- User Since
- Sep 24 2015, 12:00 AM (313 w, 3 d)
Aug 24 2021
Hi @yaxunl! I'm working on upgrading a large codebase from LLVM-9 to LLVM-12. I noticed on average 10% compilation speed regression that seems to be caused this change. We use Clang modules and historically provide -fopenmp compiler flag by default. The problem seems to be that compiling and importing modules is now slower, with the generated modules size increased by 2X. llvm-bcanalyzer tool shows that it's dominated by DECLS_TO_CHECK_FOR_DEFERRED_DIAGS. If I understand it right, your change is only relevant when target offloading is used. I inspected all of #pragma omp directives and can confirm that we don't use it.
Jun 14 2021
Jun 11 2021
Feb 10 2020
Yes, I'll fix this!
@eugenis: would you want me to send a patch for it or are you looking at this?
Feb 4 2020
@eugenis: good! Please push this for me. Thank you!
Looks like FreeBSD was also fixed in:
commit 8a1f4feb1b749529ef2349bb7879a38b8d31f5f5 Author: Dimitry Andric <email@example.com> Date: Tue Feb 4 22:50:05 2020 +0100
include likely fixes for FreeBSD and OpenBSD.
Sorry for breaking NetBSD builds! An attempt to fix it in D73976.
Feb 3 2020
@eugenis, thank you for the review! Would you please commit this for me?
Thank you @eugenis! I've updated the change with you feedback and added a test.
Jan 31 2020
I manually tested this on CentOS and would need guidance about the use of SI_POSIX, as well as how to make a unittest for this (if needed).
Dec 19 2019
Aug 22 2019
Could that be due to the fact that I made this change from compiler-rt.git and not the monorepo?
Oct 31 2018
Great, let me create another iteration and add the -fsanitize-minimal flag, I'll see if I can split it up into small changes.
Oct 30 2018
@vitalybuka I was thinking of this too, we could rename -fsanitize-minimal-runtime flag, or convert it into something similar how -fsanitize= or -fsanitize-trap= works:
Hi @eugenis, I think I've implemented this as you suggested on the mailing list, please let me know if this makes sense.
Oct 29 2018
@eugenis, please push this for me!
Dec 21 2017
Nov 7 2017
@djasper: no, I do not.
Nov 6 2017
@djasper: ping :)
Nov 3 2017
@djasper: would you accept and push this for me :)
Nov 2 2017
Nov 1 2017
Updated per comments.
Oct 31 2017
May 4 2017
I didn't know about CMAKE_MODULE_LINK_FLAGS! Confirming that the following achieves the same result as the proposed diff:
cmake . . . -DCMAKE_MODULE_LINKER_FLAGS='-static-libgcc -static-libstdc++'
Sep 30 2016
Jun 7 2016
Thank you for the explanation @dvyukov! I don't remember exactly how gperftools was used in a project where I noticed the symbol name conflict. I'll try to reproduce it, and will post here for the record.
Thank you for the review! I need someone to commit this for me :)
Jun 6 2016
Feb 11 2016
I need someone to commit this for me.
Feb 10 2016
@joker.eph: yes, in environments that don't set CYGWIN, after r252652 libLTO is always built, regardless of LLVM_ENABLE_PIC.
Feb 9 2016
Feb 2 2016
@rsmith: yes, I need someone to commit this.