From: Trond Myklebust <Trond.Myklebust@netapp.com>
To: linux-nfs@vger.kernel.org
Subject: [PATCH 26/30] SUNRPC: Fix autobind on cloned rpc clients
Date: Tue, 07 Oct 2008 18:19:54 -0400 [thread overview]
Message-ID: <20081007221954.20945.44434.stgit@localhost.localdomain> (raw)
In-Reply-To: <20081007221952.20945.69529.stgit-bi+AKbBUZKY6gyzm1THtWbp2dZbC/Bob@public.gmane.org>
Despite the fact that cloned rpc clients won't have the cl_autobind flag
set, they may still find themselves calling rpcb_getport_async(). For this
to happen, it suffices for a _parent_ rpc_clnt to use autobinding, in which
case any clone may find itself triggering the !xprt_bound() case in
call_bind().
The correct fix for this is to walk back up the tree of cloned rpc clients,
in order to find the parent that 'owns' the transport, either because it
has clnt->cl_autobind set, or because it originally created the
transport...
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
---
net/sunrpc/rpcb_clnt.c | 36 +++++++++++++++++++++++++++++-------
1 files changed, 29 insertions(+), 7 deletions(-)
diff --git a/net/sunrpc/rpcb_clnt.c b/net/sunrpc/rpcb_clnt.c
index 24db2b4..172935b 100644
--- a/net/sunrpc/rpcb_clnt.c
+++ b/net/sunrpc/rpcb_clnt.c
@@ -469,6 +469,28 @@ static struct rpc_task *rpcb_call_async(struct rpc_clnt *rpcb_clnt, struct rpcbi
return rpc_run_task(&task_setup_data);
}
+/*
+ * In the case where rpc clients have been cloned, we want to make
+ * sure that we use the program number/version etc of the actual
+ * owner of the xprt. To do so, we walk back up the tree of parents
+ * to find whoever created the transport and/or whoever has the
+ * autobind flag set.
+ */
+static struct rpc_clnt *rpcb_find_transport_owner(struct rpc_clnt *clnt)
+{
+ struct rpc_clnt *parent = clnt->cl_parent;
+
+ while (parent != clnt) {
+ if (parent->cl_xprt != clnt->cl_xprt)
+ break;
+ if (clnt->cl_autobind)
+ break;
+ clnt = parent;
+ parent = parent->cl_parent;
+ }
+ return clnt;
+}
+
/**
* rpcb_getport_async - obtain the port for a given RPC service on a given host
* @task: task that is waiting for portmapper request
@@ -478,10 +500,10 @@ static struct rpc_task *rpcb_call_async(struct rpc_clnt *rpcb_clnt, struct rpcbi
*/
void rpcb_getport_async(struct rpc_task *task)
{
- struct rpc_clnt *clnt = task->tk_client;
+ struct rpc_clnt *clnt;
struct rpc_procinfo *proc;
u32 bind_version;
- struct rpc_xprt *xprt = task->tk_xprt;
+ struct rpc_xprt *xprt;
struct rpc_clnt *rpcb_clnt;
static struct rpcbind_args *map;
struct rpc_task *child;
@@ -490,13 +512,13 @@ void rpcb_getport_async(struct rpc_task *task)
size_t salen;
int status;
+ clnt = rpcb_find_transport_owner(task->tk_client);
+ xprt = clnt->cl_xprt;
+
dprintk("RPC: %5u %s(%s, %u, %u, %d)\n",
task->tk_pid, __func__,
clnt->cl_server, clnt->cl_prog, clnt->cl_vers, xprt->prot);
- /* Autobind on cloned rpc clients is discouraged */
- BUG_ON(clnt->cl_parent != clnt);
-
/* Put self on the wait queue to ensure we get notified if
* some other task is already attempting to bind the port */
rpc_sleep_on(&xprt->binding, task, NULL);
@@ -578,9 +600,9 @@ void rpcb_getport_async(struct rpc_task *task)
task->tk_pid, __func__);
return;
}
- rpc_put_task(child);
- task->tk_xprt->stat.bind_count++;
+ xprt->stat.bind_count++;
+ rpc_put_task(child);
return;
bailout_nofree:
next prev parent reply other threads:[~2008-10-07 22:32 UTC|newest]
Thread overview: 42+ messages / expand[flat|nested] mbox.gz Atom feed top
2008-10-07 22:19 [PATCH 00/30] What's in the NFS queue for 2.6.27 Trond Myklebust
[not found] ` <20081007221952.20945.69529.stgit-bi+AKbBUZKY6gyzm1THtWbp2dZbC/Bob@public.gmane.org>
2008-10-07 22:19 ` [PATCH 02/30] NFS: Clean up nfs_sb_active/nfs_sb_deactive Trond Myklebust
2008-10-07 22:19 ` [PATCH 01/30] NFS: Fix nfs_file_llseek() Trond Myklebust
