2008-06-27 18:07:51

by Aneesh Kumar K.V

Subject: [PATCH] jbd2: We shouldnot cap writeback when we are doing a journal commit

A journal commit need to make sure we submit data buffers before
we submit the meta-data buffers in ordered mode. We should
not be looking at BDI_CAP_NO_WRITEBACK when doing a journal commit.

Signed-off-by: Aneesh Kumar K.V <[email protected]>
---
fs/jbd2/commit.c | 3 ---
1 files changed, 0 insertions(+), 3 deletions(-)

diff --git a/fs/jbd2/commit.c b/fs/jbd2/commit.c
index 32ca3c3..f8b3be8 100644
--- a/fs/jbd2/commit.c
+++ b/fs/jbd2/commit.c
@@ -203,9 +203,6 @@ static int journal_submit_inode_data_buffers(struct address_space *mapping)
.for_writepages = 1,
};

- if (!mapping_cap_writeback_dirty(mapping))
- return 0;

2008-06-27 18:19:52

by Eric Sandeen

Subject: Re: [PATCH] jbd2: We shouldnot cap writeback when we are doing a journal commit

Aneesh Kumar K.V wrote:
> A journal commit need to make sure we submit data buffers before
> we submit the meta-data buffers in ordered mode. We should
> not be looking at BDI_CAP_NO_WRITEBACK when doing a journal commit.
>
> Signed-off-by: Aneesh Kumar K.V <[email protected]>

and I don't think we'll ever even see that flag set...

Acked-by: Eric Sandeen <[email protected]>

> ---
> fs/jbd2/commit.c | 3 ---
> 1 files changed, 0 insertions(+), 3 deletions(-)
>
> diff --git a/fs/jbd2/commit.c b/fs/jbd2/commit.c
> index 32ca3c3..f8b3be8 100644
> --- a/fs/jbd2/commit.c
> +++ b/fs/jbd2/commit.c
> @@ -203,9 +203,6 @@ static int journal_submit_inode_data_buffers(struct address_space *mapping)
> .for_writepages = 1,
> };
>
> - if (!mapping_cap_writeback_dirty(mapping))
> - return 0;
> -
> ret = generic_writepages(mapping, &wbc);
> return ret;
> }

2008-06-27 21:04:40

by Jan Kara

Subject: Re: [PATCH] jbd2: We shouldnot cap writeback when we are doing a journal commit

On Fri 27-06-08 13:11:25, Eric Sandeen wrote:
> Aneesh Kumar K.V wrote:
> > A journal commit need to make sure we submit data buffers before
> > we submit the meta-data buffers in ordered mode. We should
> > not be looking at BDI_CAP_NO_WRITEBACK when doing a journal commit.
> >
> > Signed-off-by: Aneesh Kumar K.V <[email protected]>
>
> and I don't think we'll ever even see that flag set...
>
> Acked-by: Eric Sandeen <[email protected]>
Yes, I also don't think it really matters :). So feel free to remove that
check.

Honza
>
> > ---
> > fs/jbd2/commit.c | 3 ---
> > 1 files changed, 0 insertions(+), 3 deletions(-)
> >
> > diff --git a/fs/jbd2/commit.c b/fs/jbd2/commit.c
> > index 32ca3c3..f8b3be8 100644
> > --- a/fs/jbd2/commit.c
> > +++ b/fs/jbd2/commit.c
> > @@ -203,9 +203,6 @@ static int journal_submit_inode_data_buffers(struct address_space *mapping)
> > .for_writepages = 1,
> > };
> >
> > - if (!mapping_cap_writeback_dirty(mapping))
> > - return 0;
> > -
> > ret = generic_writepages(mapping, &wbc);
> > return ret;
> > }
>
--
Jan Kara <[email protected]>
SUSE Labs, CR

2008-06-27 23:52:39

by Mingming Cao

Subject: Re: [PATCH] jbd2: We shouldnot cap writeback when we are doing a journal commit

I fold this change to the parent patch: delalloc-new-ordered-mode.patch
in patch queue

Mingming
On Fri, 2008-06-27 at 23:04 +0200, Jan Kara wrote:
> On Fri 27-06-08 13:11:25, Eric Sandeen wrote:
> > Aneesh Kumar K.V wrote:
> > > A journal commit need to make sure we submit data buffers before
> > > we submit the meta-data buffers in ordered mode. We should
> > > not be looking at BDI_CAP_NO_WRITEBACK when doing a journal commit.
> > >
> > > Signed-off-by: Aneesh Kumar K.V <[email protected]>
> >
> > and I don't think we'll ever even see that flag set...
> >
> > Acked-by: Eric Sandeen <[email protected]>
> Yes, I also don't think it really matters :). So feel free to remove that
> check.
>
> Honza
> >
> > > ---
> > > fs/jbd2/commit.c | 3 ---
> > > 1 files changed, 0 insertions(+), 3 deletions(-)
> > >
> > > diff --git a/fs/jbd2/commit.c b/fs/jbd2/commit.c
> > > index 32ca3c3..f8b3be8 100644
> > > --- a/fs/jbd2/commit.c
> > > +++ b/fs/jbd2/commit.c
> > > @@ -203,9 +203,6 @@ static int journal_submit_inode_data_buffers(struct address_space *mapping)
> > > .for_writepages = 1,
> > > };
> > >
> > > - if (!mapping_cap_writeback_dirty(mapping))
> > > - return 0;
> > > -
> > > ret = generic_writepages(mapping, &wbc);
> > > return ret;
> > > }
> >