public inbox for libc-alpha@sourceware.org
 help / color / mirror / Atom feed
From: Adhemerval Zanella <adhemerval.zanella@linaro.org>
To: libc-alpha@sourceware.org, Florian Weimer <fweimer@redhat.com>
Subject: Re: [PATCH v3 32/37] nptl: pthread_mutex_lock, pthread_mutex_unock single-threaded optimization
Date: Fri, 26 Mar 2021 15:00:53 -0300	[thread overview]
Message-ID: <b99fc610-7a3f-0f70-ec79-5f12e077fa3d@linaro.org> (raw)
In-Reply-To: <782ad0d9a371fa66bd54df07413f3d15fba0cf5a.1615914632.git.fweimer@redhat.com>



On 16/03/2021 14:31, Florian Weimer via Libc-alpha wrote:
> This is optimization is similar in spirit to the SINGLE_THREAD_P check
> in the malloc implementation.  Doing this in generic code allows us
> to prioritize those cases which are likely to occur in single-threaded
> programs (normal and recursive mutexes).

At first I though about suggesting to move it on generic lowlevellock.h,
but it is used quite only in some specific places and once libpthread is 
moved we can use libc_lock_lock (which uses pthread_mutex) instead.

The only issue with the approach of moving this optimization to 
pthread_mutex_lock (instead of custom lock as lll_lock) is to use a
slight more memory due pthread_mutex_t object, but I think the sparse
usage within glibc should be ok.

It would be good if we disentangle and consolidate both lowlevellock.h 
and libc-lock.h so we have only header and definitions.

LGTM, thanks.

Reviewed-by: Adhemerval Zanella  <adhemerval.zanella@linaro.org>

> ---
>  nptl/pthread_mutex_cond_lock.c |  1 +
>  nptl/pthread_mutex_lock.c      | 25 ++++++++++++++++++++++---
>  nptl/pthread_mutex_unlock.c    | 17 ++++++++++++++++-
>  3 files changed, 39 insertions(+), 4 deletions(-)
> 
> diff --git a/nptl/pthread_mutex_cond_lock.c b/nptl/pthread_mutex_cond_lock.c
> index 2f0771302f..3386bd689b 100644
> --- a/nptl/pthread_mutex_cond_lock.c
> +++ b/nptl/pthread_mutex_cond_lock.c
> @@ -2,6 +2,7 @@
>  
>  #define LLL_MUTEX_LOCK(mutex) \
>    lll_cond_lock ((mutex)->__data.__lock, PTHREAD_MUTEX_PSHARED (mutex))
> +#define LLL_MUTEX_LOCK_OPTIMIZED(mutex) LLL_MUTEX_LOCK (mutex)
>  
>  /* Not actually elided so far. Needed? */
>  #define LLL_MUTEX_LOCK_ELISION(mutex)  \
> diff --git a/nptl/pthread_mutex_lock.c b/nptl/pthread_mutex_lock.c
> index f0de7b7fd6..8649a92ffb 100644
> --- a/nptl/pthread_mutex_lock.c
> +++ b/nptl/pthread_mutex_lock.c
> @@ -30,8 +30,27 @@
>  /* Some of the following definitions differ when pthread_mutex_cond_lock.c
>     includes this file.  */
>  #ifndef LLL_MUTEX_LOCK
> -# define LLL_MUTEX_LOCK(mutex) \
> +/* lll_lock with single-thread optimization.  */
> +static inline void
> +lll_mutex_lock_optimized (pthread_mutex_t *mutex)
> +{
> +  /* The single-threaded optimization is only valid for private
> +     mutexes.  For process-shared mutexes, the mutex could be in a
> +     shared mapping, so synchronization with another process is needed
> +     even without any threads.  If the lock is already marked as
> +     acquired, POSIX requires that pthread_mutex_lock deadlocks for
> +     normal mutexes, so skip the optimization in that case as
> +     well.  */
> +  int private = PTHREAD_MUTEX_PSHARED (mutex);
> +  if (private == LLL_PRIVATE && SINGLE_THREAD_P && mutex->__data.__lock == 0)
> +    mutex->__data.__lock = 1;
> +  else
> +    lll_lock (mutex->__data.__lock, private);
> +}
> +
> +# define LLL_MUTEX_LOCK(mutex)						\
>    lll_lock ((mutex)->__data.__lock, PTHREAD_MUTEX_PSHARED (mutex))

Ok.

