Re: [PATCH] debugobjects: Use GFP_ATOMIC to allocate when initializing

[Sasha Levin]

On Mon, Apr 2, 2012 at 8:35 PM, Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:
> On Mon, 2012-04-02 at 11:12 +0200, Thomas Gleixner wrote:
>> On Sun, 1 Apr 2012, Sasha Levin wrote:
>> > On Sun, Apr 1, 2012 at 7:36 PM, Thomas Gleixner <tglx@xxxxxxxxxxxxx> wrote:
>> > > On Sat, 31 Mar 2012, Sasha Levin wrote:
>>
>> Peter ????
>>
>> > >> During the init stage, we have only 1 cpu and run with IRQs disabled,
>> > >> we can't allocate with GFP_KERNEL.
>> > >>
>> > >> This also fixes the related warning on boot.
>> > >
>> > > Hrmm. This should not be the case. In early boot GFP_KERNEL
>> > > allocations should not trigger a warning when interrupts are
>> > > disabled. Which kind of warning are you observing ?
>> >
>> > There were two, both pointing at the same WARN(). I don't have the
>> > first one which was fixed in this patch handy, but the second one is:
>> >
>> > [    0.000000] WARNING: at kernel/lockdep.c:2739
>> > lockdep_trace_alloc+0x11f/0x130()
>> > [    0.000000] Pid: 0, comm: swapper/0 Not tainted
>> > 3.3.0-next-20120330-sasha-dirty #40
>> > [    0.000000] Call Trace:
>> > [    0.000000]  [<ffffffff810b4175>] warn_slowpath_common+0x75/0xb0
>> > [    0.000000]  [<ffffffff810b41c5>] warn_slowpath_null+0x15/0x20
>> > [    0.000000]  [<ffffffff8111999f>] lockdep_trace_alloc+0x11f/0x130
>> > [    0.000000]  [<ffffffff81182e0c>] __alloc_pages_nodemask+0x9c/0x260
>> > [    0.000000]  [<ffffffff826bcea7>] ? retint_restore_args+0x6/0x13
>> > [    0.000000]  [<ffffffff811cc2cd>] kmemcheck_alloc_shadow+0x4d/0xf0
>> > [    0.000000]  [<ffffffff811c5597>] allocate_slab+0x197/0x270
>> > [    0.000000]  [<ffffffff811c569b>] new_slab+0x2b/0x190
>> > [    0.000000]  [<ffffffff811c7e7a>] ? __slab_alloc+0x5a/0x6b0
>> > [    0.000000]  [<ffffffff811c82b3>] __slab_alloc+0x493/0x6b0
>> > [    0.000000]  [<ffffffff81114f61>] ? trace_hardirqs_off_caller+0xe1/0x150
>> > [    0.000000]  [<ffffffff81886f1d>] ? trace_hardirqs_off_thunk+0x3a/0x3c
>> > [    0.000000]  [<ffffffff8395eb55>] ?
>> > debug_objects_replace_static_objects+0x2a/0x1f5
>> > [    0.000000]  [<ffffffff8395eb55>] ?
>> > debug_objects_replace_static_objects+0x2a/0x1f5
>> > [    0.000000]  [<ffffffff811c900e>] kmem_cache_alloc+0x17e/0x1a0
>> > [    0.000000]  [<ffffffff8395eb55>]
>> > debug_objects_replace_static_objects+0x2a/0x1f5
>> > [    0.000000]  [<ffffffff8395ed59>] debug_objects_mem_init+0x39/0x68
>> > [    0.000000]  [<ffffffff8392af65>] start_kernel+0x33a/0x3f8
>> > [    0.000000]  [<ffffffff8392aa13>] ? kernel_init+0x280/0x280
>> > [    0.000000]  [<ffffffff8392a2b6>] x86_64_start_reservations+0xa1/0xa6
>> > [    0.000000]  [<ffffffff8392a410>] x86_64_start_kernel+0x155/0x164
>> > [    0.000000] ---[ end trace a7919e7f17c0a725 ]---
>>
>>
>
> Something like so?
>
> ---
>  kernel/lockdep.c |    7 +++++++
>  1 files changed, 7 insertions(+), 0 deletions(-)
>
> diff --git a/kernel/lockdep.c b/kernel/lockdep.c
> index ea9ee45..a028ba9 100644
> --- a/kernel/lockdep.c
> +++ b/kernel/lockdep.c
> @@ -2718,6 +2718,13 @@ static void __lockdep_trace_alloc(gfp_t gfp_mask, unsigned long flags)
>  {
>        struct task_struct *curr = current;
>
> +       /*
> +        * We do GFP_KERNEL allocations with IRQs disabled during system
> +        * bringup..
> +        */
> +       if (unlikely(system_state != SYSTEM_RUNNING))
> +               return;
> +
>        if (unlikely(!debug_locks))
>                return;

Does the trick here.
--
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/