Received-SPF: pass (google.com: domain of linux-ext4-owner@vger.kernel.org designates 2620:137:e000::3:8 as permitted sender) client-ip=2620:137:e000::3:8;
Date:   Wed, 25 Oct 2023 20:17:50 +0200
From:   Jan Kara <jack@suse.cz>
To:     "Matthew Wilcox (Oracle)" <willy@infradead.org>
Cc:     Jan Kara <jack@suse.com>, linux-ext4@vger.kernel.org,
        linux-fsdevel@vger.kernel.org,
        "Fabio M . De Francesco" <fmdefrancesco@gmail.com>
Subject: Re: [PATCH 07/10] ext2: Handle large block size directories in
 ext2_delete_entry()
Message-ID: <20231025181750.i2ibf27imrmpkqnt@quack3>
References: <20230921200746.3303942-1-willy@infradead.org>
 <20230921200746.3303942-7-willy@infradead.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20230921200746.3303942-7-willy@infradead.org>
Precedence: bulk

On Thu 21-09-23 21:07:44, Matthew Wilcox (Oracle) wrote:
> If the block size is > PAGE_SIZE, we need to calculate these offsets
> relative to the start of the folio, not the page.
> 
> Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>

So I had a fresh look at the patches and I like them so I'll take them into
my tree for the merge window. Just this patch has somewhat surprising
subject and changelog - I guess it should be standard "convert to folios"
kind of thing, shouldn't it? Because a lot of dir code is not prepared for
"large block size" so it is strange to speak about it in
ext2_delete_entry(). I can fix it up on commit.

								Honza

> ---
>  fs/ext2/dir.c | 30 +++++++++++++++++-------------
>  1 file changed, 17 insertions(+), 13 deletions(-)
> 
> diff --git a/fs/ext2/dir.c b/fs/ext2/dir.c
> index 2fc910e99234..7e75cfaa709c 100644
> --- a/fs/ext2/dir.c
> +++ b/fs/ext2/dir.c
> @@ -586,16 +586,20 @@ int ext2_add_link (struct dentry *dentry, struct inode *inode)
>   */
>  int ext2_delete_entry(struct ext2_dir_entry_2 *dir, struct page *page)
>  {
> -	struct inode *inode = page->mapping->host;
> -	char *kaddr = (char *)((unsigned long)dir & PAGE_MASK);
> -	unsigned from = offset_in_page(dir) & ~(ext2_chunk_size(inode)-1);
> -	unsigned to = offset_in_page(dir) +
> -				ext2_rec_len_from_disk(dir->rec_len);
> +	struct folio *folio = page_folio(page);
> +	struct inode *inode = folio->mapping->host;
> +	size_t from, to;
> +	char *kaddr;
>  	loff_t pos;
> -	ext2_dirent *pde = NULL;
> -	ext2_dirent *de = (ext2_dirent *)(kaddr + from);
> +	ext2_dirent *de, *pde = NULL;
>  	int err;
>  
> +	from = offset_in_folio(folio, dir);
> +	to = from + ext2_rec_len_from_disk(dir->rec_len);
> +	kaddr = (char *)dir - from;
> +	from &= ~(ext2_chunk_size(inode)-1);
> +	de = (ext2_dirent *)(kaddr + from);
> +
>  	while ((char*)de < (char*)dir) {
>  		if (de->rec_len == 0) {
>  			ext2_error(inode->i_sb, __func__,
> @@ -606,18 +610,18 @@ int ext2_delete_entry(struct ext2_dir_entry_2 *dir, struct page *page)
>  		de = ext2_next_entry(de);
>  	}
>  	if (pde)
> -		from = offset_in_page(pde);
> -	pos = page_offset(page) + from;
> -	lock_page(page);
> -	err = ext2_prepare_chunk(page, pos, to - from);
> +		from = offset_in_folio(folio, pde);
> +	pos = folio_pos(folio) + from;
> +	folio_lock(folio);
> +	err = ext2_prepare_chunk(&folio->page, pos, to - from);
>  	if (err) {
> -		unlock_page(page);
> +		folio_unlock(folio);
>  		return err;
>  	}
>  	if (pde)
>  		pde->rec_len = ext2_rec_len_to_disk(to - from);
>  	dir->inode = 0;
> -	ext2_commit_chunk(page, pos, to - from);
> +	ext2_commit_chunk(&folio->page, pos, to - from);
>  	inode->i_mtime = inode_set_ctime_current(inode);
>  	EXT2_I(inode)->i_flags &= ~EXT2_BTREE_FL;
>  	mark_inode_dirty(inode);
> -- 
> 2.40.1
> 
-- 
Jan Kara <jack@suse.com>
SUSE Labs, CR