[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20200115090846.GB31450@quack2.suse.cz>
Date: Wed, 15 Jan 2020 10:08:46 +0100
From: Jan Kara <jack@...e.cz>
To: Christoph Hellwig <hch@...radead.org>
Cc: Jan Kara <jack@...e.cz>, Ritesh Harjani <riteshh@...ux.ibm.com>,
linux-ext4@...r.kernel.org, tytso@....edu
Subject: Re: [RFC 1/2] iomap: direct-io: Move inode_dio_begin before
filemap_write_and_wait_range
On Tue 14-01-20 10:27:36, Christoph Hellwig wrote:
> On Tue, Jan 14, 2020 at 06:19:34PM +0100, Jan Kara wrote:
> > We want to detect in the writeback path whether there's direct IO (read)
> > currently running for the inode. Not for the writeback issued from
> > iomap_dio_rw() but for any arbitrary writeback that iomap_dio_rw() can be
> > racing with - so struct writeback_control won't help. Now if you want to
> > see the ugly details why this hack is needed, see my other email to Ritesh
> > in this thread with details of the race.
>
> How do we get other writeback after iomap_dio_rw wrote everything out?
You create dirty page using mmap in the range read by iomap_dio_rw() and then
background writeback happens at unfortunate time... Email [1] has the exact
traces.
> Either way I'm trying to kill i_dio_count as it has all kinds of
> problems, see the patch sent out earlier today.
OK, I'll see that patch.
Honza
[1] https://lore.kernel.org/linux-ext4/20200114094741.GC6466@quack2.suse.cz
--
Jan Kara <jack@...e.com>
SUSE Labs, CR
Powered by blists - more mailing lists