Re: 3.15 regression: wrong cgroup magic

[Andy Lutomirski <luto@amacapital.net>]

On Tue, Jun 3, 2014 at 10:23 AM, Linus Torvalds
<torvalds@linux-foundation.org> wrote:
> On Mon, Jun 2, 2014 at 6:30 PM, Linus Torvalds
> <torvalds@linux-foundation.org> wrote:
>>
>> Of course, Greg could just send it to me for my next branch (since the
>> merge window for 3.16 is already open) and tell me that it's also
>> stable material for 3.15. At _that_ point I'll happily cherry-pick it
>> into master...
>
> Ok, so that happened.
>
> Tejun, just to be anal, can you please test and verify current -git
> (preferably both 'master' and 'next') with your app that cared about
> the f_type information for statfs()..

Did you mean me?

My code seems to work now, and strace says:

fstatfs(6, {f_type=0x27e0eb, f_bsize=4096, f_blocks=0, f_bfree=0, f_bavail=0, f_
files=0, f_ffree=0, f_fsid={0, 0}, f_namelen=255, f_frsize=4096}) = 0

which is consistent with:

#define CGROUP_SUPER_MAGIC      0x27e0eb

This is on master.  I'll test next eventually.

Thanks everyone!

--Andy