Re: [PATCH 4.9 089/116] btrfs: flush write bio if we loop in extent_write_cache_pages

[Sasha Levin]

On Thu, Feb 13, 2020 at 10:01:09PM +0100, David Sterba wrote:
> On Thu, Feb 13, 2020 at 07:20:33AM -0800, Greg Kroah-Hartman wrote:
> > From: Josef Bacik <josef@xxxxxxxxxxxxxx>
> >
> > [ Upstream commit 96bf313ecb33567af4cb53928b0c951254a02759 ]
>
> This commit does not exist in my tree, the correct upstream commit of
> the backported patch should be 42ffb0bf584ae5b6b38f72259af1e0ee417ac77f.
>
> > --- a/fs/btrfs/extent_io.c
> > +++ b/fs/btrfs/extent_io.c
> > @@ -4060,6 +4060,14 @@ static int extent_write_cache_pages(struct extent_io_tree *tree,
> >  		 */
> >  		scanned = 1;
> >  		index = 0;
> > +
> > +		/*
> > +		 * If we're looping we could run into a page that is locked by a
> > +		 * writer and that writer could be waiting on writeback for a
> > +		 * page in our current bio, and thus deadlock, so flush the
> > +		 * write bio here.
> > +		 */
> > +		flush_write_bio(data);
>
> This has been modified to apply, flush_write_bio does not return a value
> in 4.9, perhaps this led to the different commit id.

Ah yes. This patch was tricky because there were two "big" things that
broke on the backport:

1. flush_write_bio() returning a value.
2. The 'data' -> 'epd' rename.

96bf313ecb3 comes from when both of the above conditions overlapped, and
I forgot that I'm basing my second backport change on something that was
already modified.

--
Thanks,
Sasha