Message-ID: <CAHQdGtSNmera4P6ocFGbLhjmESB8oV9zKE_x_zTm=5qUg7e7Ug@mail.gmail.com>
Date:	Sun, 25 Jan 2015 16:55:20 -0500
From:	Trond Myklebust <trond.myklebust@...marydata.com>
To:	Bruno Prémont <bonbons@...ux-vserver.org>
Cc:	"J. Bruce Fields" <bfields@...ldses.org>,
	"Eric W. Biederman" <ebiederm@...ssion.com>,
	Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
	Linux NFS Mailing List <linux-nfs@...r.kernel.org>
Subject: Re: [Patch] sunrpc: NULL utsname dereference on NFS umount during
 namespace cleanup

On Sun, Jan 25, 2015 at 4:06 PM, Bruno Prémont
<bonbons@...ux-vserver.org> wrote:
> On a system running home-brown container (mntns, utsns, pidns, netns)
> with NFS mount-point bind-mounted into the container I hit the following
> trace if nfs filesystem is first umount()ed in init ns and then later
> umounted from container when the container exists.
>
> [51397.767310] BUG: unable to handle kernel NULL pointer dereference at 0000000000000008
> [51397.770671] IP: [<ffffffff81828173>] rpc_new_client+0x193/0x2b0
> [51397.773967] PGD 0
> [51397.777218] Oops: 0000 [#1] SMP
> [51397.780490] Modules linked in:
> [51397.783751] CPU: 0 PID: 1711 Comm: image-starter Not tainted 3.19.0-rc2-kvm+ #7
> [51397.787123] Hardware name: Gigabyte Technology Co., Ltd. GA-A75M-UD2H/GA-A75M-UD2H, BIOS F6 09/28/2012
> [51397.790606] task: ffff8800c9fcbac0 ti: ffff8801fe754000 task.ti: ffff8801fe754000
> [51397.794149] RIP: 0010:[<ffffffff81828173>]  [<ffffffff81828173>] rpc_new_client+0x193/0x2b0
> [51397.797798] RSP: 0018:ffff8801fe757908  EFLAGS: 00010246
> [51397.801444] RAX: 0000000000000000 RBX: ffff88009dafb240 RCX: 0000000000000180
> [51397.805174] RDX: 000000000000bae0 RSI: 0000000000001770 RDI: ffff88009dafb308
> [51397.808913] RBP: ffff8801fe757948 R08: ffff88009daf92d8 R09: ffff88009dafb458
> [51397.812673] R10: ffff88009dafb458 R11: ffff88020ec15bc0 R12: ffff8801fe757a40
> [51397.816456] R13: ffffffff81b9d800 R14: ffff8800c6e31030 R15: 0000000000000000
> [51397.820270] FS:  00007f335a3a1700(0000) GS:ffff88020ec00000(0000) knlGS:00000000f7287700
> [51397.824168] CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
> [51397.828066] CR2: 0000000000000008 CR3: 00000001fc54d000 CR4: 00000000000007f0
> [51397.832017] Stack:
> [51397.835924]  000000000000000a ffffffff81b9d770 ffffffff81826450 ffff8801fe757a40
> [51397.840023]  ffff8801fe757a40 ffff8800cf08d500 ffffffff81826450 ffffffff820f4728
> [51397.844130]  ffff8801fe757978 ffffffff81828815 ffff8801fe757978 ffffffff8182aad8
> [51397.848224] Call Trace:
> [51397.852221]  [<ffffffff81826450>] ? call_start+0x20/0x20
> [51397.856273]  [<ffffffff81826450>] ? call_start+0x20/0x20
> [51397.860295]  [<ffffffff81828815>] rpc_create_xprt+0x15/0xb0
> [51397.864324]  [<ffffffff8182aad8>] ? xprt_create_transport+0x108/0x1b0
> [51397.868428]  [<ffffffff81828971>] rpc_create+0xc1/0x190
> [51397.872574]  [<ffffffff81111c86>] ? internal_add_timer+0x66/0x80
> [51397.876733]  [<ffffffff81113a99>] ? mod_timer+0x109/0x1e0
> [51397.880877]  [<ffffffff8183a19e>] rpcb_create+0x6e/0x90
> [51397.884999]  [<ffffffff8183a71a>] rpcb_getport_async+0x15a/0x330
> [51397.889118]  [<ffffffff8182f1da>] ? rpc_malloc+0x3a/0x70
> [51397.893240]  [<ffffffff811af8d2>] ? __kmalloc+0xc2/0x170
> [51397.897354]  [<ffffffff81826830>] ? call_reserveresult+0x110/0x110
> [51397.901490]  [<ffffffff81826450>] ? call_start+0x20/0x20
> [51397.905606]  [<ffffffff81826450>] ? call_start+0x20/0x20
> [51397.909662]  [<ffffffff8182648e>] call_bind+0x3e/0x40
> [51397.913709]  [<ffffffff8182fa99>] __rpc_execute+0x79/0x330
> [51397.917778]  [<ffffffff818327bd>] rpc_execute+0x5d/0xa0
> [51397.921871]  [<ffffffff818286cb>] rpc_run_task+0x6b/0x90
> [51397.925989]  [<ffffffff8182872e>] rpc_call_sync+0x3e/0xa0
