[PATCH 0/2] watchdog: avoid races in watchdog_nmi_{en|disable} functions - follow up patches

[PATCH 0/2] watchdog: avoid races in watchdog_nmi_{en|disable} functions - follow up patches

[Ulrich Obergfell]

This small series follows up to patch discussion related
to $subject patch series which is now in linux-next at:

- introduce watchdog_park_threads() and watchdog_unpark_threads()
  https://git.kernel.org/cgit/linux/kernel/git/next/linux-next.git/commit/kernel/watchdog.c?id=89f536017155c7f76c8a620e6aed03809e53cd4c

- introduce watchdog_suspend() and watchdog_resume()
  https://git.kernel.org/cgit/linux/kernel/git/next/linux-next.git/commit/kernel/watchdog.c?id=e0dd9ee3ecf3394e93343cf9542d8ca1b54eca08

- use park/unpark functions in update_watchdog_all_cpus()
  https://git.kernel.org/cgit/linux/kernel/git/next/linux-next.git/commit/kernel/watchdog.c?id=eef1a759acfb98766fef20afca9e06f014299e3f

- use suspend/resume interface in fixup_ht_bug()
  https://git.kernel.org/cgit/linux/kernel/git/next/linux-next.git/commit/kernel/watchdog.c?id=77592a82363dd508ce3dbe4eaaa2c6c45b966ebb

Changes based on patch discussion:

- Patch 1/2: rename watchdog_{suspend|resume}, add comment blocks
  http://marc.info/?l=linux-kernel&m=143844050610220&w=2
  http://marc.info/?l=linux-kernel&m=143876224114821&w=2

- Patch 2/2: use pr_debug() instead of pr_info()
  http://marc.info/?l=linux-kernel&m=143869949229461&w=2

Ulrich Obergfell (2):
  watchdog: rename watchdog_suspend() and watchdog_resume()
  watchdog: use pr_debug() in fixup_ht_bug() failure path

 arch/x86/kernel/cpu/perf_event_intel.c |  6 +++---
 include/linux/nmi.h                    |  8 ++++----
 kernel/watchdog.c                      | 26 ++++++++++++++++++++++----
 3 files changed, 29 insertions(+), 11 deletions(-)

-- 
1.7.11.7

[PATCH 1/2] watchdog: rename watchdog_suspend() and watchdog_resume()

[Ulrich Obergfell]

Rename watchdog_suspend() to lockup_detector_suspend() and
watchdog_resume() to lockup_detector_resume() to avoid
confusion with the watchdog subsystem and to be consistent
with the existing name lockup_detector_init().

Also provide comment blocks to explain the watchdog_running
and watchdog_suspended variables and their relationship.

Signed-off-by: Ulrich Obergfell <uobergfe@redhat.com>
---
 arch/x86/kernel/cpu/perf_event_intel.c |  4 ++--
 include/linux/nmi.h                    |  8 ++++----
 kernel/watchdog.c                      | 26 ++++++++++++++++++++++----
 3 files changed, 28 insertions(+), 10 deletions(-)

diff --git a/arch/x86/kernel/cpu/perf_event_intel.c b/arch/x86/kernel/cpu/perf_event_intel.c
index d4e1b0c..0357bf7 100644
--- a/arch/x86/kernel/cpu/perf_event_intel.c
+++ b/arch/x86/kernel/cpu/perf_event_intel.c
@@ -3368,7 +3368,7 @@ static __init int fixup_ht_bug(void)
 		return 0;
 	}
 
-	if (watchdog_suspend() != 0) {
+	if (lockup_detector_suspend() != 0) {
 		pr_info("failed to disable PMU erratum BJ122, BV98, HSD29 workaround\n");
 		return 0;
 	}
@@ -3379,7 +3379,7 @@ static __init int fixup_ht_bug(void)
 	x86_pmu.commit_scheduling = NULL;
 	x86_pmu.stop_scheduling = NULL;
 
