Re: [PATCH v2] nfsd: Always lock state exclusively.

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

On Tue, Jun 14, 2016 at 11:53:27AM -0400, Oleg Drokin wrote:
> 
> On Jun 14, 2016, at 11:38 AM, J . Bruce Fields wrote:
> 
> > On Sun, Jun 12, 2016 at 09:26:27PM -0400, Oleg Drokin wrote:
> >> It used to be the case that state had an rwlock that was locked for write
> >> by downgrades, but for read for upgrades (opens). Well, the problem is
> >> if there are two competing opens for the same state, they step on
> >> each other toes potentially leading to leaking file descriptors
> >> from the state structure, since access mode is a bitmap only set once.
> >> 
> >> Extend the holding region around in nfsd4_process_open2() to avoid
> >> racing entry into nfs4_get_vfs_file().
> >> Make init_open_stateid() return with locked stateid to be unlocked
> >> by the caller.
> >> 
> >> Now this version held up pretty well in my testing for 24 hours.
> >> It still does not address the situation if during one of the racing
> >> nfs4_get_vfs_file() calls we are getting an error from one (first?)
> >> of them. This is to be addressed in a separate patch after having a
> >> solid reproducer (potentially using some fault injection).
> >> 
> >> Signed-off-by: Oleg Drokin <green@linuxhacker.ru>
> >> ---
> >> fs/nfsd/nfs4state.c | 47 +++++++++++++++++++++++++++--------------------
> >> fs/nfsd/state.h     |  2 +-
> >> 2 files changed, 28 insertions(+), 21 deletions(-)
> >> 
> >> diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c
> >> index f5f82e1..fa5fb5a 100644
> >> --- a/fs/nfsd/nfs4state.c
> >> +++ b/fs/nfsd/nfs4state.c
> >> @@ -3487,6 +3487,10 @@ init_open_stateid(struct nfs4_ol_stateid *stp, struct nfs4_file *fp,
> >> 	struct nfs4_openowner *oo = open->op_openowner;
> >> 	struct nfs4_ol_stateid *retstp = NULL;
> >> 
> >> +	/* We are moving these outside of the spinlocks to avoid the warnings */
> >> +	mutex_init(&stp->st_mutex);
> >> +	mutex_lock(&stp->st_mutex);
> >> +
> >> 	spin_lock(&oo->oo_owner.so_client->cl_lock);
> >> 	spin_lock(&fp->fi_lock);
> >> 
> >> @@ -3502,13 +3506,14 @@ init_open_stateid(struct nfs4_ol_stateid *stp, struct nfs4_file *fp,
> >> 	stp->st_access_bmap = 0;
> >> 	stp->st_deny_bmap = 0;
> >> 	stp->st_openstp = NULL;
> >> -	init_rwsem(&stp->st_rwsem);
> >> 	list_add(&stp->st_perstateowner, &oo->oo_owner.so_stateids);
> >> 	list_add(&stp->st_perfile, &fp->fi_stateids);
> >> 
> >> out_unlock:
> >> 	spin_unlock(&fp->fi_lock);
> >> 	spin_unlock(&oo->oo_owner.so_client->cl_lock);
> >> +	if (retstp)
> >> +		mutex_lock(&retstp->st_mutex);
> >> 	return retstp;
> > 
> > You're returning with both stp->st_mutex and retstp->st_mutex locked.
> > Did you mean to drop that first lock in the (retstp) case, or am I
> > missing something?
> 
> Well, I think it's ok (perhaps worthy of a comment) it's that if we matched a different
> retstp state, then stp is not used and either released right away or even
> if reused, it would be reinitialized in another call to init_open_stateid(),
> so it's fine?

Oh, I see, you're right.

Though I wouldn't have been surprised if that triggered some kind of
warning--I guess it's OK here, but typically if I saw a structure freed
that had a locked lock in it I'd be a little suspicious that somebody
made a mistake.

--b.