[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20220503114225.GA2799075@lothringen>
Date: Tue, 3 May 2022 13:42:25 +0200
From: Frederic Weisbecker <frederic@...nel.org>
To: Mark Rutland <mark.rutland@....com>
Cc: LKML <linux-kernel@...r.kernel.org>,
Peter Zijlstra <peterz@...radead.org>,
Phil Auld <pauld@...hat.com>,
Alex Belits <abelits@...vell.com>,
Nicolas Saenz Julienne <nsaenz@...nel.org>,
Xiongfeng Wang <wangxiongfeng2@...wei.com>,
Neeraj Upadhyay <quic_neeraju@...cinc.com>,
Thomas Gleixner <tglx@...utronix.de>,
Yu Liao <liaoyu15@...wei.com>,
Boqun Feng <boqun.feng@...il.com>,
"Paul E . McKenney" <paulmck@...nel.org>,
Marcelo Tosatti <mtosatti@...hat.com>,
Paul Gortmaker <paul.gortmaker@...driver.com>,
Uladzislau Rezki <uladzislau.rezki@...y.com>,
Joel Fernandes <joel@...lfernandes.org>
Subject: Re: [PATCH 10/21] context_tracking: Take idle eqs entrypoints over
RCU
On Tue, May 03, 2022 at 12:02:51PM +0100, Mark Rutland wrote:
> Hi Frederic,
>
> On Tue, May 03, 2022 at 12:00:40PM +0200, Frederic Weisbecker wrote:
> > The RCU dynticks counter is going to be merged into the context tracking
> > subsystem. Start with moving the idle extended quiescent states
> > entrypoints to context tracking. For now those are dumb redirections to
> > existing RCU calls.
>
> I was a bit confused looking at this, because that redirection only exists for
> CONFIG_CONTEXT_TRACKING, and is empty otherwise.
>
> I see this patch makes TREE_RCU select CONTEXT_TRACKING, which means that
> works. Since that also means building the rest of the context tracking code, I
> think it'd be worth mentioning that in the commit message.
Not exactly, it only builds a minimal version of CONTEXT_TRACKING, which maps
to roughly the vanilla RCU-idle code.
The user tracking part (which is the vanilla CONTEXT_TRACKING) now becomes
CONFIG_CONTEXT_TRACKING_USER
> Do all architectures which can use TREE_RCU today already support context
> tracking? If not, do those work by default?
No need to, the old context tracking becomes CONFIG_CONTEXT_TRACKING_USER
which is not needed by TREE_RCU.
Ok you're right, this is all very confusing :o)
Let's clarify this, I'm going to do the following on the next version:
* The current CONFIG_CONTEXT_TRACKING becomes CONFIG_CONTEXT_TRACKING_USER.
This part needs support from the architecture.
* The RCU-idle functions that have been moved to context tracking will become
CONFIG_CONTEXT_TRACKING_IDLE. This part doesn't need support from the
architecture and will be set by TREE_RCU.
Would that look clearer?
Thanks.
Powered by blists - more mailing lists