Now that "struct proc_ops" exist we can start putting there stuff which
could not fly with VFS "struct file_operations"...
Most of fs/proc/inode.c file is dedicated to make open/read/.../close
reliable in the event of disappearing /proc entries which usually happens
if module is getting removed. Files like /proc/cpuinfo which never
disappear simply do not need such protection.
Save 2 atomic ops, 1 allocation, 1 free per open/read/close sequence for such
"permanent" files.
Enable "permanent" flag for
/proc/cpuinfo
/proc/kmsg
/proc/modules
/proc/slabinfo
/proc/stat
/proc/sysvipc/*
/proc/swaps
More will come once I figure out foolproof way to prevent out module
authors from marking their stuff "permanent" for performance reasons
when it is not.
This should help with scalability: benchmark is "read /proc/cpuinfo R times
by N threads scattered over the system".
N R t, s (before) t, s (after)
-----------------------------------------------------
64 4096 1.582458 1.530502 -3.2%
256 4096 6.371926 6.125168 -3.9%
1024 4096 25.64888 24.47528 -4.6%
Benchmark source:
#include <chrono>
#include <iostream>
#include <thread>
#include <vector>
#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <unistd.h>
const int NR_CPUS = sysconf(_SC_NPROCESSORS_ONLN);
int N;
const char *filename;
int R;
int xxx = 0;
int glue(int n)
{
cpu_set_t m;
CPU_ZERO(&m);
CPU_SET(n, &m);
return sched_setaffinity(0, sizeof(cpu_set_t), &m);
}
void f(int n)
{
glue(n % NR_CPUS);
while (*(volatile int *)&xxx == 0) {
}
for (int i = 0; i < R; i++) {
int fd = open(filename, O_RDONLY);
char buf[4096];
ssize_t rv = read(fd, buf, sizeof(buf));
asm volatile ("" :: "g" (rv));
close(fd);
}
}
int main(int argc, char *argv[])
{
if (argc < 4) {
std::cerr << "usage: " << argv[0] << ' ' << "N /proc/filename R
";
return 1;
}
N = atoi(argv[1]);
filename = argv[2];
R = atoi(argv[3]);
for (int i = 0; i < NR_CPUS; i++) {
if (glue(i) == 0)
break;
}
std::vector<std::thread> T;
T.reserve(N);
for (int i = 0; i < N; i++) {
T.emplace_back(f, i);
}
auto t0 = std::chrono::system_clock::now();
{
*(volatile int *)&xxx = 1;
for (auto& t: T) {
t.join();
}
}
auto t1 = std::chrono::system_clock::now();
std::chrono::duration<double> dt = t1 - t0;
std::cout << dt.count() << '
';
return 0;
}
P.S.:
Explicit randomization marker is added because adding non-function pointer
will silently disable structure layout randomization.
[akpm@linux-foundation.org: coding style fixes]
Reported-by: kbuild test robot <lkp@intel.com>
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Joe Perches <joe@perches.com>
Link: http://lkml.kernel.org/r/20200222201539.GA22576@avx2
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Replace all the occurrences of FIELD_SIZEOF() with sizeof_field() except
at places where these are defined. Later patches will remove the unused
definition of FIELD_SIZEOF().
This patch is generated using following script:
EXCLUDE_FILES="include/linux/stddef.h|include/linux/kernel.h"
git grep -l -e "\bFIELD_SIZEOF\b" | while read file;
do
if [[ "$file" =~ $EXCLUDE_FILES ]]; then
continue
fi
sed -i -e 's/\bFIELD_SIZEOF\b/sizeof_field/g' $file;
done
Signed-off-by: Pankaj Bharadiya <pankaj.laxminarayan.bharadiya@intel.com>
Link: https://lore.kernel.org/r/20190924105839.110713-3-pankaj.laxminarayan.bharadiya@intel.com
Co-developed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Kees Cook <keescook@chromium.org>
Acked-by: David Miller <davem@davemloft.net> # for net
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>
Rewrite, based on the patch from Waiman Long:
The mixing in of a sequence number into the IPC IDs is probably to avoid
ID reuse in userspace as much as possible. With ipcmni_extend mode, the
number of usable sequence numbers is greatly reduced leading to higher
chance of ID reuse.
To address this issue, we need to conserve the sequence number space as
much as possible. Right now, the sequence number is incremented for
every new ID created. In reality, we only need to increment the
sequence number when new allocated ID is not greater than the last one
allocated. It is in such case that the new ID may collide with an
existing one. This is being done irrespective of the ipcmni mode.
In order to avoid any races, the index is first allocated and then the
pointer is replaced.
Changes compared to the initial patch:
- Handle failures from idr_alloc().
- Avoid that concurrent operations can see the wrong sequence number.
(This is achieved by using idr_replace()).
- IPCMNI_SEQ_SHIFT is not a constant, thus renamed to
ipcmni_seq_shift().
- IPCMNI_SEQ_MAX is not a constant, thus renamed to ipcmni_seq_max().
