Hi All, I would like to propose a discussion on employing checksum offload in filesystems. It would be good to co-locate this with the storage track, as the finer details lie in the block layer and NVMe driver. For Btrfs, I had a fleeting chat with Joseph during the last LSFMM. It seemed there will be value in optimizing writes induced by the separate checksum tree. Anuj and I have developed an RFC. This may help us have a clearer discussion and decide the path forward. https://lore.kernel.org/linux-block/20250129140207.22718-1-joshi.k@xxxxxxxxxxx/ The proposed RFC maintains a generic infrastructure, allowing other filesystems to adopt it easily. XFS/Ext4 have native checksumming for metadata but not for data. With this approach, they could just instruct the SSD to checksum the data. But I am unsure if there are FS-specific trade-offs. Maybe that can also be up for the discussion.
