2007-02-14 16:34:06 +08:00
|
|
|
/*
|
|
|
|
* Copyright (C) 2007
|
|
|
|
*
|
|
|
|
* Author: Eric Biederman <ebiederm@xmision.com>
|
|
|
|
*
|
|
|
|
* This program is free software; you can redistribute it and/or
|
|
|
|
* modify it under the terms of the GNU General Public License as
|
|
|
|
* published by the Free Software Foundation, version 2 of the
|
|
|
|
* License.
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include <linux/module.h>
|
|
|
|
#include <linux/ipc.h>
|
|
|
|
#include <linux/nsproxy.h>
|
|
|
|
#include <linux/sysctl.h>
|
|
|
|
#include <linux/uaccess.h>
|
2008-02-08 20:18:22 +08:00
|
|
|
#include <linux/ipc_namespace.h>
|
2008-04-29 16:00:45 +08:00
|
|
|
#include <linux/msg.h>
|
|
|
|
#include "util.h"
|
2007-02-14 16:34:06 +08:00
|
|
|
|
2014-06-07 05:38:07 +08:00
|
|
|
static void *get_ipc(struct ctl_table *table)
|
2007-02-14 16:34:06 +08:00
|
|
|
{
|
|
|
|
char *which = table->data;
|
|
|
|
struct ipc_namespace *ipc_ns = current->nsproxy->ipc_ns;
|
|
|
|
which = (which - (char *)&init_ipc_ns) + (char *)ipc_ns;
|
|
|
|
return which;
|
|
|
|
}
|
|
|
|
|
2009-04-03 07:58:27 +08:00
|
|
|
#ifdef CONFIG_PROC_SYSCTL
|
2014-06-07 05:38:07 +08:00
|
|
|
static int proc_ipc_dointvec(struct ctl_table *table, int write,
|
2007-02-14 16:34:06 +08:00
|
|
|
void __user *buffer, size_t *lenp, loff_t *ppos)
|
|
|
|
{
|
|
|
|
struct ctl_table ipc_table;
|
2011-07-27 07:08:48 +08:00
|
|
|
|
2007-02-14 16:34:06 +08:00
|
|
|
memcpy(&ipc_table, table, sizeof(ipc_table));
|
|
|
|
ipc_table.data = get_ipc(table);
|
|
|
|
|
2009-09-24 06:57:19 +08:00
|
|
|
return proc_dointvec(&ipc_table, write, buffer, lenp, ppos);
|
2007-02-14 16:34:06 +08:00
|
|
|
}
|
|
|
|
|
2014-06-07 05:38:07 +08:00
|
|
|
static int proc_ipc_dointvec_minmax(struct ctl_table *table, int write,
|
2011-07-27 07:08:48 +08:00
|
|
|
void __user *buffer, size_t *lenp, loff_t *ppos)
|
|
|
|
{
|
|
|
|
struct ctl_table ipc_table;
|
|
|
|
|
|
|
|
memcpy(&ipc_table, table, sizeof(ipc_table));
|
|
|
|
ipc_table.data = get_ipc(table);
|
|
|
|
|
|
|
|
return proc_dointvec_minmax(&ipc_table, write, buffer, lenp, ppos);
|
|
|
|
}
|
|
|
|
|
2014-06-07 05:38:07 +08:00
|
|
|
static int proc_ipc_dointvec_minmax_orphans(struct ctl_table *table, int write,
|
2011-07-27 07:08:48 +08:00
|
|
|
void __user *buffer, size_t *lenp, loff_t *ppos)
|
|
|
|
{
|
|
|
|
struct ipc_namespace *ns = current->nsproxy->ipc_ns;
|
|
|
|
int err = proc_ipc_dointvec_minmax(table, write, buffer, lenp, ppos);
|
|
|
|
|
|
|
|
if (err < 0)
|
|
|
|
return err;
|
|
|
|
if (ns->shm_rmid_forced)
|
|
|
|
shm_destroy_orphaned(ns);
|
|
|
|
return err;
