Reduce CONFIG_ZONE_DMA ifdefs

Reduce CONFIG_ZONE_DMA ifdefs

[Christoph Lameter]

Add a DMA_ZONE constant that can be used to avoid #ifdef DMAs. I hope this 
will make it acceptable to remove ZONE_DMA dependent code such as the 
bouncing logic and also allow us to deal with the GFP_DMA issues in the 
SCSI layer.

Signed-off-by: Christoph Lameter <clameter@sgi.com>

Index: linux-2.6.19-rc1-mm1/include/linux/mmzone.h
===================================================================
--- linux-2.6.19-rc1-mm1.orig/include/linux/mmzone.h	2006-10-17 13:08:22.000000000 -0500
+++ linux-2.6.19-rc1-mm1/include/linux/mmzone.h	2006-10-17 13:08:51.018160800 -0500
@@ -149,6 +149,12 @@ enum zone_type {
  * match the requested limits. See gfp_zone() in include/linux/gfp.h
  */
 
+#ifdef CONFIG_ZONE_DMA
+#define DMA_ZONE 1
+#else
+#define DMA_ZONE 0
+#endif
+
 /*
  * Count the active zones.  Note that the use of defined(X) outside
  * #if and family is not necessarily defined so ensure we cannot use
Index: linux-2.6.19-rc1-mm1/mm/slab.c
===================================================================
--- linux-2.6.19-rc1-mm1.orig/mm/slab.c	2006-10-17 13:03:49.472950935 -0500
+++ linux-2.6.19-rc1-mm1/mm/slab.c	2006-10-17 13:08:08.215683664 -0500
@@ -1458,14 +1458,14 @@ void __init kmem_cache_init(void)
 					ARCH_KMALLOC_FLAGS|SLAB_PANIC,
 					NULL, NULL);
 		}
-#ifdef CONFIG_ZONE_DMA
-		sizes->cs_dmacachep = kmem_cache_create(names->name_dma,
+		if (DMA_ZONE)
+			sizes->cs_dmacachep = kmem_cache_create(
+					names->name_dma,
 					sizes->cs_size,
 					ARCH_KMALLOC_MINALIGN,
 					ARCH_KMALLOC_FLAGS|SLAB_CACHE_DMA|
 						SLAB_PANIC,
 					NULL, NULL);
-#endif
 		sizes++;
 		names++;
 	}
@@ -2298,10 +2298,8 @@ kmem_cache_create (const char *name, siz
 	cachep->slab_size = slab_size;
 	cachep->flags = flags;
 	cachep->gfpflags = 0;
-#ifdef CONFIG_ZONE_DMA
-	if (flags & SLAB_CACHE_DMA)
+	if (DMA_ZONE && (flags & SLAB_CACHE_DMA))
 		cachep->gfpflags |= GFP_DMA;
-#endif
 	cachep->buffer_size = size;
 
 	if (flags & CFLGS_OFF_SLAB) {
@@ -2624,12 +2622,12 @@ static void cache_init_objs(struct kmem_
 
 static void kmem_flagcheck(struct kmem_cache *cachep, gfp_t flags)
 {
-#ifdef CONFIG_ZONE_DMA
-	if (flags & SLAB_DMA)
-		BUG_ON(!(cachep->gfpflags & GFP_DMA));
-	else
-		BUG_ON(cachep->gfpflags & GFP_DMA);
-#endif
+	if (DMA_ZONE) {
+		if (flags & SLAB_DMA)
+			BUG_ON(!(cachep->gfpflags & GFP_DMA));
+		else
+			BUG_ON(cachep->gfpflags & GFP_DMA);
+	}
 }
 
 static void *slab_get_obj(struct kmem_cache *cachep, struct slab *slabp,

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: Reduce CONFIG_ZONE_DMA ifdefs

[Andrew Morton]

On Tue, 17 Oct 2006 11:25:07 -0700 (PDT)
Christoph Lameter <clameter@sgi.com> wrote:

> Add a DMA_ZONE constant that can be used to avoid #ifdef DMAs. I hope this 
> will make it acceptable to remove ZONE_DMA dependent code such as the 
> bouncing logic and also allow us to deal with the GFP_DMA issues in the 
> SCSI layer.
> 
> Signed-off-by: Christoph Lameter <clameter@sgi.com>
> 
> Index: linux-2.6.19-rc1-mm1/include/linux/mmzone.h
> ===================================================================
> --- linux-2.6.19-rc1-mm1.orig/include/linux/mmzone.h	2006-10-17 13:08:22.000000000 -0500
> +++ linux-2.6.19-rc1-mm1/include/linux/mmzone.h	2006-10-17 13:08:51.018160800 -0500
> @@ -149,6 +149,12 @@ enum zone_type {
>   * match the requested limits. See gfp_zone() in include/linux/gfp.h
>   */
>  
> +#ifdef CONFIG_ZONE_DMA
> +#define DMA_ZONE 1
> +#else
> +#define DMA_ZONE 0
> +#endif

This can be done in the config system.  See CONFIG_BASE_SMALL for an
example.

That would give the thing a nice name, too - say, CONFIG_HAVE_ZONE_DMA.  It
makes it obvious what's going on.

