Re: [PATCH 6/7] btrfs-progs: Add test for super block recovery

[David Sterba <dsterba@suse.cz>]

On Tue, Jan 23, 2018 at 05:29:57PM +0200, Nikolay Borisov wrote:
> On 23.01.2018 17:07, David Sterba wrote:
> > On Tue, Dec 05, 2017 at 12:04:49PM +0200, Nikolay Borisov wrote:
> >>>> --- /dev/null
> >>>> +++ b/tests/fsck-tests/029-superblock-recovery/test.sh
> >>>> @@ -0,0 +1,64 @@
> >>>> +#!/bin/bash
> >>>> +# Test that any superblock is correctly detected
> >>>> +# and fixed by btrfs rescue
> >>>> +
> >>>> +source "$TOP/tests/common"
> >>>> +
> >>>> +check_prereq btrfs
> >>>> +check_prereq mkfs.btrfs
> >>>> +check_prereq btrfs-select-super
> >>>> +
> >>>> +setup_root_helper
> >>>> +
> >>>> +rm -f dev1
> >>>> +run_check truncate -s 260G dev1
> >>>> +loop=$(run_check_stdout $SUDO_HELPER losetup --find --show dev1)
> >>>
> >>> We have function to do it already.
> >>> prepare_test_dev will use loopback device as fallback if $TEST_DEV is
> >>> not specified.
> >>> Tt can handle size well, and it also uses sparse file so no need to
> >>> worry about disk usage.
> >>
> >> Then the test suite is not very consistent, since I copied this loopback
> >> handling from some other test.
> > 
> > From which one? Some tests have special needs and may set up the loop
> > device on their own. The rest should use the helpers in tests/common and
> > I don't see anything that would not be provided by them.
> 
> Looking back at the code it seems it could have been from
> 021-image-multi-devices.

That was my guess too. There are now multi-loop device helpers, this
test has been added before that.

> Anyway, I haven't resubmitted the sequence since I was waiting for
> feedback. Qu already mentioned this could be refactored.

I've applied your patch without changes and updated it in another commit
as it changed more than just some trivial bits.