linux-ext4 - Re: [PATCH 07/10] ext4: misc fast commit fixes

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Tue, 3 Nov 2020 17:52:37 +0100
From:   Jan Kara <jack@...e.cz>
To:     Harshad Shirwadkar <harshadshirwadkar@...il.com>
Cc:     linux-ext4@...r.kernel.org, tytso@....edu, jack@...e.cz
Subject: Re: [PATCH 07/10] ext4: misc fast commit fixes

On Sat 31-10-20 13:05:15, Harshad Shirwadkar wrote:
> This patch adds a small number of misc fast commit fixes. Along with
> functional fixes such as setting the right buffer flags, there also
> typo fixes and comment additions.
> 
> Suggested-by: Jan Kara <jack@...e.cz>
> Signed-off-by: Harshad Shirwadkar <harshadshirwadkar@...il.com>

Again, please don't merge logically separate fixes into one commit.

> diff --git a/fs/ext4/ext4_jbd2.h b/fs/ext4/ext4_jbd2.h
> index 00dc668e052b..10855cd230c7 100644
> --- a/fs/ext4/ext4_jbd2.h
> +++ b/fs/ext4/ext4_jbd2.h
> @@ -422,9 +422,13 @@ static inline int ext4_journal_force_commit(journal_t *journal)
>  static inline int ext4_jbd2_inode_add_write(handle_t *handle,
>  		struct inode *inode, loff_t start_byte, loff_t length)
>  {
> -	if (ext4_handle_valid(handle))
> +	if (ext4_handle_valid(handle)) {
> +		ext4_fc_track_range(handle, inode,
> +			start_byte >> inode->i_sb->s_blocksize_bits,
> +			(start_byte + length) >> inode->i_sb->s_blocksize_bits);
>  		return jbd2_journal_inode_ranged_write(handle,
>  				EXT4_I(inode)->jinode, start_byte, length);
> +	}

Why this change? A good changelog would tell me... I'm suspicious here
because ext4_jbd2_inode_add_write() gets called only in data=ordered mode
but fastcommit can run also in data=writeback mode...

I suppose this is for the mmap coverage we were speaking about. Now that
I'm speaking about it again maybe the ext4_fc_track_range() call in
ext4_map_blocks() is actually enough? I mean once we allocate blocks for a
range (either from page fault, write, or writeback of delalloc), they will
become properly tracked in ext4_map_blocks() and that's all we need? But
then I'm missing why we have so many ext4_fc_track_range() calls around the
code... Can you please explain?

								Honza

-- 
Jan Kara <jack@...e.com>
SUSE Labs, CR