On Thu, Apr 08, 2021 at 11:49:37AM -0700, Boris Burkov wrote: > On Thu, Apr 08, 2021 at 11:41:42AM -0700, Eric Biggers wrote: > > On Thu, Apr 08, 2021 at 11:30:12AM -0700, Boris Burkov wrote: > > > > > > Note that there is a bit of a kludge here: since btrfs_corrupt_block > > > doesn't handle streaming corruption bytes from stdin (I could change > > > that, but it feels like overkill for this purpose), I just read the > > > first corruption byte and duplicate it for the desired length. That is > > > how the test is using the interface in practice, anyway. > > > > If that's the problem, couldn't you just write the data to a temporary file? > > Sorry, I was a bit too vague. It doesn't have a file or stdin interface, > as far as I know. > > btrfs-corrupt-block has your typical "kitchen sink of flags" interface and > doesn't currently read input from streams/files. I extended that > interface in the simplest way to support arbitrary corruption, which > didn't fit with the stream based corruption this test does. > > my options seem to be: > shoehorn the "byte, length" interface into this test or > shoehorn the "stream corruption input in" interface into > btrfs-corrupt-block. > > I have no problem with either, the former was just less work because I > already wrote it that way. If the junk I did here is a deal-breaker, I > don't mind modifying btrfs-corrupt-block. > If it's a lot of trouble to handle arbitrary data, then I think you should change _fsv_scratch_corrupt_merkle_tree() to actually take a (byte, length) pair instead of data on stdin. Otherwise, _fsv_scratch_corrupt_merkle_tree() would claim to do one thing but actually would do a different thing on one specific filesystem. - Eric
