Various SMP balancing algorithms require that the bandwidth period
run in sync.
Possible improvements are moving the rt_bandwidth thing into root_domain
and keeping a span per rt_bandwidth which marks throttled cpus.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The following commits cause a number of regressions:
commit 58e2d4ca58
Author: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com>
Date: Fri Jan 25 21:08:00 2008 +0100
sched: group scheduling, change how cpu load is calculated
commit 6b2d770026
Author: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com>
Date: Fri Jan 25 21:08:00 2008 +0100
sched: group scheduler, fix fairness of cpu bandwidth allocation for task groups
Namely:
- very frequent wakeups on SMP, reported by PowerTop users.
- cacheline trashing on (large) SMP
- some latencies larger than 500ms
While there is a mergeable patch to fix the latter, the former issues
are not fixable in a manner suitable for .25 (we're at -rc3 now).
Hence we revert them and try again in v2.6.26.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
CC: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com>
Tested-by: Alexey Zaytsev <alexey.zaytsev@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
proc_doulongvec_minmax() calls copy_to_user()/copy_from_user(), so we can't
hold hugetlb_lock over the call. Use a dummy variable to store the sysctl
result, like in hugetlb_sysctl_handler(), then grab the lock to update
nr_overcommit_huge_pages.
Signed-off-by: Nishanth Aravamudan <nacc@us.ibm.com>
Reported-by: Miles Lane <miles.lane@gmail.com>
Cc: Adam Litke <agl@us.ibm.com>
Cc: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Change the rt_ratio interface to rt_runtime_us, to match rt_period_us.
This avoids picking a granularity for the ratio.
Extend the /sys/kernel/uids/<uid>/ interface to allow setting
the group's rt_runtime.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Some time ago the xxx_vnr() calls (e.g. pid_vnr or find_task_by_vpid) were
_all_ converted to operate on the current pid namespace. After this each call
like xxx_nr_ns(foo, current->nsproxy->pid_ns) is nothing but a xxx_vnr(foo)
one.
Switch all the xxx_nr_ns() callers to use the xxx_vnr() calls where
appropriate.
Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Reviewed-by: Oleg Nesterov <oleg@tv-sign.ru>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Balbir Singh <balbir@in.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
When I replaced hugetlb_dynamic_pool with nr_overcommit_hugepages I used
proc_doulongvec_minmax() directly. However, hugetlb.c's locking rules
require that all counter modifications occur under the hugetlb_lock. Add a
callback into the hugetlb code similar to the one for nr_hugepages. Grab
the lock around the manipulation of nr_overcommit_hugepages in
proc_doulongvec_minmax().
Signed-off-by: Nishanth Aravamudan <nacc@us.ibm.com>
Acked-by: Adam Litke <agl@us.ibm.com>
Cc: David Gibson <david@gibson.dropbear.id.au>
Cc: William Lee Irwin III <wli@holomorphy.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Adds a new sysctl, 'oom_dump_tasks', that enables the kernel to produce a
dump of all system tasks (excluding kernel threads) when performing an
OOM-killing. Information includes pid, uid, tgid, vm size, rss, cpu,
oom_adj score, and name.
This is helpful for determining why there was an OOM condition and which
rogue task caused it.
It is configurable so that large systems, such as those with several
thousand tasks, do not incur a performance penalty associated with dumping
data they may not desire.
If an OOM was triggered as a result of a memory controller, the tasklist
shall be filtered to exclude tasks that are not a member of the same
cgroup.
Cc: Andrea Arcangeli <andrea@suse.de>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Signed-off-by: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
NR_OPEN (historically set to 1024*1024) actually forbids processes to open
more than 1024*1024 handles.
Unfortunatly some production servers hit the not so 'ridiculously high
value' of 1024*1024 file descriptors per process.
Changing NR_OPEN is not considered safe because of vmalloc space potential
exhaust.
This patch introduces a new sysctl (/proc/sys/fs/nr_open) wich defaults to
1024*1024, so that admins can decide to change this limit if their workload
needs it.
