Re: [PATCH v2 0/7] Tighten up locking around stateids in knfsd

["J. Bruce Fields" <bfields@fieldses.org>]

What tree are these patches against?

--b.

On Fri, Nov 03, 2017 at 08:00:09AM -0400, Trond Myklebust wrote:
> This patchset aims to tighten up the locking rules around stateids to
> ensure that knfsd does not reuse stateids that have already been closed
> or invalidated. The aim is to ensure we enforce the RFC5661 and RFC7530
> rules concerning stateid initialisation and updates.
> 
> v2:
> - Add a fix for when nfs4_get_vfs_file() fails in nfsd4_process_open2()
> - Add a fix for byte range lock creation
> 
> Note that byte range locks are not completely fixed. The remaining task
> of ensuring that locks don't conflict with CLOSE has been left as an
> exercise for the reviewer.
> 
> Trond Myklebust (7):
>   nfsd: Fix stateid races between OPEN and CLOSE
>   nfsd: Fix another OPEN stateid race
>   nfsd: CLOSE SHOULD return the invalid special stateid for NFSv4.x
>     (x>0)
>   nfsd: Ensure we don't recognise lock stateids after freeing them
>   nfsd: Fix race in lock stateid creation
>   nfsd: Ensure we check stateid validity in the seqid operation checks
>   nfsd: Fix races with check_stateid_generation()
> 
>  fs/nfsd/nfs4state.c | 254 ++++++++++++++++++++++++++++++++++------------------
>  1 file changed, 166 insertions(+), 88 deletions(-)
> 
> -- 
> 2.13.6