- User Since
- Feb 23 2020, 3:46 PM (59 w, 1 d)
Updating D100334: [CSSPGO] Exclude pseudo probes from slot index
Fri, Apr 9
Updating D100235: [CSSPGO][llvm-profgen] Fixing an obselete iterator issue.
Thu, Apr 8
Thanks for fixing the name volatileness issue which was overlooked previously.
Wed, Apr 7
Removing removeRedundantPseudoProbes calls.
Updating D100075: [CSSPGO] Unblock optimizations with pseudo probe instrumentation part 2.
Fixing test issue.
Tue, Apr 6
Mon, Apr 5
Fri, Apr 2
Thanks for reporting this issue. I'm wondering if you have tried D99547 which fixed a similar non-determinism issue.
Thu, Apr 1
Tue, Mar 30
Mon, Mar 29
Addressing comments from Wei and Wenlei.
Moving call graph build into profiledCallGraph.h.
Sun, Mar 28
Addressing Wenlei's and Wei's comment.
Sat, Mar 27
Thanks for trying it. You’ll also need to apply Wenlei’s patch D99146.
Fri, Mar 26
Addressing Wenlei's comments.
Thu, Mar 25
LGTM, thanks for the fix.
Wed, Mar 24
Mon, Mar 22
Sun, Mar 21
Sat, Mar 20
Using PSI looks good, thanks. Can you please add a test for -csprof-merge-cold-context? Otherwise lgtm.
Fri, Mar 19
Thu, Mar 18
Addresssing David's comment.
Thanks for fixing the assert failure!
Thanks for working on this. This gets us closer to a global inlining decision making for thinLTO.
Wed, Mar 17
Thanks for the suggestion. I was thinking about capping on function size so that small callees such as getters/setters can always be inlined. What do you think?