[akpm@linux-foundation.org: export it for sparc64]
Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Cc: Richard Henderson <rth@twiddle.net>
Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The capability bounding set is a set beyond which capabilities cannot grow.
Currently cap_bset is per-system. It can be manipulated through sysctl,
but only init can add capabilities. Root can remove capabilities. By
default it includes all caps except CAP_SETPCAP.
This patch makes the bounding set per-process when file capabilities are
enabled. It is inherited at fork from parent. Noone can add elements,
CAP_SETPCAP is required to remove them.
One example use of this is to start a safer container. For instance, until
device namespaces or per-container device whitelists are introduced, it is
best to take CAP_MKNOD away from a container.
The bounding set will not affect pP and pE immediately. It will only
affect pP' and pE' after subsequent exec()s. It also does not affect pI,
and exec() does not constrain pI'. So to really start a shell with no way
of regain CAP_MKNOD, you would do
prctl(PR_CAPBSET_DROP, CAP_MKNOD);
cap_t cap = cap_get_proc();
cap_value_t caparray[1];
caparray[0] = CAP_MKNOD;
cap_set_flag(cap, CAP_INHERITABLE, 1, caparray, CAP_DROP);
cap_set_proc(cap);
cap_free(cap);
The following test program will get and set the bounding
set (but not pI). For instance
./bset get
(lists capabilities in bset)
./bset drop cap_net_raw
(starts shell with new bset)
(use capset, setuid binary, or binary with
file capabilities to try to increase caps)
************************************************************
cap_bound.c
************************************************************
#include <sys/prctl.h>
#include <linux/capability.h>
#include <sys/types.h>
#include <unistd.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#ifndef PR_CAPBSET_READ
#define PR_CAPBSET_READ 23
#endif
#ifndef PR_CAPBSET_DROP
#define PR_CAPBSET_DROP 24
#endif
int usage(char *me)
{
printf("Usage: %s get\n", me);
printf(" %s drop <capability>\n", me);
return 1;
}
#define numcaps 32
char *captable[numcaps] = {
"cap_chown",
"cap_dac_override",
"cap_dac_read_search",
"cap_fowner",
"cap_fsetid",
"cap_kill",
"cap_setgid",
"cap_setuid",
"cap_setpcap",
"cap_linux_immutable",
"cap_net_bind_service",
"cap_net_broadcast",
"cap_net_admin",
"cap_net_raw",
"cap_ipc_lock",
"cap_ipc_owner",
"cap_sys_module",
"cap_sys_rawio",
"cap_sys_chroot",
"cap_sys_ptrace",
"cap_sys_pacct",
"cap_sys_admin",
"cap_sys_boot",
"cap_sys_nice",
"cap_sys_resource",
"cap_sys_time",
"cap_sys_tty_config",
"cap_mknod",
"cap_lease",
"cap_audit_write",
"cap_audit_control",
"cap_setfcap"
};
int getbcap(void)
{
int comma=0;
unsigned long i;
int ret;
printf("i know of %d capabilities\n", numcaps);
printf("capability bounding set:");
for (i=0; i<numcaps; i++) {
ret = prctl(PR_CAPBSET_READ, i);
if (ret < 0)
perror("prctl");
else if (ret==1)
printf("%s%s", (comma++) ? ", " : " ", captable[i]);
}
printf("\n");
return 0;
}
int capdrop(char *str)
{
unsigned long i;
int found=0;
for (i=0; i<numcaps; i++) {
if (strcmp(captable[i], str) == 0) {
found=1;
break;
}
}
if (!found)
return 1;
if (prctl(PR_CAPBSET_DROP, i)) {
perror("prctl");
return 1;
}
return 0;
}
int main(int argc, char *argv[])
{
if (argc<2)
return usage(argv[0]);
if (strcmp(argv[1], "get")==0)
return getbcap();
if (strcmp(argv[1], "drop")!=0 || argc<3)
return usage(argv[0]);
if (capdrop(argv[2])) {
printf("unknown capability\n");
return 1;
}
return execl("/bin/bash", "/bin/bash", NULL);
}
************************************************************
[serue@us.ibm.com: fix typo]
Signed-off-by: Serge E. Hallyn <serue@us.ibm.com>
Signed-off-by: Andrew G. Morgan <morgan@kernel.org>
Cc: Stephen Smalley <sds@tycho.nsa.gov>
Cc: James Morris <jmorris@namei.org>
Cc: Chris Wright <chrisw@sous-sol.org>
Cc: Casey Schaufler <casey@schaufler-ca.com>a
Signed-off-by: "Serge E. Hallyn" <serue@us.ibm.com>
Tested-by: Jiri Slaby <jirislaby@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Add vm.highmem_is_dirtyable toggle
A 32 bit machine with HIGHMEM64 enabled running DCC has an MMAPed file of
approximately 2Gb size which contains a hash format that is written
randomly by the dbclean process. On 2.6.16 this process took a few
minutes. With lowmem only accounting of dirty ratios, this takes about 12
hours of 100% disk IO, all random writes.
Include a toggle in /proc/sys/vm/highmem_is_dirtyable which can be set to 1 to
add the highmem back to the total available memory count.
[akpm@linux-foundation.org: Fix the CONFIG_DETECT_SOFTLOCKUP=y build]
Signed-off-by: Bron Gondwana <brong@fastmail.fm>
Cc: Ethan Solomita <solo@google.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: WU Fengguang <wfg@mail.ustc.edu.cn>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
execve arguments can be quite large. There is no limit on the number of
arguments and a 4G limit on the size of an argument.
this patch prints those aruguments in bite sized pieces. a userspace size
limitation of 8k was discovered so this keeps messages around 7.5k
single arguments larger than 7.5k in length are split into multiple records
and can be identified as aX[Y]=
Signed-off-by: Eric Paris <eparis@redhat.com>
various changes to the in_p/out_p delay details:
- add the io_delay=none method
- make each method selectable from the kernel config
- simplify the delay code a bit by getting rid of an indirect function call
- add the /proc/sys/kernel/io_delay_type sysctl
- change 'io_delay=standard|alternate' to io_delay=0x80 and io_delay=0xed
- make the io delay config not depend on CONFIG_DEBUG_KERNEL
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: "David P. Reed" <dpreed@reed.com>
I have removed all the entries from this table (core_table,
ipv4_table and tr_table), so now we can safely drop it.
Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch implements the basic infrastructure for per namespace sysctls.
A list of lists of sysctl headers is added, allowing each namespace to have
it's own list of sysctl headers.
Each list of sysctl headers has a lookup function to find the first
sysctl header in the list, allowing the lists to have a per namespace
instance.
register_sysct_root is added to tell sysctl.c about additional
lists of sysctl_headers. As all of the users are expected to be in
kernel no unregister function is provided.
sysctl_head_next is updated to walk through the list of lists.
__register_sysctl_paths is added to add a new sysctl table on
a non-default sysctl list.
The only intrusive part of this patch is propagating the information
to decided which list of sysctls to use for sysctl_check_table.
Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Cc: Serge Hallyn <serue@us.ibm.com>
Cc: Daniel Lezcano <dlezcano@fr.ibm.com>
Cc: Cedric Le Goater <clg@fr.ibm.com>
Cc: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
By doing this we allow users of register_sysctl_paths that build
and dynamically allocate their ctl_table to be simpler. This allows
them to just remember the ctl_table_header returned from
register_sysctl_paths from which they can now find the
ctl_table array they need to free.
Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Cc: Serge Hallyn <serue@us.ibm.com>
Cc: Daniel Lezcano <dlezcano@fr.ibm.com>
Cc: Cedric Le Goater <clg@fr.ibm.com>
Cc: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
There are a number of modules that register a sysctl table
somewhere deeply nested in the sysctl hierarchy, such as
fs/nfs, fs/xfs, dev/cdrom, etc.
They all specify several dummy ctl_tables for the path name.
This patch implements register_sysctl_path that takes
an additional path name, and makes up dummy sysctl nodes
for each component.
This patch was originally written by Olaf Kirch and
brought to my attention and reworked some by Olaf Hering.
I have changed a few additional things so the bugs are mine.
After converting all of the easy callers Olaf Hering observed
allyesconfig ARCH=i386, the patch reduces the final binary size by 9369 bytes.
.text +897
.data -7008
text data bss dec hex filename
26959310 4045899 4718592 35723801 2211a19 ../vmlinux-vanilla
26960207 4038891 4718592 35717690 221023a ../O-allyesconfig/vmlinux
So this change is both a space savings and a code simplification.
CC: Olaf Kirch <okir@suse.de>
CC: Olaf Hering <olaf@aepfle.de>
Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Cc: Serge Hallyn <serue@us.ibm.com>
Cc: Daniel Lezcano <dlezcano@fr.ibm.com>
Cc: Cedric Le Goater <clg@fr.ibm.com>
Cc: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
LatencyTOP kernel infrastructure; it measures latencies in the
scheduler and tracks it system wide and per process.
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Very simple time limit on the realtime scheduling classes.
Allow the rq's realtime class to consume sched_rt_ratio of every
sched_rt_period slice. If the class exceeds this quota the fair class
will preempt the realtime class.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
this patch extends the soft-lockup detector to automatically
detect hung TASK_UNINTERRUPTIBLE tasks. Such hung tasks are
printed the following way:
------------------>
INFO: task prctl:3042 blocked for more than 120 seconds.
"echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message
prctl D fd5e3793 0 3042 2997
f6050f38 00000046 00000001 fd5e3793 00000009 c06d8264 c06dae80 00000286
f6050f40 f6050f00 f7d34d90 f7d34fc8 c1e1be80 00000001 f6050000 00000000
f7e92d00 00000286 f6050f18 c0489d1a f6050f40 00006605 00000000 c0133a5b
Call Trace:
[<c04883a5>] schedule_timeout+0x6d/0x8b
[<c04883d8>] schedule_timeout_uninterruptible+0x15/0x17
[<c0133a76>] msleep+0x10/0x16
[<c0138974>] sys_prctl+0x30/0x1e2
[<c0104c52>] sysenter_past_esp+0x5f/0xa5
=======================
2 locks held by prctl/3042:
#0: (&sb->s_type->i_mutex_key#5){--..}, at: [<c0197d11>] do_fsync+0x38/0x7a
#1: (jbd_handle){--..}, at: [<c01ca3d2>] journal_start+0xc7/0xe9
<------------------
the current default timeout is 120 seconds. Such messages are printed
up to 10 times per bootup. If the system has crashed already then the
messages are not printed.
if lockdep is enabled then all held locks are printed as well.
this feature is a natural extension to the softlockup-detector (kernel
locked up without scheduling) and to the NMI watchdog (kernel locked up
with IRQs disabled).
[ Gautham R Shenoy <ego@in.ibm.com>: CPU hotplug fixes. ]
[ Andrew Morton <akpm@linux-foundation.org>: build warning fix. ]
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
The current load balancing scheme isn't good enough for precise
group fairness.
For example: on a 8-cpu system, I created 3 groups as under:
a = 8 tasks (cpu.shares = 1024)
b = 4 tasks (cpu.shares = 1024)
c = 3 tasks (cpu.shares = 1024)
a, b and c are task groups that have equal weight. We would expect each
of the groups to receive 33.33% of cpu bandwidth under a fair scheduler.
