linux-pm.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH] Thermal: x86_pkg_temp: change spin lock
@ 2013-09-24 18:05 Srinivas Pandruvada
  2013-09-25 13:47 ` Zhang Rui
  0 siblings, 1 reply; 2+ messages in thread
From: Srinivas Pandruvada @ 2013-09-24 18:05 UTC (permalink / raw)
  To: eduardo.valentin, rui.zhang; +Cc: linux-pm, Srinivas Pandruvada

x86_pkg_temp receives thermal notifications via a callback from a
therm_throt driver, where thermal interrupts are processed.
This callback is pkg_temp_thermal_platform_thermal_notify. Here to
avoid multiple interrupts from cores in a package, we disable the
source and also set a variable to avoid scheduling delayed work function.
This variable is protected via spin_lock_irqsave. On one buggy platform,
we still receiving interrupts even if the source is disabled. This
can cause deadlock/lockdep warning, when interrupt is generated while under
spinlock in work function.
Change spin_lock to spin_lock_irqsave and spin_unlock to
spin_unlock_irqrestore as the data it is trying to protect can also
be modified in a notification call called from interrupt handler.

Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
---
 drivers/thermal/x86_pkg_temp_thermal.c | 14 ++++++++------
 1 file changed, 8 insertions(+), 6 deletions(-)

diff --git a/drivers/thermal/x86_pkg_temp_thermal.c b/drivers/thermal/x86_pkg_temp_thermal.c
index f36950e..7722cb9 100644
--- a/drivers/thermal/x86_pkg_temp_thermal.c
+++ b/drivers/thermal/x86_pkg_temp_thermal.c
@@ -316,18 +316,19 @@ static void pkg_temp_thermal_threshold_work_fn(struct work_struct *work)
 	int phy_id = topology_physical_package_id(cpu);
 	struct phy_dev_entry *phdev = pkg_temp_thermal_get_phy_entry(cpu);
 	bool notify = false;
+	unsigned long flags;
 
 	if (!phdev)
 		return;
 
-	spin_lock(&pkg_work_lock);
+	spin_lock_irqsave(&pkg_work_lock, flags);
 	++pkg_work_cnt;
 	if (unlikely(phy_id > max_phy_id)) {
-		spin_unlock(&pkg_work_lock);
+		spin_unlock_irqrestore(&pkg_work_lock, flags);
 		return;
 	}
 	pkg_work_scheduled[phy_id] = 0;
-	spin_unlock(&pkg_work_lock);
+	spin_unlock_irqrestore(&pkg_work_lock, flags);
 
 	enable_pkg_thres_interrupt();
 	rdmsrl(MSR_IA32_PACKAGE_THERM_STATUS, msr_val);
@@ -397,6 +398,7 @@ static int pkg_temp_thermal_device_add(unsigned int cpu)
 	int thres_count;
 	u32 eax, ebx, ecx, edx;
 	u8 *temp;
+	unsigned long flags;
 
 	cpuid(6, &eax, &ebx, &ecx, &edx);
 	thres_count = ebx & 0x07;
@@ -420,19 +422,19 @@ static int pkg_temp_thermal_device_add(unsigned int cpu)
 		goto err_ret_unlock;
 	}
 
-	spin_lock(&pkg_work_lock);
+	spin_lock_irqsave(&pkg_work_lock, flags);
 	if (topology_physical_package_id(cpu) > max_phy_id)
 		max_phy_id = topology_physical_package_id(cpu);
 	temp = krealloc(pkg_work_scheduled,
 			(max_phy_id+1) * sizeof(u8), GFP_ATOMIC);
 	if (!temp) {
-		spin_unlock(&pkg_work_lock);
+		spin_unlock_irqrestore(&pkg_work_lock, flags);
 		err = -ENOMEM;
 		goto err_ret_free;
 	}
 	pkg_work_scheduled = temp;
 	pkg_work_scheduled[topology_physical_package_id(cpu)] = 0;
-	spin_unlock(&pkg_work_lock);
+	spin_unlock_irqrestore(&pkg_work_lock, flags);
 
 	phy_dev_entry->phys_proc_id = topology_physical_package_id(cpu);
 	phy_dev_entry->first_cpu = cpu;
-- 
1.8.3.1


^ permalink raw reply related	[flat|nested] 2+ messages in thread

* Re: [PATCH] Thermal: x86_pkg_temp: change spin lock
  2013-09-24 18:05 [PATCH] Thermal: x86_pkg_temp: change spin lock Srinivas Pandruvada
@ 2013-09-25 13:47 ` Zhang Rui
  0 siblings, 0 replies; 2+ messages in thread
From: Zhang Rui @ 2013-09-25 13:47 UTC (permalink / raw)
  To: Srinivas Pandruvada; +Cc: eduardo.valentin, linux-pm

