Hello, Linus.

On Fri, Mar 25, 2011 at 10:02:48AM -0700, Linus Torvalds wrote:

Ok, guys, can you verify my merge? In one branch, we had Tejun change
it to use "alloc_workqueue()", and in another, ext4 switched from
create_workqueue() to create_singlethread_workqueue().

My resolution was to use WQ_UNBOUND in allow_workqueue() in the
resulting merge, which I think should be semantically the correct
resolve. But the commit message in the create_singlethread_workqueue()
change seems to imply that the single-threadedness isn't a
_correctness_ issue as much as it is just a "we don't need multiple
threads" issue, so maybe the WQ_UNBOUND isn't necessary. So WQ_UNBOUND
may or may not be a good idea.

Regardless, please give it a look, ok?

The merg should be safe but WQ_UNBOUND isn't necessary there, so I
think the version from commit fd89d5f2030a ("ext4: convert to
alloc_workqueue()") would be better.  BTW, Stephen already spotted
this earlier today - https://lkml.org/lkml/2011/3/24/652

Thanks.

-- 
tejun