Re: [PATCH for 4.12] Revert "pinctrl: rockchip: avoid hardirq-unsafe functions in irq_chip"

[Heiko Stuebner <heiko@xxxxxxxxx>]

Am Freitag, 23. Juni 2017, 13:59:11 CEST schrieb Brian Norris:
> This reverts commit 88bb94216f59e10802aaf78c858a4146085faf18.
> 
> It introduced a new CONFIG_DEBUG_ATOMIC_SLEEP warning in v4.12-rc1:
> 
> [ 7226.716713] BUG: sleeping function called from invalid context at kernel/locking/mutex.c:238
> [ 7226.716716] in_atomic(): 0, irqs_disabled(): 0, pid: 1708, name: bash
> [ 7226.716722] CPU: 1 PID: 1708 Comm: bash Not tainted 4.12.0-rc6+ #1213
> [ 7226.716724] Hardware name: Google Kevin (DT)
> [ 7226.716726] Call trace:
> [ 7226.716738] [<ffffff8008089928>] dump_backtrace+0x0/0x24c
> [ 7226.716743] [<ffffff8008089b94>] show_stack+0x20/0x28
> [ 7226.716749] [<ffffff8008371370>] dump_stack+0x90/0xb0
> [ 7226.716755] [<ffffff80080cd2a0>] ___might_sleep+0x10c/0x124
> [ 7226.716760] [<ffffff80080cd330>] __might_sleep+0x78/0x88
> [ 7226.716765] [<ffffff800879e210>] mutex_lock+0x2c/0x64
> [ 7226.716771] [<ffffff80083ad678>] rockchip_irq_bus_lock+0x30/0x3c
> [ 7226.716777] [<ffffff80080f6d40>] __irq_get_desc_lock+0x78/0x98
> [ 7226.716782] [<ffffff80080f7e6c>] irq_set_irq_wake+0x44/0x12c
> [ 7226.716787] [<ffffff8008486e18>] dev_pm_arm_wake_irq+0x4c/0x58
> [ 7226.716792] [<ffffff800848b80c>] device_wakeup_arm_wake_irqs+0x3c/0x58
> [ 7226.716796] [<ffffff80084896fc>] dpm_suspend_noirq+0xf8/0x3a0
> [ 7226.716800] [<ffffff80080f1384>] suspend_devices_and_enter+0x1a4/0x9a8
> [ 7226.716803] [<ffffff80080f21ec>] pm_suspend+0x664/0x6a4
> [ 7226.716807] [<ffffff80080f04d8>] state_store+0xd4/0xf8
> ...
> 
> It was reported on -rc1, and it's still not fixed in -rc6, so it should
> just be reverted.
> 
> Cc: John Keeping <john@xxxxxxxxxxxx>
> Cc: Heiko Stuebner <heiko@xxxxxxxxx>
> Cc: Linus Walleij <linus.walleij@xxxxxxxxxx>
> Signed-off-by: Brian Norris <briannorris@xxxxxxxxxxxx>

Reviewed-by: Heiko Stuebner <heiko@xxxxxxxxx>
and should ideally be included for 4.12 still, as Brian's subject suggests.

a bit more below...

> ---
> 
> + Thomas, in case he has thoughts
> 
> Subject was "[4.12 REGRESSION] pinctrl: rockchip: sleeping function
> called from atomic context"
> 
> On Fri, May 26, 2017 at 07:19:00PM -0700, Brian Norris wrote:
> > Any thoughts? Revert the offending patch? I can spend a little more time
> > next week trying to debug what's actually going on if needed.
> > 
> > On Wed, May 17, 2017 at 03:56:34PM -0700, Brian Norris wrote:
> 
> > > The thing is, the documentation (and apparent design) suggest that
> > > calling sleeping functions from ->irq_bus_lock() is perfectly valid. I'm
> > > not 100% following the ___might_sleep() logic, but is this complaining
> > > because of the RCU read locking in device_wakeup_arm_wake_irqs()? I have
> > > CONFIG_PREEMPT_RCU and CONFIG_PREEMPT enabled, FWIW.
> 
> I've seen no reply that indicates anyone wants to fix the patch.

