Re: [PATCH v9 06/10] iomap: fix iomap_dio_zero() for fs bs > system page size

[Dave Chinner <david@fromorbit.com>]

On Thu, Jul 04, 2024 at 05:37:32PM +0200, Hannes Reinecke wrote:
> On 7/4/24 13:23, Pankaj Raghav (Samsung) wrote:
> > From: Pankaj Raghav <p.raghav@samsung.com>
> > 
> > iomap_dio_zero() will pad a fs block with zeroes if the direct IO size
> > < fs block size. iomap_dio_zero() has an implicit assumption that fs block
> > size < page_size. This is true for most filesystems at the moment.
> > 
> > If the block size > page size, this will send the contents of the page
> > next to zero page(as len > PAGE_SIZE) to the underlying block device,
> > causing FS corruption.
> > 
> > iomap is a generic infrastructure and it should not make any assumptions
> > about the fs block size and the page size of the system.
> > 
> > Signed-off-by: Pankaj Raghav <p.raghav@samsung.com>
> > ---
> >   fs/iomap/buffered-io.c |  4 ++--
> >   fs/iomap/direct-io.c   | 45 ++++++++++++++++++++++++++++++++++++------
> >   2 files changed, 41 insertions(+), 8 deletions(-)
> > 
> > diff --git a/fs/iomap/buffered-io.c b/fs/iomap/buffered-io.c
> > index f420c53d86acc..d745f718bcde8 100644
> > --- a/fs/iomap/buffered-io.c
> > +++ b/fs/iomap/buffered-io.c
> > @@ -2007,10 +2007,10 @@ iomap_writepages(struct address_space *mapping, struct writeback_control *wbc,
> >   }
> >   EXPORT_SYMBOL_GPL(iomap_writepages);
> > -static int __init iomap_init(void)
> > +static int __init iomap_buffered_init(void)
> >   {
> >   	return bioset_init(&iomap_ioend_bioset, 4 * (PAGE_SIZE / SECTOR_SIZE),
> >   			   offsetof(struct iomap_ioend, io_bio),
> >   			   BIOSET_NEED_BVECS);
> >   }
> > -fs_initcall(iomap_init);
> > +fs_initcall(iomap_buffered_init);
> > diff --git a/fs/iomap/direct-io.c b/fs/iomap/direct-io.c
> > index f3b43d223a46e..c02b266bba525 100644
> > --- a/fs/iomap/direct-io.c
> > +++ b/fs/iomap/direct-io.c
> > @@ -11,6 +11,7 @@
> >   #include <linux/iomap.h>
> >   #include <linux/backing-dev.h>
> >   #include <linux/uio.h>
> > +#include <linux/set_memory.h>
> >   #include <linux/task_io_accounting_ops.h>
> >   #include "trace.h"
> > @@ -27,6 +28,13 @@
> >   #define IOMAP_DIO_WRITE		(1U << 30)
> >   #define IOMAP_DIO_DIRTY		(1U << 31)
> > +/*
> > + * Used for sub block zeroing in iomap_dio_zero()
> > + */
> > +#define IOMAP_ZERO_PAGE_SIZE (SZ_64K)
> > +#define IOMAP_ZERO_PAGE_ORDER (get_order(IOMAP_ZERO_PAGE_SIZE))
> > +static struct page *zero_page;
> > +
> 
> There are other users of ZERO_PAGE, most notably in fs/direct-io.c and
> block/blk-lib.c. Any chance to make this available to them?

Please, no.

We need to stop feature creeping this patchset and bring it to a
close. If changing code entirely unrelated to this patchset is
desired, please do it as a separate independent set of patches.

-Dave.
-- 
Dave Chinner
david@fromorbit.com