[PATCH] D45878: [DEBUG INFO] Fixing cases where debug info (-g) causes changes in the program.

[Jonas Paulsson via Phabricator via llvm-commits]

jonpa added a comment.

In https://reviews.llvm.org/D45878#1082011, @mattd wrote:

> Hi jonpa,  thanks for working on all of these cases.  I had a similar fix for MachineSink.cpp;  see https://reviews.llvm.org/D45637 for details.  Feel free to either lift that code or I can commit it separately, pending approval.  I want to avoid duplicating effort, as I already have an accompany test for that logic as well.  My solution creates a routine around the logic for sinking DBG_VALUE instructions, as that same logic is also used in another routine in that file.


Interesting, I see you fixed the same post-RA issue as I found :-)

I don't really have an opinion, your patch looks nice to me, except I wonder what you think about moving this out of MachineSink into splice() per my discussion?


https://reviews.llvm.org/D45878