From: Bryan Schumaker <[email protected]>
unhash_delegation() will grab the recall lock before calling
list_del_init() in each of these places. This patch removes the
redundant calls.
Signed-off-by: Bryan Schumaker <[email protected]>
---
fs/nfsd/nfs4state.c | 3 ---
1 files changed, 0 insertions(+), 3 deletions(-)
diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c
index 6ab6779..43d5c22 100644
--- a/fs/nfsd/nfs4state.c
+++ b/fs/nfsd/nfs4state.c
@@ -1058,7 +1058,6 @@ expire_client(struct nfs4_client *clp)
spin_unlock(&recall_lock);
while (!list_empty(&reaplist)) {
dp = list_entry(reaplist.next, struct nfs4_delegation, dl_recall_lru);
- list_del_init(&dp->dl_recall_lru);
unhash_delegation(dp);
}
while (!list_empty(&clp->cl_openowners)) {
@@ -3125,7 +3124,6 @@ nfs4_laundromat(void)
spin_unlock(&recall_lock);
list_for_each_safe(pos, next, &reaplist) {
dp = list_entry (pos, struct nfs4_delegation, dl_recall_lru);
- list_del_init(&dp->dl_recall_lru);
unhash_delegation(dp);
}
test_val = nfsd4_lease;
@@ -4667,7 +4665,6 @@ __nfs4_state_shutdown(void)
spin_unlock(&recall_lock);
list_for_each_safe(pos, next, &reaplist) {
dp = list_entry (pos, struct nfs4_delegation, dl_recall_lru);
- list_del_init(&dp->dl_recall_lru);
unhash_delegation(dp);
}
--
1.7.8
On Wed Dec 14 09:07:44 2011, J. Bruce Fields wrote:
> On Wed, Dec 14, 2011 at 08:42:18AM -0500, Bryan Schumaker wrote:
>> On Tue Dec 13 17:12:09 2011, J. Bruce Fields wrote:
>>> On Tue, Dec 13, 2011 at 04:35:58PM -0500, [email protected] wrote:
>>>> From: Bryan Schumaker <[email protected]>
>>>>
>>>> unhash_delegation() will grab the recall lock before calling
>>>> list_del_init() in each of these places. This patch removes the
>>>> redundant calls.
>>>
>>> Looks right, thanks. (How did you happen across this?)
>>
>> I was looking over the fault injection stuff to see if I can figure out
>> where an occasional oops was coming from. It seems to happen when I
>> forget delegations, so I was looking at how other functions release
>> delegations to see if there was something I missed. The oops doesn't
>> happen often enough for me to know if this patch fixes it yet, though.
>
> Hm, OK. Could you post the oops?
I don't see it in the logs on the server, but I'll keep looking. I'll
post it the next time I come across it.
- Bryan
>
> --b.
On Tue, Dec 13, 2011 at 04:35:58PM -0500, [email protected] wrote:
> From: Bryan Schumaker <[email protected]>
>
> unhash_delegation() will grab the recall lock before calling
> list_del_init() in each of these places. This patch removes the
> redundant calls.
Looks right, thanks. (How did you happen across this?)
--b.
