Btrfs Sees Minor Performance Optimizations With Linux 6.12
The Btrfs changes for Linux 6.12 have been merged already and includes ongoing code refactoring and clean-ups. Plus there are some minor performance optimizations and fixes along with more folio API conversions.
Btrfs with Linux 6.12 now wakes up the cleaner thread in the SYNC ioctl so that it can potentially speed-up the cleaning of deleted sub-volumes, error reporting improvements, preparations for compression in sub-page mode, and other changes.
Btrfs maintainer David Sterba commented on the Btrfs pull:
"This brings mostly refactoring, cleanups, minor performance optimizations and usual fixes. The folio API conversions are most noticeable.
There's one less visible change that could have a high impact. The extent lock scope for read is reduced, not held for the entire operation. In the buffered read case it's left to page or inode lock, some direct io read synchronization is still needed.
This used to prevent deadlocks induced by page faults during direct io, so there was a 4K limitation on the requests, e.g. for io_uring. In the future this will allow smoother integration with iomap where the extent read lock was a major obstacle."
Nice incremental improvements continuing to head into Btrfs.
24 Comments
