
Introduced back in Linux 6.17 was FALLOC_FL_WRITE_ZEROES as a fallocate() flag for efficiently zeroing out a specified file or range. Via NVMe Deallocate or similar SCSI standards can allow for offloading the zeroing out of the file/range for better efficiency and performance over user-space manually zeroing out or pre-initializing a file.
More details on the FALLOC_FL_WRITE_ZEROES functionality can be found in the original article on the topic, Linux 6.17 Will Allow Writing Zeroes More Efficiently To SSDs . When merged last year the EXT4 file-system was the first one implementing FALLOC_FL_WRITE_ZEROES support while now XFS is ready to go with Linux 7.3.
The rest of the XFS updates for Linux 7.3 are routine bug fixes and code improvements.