>
> Signed-off-by: Bryan Schumaker <[email protected]>
> ---
> fs/nfsd/nfs4state.c | 3 ---
> 1 files changed, 0 insertions(+), 3 deletions(-)
>
> diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c
> index 6ab6779..43d5c22 100644
> --- a/fs/nfsd/nfs4state.c
> +++ b/fs/nfsd/nfs4state.c
> @@ -1058,7 +1058,6 @@ expire_client(struct nfs4_client *clp)
> spin_unlock(&recall_lock);
> while (!list_empty(&reaplist)) {
> dp = list_entry(reaplist.next, struct nfs4_delegation, dl_recall_lru);
> - list_del_init(&dp->dl_recall_lru);
> unhash_delegation(dp);
> }
> while (!list_empty(&clp->cl_openowners)) {
> @@ -3125,7 +3124,6 @@ nfs4_laundromat(void)
> spin_unlock(&recall_lock);
> list_for_each_safe(pos, next, &reaplist) {
> dp = list_entry (pos, struct nfs4_delegation, dl_recall_lru);
> - list_del_init(&dp->dl_recall_lru);
> unhash_delegation(dp);
> }
> test_val = nfsd4_lease;
> @@ -4667,7 +4665,6 @@ __nfs4_state_shutdown(void)
> spin_unlock(&recall_lock);
> list_for_each_safe(pos, next, &reaplist) {
> dp = list_entry (pos, struct nfs4_delegation, dl_recall_lru);
> - list_del_init(&dp->dl_recall_lru);
> unhash_delegation(dp);
> }
>
> --
> 1.7.8
>
On Wed, Dec 14, 2011 at 08:42:18AM -0500, Bryan Schumaker wrote:
> On Tue Dec 13 17:12:09 2011, J. Bruce Fields wrote:
> > On Tue, Dec 13, 2011 at 04:35:58PM -0500, [email protected] wrote:
> >> From: Bryan Schumaker <[email protected]>
> >>
> >> unhash_delegation() will grab the recall lock before calling
> >> list_del_init() in each of these places. This patch removes the
> >> redundant calls.
> >
> > Looks right, thanks. (How did you happen across this?)
>
> I was looking over the fault injection stuff to see if I can figure out
> where an occasional oops was coming from. It seems to happen when I
> forget delegations, so I was looking at how other functions release
> delegations to see if there was something I missed. The oops doesn't
> happen often enough for me to know if this patch fixes it yet, though.
Hm, OK. Could you post the oops?
--b.
On Tue Dec 13 17:12:09 2011, J. Bruce Fields wrote:
> On Tue, Dec 13, 2011 at 04:35:58PM -0500, [email protected] wrote:
>> From: Bryan Schumaker <[email protected]>
>>
>> unhash_delegation() will grab the recall lock before calling
>> list_del_init() in each of these places. This patch removes the
>> redundant calls.
>
> Looks right, thanks. (How did you happen across this?)
I was looking over the fault injection stuff to see if I can figure out
where an occasional oops was coming from. It seems to happen when I
forget delegations, so I was looking at how other functions release
delegations to see if there was something I missed. The oops doesn't
happen often enough for me to know if this patch fixes it yet, though.
- Bryan
>
> --b.
>
>>
>> Signed-off-by: Bryan Schumaker <[email protected]>
>> ---
>> fs/nfsd/nfs4state.c | 3 ---
>> 1 files changed, 0 insertions(+), 3 deletions(-)
>>
>> diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c
>> index 6ab6779..43d5c22 100644
>> --- a/fs/nfsd/nfs4state.c
>> +++ b/fs/nfsd/nfs4state.c
>> @@ -1058,7 +1058,6 @@ expire_client(struct nfs4_client *clp)
>> spin_unlock(&recall_lock);
>> while (!list_empty(&reaplist)) {
>> dp = list_entry(reaplist.next, struct nfs4_delegation, dl_recall_lru);
>> - list_del_init(&dp->dl_recall_lru);
>> unhash_delegation(dp);
>> }
>> while (!list_empty(&clp->cl_openowners)) {
>> @@ -3125,7 +3124,6 @@ nfs4_laundromat(void)
>> spin_unlock(&recall_lock);
>> list_for_each_safe(pos, next, &reaplist) {
>> dp = list_entry (pos, struct nfs4_delegation, dl_recall_lru);
>> - list_del_init(&dp->dl_recall_lru);
>> unhash_delegation(dp);
>> }
>> test_val = nfsd4_lease;
>> @@ -4667,7 +4665,6 @@ __nfs4_state_shutdown(void)
>> spin_unlock(&recall_lock);
>> list_for_each_safe(pos, next, &reaplist) {
>> dp = list_entry (pos, struct nfs4_delegation, dl_recall_lru);
>> - list_del_init(&dp->dl_recall_lru);
>> unhash_delegation(dp);
>> }
>>
>> --
>> 1.7.8
>>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
On Wed Dec 14 11:35:32 2011, Casey Bodley wrote:
> On Wed, Dec 14, 2011 at 9:33 AM, Bryan Schumaker <[email protected]> wrote:
>> On Wed Dec 14 09:07:44 2011, J. Bruce Fields wrote:
>>> On Wed, Dec 14, 2011 at 08:42:18AM -0500, Bryan Schumaker wrote:
>>>> On Tue Dec 13 17:12:09 2011, J. Bruce Fields wrote:
>>>>> On Tue, Dec 13, 2011 at 04:35:58PM -0500, [email protected] wrote:
>>>>>> From: Bryan Schumaker <[email protected]>
>>>>>>
>>>>>> unhash_delegation() will grab the recall lock before calling
>>>>>> list_del_init() in each of these places. This patch removes the
>>>>>> redundant calls.
