On 08/18/2017 08:34 AM, Nikolay Borisov wrote:
>
>
> On 17.08.2017 23:59, Goffredo Baroncelli wrote:
>> Hi,
>>
>> On 08/17/2017 08:43 PM, Chris Murphy wrote:
>>> # btrfs sub create test1
>>> Create subvolume './test1'
>>> # btrfs sub create test1/test2
>>> Create subvolume 'test1/test2'
>>> # btrfs sub snap test1 test1.snap
>>> Create a snapshot of 'test1' in './test1.snap'
>>> # btrfs fi du -s test1
>>> Total Exclusive Set shared Filename
>>> 0.00B 0.00B 0.00B test1
>>> # btrfs fi du -s test1.snap
>>> Total Exclusive Set shared Filename
>>> ERROR: cannot check space of 'test1.snap': Inappropriate ioctl for device
>>> #
>>>
>>
>> tanks for the test case. Now I was able to reproduce the problem. The bug(s)
>> are two:
>>
>> 1) to get the treeid of the files/directory, the function
>> lookup_path_rootid() is used, which behaves strangely when it is called on a
>> directory that is BTRFS_EMPTY_SUBVOL_DIR_OBJECTID. In fact this function is
>> commented as following:
>> [....]
>> /*
>> * For a given:
>> * - file or directory return the containing tree root id
>> * - subvolume return its own tree id
>> * - BTRFS_EMPTY_SUBVOL_DIR_OBJECTID (directory with ino == 2) the result is
>> * undefined and function returns -1
>> */
>> int lookup_path_rootid(int fd, u64 *rootid)
>> {
>> [....]
>>
>> The caller (du_add_file()) doesn't consider this case.
>>
>> 2) in the function du_walk_dir(), an error returned by du_add_file() is
>> ignored, but the return value is not reset. So if the last entry has ino
>> BTRFS_EMPTY_SUBVOL_DIR_OBJECTID (directory with ino == 2) an error is
>> returned. But if there is another directory item no error is returned !
>>
>> See the following tests cases:
>>
>> # btrfs sub create test1
>> # btrfs sub create test1/test2
>> # btrfs sub snap test1 test1.snap
>> # btrfs fi du -s test1
>> Total Exclusive Set shared Filename
>> 0.00B 0.00B 0.00B test1
>> # btrfs fi du -s test1.snap
>> Total Exclusive Set shared Filename
>> ERROR: cannot check space of 'test1.snap': Inappropriate ioctl for device
>>
>> But if you add *another* file/dir in test1.snap you got:
>>
>> # mkdir test1.snap/dir
>> # btrfs fi du -s test1.snap
>> Total Exclusive Set shared Filename
>> 0.00B 0.00B 0.00B test1.snap
>>
>> The error disappea> Patches will follow shortly
>
> It would be awesome if you manage to introduce xfstests for this case
I am not sure if this is the right thing to do: the bugs are related to
a) an incorrect error handling of the function lookup_path_rootid()
b) the strange behavior of the BTRFS_EMPTY_SUBVOL_DIR_OBJECTID directory
For a) it is impossible to check for each lookup_path_rootid() call from a
xfstest; for b) I am quite sure that there are a lot of corner case not
properly handled in btrfs progs (and still btrfs fi-du handle it wrongly)
BR
G.Baroncelli
>
>>
>>>
>>> # uname -r
>>> 4.13.0-0.rc4.git1.1.fc27.x86_64
>>> # rpm -q btrfs-progs
>>> btrfs-progs-4.12-1.fc27.x86_64
>>>
>>>
>>> ----
>>> Chris Murphy
>>>
>>
>>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
> the body of a message to majord...@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
>
--
gpg @keyserver.linux.it: Goffredo Baroncelli <kreijackATinwind.it>
Key fingerprint BBF5 1610 0B64 DAC6 5F7D 17B2 0EDA 9B37 8B82 E0B5
--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