If that doesn't work out, a better name would be HAVE_ZONE_DMA or
something.  "DMA_ZONE" sounds like the number of the dma zone.


> -#ifdef CONFIG_ZONE_DMA
> -#ifdef CONFIG_ZONE_DMA
> -#ifdef CONFIG_ZONE_DMA

Only three.  Drat.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: Reduce CONFIG_ZONE_DMA ifdefs

[Christoph Lameter]

On Tue, 17 Oct 2006, Andrew Morton wrote:

> That would give the thing a nice name, too - say, CONFIG_HAVE_ZONE_DMA.  It
> makes it obvious what's going on.

Ok.

> > -#ifdef CONFIG_ZONE_DMA
> > -#ifdef CONFIG_ZONE_DMA
> > -#ifdef CONFIG_ZONE_DMA
> 
> Only three.  Drat.

The problem is that some of the definitions like ZONE_DMA become invalid 
if CONFIG_ZONE_DMA is off and I think we need to keep that to make sure 
code does not refer to invalid zones. Around those areas the #ifdef cannot 
be dropped. The slab does not use ZONE_DMA directly. It only needs to deal 
with GFP_DMA. For that it works nicely.

I hope such a scheme would also allow the switching off of the bounce 
buffer logic and various other GFP_DMA related code all over the kernel. 

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: Reduce CONFIG_ZONE_DMA ifdefs

[Christoph Lameter]

V1->V2 use CONFIG_ZONE_DMA_FLAG that was defined in mm/Kconfig.

Reduce #ifdef CONFIG_ZONE_DMA

This reduces the #ifdefs in the slab allocator by adding a new
CONFIG_ZONE_DMA_FLAG in mm/Kconfig. The definitions for the
page allocator are already minimal and orthogonal to
CONFIG_ZONE_DMA32 and CONFIG_HIGHMEM.

Signed-off-by: Christoph Lameter <clameter@sgi.com>

Index: linux-2.6.19-rc2-mm2/mm/slab.c
===================================================================
--- linux-2.6.19-rc2-mm2.orig/mm/slab.c	2006-10-23 17:13:10.372360786 -0500
+++ linux-2.6.19-rc2-mm2/mm/slab.c	2006-10-23 17:28:49.336065747 -0500
@@ -1458,14 +1458,14 @@ void __init kmem_cache_init(void)
 					ARCH_KMALLOC_FLAGS|SLAB_PANIC,
 					NULL, NULL);
 		}
-#ifdef CONFIG_ZONE_DMA
-		sizes->cs_dmacachep = kmem_cache_create(names->name_dma,
+		if (CONFIG_ZONE_DMA_FLAG)
+			sizes->cs_dmacachep = kmem_cache_create(
+					names->name_dma,
 					sizes->cs_size,
 					ARCH_KMALLOC_MINALIGN,
 					ARCH_KMALLOC_FLAGS|SLAB_CACHE_DMA|
 						SLAB_PANIC,
 					NULL, NULL);
-#endif
 		sizes++;
 		names++;
 	}
@@ -2297,10 +2297,8 @@ kmem_cache_create (const char *name, siz
 	cachep->slab_size = slab_size;
 	cachep->flags = flags;
 	cachep->gfpflags = 0;
-#ifdef CONFIG_ZONE_DMA
-	if (flags & SLAB_CACHE_DMA)
+	if (CONFIG_ZONE_DMA_FLAG && (flags & SLAB_CACHE_DMA))
 		cachep->gfpflags |= GFP_DMA;
-#endif
 	cachep->buffer_size = size;
 
 	if (flags & CFLGS_OFF_SLAB) {
@@ -2623,12 +2621,12 @@ static void cache_init_objs(struct kmem_
 
 static void kmem_flagcheck(struct kmem_cache *cachep, gfp_t flags)
 {
-#ifdef CONFIG_ZONE_DMA
-	if (flags & SLAB_DMA)
-		BUG_ON(!(cachep->gfpflags & GFP_DMA));
-	else
-		BUG_ON(cachep->gfpflags & GFP_DMA);
-#endif
+	if (CONFIG_ZONE_DMA_FLAG) {
+		if (flags & SLAB_DMA)
+			BUG_ON(!(cachep->gfpflags & GFP_DMA));
+		else
+			BUG_ON(cachep->gfpflags & GFP_DMA);
+	}
 }
 
 static void *slab_get_obj(struct kmem_cache *cachep, struct slab *slabp,
Index: linux-2.6.19-rc2-mm2/mm/Kconfig
===================================================================
--- linux-2.6.19-rc2-mm2.orig/mm/Kconfig	2006-10-23 17:13:10.382127229 -0500
+++ linux-2.6.19-rc2-mm2/mm/Kconfig	2006-10-23 17:52:25.537437185 -0500
@@ -242,3 +242,9 @@ config READAHEAD_SMOOTH_AGING
 		- have the danger of readahead thrashing(i.e. memory tight)
 
 	  This feature is only available on non-NUMA systems.
+
+config ZONE_DMA_FLAG
+	int
+	default "0" if !ZONE_DMA
+	default "1"
+

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>