-	watchdog_resume();
+	lockup_detector_resume();
 
 	get_online_cpus();
 
diff --git a/include/linux/nmi.h b/include/linux/nmi.h
index 46e28e9e..78488e0 100644
--- a/include/linux/nmi.h
+++ b/include/linux/nmi.h
@@ -84,15 +84,15 @@ extern int proc_watchdog_thresh(struct ctl_table *, int ,
 				void __user *, size_t *, loff_t *);
 extern int proc_watchdog_cpumask(struct ctl_table *, int,
 				 void __user *, size_t *, loff_t *);
-extern int watchdog_suspend(void);
-extern void watchdog_resume(void);
+extern int lockup_detector_suspend(void);
+extern void lockup_detector_resume(void);
 #else
-static inline int watchdog_suspend(void)
+static inline int lockup_detector_suspend(void)
 {
 	return 0;
 }
 
-static inline void watchdog_resume(void)
+static inline void lockup_detector_resume(void)
 {
 }
 #endif
diff --git a/kernel/watchdog.c b/kernel/watchdog.c
index 69666f4..64ed1c3 100644
--- a/kernel/watchdog.c
+++ b/kernel/watchdog.c
@@ -67,8 +67,26 @@ unsigned long *watchdog_cpumask_bits = cpumask_bits(&watchdog_cpumask);
 #define for_each_watchdog_cpu(cpu) \
 	for_each_cpu_and((cpu), cpu_online_mask, &watchdog_cpumask)
 
-static int __read_mostly watchdog_suspended;
+/*
+ * The 'watchdog_running' variable is set to 1 when the watchdog threads
+ * are registered/started and is set to 0 when the watchdog threads are
+ * unregistered/stopped, so it is an indicator whether the threads exist.
+ */
 static int __read_mostly watchdog_running;
+/*
+ * If a subsystem has a need to deactivate the watchdog temporarily, it
+ * can use the suspend/resume interface to achieve this. The content of
+ * the 'watchdog_suspended' variable reflects this state. Existing threads
+ * are parked/unparked by the lockup_detector_{suspend|resume} functions
+ * (see comment blocks pertaining to those functions for further details).
+ *
+ * 'watchdog_suspended' also prevents threads from being registered/started
+ * or unregistered/stopped via parameters in /proc/sys/kernel, so the state
+ * of 'watchdog_running' cannot change while the watchdog is deactivated
+ * temporarily (see related code in 'proc' handlers).
+ */
+static int __read_mostly watchdog_suspended;
+
 static u64 __read_mostly sample_period;
 
 static DEFINE_PER_CPU(unsigned long, watchdog_touch_ts);
@@ -669,7 +687,7 @@ static void watchdog_unpark_threads(void)
 /*
  * Suspend the hard and soft lockup detector by parking the watchdog threads.
  */
-int watchdog_suspend(void)
+int lockup_detector_suspend(void)
 {
 	int ret = 0;
 
@@ -679,7 +697,7 @@ int watchdog_suspend(void)
 	 * the 'watchdog_suspended' variable). If the watchdog threads are
 	 * running, the first caller takes care that they will be parked.
 	 * The state of 'watchdog_running' cannot change while a suspend
-	 * request is active (see related changes in 'proc' handlers).
+	 * request is active (see related code in 'proc' handlers).
 	 */
 	if (watchdog_running && !watchdog_suspended)
 		ret = watchdog_park_threads();
@@ -695,7 +713,7 @@ int watchdog_suspend(void)
 /*
  * Resume the hard and soft lockup detector by unparking the watchdog threads.
  */
-void watchdog_resume(void)
+void lockup_detector_resume(void)
 {
 	mutex_lock(&watchdog_proc_mutex);
 
-- 
1.7.11.7

Re: [PATCH 1/2] watchdog: rename watchdog_suspend() and watchdog_resume()

[Aaron Tomlin]

[-- Attachment #1: Type: text/plain, Size: 4913 bytes --]

