- User Since
- Jul 7 2012, 2:54 PM (311 w, 1 d)
Thu, Jun 21
I had to revert this because the embedded function analysis runner inside the legacy pass manager for module passes appears to leak. =/ At least we found LeakSanitizer reports on the bots because of this commit.
It'd be nice to get a test case here, but if its not feasible, the change is obviously correct (or at least an improvement!!!)
Actually include some of the new tests and fix minor bugs they found.
Ok folks, I know this is a blast from the past, but can we actually review this? I'm willing to commandeer it, rebase it, and get it landed if you all can review this and confirm that it is the right direction.
Address comments from review, beefing up tests along the way to cover this
stuff much more carefully.
Thanks for the comments! Some responses below...
Wed, Jun 20
Rebase now that all the prior patches are landed and ping.
Rebase and ping.
I think this looks really nice.
BTW, one question I have is what the best name for the file is here. function.h seems kinda lame, but not sure what else to call it. Suggestions very welcome here.
Still a few things left, but really close.
Really nice approach for THREADS=0 w/ the deferred future that checks that "enough" of the queue has been processed. I like it.
Mon, Jun 18
Some suggestions about the way you package things and manage std::future stuff below.
Sat, Jun 16
Before I dig in too much to the code, I want to better understand the underlying goal...
Fri, Jun 8
Thu, Jun 7
Wed, Jun 6
Thanks for the review! Based on LGTM, I'll land once the dependent patch lands. Give a shout if you still have concerns though.
Fix typo spotted in review.
This time *actually* inculde the updated comments in the header file that Hal requested. Soryr this were missing in the prior update.
Thanks for comments, should all be addressed
Update to address code review comments.
Tue, Jun 5
Add Hal to this review since I was discussing it with him.
Rebase and ping again. Really looking for review on this one as well.
Rebase and ping yet again. Would really like to get a review here...
Update with a new approach and some new testing.
I think a significantly different approach to invalidating this is necessary. Also, i'd like to much more directly test the SCEV updates. Let me create a new patch here and see what you think.
Sun, Jun 3
Fri, Jun 1
Rebase and ping.
Rebase and ping.
Thanks for looking at this!
Thu, May 31
Adding James as a reviewer since I know he's interested in this stuff...
Wed, May 30
Tue, May 29
Rebase now that all the loop-instsimplify stuff has landed.
Thanks all, landing!
All suggestions implemented, submitting now, thanks!
Move the new cleanup pass additions above the loop-rotate specific comment.
This position was the whole point of the other comment I added here....
Thanks for review all! (Waiting on the patch before this in order to land this one.)
Sat, May 26
Update to rebase on the patch that adds loop-instsimplify and loop-simplifycfg
to the loop pass pipeline as cleanup passes to minimize the cleanup necessary
in the unswitch pass itself.
Rebase and teach the old PM to do this as well when using SimpleLoopUnswitch.
Fix some formatting goofs that snuck in w/o a run of clang-format to clean them up.
May 25 2018
Thanks for the ping!
May 24 2018
We have an MI combine pass IIRC -- why not put this there? (Genuine question, haven't looked at these enough to know if that doesn't work...)
Updated patch just to show where the code ended up.
Ok, after some discussion, I'm pretty convinced we just need to stop simplifying here.
Thanks for the review!
Update with fixes from code review.
May 23 2018
FWIW, I don't think this is really obvious / trivial enough to land w/o review by someone familiar w/ Orc and the users.