Link: http://lkml.kernel.org/r/20190329204930.21620-2-longman@redhat.com
Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Waiman Long <longman@redhat.com>
Suggested-by: Matthew Wilcox <willy@infradead.org>
Acked-by: Waiman Long <longman@redhat.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Davidlohr Bueso <dbueso@suse.de>
Cc: "Eric W . Biederman" <ebiederm@xmission.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Kees Cook <keescook@chromium.org>
Cc: "Luis R. Rodriguez" <mcgrof@kernel.org>
Cc: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The maximum number of unique System V IPC identifiers was limited to
32k. That limit should be big enough for most use cases.
However, there are some users out there requesting for more, especially
those that are migrating from Solaris which uses 24 bits for unique
identifiers. To satisfy the need of those users, a new boot time kernel
option "ipcmni_extend" is added to extend the IPCMNI value to 16M. This
is a 512X increase which should be big enough for users out there that
need a large number of unique IPC identifier.
The use of this new option will change the pattern of the IPC
identifiers returned by functions like shmget(2). An application that
depends on such pattern may not work properly. So it should only be
used if the users really need more than 32k of unique IPC numbers.
This new option does have the side effect of reducing the maximum number
of unique sequence numbers from 64k down to 128. So it is a trade-off.
The computation of a new IPC id is not done in the performance critical
path. So a little bit of additional overhead shouldn't have any real
performance impact.
Link: http://lkml.kernel.org/r/20190329204930.21620-1-longman@redhat.com
Signed-off-by: Waiman Long <longman@redhat.com>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Davidlohr Bueso <dbueso@suse.de>
Cc: "Eric W . Biederman" <ebiederm@xmission.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Kees Cook <keescook@chromium.org>
Cc: "Luis R. Rodriguez" <mcgrof@kernel.org>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This patch changes rhashtables to use a bit_spin_lock on BIT(1) of the
bucket pointer to lock the hash chain for that bucket.
The benefits of a bit spin_lock are:
- no need to allocate a separate array of locks.
- no need to have a configuration option to guide the
choice of the size of this array
- locking cost is often a single test-and-set in a cache line
that will have to be loaded anyway. When inserting at, or removing
from, the head of the chain, the unlock is free - writing the new
address in the bucket head implicitly clears the lock bit.
For __rhashtable_insert_fast() we ensure this always happens
when adding a new key.
- even when lockings costs 2 updates (lock and unlock), they are
in a cacheline that needs to be read anyway.
The cost of using a bit spin_lock is a little bit of code complexity,
which I think is quite manageable.
Bit spin_locks are sometimes inappropriate because they are not fair -
if multiple CPUs repeatedly contend of the same lock, one CPU can
easily be starved. This is not a credible situation with rhashtable.
Multiple CPUs may want to repeatedly add or remove objects, but they
will typically do so at different buckets, so they will attempt to
acquire different locks.
As we have more bit-locks than we previously had spinlocks (by at
least a factor of two) we can expect slightly less contention to
go with the slightly better cache behavior and reduced memory
consumption.
To enhance type checking, a new struct is introduced to represent the
pointer plus lock-bit
that is stored in the bucket-table. This is "struct rhash_lock_head"
and is empty. A pointer to this needs to be cast to either an
unsigned lock, or a "struct rhash_head *" to be useful.
Variables of this type are most often called "bkt".
Previously "pprev" would sometimes point to a bucket, and sometimes a
->next pointer in an rhash_head. As these are now different types,
pprev is NULL when it would have pointed to the bucket. In that case,
'blk' is used, together with correct locking protocol.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In sysvipc we have an ids->tables_initialized regarding the rhashtable,
introduced in 0cfb6aee70 ("ipc: optimize semget/shmget/msgget for lots
of keys")
It's there, specifically, to prevent nil pointer dereferences, from using
an uninitialized api. Considering how rhashtable_init() can fail
(probably due to ENOMEM, if anything), this made the overall ipc
initialization capable of failure as well. That alone is ugly, but fine,
however I've spotted a few issues regarding the semantics of
tables_initialized (however unlikely they may be):
- There is inconsistency in what we return to userspace: ipc_addid()
returns ENOSPC which is certainly _wrong_, while ipc_obtain_object_idr()
returns EINVAL.
- After we started using rhashtables, ipc_findkey() can return nil upon
!tables_initialized, but the caller expects nil for when the ipc
structure isn't found, and can therefore call into ipcget() callbacks.
Now that rhashtable initialization cannot fail, we can properly get rid of
the hack altogether.
[manfred@colorfullife.com: commit id extended to 12 digits]
Link: http://lkml.kernel.org/r/20180712185241.4017-10-manfred@colorfullife.com
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Cc: Kees Cook <keescook@chromium.org>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Cc: Michal Hocko <mhocko@suse.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
ipc/util.c contains multiple functions to get the ipc object pointer given
an id number.
There are two sets of function: One set verifies the sequence counter part
of the id number, other functions do not check the sequence counter.
The standard for function names in ipc/util.c is
- ..._check() functions verify the sequence counter
- ..._idr() functions do not verify the sequence counter
ipc_lock() is an exception: It does not verify the sequence counter value,
but this is not obvious from the function name.
Furthermore, shm.c is the only user of this helper. Thus, we can simply
move the logic into shm_lock() and get rid of the function altogether.
[manfred@colorfullife.com: most of changelog]
Link: http://lkml.kernel.org/r/20180712185241.4017-7-manfred@colorfullife.com
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Cc: Kees Cook <keescook@chromium.org>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Cc: Michal Hocko <mhocko@suse.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
ipc_addid() initializes kern_ipc_perm.seq after having called idr_alloc()
(within ipc_idr_alloc()).
Thus a parallel semop() or msgrcv() that uses ipc_obtain_object_check()
may see an uninitialized value.
The patch moves the initialization of kern_ipc_perm.seq before the calls
of idr_alloc().
Notes:
1) This patch has a user space visible side effect:
If /proc/sys/kernel/*_next_id is used (i.e.: checkpoint/restore) and
if semget()/msgget()/shmget() fails in the final step of adding the id
to the rhash tree, then .._next_id is cleared. Before the patch, is
remained unmodified.
There is no change of the behavior after a successful ..get() call: It
always clears .._next_id, there is no impact to non checkpoint/restore
code as that code does not use .._next_id.
2) The patch correctly documents that after a call to ipc_idr_alloc(),
the full tear-down sequence must be used. The callers of ipc_addid()
do not fullfill that, i.e. more bugfixes are required.
The patch is a squash of a patch from Dmitry and my own changes.
Link: http://lkml.kernel.org/r/20180712185241.4017-3-manfred@colorfullife.com
Reported-by: syzbot+2827ef6b3385deb07eaf@syzkaller.appspotmail.com
Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Cc: Davidlohr Bueso <dbueso@suse.de>
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Cc: Michal Hocko <mhocko@suse.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Due to the use of rhashtables in net namespaces,
rhashtable.h is included in lots of the kernel,
so a small changes can required a large recompilation.
This makes development painful.
This patch splits out rhashtable-types.h which just includes
the major type declarations, and does not include (non-trivial)
inline code. rhashtable.h is no longer included by anything
in the include/ directory.
Common include files only include rhashtable-types.h so a large
recompilation is only triggered when that changes.
Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Capture the pid namespace when /proc/sysvipc/msg /proc/sysvipc/shm
and /proc/sysvipc/sem are opened, and make it available through
the new helper ipc_seq_pid_ns.
This makes it possible to report the pids in these files in the
pid namespace of the opener of the files.
Implement ipc_update_pid. A simple impline helper that will only update
a struct pid pointer if the new value does not equal the old value. This
removes the need for wordy code sequences like:
old = object->pid;
object->pid = new;
put_pid(old);
and
old = object->pid;
if (old != new) {
object->pid = new;
put_pid(old);
}
Allowing the following to be written instead:
ipc_update_pid(&object->pid, new);
Which is easier to read and ensures that the pid reference count is
not touched the old and the new values are the same. Not touching
the reference count in this case is important to help avoid issues
like af_unix experienced, where multiple threads of the same
process managed to bounce the struct pid between cpu cache lines,
but updating the pids reference count.
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
For a custom microbenchmark on a 3.30GHz Xeon SandyBridge, which calls
IPC_STAT over and over, it was calculated that, on avg the cost of
ipc_get_maxid() for increasing amounts of keys was:
10 keys: ~900 cycles
100 keys: ~15000 cycles
1000 keys: ~150000 cycles
10000 keys: ~2100000 cycles
This is unsurprising as maxid is currently O(n).
By having the max_id available in O(1) we save all those cycles for each
semctl(_STAT) command, the idr_find can be expensive -- which some real
(customer) workloads actually poll on.
Note that this used to be the case, until commit 7ca7e564e0 ("ipc:
store ipcs into IDRs"). The cost is the extra idr_find when doing
RMIDs, but we simply go backwards, and should not take too many
iterations to find the new value.
[akpm@linux-foundation.org: coding-style fixes]
Link: http://lkml.kernel.org/r/20170831172049.14576-5-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Cc: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Patch series "sysvipc: ipc-key management improvements".
Here are a few improvements I spotted while eyeballing Guillaume's
rhashtable implementation for ipc keys. The first and fourth patches
are the interesting ones, the middle two are trivial.
This patch (of 4):
The next_id object-allocation functionality was introduced in commit
03f5956680 ("ipc: add sysctl to specify desired next object id").
Given that these new entries are _only_ exported under the
CONFIG_CHECKPOINT_RESTORE option, there is no point for the common case
to even know about ->next_id. As such rewrite ipc_buildid() such that
it can do away with the field as well as unnecessary branches when
adding a new identifier. The end result also better differentiates both
cases, so the code ends up being cleaner; albeit the small duplications
regarding the default case.
[akpm@linux-foundation.org: coding-style fixes]
Link: http://lkml.kernel.org/r/20170831172049.14576-2-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Cc: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