> +# define LLL_MUTEX_LOCK_OPTIMIZED(mutex) lll_mutex_lock_optimized (mutex)
>  # define LLL_MUTEX_TRYLOCK(mutex) \
>    lll_trylock ((mutex)->__data.__lock)
>  # define LLL_ROBUST_MUTEX_LOCK_MODIFIER 0
> @@ -64,7 +83,7 @@ __pthread_mutex_lock (pthread_mutex_t *mutex)
>        FORCE_ELISION (mutex, goto elision);
>      simple:
>        /* Normal mutex.  */
> -      LLL_MUTEX_LOCK (mutex);
> +      LLL_MUTEX_LOCK_OPTIMIZED (mutex);
>        assert (mutex->__data.__owner == 0);
>      }
>  #if ENABLE_ELISION_SUPPORT

Ok.

> @@ -99,7 +118,7 @@ __pthread_mutex_lock (pthread_mutex_t *mutex)
>  	}
>  
>        /* We have to get the mutex.  */
> -      LLL_MUTEX_LOCK (mutex);
> +      LLL_MUTEX_LOCK_OPTIMIZED (mutex);
>  
>        assert (mutex->__data.__owner == 0);
>        mutex->__data.__count = 1;

Ok.

> diff --git a/nptl/pthread_mutex_unlock.c b/nptl/pthread_mutex_unlock.c
> index 3b5ccdacf9..655093ee9a 100644
> --- a/nptl/pthread_mutex_unlock.c
> +++ b/nptl/pthread_mutex_unlock.c
> @@ -28,6 +28,21 @@ static int
>  __pthread_mutex_unlock_full (pthread_mutex_t *mutex, int decr)
>       __attribute_noinline__;
>  
> +/* lll_lock with single-thread optimization.  */
> +static inline void
> +lll_mutex_unlock_optimized (pthread_mutex_t *mutex)
> +{
> +  /* The single-threaded optimization is only valid for private
> +     mutexes.  For process-shared mutexes, the mutex could be in a
> +     shared mapping, so synchronization with another process is needed
> +     even without any threads.  */
> +  int private = PTHREAD_MUTEX_PSHARED (mutex);
> +  if (private == LLL_PRIVATE && SINGLE_THREAD_P)
> +    mutex->__data.__lock = 0;
> +  else
> +    lll_unlock (mutex->__data.__lock, private);
> +}
> +
>  int
>  attribute_hidden
>  __pthread_mutex_unlock_usercnt (pthread_mutex_t *mutex, int decr)
> @@ -51,7 +66,7 @@ __pthread_mutex_unlock_usercnt (pthread_mutex_t *mutex, int decr)
>  	--mutex->__data.__nusers;
>  
>        /* Unlock.  */
> -      lll_unlock (mutex->__data.__lock, PTHREAD_MUTEX_PSHARED (mutex));
> +      lll_mutex_unlock_optimized (mutex);
>  
>        LIBC_PROBE (mutex_release, 1, mutex);
>  
> 

Ok.

  reply	other threads:[~2021-03-26 18:00 UTC|newest]

