From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-6.9 required=3.0 tests=DKIM_SIGNED,DKIM_VALID, DKIM_VALID_AU,HEADER_FROM_DIFFERENT_DOMAINS,INCLUDES_PATCH,MAILING_LIST_MULTI, SIGNED_OFF_BY,SPF_HELO_NONE,SPF_PASS,URIBL_BLOCKED autolearn=unavailable autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id 44578C2BABC for ; Tue, 7 Apr 2020 02:05:30 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id 1B1732082D for ; Tue, 7 Apr 2020 02:05:30 +0000 (UTC) Authentication-Results: mail.kernel.org; dkim=pass (1024-bit key) header.d=joelfernandes.org header.i=@joelfernandes.org header.b="Ug+nh1/D" Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1726339AbgDGCF3 (ORCPT ); Mon, 6 Apr 2020 22:05:29 -0400 Received: from mail-qt1-f195.google.com ([209.85.160.195]:41950 "EHLO mail-qt1-f195.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1726332AbgDGCF2 (ORCPT ); Mon, 6 Apr 2020 22:05:28 -0400 Received: by mail-qt1-f195.google.com with SMTP id i3so1560468qtv.8 for ; Mon, 06 Apr 2020 19:05:28 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=joelfernandes.org; s=google; h=date:from:to:cc:subject:message-id:references:mime-version :content-disposition:in-reply-to; bh=QE9feUnP6i9X/aj/z/7/gJs0TCfrmf+hqNm1oT5H5Wo=; b=Ug+nh1/Djrrr/Q9qUloEDJX7nwoQbukp++5BAg8jRhC6bKO5v2srUpBPJ5CSdxM8+y DZM3V/FCvDfVTAdln6l0g4IpB2p5Q+zSAwEMPnG4Mpn7Id6AxXESijE8ghTVWJ6YUHrQ ldTC1LqkITSR7e14Q2G+OgeqG3MEEF1b1wsaA= X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:date:from:to:cc:subject:message-id:references :mime-version:content-disposition:in-reply-to; bh=QE9feUnP6i9X/aj/z/7/gJs0TCfrmf+hqNm1oT5H5Wo=; b=DJolLnwkLJrVRbEkrB44rVlrn5EjLZZgSr40KXZV1Npo1domnUosS1HL0mynkkRk1E TLdRs0DVGnVhnU49801B6Wm8HMEQMVjyLoFcqxsxjn3PePz7KH2YjSddOFmp1Ig/Vllo KIDjy70/mOQAYgaCvZZEVoKLEiwjI+kXVSgro+LjNHdvlOCJ8RAaowS+RkxJVeyhGQbW tx+pdQ7kzDaIK/fGZ2STMsidvgK53bRoz3Q56os9soPfhFBJNhc+CGW0M+Cx2PDkJl3J 4xKpXyk1Op8CD9uuOOM5tZtJPvRkoj4/D6oGxt7aI4vKu4SgrOeMevQxvvkqFoSFO0cv 9X9Q== X-Gm-Message-State: AGi0PubiViPjOtHY+fOjoJNGYd/7OUzvaoxIwM/9NMmg6LBsy+rEUF9I o/wkFThtfPOjrP9cC/Jaqmf5RA== X-Google-Smtp-Source: APiQypKtWDj3oedlvF3bwvLFauAg3ZIkbXpPPfDmPdit0eD4VUDvGdBp5v89JmUTbXIWhz19qDHGug== X-Received: by 2002:aed:37a8:: with SMTP id j37mr213233qtb.272.1586225127531; Mon, 06 Apr 2020 19:05:27 -0700 (PDT) Received: from localhost ([2620:15c:6:12:9c46:e0da:efbf:69cc]) by smtp.gmail.com with ESMTPSA id n5sm15879583qkn.4.2020.04.06.19.05.26 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Mon, 06 Apr 2020 19:05:26 -0700 (PDT) Date: Mon, 6 Apr 2020 22:05:25 -0400 From: Joel Fernandes To: "Uladzislau Rezki (Sony)" Cc: LKML , "Paul E . McKenney" , RCU , linux-mm@kvack.org, Andrew Morton , Steven Rostedt , Oleksiy Avramchenko Subject: Re: [PATCH 1/3] rcu/tree: use more permissive parameters when attaching a head Message-ID: <20200407020525.GA113967@google.com> References: <20200402123253.10382-1-urezki@gmail.