On Wed, 6 Jul 2005, Anton Blanchard wrote:
>
> > That is the issue. It should fall back to kmalloc. One piece of the
> > patchset dropped out between Andrew and Linus.
>
> That helped, but I still had lots of free memory in the request_queue
> slab. Giving kmem_cache_alloc_node the same treatm
> That is the issue. It should fall back to kmalloc. One piece of the
> patchset dropped out between Andrew and Linus.
That helped, but I still had lots of free memory in the request_queue
slab. Giving kmem_cache_alloc_node the same treatment fixed it, although
I wonder if we just shouldnt be p
On Wed, 6 Jul 2005, Anton Blanchard wrote:
> Its using the default. Looking at include/asm-generic/topology.h:
>
> #ifndef pcibus_to_node
> #define pcibus_to_node(node)(-1)
> #endif
>
> I wonder what kmalloc_node does when you pass it -1.
That is the issue. It should fall back to kmalloc. O
> What does pcibus_to_node return for the pcibus device that you are trying
> to allocate for?
Its using the default. Looking at include/asm-generic/topology.h:
#ifndef pcibus_to_node
#define pcibus_to_node(node)(-1)
#endif
I wonder what kmalloc_node does when you pass it -1.
Anton
Inde
On Wed, 6 Jul 2005, Anton Blanchard wrote:
> Its a shared processor partition (meaning the partition can get
> scheduled anywhere), so we throw all the cpus and memory into node 0.
>
> This machine only has 1 cpu (2 threads) and they are both in node0:
>
> # ls -l /sys/devices/system/node/node0/
> Ummm. But it has multiple memory nodes? We ran into trouble with the funky
> memory arrangement before.
Its a shared processor partition (meaning the partition can get
scheduled anywhere), so we throw all the cpus and memory into node 0.
This machine only has 1 cpu (2 threads) and they are bot
On Wed, 6 Jul 2005, Anton Blanchard wrote:
> FYI the kernel has NUMA enabled but the machine has only 1 NUMA node.
Ummm. But it has multiple memory nodes? We ran into trouble with the funky
memory arrangement before.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
th
Hi,
I just booted current -git on a POWER5 machine and quickly ran out of
memory. The machine has 256MB of memory, and looking at the slab stats
we are using almost all of it in unused slabs:
OBJS ACTIVE USE OBJ SIZE SLABS OBJ/SLAB CACHE SIZE NAME
367860 7202 1%0.12K 12262 30
8 matches
Mail list logo
