diff mbox

ARM: SAMSUNG: Use spin_lock_{irqsave,irqrestore} in clk_set_rate

Message ID 1346998102-5317-1-git-send-email-tushar.behera@linaro.org
State Accepted
Commit d6838a62b4d36d3e2791cffe155586973b20a381
Headers show

Commit Message

Tushar Behera Sept. 7, 2012, 6:08 a.m. UTC
The spinlock clocks_lock can be held during ISR, hence it is not safe to
hold that lock with disabling interrupts.

It fixes following potential deadlock.

Comments

Tushar Behera Sept. 14, 2012, 5:02 a.m. UTC | #1
Ping !

On 09/07/2012 11:38 AM, Tushar Behera wrote:
> The spinlock clocks_lock can be held during ISR, hence it is not safe to
> hold that lock with disabling interrupts.
> 
> It fixes following potential deadlock.
> 
> =========================================================
> [ INFO: possible irq lock inversion dependency detected ]
> 3.6.0-rc4+ #2 Not tainted
> ---------------------------------------------------------
> swapper/0/1 just changed the state of lock:
>  (&(&host->lock)->rlock){-.....}, at: [<c027fb0d>] sdhci_irq+0x15/0x564
> but this lock took another, HARDIRQ-unsafe lock in the past:
>  (clocks_lock){+.+...}
> 
> and interrupts could create inverse lock ordering between them.
> 
> other info that might help us debug this:
>  Possible interrupt unsafe locking scenario:
> 
>        CPU0                    CPU1
>        ----                    ----
>   lock(clocks_lock);
>                                local_irq_disable();
>                                lock(&(&host->lock)->rlock);
>                                lock(clocks_lock);
>   <Interrupt>
>     lock(&(&host->lock)->rlock);
> 
>  *** DEADLOCK ***
> 
> Signed-off-by: Tushar Behera <tushar.behera@linaro.org>
> ---
>  arch/arm/plat-samsung/clock.c |    5 +++--
>  1 files changed, 3 insertions(+), 2 deletions(-)
> 
> diff --git a/arch/arm/plat-samsung/clock.c b/arch/arm/plat-samsung/clock.c
> index 65c5eca..9b71719 100644
> --- a/arch/arm/plat-samsung/clock.c
> +++ b/arch/arm/plat-samsung/clock.c
> @@ -144,6 +144,7 @@ long clk_round_rate(struct clk *clk, unsigned long rate)
>  
>  int clk_set_rate(struct clk *clk, unsigned long rate)
>  {
> +	unsigned long flags;
>  	int ret;
>  
>  	if (IS_ERR(clk))
> @@ -159,9 +160,9 @@ int clk_set_rate(struct clk *clk, unsigned long rate)
>  	if (clk->ops == NULL || clk->ops->set_rate == NULL)
>  		return -EINVAL;
>  
> -	spin_lock(&clocks_lock);
> +	spin_lock_irqsave(&clocks_lock, flags);
>  	ret = (clk->ops->set_rate)(clk, rate);
> -	spin_unlock(&clocks_lock);
> +	spin_unlock_irqrestore(&clocks_lock, flags);
>  
>  	return ret;
>  }
>
Kukjin Kim Sept. 17, 2012, 11:25 a.m. UTC | #2
Tushar Behera wrote:
> 
> The spinlock clocks_lock can be held during ISR, hence it is not safe to
> hold that lock with disabling interrupts.
> 
> It fixes following potential deadlock.
> 
> =========================================================
> [ INFO: possible irq lock inversion dependency detected ]
> 3.6.0-rc4+ #2 Not tainted
> ---------------------------------------------------------
> swapper/0/1 just changed the state of lock:
>  (&(&host->lock)->rlock){-.....}, at: [<c027fb0d>] sdhci_irq+0x15/0x564
> but this lock took another, HARDIRQ-unsafe lock in the past:
>  (clocks_lock){+.+...}
> 
> and interrupts could create inverse lock ordering between them.
> 
> other info that might help us debug this:
>  Possible interrupt unsafe locking scenario:
> 
>        CPU0                    CPU1
>        ----                    ----
>   lock(clocks_lock);
>                                local_irq_disable();
>                                lock(&(&host->lock)->rlock);
>                                lock(clocks_lock);
>   <Interrupt>
>     lock(&(&host->lock)->rlock);
> 
>  *** DEADLOCK ***
> 
> Signed-off-by: Tushar Behera <tushar.behera@linaro.org>
> ---
>  arch/arm/plat-samsung/clock.c |    5 +++--
>  1 files changed, 3 insertions(+), 2 deletions(-)
> 
> diff --git a/arch/arm/plat-samsung/clock.c b/arch/arm/plat-samsung/clock.c
> index 65c5eca..9b71719 100644
> --- a/arch/arm/plat-samsung/clock.c
> +++ b/arch/arm/plat-samsung/clock.c
> @@ -144,6 +144,7 @@ long clk_round_rate(struct clk *clk, unsigned long
> rate)
> 
>  int clk_set_rate(struct clk *clk, unsigned long rate)
>  {
> +	unsigned long flags;
>  	int ret;
> 
>  	if (IS_ERR(clk))
> @@ -159,9 +160,9 @@ int clk_set_rate(struct clk *clk, unsigned long rate)
>  	if (clk->ops == NULL || clk->ops->set_rate == NULL)
>  		return -EINVAL;
> 
> -	spin_lock(&clocks_lock);
> +	spin_lock_irqsave(&clocks_lock, flags);
>  	ret = (clk->ops->set_rate)(clk, rate);
> -	spin_unlock(&clocks_lock);
> +	spin_unlock_irqrestore(&clocks_lock, flags);
> 
>  	return ret;
>  }
> --
> 1.7.4.1

