Received-SPF: pass (google.com: domain of linux-ext4-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Thu, 1 Dec 2022 12:55:00 +0100
From:   Jan Kara <jack@suse.cz>
To:     "Ritesh Harjani (IBM)" <ritesh.list@gmail.com>
Cc:     Jan Kara <jack@suse.cz>, Ted Tso <tytso@mit.edu>,
        linux-ext4@vger.kernel.org, Christoph Hellwig <hch@infradead.org>
Subject: Re: [PATCH 0/9] ext4: Stop using ext4_writepage() for writeout of
 ordered data
Message-ID: <20221201115500.kbxtteft3v4pzqqx@quack3>
References: <20221130162435.2324-1-jack@suse.cz>
 <20221201114205.mg6song3ulrqvt54@riteshh-domain>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20221201114205.mg6song3ulrqvt54@riteshh-domain>
Precedence: bulk

On Thu 01-12-22 17:12:05, Ritesh Harjani (IBM) wrote:
> On 22/11/30 05:35PM, Jan Kara wrote:
> > Hello,
> >
> > this patch series modifies ext4 so that we stop using ext4_writepage() for
> > writeout of ordered data during transaction commit (through
> > generic_writepages() from jbd2_journal_submit_inode_data_buffers()). Instead we
> > directly call ext4_writepages() from the
> > ext4_journal_submit_inode_data_buffers().
> 
> Hello Jan,
> 
> Do you think we should add a WARN_ON_ONCE() or something in
> ext4_do_writepages() function where we might try to start a transaction
> at [J]. Since we can now enter into ext4_do_writepages() from two places:
> 1. writeback
> 2. jbd2_journal_commit_transaction()
> 	mpage_submit_page
> 	mpage_prepare_extent_to_map
> 	ext4_do_writepages
> 	ext4_normal_submit_inode_data_buffers
> 	ext4_journal_submit_inode_data_buffers
> 	journal_submit_data_buffers
> 	jbd2_journal_commit_transaction
> 	kjournald2
> 
> So IIUC, we will call mpage_submit_page() in the first call to
> mpage_prepare_extent_to_map() [1] itself. That may set mpd->scanned_until_end = 1
> at the end of it. So then we should never enter into the while loop where we
> start a journal txn.

Yes, correct. But I agree a WARN_ON_ONCE in the loop might be useful just
in case. I'll add it. Thanks for the idea.

								Honza


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