Re: 2.6.37-rc3-mmotm1123 - arch/x86/kernel/cpu/mcheck/mce.c:1628invoked rcu_dereference_check() without protection!

[huang ying]

On Thu, Nov 25, 2010 at 2:18 PM,  <Valdis.Kletnieks@xxxxxx> wrote:
> (Appears to have been present in -rc2-mmotm1117 as well, and maybe earlier. I
> suspect mcelog wasn't getting run at boot due to my fighting with systemd, so I
> didn't see it).
>
> [ Â146.195836] ===================================================
> [ Â146.195839] [ INFO: suspicious rcu_dereference_check() usage. ]
> [ Â146.195841] ---------------------------------------------------
> [ Â146.195844] arch/x86/kernel/cpu/mcheck/mce.c:1628 invoked rcu_dereference_check() without protection!

The warning comes from the rcu_dereference_check_mce() in mce_poll().
It seems that we do not need rcu_dereference_check_mce() here. Because
we only check the value of mcelog.next, but do not use that value to
access the content of the array.

Best Regards,
Huang Ying

> [ Â146.195847]
> [ Â146.195847] other info that might help us debug this:
> [ Â146.195848]
> [ Â146.195850]
> [ Â146.195850] rcu_scheduler_active = 1, debug_locks = 1
> [ Â146.195853] no locks held by mcelog/3806.
> [ Â146.195855]
> [ Â146.195855] stack backtrace:
> [ Â146.195858] Pid: 3806, comm: mcelog Tainted: G Â Â Â ÂW Â 2.6.37-rc3-mmotm1123 #4
> [ Â146.195860] Call Trace:
> [ Â146.195869] Â[<ffffffff81069538>] lockdep_rcu_dereference+0x9d/0xa5
> [ Â146.195875] Â[<ffffffff810139ea>] mce_poll+0x66/0x74
> [ Â146.195880] Â[<ffffffff8110a62e>] do_sys_poll+0x23b/0x3a5
> [ Â146.195885] Â[<ffffffff811095ad>] ? __pollwait+0x0/0xc7
> [ Â146.195889] Â[<ffffffff81109674>] ? pollwake+0x0/0x4f
> [ Â146.195893] Â[<ffffffff81109674>] ? pollwake+0x0/0x4f
> [ Â146.195898] Â[<ffffffff81094c0e>] ? __rcu_read_unlock+0x28f/0x2ab
> [ Â146.195903] Â[<ffffffff810d565f>] ? __do_fault+0x256/0x3e5
> [ Â146.195908] Â[<ffffffff81068775>] ? look_up_lock_class+0x5f/0xc4
> [ Â146.195912] Â[<ffffffff810d565f>] ? __do_fault+0x256/0x3e5
> [ Â146.195916] Â[<ffffffff81068c42>] ? register_lock_class+0x1e/0x2d3
> [ Â146.195921] Â[<ffffffff810a5c05>] ? trace_preempt_on+0x15/0x28
> [ Â146.195925] Â[<ffffffff810d5755>] ? __do_fault+0x34c/0x3e5
> [ Â146.195931] Â[<ffffffff8155adbc>] ? _raw_spin_unlock+0x30/0x69
> [ Â146.195935] Â[<ffffffff8106a45e>] ? mark_lock+0x2d/0x1d8
> [ Â146.195940] Â[<ffffffff810ba69c>] ? unlock_page+0x25/0x2a
> [ Â146.195944] Â[<ffffffff8106ab57>] ? __lock_acquire+0x54e/0xd4e
> [ Â146.195948] Â[<ffffffff810d790d>] ? handle_mm_fault+0x193/0x7c3
> [ Â146.195953] Â[<ffffffff8104acaa>] ? sigprocmask+0x2f/0xc5
> [ Â146.195957] Â[<ffffffff8104acaa>] ? sigprocmask+0x2f/0xc5
> [ Â146.195962] Â[<ffffffff810d4cc0>] ? might_fault+0x4e/0x9e
> [ Â146.195966] Â[<ffffffff8155aea0>] ? _raw_spin_unlock_irq+0x2b/0x70
> [ Â146.195970] Â[<ffffffff810a5c05>] ? trace_preempt_on+0x15/0x28
> [ Â146.195974] Â[<ffffffff8104ad35>] ? sigprocmask+0xba/0xc5
> [ Â146.195978] Â[<ffffffff8155aeab>] ? _raw_spin_unlock_irq+0x36/0x70
> [ Â146.195983] Â[<ffffffff8155e378>] ? sub_preempt_count+0x35/0x49
> [ Â146.195987] Â[<ffffffff8155aed7>] ? _raw_spin_unlock_irq+0x62/0x70
> [ Â146.195992] Â[<ffffffff8110a973>] sys_ppoll+0xd3/0x13c
> [ Â146.195996] Â[<ffffffff8106bc89>] ? trace_hardirqs_on_caller+0x117/0x13b
> [ Â146.196087] Â[<ffffffff8155a60b>] ? trace_hardirqs_on_thunk+0x3a/0x3f
> [ Â146.196094] Â[<ffffffff8100277b>] system_call_fastpath+0x16/0x1b
>
>
--
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/