From: Trond Myklebust <trond.myklebust@fys.uio.no>
To: Neil Brown <neilb@suse.de>
Cc: linux-nfs@vger.kernel.org
Subject: Re: Another possible race in put_rpccred / rpcauth_unhash_cred
Date: Fri, 20 Nov 2009 11:48:46 -0500 [thread overview]
Message-ID: <1258735726.2494.27.camel@localhost> (raw)
In-Reply-To: <19203.18009.945354.446817-wvvUuzkyo1EYVZTmpyfIwg@public.gmane.org>
On Wed, 2009-11-18 at 11:56 +1100, Neil Brown wrote:
> Hi again Trond,
>
> I've been looking at a customer report of an oops in
> rpcauth_lookup_credcache.
> https://bugzilla.novell.com/show_bug.cgi?id=547137
>
> It appears that one of the hash chains in the cache has become
> corrupted.
How about the following patch?
Cheers
Trond
--------------------------------------------------------------------------
RPC: Fix two potential races in put_rpccred
From: Trond Myklebust <Trond.Myklebust@netapp.com>
It is possible for rpcauth_destroy_credcache() to cause the rpc credentials
to be unhashed while put_rpccred is waiting for the rpc_credcache_lock on
another cpu.
Should this happen, then we can end up calling hlist_del_rcu(&cred->cr_hash)
a second time, thus causing list corruption.
Should the credential actually be hashed, it is possible for
rpcauth_lookup_credcache to find and reference it before we get round to
unhashing it. We should therefore check for whether the credential is
referenced after we've unhashed it.
See https://bugzilla.novell.com/show_bug.cgi?id=547137
Reported-by: Neil Brown <neilb@suse.de>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
---
net/sunrpc/auth.c | 39 +++++++++++++++++++++++----------------
1 files changed, 23 insertions(+), 16 deletions(-)
diff --git a/net/sunrpc/auth.c b/net/sunrpc/auth.c
index 54a4e04..7ee6f7e 100644
--- a/net/sunrpc/auth.c
+++ b/net/sunrpc/auth.c
@@ -123,16 +123,19 @@ rpcauth_unhash_cred_locked(struct rpc_cred *cred)
clear_bit(RPCAUTH_CRED_HASHED, &cred->cr_flags);
}
-static void
+static int
rpcauth_unhash_cred(struct rpc_cred *cred)
{
spinlock_t *cache_lock;
+ int ret;
cache_lock = &cred->cr_auth->au_credcache->lock;
spin_lock(cache_lock);
- if (atomic_read(&cred->cr_count) == 0)
+ ret = atomic_read(&cred->cr_count) == 0;
+ if (ret)
rpcauth_unhash_cred_locked(cred);
spin_unlock(cache_lock);
+ return ret;
}
/*
@@ -446,31 +449,35 @@ void
put_rpccred(struct rpc_cred *cred)
{
/* Fast path for unhashed credentials */
- if (test_bit(RPCAUTH_CRED_HASHED, &cred->cr_flags) != 0)
- goto need_lock;
-
- if (!atomic_dec_and_test(&cred->cr_count))
+ if (test_bit(RPCAUTH_CRED_HASHED, &cred->cr_flags) == 0) {
+ if (atomic_dec_and_test(&cred->cr_count))
+ cred->cr_ops->crdestroy(cred);
return;
- goto out_destroy;
-need_lock:
+ }
+
if (!atomic_dec_and_lock(&cred->cr_count, &rpc_credcache_lock))
return;
if (!list_empty(&cred->cr_lru)) {
number_cred_unused--;
list_del_init(&cred->cr_lru);
}
- if (test_bit(RPCAUTH_CRED_UPTODATE, &cred->cr_flags) == 0)
- rpcauth_unhash_cred(cred);
if (test_bit(RPCAUTH_CRED_HASHED, &cred->cr_flags) != 0) {
- cred->cr_expire = jiffies;
- list_add_tail(&cred->cr_lru, &cred_unused);
- number_cred_unused++;
- spin_unlock(&rpc_credcache_lock);
- return;
+ if (test_bit(RPCAUTH_CRED_UPTODATE, &cred->cr_flags) != 0) {
+ cred->cr_expire = jiffies;
+ list_add_tail(&cred->cr_lru, &cred_unused);
+ number_cred_unused++;
+ goto out_nodestroy;
+ }
+ if (!rpcauth_unhash_cred(cred)) {
+ /* We were hashed and someone looked us up... */
+ goto out_nodestroy;
+ }
}
spin_unlock(&rpc_credcache_lock);
-out_destroy:
cred->cr_ops->crdestroy(cred);
+ return;
+out_nodestroy:
+ spin_unlock(&rpc_credcache_lock);
}
EXPORT_SYMBOL_GPL(put_rpccred);
prev parent reply other threads:[~2009-11-20 16:48 UTC|newest]
Thread overview: 3+ messages / expand[flat|nested] mbox.gz Atom feed top
2009-11-18 0:56 Another possible race in put_rpccred / rpcauth_unhash_cred Neil Brown
[not found] ` <19203.18009.945354.446817-wvvUuzkyo1EYVZTmpyfIwg@public.gmane.org>
2009-11-19 15:33 ` Trond Myklebust
2009-11-20 16:48 ` Trond Myklebust [this message]
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=1258735726.2494.27.camel@localhost \
--to=trond.myklebust@fys.uio.no \
--cc=linux-nfs@vger.kernel.org \
--cc=neilb@suse.de \
/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