> [51397.930108]  [<ffffffff8127fe29>] nsm_mon_unmon+0xb9/0xd0
> [51397.934191]  [<ffffffff8110e2a0>] ? call_rcu_bh+0x20/0x20
> [51397.938235]  [<ffffffff8128018c>] nsm_unmonitor+0x8c/0x140
> [51397.942309]  [<ffffffff8127bc43>] nlm_destroy_host_locked+0x63/0xa0
> [51397.946442]  [<ffffffff8127c03c>] nlmclnt_release_host+0x7c/0x130
> [51397.950591]  [<ffffffff81279645>] nlmclnt_done+0x15/0x30
> [51397.954773]  [<ffffffff81241862>] nfs_destroy_server+0x12/0x20
> [51397.958934]  [<ffffffff81242372>] nfs_free_server+0x22/0xa0
> [51397.963053]  [<ffffffff8124cadd>] nfs_kill_super+0x1d/0x30
> [51397.967158]  [<ffffffff811c2e2c>] deactivate_locked_super+0x4c/0x70
> [51397.971286]  [<ffffffff811c33f9>] deactivate_super+0x49/0x70
> [51397.975398]  [<ffffffff811ddafe>] cleanup_mnt+0x3e/0x90
> [51397.979499]  [<ffffffff811ddb9d>] __cleanup_mnt+0xd/0x10
> [51397.983598]  [<ffffffff810e04cc>] task_work_run+0xbc/0xe0
> [51397.987697]  [<ffffffff810c8f95>] do_exit+0x295/0xaf0
> [51397.991812]  [<ffffffff811c2239>] ? ____fput+0x9/0x10
> [51397.995937]  [<ffffffff810e04b4>] ? task_work_run+0xa4/0xe0
> [51398.000070]  [<ffffffff810c986a>] do_group_exit+0x3a/0xa0
> [51398.004201]  [<ffffffff810c98df>] SyS_exit_group+0xf/0x10
> [51398.008315]  [<ffffffff8185e8d2>] system_call_fastpath+0x12/0x17
> [51398.012438] Code: 43 78 48 8d bb c8 00 00 00 48 89 7b 70 48 8b 30 e8 63 2d 01 00 c7 03 01 00 00 00 65 48 8b 04 25 00 aa 00 00 48 8b 80 c0 09 00 00 <4c> 8b 68 08 49 83 c5 45 4
> [51398.022378] RIP  [<ffffffff81828173>] rpc_new_client+0x193/0x2b0
> [51398.026732]  RSP <ffff8801fe757908>
> [51398.031025] CR2: 0000000000000008
> [51398.035326] ---[ end trace b701b037bc457620 ]---
> [51398.058223] Fixing recursive fault but reboot is needed!
>
>
> The code executed in rpc_new_client() tries to dereference the
> struct new_utsname * returned by utsname() which has already been
> released at this time.
>
> Cc: <stable@...r.kernel.org> # 3.18
> Signed-off-by: Bruno Prémont <bonbons@...ux-vserver.org>
> ---
> I've seen this trace on 3.18.x as well as 3.19-rc.
>
>
> This patch fixes the NULL dereference but I'm not sure this is the
> right fix.
> Should init uts_ns be referred to or what is the impact of using a random
> string to satisfy rpc_clnt_set_nodename()?
>
> Thanks,
> Bruno
> ---
> diff --git a/net/sunrpc/clnt.c b/net/sunrpc/clnt.c
> index 05da12a..0246e94 100644
> --- a/net/sunrpc/clnt.c
> +++ b/net/sunrpc/clnt.c
> @@ -365,6 +365,7 @@ static struct rpc_clnt * rpc_new_client(const struct rpc_create_args *args,
>         const struct rpc_version *version;
>         struct rpc_clnt *clnt = NULL;
>         const struct rpc_timeout *timeout;
> +       const struct new_utsname *uts_name;
>         int err;
>
>         /* sanity check the name before trying to print it */
> @@ -421,7 +422,8 @@ static struct rpc_clnt * rpc_new_client(const struct rpc_create_args *args,
>         atomic_set(&clnt->cl_count, 1);
>
>         /* save the nodename */
> -       rpc_clnt_set_nodename(clnt, utsname()->nodename);
> +       uts_name = utsname();
> +       rpc_clnt_set_nodename(clnt, uts_name ? uts_name->nodename : "");
>
>         err = rpc_client_register(clnt, args->authflavor, args->client_name);
>         if (err)

We should rather change rpcb_create() to pass the nodename from the
parent. The point is that the rpc_clnt->cl_nodename is used in various
different contexts (for instance in the AUTH_SYS credential) where it
isn't always appropriate to have it set to an empty string.

-- 
Trond Myklebust
Linux NFS client maintainer, PrimaryData
trond.myklebust@...marydata.com
--
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/

Powered by blists - more mailing lists