public inbox for libc-alpha@sourceware.org
 help / color / mirror / Atom feed
From: Florian Weimer <fweimer@redhat.com>
To: libc-alpha@sourceware.org
Subject: [PATCH v4 33/37] x86: Remove low-level lock optimization
Date: Fri, 16 Apr 2021 11:24:04 +0200	[thread overview]
Message-ID: <24b83fba24f43c3a200addb13b6e6c6aa2f28ff3.1618564630.git.fweimer@redhat.com> (raw)
In-Reply-To: <cover.1618564630.git.fweimer@redhat.com>

The current approach is to do this optimizations at a higher level,
in generic code, so that single-threaded cases can be specifically
targeted.

Furthermore, using IS_IN (libc) as a compile-time indicator that
all locks are private is no longer correct once process-shared lock
implementations are moved into libc.

The generic <lowlevellock.h> is not compatible with assembler code
(obviously), so it's necessary to remove two long-unused #includes.

Reviewed-by: Adhemerval Zanella  <adhemerval.zanella@linaro.org>
---
 sysdeps/i386/nptl/pthread_spin_lock.S      |  1 -
 sysdeps/unix/sysv/linux/x86/lowlevellock.h | 87 ----------------------
 sysdeps/x86_64/nptl/pthread_spin_lock.S    |  1 -
 3 files changed, 89 deletions(-)
 delete mode 100644 sysdeps/unix/sysv/linux/x86/lowlevellock.h

diff --git a/sysdeps/i386/nptl/pthread_spin_lock.S b/sysdeps/i386/nptl/pthread_spin_lock.S
index eec157fde7..985b236aae 100644
--- a/sysdeps/i386/nptl/pthread_spin_lock.S
+++ b/sysdeps/i386/nptl/pthread_spin_lock.S
@@ -16,7 +16,6 @@
    <https://www.gnu.org/licenses/>.  */
 
 #include <sysdep.h>
-#include <lowlevellock.h>
 
 ENTRY (pthread_spin_lock)
 	mov	4(%esp), %eax
diff --git a/sysdeps/unix/sysv/linux/x86/lowlevellock.h b/sysdeps/unix/sysv/linux/x86/lowlevellock.h
deleted file mode 100644
index 7205bcadfd..0000000000
--- a/sysdeps/unix/sysv/linux/x86/lowlevellock.h
+++ /dev/null
@@ -1,87 +0,0 @@
-/* Low-level lock implementation, x86 version.
-   Copyright (C) 2019-2021 Free Software Foundation, Inc.
-   This file is part of the GNU C Library.
-
-   The GNU C Library is free software; you can redistribute it and/or
-   modify it under the terms of the GNU Lesser General Public
-   License as published by the Free Software Foundation; either
-   version 2.1 of the License, or (at your option) any later version.
-
-   The GNU C Library is distributed in the hope that it will be useful,
-   but WITHOUT ANY WARRANTY; without even the implied warranty of
-   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
-   Lesser General Public License for more details.
-
-   You should have received a copy of the GNU Lesser General Public
-   License along with the GNU C Library; if not, see
-   <https://www.gnu.org/licenses/>.  */
-
-#ifndef _X86_64_LOWLEVELLOCK_H
-#define _X86_64_LOWLEVELLOCK_H	1
-
-#ifndef __ASSEMBLER__
-#include <sysdeps/nptl/lowlevellock.h>
-#include <single-thread.h>
-
-/* The lll_trylock, lll_lock, and lll_unlock implements a single-thread
-   optimization using the cmpxchgl instruction.  It checks if the process
-   is single thread and avoid a more expensive atomic instruction.  */
-
-/* The single-thread optimization only works for libc itself, we need
-   atomicity for libpthread in case of shared futexes.  */
-#if !IS_IN(libc)
-# define is_single_thread 0
-#else
-# define is_single_thread SINGLE_THREAD_P
-#endif
-
-/* In the __lllc_as we simply return the value in %eax after the cmpxchg
-   instruction.  In case the operation succeeded this value is zero.  In case
-   the operation failed, the cmpxchg instruction has loaded the current value
-   of the memory work which is guaranteed to be nonzero.  */
-static inline int
-__attribute__ ((always_inline))
-__lll_cas_lock (int *futex)
-{
-  int ret;
-  asm volatile ("cmpxchgl %2, %1"
-		: "=a" (ret), "=m" (*futex)
-		: "r" (1), "m" (*futex), "0" (0)
-		: "memory");
-  return ret;
-}
-
-#undef lll_trylock
-#define lll_trylock(lock)						     \
-  ({									     \
-    int __ret;								     \
-    if (is_single_thread)						     \
-      __ret = __lll_cas_lock (&(lock));					     \
-    else								     \
-      __ret = __lll_trylock (&(lock));					     \
-    __ret;								     \
-  })
-
-#undef lll_lock
-#define lll_lock(lock, private)						     \
-  ((void)								     \
-   ({									     \
-     if (is_single_thread)						     \
-       __lll_cas_lock (&(lock));					     \
-     else								     \
-       __lll_lock (&(lock), private);					     \
-   }))
-
-#undef lll_unlock
-#define lll_unlock(lock, private)					     \
-  ((void)								     \
-   ({									     \
-     if (is_single_thread)						     \
-       (lock)--;							     \
-     else								     \
-       __lll_unlock (&(lock), private);					     \
-   }))
-
-#endif  /* !__ASSEMBLER__ */
-
-#endif	/* lowlevellock.h */
diff --git a/sysdeps/x86_64/nptl/pthread_spin_lock.S b/sysdeps/x86_64/nptl/pthread_spin_lock.S
index 66cbbc5252..ff5e7525ca 100644
--- a/sysdeps/x86_64/nptl/pthread_spin_lock.S
+++ b/sysdeps/x86_64/nptl/pthread_spin_lock.S
@@ -15,7 +15,6 @@
    License along with the GNU C Library; if not, see
    <https://www.gnu.org/licenses/>.  */
 
