Re: Error in generic/397 test script?

[Eric Biggers <ebiggers@kernel.org>]

On Mon, Jan 20, 2025 at 02:20:06PM +0000, David Howells wrote:
> generic/429 can also hang:
> 
> 
> 	show_file_contents()
> 	{
> 		echo "--- Contents of files using plaintext names:"
> 		cat $SCRATCH_MNT/edir/@@@ |& _filter_scratch
> 		cat $SCRATCH_MNT/edir/abcd |& _filter_scratch
> 		echo "--- Contents of files using no-key names:"
> 		cat ${nokey_names[@]} |& _filter_scratch | _filter_nokey_filenames edir
> 	}
> 	...
> 	nokey_names=( $(find $SCRATCH_MNT/edir -mindepth 1 | sort) )
> 	printf '%s\n' "${nokey_names[@]}" | \
> 		_filter_scratch | _filter_nokey_filenames edir
> 	show_file_contents
> 
> on the 'cat ...' at the end of show_file_contents().  A check that
> ${nokey_names[0]} is not nothing might be in order.
> 
> However, in this case (in which I'm running these against ceph), I don't think
> that the find should return nothing, so it's not a bug in the test script per
> se.

Similar to what I explained with generic/397, this can only happen if there is a
kernel bug.  Feel free to send a patch that updates the test to not hang in this
scenario, but the kernel bug (which is presumably in unmerged patches and not
yet upstream) will need to be fixed too.

- Eric