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: [f2fs-dev] [PATCH v2] f2fs: compress: add compress_inode to cache
 compressed blocks
To:     Eric Biggers <ebiggers@kernel.org>
CC:     <jaegeuk@kernel.org>, <linux-kernel@vger.kernel.org>,
        <linux-f2fs-devel@lists.sourceforge.net>
References: <20201126103709.80006-1-yuchao0@huawei.com>
 <X7/sCLnVPOjkC9DI@sol.localdomain>
 <7ecb947e-2f8c-abd7-c116-c82c474fded7@huawei.com>
 <X8U1zbBa4IaaSYXV@sol.localdomain>
From:   Chao Yu <yuchao0@huawei.com>
Message-ID: <11c11980-1cef-eac5-d42a-4acd4d87565c@huawei.com>
Date:   Tue, 1 Dec 2020 10:39:15 +0800
User-Agent: Mozilla/5.0 (Windows NT 10.0; WOW64; rv:52.0) Gecko/20100101
 Thunderbird/52.9.1
MIME-Version: 1.0
In-Reply-To: <X8U1zbBa4IaaSYXV@sol.localdomain>
Content-Type: text/plain; charset="windows-1252"; format=flowed
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 2020/12/1 2:11, Eric Biggers wrote:
> On Fri, Nov 27, 2020 at 09:01:47AM +0800, Chao Yu wrote:
>> On 2020/11/27 1:55, Eric Biggers wrote:
>>> On Thu, Nov 26, 2020 at 06:37:09PM +0800, Chao Yu wrote:
>>>> Support to use address space of inner inode to cache compressed block,
>>>> in order to improve cache hit ratio of random read.
>>>>
>>>> Signed-off-by: Chao Yu <yuchao0@huawei.com>
>>>
>>> If the file is also encrypted, are the compressed pages that are cached still
>>> encrypted, or are they decrypted?
>>
>> In current implementation, they are decrypted in cache.
>>
> 
> One of the things the FS_IOC_REMOVE_ENCRYPTION key ioctl is supposed to
> accomplish is evicting all the pagecache pages for all encrypted files that were
> using a particular key.  This happens as a consequence of the ioctl evicting the
> inodes that were using that key.  If the user is also using the init_on_free=1
> kernel command-line option to enable automatic zeroing of all freed memory, that
> should cause those inode's pagecache pages (which contain decrypted data) to be
> zeroed, so that they can't be compromised later by an online attack.
> 
> This new filesystem-wide cache containing decrypted pages might break that.  It
> sounds like when an inode is evicted, its cached pages won't necessarily be
> evicted from this new filesystem-wide cache.

Correct.

> 
> Can you ensure that pages get evicted from this new cache when the inode to
> which they belong is evicted?

Not yet with current implementation, let me consider to find a way to make sure
all cached compressed pages being evicted during f2fs_evict_inode().

Thanks,

> 
> - Eric
> .
>