Date:   Thu, 2 Feb 2023 12:54:12 -0300
From:   Marcelo Tosatti <mtosatti@redhat.com>
To:     Christoph Lameter <cl@gentwo.de>
Cc:     Aaron Tomlin <atomlin@atomlin.com>,
        Frederic Weisbecker <frederic@kernel.org>,
        Andrew Morton <akpm@linux-foundation.org>,
        linux-kernel@vger.kernel.org, linux-mm@kvack.org
Subject: Re: [PATCH 3/5] mm/vmstat: use cmpxchg loop in cpu_vm_stats_fold
Message-ID: <Y9vcpO5Dh2G3Y6ib@tpad>
References: <20230201195013.881721887@redhat.com>
 <20230201195104.460373427@redhat.com>
 <fb371ac-cf69-28a6-eca0-6caccda4e0b8@gentwo.de>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <fb371ac-cf69-28a6-eca0-6caccda4e0b8@gentwo.de>
Precedence: bulk

On Thu, Feb 02, 2023 at 03:38:58PM +0100, Christoph Lameter wrote:
> On Wed, 1 Feb 2023, Marcelo Tosatti wrote:
> 
> > In preparation to switch vmstat shepherd to flush
> > per-CPU counters remotely, use a cmpxchg loop
> > instead of a pair of read/write instructions.
> 
> You are mixing full atomic cmpxchg and  per cpu atomic cmpxchg? That does
> not work.

OK, missing locked on the local functions. Can fix that.

> I thought you would only run this while the kernel is not active on the
> remote cpu? Then you dont need any cmpxchg and you can leave the function
> as is.

The remote cpu can enter kernel mode while this function executes.

There is no mode which indicates userspace cannot enter the kernel.