During LSFMM we have discussed how to test lower-backend of linux IO-stack. Main idea was that xfstests is the most obvious solution which cover most use-cases filesystem do care about. There were reasonable objections that xfstests is not ideal place for that, especially when they mixed with pure FS specific ones. My opinion is that it is worth to have blkdev tests here, in xfstests, instead of not have it at all. But tests have to be placed to separate namespace "tests/blockdev" to prevent confusion. Please review my RFC and tell me: Is my proposal too ugly? And we have to have completely separate framework for basic blockdev sanity checks. Assumptions: - All new feature and regression tests should goes to "tests/blockdev" - All tests belongs to 'blockdev' group - Pure blkdev tests (tests which behavior is independent from fs) is not added to 'auto' group. So it will not affect normal fs testing flow. TOC: 0001-add-lio-target-helpers.patch 0002-add-blockdev-001-check-page-cache-coherency-after-BLKDISCARD 0003-new-blockdev-002-check-information-leak-for-lio-file 0004-new-blockdev-003-basic-blockdev-T10-DIF-TYPE1-integr 0005-new-blockdev-004-Check-that-live-fs-survives-blkdev 0006-new-blockdev-005-Check-that-busy-fs-survives-blkdev