[RFC] Snappy compressor for Linux Kernel (specifically, zram)

Greg KH greg at kroah.com
Fri Apr 15 02:21:51 UTC 2011


On Fri, Apr 15, 2011 at 04:07:36AM +0300, Zeev Tarantov wrote:
> Hello,
> 
> After Google open sources a fast compression library they wrote
> (http://code.google.com/p/snappy/) I've ported it to C with the
> intention of using it with zram within linux, instead of LZO. The code
> is BSD licensed and copyrighted by Google. I am not affiliated with
> them.
> I am interested in getting this into the mainline tree, if there is
> interest from users.
> Since zram itself is in staging directory and the code I ported
> probably needs a bit more work before it's up to the kernel's
> standards (I'm new to kernel development and it shows signs of being
> originally written in C++ for userspace), I thought maybe getting it
> into staging would be the right way.
> 
> I have a comparison between LZO and Snappy of perf profiles of
> untaring a big tarball onto an ext4 fs on zram, dropping caches and
> reading the files back.
> They show a compelling difference in cpu usage:
> https://github.com/zeevt/csnappy/
> 
> I have patches that apply to 2.6.39-rc3 at:
> https://github.com/zeevt/csnappy/raw/master/linux_2_6_39-rc3_csnappy.patch
> https://github.com/zeevt/csnappy/raw/master/linux_2_6_39-rc3_zram.patch
> 
> I'm interested in what you think about the idea, whether this is the
> right mailing list for it, and how to get this submitted, reviewed and
> pulled. And improved, certainly.

Why is this needed to be added to the kernel?  What does it provide that
users or other parts of the kernel needs?

As for the right list, we can help you out, but you have read
Documentation/SubmittingPatches and Documentation/CodingStyle in the
kernel tree already, right?

thanks,

greg k-h



More information about the devel mailing list