linux-toradex.git/ipc/sem.c, branch v4.14-rc5

Merge branch 'work.ipc' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs

2017-09-15T00:37:26+00:00

Pull ipc compat cleanup and 64-bit time_t from Al Viro:
 "IPC copyin/copyout sanitizing, including 64bit time_t work from Deepa
  Dinamani"

* 'work.ipc' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  utimes: Make utimes y2038 safe
  ipc: shm: Make shmid_kernel timestamps y2038 safe
  ipc: sem: Make sem_array timestamps y2038 safe
  ipc: msg: Make msg_queue timestamps y2038 safe
  ipc: mqueue: Replace timespec with timespec64
  ipc: Make sys_semtimedop() y2038 safe
  get rid of SYSVIPC_COMPAT on ia64
  semtimedop(): move compat to native
  shmat(2): move compat to native
  msgrcv(2), msgsnd(2): move compat to native
  ipc(2): move compat to native
  ipc: make use of compat ipc_perm helpers
  semctl(): move compat to native
  semctl(): separate all layout-dependent copyin/copyout
  msgctl(): move compat to native
  msgctl(): split the actual work from copyin/copyout
  ipc: move compat shmctl to native
  shmctl: split the work from copyin/copyout

ipc: optimize semget/shmget/msgget for lots of keys

2017-09-09T01:26:51+00:00

ipc_findkey() used to scan all objects to look for the wanted key.  This
is slow when using a high number of keys.  This change adds an rhashtable
of kern_ipc_perm objects in ipc_ids, so that one lookup cease to be O(n).

This change gives a 865% improvement of benchmark reaim.jobs_per_min on a
56 threads Intel(R) Xeon(R) CPU E5-2695 v3 @ 2.30GHz with 256G memory [1]

Other (more micro) benchmark results, by the author: On an i5 laptop, the
following loop executed right after a reboot took, without and with this
change:

    for (int i = 0, k=0x424242; i < KEYS; ++i)
        semget(k++, 1, IPC_CREAT | 0600);

                 total       total          max single  max single
   KEYS        without        with        call without   call with

      1            3.5         4.9   Âµs            3.5         4.9
     10            7.6         8.6   Âµs            3.7         4.7
     32           16.2        15.9   Âµs            4.3         5.3
    100           72.9        41.8   Âµs            3.7         4.7
   1000        5,630.0       502.0   Âµs             *           *
  10000    1,340,000.0     7,240.0   Âµs             *           *
  31900   17,600,000.0    22,200.0   Âµs             *           *

 *: unreliable measure: high variance

The duration for a lookup-only usage was obtained by the same loop once
the keys are present:

                 total       total          max single  max single
   KEYS        without        with        call without   call with

      1            2.1         2.5   Âµs            2.1         2.5
     10            4.5         4.8   Âµs            2.2         2.3
     32           13.0        10.8   Âµs            2.3         2.8
    100           82.9        25.1   Âµs             *          2.3
   1000        5,780.0       217.0   Âµs             *           *
  10000    1,470,000.0     2,520.0   Âµs             *           *
  31900   17,400,000.0     7,810.0   Âµs             *           *

Finally, executing each semget() in a new process gave, when still
summing only the durations of these syscalls:

creation:
                 total       total
   KEYS        without        with

      1            3.7         5.0   Âµs
     10           32.9        36.7   Âµs
     32          125.0       109.0   Âµs
    100          523.0       353.0   Âµs
   1000       20,300.0     3,280.0   Âµs
  10000    2,470,000.0    46,700.0   Âµs
  31900   27,800,000.0   219,000.0   Âµs

lookup-only:
                 total       total
   KEYS        without        with

      1            2.5         2.7   Âµs
     10           25.4        24.4   Âµs
     32          106.0        72.6   Âµs
    100          591.0       352.0   Âµs
   1000       22,400.0     2,250.0   Âµs
  10000    2,510,000.0    25,700.0   Âµs
  31900   28,200,000.0   115,000.0   Âµs

[1] http://lkml.kernel.org/r/20170814060507.GE23258@yexl-desktop

Link: http://lkml.kernel.org/r/20170815194954.ck32ta2z35yuzpwp@debix
Signed-off-by: Guillaume Knispel 
Reviewed-by: Marc Pardo 
Cc: Davidlohr Bueso 
Cc: Kees Cook 
Cc: Manfred Spraul 
Cc: Alexey Dobriyan 
Cc: "Eric W. Biederman" 
Cc: "Peter Zijlstra (Intel)" 
Cc: Ingo Molnar 
Cc: Sebastian Andrzej Siewior 
Cc: Serge Hallyn 
Cc: Andrey Vagin 
Cc: Guillaume Knispel 
Cc: Marc Pardo 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds

ipc/sem: play nicer with large nsops allocations

2017-09-09T01:26:51+00:00

Replacing semop()'s kmalloc for kvmalloc was originally proposed by
Manfred on the premise that it can be called for large (than order-1)
sizes.  For example, while Oracle recommends setting SEMOPM to a _minimum_
of 100, some distros[1] encourage the setting to be a factor of the amount
of db tasks (PROCESSES), which can get fishy for large systems (easily
going beyond 1000).

