[cfe-commits] [PATCH] Fixing bug for function calls w/ diff addr spaces

[Anton Lokhmotov]

Clang freshly built from r158193 without the patch applied happily compiles
the test case.  The bug must have been fixed elsewhere...

Anton.

> -----Original Message-----
> From: Anton Lokhmotov [mailto:Anton.Lokhmotov at arm.com]
> Sent: 15 June 2012 15:43
> To: 'lattner at apple.com'
> Cc: 'cfe-commits at cs.uiuc.edu'
> Subject: RE: [cfe-commits] [PATCH] Fixing bug for function calls w/
> diff addr spaces
> 
> 
> 
> > -----Original Message-----
> > From: Anton Lokhmotov [mailto:Anton.Lokhmotov at arm.com]
> > Sent: 15 June 2012 15:34
> > To: 'lattner at apple.com'
> > Cc: cfe-commits at cs.uiuc.edu
> > Subject: Re: [cfe-commits] [PATCH] Fixing bug for function calls w/
> > diff addr spaces
> >
> > Hi Tanya,
> >
> > The code itself looks good to me.  But I'm worried why Clang thinks
> > that dereferencing a pointer to a struct results in a value in the
> > address space that the pointer points to.  A value is just a value,
> it
> > shouldn't live in a particular address space:
> >
> > __kernel void entry(__global int * buffer)
> > {
> >   const size_t gid = get_global_id();
> >   int value = buffer[gid] * 2; // no address space
> >   buffer[gid] = value;
> > }
> >
> > Can someone explain this please? (Assume that I know nothing - it's
> not
> > that far from truth!)
> >
> > Anton.