Jeff Law <l...@redhat.com> wrote:
>On 11/22/13 05:10, Richard Biener wrote:
>
>>> + if (totally_clobbered_loops)
>>> + {
>>> + /* Release the current loop structures, they are totally
>>> + clobbered at this point. */
>>> + loop_optimizer_finalize ();
>>> + current_loops = NULL;
>>
>> This is definitely a no-go and should be immediately reverted. If
>you
>> wreck a particular loop simply mark it for removal by setting
>> ->header and ->latch to NULL. Loop fixup will then re-discover
>> it (or really remove it).
>>
>> With the code above you discard all loops in the function including
>> meta-information on openmp loops, #pragma ivdeps info, etc.
>I realize it discards all that stuff, but that's still the safeest
>thing
>to do.
>
>If we can selectively discard, that seems like a follow-up item. The
>loop stuff is new and something I know little about. Thus I chose the
>safest route.
Well. Safe here is arguably a bit optimistic ;-)
>So the issue here is we can create irreducible regions & new nested
>loops. Does just setting the header,latch fields for the current loop
>handle those cases?
Yes.
Richard.
>jeff
