linux/kernel/cpuset.c, branch v2.6.26

cpusets, hotplug, scheduler: fix scheduler domain breakage

2008-07-13T09:37:02Z

Commit f18f982ab ("sched: CPU hotplug events must not destroy scheduler domains created by the cpusets") introduced a hotplug-related problem as described below: Upon CPU_DOWN_PREPARE, update_sched_domains() -> detach_destroy_domains(&cpu_online_map) does the following: /* * Force a reinitialization of the sched domains hierarchy. The domains * and groups cannot be updated in place without racing with the balancing * code, so we temporarily attach all running cpus to the NULL domain * which will prevent rebalancing while the sched domains are recalculated. */ The sched-domains should be rebuilt when a CPU_DOWN ops. has been completed, effectively either upon CPU_DEAD{_FROZEN} (upon success) or CPU_DOWN_FAILED{_FROZEN} (upon failure -- restore the things to their initial state). That's what update_sched_domains() also does but only for !CPUSETS case. With f18f982ab, sched-domains' reinitialization is delegated to CPUSETS code: cpuset_handle_cpuhp() -> common_cpu_mem_hotplug_unplug() -> rebuild_sched_domains() Being called for CPU_UP_PREPARE and if its callback is called after update_sched_domains()), it just negates all the work done by update_sched_domains() -- i.e. a soon-to-be-offline cpu is included in the sched-domains and that makes it visible for the load-balancer while the CPU_DOWN ops. is in progress. __migrate_live_tasks() moves the tasks off a 'dead' cpu (it's already "offline" when this function is called). try_to_wake_up() is called for one of these tasks from another CPU -> the load-balancer (wake_idle()) picks up a "dead" CPU and places the task on it. Then e.g. BUG_ON(rq->nr_running) detects this a bit later -> oops. Signed-off-by: Dmitry Adamushko Tested-by: Vegard Nossum Cc: Paul Menage Cc: Max Krasnyansky Cc: Paul Jackson Cc: Peter Zijlstra Cc: miaox@cn.fujitsu.com Cc: rostedt@goodmis.org Cc: Linus Torvalds Signed-off-by: Ingo Molnar

Merge branch 'core-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip

2008-06-20T19:37:13Z

* 'core-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip: softlockup: fix NMI hangs due to lock race - 2.6.26-rc regression rcupreempt: remove export of rcu_batches_completed_bh cpuset: limit the input of cpuset.sched_relax_domain_level

cpuset: limit the input of cpuset.sched_relax_domain_level

2008-06-19T07:45:36Z

We allow the inputs to be [-1 ... SD_LV_MAX), and return -EINVAL for inputs outside this range. Signed-off-by: Li Zefan Acked-by: Paul Menage Acked-by: Paul Jackson Acked-by: Hidetoshi Seto Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner

sched: CPU hotplug events must not destroy scheduler domains created by the cpusets

2008-06-19T07:14:51Z

First issue is not related to the cpusets. We're simply leaking doms_cur. It's allocated in arch_init_sched_domains() which is called for every hotplug event. So we just keep reallocation doms_cur without freeing it. I introduced free_sched_domains() function that cleans things up. Second issue is that sched domains created by the cpusets are completely destroyed by the CPU hotplug events. For all CPU hotplug events scheduler attaches all CPUs to the NULL domain and then puts them all into the single domain thereby destroying domains created by the cpusets (partition_sched_domains). The solution is simple, when cpusets are enabled scheduler should not create default domain and instead let cpusets do that. Which is exactly what the patch does. Signed-off-by: Max Krasnyansky Cc: pj@sgi.com Cc: menage@google.com Cc: rostedt@goodmis.org Acked-by: Peter Zijlstra Signed-off-by: Thomas Gleixner

cpusets: fix bug when adding nonexistent cpu or mem

2008-06-06T18:29:11Z

Adding a nonexistent cpu to a cpuset will be omitted quietly. It should return -EINVAL. Example: (real_nr_cpus <= 4 < NR_CPUS or cpu#4 was just offline) # cat cpus 0-1 # /bin/echo 4 > cpus # /bin/echo $? 0 # cat cpus # The same occurs when add a nonexistent mem. This patch will fix this bug. And when *buf == "", the check is unneeded. Signed-off-by: Lai Jiangshan Acked-by: Paul Jackson Cc: David Rientjes Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

Fix cpuset sched_relax_domain_level control file

2008-05-08T17:46:56Z

Due to a merge conflict, the sched_relax_domain_level control file was marked as being handled by cpuset_read/write_u64, but the code to handle it was actually in cpuset_common_file_read/write. Since the value being written/read is in fact a signed integer, it should be treated as such; this patch adds cpuset_read/write_s64 functions, and uses them to handle the sched_relax_domain_level file. With this patch, the sched_relax_domain_level can be read and written, and the correct contents seen/updated. Signed-off-by: Paul Menage Cc: Hidetoshi Seto Cc: Paul Jackson Cc: Ingo Molnar Reviewed-by: Li Zefan Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

Cpuset hardwall flag: add a mem_hardwall flag to cpusets

2008-04-29T15:06:11Z

This flag provides the hardwalling properties of mem_exclusive, without enforcing the exclusivity. Either mem_hardwall or mem_exclusive is sufficient to prevent GFP_KERNEL allocations from passing outside the cpuset's assigned nodes. Signed-off-by: Paul Menage Acked-by: Paul Jackson Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

Cpuset hardwall flag: switch cpusets to use the bulk cgroup_add_files() API

2008-04-29T15:06:11Z

Currently the cpusets mem_exclusive flag is overloaded to mean both "no-overlapping" and "no GFP_KERNEL allocations outside this cpuset". These patches add a new mem_hardwall flag with just the allocation restriction part of the mem_exclusive semantics, without breaking backwards-compatibility for those who continue to use just mem_exclusive. Additionally, the cgroup control file registration for cpusets is cleaned up to reduce boilerplate. This patch: This change tidies up the cpusets control file definitions, and reduces the amount of boilerplate required to add/change control files in the future. Signed-off-by: Paul Menage Reviewed-by: Li Zefan Acked-by: Paul Jackson Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

kernel/cpuset.c: make 3 functions static

2008-04-29T15:06:11Z

Make the following needlessly global functions static: - cpuset_test_cpumask() - cpuset_change_cpumask() - cpuset_do_move_task() Signed-off-by: Adrian Bunk Acked-by: Paul Jackson Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

CGroup API files: update cpusets to use cgroup structured file API

2008-04-29T15:06:08Z

Many of the cpusets control files are simple integer values, which don't require the overhead of memory allocations for reads and writes. Move the handlers for these control files into cpuset_read_u64() and cpuset_write_u64(). [akpm@linux-foundation.org: ad dmissing `break'] Signed-off-by: Paul Menage Cc: "Li Zefan" Cc: Balbir Singh Cc: Paul Jackson Cc: Pavel Emelyanov Cc: KAMEZAWA Hiroyuki Cc: "YAMAMOTO Takashi" Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds