On Thu, Nov 03, 2022 at 11:23:21AM +0100, Andrey Albershteyn wrote: > On Wed, Nov 02, 2022 at 03:41:38PM -0700, Eric Biggers wrote: > > Thank you for reporting this. How did you notice this? This test actually is > > currently passing, because there is another test bug that makes > > _fsv_have_hash_algorithm always return false here. That really needs to be > > fixed first, as otherwise your fix doesn't really do anything. > > Hmm, it fails for me. _fsv_have_hash_algorithm also seems to work > (returns 0), what bug do you mean? Don't see anything obvious. > > I was running -g verity on ext4/xfs (I'm working on adding fsverity > support to xfs). That's exciting that you're working on fsverity support for xfs! I wasn't aware that someone was working on that. If the test is failing even on ext4 for you, then you must have changed something, either in the kernel or in the tests, that exposed the issue. Anyway, for _fsv_have_hash_algorithm() to work properly, it needs to set the fs.verity.require_signatures sysctl to 0 temporarily (if it was 1). I'll send out a patch if you don't get to it first. - Eric
