On Wed, Sep 16, 2020 at 07:19:09PM +0800, xiakaixu1987@xxxxxxxxx wrote:
> From: Kaixu Xia <kaixuxia@xxxxxxxxxxx>
>
> We can do the assert directly for the return value of xfs_mod_fdblocks()
> function, and the variable error is unnecessary, so remove it.
>
> Signed-off-by: Kaixu Xia <kaixuxia@xxxxxxxxxxx>
> ---
> fs/xfs/xfs_trans.c | 7 ++-----
> 1 file changed, 2 insertions(+), 5 deletions(-)
>
> diff --git a/fs/xfs/xfs_trans.c b/fs/xfs/xfs_trans.c
> index d33d0ba6f3bd..caa207220e2c 100644
> --- a/fs/xfs/xfs_trans.c
> +++ b/fs/xfs/xfs_trans.c
> @@ -573,7 +573,6 @@ xfs_trans_unreserve_and_mod_sb(
> int64_t rtxdelta = 0;
> int64_t idelta = 0;
> int64_t ifreedelta = 0;
> - int error;
>
> /* calculate deltas */
> if (tp->t_blk_res > 0)
> @@ -596,10 +595,8 @@ xfs_trans_unreserve_and_mod_sb(
> }
>
> /* apply the per-cpu counters */
> - if (blkdelta) {
> - error = xfs_mod_fdblocks(mp, blkdelta, rsvd);
> - ASSERT(!error);
> - }
> + if (blkdelta)
> + ASSERT(!xfs_mod_fdblocks(mp, blkdelta, rsvd));
Um.... did you test this with ASSERTs disabled? Because this compiles
the free block counter update out of the function on non-debug kernels,
which (AFAICT) will cause fs corruption...
--D
>
> if (idelta) {
> percpu_counter_add_batch(&mp->m_icount, idelta,
> --
> 2.20.0
>
