On Tue, Dec 04, 2018 at 10:29:49PM +0000, Chris Mason wrote: > I think (hope) this is: > > https://bugzilla.kernel.org/show_bug.cgi?id=201685 > > Which was just nailed down to a blkmq bug. It triggers when you have > scsi devices using elevator=none over blkmq.
Thanks a lot Chris. Really. Good news: I confirm I recompiled and used blkmq and no-op (at that time). Also, the massive write of btrfs defrag can explain the massive trigger of the bug, and next corruption. Thanks again, Andrea
