On 4/15/12 9:46 PM, Eric Sandeen wrote:
> On 4/13/12 4:49 AM, tmarek@xxxxxxxxxx wrote:
>> From: Tom Marek <tmarek@xxxxxxxxxx>
>> Tests number 062 was supposed to work with ext4 fs but there were some
>> problems in it - Tests haven't considered existence of lost+found directory
>> ext4. Also when scratch was mounted with SELinux context test failed because
>> fgetattr returns SELinux extended attributes. And when fgetattr is run with
>> recursive flag it's output might change between file systems due to different
>> file ordering. Fix this by sorting the output manually. Also all lines
>> containing SELinux and lost+found were removed from output.
>> Signed-off-by: Tom Marek <tmarek@xxxxxxxxxx>
> Have you run the latest xfstests? This looks a lot like:
> commit 2fb1c931a6090f646afa44e4ce3f1f9815af9067
> Author: Eric Sandeen <sandeen@xxxxxxxxxxx>
> Date: Fri Jan 27 12:28:20 2012 -0600
> 062: Sort recursive getfattr output
> Test 062 was made "generic" a while back, but it fails on any filesystem
> which returns getfattr -R results (aka readdir results) in something
> other than inode-order.
> With a little awk-fu we can sort the records from getfattr -R so that
> the output is the same for xfs as well as ext4, etc.
> Also filter out lost+found which extN creates at mkfs time, but
> some other filesystems do not.
> Signed-off-by: Eric Sandeen <sandeen@xxxxxxxxxx>
> Reviewed-by: Christoph Hellwig <hch@xxxxxx>
Oh, and as for selinux, scratch should be getting mounted with an fs-wide
selinux context mount option, I think?