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] [thread-next>] [day] [month] [year] [list]
Date:   Wed, 27 Nov 2019 09:25:08 -0500
From:   "Theodore Y. Ts'o" <tytso@....edu>
To:     Daniel Phillips <daniel@...nq.net>
Cc:     linux-ext4@...r.kernel.org, linux-kernel@...r.kernel.org,
        linux-fsdevel@...r.kernel.org,
        OGAWA Hirofumi <hirofumi@...l.parknet.co.jp>
Subject: Re: [RFC] Thing 1: Shardmap fox Ext4

A couple of quick observations about Shardmap.

(1) It's licensed[1] under the GPLv3, so it's not compatible with the
kernel license.  That doesn't matter much for ext4, because...

[1] https://github.com/danielbot/Shardmap/blob/master/LICENSE


(2) It's implemented as userspace code (e.g., it uses open(2),
mmap(2), et. al) and using C++, so it would need to be reimplemented
from scratch for use in the kernel.


(3) It's not particularly well documented, making the above more
challenging, but it appears to be a variation of an extensible hashing
scheme, which was used by dbx and Berkley DB.


(4) Because of (2), we won't be able to do any actual benchmarks for a
while.  I just checked the latest version of Tux3[2], and it appears
to be be still using a linear search scheme for its directory ---
e.g., an O(n) lookup ala ext2.  So I'm guessing Shardmap may have been
*designed* for Tux3, but it has not yet been *implemented* for Tux3?

[2] https://github.com/OGAWAHirofumi/linux-tux3/blob/hirofumi/fs/tux3/dir.c#L283


(5) The claim is made that readdir() accesses files sequentially; but
there is also mention in Shardmap of compressing shards (e.g.,
rewriting them) to squeeze out deleted and tombstone entries.  This
pretty much guarantees that it will not be possible to satisfy POSIX
requirements of telldir(2)/seekdir(3) (using a 32-bit or 64-bitt
cookie), NFS (which also requires use of a 32-bit or 64-bit cookie
while doing readdir scan), or readdir() semantics in the face of
directory entries getting inserted or removed from the directory.

(To be specific, POSIX requires readdir returns each entry in a
directory once and only once, and in the case of a directory entry
which is removed or inserted, that directory entry must be returned
exactly zero or one times.  This is true even if telldir(2) ort
seekdir(2) is used to memoize a particular location in the directory,
which means you have a 32-bit or 64-bit cookie to define a particular
location in the readdir(2) stream.  If the file system wants to be
exportable via NFS, it must meet similar requirements ---- except the
32-bit or 64-bit cookie MUST survive a reboot.)

Regards,

						- Ted

Powered by blists - more mailing lists