[llvm-commits] [test-suite] r155752 - in /test-suite/trunk/SingleSource/UnitTests/Vector: constpool.c constpool.reference_output

David Blaikie dblaikie at gmail.com
Fri May 4 11:22:09 PDT 2012


On Fri, May 4, 2012 at 10:50 AM, Jakob Stoklund Olesen <stoklund at 2pi.dk> wrote:
>
> On May 4, 2012, at 10:43 AM, David Blaikie <dblaikie at gmail.com> wrote:
>
>> On Fri, May 4, 2012 at 10:18 AM, Jakob Stoklund Olesen <stoklund at 2pi.dk> wrote:
>>>
>>> Does r156181 help?
>>
>> That fixes the first 4, but not the other 4000 after it in the same
>> file (sorry I didn't make this clear in my original email). (I've been
>> running this from buildbots so I haven't confirmed specifically, but
>> that's my guess - I'm doing a local test-suite run too, just to
>> confirm)
>
> The other tests in SingleSource/UnitTests/Vector are using the same syntax.

I'm not sure which syntax you're referring to - but I couldn't find
any other tests using [] on vector types. They all seem to go through
a union like the changes you made to the start of constpool.c - but
unlike the body of the 'f' function in constpool.c that's still using
[] on vector types which GCC 4.4 doesn't seem to allow.

[and my local repro using the test-suite Makefiles doesn't quite work
because I had clang in my path so that was picked up as the host
compiler - reconfiguring now. But manually compiling the current ToT
constpool.c still produces 4001 of those "subscripted value is neither
array nor pointer" errors - and none of the other .c files in that
directory do that, only a few warnings]



More information about the llvm-commits mailing list