|
|
|
|
}
|
|
|
|
|
2014-06-07 05:38:07 +08:00
|
|
|
static int proc_ipc_doulongvec_minmax(struct ctl_table *table, int write,
|
2009-09-24 06:57:19 +08:00
|
|
|
void __user *buffer, size_t *lenp, loff_t *ppos)
|
2007-02-14 16:34:06 +08:00
|
|
|
{
|
|
|
|
struct ctl_table ipc_table;
|
|
|
|
memcpy(&ipc_table, table, sizeof(ipc_table));
|
|
|
|
ipc_table.data = get_ipc(table);
|
|
|
|
|
2009-09-24 06:57:19 +08:00
|
|
|
return proc_doulongvec_minmax(&ipc_table, write, buffer,
|
2007-02-14 16:34:06 +08:00
|
|
|
lenp, ppos);
|
|
|
|
}
|
|
|
|
|
2014-12-13 08:58:17 +08:00
|
|
|
static int proc_ipc_auto_msgmni(struct ctl_table *table, int write,
|
2009-09-24 06:57:19 +08:00
|
|
|
void __user *buffer, size_t *lenp, loff_t *ppos)
|
2008-07-25 16:48:08 +08:00
|
|
|
{
|
|
|
|
struct ctl_table ipc_table;
|
2014-12-13 08:58:17 +08:00
|
|
|
int dummy = 0;
|
2008-07-25 16:48:08 +08:00
|
|
|
|
|
|
|
memcpy(&ipc_table, table, sizeof(ipc_table));
|
2014-12-13 08:58:17 +08:00
|
|
|
ipc_table.data = &dummy;
|
|
|
|
|
|
|
|
if (write)
|
|
|
|
pr_info_once("writing to auto_msgmni has no effect");
|
|
|
|
|
|
|
|
return proc_dointvec_minmax(&ipc_table, write, buffer, lenp, ppos);
|
2008-07-25 16:48:08 +08:00
|
|
|
}
|
|
|
|
|
2018-10-31 06:07:24 +08:00
|
|
|
static int proc_ipc_sem_dointvec(struct ctl_table *table, int write,
|
|
|
|
void __user *buffer, size_t *lenp, loff_t *ppos)
|
|
|
|
{
|
|
|
|
int ret, semmni;
|
|
|
|
struct ipc_namespace *ns = current->nsproxy->ipc_ns;
|
|
|
|
|
|
|
|
semmni = ns->sem_ctls[3];
|
|
|
|
ret = proc_ipc_dointvec(table, write, buffer, lenp, ppos);
|
|
|
|
|
|
|
|
if (!ret)
|
|
|
|
ret = sem_check_semmni(current->nsproxy->ipc_ns);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Reset the semmni value if an error happens.
|
|
|
|
*/
|
|
|
|
if (ret)
|
|
|
|
ns->sem_ctls[3] = semmni;
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2007-02-14 16:34:06 +08:00
|
|
|
#else
|
|
|
|
#define proc_ipc_doulongvec_minmax NULL
|
|
|
|
#define proc_ipc_dointvec NULL
|
2011-07-27 07:08:48 +08:00
|
|
|
#define proc_ipc_dointvec_minmax NULL
|
|
|
|
#define proc_ipc_dointvec_minmax_orphans NULL
|
2014-12-13 08:58:17 +08:00
|
|
|
#define proc_ipc_auto_msgmni NULL
|
2018-10-31 06:07:24 +08:00
|
|
|
#define proc_ipc_sem_dointvec NULL
|
2007-02-14 16:34:06 +08:00
|
|
|
#endif
|
|
|
|
|
2008-07-25 16:48:08 +08:00
|
|
|
static int zero;
|
|
|
|
static int one = 1;
|
2013-01-05 07:34:50 +08:00
|
|
|
static int int_max = INT_MAX;
|
2019-05-15 06:46:29 +08:00
|
|
|
int ipc_mni = IPCMNI;
|
|
|
|
int ipc_mni_shift = IPCMNI_SHIFT;
|
ipc: do cyclic id allocation for the ipc object.
