freebsd-dev/sys/vm
John Baldwin a3870a1826 Very rough first cut at NUMA support for the physical page allocator. For
now it uses a very dumb first-touch allocation policy.  This will change in
the future.
- Each architecture indicates the maximum number of supported memory domains
  via a new VM_NDOMAIN parameter in <machine/vmparam.h>.
- Each cpu now has a PCPU_GET(domain) member to indicate the memory domain
  a CPU belongs to.  Domain values are dense and numbered from 0.
- When a platform supports multiple domains, the default freelist
  (VM_FREELIST_DEFAULT) is split up into N freelists, one for each domain.
  The MD code is required to populate an array of mem_affinity structures.
  Each entry in the array defines a range of memory (start and end) and a
  domain for the range.  Multiple entries may be present for a single
  domain.  The list is terminated by an entry where all fields are zero.
  This array of structures is used to split up phys_avail[] regions that
  fall in VM_FREELIST_DEFAULT into per-domain freelists.
- Each memory domain has a separate lookup-array of freelists that is
  used when fulfulling a physical memory allocation.  Right now the
  per-domain freelists are listed in a round-robin order for each domain.
  In the future a table such as the ACPI SLIT table may be used to order
  the per-domain lookup lists based on the penalty for each memory domain
  relative to a specific domain.  The lookup lists may be examined via a
  new vm.phys.lookup_lists sysctl.
- The first-touch policy is implemented by using PCPU_GET(domain) to
  pick a lookup list when allocating memory.

Reviewed by:	alc
2010-07-27 20:33:50 +00:00
..
default_pager.c Implement global and per-uid accounting of the anonymous memory. Add 2009-06-23 20:45:22 +00:00
device_pager.c Eliminate page queues locking around most calls to vm_page_free(). 2010-05-06 18:58:32 +00:00
memguard.c Start copyright notice with /*- 2010-04-07 16:29:10 +00:00
memguard.h Start copyright notice with /*- 2010-04-07 16:29:10 +00:00
phys_pager.c Handle busy status of the page in a way expected for pager_getpage(). 2010-05-03 19:19:58 +00:00
pmap.h Roughly half of a typical pmap_mincore() implementation is machine- 2010-05-24 14:26:57 +00:00
redzone.c Fix REDZONE(9) on amd64 and perhaps other 64 bit targets -- ensure the space 2008-08-13 17:32:48 +00:00
redzone.h
sg_pager.c Eliminate page queues locking around most calls to vm_page_free(). 2010-05-06 18:58:32 +00:00
swap_pager.c Call vm_page_deactivate() rather than vm_page_dontneed() in 2010-05-09 16:27:42 +00:00
swap_pager.h - Move 'struct swdevt' back into swap_pager.h and expose it to userland. 2007-02-07 17:43:11 +00:00
uma_core.c Add a new column to the output of vmstat -z to indicate the number 2010-06-15 19:28:37 +00:00
uma_dbg.c - Make the keg abstraction more complete. Permit a zone to have multiple 2009-01-25 09:11:24 +00:00
uma_dbg.h
uma_int.h Add a new column to the output of vmstat -z to indicate the number 2010-06-15 19:28:37 +00:00
uma.h Add a new column to the output of vmstat -z to indicate the number 2010-06-15 19:28:37 +00:00
vm_contig.c With the demise of page coloring, the page queue macros no longer serve any 2010-07-02 15:02:51 +00:00
vm_extern.h o) Add a VM find-space option, VMFS_TLB_ALIGNED_SPACE, which searches the 2010-04-18 22:32:07 +00:00
vm_fault.c Use vm_page_prev() instead of vm_page_lookup() in the implementation of 2010-07-02 19:59:18 +00:00
vm_glue.c Eliminate page queues locking around most calls to vm_page_free(). 2010-05-06 18:58:32 +00:00
vm_init.c Introduce exec_alloc_args(). The objective being to encapsulate the 2010-07-27 17:31:03 +00:00
vm_kern.c The pages allocated by kmem_alloc_attr() and kmem_malloc() are unmanaged. 2010-05-03 07:08:16 +00:00
vm_kern.h Enable the creation of a kmem map larger than 4GB. 2008-07-05 19:34:33 +00:00
vm_map.c Introduce a helper function vm_page_find_least(). Use it in several places, 2010-07-04 11:13:33 +00:00
vm_map.h o) Add a VM find-space option, VMFS_TLB_ALIGNED_SPACE, which searches the 2010-04-18 22:32:07 +00:00
vm_meter.c Add a new type of VM object: OBJT_SG. An OBJT_SG object is very similar to 2009-07-24 13:50:29 +00:00
vm_mmap.c Fix commented out resource limit check in mlockall(2). It's still racy, 2010-07-27 19:26:18 +00:00
vm_object.c Several cleanups for the r209686: 2010-07-04 19:02:32 +00:00
vm_object.h VI_OBJDIRTY vnode flag mirrors the state of OBJ_MIGHTBEDIRTY vm object 2009-12-21 12:29:38 +00:00
vm_page.c Redo the page table page allocation on MIPS, as suggested by 2010-07-21 09:27:00 +00:00
vm_page.h Redo the page table page allocation on MIPS, as suggested by 2010-07-21 09:27:00 +00:00
vm_pageout.c Push down the acquisition of the page queues lock into 2010-07-02 20:56:22 +00:00
vm_pageout.h Make vm_contig_grow_cache() extern, and use it when vm_phys_alloc_contig() 2010-06-04 06:35:36 +00:00
vm_pager.c Add a new type of VM object: OBJT_SG. An OBJT_SG object is very similar to 2009-07-24 13:50:29 +00:00
vm_pager.h Reimplement vm_object_page_clean(), using the fact that vm object memq 2010-07-04 11:26:56 +00:00
vm_param.h On Alan's advice, rather than do a wholesale conversion on a single 2010-04-30 00:46:43 +00:00
vm_phys.c Very rough first cut at NUMA support for the physical page allocator. For 2010-07-27 20:33:50 +00:00
vm_phys.h Very rough first cut at NUMA support for the physical page allocator. For 2010-07-27 20:33:50 +00:00
vm_reserv.c Previously, when vm_page_free_toq() was performed on a page belonging to 2009-04-11 09:09:00 +00:00
vm_reserv.h Introduce vm_reserv_reclaim_contig(). This function is used by 2008-04-06 18:09:28 +00:00
vm_unix.c Remove execute permission from the memory allocated by sbrk(). 2009-04-11 22:34:08 +00:00
vm_zeroidle.c Split P_NOLOAD into a per-thread flag (TDF_NOLOAD). 2009-11-03 16:46:52 +00:00
vm.h Replace VM_PROT_OVERRIDE_WRITE by VM_PROT_COPY. VM_PROT_OVERRIDE_WRITE has 2009-11-26 05:16:07 +00:00
vnode_pager.c Minimize the use of the page queues lock for synchronizing access to the 2010-06-02 15:46:37 +00:00
vnode_pager.h Do not sleep for vnode lock while holding map lock in vm_fault. Try to 2009-02-08 20:23:46 +00:00