The current code groups up to 16 nodes in a level and then puts an
ALLNODES domain spanning the entire tree on top of that. This doesn't
reflect the numa topology and esp for the smaller not-fully-connected
machines out there today this might make a difference.
Therefore, build a proper numa topology based on node_distance().
Since there's no fixed numa layers anymore, the static SD_NODE_INIT
and SD_ALLNODES_INIT aren't usable anymore, the new code tries to
construct something similar and scales some values either on the
number of cpus in the domain and/or the node_distance() ratio.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Anton Blanchard <anton@samba.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Chris Metcalf <cmetcalf@tilera.com>
Cc: David Howells <dhowells@redhat.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
Cc: linux-alpha@vger.kernel.org
Cc: linux-ia64@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Cc: linux-mips@linux-mips.org
Cc: linuxppc-dev@lists.ozlabs.org
Cc: linux-sh@vger.kernel.org
Cc: Matt Turner <mattst88@gmail.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Paul Mundt <lethal@linux-sh.org>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Richard Henderson <rth@twiddle.net>
Cc: sparclinux@vger.kernel.org
Cc: Tony Luck <tony.luck@intel.com>
Cc: x86@kernel.org
Cc: Dimitri Sivanich <sivanich@sgi.com>
Cc: Greg Pearson <greg.pearson@hp.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: bob.picco@oracle.com
Cc: chris.mason@oracle.com
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Link: http://lkml.kernel.org/n/tip-r74n3n8hhuc2ynbrnp3vt954@git.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
This moves the 'cpu sysdev_class' over to a regular 'cpu' subsystem
and converts the devices to regular devices. The sysdev drivers are
implemented as subsystem interfaces now.
After all sysdev classes are ported to regular driver core entities, the
sysdev implementation will be entirely removed from the kernel.
Userspace relies on events and generic sysfs subsystem infrastructure
from sysdev devices, which are made available with this conversion.
Cc: Haavard Skinnemoen <hskinnemoen@gmail.com>
Cc: Hans-Christian Egtvedt <egtvedt@samfundet.no>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Paul Mundt <lethal@linux-sh.org>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Chris Metcalf <cmetcalf@tilera.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Borislav Petkov <bp@amd64.org>
Cc: Tigran Aivazian <tigran@aivazian.fsnet.co.uk>
Cc: Len Brown <lenb@kernel.org>
Cc: Zhang Rui <rui.zhang@intel.com>
Cc: Dave Jones <davej@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Russell King <rmk+kernel@arm.linux.org.uk>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Arjan van de Ven <arjan@linux.intel.com>
Cc: "Rafael J. Wysocki" <rjw@sisk.pl>
Cc: "Srivatsa S. Bhat" <srivatsa.bhat@linux.vnet.ibm.com>
Signed-off-by: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
We have two identical definitions of RECLAIM_DISTANCE, looks like
the patch got applied twice. Remove one.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
On big POWER7 boxes we see large amounts of CPU time in system
processes like workqueue and watchdog kernel threads.
We currently rebalance the entire machine each time a task goes
idle and this is very expensive on large machines. Disable newidle
balancing at the node level and rely on the scheduler tick to
rebalance across nodes.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
The largest POWER7 boxes have 32 nodes. SD_NODES_PER_DOMAIN groups
nodes into chunks of 16 and adds a global balancing domain
(SD_ALLNODES) above it.
If we bump SD_NODES_PER_DOMAIN to 32, then we avoid this extra
level of balancing on our largest boxes.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
When chasing a performance issue on ppc64, I noticed tasks
communicating via a pipe would often end up on different nodes.
It turns out SD_WAKE_AFFINE is not set in our node defition. Commit
9fcd18c9e6 (sched: re-tune balancing) enabled SD_WAKE_AFFINE
in the node definition for x86 and we need a similar change for
ppc64.
I used lmbench lat_ctx and perf bench pipe to verify this fix. Each
benchmark was run 10 times and the average taken.
lmbench lat_ctx:
before: 66565 ops/sec
after: 204700 ops/sec
3.1x faster
perf bench pipe:
before: 5.6570 usecs
after: 1.3470 usecs
4.2x faster
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
The header asm/hvcall.h was previously included indirectly via
smp.h. On non-SMP systems, however, these declarations are excluded
and the build breaks. This is easily fixed by including asm/hvcall.h
directly.
The VPHN feature is only meaningful on NUMA systems that implement
the SPLPAR option, so exclude the VPHN code on systems without
SPLPAR enabled.
Also, expose unmap_cpu_from_node() on systems with SPLPAR enabled,
even if CONFIG_HOTPLUG_CPU is disabled.
Lastly, map_cpu_to_node() is now needed by VPHN to manipulate the
node masks after boot time, so remove the __cpuinit annotation to
fix a section mismatch.
Signed-off-by: Jesse Larrew <jlarrew@linux.vnet.ibm.com>
Tie the polling mechanism into the ibm,suspend-me rtas call to
stop/restart polling before/after a suspend, hibernate, migrate,
or checkpoint restart operation. This ensures that the system has a
chance to disable the polling if the partition is migrated to a system
that does not support VPHN (and vice versa).
Signed-off-by: Jesse Larrew <jlarrew@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
* 'next-devicetree' of git://git.secretlab.ca/git/linux-2.6: (63 commits)
of/platform: Register of_platform_drivers with an "of:" prefix
of/address: Clean up function declarations
of/spi: call of_register_spi_devices() from spi core code
of: Provide default of_node_to_nid() implementation.
of/device: Make of_device_make_bus_id() usable by other code.
of/irq: Fix endian issues in parsing interrupt specifiers
of: Fix phandle endian issues
of/flattree: fix of_flat_dt_is_compatible() to match the full compatible string
of: remove of_default_bus_ids
of: make of_find_device_by_node generic
microblaze: remove references to of_device and to_of_device
sparc: remove references to of_device and to_of_device
powerpc: remove references to of_device and to_of_device
of/device: Replace of_device with platform_device in includes and core code
of/device: Protect against binding of_platform_drivers to non-OF devices
of: remove asm/of_device.h
of: remove asm/of_platform.h
of/platform: remove all of_bus_type and of_platform_bus_type references
of: Merge of_platform_bus_type with platform_bus_type
drivercore/of: Add OF style matching to platform bus
...
Fix up trivial conflicts in arch/microblaze/kernel/Makefile due to just
some obj-y removals by the devicetree branch, while the microblaze
updates added a new file.
of_node_to_nid() is only relevant in a few architectures. Don't force
everyone to implement it anyway.
Signed-off-by: Grant Likely <grant.likely@secretlab.ca>
Form 1 affinity allows multiple entries in ibm,associativity-reference-points
which represent affinity domains in decreasing order of importance. The
Linux concept of a node is always the first entry, but using the other
values as an input to node_distance() allows the memory allocator to make
better decisions on which node to go first when local memory has been
exhausted.
We keep things simple and create an array indexed by NUMA node, capped at
4 entries. Each time we lookup an associativity property we initialise
the array which is overkill, but since we should only hit this path during
boot it didn't seem worth adding a per node valid bit.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
I noticed /proc/sys/vm/zone_reclaim_mode was 0 on a ppc64 NUMA box. It gets
enabled via this:
/*
* If another node is sufficiently far away then it is better
* to reclaim pages in a zone before going off node.
*/
if (distance > RECLAIM_DISTANCE)
zone_reclaim_mode = 1;
Since we use the default value of 20 for REMOTE_DISTANCE and 20 for
RECLAIM_DISTANCE it never kicks in.
The local to remote bandwidth ratios can be quite large on System p
machines so it makes sense for us to reclaim clean pagecache locally before
going off node.
The patch below sets a smaller value for RECLAIM_DISTANCE and thus enables
zone reclaim.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Convert NUMA code to new cpumask API. We shift the node to cpumask
setup code until after we complete bootmem allocation so we can
dynamically allocate the cpumasks.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Dynamically allocate cpu_sibling_map and cpu_core_map cpumasks.
We don't need to set_cpu_online() the boot cpu in smp_prepare_boot_cpu,
init/main.c does it for us.
We also postpone setting of the boot cpu in cpu_sibling_map and cpu_core_map
until when the memory allocator is available (smp_prepare_cpus), similar
to x86.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
I noticed /proc/sys/vm/zone_reclaim_mode was 0 on a ppc64 NUMA box. It gets
enabled via this:
/*
* If another node is sufficiently far away then it is better
* to reclaim pages in a zone before going off node.
*/
if (distance > RECLAIM_DISTANCE)
zone_reclaim_mode = 1;
Since we use the default value of 20 for REMOTE_DISTANCE and 20 for
RECLAIM_DISTANCE it never kicks in.
The local to remote bandwidth ratios can be quite large on System p
machines so it makes sense for us to reclaim clean pagecache locally before
going off node.
The patch below sets a smaller value for RECLAIM_DISTANCE and thus enables
zone reclaim.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Clean up SD_NODE_INITS so we can easily compare it to x86. Similar to the
work in 47734f89be (sched: Clean up topology.h)
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
pcibus_to_node can return -1 if we cannot determine which node a pci bus
is on. If passed -1, cpumask_of_node will negatively index the lookup array
and pull in random data:
# cat /sys/devices/pci0000:00/0000:00:01.0/local_cpus
00000000,00000003,00000000,00000000
# cat /sys/devices/pci0000:00/0000:00:01.0/local_cpulist
64-65
Change cpumask_of_node to check for -1 and return cpu_all_mask in this
case:
# cat /sys/devices/pci0000:00/0000:00:01.0/local_cpus
ffffffff,ffffffff,ffffffff,ffffffff
# cat /sys/devices/pci0000:00/0000:00:01.0/local_cpulist
0-127
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Sysbench thinks SD_BALANCE_WAKE is too agressive and kbuild doesn't
really mind too much, SD_BALANCE_NEWIDLE picks up most of the
slack.
