Re: [PATCH v2 6/8] mm/lru: remove rcu_read_lock to fix performance regression

[Mel Gorman]

On Wed, Nov 13, 2019 at 10:40:58AM +0800, Alex Shi wrote:
> 
> 
> ?? 2019/11/12 ????10:38, Matthew Wilcox ????:
> > On Tue, Nov 12, 2019 at 10:06:26PM +0800, Alex Shi wrote:
> >> Intel 0day report there are performance regression on this patchset.
> >> The detailed info points to rcu_read_lock + PROVE_LOCKING which causes
> >> queued_spin_lock_slowpath waiting too long time to get lock.
> >> Remove rcu_read_lock is safe here since we had a spinlock hold.
> > Argh.  You have not sent these patches in a properly reviewable form!
> > I wasted all that time reviewing the earlier patch in this series only to
> > find out that you changed it here.  FIX THE PATCH, don't send a fix-patch
> > on top of it!
> > 
> 
> Hi Matthew,
> 
> Very sorry for your time! The main reasons I use a separate patch since a, Intel 0day asking me to credit their are founding, and I don't know how to give a clearly/elegant explanation for a non-exist regression in a fixed patch. b, this regression is kindly pretty tricky.  Maybe it's better saying thanks in version change log of cover-letter?
> 

Add something like this to the patch

[lkp@xxxxxxxxx: Fix RCU-related regression reported by LKP robot]
Signed-off-by: Alex Shi <alex.shi@xxxxxxxxxxxxxxxxx>
...

-- 
Mel Gorman
SUSE Labs