On Thu, May 03, 2018 at 01:57:03PM -0700, Matthew Wilcox wrote: > I was running xfstests xfs/350 at the time. It's a superblock fuzzing test. Described as: tests/xfs/group:350 dangerous_fuzzers dangerous_scrub dangerous_repair Are you trying to run xfstests as a regression test suite (i.e. check -g auto) , or as a fuzzer to find on-disk format parsing issues (check -g dangerous_fuzzers)? > No kernel patches applied > beyond linux-next (I was hoping to get a baseline). > > 5726 XFS: Assertion failed: sbp->sb_rsumino != NULLFSINO, file: ../fs/xfs/xfs_rtalloc.c, line: 1223 And you're running a debug kernel, so assert failures are expected. Essentially, the fuzzed superblock has a non-null realtime bitmap inode, but a null realtime summary inode. Which on a non-debug kernel will result in mount failing, but on a CONFIG_XFS_DEBUG=y kernel it fires this assert.... We're slowly converting asserts like this to explicit checks that return -EFSCORRUPTED as we come across them - this is obviously code that hasn't been converted yet.... Cheers, Dave. -- Dave Chinner david@xxxxxxxxxxxxx -- To unsubscribe from this list: send the line "unsubscribe linux-xfs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