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20200402123253.10382-1-urezki@gmail.com> Sender: rcu-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: rcu@vger.kernel.org On Thu, Apr 02, 2020 at 02:32:51PM +0200, Uladzislau Rezki (Sony) wrote: > It is documneted that a headless object can be reclaimed from > might_sleep() context only. Because of that when a head is > dynamically attached it makes sense to drop the lock and do > an allocation with much more permissve flags comparing if it > is done from atomic context. > > That is why use GFP_KERNEL flag plus some extra ones which > would make an allocation most likely to be succeed. The big > advantage of doing so is a direct reclaim process. > > Tested such approach on my local tiny system with 145MB of > ram(the minimum amount the KVM system is capable of booting) > and 4xCPUs. For stressing the rcuperf module was used. During > tests with difference combinations i did not observe any hit > of our last emergency case, when synchronize_rcu() is involved. > > Please note, the "dynamically attaching" path was enabled only, > apart of that all types of objects were considered as headless > variant during testing. > > Signed-off-by: Uladzislau Rezki (Sony) > Suggested-by: Joel Fernandes (Google) Beautifully done, thanks Vlad! I agree with this and the other 2 changes and I applied it to my rcu/kfree branch for further testing by both of us. thanks, - Joel > --- > kernel/rcu/tree.c | 23 ++++++++++++++++------- > 1 file changed, 16 insertions(+), 7 deletions(-) > > diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c > index 6172e6296dd7..24f620a06219 100644 > --- a/kernel/rcu/tree.c > +++ b/kernel/rcu/tree.c > @@ -3148,13 +3148,10 @@ static inline struct rcu_head *attach_rcu_head_to_object(void *obj) > { > unsigned long *ptr; > > + /* Try hard to get the memory. */ > ptr = kmalloc(sizeof(unsigned long *) + > - sizeof(struct rcu_head), GFP_NOWAIT | __GFP_NOWARN); > - > - if (!ptr) > - ptr = kmalloc(sizeof(unsigned long *) + > - sizeof(struct rcu_head), GFP_ATOMIC | __GFP_NOWARN); > - > + sizeof(struct rcu_head), GFP_KERNEL | > + __GFP_ATOMIC | __GFP_HIGH | __GFP_RETRY_MAYFAIL); > if (!ptr) > return NULL; > > @@ -3222,9 +3219,20 @@ void kvfree_call_rcu(struct rcu_head *head, rcu_callback_t func) > if (!success) { > /* Is headless object? */ > if (head == NULL) { > + /* Drop the lock. */ > + if (krcp->initialized) > + spin_unlock(&krcp->lock); > + local_irq_restore(flags); > + > head = attach_rcu_head_to_object(ptr); > if (head == NULL) > - goto unlock_return; > + goto inline_return; > + > + /* Take it back. */ > + local_irq_save(flags); > + krcp = this_cpu_ptr(&krc); > + if (krcp->initialized) > + spin_lock(&krcp->lock); > > /* > * Tag the headless object. Such objects have a back-pointer > @@ -3263,6 +3271,7 @@ void kvfree_call_rcu(struct rcu_head *head, rcu_callback_t func) > spin_unlock(&krcp->lock); > local_irq_restore(flags); > > +inline_return: > /* > * High memory pressure, so inline kvfree() after > * synchronize_rcu(). We can do it from might_sleep() > -- > 2.20.1 >