we're recomputing this for a *OSRExitDescriptor* for every one of its *OSRExits*. This is having a multiplicative effect on offsets. I'm amazed this didn't break any of our tests. I think I introduced this in: https://bugs.webkit.org/show_bug.cgi?id=149970
Created attachment 264287 [details] patch
Nice catch!
Comment on attachment 264287 [details] patch Can haz regression test?
(In reply to comment #3) > Comment on attachment 264287 [details] > patch > > Can haz regression test? I'll try to see if I can come up with one before landing this. We will definitely have one once FTL try/catch lands but it will be nice if I can come up with a small and simple reproducible program. This might be a bit tricky to get right. If I can't come up with a small program, I think coming up with a small program for FTL try/catch will be easier because of the nature of OSR exits from from calls and get/put_by_ids.
(In reply to comment #4) > (In reply to comment #3) > > Comment on attachment 264287 [details] > > patch > > > > Can haz regression test? > I'll try to see if I can come up with one before landing this. > We will definitely have one once FTL try/catch lands but > it will be nice if I can come up with a small and simple > reproducible program. This might be a bit tricky > to get right. If I can't come up with a small program, > I think coming up with a small program for FTL try/catch > will be easier because of the nature of OSR exits from > from calls and get/put_by_ids. I couldn't get a test case to reproduce this easily. I'm going to defer until FTL try/catch where we definitely will have a test case.
Okeedokee.
Comment on attachment 264287 [details] patch Clearing flags on attachment: 264287 Committed r191744: <http://trac.webkit.org/changeset/191744>
All reviewed patches have been landed. Closing bug.