[1/2] ext4, project: expand inode extra size if possible

when upgrading from old format, try to set project id
to old file first time, it will return EOVERFLOW, but if
that file is dirtied(touch etc), changing project id will
be allowed, this might be confusing for users, we could
try to expand @i_extra_iszie here too.

Reported-by: zhangyi(F) <yi.zhang@huawei.com>
Signed-off-by: Wang Shilong <wshilong@ddn.com>
---
 fs/ext4/ext4.h  |  3 +++
 fs/ext4/inode.c |  7 +++----
 fs/ext4/ioctl.c | 17 +++++++++++++++--
 3 files changed, 21 insertions(+), 6 deletions(-)

on 2017/6/30 at 11:51, Wang Shilong wrote:
> when upgrading from old format, try to set project id
> to old file first time, it will return EOVERFLOW, but if
> that file is dirtied(touch etc), changing project id will
> be allowed, this might be confusing for users, we could
> try to expand @i_extra_iszie here too.
> 
> Reported-by: zhangyi(F) <yi.zhang@huawei.com>
> Signed-off-by: Wang Shilong <wshilong@ddn.com>
> ---
>  fs/ext4/ext4.h  |  3 +++
>  fs/ext4/inode.c |  7 +++----
>  fs/ext4/ioctl.c | 17 +++++++++++++++--
>  3 files changed, 21 insertions(+), 6 deletions(-)
> 
> diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h
> index 3219154..640f006 100644
> --- a/fs/ext4/ext4.h
> +++ b/fs/ext4/ext4.h
> @@ -2453,6 +2453,9 @@ int ext4_walk_page_buffers(handle_t *handle,
>  			   int *partial,
>  			   int (*fn)(handle_t *handle,
>  				     struct buffer_head *bh));
> +int ext4_expand_extra_isize(struct inode *inode,
> +			    unsigned int new_extra_isize,
> +			    struct ext4_iloc iloc, handle_t *handle);
>  int do_journal_get_write_access(handle_t *handle,
>  				struct buffer_head *bh);
>  #define FALL_BACK_TO_NONDELALLOC 1
> diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
> index 5cf82d0..d53fae6 100644
> --- a/fs/ext4/inode.c
> +++ b/fs/ext4/inode.c
> @@ -5632,10 +5632,9 @@ ext4_reserve_inode_write(handle_t *handle, struct inode *inode,
>   * Expand an inode by new_extra_isize bytes.
>   * Returns 0 on success or negative error number on failure.
>   */
> -static int ext4_expand_extra_isize(struct inode *inode,
> -				   unsigned int new_extra_isize,
> -				   struct ext4_iloc iloc,
> -				   handle_t *handle)
> +int ext4_expand_extra_isize(struct inode *inode,
> +			    unsigned int new_extra_isize,
> +			    struct ext4_iloc iloc, handle_t *handle)
>  {
>  	struct ext4_inode *raw_inode;
>  	struct ext4_xattr_ibody_header *header;
> diff --git a/fs/ext4/ioctl.c b/fs/ext4/ioctl.c
> index 0c21e22..819f59d 100644
> --- a/fs/ext4/ioctl.c
> +++ b/fs/ext4/ioctl.c
> @@ -319,6 +319,7 @@ static int ext4_ioctl_setproject(struct file *filp, __u32 projid)
>  	struct ext4_iloc iloc;
>  	struct ext4_inode *raw_inode;
>  	struct dquot *transfer_to[MAXQUOTAS] = { };
> +	bool need_expand = false;
>  
>  	if (!ext4_has_feature_project(sb)) {
>  		if (projid != EXT4_DEF_PROJID)
> @@ -350,7 +351,10 @@ static int ext4_ioctl_setproject(struct file *filp, __u32 projid)
>  		goto out_unlock;
>  
>  	raw_inode = ext4_raw_inode(&iloc);
> -	if (!EXT4_FITS_IN_INODE(raw_inode, ei, i_projid)) {
> +	if (!EXT4_FITS_IN_INODE(raw_inode, ei, i_projid) &&
> +	    !ext4_test_inode_state(inode, EXT4_STATE_NO_EXPAND)) {
> +		need_expand = true;
> +	} else if (!EXT4_FITS_IN_INODE(raw_inode, ei, i_projid)) {
>  		err = -EOVERFLOW;
>  		brelse(iloc.bh);
>  		goto out_unlock;
> @@ -361,12 +365,21 @@ static int ext4_ioctl_setproject(struct file *filp, __u32 projid)
>  
>  	handle = ext4_journal_start(inode, EXT4_HT_QUOTA,
>  		EXT4_QUOTA_INIT_BLOCKS(sb) +
> -		EXT4_QUOTA_DEL_BLOCKS(sb) + 3);
> +		EXT4_QUOTA_DEL_BLOCKS(sb) + 3 +
> +		need_expand ? EXT4_DATA_TRANS_BLOCKS(sb) : 0);
>  	if (IS_ERR(handle)) {
>  		err = PTR_ERR(handle);
>  		goto out_unlock;
>  	}
>  
> +	if (need_expand) {
> +		err = ext4_expand_extra_isize(inode,
> +				      EXT4_SB(sb)->s_want_extra_isize,
> +				      iloc, handle);
> +		if (err)
> +			goto out_stop;
> +	}
> +

