Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII;
 format=flowed
Content-Transfer-Encoding: 7bit
Date:   Tue, 08 Jan 2019 17:07:23 +0100
From:   Roman Penyaev <rpenyaev@suse.de>
To:     Davidlohr Bueso <dbueso@suse.de>
Cc:     Jason Baron <jbaron@akamai.com>, Al Viro <viro@zeniv.linux.org.uk>,
        Andrew Morton <akpm@linux-foundation.org>,
        Linus Torvalds <torvalds@linux-foundation.org>,
        linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH 1/1] epoll: remove wrong assert that ep_poll_callback is
 always called with irqs off
In-Reply-To: <725cdce88418c2ec62ef6014d388dbeb@suse.de>
References: <20190108100121.20247-1-rpenyaev@suse.de>
 <f582f40b9697dced81d16b3b35a4b071@suse.de>
 <725cdce88418c2ec62ef6014d388dbeb@suse.de>
Message-ID: <2cec506322600c3d2b59f2ca9bcd59dc@suse.de>
User-Agent: Roundcube Webmail
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On 2019-01-08 16:16, Davidlohr Bueso wrote:
> On 2019-01-08 04:42, Roman Penyaev wrote:
>> What we can do:
>> 
>> a) disable irqs if we are not in interrupt.
>> b) revert the patch completely.
>> 
>> David, is it really crucial in terms of performance to avoid double
>> local_irq_save() on Xen on this ep_poll_callback() hot path?
> 
> Note that such optimizations are also relevant for baremetal, ie: x86
> PUSHF + POPF can be pretty expensive because of insn dependencies.
> 
>> 
>> For example why not to do the following:
>> 
>>   if (!in_interrupt())
>>        local_irq_save(flags);
>>   read_lock(ep->lock);
>> 
>> with huge comment explaining performance number.
>> 
>> Or just give up and simply revert the original patch completely
>> and always call read_lock_irqsave().
> 
> Yeah so the reason why I had done the other epoll lock irq
> optimizations was because they were painfully obvious.
> ep_poll_callback(), however is a different beast, as you've
> encountered. I vote for not shooting ourselves in the foot and just
> dropping this patch -- most large performance benefits will come from
> microbenches anyway.

Then I will send another patch which changes read_lock() on 
read_lock_irqsave(),
since simple revert of the original patch won't work.

Thanks.

--
Roman