Re: [PATCH 1/2] sched/wait: Break up long wake list walk

[Tim Chen]

On 08/23/2017 04:30 PM, Linus Torvalds wrote:
> On Wed, Aug 23, 2017 at 11:17 AM, Linus Torvalds
> <torvalds@xxxxxxxxxxxxxxxxxxxx> wrote:
>> On Wed, Aug 23, 2017 at 8:58 AM, Tim Chen <tim.c.chen@xxxxxxxxxxxxxxx> wrote:
>>>
>>> Will you still consider the original patch as a fail safe mechanism?
>>
>> I don't think we have much choice, although I would *really* want to
>> get this root-caused rather than just papering over the symptoms.
> 
> Oh well. Apparently we're not making progress on that, so I looked at
> the patch again.
> 
> Can we fix it up a bit? In particular, the "bookmark_wake_function()"
> thing added no value, and definitely shouldn't have been exported.
> Just use NULL instead.
> 
> And the WAITQUEUE_WALK_BREAK_CNT thing should be internal to
> __wake_up_common(), not in some common header file. Again, there's no
> value in exporting it to anybody else.
> 
> And doing
> 
>                 if (curr->flags & WQ_FLAG_BOOKMARK)
> 
> looks odd, when we just did
> 
>                 unsigned flags = curr->flags;
> 
> one line earlier, so that can be just simplified.
> 
> So can you test that simplified version of the patch? I'm attaching my
> suggested edited patch, but you may just want to do those changes
> directly to your tree instead.

These changes look fine.  We are testing them now.
Does the second patch in the series look okay to you?

Thanks.

Tim