linux-ext4.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Eric Sandeen <sandeen@redhat.com>
To: Dmitry Monakhov <dmonakhov@openvz.org>
Cc: linux-ext4@vger.kernel.org
Subject: Re: [PATCH 1/4] ext4: delalloc quota fixes
Date: Tue, 24 Nov 2009 09:24:06 -0600	[thread overview]
Message-ID: <4B0BFA96.30803@redhat.com> (raw)
In-Reply-To: <1259017138-19165-1-git-send-email-dmonakhov@openvz.org>

Dmitry Monakhov wrote:
> This patch fix most visible problems with delalloc+quota issues
> - ext4_get_reserved_space() must return bytes instead of blocks.
> - Claim allocated meta blocks. Do it as we do for data blocks
>   but delay it untill proper moment.
> - Move space claiming to ext4_da_update_reserve_space()
>   Only here we do know what we are dealing with data or metadata
> 
> The most useful test case for delalloc+quota is concurent tasks
> with write+truncate vs chown for a same file.
> 
> Signed-off-by: Dmitry Monakhov <dmonakhov@openvz.org>
> ---
>  fs/ext4/inode.c   |   12 ++++++++----
>  fs/ext4/mballoc.c |    6 ------
>  2 files changed, 8 insertions(+), 10 deletions(-)
> 
> diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
> index ab22297..e642cdb 100644
> --- a/fs/ext4/inode.c
> +++ b/fs/ext4/inode.c
> @@ -1057,7 +1057,7 @@ qsize_t ext4_get_reserved_space(struct inode *inode)
>  		EXT4_I(inode)->i_reserved_meta_blocks;
>  	spin_unlock(&EXT4_I(inode)->i_block_reservation_lock);
>  
> -	return total;
> +	return (total << inode->i_blkbits);
>  }

Ow, whoops, yes this looks like a bug!  (maybe should be in its own patch?)

>  /*
>   * Calculate the number of metadata blocks need to reserve
> @@ -1096,7 +1096,7 @@ static int ext4_calc_metadata_amount(struct inode *inode, int blocks)
>  static void ext4_da_update_reserve_space(struct inode *inode, int used)
>  {
>  	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
> -	int total, mdb, mdb_free;
> +	int total, mdb, mdb_free, mdb_claim = 0;
>  
>  	spin_lock(&EXT4_I(inode)->i_block_reservation_lock);
>  	/* recalculate the number of metablocks still need to be reserved */
> @@ -1109,7 +1109,9 @@ static void ext4_da_update_reserve_space(struct inode *inode, int used)
>  
>  	if (mdb_free) {
>  		/* Account for allocated meta_blocks */
> -		mdb_free -= EXT4_I(inode)->i_allocated_meta_blocks;
> +		mdb_claim = EXT4_I(inode)->i_allocated_meta_blocks;
> +		BUG_ON(mdb_free < mdb_claim);

Did you see this happening in testing?

> +		mdb_free -= mdb_claim;
>  
>  		/* update fs dirty blocks counter */
>  		percpu_counter_sub(&sbi->s_dirtyblocks_counter, mdb_free);
> @@ -1119,7 +1121,9 @@ static void ext4_da_update_reserve_space(struct inode *inode, int used)
>  
>  	/* update per-inode reservations */
>  	BUG_ON(used  > EXT4_I(inode)->i_reserved_data_blocks);
> -	EXT4_I(inode)->i_reserved_data_blocks -= used;
> + 	EXT4_I(inode)->i_reserved_data_blocks -= used;

looks like a little whitespace damage here

> +	percpu_counter_sub(&sbi->s_dirtyblocks_counter, used + mdb_claim);
> +	vfs_dq_claim_block(inode, used + mdb_claim);
>  	spin_unlock(&EXT4_I(inode)->i_block_reservation_lock);
>  
>  	/*
> diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c
> index bba1282..d4c52db 100644
> --- a/fs/ext4/mballoc.c
> +++ b/fs/ext4/mballoc.c
> @@ -2750,12 +2750,6 @@ ext4_mb_mark_diskspace_used(struct ext4_allocation_context *ac,
>  	if (!(ac->ac_flags & EXT4_MB_DELALLOC_RESERVED))
>  		/* release all the reserved blocks if non delalloc */
>  		percpu_counter_sub(&sbi->s_dirtyblocks_counter, reserv_blks);
> -	else {
> -		percpu_counter_sub(&sbi->s_dirtyblocks_counter,
> -						ac->ac_b_ex.fe_len);
> -		/* convert reserved quota blocks to real quota blocks */
> -		vfs_dq_claim_block(ac->ac_inode, ac->ac_b_ex.fe_len);
> -	}

Maybe Mingming can review this change better, I don't have all the quota paths
in my head yet ...

-Eric

>  	if (sbi->s_log_groups_per_flex) {
>  		ext4_group_t flex_group = ext4_flex_group(sbi,


  parent reply	other threads:[~2009-11-24 15:24 UTC|newest]

Thread overview: 18+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2009-11-23 18:13 ext4+quota patch series Dmitry Monakhov
2009-11-23 18:15 ` Eric Sandeen
2009-11-23 19:18   ` Dmitry Monakhov
2009-11-23 19:35     ` Eric Sandeen
2009-11-23 18:30 ` [PATCH 1/4] ext4: delalloc quota fixes Dmitry Monakhov
2009-11-23 22:43   ` Dmitry Monakhov
2009-11-23 22:58   ` Dmitry Monakhov
2009-11-23 22:58     ` [PATCH 2/4] ext4: fix race chown vs truncate Dmitry Monakhov
2009-11-23 22:58       ` [PATCH 3/4] ext4: quota macros cleanup Dmitry Monakhov
2009-11-23 22:58         ` [PATCH 4/4] ext4: fix incorrect block reservation on quota transfer Dmitry Monakhov
2009-11-24 15:24     ` Eric Sandeen [this message]
2009-11-24 19:38       ` [PATCH 1/4] ext4: delalloc quota fixes Dmitry Monakhov
2009-12-08  0:00     ` Mingming
2009-12-08  6:34       ` Dmitry Monakhov
2009-11-23 18:32 ` [PATCH 2/4] ext4: fix race chown vs truncate Dmitry Monakhov
2009-11-23 18:42   ` Dmitry Monakhov
2009-11-23 18:33 ` [PATCH 3/4] ext4: quota macros cleanup Dmitry Monakhov
2009-11-23 18:34 ` [PATCH 4/4] ext4: fix incorrect block reservation on quota transfer Dmitry Monakhov

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=4B0BFA96.30803@redhat.com \
    --to=sandeen@redhat.com \
    --cc=dmonakhov@openvz.org \
    --cc=linux-ext4@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).