svn commit: r187395 - head/sys/gnu/fs/ext2fs

[Stanislav Sedov]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Sun, 18 Jan 2009 16:06:44 +0200
Kostik Belousov <kostikbel at gmail.com> mentioned:

> 
> Please see a discussion on the fs@ and reasoning why I declined to commit
> the similar patch.
> 

I'm sorry I've missed that discussion entirely, but I can admit I had all the
same concerns while reviewing the patch, and Alex's comments and Linux code
reading made me fully confident that the modifications are safe and correct.
Let me explain why I think so.

The extra size added in inodes are used to store additional info like extended
attributes, ACLs and so on. Each inode now has a field just after the legacy
inode format struct that shows hom much additional space has been added to
this particular inode. By analyzing this field the operating system can interpret
additional data located in inode, if it understand its format (there might be
application and/or OS specific data too).

Our implementation just ignore this additional fields after sizeof(ext2fs_inode),
both while reading and writing. Thus we don't interet this data yet we don't
overwrite it.  

- -- 
Stanislav Sedov
ST4096-RIPE
-----BEGIN PGP SIGNATURE-----

iEYEARECAAYFAklzQDIACgkQK/VZk+smlYE6TQCcCe7yDdJYU2cxYwiosHjYmK6s
NT8An0+9Jom3CgHFhVy4PrvztjNXgpT9
=Cxan
-----END PGP SIGNATURE-----

!DSPAM:49734032967001169152874!