On Tue, 2013-09-24 at 11:05 -0700, Srinivas Pandruvada wrote:
> x86_pkg_temp receives thermal notifications via a callback from a
> therm_throt driver, where thermal interrupts are processed.
> This callback is pkg_temp_thermal_platform_thermal_notify. Here to
> avoid multiple interrupts from cores in a package, we disable the
> source and also set a variable to avoid scheduling delayed work function.
> This variable is protected via spin_lock_irqsave. On one buggy platform,
> we still receiving interrupts even if the source is disabled. This
> can cause deadlock/lockdep warning, when interrupt is generated while under
> spinlock in work function.
> Change spin_lock to spin_lock_irqsave and spin_unlock to
> spin_unlock_irqrestore as the data it is trying to protect can also
> be modified in a notification call called from interrupt handler.
> 
> Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>

applied to thermal .next.

thanks,
rui
> ---
>  drivers/thermal/x86_pkg_temp_thermal.c | 14 ++++++++------
>  1 file changed, 8 insertions(+), 6 deletions(-)
> 
> diff --git a/drivers/thermal/x86_pkg_temp_thermal.c b/drivers/thermal/x86_pkg_temp_thermal.c
> index f36950e..7722cb9 100644
> --- a/drivers/thermal/x86_pkg_temp_thermal.c
> +++ b/drivers/thermal/x86_pkg_temp_thermal.c
> @@ -316,18 +316,19 @@ static void pkg_temp_thermal_threshold_work_fn(struct work_struct *work)
>  	int phy_id = topology_physical_package_id(cpu);
>  	struct phy_dev_entry *phdev = pkg_temp_thermal_get_phy_entry(cpu);
>  	bool notify = false;
> +	unsigned long flags;
>  
>  	if (!phdev)
>  		return;
>  
> -	spin_lock(&pkg_work_lock);
> +	spin_lock_irqsave(&pkg_work_lock, flags);
>  	++pkg_work_cnt;
>  	if (unlikely(phy_id > max_phy_id)) {
> -		spin_unlock(&pkg_work_lock);
> +		spin_unlock_irqrestore(&pkg_work_lock, flags);
>  		return;
>  	}
>  	pkg_work_scheduled[phy_id] = 0;
> -	spin_unlock(&pkg_work_lock);
> +	spin_unlock_irqrestore(&pkg_work_lock, flags);
>  
>  	enable_pkg_thres_interrupt();
>  	rdmsrl(MSR_IA32_PACKAGE_THERM_STATUS, msr_val);
> @@ -397,6 +398,7 @@ static int pkg_temp_thermal_device_add(unsigned int cpu)
>  	int thres_count;
>  	u32 eax, ebx, ecx, edx;
>  	u8 *temp;
> +	unsigned long flags;
>  
>  	cpuid(6, &eax, &ebx, &ecx, &edx);
>  	thres_count = ebx & 0x07;
> @@ -420,19 +422,19 @@ static int pkg_temp_thermal_device_add(unsigned int cpu)
>  		goto err_ret_unlock;
>  	}
>  
> -	spin_lock(&pkg_work_lock);
> +	spin_lock_irqsave(&pkg_work_lock, flags);
>  	if (topology_physical_package_id(cpu) > max_phy_id)
>  		max_phy_id = topology_physical_package_id(cpu);
>  	temp = krealloc(pkg_work_scheduled,
>  			(max_phy_id+1) * sizeof(u8), GFP_ATOMIC);
>  	if (!temp) {
> -		spin_unlock(&pkg_work_lock);
> +		spin_unlock_irqrestore(&pkg_work_lock, flags);
>  		err = -ENOMEM;
>  		goto err_ret_free;
>  	}
>  	pkg_work_scheduled = temp;
>  	pkg_work_scheduled[topology_physical_package_id(cpu)] = 0;
> -	spin_unlock(&pkg_work_lock);
> +	spin_unlock_irqrestore(&pkg_work_lock, flags);
>  
>  	phy_dev_entry->phys_proc_id = topology_physical_package_id(cpu);
>  	phy_dev_entry->first_cpu = cpu;



^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2013-09-25 13:48 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2013-09-24 18:05 [PATCH] Thermal: x86_pkg_temp: change spin lock Srinivas Pandruvada
2013-09-25 13:47 ` Zhang Rui

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).