ext4_expand_extra_isize should be invoked after ext4_reserve_inode_write.

And I think it is better to restructure ext4_expand_extra_isize by moving NO_EXPAND check,
nojournal check and journal credits extend into it, and then we just if i_projid is in
the inode or not, if not, invoke ext4_expand_extra_isize. (don't forget to do cleanup for
ext4_mark_inode_dirty), And then we can remove many check in the above code.

Thanks
Miao

>  	err = ext4_reserve_inode_write(handle, inode, &iloc);
>  	if (err)
>  		goto out_stop;
>

on 2017/7/3 at 9:16, Wang Shilong wrote:
> 
> ________________________________________
> From: Miao Xie [miaoxie@huawei.com]
> Sent: Friday, June 30, 2017 16:48
> To: Wang Shilong; linux-ext4@vger.kernel.org
> Cc: tytso@mit.edu; Li Xi; yi.zhang@huawei.com; adilger@dilger.ca; Wang Shilong; Shuichi Ihara
> Subject: Re: [PATCH 1/2] ext4, project: expand inode extra size if possible
> 
> on 2017/6/30 at 11:51, Wang Shilong wrote:
>> when upgrading from old format, try to set project id
>> to old file first time, it will return EOVERFLOW, but if
>> that file is dirtied(touch etc), changing project id will
> <...SNIP...>
> 
> ext4_expand_extra_isize should be invoked after ext4_reserve_inode_write.
> 
> And I think it is better to restructure ext4_expand_extra_isize by moving NO_EXPAND check,
> nojournal check and journal credits extend into it, and then we just if i_projid is in
> 
> ---->I agreed we could move NO_EXPAND check, but i don't think it good idea to move journal
> credits extend to it, jbd2_extend_journal() might fail, and we'd better avoid it.
> 
> For changing projectid, we could know how many credits before start transaction..

I found most check in set_projectid is the same as ext4_mark_inode_dirty, so I think it's better
to move those checks into ext4_expand_extra_isize to avoid the reduplicated code.

And I don't think jbd2_journal_extend's failure is a big problem, because we just invoke it once
most of the time, and even if we fail to extend the inode because of jbd2_journal_extend's failure,
the metadata is still safe, so it is unnecessary to change many codes in set_projectid, or
we will impact readability of the code.

> 
> the inode or not, if not, invoke ext4_expand_extra_isize. (don't forget to do cleanup for
> ext4_mark_inode_dirty), And then we can remove many check in the above code.
> 
> 
> ---->what do you mean cleanup for ext4_mark_inode_dirty()? I supposed you mean don't
> call ext4_mark_iloc_dirty() if extend fail? i think that is expected, even inode extend fail,
> if ext4_mark_inode_dirty() is called, it means inode is dirtied already before we call the
> function.

I means if we move the checks into ext4_expand_extra_isize, we should do cleanup for ext4_mark_inode_dirty.
(ext4_mark_iloc_dirty() shoud be invoked even if extend fail, because we need update the on-disk inode)

Thanks
Miao

> Thanks,
> Shilong
> 
> 
> .
>