[parted-devel] [PATCH] [RFC] Do not automatically update GPT label in interactive mode

[Jim Meyering]

Petr Uzel <petr.uzel at suse.cz> wrote:
> Here is a simple shell script that illustrates the incorrect behavior.
> The patch I've sent before fixes the issue (at least for me ;) ).
>
> ============================
> I'd like to transform it to the form of parted-testsuite, but it will
> take me some time as I'm not familiar much with it - any hints would
> be highly appreciated ;)

Great.
Thanks for the example.
Here's how I'd do it, using only a backing file.
No need for lvm.
First I do it at the shell/command-line level:

  dev=file
  dd if=/dev/null of=$dev bs=1 seek=30M
  ./parted -s $dev mklabel gpt
  dd if=$dev of=before bs=512 count=1 skip=1 2>/dev/null
  ./parted -m -s $dev p

  # make the backing file larger.  one byte is enough
  printf x >> $dev

  ./parted -m -s $dev p
  Error: The backup GPT table is not at the end of the disk, as it should be...
  ...
  dd if=$dev of=after bs=512 count=1 skip=1 2>/dev/null
  cmp before after
  before after differ: byte 17, line 1
  [Exit 1]

Then I made that into a test like any of the others:
[I chose a low numbered name for the new file, because this is a
 very fundamental property of partition tables (no file system
 aspects involved) ]

In the process, I discovered a GPT-related bug that
would provoke a failed assertion.  So I fixed that.  Patch below.
Would you like to write the test case for *it*?
If so, please name it t0201-gpt.sh, since it'll be
very similar to the one below (which has a FIXME
comment I plan to address before pushing).

My plan for tomorrow:

  - apply your patch, and test a little
  - address FIXME comment in the test script below
  - ensure that the test passes
  - push your patch and the test-adding one

If you write the test to exercise my patch,
(or tell me you'd rather not, in which case I'll do it),
I'll push those two change sets, too.