[PATCH] Makefiles: Disable unused-variable warning (was: Re: [PATCH 1/6] memcg: fix unused variable warning)

[PATCH] Makefiles: Disable unused-variable warning (was: Re: [PATCH 1/6] memcg: fix unused variable warning)

[Michal Hocko]

On Sat 24-12-11 05:00:14, Kirill A. Shutemov wrote:
> From: "Kirill A. Shutemov" <kirill@shutemov.name>
> 
> mm/memcontrol.c: In function ‘memcg_check_events’:
> mm/memcontrol.c:784:22: warning: unused variable ‘do_numainfo’ [-Wunused-variable]
> 
> Signed-off-by: Kirill A. Shutemov <kirill@shutemov.name>
> ---
>  mm/memcontrol.c |    7 ++++---
>  1 files changed, 4 insertions(+), 3 deletions(-)
> 
> diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> index d643bd6..a5e92bd 100644
> --- a/mm/memcontrol.c
> +++ b/mm/memcontrol.c
> @@ -781,14 +781,15 @@ static void memcg_check_events(struct mem_cgroup *memcg, struct page *page)
>  	/* threshold event is triggered in finer grain than soft limit */
>  	if (unlikely(mem_cgroup_event_ratelimit(memcg,
>  						MEM_CGROUP_TARGET_THRESH))) {
> -		bool do_softlimit, do_numainfo;
> +		bool do_softlimit;
>  
> -		do_softlimit = mem_cgroup_event_ratelimit(memcg,
> -						MEM_CGROUP_TARGET_SOFTLIMIT);
>  #if MAX_NUMNODES > 1
> +		bool do_numainfo;
>  		do_numainfo = mem_cgroup_event_ratelimit(memcg,
>  						MEM_CGROUP_TARGET_NUMAINFO);
>  #endif
> +		do_softlimit = mem_cgroup_event_ratelimit(memcg,
> +						MEM_CGROUP_TARGET_SOFTLIMIT);

I don't like this very much. Maybe we should get rid of both do_* and
do it with flags? But maybe it is not worth the additional code at
all...

Anyway, I am wondering why unused-but-set-variable is disabled while
unused-variable is enabled. Shouldn't we just disable it as well rather
than workaround this in the code? The warning is just pure noise in this
case.
What about something like:
---
From e1136891fe86eacf9212b2144f80ff6b75b10194 Mon Sep 17 00:00:00 2001
From: Michal Hocko <mhocko@suse.cz>
Date: Tue, 27 Dec 2011 14:53:06 +0100
Subject: [PATCH] Makefiles: Disable unused-variable warning

We are already disabling unused-but-set-variable and Wunused-variable
produces some noise as well.

Signed-off-by: Michal Hocko <mhocko@suse.cz>
---
 Makefile |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)

diff --git a/Makefile b/Makefile
index ea51081..25c76f3 100644
--- a/Makefile
+++ b/Makefile
@@ -578,6 +578,7 @@ endif
 # This warning generated too much noise in a regular build.
 # Use make W=1 to enable this warning (see scripts/Makefile.build)
 KBUILD_CFLAGS += $(call cc-disable-warning, unused-but-set-variable)
+KBUILD_CFLAGS += $(call cc-disable-warning, unused-variable)
 
 ifdef CONFIG_FRAME_POINTER
 KBUILD_CFLAGS	+= -fno-omit-frame-pointer -fno-optimize-sibling-calls
-- 
1.7.7.3



-- 
Michal Hocko
SUSE Labs
SUSE LINUX s.r.o.
Lihovarska 1060/12
190 00 Praha 9    
Czech Republic

Re: [PATCH] Makefiles: Disable unused-variable warning (was: Re: [PATCH 1/6] memcg: fix unused variable warning)

[Kirill A. Shutemov]

