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] |
On 05/24/2010 08:15 PM, Jeff Law wrote:OK. Then the trivial initialization to zero should be safe. Thanks for the confirmation about the behaviour of try_fit_stack_local.
On 05/24/10 12:05, Sebastian Pop wrote:The last two? It's designed so that try_fit_stack_local must always
On Mon, May 24, 2010 at 11:18, Jeff Law<law@redhat.com> wrote:Right. And thus what should be the behaviour of the two callers that
Or is there another execution path that's causing the problem?
I think that at -O3 the function try_fit_stack_local gets inlined, and that initializes poffset only at the very bottom, and other cases that return false do not initialize it. I think that explains why at -O2 it does not triggers the uninitialized variable warning.
don't check the return value of try_fit_stack_local when it return
false? Or are we sure it can never return false from those calls?
succeed when we get there.
Thanks, Jeff
Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
---|---|---|
Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |