Re: PM freeze does not work on AT91

[Alexander Stein <alexander.stein@systec-electronic.com>]

On Tuesday 15 July 2014 17:31:22, Zhang Rui wrote:
> On Tue, 2014-07-15 at 10:07 +0200, Alexander Stein wrote:
> > Hi,
> > 
> > I tried using the PM features in linux on my at91sam9263 custom board. Using 'echo mem > state' and 'echo standby > mem' works in general. But when using 'echo freeze > state' I get the following oops.
> > 
> > sh-3.2:/sys/power# echo freeze > state
> > [   22.559000] PM: Syncing filesystems ... done.
> > [   23.073000] Freezing user space processes ... (elapsed 0.001 seconds) done.
> > [   23.092000] Freezing remaining freezable tasks ... (elapsed 0.001 seconds) done.
> > [   23.101000] Unable to handle kernel NULL pointer dereference at virtual address 00000000
> > [   23.109000] pgd = c283c000
> > [   23.112000] [00000000] *pgd=23384831, *pte=00000000, *ppte=00000000
> > [   23.118000] Internal error: Oops: 17 [#1] PREEMPT ARM
> > [   23.118000] Modules linked in:
> > [   23.118000] CPU: 0 PID: 106 Comm: sh Not tainted 3.16.0-rc5+ #3
> > [   23.118000] task: c3369c00 ti: c33e0000 task.ti: c33e0000
> > [   23.118000] PC is at suspend_devices_and_enter+0x6c/0x32c
> > [   23.118000] LR is at pm_suspend+0x160/0x27c
> > [   23.118000] pc : [<c004111c>]    lr : [<c004153c>]    psr: 60000013
> > [   23.118000] sp : c33e1eb8  ip : 000008e0  fp : 00000000
> > [   23.118000] r10: c280480c  r9 : c2804800  r8 : 00000007
> > [   23.118000] r7 : c05527f8  r6 : 00000001  r5 : 00000001  r4 : c056ca00
> > [   23.118000] r3 : 00000000  r2 : 01ffff3f  r1 : 60000013  r0 : 00000001
> > [   23.118000] Flags: nZCv  IRQs on  FIQs on  Mode SVC_32  ISA ARM  Segment user
> > [   23.118000] Control: 0005317f  Table: 2283c000  DAC: 00000015
> > [   23.118000] Process sh (pid: 106, stack limit = 0xc33e01c0)
> > [   23.118000] Stack: (0xc33e1eb8 to 0xc33e2000)
> > [   23.118000] 1ea0:                                                       00000000 c0391454
> > [   23.118000] 1ec0: c049e549 c33e1edc 60000013 c33e1edc 00000000 c056ca00 00000000 00000001
> > [   23.118000] 1ee0: c05527f8 00000007 c2804800 c004153c c055e2bc c3064000 00000007 c282a0c0
> > [   23.118000] 1f00: c33e1f80 c0040304 c3064000 c282a0c0 c282a0c0 c01abd94 00000007 c00f0e54
> > [   23.118000] 1f20: c00f0e18 00000000 00000000 c00f0050 00000000 00000000 c33c93c0 b6fa6000
> > [   23.118000] 1f40: c33e1f80 00000007 00000007 b6fa6000 00000000 c009a19c c33c93c0 b6fa6000
> > [   23.118000] 1f60: 00000007 00000000 00000000 c33c93c0 c33c93c0 00000007 b6fa6000 c009a70c
> > [   23.118000] 1f80: 00000000 00000000 00000007 00000007 b6fa6000 b6f76b38 00000004 c0009764
> > [   23.118000] 1fa0: c33e0000 c0009600 00000007 b6fa6000 00000001 b6fa6000 00000007 00000000
> > [   23.118000] 1fc0: 00000007 b6fa6000 b6f76b38 00000004 00000007 b6fa6000 00000007 00000000
> > [   23.118000] 1fe0: 00000000 be83ba9c b6eaeadc b6f0317c 60000010 00000001 00000000 00000000
