This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: PR debug/13367

> On Mon, Dec 29, 2003 at 02:28:45PM +0100, Jan Hubicka wrote:
> > the problem is caused by fact that we inline even tought we are not
> > supposed to inline at all in non-optimizing compilation.
> 
> False.  The always_inline attribute forces inlining even at -O0.
> This is required for various altivec/sse interface functions to
> work properly.

I've tested this behaviour on older version of GCC that didn't but I see
that GCC 3.3 really inline this way.

SSE interface does not require it.  When compiled without inlining it
still work, just all the operations are done via function call, but I
see your point.  I will fix it accordingly then.  (making
cgraph_possibly_inlined_p to return true for always_inline functions)

Thanks for noticing this.
Honza
> 
> 
> r~
Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]