On Fri 2015-08-07 11:58 +0200, Ulrich Obergfell wrote:
> Rename watchdog_suspend() to lockup_detector_suspend() and
> watchdog_resume() to lockup_detector_resume() to avoid
> confusion with the watchdog subsystem and to be consistent
> with the existing name lockup_detector_init().
> 
> Also provide comment blocks to explain the watchdog_running
> and watchdog_suspended variables and their relationship.
> 
> Signed-off-by: Ulrich Obergfell <uobergfe@redhat.com>
> ---
>  arch/x86/kernel/cpu/perf_event_intel.c |  4 ++--
>  include/linux/nmi.h                    |  8 ++++----
>  kernel/watchdog.c                      | 26 ++++++++++++++++++++++----
>  3 files changed, 28 insertions(+), 10 deletions(-)
> 
> diff --git a/arch/x86/kernel/cpu/perf_event_intel.c b/arch/x86/kernel/cpu/perf_event_intel.c
> index d4e1b0c..0357bf7 100644
> --- a/arch/x86/kernel/cpu/perf_event_intel.c
> +++ b/arch/x86/kernel/cpu/perf_event_intel.c
> @@ -3368,7 +3368,7 @@ static __init int fixup_ht_bug(void)
>  		return 0;
>  	}
>  
> -	if (watchdog_suspend() != 0) {
> +	if (lockup_detector_suspend() != 0) {
>  		pr_info("failed to disable PMU erratum BJ122, BV98, HSD29 workaround\n");
>  		return 0;
>  	}
> @@ -3379,7 +3379,7 @@ static __init int fixup_ht_bug(void)
>  	x86_pmu.commit_scheduling = NULL;
>  	x86_pmu.stop_scheduling = NULL;
>  
> -	watchdog_resume();
> +	lockup_detector_resume();
>  
>  	get_online_cpus();
>  
> diff --git a/include/linux/nmi.h b/include/linux/nmi.h
> index 46e28e9e..78488e0 100644
> --- a/include/linux/nmi.h
> +++ b/include/linux/nmi.h
> @@ -84,15 +84,15 @@ extern int proc_watchdog_thresh(struct ctl_table *, int ,
>  				void __user *, size_t *, loff_t *);
>  extern int proc_watchdog_cpumask(struct ctl_table *, int,
>  				 void __user *, size_t *, loff_t *);
> -extern int watchdog_suspend(void);
> -extern void watchdog_resume(void);
> +extern int lockup_detector_suspend(void);
> +extern void lockup_detector_resume(void);
>  #else
> -static inline int watchdog_suspend(void)
> +static inline int lockup_detector_suspend(void)
>  {
>  	return 0;
>  }
>  
> -static inline void watchdog_resume(void)
> +static inline void lockup_detector_resume(void)
>  {
>  }
>  #endif
> diff --git a/kernel/watchdog.c b/kernel/watchdog.c
> index 69666f4..64ed1c3 100644
> --- a/kernel/watchdog.c
> +++ b/kernel/watchdog.c
> @@ -67,8 +67,26 @@ unsigned long *watchdog_cpumask_bits = cpumask_bits(&watchdog_cpumask);
>  #define for_each_watchdog_cpu(cpu) \
>  	for_each_cpu_and((cpu), cpu_online_mask, &watchdog_cpumask)
>  
> -static int __read_mostly watchdog_suspended;
> +/*
> + * The 'watchdog_running' variable is set to 1 when the watchdog threads
> + * are registered/started and is set to 0 when the watchdog threads are
> + * unregistered/stopped, so it is an indicator whether the threads exist.
> + */
>  static int __read_mostly watchdog_running;
> +/*
> + * If a subsystem has a need to deactivate the watchdog temporarily, it
> + * can use the suspend/resume interface to achieve this. The content of
> + * the 'watchdog_suspended' variable reflects this state. Existing threads
> + * are parked/unparked by the lockup_detector_{suspend|resume} functions
> + * (see comment blocks pertaining to those functions for further details).
> + *
> + * 'watchdog_suspended' also prevents threads from being registered/started
> + * or unregistered/stopped via parameters in /proc/sys/kernel, so the state
> + * of 'watchdog_running' cannot change while the watchdog is deactivated
> + * temporarily (see related code in 'proc' handlers).
> + */
> +static int __read_mostly watchdog_suspended;
> +
>  static u64 __read_mostly sample_period;
>  
>  static DEFINE_PER_CPU(unsigned long, watchdog_touch_ts);
> @@ -669,7 +687,7 @@ static void watchdog_unpark_threads(void)
>  /*
>   * Suspend the hard and soft lockup detector by parking the watchdog threads.
>   */
> -int watchdog_suspend(void)
> +int lockup_detector_suspend(void)
>  {
>  	int ret = 0;
>  
> @@ -679,7 +697,7 @@ int watchdog_suspend(void)
>  	 * the 'watchdog_suspended' variable). If the watchdog threads are
>  	 * running, the first caller takes care that they will be parked.
>  	 * The state of 'watchdog_running' cannot change while a suspend
> -	 * request is active (see related changes in 'proc' handlers).
> +	 * request is active (see related code in 'proc' handlers).
>  	 */
>  	if (watchdog_running && !watchdog_suspended)
>  		ret = watchdog_park_threads();
> @@ -695,7 +713,7 @@ int watchdog_suspend(void)
>  /*
>   * Resume the hard and soft lockup detector by unparking the watchdog threads.
>   */
> -void watchdog_resume(void)
> +void lockup_detector_resume(void)
>  {
>  	mutex_lock(&watchdog_proc_mutex);
>  