>>>>>
>>>>> Looks right, thanks. (How did you happen across this?)
>>>>
>>>> I was looking over the fault injection stuff to see if I can figure out
>>>> where an occasional oops was coming from. It seems to happen when I
>>>> forget delegations, so I was looking at how other functions release
>>>> delegations to see if there was something I missed. The oops doesn't
>>>> happen often enough for me to know if this patch fixes it yet, though.
>>>
>>> Hm, OK. Could you post the oops?
>>
>> I don't see it in the logs on the server, but I'll keep looking. I'll
>> post it the next time I come across it.
>
> I'm not sure if this is the same oops, but I'm able to reproduce it
> consistently against Bruce's 3.2.0-rc1 tree with fault injection
> enabled. I use the windows client to read a file, then 'echo 0 >
> /sys/kernel/debug/nfsd/forget_delegations'. I just tested with the
> patch above, and was able to reproduce the same oops:
>
> [ 1200.251735] NFSD Fault Injection: forget_delegations (all)
> [ 1200.251830] general protection fault: 0000 [#1] SMP
> [ 1200.252450] CPU 1
> [ 1200.252597] Modules linked in:
> [ 1200.253062]
> [ 1200.253256] Pid: 2004, comm: bash Not tainted 3.2.0-rc1+ #9 VMware,
> Inc. VMware Virtual Platform/440BX Desktop Reference Platform
> [ 1200.254102] RIP: 0010:[<ffffffff811ae7f5>] [<ffffffff811ae7f5>]
> nfsd_process_n_delegations+0x3e/0xbe
> [ 1200.254746] RSP: 0018:ffff88003b579e08 EFLAGS: 00010287
> [ 1200.255067] RAX: dead000000100100 RBX: 0000000000000000 RCX: 00000001802c002b
> [ 1200.255458] RDX: ffff88003d2d3328 RSI: ffffffff811aa49e RDI: ffff88003d2d3308
> [ 1200.255842] RBP: ffff88003b579e58 R08: ffff88003b558000 R09: ffff88003d2d3480
> [ 1200.256226] R10: ffff88003b68b750 R11: 0000000000000040 R12: dead0000001000f8
> [ 1200.256610] R13: 0000000000000001 R14: 000000000000002b R15: 00000000000000d5
> [ 1200.257029] FS: 00007fbad1d44700(0000) GS:ffff88003fa20000(0000)
> knlGS:0000000000000000
> [ 1200.257512] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> [ 1200.257850] CR2: 00007fbad1d49000 CR3: 000000003b593000 CR4: 00000000000006e0
> [ 1200.258272] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
> [ 1200.258693] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
> [ 1200.259083] Process bash (pid: 2004, threadinfo ffff88003b578000,
> task ffff88003cc3dfe0)
> [ 1200.259561] Stack:
> [ 1200.259768] ffffffff811aa49e ffff88003d2d3308 ffffffff81be1560
> ffff88003d2d3328
> [ 1200.260553] ffff88003b579e88 ffffffff81a2b6a0 0000000000000000
> 0000000000000002
> [ 1200.261337] ffff88003d8c1550 ffff88003d8c1528 ffff88003b579e88
> ffffffff811ae897
> [ 1200.262121] Call Trace:
> [ 1200.262346] [<ffffffff811aa49e>] ? nfs4_put_delegation+0x55/0x55
> [ 1200.262701] [<ffffffff811ae897>] nfsd_forget_delegations+0x22/0x40
> [ 1200.263119] [<ffffffff8119d3e7>] nfsd_inject_set+0x3f/0x46
