linux-kernel - Re: NFSv4 regression, kernel BUG at fs/nfsd/nfs4state.c:1044!

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Tue, 12 Jun 2012 08:36:44 -0400
From:	"J. Bruce Fields" <bfields@...ldses.org>
To:	"J. Bruce Fields" <bfields@...hat.com>
Cc:	linux-nfs@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: NFSv4 regression, kernel BUG at fs/nfsd/nfs4state.c:1044!

On Tue, Jun 12, 2012 at 07:18:03AM -0400, J. Bruce Fields wrote:
> On Tue, Jun 12, 2012 at 10:13:39AM +0000, Jamie Heilman wrote:
> > J. Bruce Fields wrote:
> > > On Sun, Jun 10, 2012 at 09:03:42AM +0000, Jamie Heilman wrote:
> > > > Upgrading my NFSv4 server from 3.3.8 to 3.4.2 I've managed to
> > > > reproduce this reliably:
> > > > 
> > > > ------------[ cut here ]------------
> > > > kernel BUG at fs/nfsd/nfs4state.c:1044!
> > > 
> > > That's 
> > > 
> > > 	free_client(struct nfs4_client *clp)
> > > 	{
> > > --->        	BUG_ON(!spin_is_locked(&client_lock));
> > > 
> > > 
> > > > invalid opcode: 0000 [#1] 
> > > > Modules linked in: md5 cpufreq_conservative cpufreq_stats cpufreq_powersave cpufreq_ondemand autofs4 quota_v2 quota_tree nfsd nfs_acl exportfs nfs lockd fscache auth_rpcgss sunrpc xt_mark xt_dscp cls_fw sch_htb iptable_nat nf_nat ipt_REJECT xt_multiport xt_mac xt_tcpudp iptable_mangle nf_conntrack_ipv4 nf_defrag_ipv4 xt_state nf_conntrack xt_LOG xt_limit iptable_filter ip_tables x_tables dm_crypt dm_mod snd_hda_codec_via tpm_tis tpm via_rhine snd_hda_intel snd_hda_codec tpm_bios mii via_velocity snd_hwdep snd_pcm snd_timer snd soundcore snd_page_alloc evdev crc_ccitt via_agp agpgart button
> > > > 
> > > > Pid: 1804, comm: nfsd Not tainted 3.4.2 #3 To Be Filled By O.E.M. To Be Filled By O.E.M./To be filled by O.E.M.
> > > > EIP: 0060:[<f86ac683>] EFLAGS: 00010246 CPU: 0
> > > > EIP is at free_client.isra.47+0x3/0x5 [nfsd]
> > > > EAX: 00000000 EBX: e9d48800 ECX: e9d48030 EDX: f86b20f8
> > > > ESI: e9d48810 EDI: e9cf9ec0 EBP: e9cf9eb4 ESP: e9cf9eb4
> > > >  DS: 007b ES: 007b FS: 0000 GS: 00e0 SS: 0068
> > > > CR0: 8005003b CR2: b760bee0 CR3: 34d04000 CR4: 000006b0
> > > > DR0: 00000000 DR1: 00000000 DR2: 00000000 DR3: 00000000
> > > > DR6: ffff0ff0 DR7: 00000400
> > > > Process nfsd (pid: 1804, ti=e9cf8000 task=e9cf4000 task.ti=e9cf8000)
> > > > Stack:
> > > >  e9cf9ed4 f86a6c00 e9d48828 e9cf9ec0 e9cf9ec0 e9d48800 e9d48000 00000000
> > > >  e9cf9f00 f86a6d36 00000000 000555c0 f53c0001 4fd43f69 00000001 7ad0b2a0
> > > >  f4c45000 f4c44060 00000360 e9cf9f3c f869f63c f86aff20 e9cb9000 00000000
> > > > Call Trace:
> > > >  [<f86a6c00>] expire_client+0xb1/0xb9 [nfsd]
> > > 
> > > And the only free_client call there is:
> > > 
> > > 	spin_lock(&client_lock);
> > > 	unhash_client_locked(clp);
> > > 	if (atomic_read(&clp->cl_refcount) == 0)
> > > --->		free_client(clp);
> > > 	spin_unlock(&client_lock);
> > > 
> > > So, that's strange.
> > 
> > OK, other observations which may or may not have any value:
> > - I can't trigger the BUG if I build with CONFIG_SMP (which obviously
> >   changes the RCU as well)
> > - I can't trigger the BUG if I build with CONFIG_DEBUG_SPINLOCK
> 
> Oh, OK, so probably spin_is_locked() just isn't reliable in the
> !CONFIG_SMP case.  That would make sense.
> 
> So at a minimum we should change that BUG_ON to a WARN_ON_ONCE.

Doh.  Googline around--looks like we want something like this.

I've got a couple other bugfixes to collect and then I'll try to get
this to Linus by the end of the week.

--b.


commit 4cdc12ca178dfa74a9619451be41419617424a07
Author: J. Bruce Fields <bfields@...hat.com>
Date:   Tue Jun 12 08:28:48 2012 -0400

    nfsd4: BUG_ON(!is_spin_locked()) no good on UP kernels
    
    Most frequent symptom was a BUG triggering in expire_client, with the
    server locking up shortly thereafter.
    
    Introduced by 508dc6e110c6dbdc0bbe84298ccfe22de7538486 "nfsd41:
    free_session/free_client must be called under the client_lock".
    
    Cc: stable@...nel.org
    Cc: Benny Halevy <bhalevy@...ian.com>
    Signed-off-by: J. Bruce Fields <bfields@...hat.com>

diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c
index 8fdc9ec..94effd5 100644
--- a/fs/nfsd/nfs4state.c
+++ b/fs/nfsd/nfs4state.c
@@ -900,7 +900,7 @@ static void free_session(struct kref *kref)
 	struct nfsd4_session *ses;
 	int mem;
 
-	BUG_ON(!spin_is_locked(&client_lock));
+	lockdep_assert_held(&client_lock);
 	ses = container_of(kref, struct nfsd4_session, se_ref);
 	nfsd4_del_conns(ses);
 	spin_lock(&nfsd_drc_lock);
@@ -1080,7 +1080,7 @@ static struct nfs4_client *alloc_client(struct xdr_netobj name)
 static inline void
 free_client(struct nfs4_client *clp)
 {
-	BUG_ON(!spin_is_locked(&client_lock));
+	lockdep_assert_held(&client_lock);
 	while (!list_empty(&clp->cl_sessions)) {
 		struct nfsd4_session *ses;
 		ses = list_entry(clp->cl_sessions.next, struct nfsd4_session,
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/