Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Tue, 3 May 2022 18:04:13 +0200
From:   Michal Hocko <mhocko@suse.com>
To:     "Paul E. McKenney" <paulmck@kernel.org>
Cc:     liam.howlett@oracle.com, willy@infradead.org, walken.cr@gmail.com,
        hannes@cmpxchg.org, linux-mm@kvack.org,
        linux-kernel@vger.kernel.org
Subject: Re: Memory allocation on speculative fastpaths
Message-ID: <YnFSfc8BR8CadOtw@dhcp22.suse.cz>
References: <20220503155913.GA1187610@paulmck-ThinkPad-P17-Gen-1>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20220503155913.GA1187610@paulmck-ThinkPad-P17-Gen-1>
Precedence: bulk

On Tue 03-05-22 08:59:13, Paul E. McKenney wrote:
> Hello!
> 
> Just following up from off-list discussions yesterday.
> 
> The requirements to allocate on an RCU-protected speculative fastpath
> seem to be as follows:
> 
> 1.	Never sleep.
> 2.	Never reclaim.
> 3.	Leave emergency pools alone.
> 
> Any others?
> 
> If those rules suffice, and if my understanding of the GFP flags is
> correct (ha!!!), then the following GFP flags should cover this:
> 
> 	__GFP_NOMEMALLOC | __GFP_NOWARN

GFP_NOWAIT | __GFP_NOMEMALLOC | __GFP_NOWARN
 
> Or is this just a fancy way of always returning NULL or some such?  ;-)

It could fail quite easily. We would also want to guarantee (by
documenting I guess) that the page allocator never does anything that
would depend or invoke rcu_synchronize or something like that.

I believe this is the case currently.
-- 
Michal Hocko
SUSE Labs