For ipcmni_extend mode, the sequence number space is only 7 bits. So
the chance of id reuse is relatively high compared with the non-extended
mode.
To alleviate this id reuse problem, this patch enables cyclic allocation
for the index to the radix tree (idx). The disadvantage is that this
can cause a slight slow-down of the fast path, as the radix tree could
be higher than necessary.
To limit the radix tree height, I have chosen the following limits:
1) The cycling is done over in_use*1.5.
2) At least, the cycling is done over
"normal" ipcnmi mode: RADIX_TREE_MAP_SIZE elements
"ipcmni_extended": 4096 elements
Result:
- for normal mode:
No change for <= 42 active ipc elements. With more than 42
active ipc elements, a 2nd level would be added to the radix
tree.
Without cyclic allocation, a 2nd level would be added only with
more than 63 active elements.
- for extended mode:
Cycling creates always at least a 2-level radix tree.
With more than 2730 active objects, a 3rd level would be
added, instead of > 4095 active objects until the 3rd level
is added without cyclic allocation.
For a 2-level radix tree compared to a 1-level radix tree, I have
observed < 1% performance impact.
Notes:
1) Normal "x=semget();y=semget();" is unaffected: Then the idx
is e.g. a and a+1, regardless if idr_alloc() or idr_alloc_cyclic()
is used.
2) The -1% happens in a microbenchmark after this situation:
x=semget();
for(i=0;i<4000;i++) {t=semget();semctl(t,0,IPC_RMID);}
y=semget();
Now perform semget calls on x and y that do not sleep.
3) The worst-case reuse cycle time is unfortunately unaffected:
If you have 2^24-1 ipc objects allocated, and get/remove the last
possible element in a loop, then the id is reused after 128
get/remove pairs.
Performance check:
A microbenchmark that performes no-op semop() randomly on two IDs,
with only these two IDs allocated.
The IDs were set using /proc/sys/kernel/sem_next_id.
The test was run 5 times, averages are shown.
1 & 2: Base (6.22 seconds for 10.000.000 semops)
1 & 40: -0.2%
1 & 3348: - 0.8%
1 & 27348: - 1.6%
1 & 15777204: - 3.2%
Or: ~12.6 cpu cycles per additional radix tree level.
The cpu is an Intel I3-5010U. ~1300 cpu cycles/syscall is slower
than what I remember (spectre impact?).
V2 of the patch:
- use "min" and "max"
- use RADIX_TREE_MAP_SIZE * RADIX_TREE_MAP_SIZE instead of
(2<<12).
[akpm@linux-foundation.org: fix max() warning]
Link: http://lkml.kernel.org/r/20190329204930.21620-3-longman@redhat.com
Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Acked-by: Waiman Long <longman@redhat.com>
Cc: "Luis R. Rodriguez" <mcgrof@kernel.org>
Cc: Kees Cook <keescook@chromium.org>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: "Eric W . Biederman" <ebiederm@xmission.com>
Cc: Takashi Iwai <tiwai@suse.de>
Cc: Davidlohr Bueso <dbueso@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-05-15 06:46:36 +08:00
|
|
|
int ipc_min_cycle = RADIX_TREE_MAP_SIZE;
|
2008-07-25 16:48:08 +08:00
|
|
|
|
2007-02-14 16:34:06 +08:00
|
|
|
static struct ctl_table ipc_kern_table[] = {
|
|
|
|
{
|
|
|
|
.procname = "shmmax",
|
|
|
|
.data = &init_ipc_ns.shm_ctlmax,
|
2014-01-28 09:07:04 +08:00
|
|
|
.maxlen = sizeof(init_ipc_ns.shm_ctlmax),
|
2007-02-14 16:34:06 +08:00
|
|
|
.mode = 0644,
|
|
|
|
.proc_handler = proc_ipc_doulongvec_minmax,
|
|
|
|
},
|
|
|
|
{
|
|
|
|
.procname = "shmall",
|
|
|
|
.data = &init_ipc_ns.shm_ctlall,
|
2014-01-28 09:07:04 +08:00
|
|
|
.maxlen = sizeof(init_ipc_ns.shm_ctlall),
|
2007-02-14 16:34:06 +08:00
|
|
|
.mode = 0644,
|
|
|
|
.proc_handler = proc_ipc_doulongvec_minmax,
|
|
|
|
},
|
|
|
|
{
|
|
|
|
.procname = "shmmni",
|
|
|
|
.data = &init_ipc_ns.shm_ctlmni,
|
2014-01-28 09:07:04 +08:00
|
|
|
.maxlen = sizeof(init_ipc_ns.shm_ctlmni),
|
2007-02-14 16:34:06 +08:00
|
|
|
.mode = 0644,
|
2018-10-31 06:07:20 +08:00
|
|
|
.proc_handler = proc_ipc_dointvec_minmax,
|
|
|
|
.extra1 = &zero,
|
|
|
|
.extra2 = &ipc_mni,
|
2007-02-14 16:34:06 +08:00
|
|
|
},
|
2011-07-27 07:08:48 +08:00
|
|
|
{
|
|
|
|
.procname = "shm_rmid_forced",
|
|
|
|
.data = &init_ipc_ns.shm_rmid_forced,
|
|
|
|
.maxlen = sizeof(init_ipc_ns.shm_rmid_forced),
|
|
|
|
.mode = 0644,
|
|
|
|
.proc_handler = proc_ipc_dointvec_minmax_orphans,
|
|
|
|
.extra1 = &zero,
|
|
|
|
.extra2 = &one,
|
|
|
|
},
|
2007-02-14 16:34:06 +08:00
|
|
|
{
|
|
|
|
.procname = "msgmax",
|
|
|
|
.data = &init_ipc_ns.msg_ctlmax,
|
2014-01-28 09:07:04 +08:00
|
|
|
.maxlen = sizeof(init_ipc_ns.msg_ctlmax),
|
2007-02-14 16:34:06 +08:00
|
|
|
.mode = 0644,
|
2013-11-03 19:36:28 +08:00
|
|
|
.proc_handler = proc_ipc_dointvec_minmax,
|
|
|
|
.extra1 = &zero,
|
|
|
|
.extra2 = &int_max,
|
2007-02-14 16:34:06 +08:00
|
|
|
},
|
|
|
|
{
|
|
|
|
.procname = "msgmni",
|
|
|
|
.data = &init_ipc_ns.msg_ctlmni,
|
2014-01-28 09:07:04 +08:00
|
|
|
.maxlen = sizeof(init_ipc_ns.msg_ctlmni),
|
2007-02-14 16:34:06 +08:00
|
|
|
.mode = 0644,
|
2014-12-13 08:58:17 +08:00
|
|
|
.proc_handler = proc_ipc_dointvec_minmax,
|
2013-11-03 19:36:28 +08:00
|
|
|
.extra1 = &zero,
|
2018-10-31 06:07:20 +08:00
|
|
|
.extra2 = &ipc_mni,
