Re: [PATCH] nfs: clear_commit_release incorrectly handle truncated page

[Dmitry Monakhov]

Trond Myklebust <trond.myklebust@xxxxxxxxxx> writes:

> On Tue, 2010-02-02 at 18:56 +0300, Dmitry Monakhov wrote: 
>> Trond Myklebust <trond.myklebust@xxxxxxxxxx> writes:
>> 
>> > On Tue, 2010-02-02 at 18:17 +0300, Dmitry Monakhov wrote: 
>> >> Trond Myklebust <trond.myklebust@xxxxxxxxxx> writes:
>> >> 
>> >> > On Tue, 2010-02-02 at 13:36 +0300, Dmitry Monakhov wrote:
>> >> >
>> >> > Something is wrong here. nfs_release_page() returns '0' if the 
>> >> > page has an associated write request (i.e. PagePrivate is set), and so
>> >> > both invalidate_complete_page() and invalidate_complete_page2() will
>> >> > fail.
>> >> >
>> >> > So what is truncating the page?
>> >> truncate_inode_page()
>> >>   truncate_complete_page()
>> >>     if (page_has_private(page))
>> >>        do_invalidatepage()
>> >>          ->nfs_invalidate_page()
>> >
>> > do_invalidate_page() is called before remove_from_page_cache(), so
>> > page->mapping should still be set.
>
> Hmm.... There is a known problem with a reference leak in
> nfs_wb_page_cancel() (I've queued up a fix for 2.6.33 in the 'bugfixes'
> branch of my git tree already). What happens when you apply the
> following patch?
The not helps, still get the same oops(log follows).
Have you tried my testcase?

 BUG: unable to handle kernel NULL pointer dereference at 00000040
 IP: [<f80d415f>] nfs_clear_request_commit+0x3f/0xb0 [nfs]
 *pde = 00000000 
 Oops: 0000 [#1] SMP DEBUG_PAGEALLOC
 last sysfs file: /sys/devices/platform/thinkpad_acpi/leds/tpacpi::thinkvantage/uevent
 Modules linked in: binfmt_misc quota_v2 quota_tree nfsd exportfs nfs lockd sunrpc iwl3945 thinkpad_acpi psmouse led_class serio_raw iwlcore nvram raid1 raid0 linear e1000e
 
 Pid: 1035, comm: nfsiod Not tainted 2.6.33-rc6 #60 2623DDU/2623DDU
 EIP: 0060:[<f80d415f>] EFLAGS: 00010296 CPU: 0
 EIP is at nfs_clear_request_commit+0x3f/0xb0 [nfs]
 EAX: 00000000 EBX: c2561d80 ECX: c06d3700 EDX: 00000014
 ESI: f69916c0 EDI: f80dab58 EBP: f6724ef4 ESP: f6724ee8
  DS: 007b ES: 007b FS: 00d8 GS: 0000 SS: 0068
 Process nfsiod (pid: 1035, ti=f6724000 task=f69dda90 task.ti=f6724000)
 Stack:
  c04df67d f69d7440 f69916c0 f6724f34 f80d4258 f6724f44 00001b0e 00000000
 <0> ffff799c 00000400 f505f000 94c0042a f69917e0 f69917e8 00000000 f69916c0
 <0> f69916c4 f69916c0 f80dab58 f6724f3c f8075703 f6724f58 f8075871 f6724f60
 Call Trace:
  [<c04df67d>] ? schedule+0x3ad/0xa30
  [<f80d4258>] ? nfs_commit_release+0x88/0x1a0 [nfs]
  [<f8075703>] ? rpc_release_calldata+0x13/0x20 [sunrpc]
  [<f8075871>] ? rpc_free_task+0x41/0x70 [sunrpc]
  [<c01acc4c>] ? probe_workqueue_execution+0x8c/0xd0
  [<f8075940>] ? rpc_async_release+0x10/0x20 [sunrpc]
  [<c015834d>] ? worker_thread+0x10d/0x210
  [<f8075930>] ? rpc_async_release+0x0/0x20 [sunrpc]
  [<c015bb10>] ? autoremove_wake_function+0x0/0x50
  [<c0158240>] ? worker_thread+0x0/0x210
  [<c015b724>] ? kthread+0x74/0x80
  [<c015b6b0>] ? kthread+0x0/0x80
  [<c0103546>] ? kernel_thread_helper+0x6/0x10
 Code: f0 0f ba 70 28 01 19 d2 31 c0 85 d2 75 0e 8b 5d f8 8b 75 fc 89 ec 5d c3 8d 74 26 00 89 d8 ba 10 00 00 00 e8 74 0e 10 c8 8b 43 10 <8b> 70 40 9c 5b fa e8 26 67 0d c8 8d 46 30 b9 ff ff ff ff 0f bd 
 EIP: [<f80d415f>] nfs_clear_request_commit+0x3f/0xb0 [nfs] SS:ESP 0068:f6724ee8
 CR2: 0000000000000040
 ---[ end trace 4bf8ee9d233ce744 ]---
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/