On 23/02/2021 04:00, Jérôme Carretero wrote: > Tested-by: Jérôme Carretero <cJ-ko@xxxxxxxxxxx> > > > I smoke-tested by fuzzing contiguous ranges of user data (not hash > tree) and correction works up to the amount of added correction (since > erasures are identified by hashes). Thanks for testing! > Since the fix looks obvious, this should probably CC to stable? Yes, I forgot to add the tag, I think this should go to all stable kernels. (It only needs dm_bufio_set_sector_offset() that was added later than FEC, IIRC.) Milan p.s. There are also bugs in userspace veritysetup if FEC has hash area is stored on the same device, then code miscalculates parity bytes then. This will be fixed soon in next stable cryptsetup release. -- dm-devel mailing list dm-devel@xxxxxxxxxx https://listman.redhat.com/mailman/listinfo/dm-devel