|
2007-02-14 16:34:06 +08:00
|
|
|
},
|
2014-12-13 08:58:17 +08:00
|
|
|
{
|
|
|
|
.procname = "auto_msgmni",
|
|
|
|
.data = NULL,
|
|
|
|
.maxlen = sizeof(int),
|
|
|
|
.mode = 0644,
|
|
|
|
.proc_handler = proc_ipc_auto_msgmni,
|
|
|
|
.extra1 = &zero,
|
|
|
|
.extra2 = &one,
|
|
|
|
},
|
2007-02-14 16:34:06 +08:00
|
|
|
{
|
|
|
|
.procname = "msgmnb",
|
|
|
|
.data = &init_ipc_ns.msg_ctlmnb,
|
2014-01-28 09:07:04 +08:00
|
|
|
.maxlen = sizeof(init_ipc_ns.msg_ctlmnb),
|
2007-02-14 16:34:06 +08:00
|
|
|
.mode = 0644,
|
2013-11-03 19:36:28 +08:00
|
|
|
.proc_handler = proc_ipc_dointvec_minmax,
|
|
|
|
.extra1 = &zero,
|
|
|
|
.extra2 = &int_max,
|
2007-02-14 16:34:06 +08:00
|
|
|
},
|
|
|
|
{
|
|
|
|
.procname = "sem",
|
|
|
|
.data = &init_ipc_ns.sem_ctls,
|
2014-01-28 09:07:04 +08:00
|
|
|
.maxlen = 4*sizeof(int),
|
2007-02-14 16:34:06 +08:00
|
|
|
.mode = 0644,
|
2018-10-31 06:07:24 +08:00
|
|
|
.proc_handler = proc_ipc_sem_dointvec,
|
2007-02-14 16:34:06 +08:00
|
|
|
},
|
2013-01-05 07:34:50 +08:00
|
|
|
#ifdef CONFIG_CHECKPOINT_RESTORE
|
|
|
|
{
|
|
|
|
.procname = "sem_next_id",
|
|
|
|
.data = &init_ipc_ns.ids[IPC_SEM_IDS].next_id,
|
|
|
|
.maxlen = sizeof(init_ipc_ns.ids[IPC_SEM_IDS].next_id),
|
|
|
|
.mode = 0644,
|
|
|
|
.proc_handler = proc_ipc_dointvec_minmax,
|
|
|
|
.extra1 = &zero,
|
|
|
|
.extra2 = &int_max,
|
|
|
|
},
|
|
|
|
{
|
|
|
|
.procname = "msg_next_id",
|
|
|
|
.data = &init_ipc_ns.ids[IPC_MSG_IDS].next_id,
|
|
|
|
.maxlen = sizeof(init_ipc_ns.ids[IPC_MSG_IDS].next_id),
|
|
|
|
.mode = 0644,
|
|
|
|
.proc_handler = proc_ipc_dointvec_minmax,
|
|
|
|
.extra1 = &zero,
|
|
|
|
.extra2 = &int_max,
|
|
|
|
},
|
|
|
|
{
|
|
|
|
.procname = "shm_next_id",
|
|
|
|
.data = &init_ipc_ns.ids[IPC_SHM_IDS].next_id,
|
|
|
|
.maxlen = sizeof(init_ipc_ns.ids[IPC_SHM_IDS].next_id),
|
|
|
|
.mode = 0644,
|
|
|
|
.proc_handler = proc_ipc_dointvec_minmax,
|
|
|
|
.extra1 = &zero,
|
|
|
|
.extra2 = &int_max,
|
|
|
|
},
|
|
|
|
#endif
|
2007-02-14 16:34:06 +08:00
|
|
|
{}
|
|
|
|
};
|
|
|
|
|
|
|
|
static struct ctl_table ipc_root_table[] = {
|
|
|
|
{
|
|
|
|
.procname = "kernel",
|
|
|
|
.mode = 0555,
|
|
|
|
.child = ipc_kern_table,
|
|
|
|
},
|
|
|
|
{}
|
|
|
|
};
|
|
|
|
|
|
|
|
static int __init ipc_sysctl_init(void)
|
|
|
|
{
|
2007-02-14 16:34:09 +08:00
|
|
|
register_sysctl_table(ipc_root_table);
|
2007-02-14 16:34:06 +08:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2014-04-08 06:39:18 +08:00
|
|
|
device_initcall(ipc_sysctl_init);
|
2019-05-15 06:46:29 +08:00
|
|
|
|
|
|
|
static int __init ipc_mni_extend(char *str)
|
|
|
|
{
|
|
|
|
ipc_mni = IPCMNI_EXTEND;
|
|
|
|
ipc_mni_shift = IPCMNI_EXTEND_SHIFT;
|
ipc: do cyclic id allocation for the ipc object.
