On Wed, Aug 04, 2021 at 03:41:48PM +0800, Hao Xu wrote:
> The block size of localfs for nfs may be different with nfs itself.
> So it's pointless to test nfs on xattrs size, just remove the special
> judge code.
>
> Fixes: commit da3cdb3b91ca ("common/attr: set MAX_ATTR values correctly for NFS")
> Signed-off-by: Hao Xu <haoxu@xxxxxxxxxxxxxxxxx>
> ---
> common/attr | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/common/attr b/common/attr
> index 42ceab92335a..542dff34bf55 100644
> --- a/common/attr
> +++ b/common/attr
> @@ -253,7 +253,7 @@ _getfattr()
>
> # set maximum total attr space based on fs type
> case "$FSTYP" in
> -xfs|udf|pvfs2|9p|ceph|nfs)
> +xfs|udf|pvfs2|9p|ceph)
> MAX_ATTRS=1000
This makes generic/020 _notrun on nfs, I don't think that's what we
want.
I think MAX_ATTRS is a best-effort guess based on filesystem block size,
it's not a accurate hard limit. And if there's no good way to tell the
fs blocksize on nfs server side, then we could make a conservative
assumption, e.g. the blocksize is 1k, and colculate MAX_ATTRS and
MAX_ATTRVAL_SIZE based on that assumption.
Thanks,
Eryu
> ;;
> *)
> @@ -273,7 +273,7 @@ xfs|udf|btrfs)
> pvfs2)
> MAX_ATTRVAL_SIZE=8192
> ;;
> -9p|ceph|nfs)
> +9p|ceph)
> MAX_ATTRVAL_SIZE=65536
> ;;
> bcachefs)
> --
> 2.24.4
