Re: [GIT PULL] locking changes for v5.2

From: Waiman Long
Date: Tue May 07 2019 - 10:40:20 EST

On 5/6/19 3:43 PM, Ingo Molnar wrote:
> * Waiman Long <longman9394@xxxxxxxxx> wrote:
>> On 5/6/19 4:50 AM, Ingo Molnar wrote:
>>> Linus,
>>> Please pull the latest locking-core-for-linus git tree from:
>>> git:// locking-core-for-linus
>>> # HEAD: d671002be6bdd7f77a771e23bf3e95d1f16775e6 locking/lockdep: Remove unnecessary unlikely()
>>> [ Dependency note: this tree depends on commits also in the RCU tree,
>>> please disregard this pull request if you weren't able to pull the RCU
>>> tree for some reason. ]
>>> Here are the locking changes in this cycle:
>>> - rwsem unification and simpler micro-optimizations to prepare for more
>>> intrusive (and more lucrative) scalability improvements in v5.3
>>> (Waiman Long)
>> Is it possible to pull in also my "locking/rwsem: Prevent decrement of
>> reader count before increment" patch for 5.2? The rests can wait until 5.3.
> Sure - how close is this to a straight:
> git revert 70800c3c0cc5
> ?
> If it's close enough then please resubmit this as a 'Revert "..."' patch,
> which I'll queue up in locking/urgent.
As explained by Linus, it is not a straight revert.
> It also is a performance, not a correctness fix, and should probably get
> a Cc: stable as well, right?

This patch is not for performance. It is fixing a regression and it does
have a cc: stable tag.

Thanks you for your help as I would like to backport the fix downstream.