Andreas Dilger <adilger(a)dilger.ca> writes:
On Aug 14, 2015, at 2:15 PM, Jeff Moyer <jmoyer(a)redhat.com>
> The dax code doesn't currently support misaligned partitions,
> so disable O_DIRECT via dax until such time as that support
> Suggested-by: Boaz Harrosh <boaz(a)plexistor.com>
> Signed-off-by: Jeff Moyer <jmoyer(a)redhat.com>
> fs/block_dev.c | 7 +++++++
> 1 file changed, 7 insertions(+)
> diff --git a/fs/block_dev.c b/fs/block_dev.c
> index 1982437..1170f8c 100644
> --- a/fs/block_dev.c
> +++ b/fs/block_dev.c
> @@ -1241,6 +1241,13 @@ static int __blkdev_get(struct block_device *bdev, fmode_t
mode, int for_part)
> goto out_clear;
> bd_set_size(bdev, (loff_t)bdev->bd_part->nr_sects << 9);
> + /*
> + * If the partition is not aligned on a page
> + * boundary, we can't do dax I/O to it.
> + */
> + if ((bdev->bd_part->start_sect % (PAGE_SIZE / 512)) ||
> + (bdev->bd_part->nr_sects % (PAGE_SIZE / 512)))
Maybe I'm missing something, but doesn't the second condition above
disable DAX for the case that the 1/2 patch is fixing (i.e. the last
sectors at the end of a non-PAGE_SIZE-multiple device)? It seems a
shame to disable DAX for the whole device because of the last sector.
No. Patch 1/2 fixes a 512 byte read of the last sector of a properly
aligned partiton. The goal is to eventually fix things so we can enable
the dax path for misaligned partitions, but it's not going to happen in
time for 4.2. Also, keep in mind that this is just for opening the
block device itself with O_DIRECT.
Thanks for taking a look.