Reviewed-by: Aaron Tomlin <atomlin@redhat.com>

-- 
Aaron Tomlin

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 819 bytes --]

[PATCH 2/2] watchdog: use pr_debug() in fixup_ht_bug() failure path

[Ulrich Obergfell]

Signed-off-by: Ulrich Obergfell <uobergfe@redhat.com>
---
 arch/x86/kernel/cpu/perf_event_intel.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/x86/kernel/cpu/perf_event_intel.c b/arch/x86/kernel/cpu/perf_event_intel.c
index 0357bf7..abb25c3 100644
--- a/arch/x86/kernel/cpu/perf_event_intel.c
+++ b/arch/x86/kernel/cpu/perf_event_intel.c
@@ -3369,7 +3369,7 @@ static __init int fixup_ht_bug(void)
 	}
 
 	if (lockup_detector_suspend() != 0) {
-		pr_info("failed to disable PMU erratum BJ122, BV98, HSD29 workaround\n");
+		pr_debug("failed to disable PMU erratum BJ122, BV98, HSD29 workaround\n");
 		return 0;
 	}
 
-- 
1.7.11.7

Re: [PATCH 2/2] watchdog: use pr_debug() in fixup_ht_bug() failure path

[Aaron Tomlin]

[-- Attachment #1: Type: text/plain, Size: 830 bytes --]

On Fri 2015-08-07 11:58 +0200, Ulrich Obergfell wrote:
> 
> Signed-off-by: Ulrich Obergfell <uobergfe@redhat.com>
> ---
>  arch/x86/kernel/cpu/perf_event_intel.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/arch/x86/kernel/cpu/perf_event_intel.c b/arch/x86/kernel/cpu/perf_event_intel.c
> index 0357bf7..abb25c3 100644
> --- a/arch/x86/kernel/cpu/perf_event_intel.c
> +++ b/arch/x86/kernel/cpu/perf_event_intel.c
> @@ -3369,7 +3369,7 @@ static __init int fixup_ht_bug(void)
>  	}
>  
>  	if (lockup_detector_suspend() != 0) {
> -		pr_info("failed to disable PMU erratum BJ122, BV98, HSD29 workaround\n");
> +		pr_debug("failed to disable PMU erratum BJ122, BV98, HSD29 workaround\n");
>  		return 0;
>  	}
>  

Reviewed-by: Aaron Tomlin <atomlin@redhat.com>

-- 
Aaron Tomlin

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 819 bytes --]