> [ 1200.263470] [<ffffffff810fb878>] simple_attr_write+0x8a/0xb2
> [ 1200.263818] [<ffffffff810e0949>] vfs_write+0xa7/0xfa
> [ 1200.264136] [<ffffffff810e0b41>] sys_write+0x45/0x69
> [ 1200.264483] [<ffffffff81450f3b>] system_call_fastpath+0x16/0x1b
> [ 1200.264832] Code: 45 31 ed 41 54 53 48 89 fb 48 83 ec 28 49 63 c6
> 48 c1 e0 04 4c 8b a0 b0 12 be 81 48 05 b0 12 be 81 48 89 45 c0 49 83
> ec 08 eb 57
> [ 1200.270209] 8b 7c 24 28 49 8d 54 24 28 48 89 55 c8 48 83 ef 20 48 8b 47
> [ 1200.272534] RIP [<ffffffff811ae7f5>] nfsd_process_n_delegations+0x3e/0xbe
> [ 1200.273007] RSP <ffff88003b579e08>
> [ 1200.273331] ---[ end trace 0071bb3837d120e2 ]---
That's similar to the one I just got. Thanks for the test case! I had
been opening two files in order to get a read delegation, I didn't
think of trying with just one. Here is what I got:
Dec 14 11:21:01 localhost -- MARK --
Dec 14 11:45:00 localhost kernel: [ 7716.735003] NFSD Fault Injection:
forget_delegations (all)
Dec 14 11:45:00 localhost kernel: [ 7716.735814] CPU 0
Dec 14 11:45:00 localhost kernel: [ 7716.735902] Modules linked in:
dm_mod md5 nfsd exportfs nfs lockd fscache auth_rpcgss nfs_acl sunrpc
ipv6 ext2 mbcache loop snd_hda_intel snd_hda_codec snd_hwdep snd_pcm
snd_timer psmouse snd serio_raw soundcore pcspkr snd_page_alloc
virtio_balloon i2c_piix4 evdev i2c_core floppy processor button btrfs
zlib_deflate crc32c libcrc32c sr_mod cdrom pata_acpi uhci_hcd ata_piix
usbcore usb_common libata scsi_mod virtio_net virtio_pci virtio_blk
virtio_ring virtio
Dec 14 11:45:00 localhost kernel: [ 7716.737782]
Dec 14 11:45:00 localhost kernel: [ 7716.737782] Pid: 17423, comm: bash
Not tainted 3.2.0-rc1-BRYAN+ #1 Bochs Bochs
Dec 14 11:45:00 localhost kernel: [ 7716.737782] RIP:
0010:[<ffffffffa0415533>] [<ffffffffa0415533>]
nfsd_process_n_delegations+0x53/0xe0 [nfsd]
Dec 14 11:45:00 localhost kernel: [ 7716.737782] RSP:
0018:ffff880034121e18 EFLAGS: 00010202
Dec 14 11:45:00 localhost kernel: [ 7716.737782] RAX: dead000000100100
RBX: 0000000000000000 RCX: 0000000000000200
Dec 14 11:45:00 localhost kernel: [ 7716.737782] RDX: dead0000001000f8
RSI: 000060ffc0001270 RDI: ffffea0000d93400
Dec 14 11:45:00 localhost kernel: [ 7716.737782] RBP: ffff880034121e68
R08: ffffffffa040fc26 R09: 0000000000000000
Dec 14 11:45:00 localhost kernel: [ 7716.737782] R10: ffff88002b206a90
R11: 0000000000000000 R12: ffff88002eed4188
Dec 14 11:45:00 localhost kernel: [ 7716.737782] R13: 0000000000000001
R14: ffff88002eed41a8 R15: ffffffffa040fcb0
Dec 14 11:45:00 localhost kernel: [ 7716.737782] FS:
00007f09bfb8c700(0000) GS:ffff88003fc00000(0000) knlGS:0000000000000000
Dec 14 11:45:00 localhost kernel: [ 7716.737782] CS: 0010 DS: 0000 ES:
0000 CR0: 0000000080050033
Dec 14 11:45:00 localhost kernel: [ 7716.737782] CR2: 00007f09bfb92000
CR3: 000000003a90f000 CR4: 00000000000006f0