On Tue, Dec 27, 2011 at 02:57:52PM +0100, Michal Hocko wrote:
> On Sat 24-12-11 05:00:14, Kirill A. Shutemov wrote:
> > From: "Kirill A. Shutemov" <kirill@shutemov.name>
> > 
> > mm/memcontrol.c: In function ‘memcg_check_events’:
> > mm/memcontrol.c:784:22: warning: unused variable ‘do_numainfo’ [-Wunused-variable]
> > 
> > Signed-off-by: Kirill A. Shutemov <kirill@shutemov.name>
> > ---
> >  mm/memcontrol.c |    7 ++++---
> >  1 files changed, 4 insertions(+), 3 deletions(-)
> > 
> > diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> > index d643bd6..a5e92bd 100644
> > --- a/mm/memcontrol.c
> > +++ b/mm/memcontrol.c
> > @@ -781,14 +781,15 @@ static void memcg_check_events(struct mem_cgroup *memcg, struct page *page)
> >  	/* threshold event is triggered in finer grain than soft limit */
> >  	if (unlikely(mem_cgroup_event_ratelimit(memcg,
> >  						MEM_CGROUP_TARGET_THRESH))) {
> > -		bool do_softlimit, do_numainfo;
> > +		bool do_softlimit;
> >  
> > -		do_softlimit = mem_cgroup_event_ratelimit(memcg,
> > -						MEM_CGROUP_TARGET_SOFTLIMIT);
> >  #if MAX_NUMNODES > 1
> > +		bool do_numainfo;
> >  		do_numainfo = mem_cgroup_event_ratelimit(memcg,
> >  						MEM_CGROUP_TARGET_NUMAINFO);
> >  #endif
> > +		do_softlimit = mem_cgroup_event_ratelimit(memcg,
> > +						MEM_CGROUP_TARGET_SOFTLIMIT);
> 
> I don't like this very much. Maybe we should get rid of both do_* and
> do it with flags? But maybe it is not worth the additional code at
> all...

Something like this (untested):

====
From f57e1a2e1aaaa167c75b963d5bf12fcbdd3331b8 Mon Sep 17 00:00:00 2001
From: "Kirill A. Shutemov" <kirill@shutemov.name>
Date: Tue, 27 Dec 2011 20:17:13 +0200
Subject: [PATCH] memcg: cleanup memcg_check_events()

Signed-off-by: Kirill A. Shutemov <kirill@shutemov.name>
---
 mm/memcontrol.c |   42 ++++++++++++++++++++++++------------------
 1 files changed, 24 insertions(+), 18 deletions(-)

diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index d643bd6..40c2236 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -108,11 +108,12 @@ enum mem_cgroup_events_index {
  * than using jiffies etc. to handle periodic memcg event.
  */
 enum mem_cgroup_events_target {
-	MEM_CGROUP_TARGET_THRESH,
-	MEM_CGROUP_TARGET_SOFTLIMIT,
-	MEM_CGROUP_TARGET_NUMAINFO,
-	MEM_CGROUP_NTARGETS,
+	MEM_CGROUP_TARGET_THRESH	= BIT(1),
+	MEM_CGROUP_TARGET_SOFTLIMIT	= BIT(2),
+	MEM_CGROUP_TARGET_NUMAINFO	= BIT(3),
 };
+#define MEM_CGROUP_NTARGETS 3
+
 #define THRESHOLDS_EVENTS_TARGET (128)
 #define SOFTLIMIT_EVENTS_TARGET (1024)
 #define NUMAINFO_EVENTS_TARGET	(1024)
@@ -743,7 +744,7 @@ static unsigned long mem_cgroup_nr_lru_pages(struct mem_cgroup *memcg,
 	return total;
 }
 
-static bool mem_cgroup_event_ratelimit(struct mem_cgroup *memcg,
+static int mem_cgroup_event_ratelimit(struct mem_cgroup *memcg,
 				       enum mem_cgroup_events_target target)
 {
 	unsigned long val, next;
@@ -766,9 +767,9 @@ static bool mem_cgroup_event_ratelimit(struct mem_cgroup *memcg,
 			break;
 		}
 		__this_cpu_write(memcg->stat->targets[target], next);
-		return true;
+		return target;
 	}
-	return false;
+	return 0;
 }
 
 /*
@@ -777,29 +778,34 @@ static bool mem_cgroup_event_ratelimit(struct mem_cgroup *memcg,
  */
 static void memcg_check_events(struct mem_cgroup *memcg, struct page *page)
 {
+	int flags;
+
 	preempt_disable();
-	/* threshold event is triggered in finer grain than soft limit */
-	if (unlikely(mem_cgroup_event_ratelimit(memcg,
-						MEM_CGROUP_TARGET_THRESH))) {
-		bool do_softlimit, do_numainfo;
+	flags = mem_cgroup_event_ratelimit(memcg, MEM_CGROUP_TARGET_THRESH);
 
-		do_softlimit = mem_cgroup_event_ratelimit(memcg,
+	/*
+	 * Threshold event is triggered in finer grain than soft limit
+	 * and numainfo
+	 */
+	if (unlikely(flags)) {
+		flags |= mem_cgroup_event_ratelimit(memcg,
 						MEM_CGROUP_TARGET_SOFTLIMIT);
 #if MAX_NUMNODES > 1
-		do_numainfo = mem_cgroup_event_ratelimit(memcg,
+		flags |= mem_cgroup_event_ratelimit(memcg,
 						MEM_CGROUP_TARGET_NUMAINFO);
 #endif
-		preempt_enable();
+	}
+	preempt_enable();
 
