[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Mon, 15 Dec 2008 14:40:16 +0100
From: Jan Kara <jack@...e.cz>
To: Mingming Cao <cmm@...ibm.com>
Cc: Andrew Morton <akpm@...ux-foundation.org>, tytso <tytso@....edu>,
linux-ext4 <linux-ext4@...r.kernel.org>,
linux-fsdevel <linux-fsdevel@...r.kernel.org>
Subject: Re: [PATCH V4 2/3] quota: Add quota claim and release reserved
quota blocks operations
Hi Mingming,
sorry I spoke to soon in my previous email. You tried to resolve the
issue with dquot_transfer(). I just thought it would belong to the first
patch...
On Fri 12-12-08 12:47:40, Mingming Cao wrote:
> quota: Add quota reservation claim and released operations
>
> Reserved quota will be claimed at the block allocation time. Over-booked
> quota could be returned back with the release callback function.
>
> Signed-off-by: Mingming Cao <cmm@...ibm.com>
> ---
> fs/dquot.c | 101 +++++++++++++++++++++++++++++++++++++++++++++--
> include/linux/quota.h | 4 +
> include/linux/quotaops.h | 53 ++++++++++++++++++++++++
> 3 files changed, 153 insertions(+), 5 deletions(-)
>
<snip>
> Index: linux-2.6.28-rc2/fs/dquot.c
> ===================================================================
> --- linux-2.6.28-rc2.orig/fs/dquot.c 2008-12-12 12:20:45.000000000 -0800
> +++ linux-2.6.28-rc2/fs/dquot.c 2008-12-12 12:21:49.000000000 -0800
<snip>
> @@ -1448,7 +1532,8 @@ int dquot_free_inode(const struct inode
> */
> int dquot_transfer(struct inode *inode, struct iattr *iattr)
> {
> - qsize_t space;
> + qsize_t space, cur_space;
> + qsize_t rsv_space = 0;
> struct dquot *transfer_from[MAXQUOTAS];
> struct dquot *transfer_to[MAXQUOTAS];
> int cnt, ret = NO_QUOTA, chuid = (iattr->ia_valid & ATTR_UID) && inode->i_uid != iattr->ia_uid,
> @@ -1489,12 +1574,16 @@ int dquot_transfer(struct inode *inode,
> }
> }
> spin_lock(&dq_data_lock);
> - space = inode_get_bytes(inode);
> + space = cur_space = inode_get_bytes(inode);
> /* Build the transfer_from list and check the limits */
> for (cnt = 0; cnt < MAXQUOTAS; cnt++) {
> if (transfer_to[cnt] == NODQUOT)
> continue;
> transfer_from[cnt] = inode->i_dquot[cnt];
> + if (!rsv_space) {
> + rsv_space = transfer_from[cnt]->dq_dqb.dqb_rsvspace;
> + space += rsv_space;
Hmm, but you cannot simply transfer all the reserved space from one user
to another. You have to transfer just the amount reserved for this inode
(i.e. EXT4_I(inode)->i_reserved_data_blocks +
EXT4_I(inode)->i_reserved_meta_blocks). The fact is this is not available
to quota code. One possibility is to store this number in VFS inode (which I
guess people would not like because it would increase every inode size by 8
bytes). So probably more plausible is to provide a callback so that quota
code can find out how much space this inode has reserved. This is not a
performance critical operation so the cost of the function call is
acceptable here IMO.
> + }
> if (check_idq(transfer_to[cnt], 1, warntype_to + cnt) ==
> NO_QUOTA || check_bdq(transfer_to[cnt], space, 0,
> warntype_to + cnt) == NO_QUOTA)
> @@ -1518,11 +1607,13 @@ int dquot_transfer(struct inode *inode,
> warntype_from_space[cnt] =
> info_bdq_free(transfer_from[cnt], space);
> dquot_decr_inodes(transfer_from[cnt], 1);
> - dquot_decr_space(transfer_from[cnt], space);
> + dquot_decr_space(transfer_from[cnt], cur_space);
> + dquot_free_reserved_space(transfer_from[cnt], rsv_space);
> }
>
> dquot_incr_inodes(transfer_to[cnt], 1);
> - dquot_incr_space(transfer_to[cnt], space);
> + dquot_incr_space(transfer_to[cnt], cur_space);
> + dquot_resv_space(transfer_to[cnt], rsv_space);
>
> inode->i_dquot[cnt] = transfer_to[cnt];
> }
Honza
--
Jan Kara <jack@...e.cz>
SUSE Labs, CR
--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Powered by blists - more mailing lists