http://bugzilla.opensuse.org/show_bug.cgi?id=1052419 http://bugzilla.opensuse.org/show_bug.cgi?id=1052419#c7 --- Comment #7 from Peter Grandi <pg@suse.for.sabi.co.uk> --- I did not report an issue as to "all NFS writes" and your quote as to this seems to me to be malicious, I think it cannot be a mistake because a simple string search on this page shows that the first appearance of those words is by you. I have the reported the *possibility* of data updates after checksums are computed, that is as a result of unstable writes, in both the case of direct IO and the NFS kernel server. In the case of direct IO this was confirmed by Josef Bacik, as per the link in the Btrfs wiki, and in the case of the NFS kernel server it was mentioned by a user on IRC. Additionally as a rule any superficial analysis cannot easily prove (or disprove) the absence of race conditions leading to unstable writes, and however rare race conditions and unstable writes on busy servers can result in extensive data corruption. You have the authority to ignore my friendly report and make accusations of "FUD" that to me seem purely paranoid, and these remain on the record under your name. -- You are receiving this mail because: You are on the CC list for the bug.