There were some mails exchanged with John, as David Wu did find the same
issue from a different calling context - namely disable_irq_nosync from
irq-context. I just realized they never included mailing lists though.

David's stack trace looked like
BUG: sleeping function called from invalid context at 
kernel/locking/mutex.c:620
in_atomic(): 1, irqs_disabled(): 128, pid: 141, name: irq/95-fusb302
1 lock held by irq/95-fusb302/141:
  #0:  (&(&chip->irq_lock)->rlock){......}, at: [<ffffff800859e3a0>] 
fusb_irq_disable+0x20/0x68
irq event stamp: 52
hardirqs last  enabled at (51): [<ffffff80080bcc30>] queue_work_on+0x68/0x80
hardirqs last disabled at (52): [<ffffff8008c6f41c>] 
_raw_spin_lock_irqsave+0x20/0x60
softirqs last  enabled at (0): [<ffffff800809e9ec>] 
copy_process.isra.54+0x390/0x1728
softirqs last disabled at (0): [<          (null)>]           (null)
Preemption disabled at:[<ffffff800859e3a0>] fusb_irq_disable+0x20/0x68

CPU: 5 PID: 141 Comm: irq/95-fusb302 Not tainted 4.4.70 #30
Hardware name: Rockchip RK3399 Evaluation Board v3 (Android) (DT)
Call trace:
[<ffffff800808a82c>] dump_backtrace+0x0/0x1c4
[<ffffff800808aa04>] show_stack+0x14/0x1c
[<ffffff80083c3b90>] dump_stack+0xa8/0xe0
[<ffffff80080cf560>] ___might_sleep+0x214/0x224
[<ffffff80080cf5e4>] __might_sleep+0x74/0x84
[<ffffff8008c6c1ac>] mutex_lock_nested+0x48/0x3cc
[<ffffff80083fe2b0>] rockchip_irq_bus_lock+0x28/0x34
[<ffffff800810b680>] __irq_get_desc_lock+0x68/0x88
[<ffffff800810d558>] __disable_irq_nosync+0x28/0x70
[<ffffff800810d5ac>] disable_irq_nosync+0xc/0x14
[<ffffff800859e3b4>] fusb_irq_disable+0x34/0x68
[<ffffff800859e410>] cc_interrupt_handler+0x28/0x38
[<ffffff800810cd48>] irq_thread_fn+0x28/0x68
[<ffffff800810cf80>] irq_thread+0x130/0x234
[<ffffff80080c58e8>] kthread+0x104/0x10c
[<ffffff8008083080>] ret_from_fork+0x10/0x50

So it should really be reverted until we get this sorted.


