[cfe-commits] [cfe-dev] Clang and OpenCL address spaces

John McCall rjmccall at apple.com
Mon Oct 18 13:44:56 PDT 2010


On Oct 18, 2010, at 1:38 PM, Peter Collingbourne wrote:

> On Mon, Oct 18, 2010 at 10:43:24AM -0700, John McCall wrote:
>> On Oct 18, 2010, at 10:33 AM, Peter Collingbourne wrote:
>>>> From my understanding of the OpenCL specification, setting the address
>>> space on an automatic variable is supposed to not only change the
>>> address space but also make the variable static (as the term is defined
>>> in standard C).  Adding 'static' to the qualifier macro definitions
>>> won't work too well because the qualifiers can also appear in other
>>> places.
>> 
>> If that's how OpenCL is specified to work, then changing the front-end to
>> automatically add 'static' to local variables of address-space-qualified type
>> is a very reasonable solution.  We would accept a patch for this.
> 
> Patch attached.  OK to commit?

Generally okay.  Two requests:
1) Please quote something appropriate from the OpenCL spec here.
2) This should almost certainly be conditioned on not having an explicit
  storage class;  if the user wrote 'auto' or 'register', we should not just ignore
  that (unless the spec says to, but that would be wierd).

John.



More information about the cfe-commits mailing list