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 05/19/14 22:00, Maxim Kuvyrkov wrote:
I'd say #2. While we may not have any evidence those options aren't being used, we really wouldn't want to break a build which happened to use them.Hi, This patch removes msched-prefer-non-data-spec-insns and msched-prefer-non-control-spec-insns IA64 options and handling thereof. I have implemented these option during my youthful indiscrimination and, as far as I can tell, no one uses them. Yet, handling of these options complicate logic of haifa-sched.c:choose_ready(), and other functions. We have 3 choice for future of these options (of which I prefer the first one): 1. Remove options and their handling in haifa-sched and sel-sched. 2. Remove handling, but keep dummy options for the sake of ... someone ... I guess ? 3. Keep options, and reimplement their handling through targetm.sched.first_cycle_multipass_dfa_lookahead_guard hook -- requires some additional mental cycles on my part, therefore the least preferable option to me. Can we go with option (1), please? Tested on x86_64-linux-gnu and also build ia64-linux-gnu cc1.
Now if you happened to know that if those options were enabled that the compiler consistently aborted/segfaulted or the like, then, well, we could make the case that nobody could be reasonably using those options.
So I'd leave just the deprecated options. The removal of the support bits in the scheduler and ia64 port looks good. So pre-approved with the options left in as dummies.
Jeff
Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
---|---|---|
Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |