Don't eliminate frame pointer at -O0
#2480
Labels
Comments
|
I agree, we shouldn't do frame-pointer elimination at -O0. (Is this possibly a consequence of a LLVM API change at some point?) Just to recap, in #2096 I was merely arguing for a better backtrace mechanism, rather than against disabling fp elim. |
kinke
added a commit
to kinke/ldc
that referenced
this issue
Jan 4, 2018
|
Related: #2222 |
kinke
added a commit
to kinke/ldc
that referenced
this issue
Jan 12, 2018
kinke
added a commit
that referenced
this issue
Jan 13, 2018
|
Fixed by #2483. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
For previous discussion on this topic: #2096
I think we should not eliminate the frame pointer at
-O0. It is not just our own stack traces that are (no longer) affected. ASan also does not show correct/complete stack traces with frame pointer elimination.Clang does not even eliminate the frame pointer at
-O3on macOS. We don't have to mimic that, but at-O0I feel we really should keep the fp.The text was updated successfully, but these errors were encountered: