Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > linux.kernel > #1270765 > unrolled thread

Re: [PATCH 6/7] mm/gfp: make gfp_zonelist return directly and bool

Started byYaowei Bai <baiyaowei@cmss.chinamobile.com>
First post2015-11-17 03:10 +0100
Last post2015-11-17 09:40 +0100
Articles 3 — 2 participants

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.


Contents

  Re: [PATCH 6/7] mm/gfp: make gfp_zonelist return directly and bool Yaowei Bai <baiyaowei@cmss.chinamobile.com> - 2015-11-17 03:10 +0100
    Re: [PATCH 6/7] mm/gfp: make gfp_zonelist return directly and bool David Rientjes <rientjes@google.com> - 2015-11-17 06:50 +0100
      Re: [PATCH 6/7] mm/gfp: make gfp_zonelist return directly and bool Yaowei Bai <baiyaowei@cmss.chinamobile.com> - 2015-11-17 09:40 +0100

#1270765 — Re: [PATCH 6/7] mm/gfp: make gfp_zonelist return directly and bool

FromYaowei Bai <baiyaowei@cmss.chinamobile.com>
Date2015-11-17 03:10 +0100
SubjectRe: [PATCH 6/7] mm/gfp: make gfp_zonelist return directly and bool
Message-ID<qvDSy-75S-3@gated-at.bofh.it>
On Mon, Nov 16, 2015 at 02:05:46AM -0800, David Rientjes wrote:
> On Mon, 16 Nov 2015, Yaowei Bai wrote:
> 
> > diff --git a/include/linux/gfp.h b/include/linux/gfp.h
> > index 6523109..1da03f5 100644
> > --- a/include/linux/gfp.h
> > +++ b/include/linux/gfp.h
> > @@ -375,12 +375,9 @@ static inline enum zone_type gfp_zone(gfp_t flags)
> >   * virtual kernel addresses to the allocated page(s).
> >   */
> >  
> > -static inline int gfp_zonelist(gfp_t flags)
> > +static inline bool gfp_zonelist(gfp_t flags)
> >  {
> > -	if (IS_ENABLED(CONFIG_NUMA) && unlikely(flags & __GFP_THISNODE))
> > -		return 1;
> > -
> > -	return 0;
> > +	return IS_ENABLED(CONFIG_NUMA) && unlikely(flags & __GFP_THISNODE);
> >  }
> >  
> >  /*
> 
> This function is used to index into a pgdat's node_zonelists[] array, bool 
> makes no sense.

Yes, you'r right, but i think hardcoding the index here is not a good idea.
How about this:

diff --git a/include/linux/gfp.h b/include/linux/gfp.h
index 6523109..14a6249 100644
--- a/include/linux/gfp.h
+++ b/include/linux/gfp.h
@@ -378,9 +378,9 @@ static inline enum zone_type gfp_zone(gfp_t flags)
 static inline int gfp_zonelist(gfp_t flags)
 {
        if (IS_ENABLED(CONFIG_NUMA) && unlikely(flags & __GFP_THISNODE))
-               return 1;
+               return ZONELIST_NOFALLBACK;
 
-       return 0;
+       return ZONELIST_FALLBACK;
 }
 
 /*
diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h
index e23a9e7..9664d6c 100644
--- a/include/linux/mmzone.h
+++ b/include/linux/mmzone.h
@@ -576,8 +576,6 @@ static inline bool zone_is_empty(struct zone *zone)
 /* Maximum number of zones on a zonelist */
 #define MAX_ZONES_PER_ZONELIST (MAX_NUMNODES * MAX_NR_ZONES)
 
-#ifdef CONFIG_NUMA
-
 /*
  * The NUMA zonelists are doubled because we need zonelists that restrict the
  * allocations to a single node for __GFP_THISNODE.
@@ -585,10 +583,13 @@ static inline bool zone_is_empty(struct zone *zone)
  * [0] : Zonelist with fallback
  * [1] : No fallback (__GFP_THISNODE)
  */
-#define MAX_ZONELISTS 2
-#else
-#define MAX_ZONELISTS 1
+enum {
+       ZONELIST_FALLBACK,
+#ifdef CONFIG_NUMA
+       ZONELIST_NOFALLBACK,
 #endif
+       MAX_ZONELISTS
+};
 
 /*
  * This struct contains information about a zone in a zonelist. It is stored


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [next] | [standalone]


#1270883

FromDavid Rientjes <rientjes@google.com>
Date2015-11-17 06:50 +0100
Message-ID<qvHjr-QF-3@gated-at.bofh.it>
In reply to#1270765
On Tue, 17 Nov 2015, Yaowei Bai wrote:

> diff --git a/include/linux/gfp.h b/include/linux/gfp.h
> index 6523109..14a6249 100644
> --- a/include/linux/gfp.h
> +++ b/include/linux/gfp.h
> @@ -378,9 +378,9 @@ static inline enum zone_type gfp_zone(gfp_t flags)
>  static inline int gfp_zonelist(gfp_t flags)
>  {
>         if (IS_ENABLED(CONFIG_NUMA) && unlikely(flags & __GFP_THISNODE))
> -               return 1;
> +               return ZONELIST_NOFALLBACK;
>  
> -       return 0;
> +       return ZONELIST_FALLBACK;
>  }
>  
>  /*
> diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h
> index e23a9e7..9664d6c 100644
> --- a/include/linux/mmzone.h
> +++ b/include/linux/mmzone.h
> @@ -576,8 +576,6 @@ static inline bool zone_is_empty(struct zone *zone)
>  /* Maximum number of zones on a zonelist */
>  #define MAX_ZONES_PER_ZONELIST (MAX_NUMNODES * MAX_NR_ZONES)
>  
> -#ifdef CONFIG_NUMA
> -
>  /*
>   * The NUMA zonelists are doubled because we need zonelists that restrict the
>   * allocations to a single node for __GFP_THISNODE.
> @@ -585,10 +583,13 @@ static inline bool zone_is_empty(struct zone *zone)
>   * [0] : Zonelist with fallback
>   * [1] : No fallback (__GFP_THISNODE)
>   */
> -#define MAX_ZONELISTS 2
> -#else
> -#define MAX_ZONELISTS 1
> +enum {
> +       ZONELIST_FALLBACK,
> +#ifdef CONFIG_NUMA
> +       ZONELIST_NOFALLBACK,
>  #endif
> +       MAX_ZONELISTS
> +};
>  
>  /*
>   * This struct contains information about a zone in a zonelist. It is stored

This is a different change than the original.  I don't see a benefit from 
it, but I have no strong feelings on it.  If someone else finds value in 
this, please update the comment when defining the enum as well.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [prev] | [next] | [standalone]


#1270959

FromYaowei Bai <baiyaowei@cmss.chinamobile.com>
Date2015-11-17 09:40 +0100
Message-ID<qvJXY-2zA-21@gated-at.bofh.it>
In reply to#1270883
On Mon, Nov 16, 2015 at 09:44:11PM -0800, David Rientjes wrote:
> On Tue, 17 Nov 2015, Yaowei Bai wrote:
> 
> > diff --git a/include/linux/gfp.h b/include/linux/gfp.h
> > index 6523109..14a6249 100644
> > --- a/include/linux/gfp.h
> > +++ b/include/linux/gfp.h
> > @@ -378,9 +378,9 @@ static inline enum zone_type gfp_zone(gfp_t flags)
> >  static inline int gfp_zonelist(gfp_t flags)
> >  {
> >         if (IS_ENABLED(CONFIG_NUMA) && unlikely(flags & __GFP_THISNODE))
> > -               return 1;
> > +               return ZONELIST_NOFALLBACK;
> >  
> > -       return 0;
> > +       return ZONELIST_FALLBACK;
> >  }
> >  
> >  /*
> > diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h
> > index e23a9e7..9664d6c 100644
> > --- a/include/linux/mmzone.h
> > +++ b/include/linux/mmzone.h
> > @@ -576,8 +576,6 @@ static inline bool zone_is_empty(struct zone *zone)
> >  /* Maximum number of zones on a zonelist */
> >  #define MAX_ZONES_PER_ZONELIST (MAX_NUMNODES * MAX_NR_ZONES)
> >  
> > -#ifdef CONFIG_NUMA
> > -
> >  /*
> >   * The NUMA zonelists are doubled because we need zonelists that restrict the
> >   * allocations to a single node for __GFP_THISNODE.
> > @@ -585,10 +583,13 @@ static inline bool zone_is_empty(struct zone *zone)
> >   * [0] : Zonelist with fallback
> >   * [1] : No fallback (__GFP_THISNODE)
> >   */
> > -#define MAX_ZONELISTS 2
> > -#else
> > -#define MAX_ZONELISTS 1
> > +enum {
> > +       ZONELIST_FALLBACK,
> > +#ifdef CONFIG_NUMA
> > +       ZONELIST_NOFALLBACK,
> >  #endif
> > +       MAX_ZONELISTS
> > +};
> >  
> >  /*
> >   * This struct contains information about a zone in a zonelist. It is stored
> 
> This is a different change than the original. 

The original patch doesn't make sense as you said so let's drop it.

> I don't see a benefit from 
> it, but I have no strong feelings on it.  If someone else finds value in 
> this, please update the comment when defining the enum as well.

OK, i'll send a update patch to review first and if nobody disagrees with it i will
resend this patchset.


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [prev] | [standalone]


Back to top | Article view | linux.kernel


csiph-web