linux-toradex.git/kernel/sched/core.c, branch v3.5-rc3

sched: Fix the relax_domain_level boot parameter

2012-06-06T15:07:41+00:00

It does not get processed because sched_domain_level_max is 0 at the
time that setup_relax_domain_level() is run.

Simply accept the value as it is, as we don't know the value of
sched_domain_level_max until sched domain construction is completed.

Fix sched_relax_domain_level in cpuset.  The build_sched_domain() routine calls
the set_domain_attribute() routine prior to setting the sd->level, however,
the set_domain_attribute() routine relies on the sd->level to decide whether
idle load balancing will be off/on.

Signed-off-by: Dimitri Sivanich 
Signed-off-by: Peter Zijlstra 
Link: http://lkml.kernel.org/r/20120605184436.GA15668@sgi.com
Signed-off-by: Ingo Molnar

sched: Validate assumptions in sched_init_numa()

2012-06-06T14:52:30+00:00

Add some code to validate assumptions we're making and output
warnings if they are not.

If this trigger we want to know about it.

Signed-off-by: Peter Zijlstra 
Cc: Alex Shi 
Link: http://lkml.kernel.org/n/tip-6uc3wk5s9udxtdl9cnku0vtt@git.kernel.org
Signed-off-by: Ingo Molnar

sched: Always initialize cpu-power

2012-06-06T14:52:27+00:00

Often when we run into mis-shapen topologies the balance iteration
fails to update the cpu power properly and we'll end up in /0 traps.

Always initialize the cpu-power to a semi-sane value so that we can
at least boot the machine, even if the load-balancer might not
function correctly.

Signed-off-by: Peter Zijlstra 
Link: http://lkml.kernel.org/n/tip-3lbhyj25sr169ha7z3qht5na@git.kernel.org
Signed-off-by: Ingo Molnar

sched: Fix domain iteration

2012-06-06T14:52:26+00:00

Weird topologies can lead to asymmetric domain setups. This needs
further consideration since these setups are typically non-minimal
too.

For now, make it work by adding an extra mask selecting which CPUs
are allowed to iterate up.

The topology that triggered it is the one from David Rientjes:

	10 20 20 30
	20 10 20 20
	20 20 10 20
	30 20 20 10

resulting in boxes that wouldn't even boot.

Reported-by: David Rientjes 
Signed-off-by: Peter Zijlstra 
Link: http://lkml.kernel.org/n/tip-3p86l9cuaqnxz7uxsojmz5rm@git.kernel.org
Signed-off-by: Ingo Molnar

sched/numa: Load balance between remote nodes

2012-06-06T14:52:25+00:00

Commit cb83b629b ("sched/numa: Rewrite the CONFIG_NUMA sched
domain support") removed the NODE sched domain and started checking
if the node distance in SLIT table is farther than REMOTE_DISTANCE,
if so, it will lose the load balance chance at exec/fork/wake_affine
points.

But actually, even the node distance is farther than REMOTE_DISTANCE.

Modern CPUs also has QPI like connections, which ensures that memory
access is not too slow between nodes. So the above change in behavior
on NUMA machine causes a performance regression on various benchmarks:
hackbench, tbench, netperf, oltp, etc.

This patch will recover the scheduler behavior to old mode on all my
Intel platforms: NHM EP/EX, WSM EP, SNB EP/EP4S, and thus fixes the
perfromance regressions. (all of them just have 2 kinds distance, 10, 21)

Signed-off-by: Alex Shi 
Signed-off-by: Peter Zijlstra 
Link: http://lkml.kernel.org/r/1338965571-9812-1-git-send-email-alex.shi@intel.com
Signed-off-by: Ingo Molnar

sched: Remove NULL assignment of dattr_cur

2012-05-30T12:02:27+00:00

Remove explicit NULL assignment of static pointer
dattr_cur from init_sched_domains().

Signed-off-by: Kamalesh Babulal 
Signed-off-by: Peter Zijlstra 
Link: http://lkml.kernel.org/r/20120523091411.GG5005@linux.vnet.ibm.com
Signed-off-by: Ingo Molnar

sched: Remove the last NULL entry from sched_feat_names

2012-05-30T12:02:27+00:00

No need to have the last NULL entry.

Signed-off-by: Hiroshi Shimamoto 
Signed-off-by: Peter Zijlstra 
Link: http://lkml.kernel.org/r/4FBF29E7.5020805@ct.jp.nec.com
Signed-off-by: Ingo Molnar

sched: Make sched_feat_names const

2012-05-30T12:02:26+00:00

The strings sched_feat_names are never changed.

Signed-off-by: Hiroshi Shimamoto 
Signed-off-by: Peter Zijlstra 
Link: http://lkml.kernel.org/r/4FBF29B2.9030904@ct.jp.nec.com
Signed-off-by: Ingo Molnar

sched: Move nr_cpus_allowed out of 'struct sched_rt_entity'

2012-05-30T12:02:25+00:00

Since nr_cpus_allowed is used outside of sched/rt.c and wants to be
used outside of there more, move it to a more natural site.

Signed-off-by: Peter Zijlstra 
Link: http://lkml.kernel.org/n/tip-kr61f02y9brwzkh6x53pdptm@git.kernel.org
Signed-off-by: Ingo Molnar

sched: Fix SD_OVERLAP

2012-05-30T12:02:24+00:00

SD_OVERLAP exists to allow overlapping groups, overlapping groups
appear in NUMA topologies that aren't fully connected.

The typical result of not fully connected NUMA is that each cpu (or
rather node) will have different spans for a particular distance.
However due to how sched domains are traversed -- only the first cpu
in the mask goes one level up -- the next level only cares about the
spans of the cpus that went up.

Due to this two things were observed to be broken:

 - build_overlap_sched_groups() -- since its possible the cpu we're
   building the groups for exists in multiple (or all) groups, the
   selection criteria of the first group didn't ensure there was a cpu
   for which is was true that cpumask_first(span) == cpu. Thus load-
   balancing would terminate.

 - update_group_power() -- assumed that the cpu span of the first
   group of the domain was covered by all groups of the child domain.
   The above explains why this isn't true, so deal with it.

Signed-off-by: Peter Zijlstra 
Cc: David Rientjes 
Link: http://lkml.kernel.org/r/1337788843.9783.14.camel@laptop
Signed-off-by: Ingo Molnar