On a dual socket, quad core, dual thread nehalem system:
sysbench (--num_threads=16):
SD_BALANCE_WAKE-: 13982 tx/s
SD_BALANCE_WAKE+: 15688 tx/s
kbuild (-j16):
SD_BALANCE_WAKE-: 47.648295846 seconds time elapsed ( +- 0.312% )
SD_BALANCE_WAKE+: 47.608607360 seconds time elapsed ( +- 0.026% )
(same within noise)
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Make the idle balancer more agressive, to improve a
x264 encoding workload provided by Jason Garrett-Glaser:
NEXT_BUDDY NO_LB_BIAS
encoded 600 frames, 252.82 fps, 22096.60 kb/s
encoded 600 frames, 250.69 fps, 22096.60 kb/s
encoded 600 frames, 245.76 fps, 22096.60 kb/s
NO_NEXT_BUDDY LB_BIAS
encoded 600 frames, 344.44 fps, 22096.60 kb/s
encoded 600 frames, 346.66 fps, 22096.60 kb/s
encoded 600 frames, 352.59 fps, 22096.60 kb/s
NO_NEXT_BUDDY NO_LB_BIAS
encoded 600 frames, 425.75 fps, 22096.60 kb/s
encoded 600 frames, 425.45 fps, 22096.60 kb/s
encoded 600 frames, 422.49 fps, 22096.60 kb/s
Peter pointed out that this is better done via newidle_idx,
not via LB_BIAS, newidle balancing should look for where
there is load _now_, not where there was load 2 ticks ago.
Worst-case latencies are improved as well as no buddies
means less vruntime spread. (as per prior lkml discussions)
This change improves kbuild-peak parallelism as well.
Reported-by: Jason Garrett-Glaser <darkshikari@gmail.com>
Signed-off-by: Mike Galbraith <efault@gmx.de>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <1253011667.9128.16.camel@marge.simson.net>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
When merging select_task_rq_fair() and sched_balance_self() we lost
the use of wake_idx, restore that and set them to 0 to make wake
balancing more aggressive.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The problem with wake_idle() is that is doesn't respect things like
cpu_power, which means it doesn't deal well with SMT nor the recent
RT interaction.
To cure this, it needs to do what sched_balance_self() does, which
leads to the possibility of merging select_task_rq_fair() and
sched_balance_self().
Modify sched_balance_self() to:
- update_shares() when walking up the domain tree,
(it only called it for the top domain, but it should
have done this anyway), which allows us to remove
this ugly bit from try_to_wake_up().
- do wake_affine() on the smallest domain that contains
both this (the waking) and the prev (the wakee) cpu for
WAKE invocations.
Then use the top-down balance steps it had to replace wake_idle().
This leads to the dissapearance of SD_WAKE_BALANCE and
SD_WAKE_IDLE_FAR, with SD_WAKE_IDLE replaced with SD_BALANCE_WAKE.
SD_WAKE_AFFINE needs SD_BALANCE_WAKE to be effective.
Touch all topology bits to replace the old with new SD flags --
platforms might need re-tuning, enabling SD_BALANCE_WAKE
conditionally on a NUMA distance seems like a good additional
feature, magny-core and small nehalem systems would want this
enabled, systems with slow interconnects would not.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Everyone defines it, and only one person uses it
(arch/mips/sgi-ip27/ip27-nmi.c). So just open code it there.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Cc: linux-mips@linux-mips.org
Impact: New API
The old topology_core_siblings() and topology_thread_siblings() return
a cpumask_t; these new ones return a (const) struct cpumask *.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Mike Travis <travis@sgi.com>
Impact: New APIs
The old node_to_cpumask/node_to_pcibus returned a cpumask_t: these
return a pointer to a struct cpumask. Part of removing cpumasks from
the stack.
(Also replaces powerpc internal uses of node_to_cpumask).
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Mathieu Desnoyers reported this build failure on powerpc:
kernel/sched.c: In function 'sd_init_NODE':
kernel/sched.c:7319: error: non-static initialization of a flexible array member
kernel/sched.c:7319: error: (near initialization for '(anonymous)')
this happens because .span changed to cpumask_var_t, hence
the static CPU_MASK_NONE initializers in the SD_*_INIT
templates are not type-correct anymore.
Remove them, as they default to empty anyway.
Also remove them from IA64, MIPS and SH.
Reported-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
from include/asm-powerpc. This is the result of a
mkdir arch/powerpc/include/asm
git mv include/asm-powerpc/* arch/powerpc/include/asm
Followed by a few documentation/comment fixups and a couple of places
where <asm-powepc/...> was being used explicitly. Of the latter only
one was outside the arch code and it is a driver only built for powerpc.
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Peter Zijlstra