Heiko



>  drivers/pinctrl/pinctrl-rockchip.c | 44 ++++----------------------------------
>  1 file changed, 4 insertions(+), 40 deletions(-)
> 
> diff --git a/drivers/pinctrl/pinctrl-rockchip.c b/drivers/pinctrl/pinctrl-rockchip.c
> index f141aa0430b1..9dd981ddbb17 100644
> --- a/drivers/pinctrl/pinctrl-rockchip.c
> +++ b/drivers/pinctrl/pinctrl-rockchip.c
> @@ -143,9 +143,6 @@ struct rockchip_drv {
>   * @gpio_chip: gpiolib chip
>   * @grange: gpio range
>   * @slock: spinlock for the gpio bank
> - * @irq_lock: bus lock for irq chip
> - * @new_irqs: newly configured irqs which must be muxed as GPIOs in
> - *	irq_bus_sync_unlock()
>   */
>  struct rockchip_pin_bank {
>  	void __iomem			*reg_base;
> @@ -168,8 +165,6 @@ struct rockchip_pin_bank {
>  	struct pinctrl_gpio_range	grange;
>  	raw_spinlock_t			slock;
>  	u32				toggle_edge_mode;
> -	struct mutex			irq_lock;
> -	u32				new_irqs;
>  };
>  
>  #define PIN_BANK(id, pins, label)			\
> @@ -2134,12 +2129,11 @@ static int rockchip_irq_set_type(struct irq_data *d, unsigned int type)
>  	int ret;
>  
>  	/* make sure the pin is configured as gpio input */
> -	ret = rockchip_verify_mux(bank, d->hwirq, RK_FUNC_GPIO);
> +	ret = rockchip_set_mux(bank, d->hwirq, RK_FUNC_GPIO);
>  	if (ret < 0)
>  		return ret;
>  
> -	bank->new_irqs |= mask;
> -
> +	clk_enable(bank->clk);
>  	raw_spin_lock_irqsave(&bank->slock, flags);
>  
>  	data = readl_relaxed(bank->reg_base + GPIO_SWPORT_DDR);
> @@ -2197,6 +2191,7 @@ static int rockchip_irq_set_type(struct irq_data *d, unsigned int type)
>  	default:
>  		irq_gc_unlock(gc);
>  		raw_spin_unlock_irqrestore(&bank->slock, flags);
> +		clk_disable(bank->clk);
>  		return -EINVAL;
>  	}
>  
> @@ -2205,6 +2200,7 @@ static int rockchip_irq_set_type(struct irq_data *d, unsigned int type)
>  
>  	irq_gc_unlock(gc);
>  	raw_spin_unlock_irqrestore(&bank->slock, flags);
> +	clk_disable(bank->clk);
>  
>  	return 0;
>  }
> @@ -2248,34 +2244,6 @@ static void rockchip_irq_disable(struct irq_data *d)
>  	clk_disable(bank->clk);
>  }
>  
> -static void rockchip_irq_bus_lock(struct irq_data *d)
> -{
> -	struct irq_chip_generic *gc = irq_data_get_irq_chip_data(d);
> -	struct rockchip_pin_bank *bank = gc->private;
> -
> -	clk_enable(bank->clk);
> -	mutex_lock(&bank->irq_lock);
> -}
> -
> -static void rockchip_irq_bus_sync_unlock(struct irq_data *d)
> -{
> -	struct irq_chip_generic *gc = irq_data_get_irq_chip_data(d);
> -	struct rockchip_pin_bank *bank = gc->private;
> -
> -	while (bank->new_irqs) {
> -		unsigned int irq = __ffs(bank->new_irqs);
> -		int ret;
> -
> -		ret = rockchip_set_mux(bank, irq, RK_FUNC_GPIO);
> -		WARN_ON(ret < 0);
> -
> -		bank->new_irqs &= ~BIT(irq);
> -	}
> -
> -	mutex_unlock(&bank->irq_lock);
> -	clk_disable(bank->clk);
> -}
> -
>  static int rockchip_interrupts_register(struct platform_device *pdev,
>  						struct rockchip_pinctrl *info)
>  {
> @@ -2342,9 +2310,6 @@ static int rockchip_interrupts_register(struct platform_device *pdev,
>  		gc->chip_types[0].chip.irq_suspend = rockchip_irq_suspend;
>  		gc->chip_types[0].chip.irq_resume = rockchip_irq_resume;
>  		gc->chip_types[0].chip.irq_set_type = rockchip_irq_set_type;
> -		gc->chip_types[0].chip.irq_bus_lock = rockchip_irq_bus_lock;
> -		gc->chip_types[0].chip.irq_bus_sync_unlock =
> -						rockchip_irq_bus_sync_unlock;
>  		gc->wake_enabled = IRQ_MSK(bank->nr_pins);
>  
>  		irq_set_chained_handler_and_data(bank->irq,
> @@ -2518,7 +2483,6 @@ static struct rockchip_pin_ctrl *rockchip_pinctrl_get_soc_data(
>  		int bank_pins = 0;
>  
>  		raw_spin_lock_init(&bank->slock);
> -		mutex_init(&bank->irq_lock);
>  		bank->drvdata = d;
>  		bank->pin_base = ctrl->nr_pins;
>  		ctrl->nr_pins += bank->nr_pins;
> 


--
To unsubscribe from this list: send the line "unsubscribe linux-gpio" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html