> > [   23.118000] [<c004111c>] (suspend_devices_and_enter) from [<c004153c>] (pm_suspend+0x160/0x27c)
> > [   23.118000] [<c004153c>] (pm_suspend) from [<c0040304>] (state_store+0x40/0x68)
> > [   23.118000] [<c0040304>] (state_store) from [<c01abd94>] (kobj_attr_store+0x14/0x20)
> > [   23.118000] [<c01abd94>] (kobj_attr_store) from [<c00f0e54>] (sysfs_kf_write+0x3c/0x48)
> > [   23.118000] [<c00f0e54>] (sysfs_kf_write) from [<c00f0050>] (kernfs_fop_write+0x100/0x158)
> > [   23.118000] [<c00f0050>] (kernfs_fop_write) from [<c009a19c>] (vfs_write+0xb4/0x188)
> > [   23.118000] [<c009a19c>] (vfs_write) from [<c009a70c>] (SyS_write+0x3c/0x7c)
> > [   23.118000] [<c009a70c>] (SyS_write) from [<c0009600>] (ret_fast_syscall+0x0/0x2c)
> > [   23.118000] Code: eafffff3 1afffff2 e59f32ac e5933008 (e5933000)
> > [   23.358000] ---[ end trace 9206ef60cdd758d9 ]---
> > 
> > The systems hangs here now.
> 
> can you please check if the patch below fixes the problem for you?
> 
> thanks,
> rui
> 
> From 0617bbe7890aec961828904659cf207fd817a5cf Mon Sep 17 00:00:00 2001
> From: Zhang Rui <rui.zhang@intel.com>
> Date: Tue, 15 Jul 2014 17:20:16 +0800
> Subject: [PATCH] PM: fix a NULL pointer dereference issue
> 
> This patch fixes a NULL pointer dereference issue introduced by
> commit 1f0b63866fc1be700260547be8edf8e6f0af37f2.
> 
> Reference: http://marc.info/?l=linux-pm&m=140541182017443&w=2
> Signed-off-by: Zhang Rui <rui.zhang@intel.com>
> ---
>  kernel/power/suspend.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/kernel/power/suspend.c b/kernel/power/suspend.c
> index 4dd8822..ed35a47 100644
> --- a/kernel/power/suspend.c
> +++ b/kernel/power/suspend.c
> @@ -306,7 +306,7 @@ int suspend_devices_and_enter(suspend_state_t state)
>  		error = suspend_ops->begin(state);
>  		if (error)
>  			goto Close;
> -	} else if (state == PM_SUSPEND_FREEZE && freeze_ops->begin) {
> +	} else if (state == PM_SUSPEND_FREEZE && freeze_ops && freeze_ops->begin) {
>  		error = freeze_ops->begin();
>  		if (error)
>  			goto Close;
> @@ -335,7 +335,7 @@ int suspend_devices_and_enter(suspend_state_t state)
>   Close:
>  	if (need_suspend_ops(state) && suspend_ops->end)
>  		suspend_ops->end();
> -	else if (state == PM_SUSPEND_FREEZE && freeze_ops->end)
> +	else if (state == PM_SUSPEND_FREEZE && freeze_ops && freeze_ops->end)
>  		freeze_ops->end();
>  
>  	return error;
> 

Oh, I forgot to add my testing patch which is the same as yours. Nevertheless I've taken your patch and tried again:
$ echo freeze > /sys/power/state
[   20.920000] PM: Syncing filesystems ... done.
[   21.391000] Freezing user space processes ... (elapsed 0.001 seconds) done.
[   21.405000] Freezing remaining freezable tasks ... (elapsed 0.001 seconds) done.
[   21.414000] Suspending console(s) (use no_console_suspend to debug)

The baord is freezed now here. It doesn't wakeup by GPIO keys, etc.

Regards,
Alexander