After two months of development, Linus Torvalds presented the release of the nucleus linux 6.13 . Among the most noticeable changes: the regime of lazy displacement of tasks in the planner, support for the atomic entry in XFS and EXT4, the MultigRain TimeStamps mechanism, the adaptive mode of turning the polying in the AUTOFDO optimization, support for the ARM65 Guarded Control. Separate stacks in BPF, removal of Reiserfs, mounting mode of TMPFS accounting of the register register, support for POSIX expanding in SMB3.
The new version adopted 14172 corrections of 2086 developers,
The size of the patch is 46 MB (the changes affected 15375 files, 598707 lines of code were added, 406294 lines were removed). In the last issue, there were 14607 corrections from 2167 developers, the size of the patch was 37 MB. About 52% of all presented at 6.13
changes are associated with devices drivers, about 13% of the changes have
Attitude to updating the code specific to hardware architectures, 11%
associated with a network stack, 4% – with file systems and 3% with internal
nucleus subsystems.
the main innovation In the nucleus 6.13:
- disk subsystem, input/output and file systems
- added mechanism “< a href=”https://docs.kernel.org/filesystems/multigrain-ts.html”> multigrain timeStamps , which allows you to receive information about the time of change or access to files with more than a millisecond for accuracy, but without negative impact on performance. Increasing the accuracy of the labels leads to additional overhead costs due to an increase in the intensity of metadata recording to the disk, therefore, in the proposed implementation, more accurate marks are not created for all files, but only for those for which the processes request such marks through the Getattr () call.
- Added support , in which the data size exceeds the size of the sector, is carried out atomarly on storage devices providing A similar opportunity. Currently, the atomic recording is implemented for XFS, EXT4 in O_DIRECT (Direct I/O) and MD RAID 0/1/10.
- proposed new mechanism of calculation links for files, providing an increase in performance by 3-5% in working loads having more than 255 flows.
- remote File system linging Reiserfs, which was outdated in the year before last.
- Added