> > commit a666e5c05e7c ("dm: fix deadlock when swapping to encrypted device") > > Since there is no further information or clue for the deadlock, I assume there is swap I/O hitting encrypted device. And btw, the idea of limiting swap I/O on encrypted device looks interesting, it may also be helpful for Bug #1177529, which causes deadlock for swap I/O on dm-thin device. This patch breaks our KABI, let me do more KABI fix and testing before submitting it to our tree. Thanks. Coly Li