optional-zone_dma-in-the-vm-no-gfp_dma-check-in-the-slab-if-
no-config_zone_dma-is-set-reduce-config_zone_dma-ifdefs converts some
ifdefs into ifs.
One of them causes cs_dmacachep to be referenced, when that field
doesn't exist, because it's ifdefed on CONFIG_ZONE_DMA.
I'd suggest reverting that chunk.
Signed-off-by: Jeff Dike <[email protected]>
--
mm/slab.c | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)
Index: linux-2.6.18-mm/mm/slab.c
===================================================================
--- linux-2.6.18-mm.orig/mm/slab.c 2007-01-29 14:32:13.000000000 -0500
+++ linux-2.6.18-mm/mm/slab.c 2007-01-29 15:04:18.000000000 -0500
@@ -1492,14 +1492,15 @@ void __init kmem_cache_init(void)
ARCH_KMALLOC_FLAGS|SLAB_PANIC,
NULL, NULL);
}
- if (CONFIG_ZONE_DMA_FLAG)
- sizes->cs_dmacachep = kmem_cache_create(
+#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++;
}
--
Work email - jdike at linux dot intel dot com
On Mon, 29 Jan 2007, Jeff Dike wrote:
> +#if CONFIG_ZONE_DMA_FLAG
> + sizes->cs_dmacachep = kmem_cache_create(
This would need to be
#idef CONFIG_ZONE_DMA
There have been some recent changes though so I am not sure what your
codebase is.
I also hit the same issue. Here is the patch:
Fix slab build failure if !CONFIG_ZONE_DMA
I also needed this to get 2.6.20-rc6-mm2 to build. Fixes the fix by the
complainer about the fixes. #ifdef cannot be avoided since cs_dmacachep
is no longer defined.
Signed-off-by: Christoph Lameter <[email protected]>
Index: linux-2.6.20-rc6-mm2/mm/slab.c
===================================================================
--- linux-2.6.20-rc6-mm2.orig/mm/slab.c 2007-01-29 14:27:58.000000000 -0600
+++ linux-2.6.20-rc6-mm2/mm/slab.c 2007-01-29 14:33:04.270329009 -0600
@@ -1503,6 +1503,7 @@ void __init kmem_cache_init(void)
ARCH_KMALLOC_FLAGS|SLAB_PANIC,
NULL, NULL);
}
+#ifdef CONFIG_ZONE_DMA
if (CONFIG_ZONE_DMA_FLAG)
sizes->cs_dmacachep = kmem_cache_create(
names->name_dma,
@@ -1511,6 +1512,7 @@ void __init kmem_cache_init(void)
ARCH_KMALLOC_FLAGS|SLAB_CACHE_DMA|
SLAB_PANIC,
NULL, NULL);
+#endif
sizes++;
names++;
}
On Mon, 29 Jan 2007 12:41:33 -0800 (PST)
Christoph Lameter <[email protected]> wrote:
> +#ifdef CONFIG_ZONE_DMA
> if (CONFIG_ZONE_DMA_FLAG)
We don't need the `if (CONFIG_ZONE_DMA_FLAG)' any more, do we?
On Mon, 29 Jan 2007, Andrew Morton wrote:
> On Mon, 29 Jan 2007 12:41:33 -0800 (PST)
> Christoph Lameter <[email protected]> wrote:
>
> > +#ifdef CONFIG_ZONE_DMA
> > if (CONFIG_ZONE_DMA_FLAG)
>
> We don't need the `if (CONFIG_ZONE_DMA_FLAG)' any more, do we?
Correct.
On Mon, 29 Jan 2007, Christoph Lameter wrote:
> On Mon, 29 Jan 2007, Andrew Morton wrote:
>
> > On Mon, 29 Jan 2007 12:41:33 -0800 (PST)
> > Christoph Lameter <[email protected]> wrote:
> >
> > > +#ifdef CONFIG_ZONE_DMA
> > > if (CONFIG_ZONE_DMA_FLAG)
> >
> > We don't need the `if (CONFIG_ZONE_DMA_FLAG)' any more, do we?
>
> Correct.
>
Here is the modified patch:
Fix slab build failure if !CONFIG_ZONE_DMA
I also needed this to get 2.6.20-rc6-mm2 to build. Fixes the fix of the
complainer about the fixes.
Signed-off-by: Christoph Lameter <[email protected]>
Index: linux-2.6.20-rc6-mm2/mm/slab.c
===================================================================
--- linux-2.6.20-rc6-mm2.orig/mm/slab.c 2007-01-30 13:56:10.683074081 -0600
+++ linux-2.6.20-rc6-mm2/mm/slab.c 2007-01-30 13:57:13.152420752 -0600
@@ -1504,14 +1504,15 @@ void __init kmem_cache_init(void)
ARCH_KMALLOC_FLAGS|SLAB_PANIC,
NULL, NULL);
}
- if (CONFIG_ZONE_DMA_FLAG)
- sizes->cs_dmacachep = kmem_cache_create(
+#ifdef CONFIG_ZONE_DMA
+ 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++;
}