+	if (unlikely(flags)) {
 		mem_cgroup_threshold(memcg);
-		if (unlikely(do_softlimit))
+		if (unlikely(flags & MEM_CGROUP_TARGET_SOFTLIMIT))
 			mem_cgroup_update_tree(memcg, page);
 #if MAX_NUMNODES > 1
-		if (unlikely(do_numainfo))
+		if (unlikely(flags & MEM_CGROUP_TARGET_NUMAINFO))
 			atomic_inc(&memcg->numainfo_events);
 #endif
-	} else
-		preempt_enable();
+	}
 }
 
 struct mem_cgroup *mem_cgroup_from_cont(struct cgroup *cont)
-- 
1.7.7.3

-- 
 Kirill A. Shutemov

Re: [PATCH] Makefiles: Disable unused-variable warning (was: Re: [PATCH 1/6] memcg: fix unused variable warning)

[Michal Hocko]

On Tue 27-12-11 20:26:13, Kirill A. Shutemov wrote:
> On Tue, Dec 27, 2011 at 02:57:52PM +0100, Michal Hocko wrote:
> > On Sat 24-12-11 05:00:14, Kirill A. Shutemov wrote:
> > > From: "Kirill A. Shutemov" <kirill@shutemov.name>
> > > 
> > > mm/memcontrol.c: In function ‘memcg_check_events’:
> > > mm/memcontrol.c:784:22: warning: unused variable ‘do_numainfo’ [-Wunused-variable]
> > > 
> > > Signed-off-by: Kirill A. Shutemov <kirill@shutemov.name>
> > > ---
> > >  mm/memcontrol.c |    7 ++++---
> > >  1 files changed, 4 insertions(+), 3 deletions(-)
> > > 
> > > diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> > > index d643bd6..a5e92bd 100644
> > > --- a/mm/memcontrol.c
> > > +++ b/mm/memcontrol.c
> > > @@ -781,14 +781,15 @@ static void memcg_check_events(struct mem_cgroup *memcg, struct page *page)
> > >  	/* threshold event is triggered in finer grain than soft limit */
> > >  	if (unlikely(mem_cgroup_event_ratelimit(memcg,
> > >  						MEM_CGROUP_TARGET_THRESH))) {
> > > -		bool do_softlimit, do_numainfo;
> > > +		bool do_softlimit;
> > >  
> > > -		do_softlimit = mem_cgroup_event_ratelimit(memcg,
> > > -						MEM_CGROUP_TARGET_SOFTLIMIT);
> > >  #if MAX_NUMNODES > 1
> > > +		bool do_numainfo;
> > >  		do_numainfo = mem_cgroup_event_ratelimit(memcg,
> > >  						MEM_CGROUP_TARGET_NUMAINFO);
> > >  #endif
> > > +		do_softlimit = mem_cgroup_event_ratelimit(memcg,
> > > +						MEM_CGROUP_TARGET_SOFTLIMIT);
> > 
> > I don't like this very much. Maybe we should get rid of both do_* and
> > do it with flags? But maybe it is not worth the additional code at
> > all...
> 
> Something like this (untested):
> ====
> From f57e1a2e1aaaa167c75b963d5bf12fcbdd3331b8 Mon Sep 17 00:00:00 2001
> From: "Kirill A. Shutemov" <kirill@shutemov.name>
> Date: Tue, 27 Dec 2011 20:17:13 +0200
> Subject: [PATCH] memcg: cleanup memcg_check_events()
> 
> Signed-off-by: Kirill A. Shutemov <kirill@shutemov.name>

The patch looks correct but I am still not sure this is worth fixing in
the code rather than disabling Wunused-variable.

> ---
>  mm/memcontrol.c |   42 ++++++++++++++++++++++++------------------
>  1 files changed, 24 insertions(+), 18 deletions(-)
> 
> diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> index d643bd6..40c2236 100644
> --- a/mm/memcontrol.c
> +++ b/mm/memcontrol.c
> @@ -108,11 +108,12 @@ enum mem_cgroup_events_index {
>   * than using jiffies etc. to handle periodic memcg event.
>   */
>  enum mem_cgroup_events_target {
> -	MEM_CGROUP_TARGET_THRESH,
> -	MEM_CGROUP_TARGET_SOFTLIMIT,
> -	MEM_CGROUP_TARGET_NUMAINFO,
> -	MEM_CGROUP_NTARGETS,
> +	MEM_CGROUP_TARGET_THRESH	= BIT(1),
> +	MEM_CGROUP_TARGET_SOFTLIMIT	= BIT(2),
> +	MEM_CGROUP_TARGET_NUMAINFO	= BIT(3),
>  };
> +#define MEM_CGROUP_NTARGETS 3
> +
>  #define THRESHOLDS_EVENTS_TARGET (128)
>  #define SOFTLIMIT_EVENTS_TARGET (1024)
>  #define NUMAINFO_EVENTS_TARGET	(1024)
> @@ -743,7 +744,7 @@ static unsigned long mem_cgroup_nr_lru_pages(struct mem_cgroup *memcg,
>  	return total;
>  }
>  
> -static bool mem_cgroup_event_ratelimit(struct mem_cgroup *memcg,
> +static int mem_cgroup_event_ratelimit(struct mem_cgroup *memcg,
>  				       enum mem_cgroup_events_target target)
>  {
>  	unsigned long val, next;
> @@ -766,9 +767,9 @@ static bool mem_cgroup_event_ratelimit(struct mem_cgroup *memcg,
>  			break;
>  		}
>  		__this_cpu_write(memcg->stat->targets[target], next);
> -		return true;
> +		return target;
>  	}
> -	return false;
> +	return 0;
>  }
>  
>  /*
> @@ -777,29 +778,34 @@ static bool mem_cgroup_event_ratelimit(struct mem_cgroup *memcg,
>   */
>  static void memcg_check_events(struct mem_cgroup *memcg, struct page *page)
>  {
> +	int flags;
> +
>  	preempt_disable();
> -	/* threshold event is triggered in finer grain than soft limit */
> -	if (unlikely(mem_cgroup_event_ratelimit(memcg,
> -						MEM_CGROUP_TARGET_THRESH))) {
> -		bool do_softlimit, do_numainfo;
> +	flags = mem_cgroup_event_ratelimit(memcg, MEM_CGROUP_TARGET_THRESH);
>  
> -		do_softlimit = mem_cgroup_event_ratelimit(memcg,
> +	/*
> +	 * Threshold event is triggered in finer grain than soft limit
> +	 * and numainfo
> +	 */
> +	if (unlikely(flags)) {
> +		flags |= mem_cgroup_event_ratelimit(memcg,
>  						MEM_CGROUP_TARGET_SOFTLIMIT);
>  #if MAX_NUMNODES > 1
> -		do_numainfo = mem_cgroup_event_ratelimit(memcg,
> +		flags |= mem_cgroup_event_ratelimit(memcg,
>  						MEM_CGROUP_TARGET_NUMAINFO);
>  #endif
> -		preempt_enable();
> +	}
> +	preempt_enable();
>  
> +	if (unlikely(flags)) {
>  		mem_cgroup_threshold(memcg);
> -		if (unlikely(do_softlimit))
> +		if (unlikely(flags & MEM_CGROUP_TARGET_SOFTLIMIT))
>  			mem_cgroup_update_tree(memcg, page);
>  #if MAX_NUMNODES > 1
> -		if (unlikely(do_numainfo))
> +		if (unlikely(flags & MEM_CGROUP_TARGET_NUMAINFO))
>  			atomic_inc(&memcg->numainfo_events);
>  #endif
> -	} else
> -		preempt_enable();
> +	}
>  }
>  
>  struct mem_cgroup *mem_cgroup_from_cont(struct cgroup *cont)
> -- 
> 1.7.7.3

-- 
Michal Hocko
SUSE Labs
SUSE LINUX s.r.o.
Lihovarska 1060/12
190 00 Praha 9    
Czech Republic

Re: [PATCH] Makefiles: Disable unused-variable warning (was: Re: [PATCH 1/6] memcg: fix unused variable warning)

[Kirill A. Shutemov]

