linux-kernel - Re: [PATCH 2/2] zram: remove zlib from the list of recommended algorithms

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date:   Thu, 14 Sep 2017 17:02:49 +0900
From:   Minchan Kim <minchan@...nel.org>
To:     Adam Borowski <kilobyte@...band.pl>
Cc:     Sergey Senozhatsky <sergey.senozhatsky.work@...il.com>,
        Andrew Morton <akpm@...ux-foundation.org>,
        linux-kernel@...r.kernel.org,
        Sergey Senozhatsky <sergey.senozhatsky@...il.com>
Subject: Re: [PATCH 2/2] zram: remove zlib from the list of recommended
 algorithms

On Wed, Sep 13, 2017 at 12:38:56PM +0200, Adam Borowski wrote:
> On Wed, Sep 13, 2017 at 04:12:46PM +0900, Minchan Kim wrote:
> > On Tue, Sep 12, 2017 at 02:00:05PM +0900, Sergey Senozhatsky wrote:
> > > ZSTD tends to outperform deflate/inflate, thus we remove
> > > zlib from the list of recommended algorithms and recommend
> > > zstd instead.
> >  
> > I did test with my sample data and compared zstd with deflate.
> > zstd's compress ratio is lower a little bit but compression
> > speed is much faster 3 times more and decompress speed is too
> > 2 times more. With different data, it is different but overall,
> > zstd would be better for speed at the cost of a little lower compress
> > ratio(about 5%) so I believe it's worth to replace deflate.
> 
> Both zlib and zstd have the compression level adjustable, zstd in a far
> greater range (from lzo-like at lowest levels to mid-range lzma at the
> highest).  Thus, any such comparison needs to mention the used level.
> Ie, if you selected a setting where speed is same, compression ratio
> will be a lot better.

Unfortunately, it seems crypto doesn't support configurable level yet
so when I read the source properly, zstd default level is 3 while
deflate is 6.

> 
> For compressing RAM it's reasonable to keep to fastest levels, and a
> non-adjustable level reduces complexity, but if your use case wants high but
> slow compression, now is a good time to mention this.

What I expect zstd is same comp ratio with zlib but much faster speed.
And lastly, predefined dictionary for 4K comp/decomp. :)

Thanks.