[Ocfs2-devel] [PATCH 25/29] ocfs2: Implementation of local and global quota file handling

[Jan Kara]

On Tue 28-10-08 12:36:52, Joel Becker wrote:
> On Sat, Oct 25, 2008 at 12:08:18AM +0200, Jan Kara wrote:
> > For each quota type each node has local quota file. In this file it stores
> > changes users have made to disk usage via this node. Once in a while this
> > information is synced to global file (and thus with other nodes) so that
> > limits enforcement at least aproximately works.
> > diff --git a/fs/ocfs2/ocfs2_fs.h b/fs/ocfs2/ocfs2_fs.h
> > index dd17137..04bebd2 100644
> > --- a/fs/ocfs2/ocfs2_fs.h
> > +++ b/fs/ocfs2/ocfs2_fs.h
> > @@ -878,6 +878,101 @@ static inline int ocfs2_xattr_get_type(struct ocfs2_xattr_entry *xe)
> >  	return xe->xe_type & OCFS2_XATTR_TYPE_MASK;
> >  }
> >  
> > +/*
> > + *  On disk structures for global quota file
> > + */
> > +
> > +/* Magic numbers and known versions for global quota files */
> > +#define OCFS2_GLOBAL_QMAGICS {\
> > +	0x0cf52470, /* USRQUOTA */ \
> > +	0x0cf52471  /* GRPQUOTA */ \
> > +}
> > +
> > +#define OCFS2_GLOBAL_QVERSIONS {\
> > +	0, \
> > +	0, \
> > +}
> > +
> > +/* Generic header of all quota files */
> > +struct ocfs2_disk_dqheader {
> > +	__le32 dqh_magic;	/* Magic number identifying file */
> > +	__le32 dqh_version;	/* Quota format version */
> > +};
> > +
> > +#define OCFS2_GLOBAL_INFO_OFF (sizeof(struct ocfs2_disk_dqheader))
> > +
> > +/* Information header of global quota file (immediately follows the generic
> > + * header) */
> > +struct ocfs2_global_disk_dqinfo {
> > +/*00*/	__le32 dqi_bgrace;
> > +	__le32 dqi_igrace;
> > +	__le32 dqi_syncms;
> > +	__le32 dqi_blocks;
> > +/*10*/	__le32 dqi_free_blk;
> > +	__le32 dqi_free_entry;
> > +};
> 
> 	The way I read this, the first block of a quota file contains
> first the disk_dqheader, and then the local info
> (global_disk_dqinfo,local_disk_dqinfo).  Is that right?  I'd love to see
> the ecc field (__le64 for now) in the disk_dqheader.
  Yes. Actually, this set of patches already counts with ECC, but it's
hidden ;). See below.

> > +/* Structure with global user / group information. We reserve some space
> > + * for future use. */
> > +struct ocfs2_global_disk_dqblk {
> > +/*00*/	__le32 dqb_id;          /* ID the structure belongs to */
> > +	__le32 dqb_use_count;   /* Number of nodes having reference to this structure */
> > +	__le64 dqb_ihardlimit;  /* absolute limit on allocated inodes */
> > +/*10*/	__le64 dqb_isoftlimit;  /* preferred inode limit */
> > +	__le64 dqb_curinodes;   /* current # allocated inodes */
> > +/*20*/	__le64 dqb_bhardlimit;  /* absolute limit on disk space */
> > +	__le64 dqb_bsoftlimit;  /* preferred limit on disk space */
> > +/*30*/	__le64 dqb_curspace;    /* current space occupied */
> > +	__le64 dqb_btime;       /* time limit for excessive disk use */
> > +/*40*/	__le64 dqb_itime;       /* time limit for excessive inode use */
> > +	__le64 dqb_pad1;
> > +/*50*/	__le64 dqb_pad2;
> > +};
> 
> 	We also need space for the ecc data here.  It looks like ecc
> space wasn't added to this series.  Rather than comment on each
> structure, just let me know when that's ready.
  What I did is, that in the end of each quotafile block there are 8 bytes
reserved for OCFS2 use (both in local and global quota files). So these
bytes are intended to be used for ECC. I hope that's fine.

								Honza

-- 
Jan Kara <jack at suse.cz>
SUSE Labs, CR