Greater than a decade in the past, in 2012, Microsoft launched its 64-bit exFAT (Prolonged File Allocation Desk) file system because the successor to 32-bit FAT32. Seven years later, in 2019, the Redmond large made an enormous revelation when it introduced assist for the exFAT format in Linux.
Quick ahead a decade, in 2022, Sony engineer Yuezhang Mo discovered that lowering repeated traversal of listing entries improved the efficiency of exFAT by as much as 57%. This was pushed in Linux kernel model 6.2.
Mo is again to serving to exFAT once more, and this time, the newest pull request accommodates a bunch of exFAT-related enhancements, and a efficiency increase is one in all them. He writes:
Description for this pull request:
- Repair the random stack corruption and incorrect error returns in exfat_get_block().
- Optimize exfat_get_block() by enhancing checking nook instances.
- Repair an infinite loop by self-linked chain in exfat_find_last_cluster.
- Take away useless EXFAT_CLUSTERS_UNTRACKED codes.
- Add lacking shutdown examine.
- Enhance the delete efficiency with discard mount choice.
For these questioning, the “discard mount” choice refers back to the operation of informing the underlying storage system about blocks which are not in use. So, when a file system is mounted with the discard choice, it permits real-time discard operations. These operations routinely notify the storage system every time blocks transition from “used” to “free.”
The patch explains:
If the discard mount choice is enabled, the file’s clusters are discarded when the clusters are freed. Discarding clusters one after the other will considerably cut back efficiency. Poor efficiency might trigger delicate lockup when a lot of clusters are freed.
This commit improves efficiency by discarding contiguous clusters in batches.
The efficiency enchancment was measured utilizing an 80 GB file, and it took 4 minutes and 46 seconds or near 287 seconds to delete that. With the brand new optimization, the duty was accomplished in underneath 2 seconds, which equates to an enchancment of 172 occasions:
Measure the efficiency by:
# truncate -s 80G /mnt/file
# time rm /mnt/file
With out this commit:
- actual 4m46.183s
- consumer 0m0.000s
- sys 0m12.863s
With this commit:
- actual 0m1.661s
- consumer 0m0.000s
- sys 0m0.017s
You’ll be able to view the pull request right here on the Linux Kernel Mailing Listing (LKML) web site.