[1] An Example of Semaphore Settings
https://access.redhat.com/documentation/en-US/Red_Hat_Enterprise_Linux/5/html/Tuning_and_Optimizing_Red_Hat_Enterprise_Linux_for_Oracle_9i_and_10g_Databases/sect-Oracle_9i_and_10g_Tuning_Guide-Setting_Semaphores-An_Example_of_Semaphore_Settings.html

So let's just convert this to kvmalloc, just like the rest of the
allocations we do in ipc.  While the fallback vmalloc obviously involves
more overhead, this by far the uncommon path, and it's better for the user
than just erroring out with kmalloc.

Link: http://lkml.kernel.org/r/20170803184136.13855-2-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso 
Cc: Manfred Spraul 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds

ipc/sem: drop sem_checkid helper

2017-09-09T01:26:51+00:00

... 'tis not used.

Link: http://lkml.kernel.org/r/20170803184136.13855-1-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso 
Cc: Manfred Spraul 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds

ipc: convert sem_undo_list.refcnt from atomic_t to refcount_t

2017-09-09T01:26:51+00:00

refcount_t type and corresponding API should be used instead of atomic_t
when the variable is used as a reference counter.  This allows to avoid
accidental refcounter overflows that might lead to use-after-free
situations.

Link: http://lkml.kernel.org/r/1499417992-3238-3-git-send-email-elena.reshetova@intel.com
Signed-off-by: Elena Reshetova 
Signed-off-by: Hans Liljestrand 
Signed-off-by: Kees Cook 
Signed-off-by: David Windsor 
Cc: Peter Zijlstra 
Cc: Greg Kroah-Hartman 
Cc: "Eric W. Biederman" 
Cc: Ingo Molnar 
Cc: Alexey Dobriyan 
Cc: Serge Hallyn 
Cc: 
Cc: Davidlohr Bueso 
Cc: Manfred Spraul 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds

ipc: sem: Make sem_array timestamps y2038 safe

2017-09-04T00:24:29+00:00

time_t is not y2038 safe. Replace all uses of
time_t by y2038 safe time64_t.

Similarly, replace the calls to get_seconds() with
y2038 safe ktime_get_real_seconds().
Note that this preserves fast access on 64 bit systems,
but 32 bit systems need sequence counters.

The syscall interface themselves are not changed as part of
the patch. They will be part of a different series.

Signed-off-by: Deepa Dinamani 
Reviewed-by: Arnd Bergmann 
Signed-off-by: Al Viro

ipc: Make sys_semtimedop() y2038 safe

2017-09-04T00:21:23+00:00

struct timespec is not y2038 safe on 32 bit machines.
Replace timespec with y2038 safe struct timespec64.

Note that the patch only changes the internals without
modifying the syscall interface. This will be part
of a separate series.

Signed-off-by: Deepa Dinamani 
Reviewed-by: Arnd Bergmann 
Signed-off-by: Al Viro

Merge branch 'for-mingo' of git://git.kernel.org/pub/scm/linux/kernel/git/paulmck/linux-rcu into core/rcu

2017-08-21T07:45:19+00:00

Pull RCU updates from Paul E. McKenney:

 - Removal of spin_unlock_wait()
 - SRCU updates
 - Torture-test updates
 - Documentation updates
 - Miscellaneous fixes
 - CPU-hotplug fixes
 - Miscellaneous non-RCU fixes

Signed-off-by: Ingo Molnar

ipc: Replace spin_unlock_wait() with lock/unlock pair

2017-08-17T15:08:57+00:00

There is no agreed-upon definition of spin_unlock_wait()'s semantics,
and it appears that all callers could do just as well with a lock/unlock
pair.  This commit therefore replaces the spin_unlock_wait() call in
exit_sem() with spin_lock() followed immediately by spin_unlock().
This should be safe from a performance perspective because exit_sem()
is rarely invoked in production.

Signed-off-by: Paul E. McKenney 
Cc: Andrew Morton 
Cc: Davidlohr Bueso 
Cc: Will Deacon 
Cc: Peter Zijlstra 
Cc: Alan Stern 
Cc: Andrea Parri 
Cc: Linus Torvalds 
Acked-by: Manfred Spraul

ipc: add missing container_of()s for randstruct

2017-08-03T00:16:12+00:00

When building with the randstruct gcc plugin, the layout of the IPC
structs will be randomized, which requires any sub-structure accesses to
use container_of().  The proc display handlers were missing the needed
container_of()s since the iterator is passing in the top-level struct
kern_ipc_perm.

This would lead to crashes when running the "lsipc" program after the
system had IPC registered (e.g. after starting up Gnome):

  general protection fault: 0000 [#1] PREEMPT SMP
  ...
  RIP: 0010:shm_add_rss_swap.isra.1+0x13/0xa0
  ...
  Call Trace:
    sysvipc_shm_proc_show+0x5e/0x150
    sysvipc_proc_show+0x1a/0x30
    seq_read+0x2e9/0x3f0
  ...

Link: http://lkml.kernel.org/r/20170730205950.GA55841@beast
Fixes: 3859a271a003 ("randstruct: Mark various structs for randomization")
Signed-off-by: Kees Cook 
Reported-by: Dominik Brodowski 
Acked-by: Davidlohr Bueso 
Acked-by: Manfred Spraul 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds