linux/mm
Christoph Lameter 4c93c355d5 SLUB: Place kmem_cache_cpu structures in a NUMA aware way
The kmem_cache_cpu structures introduced are currently an array placed in the
kmem_cache struct. Meaning the kmem_cache_cpu structures are overwhelmingly
on the wrong node for systems with a higher amount of nodes. These are
performance critical structures since the per node information has
to be touched for every alloc and free in a slab.

In order to place the kmem_cache_cpu structure optimally we put an array
of pointers to kmem_cache_cpu structs in kmem_cache (similar to SLAB).

However, the kmem_cache_cpu structures can now be allocated in a more
intelligent way.

We would like to put per cpu structures for the same cpu but different
slab caches in cachelines together to save space and decrease the cache
footprint. However, the slab allocators itself control only allocations
per node. We set up a simple per cpu array for every processor with
100 per cpu structures which is usually enough to get them all set up right.
If we run out then we fall back to kmalloc_node. This also solves the
bootstrap problem since we do not have to use slab allocator functions
early in boot to get memory for the small per cpu structures.

Pro:
	- NUMA aware placement improves memory performance
	- All global structures in struct kmem_cache become readonly
	- Dense packing of per cpu structures reduces cacheline
	  footprint in SMP and NUMA.
	- Potential avoidance of exclusive cacheline fetches
	  on the free and alloc hotpath since multiple kmem_cache_cpu
	  structures are in one cacheline. This is particularly important
	  for the kmalloc array.

Cons:
	- Additional reference to one read only cacheline (per cpu
	  array of pointers to kmem_cache_cpu) in both slab_alloc()
	  and slab_free().

