Re: [PATCH v7 7/6] fs/epoll: scale nested callbacks

To: Jason Baron <jbaron@xxxxxxxxxx>
Subject: Re: [PATCH v7 7/6] fs/epoll: scale nested callbacks
From: Davidlohr Bueso <dave@xxxxxxxxxxxx>
Date: Wed, 18 Oct 2017 08:44:36 -0700
Cc: Waiman Long <longman@xxxxxxxxxx>, akpm@xxxxxxxxxxxxxxxxxxxx, Alexander Viro <viro@xxxxxxxxxxxxxxxxxx>, Jan Kara <jack@xxxxxxxx>, Jeff Layton <jlayton@xxxxxxxxxxxxxxx>, "J. Bruce Fields" <bfields@xxxxxxxxxxxx>, Tejun Heo <tj@xxxxxxxxxx>, Christoph Lameter <cl@xxxxxxxxxxxxxxxxxxxx>, linux-fsdevel@xxxxxxxxxxxxxxx, linux-kernel@xxxxxxxxxxxxxxx, Ingo Molnar <mingo@xxxxxxxxxx>, Peter Zijlstra <peterz@xxxxxxxxxxxxx>, Andi Kleen <andi@xxxxxxxxxxxxxx>, Dave Chinner <dchinner@xxxxxxxxxx>, Boqun Feng <boqun.feng@xxxxxxxxx>
In-reply-to: <77a2bec3-f663-40de-9ad8-949bf3230a76@akamai.com>
User-agent: NeoMutt/20170421 (1.8.2)

On Wed, 18 Oct 2017, Jason Baron wrote:

If there are not many EPOLL_CTL_ADDs, then I wouldn't think loop_ncalls
would be highly contented (since it should only be called from the add
path)?


So yeah it's all about the readywalks.

Thanks,
Davidlohr

References:
- [PATCH v7 0/6] vfs: Use dlock list for SB's s_inodes list
  - From: Waiman Long
- [PATCH v7 7/6] fs/epoll: scale nested callbacks
  - From: Davidlohr Bueso
- Re: [PATCH v7 7/6] fs/epoll: scale nested callbacks
  - From: Jason Baron
- Re: [PATCH v7 7/6] fs/epoll: scale nested callbacks
  - From: Davidlohr Bueso
- Re: [PATCH v7 7/6] fs/epoll: scale nested callbacks
  - From: Jason Baron