From: Trond Myklebust <trond.myklebust@primarydata.com>
To: anna.schumaker@netapp.com
Cc: linux-nfs@vger.kernel.org, Oleg Drokin <green@linuxhacker.ru>
Subject: [PATCH v5 01/25] NFSv4.1: Don't deadlock the state manager on the SEQUENCE status flags
Date: Sat, 17 Sep 2016 01:13:09 -0400 [thread overview]
Message-ID: <1474089213-104014-2-git-send-email-trond.myklebust@primarydata.com> (raw)
In-Reply-To: <1474089213-104014-1-git-send-email-trond.myklebust@primarydata.com>
As described in RFC5661, section 18.46, some of the status flags exist
in order to tell the client when it needs to acknowledge the existence of
revoked state on the server and/or to recover state.
Those flags will then remain set until the recovery procedure is done.
In order to avoid looping, the client therefore needs to ignore
those particular flags while recovering.
Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
---
fs/nfs/nfs4_fs.h | 2 +-
fs/nfs/nfs4proc.c | 5 ++++-
fs/nfs/nfs4session.h | 1 +
fs/nfs/nfs4state.c | 12 +++++++++++-
4 files changed, 17 insertions(+), 3 deletions(-)
diff --git a/fs/nfs/nfs4_fs.h b/fs/nfs/nfs4_fs.h
index f230aa62ca59..4390d73a92e5 100644
--- a/fs/nfs/nfs4_fs.h
+++ b/fs/nfs/nfs4_fs.h
@@ -439,7 +439,7 @@ extern void nfs4_schedule_path_down_recovery(struct nfs_client *clp);
extern int nfs4_schedule_stateid_recovery(const struct nfs_server *, struct nfs4_state *);
extern int nfs4_schedule_migration_recovery(const struct nfs_server *);
extern void nfs4_schedule_lease_moved_recovery(struct nfs_client *);
-extern void nfs41_handle_sequence_flag_errors(struct nfs_client *clp, u32 flags);
+extern void nfs41_handle_sequence_flag_errors(struct nfs_client *clp, u32 flags, bool);
extern void nfs41_handle_server_scope(struct nfs_client *,
struct nfs41_server_scope **);
extern void nfs4_put_lock_state(struct nfs4_lock_state *lsp);
diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index 251e48e7ba16..6b700c59eede 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -616,6 +616,7 @@ int nfs40_setup_sequence(struct nfs4_slot_table *tbl,
}
spin_unlock(&tbl->slot_tbl_lock);
+ slot->privileged = args->sa_privileged ? 1 : 0;
args->sa_slot = slot;
res->sr_slot = slot;
@@ -728,7 +729,8 @@ static int nfs41_sequence_process(struct rpc_task *task,
clp = session->clp;
do_renew_lease(clp, res->sr_timestamp);
/* Check sequence flags */
- nfs41_handle_sequence_flag_errors(clp, res->sr_status_flags);
+ nfs41_handle_sequence_flag_errors(clp, res->sr_status_flags,
+ !!slot->privileged);
nfs41_update_target_slotid(slot->table, slot, res);
break;
case 1:
@@ -875,6 +877,7 @@ int nfs41_setup_sequence(struct nfs4_session *session,
}
spin_unlock(&tbl->slot_tbl_lock);
+ slot->privileged = args->sa_privileged ? 1 : 0;
args->sa_slot = slot;
dprintk("<-- %s slotid=%u seqid=%u\n", __func__,
diff --git a/fs/nfs/nfs4session.h b/fs/nfs/nfs4session.h
index 3bb6af70973c..dae385500005 100644
--- a/fs/nfs/nfs4session.h
+++ b/fs/nfs/nfs4session.h
@@ -23,6 +23,7 @@ struct nfs4_slot {
u32 slot_nr;
u32 seq_nr;
unsigned int interrupted : 1,
+ privileged : 1,
seq_done : 1;
};
diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c
index cada00aa5096..9801b5bb5fac 100644
--- a/fs/nfs/nfs4state.c
+++ b/fs/nfs/nfs4state.c
@@ -2227,13 +2227,22 @@ static void nfs41_handle_cb_path_down(struct nfs_client *clp)
nfs4_schedule_state_manager(clp);
}
-void nfs41_handle_sequence_flag_errors(struct nfs_client *clp, u32 flags)
+void nfs41_handle_sequence_flag_errors(struct nfs_client *clp, u32 flags,
+ bool recovery)
{
if (!flags)
return;
dprintk("%s: \"%s\" (client ID %llx) flags=0x%08x\n",
__func__, clp->cl_hostname, clp->cl_clientid, flags);
+ /*
+ * If we're called from the state manager thread, then assume we're
+ * already handling the RECLAIM_NEEDED and/or STATE_REVOKED.
+ * Those flags are expected to remain set until we're done
+ * recovering (see RFC5661, section 18.46.3).