Dec 14 11:45:00 localhost kernel: [ 7716.737782] DR0: 0000000000000000
DR1: 0000000000000000 DR2: 0000000000000000
Dec 14 11:45:00 localhost kernel: [ 7716.737782] DR3: 0000000000000000
DR6: 00000000ffff0ff0 DR7: 0000000000000400
Dec 14 11:45:00 localhost kernel: [ 7716.737782] Process bash (pid:
17423, threadinfo ffff880034120000, task ffff88003c0064a0)
Dec 14 11:45:00 localhost kernel: [ 7716.737782] ffffffffa041ffb0
00000000000000f7 ffffffffa04283b0 dead0000001000f8
Dec 14 11:45:00 localhost kernel: [ 7716.737782] ffff880034121e58
0000000000000000 ffffffffa041ffb0 ffff88002eed48d0
Dec 14 11:45:00 localhost kernel: [ 7716.737782] 0000000000000002
ffff88002eed48a8 ffff880034121e88 ffffffffa04155e5
Dec 14 11:45:00 localhost kernel: [ 7716.737782] [<ffffffffa04155e5>]
nfsd_forget_delegations+0x25/0x50 [nfsd]
Dec 14 11:45:00 localhost kernel: [ 7716.737782] [<ffffffffa03fd75c>]
nfsd_inject_set+0x3c/0x70 [nfsd]
Dec 14 11:45:00 localhost kernel: [ 7716.737782] [<ffffffff81186f22>]
simple_attr_write+0xc2/0xe0
Dec 14 11:45:00 localhost kernel: [ 7716.737782] [<ffffffff811630ef>]
vfs_write+0xaf/0x180
Dec 14 11:45:00 localhost kernel: [ 7716.737782] [<ffffffff8116341a>]
sys_write+0x4a/0x90
Dec 14 11:45:00 localhost kernel: [ 7716.737782] [<ffffffff81417e42>]
system_call_fastpath+0x16/0x1b
Dec 14 11:45:00 localhost kernel: [ 7716.737782] RSP <ffff880034121e18>
Dec 14 11:45:00 localhost kernel: [ 7716.754976] ---[ end trace
f7471c4cb1a3a2f4 ]---
- Bryan
>
>>
>> - Bryan
>>>
>>> --b.
>>
>>
>> --
>> To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
>> the body of a message to [email protected]
>> More majordomo info at http://vger.kernel.org/majordomo-info.html
On Wed, Dec 14, 2011 at 9:33 AM, Bryan Schumaker <[email protected]> wrote:
> On Wed Dec 14 09:07:44 2011, J. Bruce Fields wrote:
>> On Wed, Dec 14, 2011 at 08:42:18AM -0500, Bryan Schumaker wrote:
>>> On Tue Dec 13 17:12:09 2011, J. Bruce Fields wrote:
>>>> On Tue, Dec 13, 2011 at 04:35:58PM -0500, [email protected] wrote:
>>>>> From: Bryan Schumaker <[email protected]>
>>>>>
>>>>> unhash_delegation() will grab the recall lock before calling
>>>>> list_del_init() in each of these places. ?This patch removes the
>>>>> redundant calls.
>>>>
>>>> Looks right, thanks. ?(How did you happen across this?)
>>>
>>> I was looking over the fault injection stuff to see if I can figure out
>>> where an occasional oops was coming from. ?It seems to happen when I
>>> forget delegations, so I was looking at how other functions release
>>> delegations to see if there was something I missed. ?The oops doesn't
>>> happen often enough for me to know if this patch fixes it yet, though.
>>
>> Hm, OK. ?Could you post the oops?
>
> I don't see it in the logs on the server, but I'll keep looking. ?I'll
> post it the next time I come across it.