Thread overview: 99+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-03-16 17:27 [PATCH v3 00/37] libpthread removal: NPTL forwarders are gone Florian Weimer
2021-03-16 17:27 ` [PATCH v3 01/37] nptl: Move pthread_mutex_consistent into libc Florian Weimer
2021-03-17 11:36   ` Adhemerval Zanella
2021-03-16 17:27 ` [PATCH v3 02/37] nptl: Move __pthread_cleanup_routine " Florian Weimer
2021-03-16 17:27 ` [PATCH v3 03/37] nptl: Move legacy unwinding implementation " Florian Weimer
2021-03-16 17:27 ` [PATCH v3 04/37] nptl: Move legacy cancelation handling into libc as compat symbols Florian Weimer
2021-03-16 17:27 ` [PATCH v3 05/37] nptl: Remove longjmp, siglongjmp from libpthread Florian Weimer
2021-03-17 11:38   ` Adhemerval Zanella
2021-03-16 17:28 ` [PATCH v3 06/37] x86: Restore compile-time check for shadow stack pointer in longjmp Florian Weimer
2021-03-16 17:28 ` [PATCH v3 07/37] nptl: Move __pthread_cleanup_upto into libc Florian Weimer
2021-03-16 17:28 ` [PATCH v3 08/37] nptl: Move pthread_once and __pthread_once " Florian Weimer
2021-03-17 13:30   ` Adhemerval Zanella
2021-03-17 13:37     ` Adhemerval Zanella
2021-03-17 14:45       ` Florian Weimer
2021-03-17 16:39         ` Adhemerval Zanella
2021-03-17 16:56           ` Florian Weimer
2021-03-17 17:22             ` Adhemerval Zanella
2021-03-17 17:43               ` Florian Weimer
2021-03-17 19:09                 ` Adhemerval Zanella
2021-03-16 17:29 ` [PATCH v3 09/37] nptl: Move __pthread_unwind_next " Florian Weimer
2021-03-17 19:42   ` Adhemerval Zanella
2021-03-17 19:54     ` Florian Weimer
2021-03-17 20:16       ` Adhemerval Zanella
2021-03-17 20:33         ` Florian Weimer
2021-03-17 20:44           ` Adhemerval Zanella
2021-03-16 17:29 ` [PATCH v3 10/37] csu: Move calling main out of __libc_start_main_impl Florian Weimer
2021-03-17 20:45   ` Adhemerval Zanella
2021-03-16 17:29 ` [PATCH v3 11/37] nptl: Move internal __nptl_nthreads variable into libc Florian Weimer
2021-03-18 12:42   ` Adhemerval Zanella
2021-03-16 17:29 ` [PATCH v3 12/37] nptl_db: Introduce DB_MAIN_ARRAY_VARIABLE Florian Weimer
2021-03-18 12:43   ` Adhemerval Zanella
2021-03-16 17:29 ` [PATCH v3 13/37] nptl: Move __pthread_keys global variable into libc Florian Weimer
2021-03-18 12:44   ` Adhemerval Zanella
2021-03-16 17:29 ` [PATCH v3 14/37] nptl: Move __nptl_deallocate_tsd " Florian Weimer
2021-03-18 12:46   ` Adhemerval Zanella
2021-03-18 17:16     ` Florian Weimer
2021-03-18 17:54       ` Adhemerval Zanella
2021-03-16 17:29 ` [PATCH v3 15/37] nptl: Move pthread_exit " Florian Weimer
2021-03-18 12:49   ` Adhemerval Zanella
2021-03-16 17:29 ` [PATCH v3 16/37] nptl: Move pthread_setcancelstate " Florian Weimer
2021-03-18 12:52   ` Adhemerval Zanella
2021-03-16 17:29 ` [PATCH v3 17/37] nptl: Move pthread_setcanceltype " Florian Weimer
2021-03-18 12:53   ` Adhemerval Zanella
2021-03-16 17:29 ` [PATCH v3 18/37] nptl: Invoke the set_robust_list system call directly in fork Florian Weimer
2021-03-18 12:54   ` Adhemerval Zanella
2021-03-16 17:30 ` [PATCH v3 19/37] dlfcn: Failures after dlmopen should not terminate process [BZ #24772] Florian Weimer
2021-03-19 19:56   ` Adhemerval Zanella
2021-03-27 16:57     ` Florian Weimer
2021-03-16 17:30 ` [PATCH v3 20/37] dlfcn: dlerror needs to call free from the base namespace [BZ #24773] Florian Weimer
2021-03-23 14:47   ` Adhemerval Zanella
2021-03-16 17:30 ` [PATCH v3 21/37] Remove pthread_key_create-related internals from libc-lock.h Florian Weimer
2021-03-23 16:39   ` Adhemerval Zanella
2021-03-16 17:30 ` [PATCH v3 22/37] elf: Introduce __tls_init_tp for second-phase TCB initialization Florian Weimer
2021-03-23 18:25   ` Adhemerval Zanella
2021-03-16 17:30 ` [PATCH v3 23/37] nptl: Move part of TCB initialization from libpthread to __tls_init_tp Florian Weimer
2021-03-24 13:56   ` Adhemerval Zanella
2021-03-27 17:19     ` Florian Weimer
2021-03-16 17:30 ` [PATCH v3 24/37] nptl: Move pthread_key_create, __pthread_key_create into libc Florian Weimer
2021-03-24 14:09   ` Adhemerval Zanella
2021-03-24 14:32     ` Florian Weimer
2021-03-24 14:42       ` Adhemerval Zanella
2021-03-24 15:08         ` Florian Weimer
2021-03-24 15:46           ` Adhemerval Zanella
2021-03-16 17:30 ` [PATCH v3 25/37] nptl: Move pthread_getspecific, __pthread_getspecific " Florian Weimer
2021-03-24 14:12   ` Adhemerval Zanella
2021-03-24 14:38     ` Florian Weimer
2021-03-24 14:43       ` Adhemerval Zanella
2021-03-16 17:30 ` [PATCH v3 26/37] nptl: Move pthread_setspecific, __pthread_setspecific " Florian Weimer
2021-03-24 14:26   ` Adhemerval Zanella
2021-03-16 17:30 ` [PATCH v3 27/37] nptl: Move pthread_key_delete " Florian Weimer
2021-03-24 14:45   ` Adhemerval Zanella
2021-03-16 17:31 ` [PATCH v3 28/37] nptl: Move rwlock functions with forwarders " Florian Weimer
2021-03-25 19:52   ` Adhemerval Zanella
2021-03-27 21:41     ` Florian Weimer
2021-03-16 17:31 ` [PATCH v3 29/37] nptl: Move the internal thread priority protection symbols " Florian Weimer
2021-03-25 20:21   ` Adhemerval Zanella
2021-03-16 17:31 ` [PATCH v3 30/37] pthread: Introduce __pthread_early_init Florian Weimer
2021-03-25 20:22   ` Adhemerval Zanella
2021-03-16 17:31 ` [PATCH v3 31/37] nptl: Move internal symbol __mutex_aconf into libc Florian Weimer
2021-03-25 20:24   ` Adhemerval Zanella
2021-03-16 17:31 ` [PATCH v3 32/37] nptl: pthread_mutex_lock, pthread_mutex_unock single-threaded optimization Florian Weimer
2021-03-26 18:00   ` Adhemerval Zanella [this message]
2021-03-16 17:31 ` [PATCH v3 33/37] x86: Remove low-level lock optimization Florian Weimer
2021-03-25 20:30   ` Adhemerval Zanella
2021-03-16 17:31 ` [PATCH v3 34/37] nptl: Move core mutex functions into libc Florian Weimer
2021-03-25 20:46   ` Adhemerval Zanella
2021-03-16 17:31 ` [PATCH v3 35/37] nptl: Move core condition variable " Florian Weimer
2021-03-26 17:14   ` Adhemerval Zanella
2021-03-16 17:31 ` [PATCH v3 36/37] nptl: Move setxid broadcast implementation " Florian Weimer
2021-03-26 18:15   ` Adhemerval Zanella
2021-04-06 18:41     ` Florian Weimer
2021-04-06 18:54       ` Adhemerval Zanella
2021-04-06 19:23         ` Florian Weimer
2021-04-06 19:40           ` Adhemerval Zanella
2021-03-16 17:31 ` [PATCH v3 37/37] nptl: Remove remnants of the libc/libpthread forwarder interface Florian Weimer
2021-03-26 18:19   ` Adhemerval Zanella
2021-03-18 22:06 ` [PATCH v3 00/37] libpthread removal: NPTL forwarders are gone Florian Weimer
2021-03-26 18:25 ` Adhemerval Zanella
2021-03-31 10:18   ` Florian Weimer

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=b99fc610-7a3f-0f70-ec79-5f12e077fa3d@linaro.org \
    --to=adhemerval.zanella@linaro.org \
    --cc=fweimer@redhat.com \
    --cc=libc-alpha@sourceware.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).