+ */
+ if (recovery)
+ goto out_recovery;
if (flags & SEQ4_STATUS_RESTART_RECLAIM_NEEDED)
nfs41_handle_server_reboot(clp);
@@ -2246,6 +2255,7 @@ void nfs41_handle_sequence_flag_errors(struct nfs_client *clp, u32 flags)
nfs4_schedule_lease_moved_recovery(clp);
if (flags & SEQ4_STATUS_RECALLABLE_STATE_REVOKED)
nfs41_handle_recallable_state_revoked(clp);
+out_recovery:
if (flags & SEQ4_STATUS_BACKCHANNEL_FAULT)
nfs41_handle_backchannel_fault(clp);
else if (flags & (SEQ4_STATUS_CB_PATH_DOWN |
--
2.7.4
next prev parent reply other threads:[~2016-09-17 5:13 UTC|newest]
Thread overview: 39+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-09-17 5:13 [PATCH v5 00/25] Fix delegation behaviour when server revokes some state Trond Myklebust
2016-09-17 5:13 ` Trond Myklebust [this message]
2016-09-17 5:13 ` [PATCH v5 02/25] NFSv4: Don't report revoked delegations as valid in nfs_have_delegation() Trond Myklebust
2016-09-17 5:13 ` [PATCH v5 03/25] NFSv4.1: Don't check delegations that are already marked as revoked Trond Myklebust
2016-09-17 5:13 ` [PATCH v5 04/25] NFSv4.1: Allow test_stateid to handle session errors without waiting Trond Myklebust
2016-09-17 5:13 ` [PATCH v5 05/25] NFSv4.1: Add a helper function to deal with expired stateids Trond Myklebust
2016-09-17 5:13 ` [PATCH v5 06/25] NFSv4.x: Allow callers of nfs_remove_bad_delegation() to specify a stateid Trond Myklebust
2016-09-17 5:13 ` [PATCH v5 07/25] NFSv4.1: Test delegation stateids when server declares "some state revoked" Trond Myklebust
2016-09-17 5:13 ` [PATCH v5 08/25] NFSv4.1: Deal with server reboots during delegation expiration recovery Trond Myklebust
2016-09-17 5:13 ` [PATCH v5 09/25] NFSv4.1: Don't recheck delegations that have already been checked Trond Myklebust
2016-09-17 5:13 ` [PATCH v5 10/25] NFSv4.1: Allow revoked stateids to skip the call to TEST_STATEID Trond Myklebust
2016-09-17 5:13 ` [PATCH v5 11/25] NFSv4.1: Ensure we always run TEST/FREE_STATEID on locks Trond Myklebust
2016-09-17 5:13 ` [PATCH v5 12/25] NFSv4.1: FREE_STATEID can be asynchronous Trond Myklebust
2016-09-17 5:13 ` [PATCH v5 13/25] NFSv4.1: Ensure we call FREE_STATEID if needed on close/delegreturn/locku Trond Myklebust
2016-09-17 5:13 ` [PATCH v5 14/25] NFSv4: Ensure we don't re-test revoked and freed stateids Trond Myklebust
2016-09-17 5:13 ` [PATCH v5 15/25] NFSv4: nfs_inode_find_delegation_state_and_recover() should check all stateids Trond Myklebust
2016-09-17 5:13 ` [PATCH v5 16/25] NFSv4: nfs4_do_handle_exception() handle revoke/expiry of a single stateid Trond Myklebust
2016-09-17 5:13 ` [PATCH v5 17/25] NFSv4: nfs4_handle_delegation_recall_error() handle expiration as revoke case Trond Myklebust
2016-09-17 5:13 ` [PATCH v5 18/25] NFSv4: nfs4_handle_setlk_error() " Trond Myklebust
2016-09-17 5:13 ` [PATCH v5 19/25] NFSv4.1: nfs4_layoutget_handle_exception handle revoked state Trond Myklebust
2016-09-17 5:13 ` [PATCH v5 20/25] NFSv4: Pass the stateid to the exception handler in nfs4_read/write_done_cb Trond Myklebust
2016-09-17 5:13 ` [PATCH v5 21/25] NFSv4: Fix a race in nfs_inode_reclaim_delegation() Trond Myklebust
2016-09-17 5:13 ` [PATCH v5 22/25] NFSv4: Fix a race when updating an open_stateid Trond Myklebust
2016-09-17 5:13 ` [PATCH v5 23/25] NFS: Always call nfs_inode_find_state_and_recover() when revoking a delegation Trond Myklebust
2016-09-17 5:13 ` [PATCH v5 24/25] NFSv4: Don't test open_stateid unless it is set Trond Myklebust
2016-09-17 5:13 ` [PATCH v5 25/25] NFSv4: Mark the lock and open stateids as invalid after freeing them Trond Myklebust
2016-09-17 18:04 ` [PATCH v5 00/25] Fix delegation behaviour when server revokes some state Oleg Drokin
2016-09-17 18:18 ` Trond Myklebust
2016-09-17 19:16 ` Oleg Drokin
2016-09-17 19:32 ` Trond Myklebust
2016-09-17 21:45 ` Trond Myklebust
2016-09-17 21:55 ` Oleg Drokin
2016-09-17 21:59 ` Trond Myklebust
2016-09-18 1:19 ` Oleg Drokin
2016-09-18 16:23 ` Oleg Drokin
2016-09-18 22:38 ` Trond Myklebust
2016-09-18 22:44 ` Oleg Drokin
2016-09-18 23:08 ` Trond Myklebust
2016-09-18 23:30 ` Oleg Drokin
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1474089213-104014-2-git-send-email-trond.myklebust@primarydata.com \
--to=trond.myklebust@primarydata.com \
--cc=anna.schumaker@netapp.com \
--cc=green@linuxhacker.ru \
--cc=linux-nfs@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).