[Lldb-commits] [PATCH] D58678: Improve step over performance by not stopping at branches that are function calls and stepping into and them out of each one

Jim Ingham via Phabricator via lldb-commits lldb-commits at lists.llvm.org
Thu May 9 18:03:29 PDT 2019


jingham added a comment.

The one outstanding bit of work here is that this change requires that the MSInst "IsCall" function has to mean "will return to the next instruction after call" or we might lose control of the program.  It seems obvious that that SHOULD be what it means, but we need to make sure that is always going to be what it means or we risk losing control of the program.  Greg is going to follow up on that.

If we have that assurance then this is a great change, a little because it avoids the extra stop and start and even more because it means we don't have to be so disciplined about never doing any work when we newly arrive in a frame.  I've had to squash bugs where we start to get debug info for a function we have no intention of stopping in, which can really slow down stepping in a big program to no purpose.


Repository:
  rLLDB LLDB

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D58678/new/

https://reviews.llvm.org/D58678





More information about the lldb-commits mailing list