On Thu, Dec 29, 2011 at 11:42:30AM +0100, Michal Hocko wrote:
> On Tue 27-12-11 20:26:13, Kirill A. Shutemov wrote:
> > On Tue, Dec 27, 2011 at 02:57:52PM +0100, Michal Hocko wrote:
> > > On Sat 24-12-11 05:00:14, Kirill A. Shutemov wrote:
> > > > From: "Kirill A. Shutemov" <kirill@shutemov.name>
> > > > 
> > > > mm/memcontrol.c: In function ‘memcg_check_events’:
> > > > mm/memcontrol.c:784:22: warning: unused variable ‘do_numainfo’ [-Wunused-variable]
> > > > 
> > > > Signed-off-by: Kirill A. Shutemov <kirill@shutemov.name>
> > > > ---
> > > >  mm/memcontrol.c |    7 ++++---
> > > >  1 files changed, 4 insertions(+), 3 deletions(-)
> > > > 
> > > > diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> > > > index d643bd6..a5e92bd 100644
> > > > --- a/mm/memcontrol.c
> > > > +++ b/mm/memcontrol.c
> > > > @@ -781,14 +781,15 @@ static void memcg_check_events(struct mem_cgroup *memcg, struct page *page)
> > > >  	/* threshold event is triggered in finer grain than soft limit */
> > > >  	if (unlikely(mem_cgroup_event_ratelimit(memcg,
> > > >  						MEM_CGROUP_TARGET_THRESH))) {
> > > > -		bool do_softlimit, do_numainfo;
> > > > +		bool do_softlimit;
> > > >  
> > > > -		do_softlimit = mem_cgroup_event_ratelimit(memcg,
> > > > -						MEM_CGROUP_TARGET_SOFTLIMIT);
> > > >  #if MAX_NUMNODES > 1
> > > > +		bool do_numainfo;
> > > >  		do_numainfo = mem_cgroup_event_ratelimit(memcg,
> > > >  						MEM_CGROUP_TARGET_NUMAINFO);
> > > >  #endif
> > > > +		do_softlimit = mem_cgroup_event_ratelimit(memcg,
> > > > +						MEM_CGROUP_TARGET_SOFTLIMIT);
> > > 
> > > I don't like this very much. Maybe we should get rid of both do_* and
> > > do it with flags? But maybe it is not worth the additional code at
> > > all...
> > 
> > Something like this (untested):
> > ====
> > From f57e1a2e1aaaa167c75b963d5bf12fcbdd3331b8 Mon Sep 17 00:00:00 2001
> > From: "Kirill A. Shutemov" <kirill@shutemov.name>
> > Date: Tue, 27 Dec 2011 20:17:13 +0200
> > Subject: [PATCH] memcg: cleanup memcg_check_events()
> > 
> > Signed-off-by: Kirill A. Shutemov <kirill@shutemov.name>
> 
> The patch looks correct but I am still not sure this is worth fixing in
> the code rather than disabling Wunused-variable.

Does it look better then original code from your POV?

It's a bit cleaner, I think.

> 
> > ---
> >  mm/memcontrol.c |   42 ++++++++++++++++++++++++------------------
> >  1 files changed, 24 insertions(+), 18 deletions(-)
> > 
> > diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> > index d643bd6..40c2236 100644
> > --- a/mm/memcontrol.c
> > +++ b/mm/memcontrol.c
> > @@ -108,11 +108,12 @@ enum mem_cgroup_events_index {
> >   * than using jiffies etc. to handle periodic memcg event.
> >   */
> >  enum mem_cgroup_events_target {
> > -	MEM_CGROUP_TARGET_THRESH,
> > -	MEM_CGROUP_TARGET_SOFTLIMIT,
> > -	MEM_CGROUP_TARGET_NUMAINFO,
> > -	MEM_CGROUP_NTARGETS,
> > +	MEM_CGROUP_TARGET_THRESH	= BIT(1),
> > +	MEM_CGROUP_TARGET_SOFTLIMIT	= BIT(2),
> > +	MEM_CGROUP_TARGET_NUMAINFO	= BIT(3),
> >  };
> > +#define MEM_CGROUP_NTARGETS 3
> > +
> >  #define THRESHOLDS_EVENTS_TARGET (128)
> >  #define SOFTLIMIT_EVENTS_TARGET (1024)
> >  #define NUMAINFO_EVENTS_TARGET	(1024)
> > @@ -743,7 +744,7 @@ static unsigned long mem_cgroup_nr_lru_pages(struct mem_cgroup *memcg,
> >  	return total;
> >  }
> >  
> > -static bool mem_cgroup_event_ratelimit(struct mem_cgroup *memcg,
> > +static int mem_cgroup_event_ratelimit(struct mem_cgroup *memcg,
> >  				       enum mem_cgroup_events_target target)
> >  {
> >  	unsigned long val, next;
> > @@ -766,9 +767,9 @@ static bool mem_cgroup_event_ratelimit(struct mem_cgroup *memcg,
> >  			break;
> >  		}
> >  		__this_cpu_write(memcg->stat->targets[target], next);
> > -		return true;
> > +		return target;
> >  	}
> > -	return false;
> > +	return 0;
> >  }
> >  
> >  /*
> > @@ -777,29 +778,34 @@ static bool mem_cgroup_event_ratelimit(struct mem_cgroup *memcg,
> >   */
> >  static void memcg_check_events(struct mem_cgroup *memcg, struct page *page)
> >  {
> > +	int flags;
> > +
> >  	preempt_disable();
> > -	/* threshold event is triggered in finer grain than soft limit */
> > -	if (unlikely(mem_cgroup_event_ratelimit(memcg,
> > -						MEM_CGROUP_TARGET_THRESH))) {
> > -		bool do_softlimit, do_numainfo;
> > +	flags = mem_cgroup_event_ratelimit(memcg, MEM_CGROUP_TARGET_THRESH);
> >  
> > -		do_softlimit = mem_cgroup_event_ratelimit(memcg,
> > +	/*
> > +	 * Threshold event is triggered in finer grain than soft limit
> > +	 * and numainfo
> > +	 */
> > +	if (unlikely(flags)) {
> > +		flags |= mem_cgroup_event_ratelimit(memcg,
> >  						MEM_CGROUP_TARGET_SOFTLIMIT);
> >  #if MAX_NUMNODES > 1
> > -		do_numainfo = mem_cgroup_event_ratelimit(memcg,
> > +		flags |= mem_cgroup_event_ratelimit(memcg,
> >  						MEM_CGROUP_TARGET_NUMAINFO);
> >  #endif
> > -		preempt_enable();
> > +	}
> > +	preempt_enable();
> >  
> > +	if (unlikely(flags)) {
> >  		mem_cgroup_threshold(memcg);
> > -		if (unlikely(do_softlimit))
> > +		if (unlikely(flags & MEM_CGROUP_TARGET_SOFTLIMIT))
> >  			mem_cgroup_update_tree(memcg, page);
> >  #if MAX_NUMNODES > 1
> > -		if (unlikely(do_numainfo))
> > +		if (unlikely(flags & MEM_CGROUP_TARGET_NUMAINFO))
> >  			atomic_inc(&memcg->numainfo_events);
> >  #endif
> > -	} else
> > -		preempt_enable();
> > +	}
> >  }
> >  
> >  struct mem_cgroup *mem_cgroup_from_cont(struct cgroup *cont)
> > -- 
> > 1.7.7.3
> 
> -- 
> Michal Hocko
> SUSE Labs
> SUSE LINUX s.r.o.
> Lihovarska 1060/12
> 190 00 Praha 9    
> Czech Republic

-- 
 Kirill A. Shutemov

Re: [PATCH] Makefiles: Disable unused-variable warning (was: Re: [PATCH 1/6] memcg: fix unused variable warning)

[Michal Hocko]

On Thu 29-12-11 13:08:50, Kirill A. Shutemov wrote:
> On Thu, Dec 29, 2011 at 11:42:30AM +0100, Michal Hocko wrote:
> > On Tue 27-12-11 20:26:13, Kirill A. Shutemov wrote:
> > > On Tue, Dec 27, 2011 at 02:57:52PM +0100, Michal Hocko wrote:
> > > > On Sat 24-12-11 05:00:14, Kirill A. Shutemov wrote:
> > > > > From: "Kirill A. Shutemov" <kirill@shutemov.name>
> > > > > 
> > > > > mm/memcontrol.c: In function ‘memcg_check_events’:
> > > > > mm/memcontrol.c:784:22: warning: unused variable ‘do_numainfo’ [-Wunused-variable]
> > > > > 
> > > > > Signed-off-by: Kirill A. Shutemov <kirill@shutemov.name>
> > > > > ---
> > > > >  mm/memcontrol.c |    7 ++++---
> > > > >  1 files changed, 4 insertions(+), 3 deletions(-)
> > > > > 
> > > > > diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> > > > > index d643bd6..a5e92bd 100644
> > > > > --- a/mm/memcontrol.c
> > > > > +++ b/mm/memcontrol.c
> > > > > @@ -781,14 +781,15 @@ static void memcg_check_events(struct mem_cgroup *memcg, struct page *page)
> > > > >  	/* threshold event is triggered in finer grain than soft limit */
> > > > >  	if (unlikely(mem_cgroup_event_ratelimit(memcg,
> > > > >  						MEM_CGROUP_TARGET_THRESH))) {
> > > > > -		bool do_softlimit, do_numainfo;
> > > > > +		bool do_softlimit;
> > > > >  
> > > > > -		do_softlimit = mem_cgroup_event_ratelimit(memcg,
> > > > > -						MEM_CGROUP_TARGET_SOFTLIMIT);
> > > > >  #if MAX_NUMNODES > 1
> > > > > +		bool do_numainfo;
> > > > >  		do_numainfo = mem_cgroup_event_ratelimit(memcg,
> > > > >  						MEM_CGROUP_TARGET_NUMAINFO);
> > > > >  #endif
> > > > > +		do_softlimit = mem_cgroup_event_ratelimit(memcg,
> > > > > +						MEM_CGROUP_TARGET_SOFTLIMIT);
> > > > 
> > > > I don't like this very much. Maybe we should get rid of both do_* and
> > > > do it with flags? But maybe it is not worth the additional code at
> > > > all...
> > > 
> > > Something like this (untested):
> > > ====
> > > From f57e1a2e1aaaa167c75b963d5bf12fcbdd3331b8 Mon Sep 17 00:00:00 2001
> > > From: "Kirill A. Shutemov" <kirill@shutemov.name>
> > > Date: Tue, 27 Dec 2011 20:17:13 +0200
> > > Subject: [PATCH] memcg: cleanup memcg_check_events()
> > > 
> > > Signed-off-by: Kirill A. Shutemov <kirill@shutemov.name>
> > 
> > The patch looks correct but I am still not sure this is worth fixing in
> > the code rather than disabling Wunused-variable.
> 
> Does it look better then original code from your POV?
> 
> It's a bit cleaner, I think.

Yes it is less hakish than the previous one which relied on having
#ifdef block before any other code in the block.

> 
> > 
> > > ---
> > >  mm/memcontrol.c |   42 ++++++++++++++++++++++++------------------
> > >  1 files changed, 24 insertions(+), 18 deletions(-)
> > > 
> > > diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> > > index d643bd6..40c2236 100644
> > > --- a/mm/memcontrol.c
> > > +++ b/mm/memcontrol.c
> > > @@ -108,11 +108,12 @@ enum mem_cgroup_events_index {
> > >   * than using jiffies etc. to handle periodic memcg event.
> > >   */
> > >  enum mem_cgroup_events_target {
> > > -	MEM_CGROUP_TARGET_THRESH,
> > > -	MEM_CGROUP_TARGET_SOFTLIMIT,
> > > -	MEM_CGROUP_TARGET_NUMAINFO,
> > > -	MEM_CGROUP_NTARGETS,
> > > +	MEM_CGROUP_TARGET_THRESH	= BIT(1),
> > > +	MEM_CGROUP_TARGET_SOFTLIMIT	= BIT(2),
> > > +	MEM_CGROUP_TARGET_NUMAINFO	= BIT(3),
> > >  };
> > > +#define MEM_CGROUP_NTARGETS 3
> > > +
> > >  #define THRESHOLDS_EVENTS_TARGET (128)
> > >  #define SOFTLIMIT_EVENTS_TARGET (1024)
> > >  #define NUMAINFO_EVENTS_TARGET	(1024)
> > > @@ -743,7 +744,7 @@ static unsigned long mem_cgroup_nr_lru_pages(struct mem_cgroup *memcg,
> > >  	return total;
> > >  }
> > >  
> > > -static bool mem_cgroup_event_ratelimit(struct mem_cgroup *memcg,
> > > +static int mem_cgroup_event_ratelimit(struct mem_cgroup *memcg,
> > >  				       enum mem_cgroup_events_target target)
> > >  {
> > >  	unsigned long val, next;
> > > @@ -766,9 +767,9 @@ static bool mem_cgroup_event_ratelimit(struct mem_cgroup *memcg,
> > >  			break;
> > >  		}
> > >  		__this_cpu_write(memcg->stat->targets[target], next);
> > > -		return true;
> > > +		return target;
> > >  	}
> > > -	return false;
> > > +	return 0;
> > >  }
> > >  
> > >  /*
> > > @@ -777,29 +778,34 @@ static bool mem_cgroup_event_ratelimit(struct mem_cgroup *memcg,
> > >   */
> > >  static void memcg_check_events(struct mem_cgroup *memcg, struct page *page)
> > >  {
> > > +	int flags;
> > > +
> > >  	preempt_disable();
> > > -	/* threshold event is triggered in finer grain than soft limit */
> > > -	if (unlikely(mem_cgroup_event_ratelimit(memcg,
> > > -						MEM_CGROUP_TARGET_THRESH))) {
> > > -		bool do_softlimit, do_numainfo;
> > > +	flags = mem_cgroup_event_ratelimit(memcg, MEM_CGROUP_TARGET_THRESH);
> > >  
> > > -		do_softlimit = mem_cgroup_event_ratelimit(memcg,
> > > +	/*
> > > +	 * Threshold event is triggered in finer grain than soft limit
> > > +	 * and numainfo
> > > +	 */
> > > +	if (unlikely(flags)) {
> > > +		flags |= mem_cgroup_event_ratelimit(memcg,
> > >  						MEM_CGROUP_TARGET_SOFTLIMIT);
> > >  #if MAX_NUMNODES > 1
> > > -		do_numainfo = mem_cgroup_event_ratelimit(memcg,
> > > +		flags |= mem_cgroup_event_ratelimit(memcg,
> > >  						MEM_CGROUP_TARGET_NUMAINFO);
> > >  #endif
> > > -		preempt_enable();
> > > +	}
> > > +	preempt_enable();
> > >  
> > > +	if (unlikely(flags)) {
> > >  		mem_cgroup_threshold(memcg);
> > > -		if (unlikely(do_softlimit))
> > > +		if (unlikely(flags & MEM_CGROUP_TARGET_SOFTLIMIT))
> > >  			mem_cgroup_update_tree(memcg, page);
> > >  #if MAX_NUMNODES > 1
> > > -		if (unlikely(do_numainfo))
> > > +		if (unlikely(flags & MEM_CGROUP_TARGET_NUMAINFO))
> > >  			atomic_inc(&memcg->numainfo_events);
> > >  #endif
> > > -	} else
> > > -		preempt_enable();
> > > +	}
> > >  }
> > >  
> > >  struct mem_cgroup *mem_cgroup_from_cont(struct cgroup *cont)
> > > -- 
> > > 1.7.7.3
> > 
> > -- 
> > Michal Hocko
> > SUSE Labs
> > SUSE LINUX s.r.o.
> > Lihovarska 1060/12
> > 190 00 Praha 9    
> > Czech Republic
> 
> -- 
>  Kirill A. Shutemov
> --
> To unsubscribe from this list: send the line "unsubscribe cgroups" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

-- 
Michal Hocko
SUSE Labs
SUSE LINUX s.r.o.
Lihovarska 1060/12
190 00 Praha 9    
Czech Republic

Re: [PATCH] Makefiles: Disable unused-variable warning

[Michal Marek]

Dne 27.12.2011 14:57, Michal Hocko napsal(a):
> Anyway, I am wondering why unused-but-set-variable is disabled while
> unused-variable is enabled.

unused-but-set-variable was disabled, because it was a new warning in
gcc 4.6 and produced too much noise relatively to its severity. A make
W=1 build of x86_64_defconfig gives:
$ grep -c 'Wunused-but-set-variable' log
77
$ grep -c 'Wunused-variable' log
0

More exotic configuration will probably result in a couple of unused
variable warnings, but that IMO no reason to disable them globally.

> Shouldn't we just disable it as well rather
> than workaround this in the code? The warning is just pure noise in this
> case.

If it's noise in a particular case, there is always the option to add

CFLAGS_memcontrol.o := $(call cc-disable-warning, unused-variable)

to the respective Makefile.

Michal

Re: [PATCH] Makefiles: Disable unused-variable warning

[Michal Hocko]

On Sun 08-01-12 16:01:17, Michal Marek wrote:
> Dne 27.12.2011 14:57, Michal Hocko napsal(a):
> > Anyway, I am wondering why unused-but-set-variable is disabled while
> > unused-variable is enabled.
> 
> unused-but-set-variable was disabled, because it was a new warning in
> gcc 4.6 and produced too much noise relatively to its severity. A make
> W=1 build of x86_64_defconfig gives:
> $ grep -c 'Wunused-but-set-variable' log
> 77
> $ grep -c 'Wunused-variable' log
> 0
> 
> More exotic configuration will probably result in a couple of unused
> variable warnings, but that IMO no reason to disable them globally.

OK.

> > Shouldn't we just disable it as well rather
> > than workaround this in the code? The warning is just pure noise in this
> > case.
> 
> If it's noise in a particular case, there is always the option to add
> 
> CFLAGS_memcontrol.o := $(call cc-disable-warning, unused-variable)

I would like to prevent from local cflags hacks. Moreover the code will
go away so I guess it doesn't make much sense to play tricks here.

> 
> to the respective Makefile.
> 
> Michal

Thanks

-- 
Michal Hocko
SUSE Labs
SUSE LINUX s.r.o.
Lihovarska 1060/12
190 00 Praha 9    
Czech Republic