This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH] Provide debug info for SRA optimized variables (PR debug/43983)
On Fri, May 14, 2010 at 11:10:39AM +0200, Manuel LÃpez-IbÃÃez wrote:
> On 14 May 2010 10:35, Jakub Jelinek <jakub@redhat.com> wrote:
> >
> > it does. ÂPerhaps SRA should instantiate even the uninitialized fields?
>
> Are we currently avoiding this on purpose? Why?
>
We try to avoid useless replacement in order not to create useless
load/store statements and this one happened to be considered useless
too. Fortunatly the code specifically deletes the load from s.i
because it knows it cannot contain anything meaningful and so I hope
to insert creation of an SSA name that would trigger the warning
there.
Martin