On Wed, 23 May 2012 20:38:16 +0800 "majianpeng" <majianpeng@xxxxxxxxx> wrote:
> It isn't necessary to judge.Because the atomic_read and atomic_set is excuce same cpu. So
> read_errors is zero or not, directly set it zero.
'read' imposes less bus contention than 'write'. And read_errors will most
often be 0. So if we can avoid lots of unnecessary writes we should.
i.e. code is correct as it is.
Thanks,
NeilBrown
> Signed-off-by: majianpeng <majianpeng@xxxxxxxxx>
> ---
> drivers/md/raid5.c | 3 +--
> 1 files changed, 1 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
> index f351422..1fe2def 100644
> --- a/drivers/md/raid5.c
> +++ b/drivers/md/raid5.c
> @@ -1690,8 +1690,7 @@ static void raid5_end_read_request(struct bio * bi, int error)
> clear_bit(R5_ReadError, &sh->dev[i].flags);
> clear_bit(R5_ReWrite, &sh->dev[i].flags);
> }
> - if (atomic_read(&rdev->read_errors))
> - atomic_set(&rdev->read_errors, 0);
> + atomic_set(&rdev->read_errors, 0);
> } else {
> const char *bdn = bdevname(rdev->bdev, b);
> int retry = 0;
Attachment:
signature.asc
Description: PGP signature