This is what I get with the latest scheduler git tree:
Signed-off-by: Ingo Molnar <mingo@elte.hu>
--------------------------------------------------------------------------------
Col1 | Col2 | Col3 | Col4
------|---------|-------|-------------------------------------------------------
a | 277.676 | 57.8% | 54.1% 54.1% 54.1% 54.2% 56.7% 62.2% 62.8% 64.5%
b | 116.108 | 24.2% | 47.4% 48.1% 48.7% 49.3%
c | 86.326 | 18.0% | 47.5% 47.9% 48.5%
--------------------------------------------------------------------------------
Explanation of o/p:
Col1 -> Group name
Col2 -> Cumulative execution time (in seconds) received by all tasks of that
group in a 60sec window across 8 cpus
Col3 -> CPU bandwidth received by the group in the 60sec window, expressed in
percentage. Col3 data is derived as:
Col3 = 100 * Col2 / (NR_CPUS * 60)
Col4 -> CPU bandwidth received by each individual task of the group.
Col4 = 100 * cpu_time_recd_by_task / 60
[I can share the test case that produces a similar o/p if reqd]
The deviation from desired group fairness is as below:
a = +24.47%
b = -9.13%
c = -15.33%
which is quite high.
After the patch below is applied, here are the results:
--------------------------------------------------------------------------------
Col1 | Col2 | Col3 | Col4
------|---------|-------|-------------------------------------------------------
a | 163.112 | 34.0% | 33.2% 33.4% 33.5% 33.5% 33.7% 34.4% 34.8% 35.3%
b | 156.220 | 32.5% | 63.3% 64.5% 66.1% 66.5%
c | 160.653 | 33.5% | 85.8% 90.6% 91.4%
--------------------------------------------------------------------------------
Deviation from desired group fairness is as below:
a = +0.67%
b = -0.83%
c = +0.17%
which is far better IMO. Most of other runs have yielded a deviation within
+-2% at the most, which is good.
Why do we see bad (group) fairness with current scheuler?
=========================================================
Currently cpu's weight is just the summation of individual task weights.
This can yield incorrect results. For ex: consider three groups as below
on a 2-cpu system:
CPU0 CPU1
---------------------------
A (10) B(5)
C(5)
---------------------------
Group A has 10 tasks, all on CPU0, Group B and C have 5 tasks each all
of which are on CPU1. Each task has the same weight (NICE_0_LOAD =
1024).
The current scheme would yield a cpu weight of 10240 (10*1024) for each cpu and
the load balancer will think both CPUs are perfectly balanced and won't
move around any tasks. This, however, would yield this bandwidth:
A = 50%
B = 25%
C = 25%
which is not the desired result.
What's changing in the patch?
=============================
- How cpu weights are calculated when CONFIF_FAIR_GROUP_SCHED is
defined (see below)
- API Change
- Two tunables introduced in sysfs (under SCHED_DEBUG) to
control the frequency at which the load balance monitor
thread runs.
The basic change made in this patch is how cpu weight (rq->load.weight) is
calculated. Its now calculated as the summation of group weights on a cpu,
rather than summation of task weights. Weight exerted by a group on a
cpu is dependent on the shares allocated to it and also the number of
tasks the group has on that cpu compared to the total number of
(runnable) tasks the group has in the system.
Let,
W(K,i) = Weight of group K on cpu i
T(K,i) = Task load present in group K's cfs_rq on cpu i
T(K) = Total task load of group K across various cpus
S(K) = Shares allocated to group K
NRCPUS = Number of online cpus in the scheduler domain to
which group K is assigned.
Then,
W(K,i) = S(K) * NRCPUS * T(K,i) / T(K)
A load balance monitor thread is created at bootup, which periodically
runs and adjusts group's weight on each cpu. To avoid its overhead, two
min/max tunables are introduced (under SCHED_DEBUG) to control the rate
at which it runs.
Fixes from: Peter Zijlstra <a.p.zijlstra@chello.nl>
- don't start the load_balance_monitor when there is only a single cpu.
- rename the kthread because its currently longer than TASK_COMM_LEN
Signed-off-by: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
min_sched_granularity_ns, max_sched_granularity_ns,
min_wakeup_granularity_ns and max_wakeup_granularity_ns are declared
"unsigned long".
This is incorrect since proc_dointvec_minmax() expects plain "int" guard
values.
This bug only triggers on big endian 64 bit arches.
Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