2008-10-07 22:19 ` [PATCH 07/30] NFS: Fix the NFS attribute update Trond Myklebust
2008-10-07 22:19 ` [PATCH 14/30] nfs: ERR_PTR is expected on failure from nfs_do_clone_mount Trond Myklebust
2008-10-07 22:19 ` [PATCH 16/30] nfs: BUG_ON in nfs_follow_mountpoint Trond Myklebust
2008-10-07 22:19 ` [PATCH 03/30] NFS: Add options for finer control of the lookup cache Trond Myklebust
2008-10-07 22:19 ` [PATCH 13/30] fix fs/nfs/nfsroot.c compilation Trond Myklebust
2008-10-07 22:19 ` [PATCH 10/30] NFS: Don't clear nfsi->cache_validity in nfs_check_inode_attributes() Trond Myklebust
2008-10-07 22:19 ` [PATCH 08/30] NFS: Fix nfs_post_op_update_inode_force_wcc() Trond Myklebust
2008-10-07 22:19 ` [PATCH 20/30] nfs: break up nfs_follow_referral Trond Myklebust
2008-10-07 22:19 ` [PATCH 22/30] nfs: prepare to share nfs_set_port Trond Myklebust
2008-10-07 22:19 ` [PATCH 04/30] NFS: Don't apply NFS_MOUNT_FLAGMASK to text-based mounts Trond Myklebust
2008-10-07 22:19 ` [PATCH 06/30] NFS: Clean up nfs_refresh_inode() and nfs_post_op_update_inode() Trond Myklebust
2008-10-07 22:19 ` [PATCH 09/30] NFS: Convert __nfs_revalidate_inode() to use nfs_refresh_inode() Trond Myklebust
2008-10-07 22:19 ` [PATCH 15/30] sunrpc: do not pin sunrpc module in the memory Trond Myklebust
2008-10-07 22:19 ` [PATCH 19/30] nfs: authenticated deep mounting Trond Myklebust
2008-10-07 22:19 ` [PATCH 12/30] NFS: Allow concurrent inode revalidation Trond Myklebust
2008-10-07 22:19 ` [PATCH 18/30] NFS: missing nfs_fattr_init in nfs3_proc_getacl and nfs3_proc_setacls (resend #2) Trond Myklebust
2008-10-07 22:19 ` [PATCH 21/30] nfs: replace while loop by for loops in nfs_follow_referral Trond Myklebust
2008-10-07 22:19 ` [PATCH 05/30] NFS: Add mount options for controlling the lookup cache Trond Myklebust
2008-10-07 22:19 ` [PATCH 11/30] NFS: Fix up nfs_setattr_update_inode() Trond Myklebust
2008-10-07 22:19 ` [PATCH 17/30] nfs: remove an obsolete nfs_flock comment Trond Myklebust
2008-10-07 22:19 ` [PATCH 25/30] NFS: SETCLIENTID truncates client ID and netid Trond Myklebust
[not found] ` <20081007221954.20945.76616.stgit-bi+AKbBUZKY6gyzm1THtWbp2dZbC/Bob@public.gmane.org>
2008-10-08 14:55 ` Chuck Lever
2008-10-08 17:56 ` Trond Myklebust
2008-10-15 15:36 ` Chuck Lever
2008-10-07 22:19 ` [PATCH 30/30] sunrpc: fix oops in rpc_create when the mount namespace is unshared Trond Myklebust
2008-10-07 22:19 ` [PATCH 23/30] nfs: Fix misparsing of nfsv4 fs_locations attribute Trond Myklebust
2008-10-07 22:19 ` [PATCH 24/30] NFS: remove 8 bytes of padding from struct nfs_fattr on 64 bit builds Trond Myklebust
2008-10-07 22:19 ` Trond Myklebust [this message]
2008-10-07 22:19 ` [PATCH 28/30] NFS: Client mounts hang when exported directory do not exist Trond Myklebust
2008-10-07 22:19 ` [PATCH 29/30] NFS: Don't use range_cyclic for data integrity syncs Trond Myklebust
2008-10-07 22:19 ` [PATCH 27/30] SUNRPC: Fix a memory leak in rpcb_getport_async Trond Myklebust
2008-10-08 19:31 ` [PATCH 00/30] What's in the NFS queue for 2.6.27 J. Bruce Fields
2008-10-08 19:37 ` Trond Myklebust
2008-10-08 19:39 ` J. Bruce Fields
2008-10-08 19:38 ` [PATCH 1/5] NFS: fix nfs_parse_ip_address() corner case J. Bruce Fields
2008-10-08 19:38 ` [PATCH 2/5] nfs: break up nfs_follow_referral J. Bruce Fields
2008-10-08 19:38 ` [PATCH 3/5] nfs: replace while loop by for loops in nfs_follow_referral J. Bruce Fields
2008-10-08 19:38 ` [PATCH 4/5] nfs: prepare to share nfs_set_port J. Bruce Fields
2008-10-08 19:38 ` [PATCH 5/5] nfs: Fix misparsing of nfsv4 fs_locations attribute J. Bruce Fields
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=20081007221954.20945.44434.stgit@localhost.localdomain \
--to=trond.myklebust@netapp.com \
--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