[akinobu.mita@gmail.com: fix cpu hotplug offline/online path]
Signed-off-by: Christoph Lameter <clameter@sgi.com>
Cc: "Pekka Enberg" <penberg@cs.helsinki.fi>
Cc: Akinobu Mita <akinobu.mita@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-16 09:43:01 -07:00
..
Kconfig vmemmap: generify initialisation via helpers 2007-10-16 09:42:51 -07:00
Makefile Generic Virtual Memmap support for SPARSEMEM 2007-10-16 09:42:51 -07:00
allocpercpu.c Slab allocators: Replace explicit zeroing with __GFP_ZERO 2007-07-17 10:23:02 -07:00
backing-dev.c remove mm/backing-dev.c:congestion_wait_interruptible() 2007-07-16 09:05:52 -07:00
bootmem.c [PATCH] remove EXPORT_UNUSED_SYMBOL'ed symbols 2006-12-07 08:39:44 -08:00
bounce.c Drop 'size' argument from bio_endio and bi_end_io 2007-10-10 09:25:57 +02:00
fadvise.c [PATCH] mm: change uses of f_{dentry,vfsmnt} to use f_path 2006-12-08 08:28:43 -08:00
filemap.c fs: remove some AOP_TRUNCATED_PAGE 2007-10-16 09:42:58 -07:00
filemap_xip.c mm: write iovec cleanup 2007-10-16 09:42:54 -07:00
fremap.c fix VM_CAN_NONLINEAR check in sys_remap_file_pages 2007-10-08 12:58:14 -07:00
highmem.c Create the ZONE_MOVABLE zone 2007-07-17 10:22:59 -07:00
hugetlb.c flush icache before set_pte() on ia64: flush icache at set_pte 2007-10-16 09:42:59 -07:00
internal.h Make page->private usable in compound pages 2007-05-07 12:12:53 -07:00
madvise.c speed up madvise_need_mmap_write() usage 2007-07-16 09:05:36 -07:00
memory.c flush icache before set_pte() on ia64: flush icache at set_pte 2007-10-16 09:42:59 -07:00
memory_hotplug.c Memoryless nodes: introduce mask of nodes with memory 2007-10-16 09:42:58 -07:00
mempolicy.c memoryless nodes: fixup uses of node_online_map in generic code 2007-10-16 09:42:59 -07:00
mempool.c Slab allocators: Replace explicit zeroing with __GFP_ZERO 2007-07-17 10:23:02 -07:00
migrate.c flush icache before set_pte() on ia64: flush icache at set_pte 2007-10-16 09:42:59 -07:00
mincore.c [PATCH] mincore: vma crossing fix 2007-02-15 09:57:03 -08:00
mlock.c do not limit locked memory when RLIMIT_MEMLOCK is RLIM_INFINITY 2007-07-16 09:05:37 -07:00
mmap.c fix NULL pointer dereference in __vm_enough_memory() 2007-08-22 19:52:45 -07:00
mmzone.c [PATCH] remove EXPORT_UNUSED_SYMBOL'ed symbols 2006-12-07 08:39:44 -08:00
mprotect.c flush icache before set_pte() on ia64: flush icache at set_pte 2007-10-16 09:42:59 -07:00
mremap.c mm: variable length argument support 2007-07-19 10:04:45 -07:00
msync.c Detach sched.h from mm.h 2007-05-21 09:18:19 -07:00
nommu.c fix NULL pointer dereference in __vm_enough_memory() 2007-08-22 19:52:45 -07:00
oom_kill.c Memoryless nodes: OOM: use N_HIGH_MEMORY map instead of constructing one on the fly 2007-10-16 09:42:58 -07:00
page-writeback.c memoryless nodes: fixup uses of node_online_map in generic code 2007-10-16 09:42:59 -07:00
page_alloc.c mm/page_alloc.c: make code static 2007-10-16 09:43:01 -07:00
page_io.c Drop 'size' argument from bio_endio and bi_end_io 2007-10-10 09:25:57 +02:00
pdflush.c Freezer: make kernel threads nonfreezable by default 2007-07-17 10:23:02 -07:00
prio_tree.c Linux-2.6.12-rc2 2005-04-16 15:20:36 -07:00
quicklist.c Quicklists for page table pages 2007-05-07 12:12:54 -07:00
readahead.c mm: buffered write cleanup 2007-10-16 09:42:54 -07:00
rmap.c flush icache before set_pte() on ia64: flush icache at set_pte 2007-10-16 09:42:59 -07:00
shmem.c Group short-lived and reclaimable kernel allocations 2007-10-16 09:43:00 -07:00
shmem_acl.c [PATCH] Fix typos in mm/shmem_acl.c 2006-10-11 11:14:23 -07:00
slab.c Group short-lived and reclaimable kernel allocations 2007-10-16 09:43:00 -07:00
slob.c Slab allocators: fail if ksize is called with a NULL parameter 2007-10-16 09:42:53 -07:00
slub.c SLUB: Place kmem_cache_cpu structures in a NUMA aware way 2007-10-16 09:43:01 -07:00
sparse-vmemmap.c vmemmap: generify initialisation via helpers 2007-10-16 09:42:51 -07:00
sparse.c Fix corruption of memmap on IA64 SPARSEMEM when mem_section is not a power of 2 2007-10-16 09:43:00 -07:00
swap.c mm: use pagevec to rotate reclaimable page 2007-10-16 09:42:54 -07:00
swap_state.c mm: clarify __add_to_swap_cache locking 2007-10-16 09:42:53 -07:00
swapfile.c Replace CONFIG_SOFTWARE_SUSPEND with CONFIG_HIBERNATION 2007-07-29 16:45:38 -07:00
thrash.c Bug in mm/thrash.c function grab_swap_token() 2007-05-11 08:29:32 -07:00
tiny-shmem.c [PATCH] mm/{,tiny-}shmem.c cleanups 2007-03-01 14:53:35 -08:00
truncate.c mm: merge populate and nopage into fault (fixes nonlinear) 2007-07-19 10:04:41 -07:00
util.c Slab allocators: fail if ksize is called with a NULL parameter 2007-10-16 09:42:53 -07:00
vmalloc.c Categorize GFP flags 2007-10-16 09:42:59 -07:00
vmscan.c make swappiness safer to use 2007-10-16 09:42:59 -07:00
vmstat.c Print out statistics in relation to fragmentation avoidance to /proc/pagetypeinfo 2007-10-16 09:43:00 -07:00