Re: block layout oops

[Bian Naimeng <biannm@cn.fujitsu.com>]

> I'm getting this when I try to do a block layout mount with benny's new
> pnfs-all-latest kernel, 2.6.36-rc6-pnfs:
> 
> BUG: unable to handle kernel NULL pointer dereference at 00000008
> IP: [<e0c5f9a3>] nfs4_proc_getdevicelist+0x3c/0x118 [nfs]
> *pde = 00000000 
> Oops: 0000 [#1] 
> last sysfs file: /sys/module/sunrpc/initstate
> Modules linked in: blocklayoutdriver nfs lockd fscache nfs_acl auth_rpcgss sunrpc ip6t_REJECT nf_conntrack_ipv6 ip6table_filter ip6_tables ipv6 iscsi_tcp libiscsi_tcp libiscsi scsi_transport_iscsi snd_intel8x0 snd_ac97_codec ac97_bus snd_pcm microcode snd_timer snd i2c_piix4 pcnet32 i2c_core soundcore pcspkr snd_page_alloc mii [last unloaded: speedstep_lib]
> 
> Pid: 1226, comm: mount.nfs4 Not tainted 2.6.36-rc6-pnfs #9 /VirtualBox
> EIP: 0060:[<e0c5f9a3>] EFLAGS: 00210246 CPU: 0
> EIP is at nfs4_proc_getdevicelist+0x3c/0x118 [nfs]
> EAX: 00000000 EBX: e0cb3310 ECX: 00000004 EDX: de9c3c40
> ESI: df756000 EDI: de9c3c68 EBP: de9c3c90 ESP: de9c3c30
>  DS: 007b ES: 007b FS: 0000 GS: 00e0 SS: 0069
> Process mount.nfs4 (pid: 1226, ti=de9c2000 task=df626380 task.ti=de9c2000)
> Stack:
>  de9c3c40 de9c3c68 debfb300 dec0ee00 00000000 00000000 df756000 e0caec47
> <0> c14b830c dea26588 dea26590 de9c3c88 c10a2f7f 00000008 00200246 e0cb3310
> <0> c14b8624 de9c3c90 00000000 00000000 00000000 e0cb3310 00000003 de9c3d38
> Call Trace:
>  [<e0caec47>] ? T.983+0x12/0x14 [blocklayoutdriver]
>  [<c10a2f7f>] ? __kmalloc+0xf7/0x103
>  [<e0caed2e>] ? bl_initialize_mountpoint+0xe5/0x36b [blocklayoutdriver]
>  [<e0c7233e>] ? set_pnfs_layoutdriver+0x146/0x1e2 [nfs]
>  [<e0c48ba8>] ? nfs_probe_fsinfo+0x18a/0x2c0 [nfs]
>  [<e0c483a6>] ? nfs_init_server_rpcclient+0x25/0xd1 [nfs]
>  [<e0c49511>] ? nfs_clone_server+0x11f/0x206 [nfs]
>  [<e0c51ca2>] ? nfs4_xdev_get_sb+0x42/0x1cc [nfs]
>  [<c10acf0a>] ? vfs_kern_mount+0x9b/0x15d
>  [<e0c59871>] ? nfs_follow_mountpoint+0x1df/0x31e [nfs]
>  [<c10b3950>] ? do_follow_link+0xcc/0x1ed
>  [<c10b3d54>] ? link_path_walk+0x2e3/0x3be
>  [<c10b3f18>] ? path_walk+0x52/0xaf
>  [<c10b3fe6>] ? vfs_path_lookup+0x71/0xb5
>  [<e0c5197f>] ? nfs_follow_remote_path+0x126/0x255 [nfs]
>  [<e0c51b9b>] ? nfs4_try_mount+0x61/0x94 [nfs]
>  [<e0c52586>] ? nfs4_get_sb+0x22a/0x299 [nfs]
>  [<c10acf0a>] ? vfs_kern_mount+0x9b/0x15d
>  [<c10ad01b>] ? do_kern_mount+0x38/0xc2
>  [<c10be1e9>] ? do_mount+0x5ee/0x634
>  [<c10bcd5c>] ? copy_mount_options+0x78/0xd7
>  [<c10be295>] ? sys_mount+0x66/0x94
>  [<c132425c>] ? syscall_call+0x7/0xb
> Code: c0 89 4d ac 8d 7d e8 b9 03 00 00 00 89 55 a8 8d 55 b0 f3 ab 8d 45 d8 89 45 a4 89 55 a0 8b 86 a8 01 00 00 b9 04 00 00 00 8b 7d a4 <8b> 50 08 31 c0 f3 ab 8b 4d a8 8b 7d a0 89 55 dc 89 4d d8 b9 06 
> EIP: [<e0c5f9a3>] nfs4_proc_getdevicelist+0x3c/0x118 [nfs] SS:ESP 0068:de9c3c30
> CR2: 0000000000000008


  Is this patch can fix it?

----------------------------------------------------------------------
We must specify the layout type before initialize mountpoint, so we should try set 
pnfs_curr_ld earlier.

Signed-off-by: Bian Naimeng <biannm@cn.fujitsu.com>

---
 fs/nfs/pnfs.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c
index a225cd1..72ec712 100644
--- a/fs/nfs/pnfs.c
+++ b/fs/nfs/pnfs.c
@@ -162,6 +162,7 @@ set_pnfs_layoutdriver(struct nfs_server *server, const struct nfs_fh *mntfh,
 		dprintk("%s: Could not grab reference on module\n", __func__);
 		goto out_no_driver;
 	}
+	server->pnfs_curr_ld = ld_type;
 	if (ld_type->initialize_mountpoint(server, mntfh)) {
 		printk(KERN_ERR
 		       "%s: Error initializing mount point for layout driver %u.\n",
@@ -169,7 +170,6 @@ set_pnfs_layoutdriver(struct nfs_server *server, const struct nfs_fh *mntfh,
 		module_put(ld_type->owner);
 		goto out_no_driver;
 	}
-	server->pnfs_curr_ld = ld_type;
 	dprintk("%s: pNFS module for %u set\n", __func__, id);
 	return;
 
-- 
1.7.0.1




-- 
Regards
Bian Naimeng