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: Incorrect bitfield aliasing with Tree SSA


Eric Botcazou writes:
> struct tree_struct_field_tag GTY(())
> {
>   struct tree_memory_tag common;
> 
>   /* Parent variable.  */
>   tree parent_var;
> 
>   /* Offset inside structure.  */
>   unsigned HOST_WIDE_INT offset;
> 
>   /* Size of the field.  */
>   unsigned HOST_WIDE_INT size;
> 
> };
> 
> And use parent_var to fetch the alias set somehow?

I thought of this but then I rejected it because of the recursive
case.  Consider the ref a.b.c.  Even if 'b' was addressable you would
get the alias set of type of 'a' which would be more conservative than
it has to be.  But anyway if you don't think this is major issue then
we don't have to pay the price for it.

Adam


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]