For ipcmni_extend mode, the sequence number space is only 7 bits. So
the chance of id reuse is relatively high compared with the non-extended
mode.
To alleviate this id reuse problem, this patch enables cyclic allocation
for the index to the radix tree (idx). The disadvantage is that this
can cause a slight slow-down of the fast path, as the radix tree could
be higher than necessary.
To limit the radix tree height, I have chosen the following limits:
1) The cycling is done over in_use*1.5.
2) At least, the cycling is done over
"normal" ipcnmi mode: RADIX_TREE_MAP_SIZE elements
"ipcmni_extended": 4096 elements
Result:
- for normal mode:
No change for <= 42 active ipc elements. With more than 42
active ipc elements, a 2nd level would be added to the radix
tree.
Without cyclic allocation, a 2nd level would be added only with
more than 63 active elements.
- for extended mode:
Cycling creates always at least a 2-level radix tree.
With more than 2730 active objects, a 3rd level would be
added, instead of > 4095 active objects until the 3rd level
is added without cyclic allocation.
For a 2-level radix tree compared to a 1-level radix tree, I have
observed < 1% performance impact.
Notes:
1) Normal "x=semget();y=semget();" is unaffected: Then the idx
is e.g. a and a+1, regardless if idr_alloc() or idr_alloc_cyclic()
is used.
2) The -1% happens in a microbenchmark after this situation:
x=semget();
for(i=0;i<4000;i++) {t=semget();semctl(t,0,IPC_RMID);}
y=semget();
Now perform semget calls on x and y that do not sleep.
3) The worst-case reuse cycle time is unfortunately unaffected:
If you have 2^24-1 ipc objects allocated, and get/remove the last
possible element in a loop, then the id is reused after 128
get/remove pairs.
Performance check:
A microbenchmark that performes no-op semop() randomly on two IDs,
with only these two IDs allocated.
The IDs were set using /proc/sys/kernel/sem_next_id.
The test was run 5 times, averages are shown.
1 & 2: Base (6.22 seconds for 10.000.000 semops)
1 & 40: -0.2%
1 & 3348: - 0.8%
1 & 27348: - 1.6%
1 & 15777204: - 3.2%
Or: ~12.6 cpu cycles per additional radix tree level.
The cpu is an Intel I3-5010U. ~1300 cpu cycles/syscall is slower
than what I remember (spectre impact?).
V2 of the patch:
- use "min" and "max"
- use RADIX_TREE_MAP_SIZE * RADIX_TREE_MAP_SIZE instead of
(2<<12).
[akpm@linux-foundation.org: fix max() warning]
Link: http://lkml.kernel.org/r/20190329204930.21620-3-longman@redhat.com
Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Acked-by: Waiman Long <longman@redhat.com>
Cc: "Luis R. Rodriguez" <mcgrof@kernel.org>
Cc: Kees Cook <keescook@chromium.org>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: "Eric W . Biederman" <ebiederm@xmission.com>
Cc: Takashi Iwai <tiwai@suse.de>
Cc: Davidlohr Bueso <dbueso@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-05-15 06:46:36 +08:00
|
|
|
ipc_min_cycle = IPCMNI_EXTEND_MIN_CYCLE;
|
2019-05-15 06:46:29 +08:00
|
|
|
pr_info("IPCMNI extended to %d.\n", ipc_mni);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
early_param("ipcmni_extend", ipc_mni_extend);
|