Contended CAS on every buffer allocation #13
Comments
|
Actually, it's not that. FrameBufferAllocationTest is broken, as it leaves Singleton in invalid state. The next test that uses allocator (RequestChannel) crashes... That wasn't very obvious... |
|
Closing this, as I will leave ::get usage. |
ghost
closed this as 
facebook-github-bot
pushed a commit
that referenced
this issue
Feb 13, 2020
) Summary: Take the README.md from https://github.com/facebookexperimental/mononoke/blob/7ead0e29e41aec0771531a4650b6170bc1ff6566/README.md and apply it on Eden repo. Re-add the Cargo.toml file that declares Cargo workspace. Re-add fbcode_builder/getdeps manifest for Mononoke Pull Request resolved: facebook/sapling#13 Test Plan: ./build/fbcode_builder/getdeps.py build mononoke ./build/fbcode_builder/getdeps.py test mononoke Reviewed By: ahornby Differential Revision: D19833059 Pulled By: lukaspiatkowski fbshipit-source-id: fb37e13306c0b9969a7c4e52b05e1a66a577022f
This issue was closed.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
folly::Singleton::try_get, which is not available in the latest stable version of folly and incurs a contended CAS on every buffer allocation.folly::Singleton::getto fix build, but that change broke tests (why?) when compiling with CMake.get_weak().lock(), we again have a contended CAS on buffer allocation path. * For this reason, I'd vote to get rid of Singleton hack and inject a Singleton instance into frame serialisers.The text was updated successfully, but these errors were encountered: