudABI triple:
>
> Index: test/CodeGen/X86/dwarf-eh-prepare.ll
> ===================================================================
> --- test/CodeGen/X86/dwarf-eh-prepare.ll        (revision 231492)
> +++ test/CodeGen/X86/dwarf-eh-prepare.ll        (working copy)
> @@ -1,4 +1,4 @@
> -; RUN: opt -mtriple=x86_64-linux-gnu -dwarfehprepare < %s -S | FileCheck %s
> +; RUN: opt -mtriple=x86_64-unknown-cloudabi -dwarfehprepare < %s -S |
For future reference, it would be better in cases like this to duplicate 
the whole RUN line, changing the bits you're interested in (like the 
triple). That way there's still coverage for the original case, in 
addition to your new additions. You can even add multiple 
--check-prefix='s to a single FileCheck command for cases where the 
checks need to differ slightly in some places but are mostly the same 
otherwise.


Cheers,

Jon
> FileCheck %s
>
>   ; Check basic functionality of IR-to-IR DWARF EH preparation. This should
>   ; eliminate resumes. This pass requires a TargetMachine, so we put it under X86
>
> Further testing (after I sent the patch to Eric) revealed that this
> change makes little sense. LLVM doesn't really care if the triple you
> provide is valid at all. I used a couple of random OS names but none
> of them caused the test to fail.
>
> My proposal is to stick to the original patch I sent out two days ago,
> including the change to unittests/ADT/TripleTest.cpp. Attached is a
> combined patch.
>
>
>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>

-- 
Jon Roelofs
jonathan at codesourcery.com
CodeSourcery / Mentor Embedded