I'm not sure if this is the same oops, but I'm able to reproduce it
consistently against Bruce's 3.2.0-rc1 tree with fault injection
enabled. I use the windows client to read a file, then 'echo 0 >
/sys/kernel/debug/nfsd/forget_delegations'. I just tested with the
patch above, and was able to reproduce the same oops:
[ 1200.251735] NFSD Fault Injection: forget_delegations (all)
[ 1200.251830] general protection fault: 0000 [#1] SMP
[ 1200.252450] CPU 1
[ 1200.252597] Modules linked in:
[ 1200.253062]
[ 1200.253256] Pid: 2004, comm: bash Not tainted 3.2.0-rc1+ #9 VMware,
Inc. VMware Virtual Platform/440BX Desktop Reference Platform
[ 1200.254102] RIP: 0010:[<ffffffff811ae7f5>] [<ffffffff811ae7f5>]
nfsd_process_n_delegations+0x3e/0xbe
[ 1200.254746] RSP: 0018:ffff88003b579e08 EFLAGS: 00010287
[ 1200.255067] RAX: dead000000100100 RBX: 0000000000000000 RCX: 00000001802c002b
[ 1200.255458] RDX: ffff88003d2d3328 RSI: ffffffff811aa49e RDI: ffff88003d2d3308
[ 1200.255842] RBP: ffff88003b579e58 R08: ffff88003b558000 R09: ffff88003d2d3480
[ 1200.256226] R10: ffff88003b68b750 R11: 0000000000000040 R12: dead0000001000f8
[ 1200.256610] R13: 0000000000000001 R14: 000000000000002b R15: 00000000000000d5
[ 1200.257029] FS: 00007fbad1d44700(0000) GS:ffff88003fa20000(0000)
knlGS:0000000000000000
[ 1200.257512] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 1200.257850] CR2: 00007fbad1d49000 CR3: 000000003b593000 CR4: 00000000000006e0
[ 1200.258272] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[ 1200.258693] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
[ 1200.259083] Process bash (pid: 2004, threadinfo ffff88003b578000,
task ffff88003cc3dfe0)
[ 1200.259561] Stack:
[ 1200.259768] ffffffff811aa49e ffff88003d2d3308 ffffffff81be1560
ffff88003d2d3328
[ 1200.260553] ffff88003b579e88 ffffffff81a2b6a0 0000000000000000
0000000000000002
[ 1200.261337] ffff88003d8c1550 ffff88003d8c1528 ffff88003b579e88
ffffffff811ae897
[ 1200.262121] Call Trace:
[ 1200.262346] [<ffffffff811aa49e>] ? nfs4_put_delegation+0x55/0x55
[ 1200.262701] [<ffffffff811ae897>] nfsd_forget_delegations+0x22/0x40
[ 1200.263119] [<ffffffff8119d3e7>] nfsd_inject_set+0x3f/0x46
[ 1200.263470] [<ffffffff810fb878>] simple_attr_write+0x8a/0xb2
[ 1200.263818] [<ffffffff810e0949>] vfs_write+0xa7/0xfa
[ 1200.264136] [<ffffffff810e0b41>] sys_write+0x45/0x69
[ 1200.264483] [<ffffffff81450f3b>] system_call_fastpath+0x16/0x1b
[ 1200.264832] Code: 45 31 ed 41 54 53 48 89 fb 48 83 ec 28 49 63 c6
48 c1 e0 04 4c 8b a0 b0 12 be 81 48 05 b0 12 be 81 48 89 45 c0 49 83
ec 08 eb 57
[ 1200.270209] 8b 7c 24 28 49 8d 54 24 28 48 89 55 c8 48 83 ef 20 48 8b 47
[ 1200.272534] RIP [<ffffffff811ae7f5>] nfsd_process_n_delegations+0x3e/0xbe
[ 1200.273007] RSP <ffff88003b579e08>
[ 1200.273331] ---[ end trace 0071bb3837d120e2 ]---
>
> - Bryan
>>
>> --b.
>
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
> the body of a message to [email protected]
> More majordomo info at ?http://vger.kernel.org/majordomo-info.html