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] mm/vmscan: fix infinite loop in drop_slab_node
To:     Chris Down <chris@chrisdown.name>, zangchunxin@bytedance.com
Cc:     akpm@linux-foundation.org, linux-mm@kvack.org,
        linux-kernel@vger.kernel.org,
        Muchun Song <songmuchun@bytedance.com>
References: <20200908142456.89626-1-zangchunxin@bytedance.com>
 <20200908150945.GA1301981@chrisdown.name>
From:   Vlastimil Babka <vbabka@suse.cz>
Message-ID: <07c6ebf1-e2b3-11a2-538f-4ac542a4373b@suse.cz>
Date:   Tue, 8 Sep 2020 17:42:05 +0200
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101
 Thunderbird/68.11.0
MIME-Version: 1.0
In-Reply-To: <20200908150945.GA1301981@chrisdown.name>
Content-Type: text/plain; charset=utf-8
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On 9/8/20 5:09 PM, Chris Down wrote:
> drop_caches by its very nature can be extremely performance intensive -- if 
> someone wants to abort after trying too long, they can just send a 
> TASK_KILLABLE signal, no? If exiting the loop and returning to usermode doesn't 
> reliably work when doing that, then _that's_ something to improve, but this 
> looks premature to me until that's demonstrated not to work.

Hm there might be existings scripts (even though I dislike those) running
drop_caches periodically, and they are currently not set up to be killed, so one
day it might surprise someone. Dropping should be a one-time event, not a
continual reclaim.

Maybe we could be a bit smarter and e.g. double the threshold currently
hardcoded as "10" with each iteration?

> zangchunxin@bytedance.com writes:
>>In one drop caches action, only traverse memcg once maybe is better.
>>If user need more memory, they can do drop caches again.
> 
> Can you please provide some measurements of the difference in reclamation in 
> practice?
>