Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Subject: Re: [PATCH v2] kvm: Replace vcpu->swait with rcuwait
To:     Davidlohr Bueso <dave@stgolabs.net>
Cc:     Marc Zyngier <maz@kernel.org>, tglx@linutronix.de,
        kvm@vger.kernel.org, Davidlohr Bueso <dbueso@suse.de>,
        peterz@infradead.org, torvalds@linux-foundation.org,
        bigeasy@linutronix.de, linux-kernel@vger.kernel.org,
        rostedt@goodmis.org, linux-mips@vger.kernel.org,
        Paul Mackerras <paulus@ozlabs.org>, joel@joelfernandes.org,
        will@kernel.org, kvmarm@lists.cs.columbia.edu
References: <20200324044453.15733-1-dave@stgolabs.net>
 <20200324044453.15733-4-dave@stgolabs.net>
 <20200420164132.tjzk5ebx35m66yce@linux-p48b>
 <418acdb5001a9ae836095b7187338085@misterjones.org>
 <20200420205641.6sgsllj6pmsnwrvp@linux-p48b>
 <f7cc83fe-3e91-0057-9af2-26c201456689@redhat.com>
 <20200420215014.sarodevmhphnkkn7@linux-p48b>
 <02e1b00d-a8ea-a947-bbe6-0b1380aa7ec4@redhat.com>
 <20200421180733.xrl5ta6cuo2weuva@linux-p48b>
From:   Paolo Bonzini <pbonzini@redhat.com>
Message-ID: <ab78bbc8-aa03-6e88-940e-5e1c041f48e4@redhat.com>
Date:   Tue, 21 Apr 2020 21:52:47 +0200
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101
 Thunderbird/68.6.0
MIME-Version: 1.0
In-Reply-To: <20200421180733.xrl5ta6cuo2weuva@linux-p48b>
Content-Type: text/plain; charset=windows-1252
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On 21/04/20 20:07, Davidlohr Bueso wrote:
>> 
> 
> I should have looked closer here - I was thinking about the return
> value of rcuwait_wait_event. Yes, that signal_pending check you
> mention makes the sleep semantics change bogus as interruptible is no
> longer just to avoid contributing to the load balance.
> 
> And yes, unfortunately adding prepare_to and finish_rcuwait() looks
> like the most reasonable approach to keeping the tracepoint
> semantics. I also considered extending rcuwait_wait_event() by
> another parameter to pass back to the caller if there was any wait at
> all, but that enlarges the call and is probably less generic.

Yes, at some point the usual prepare_to/finish APIs become simpler.

> I'll send another version keeping the current sleep and tracepoint 
> semantics.

Thanks---and sorry, I should have noticed that way earlier.

Paolo