Looks OK, applied.
Thanks.

Best regards,
Kgene.
--
Kukjin Kim <kgene.kim@samsung.com>, Senior Engineer,
SW Solution Development Team, Samsung Electronics Co., Ltd.
diff mbox

Patch

=========================================================
[ INFO: possible irq lock inversion dependency detected ]
3.6.0-rc4+ #2 Not tainted
---------------------------------------------------------
swapper/0/1 just changed the state of lock:
 (&(&host->lock)->rlock){-.....}, at: [<c027fb0d>] sdhci_irq+0x15/0x564
but this lock took another, HARDIRQ-unsafe lock in the past:
 (clocks_lock){+.+...}

and interrupts could create inverse lock ordering between them.

other info that might help us debug this:
 Possible interrupt unsafe locking scenario:

       CPU0                    CPU1
       ----                    ----
  lock(clocks_lock);
                               local_irq_disable();
                               lock(&(&host->lock)->rlock);
                               lock(clocks_lock);
  <Interrupt>
    lock(&(&host->lock)->rlock);

 *** DEADLOCK ***

Signed-off-by: Tushar Behera <tushar.behera@linaro.org>
---
 arch/arm/plat-samsung/clock.c |    5 +++--
 1 files changed, 3 insertions(+), 2 deletions(-)

diff --git a/arch/arm/plat-samsung/clock.c b/arch/arm/plat-samsung/clock.c
index 65c5eca..9b71719 100644
--- a/arch/arm/plat-samsung/clock.c
+++ b/arch/arm/plat-samsung/clock.c
@@ -144,6 +144,7 @@  long clk_round_rate(struct clk *clk, unsigned long rate)
 
 int clk_set_rate(struct clk *clk, unsigned long rate)
 {
+	unsigned long flags;
 	int ret;
 
 	if (IS_ERR(clk))
@@ -159,9 +160,9 @@  int clk_set_rate(struct clk *clk, unsigned long rate)
 	if (clk->ops == NULL || clk->ops->set_rate == NULL)
 		return -EINVAL;
 
-	spin_lock(&clocks_lock);
+	spin_lock_irqsave(&clocks_lock, flags);
 	ret = (clk->ops->set_rate)(clk, rate);
-	spin_unlock(&clocks_lock);
+	spin_unlock_irqrestore(&clocks_lock, flags);
 
 	return ret;
 }