-#include <lowlevellock.h>
 #include <sysdep.h>
 
 ENTRY(pthread_spin_lock)
-- 
2.30.2



  parent reply	other threads:[~2021-04-16  9:23 UTC|newest]

Thread overview: 43+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-04-16  9:20 [PATCH v4 00/37] libpthread: Remove NPTL forwarders Florian Weimer
2021-04-16  9:20 ` [PATCH v4 01/37] nptl: Move pthread_mutex_consistent into libc Florian Weimer
2021-04-16  9:20 ` [PATCH v4 02/37] nptl: Move __pthread_cleanup_routine " Florian Weimer
2021-04-16  9:20 ` [PATCH v4 03/37] nptl: Move legacy unwinding implementation " Florian Weimer
2021-04-16  9:20 ` [PATCH v4 04/37] nptl: Move legacy cancelation handling into libc as compat symbols Florian Weimer
2021-04-16  9:20 ` [PATCH v4 05/37] nptl: Remove longjmp, siglongjmp from libpthread Florian Weimer
2021-04-16  9:20 ` [PATCH v4 06/37] x86: Restore compile-time check for shadow stack pointer in longjmp Florian Weimer
2021-04-16  9:20 ` [PATCH v4 07/37] nptl: Move __pthread_cleanup_upto into libc Florian Weimer
2021-04-16  9:20 ` [PATCH v4 08/37] nptl: Move pthread_once and __pthread_once " Florian Weimer
2021-04-16  9:21 ` [PATCH v4 09/37] nptl: Move __pthread_unwind_next " Florian Weimer
2021-04-16  9:21 ` [PATCH v4 10/37] csu: Move calling main out of __libc_start_main_impl Florian Weimer
2021-04-16  9:21 ` [PATCH v4 11/37] nptl: Move internal __nptl_nthreads variable into libc Florian Weimer
2021-04-16  9:21 ` [PATCH v4 12/37] nptl_db: Introduce DB_MAIN_ARRAY_VARIABLE Florian Weimer
2021-04-16  9:21 ` [PATCH v4 13/37] nptl: Move __pthread_keys global variable into libc Florian Weimer
2021-04-16  9:21 ` [PATCH v4 14/37] nptl: Move __nptl_deallocate_tsd " Florian Weimer
2021-04-16  9:21 ` [PATCH v4 15/37] nptl: Move pthread_exit " Florian Weimer
2021-04-16  9:22 ` [PATCH v4 16/37] nptl: Move pthread_setcancelstate " Florian Weimer
2021-04-16  9:22 ` [PATCH v4 17/37] nptl: Move pthread_setcanceltype " Florian Weimer
2021-04-16  9:22 ` [PATCH v4 18/37] nptl: Invoke the set_robust_list system call directly in fork Florian Weimer
2021-04-16  9:22 ` [PATCH v4 19/37] dlfcn: Failures after dlmopen should not terminate process [BZ #24772] Florian Weimer
2021-04-16  9:22 ` [PATCH v4 20/37] dlfcn: dlerror needs to call free from the base namespace [BZ #24773] Florian Weimer
2021-04-16  9:22 ` [PATCH v4 21/37] Remove pthread_key_create-related internals from libc-lock.h Florian Weimer
2021-04-16  9:22 ` [PATCH v4 22/37] elf: Introduce __tls_init_tp for second-phase TCB initialization Florian Weimer
2021-04-16  9:22 ` [PATCH v4 23/37] nptl: Move part of TCB initialization from libpthread to __tls_init_tp Florian Weimer
2021-04-16  9:22 ` [PATCH v4 24/37] nptl: Move pthread_key_create, __pthread_key_create into libc Florian Weimer
2021-04-16  9:23 ` [PATCH v4 25/37] nptl: Move pthread_getspecific, __pthread_getspecific " Florian Weimer
2021-04-16  9:23 ` [PATCH v4 26/37] nptl: Move pthread_setspecific, __pthread_setspecific " Florian Weimer
2021-04-16  9:23 ` [PATCH v4 27/37] nptl: Move pthread_key_delete " Florian Weimer
2021-04-16  9:23 ` [PATCH v4 28/37] nptl: Move rwlock functions with forwarders " Florian Weimer
2021-04-16  9:23 ` [PATCH v4 29/37] nptl: Move the internal thread priority protection symbols " Florian Weimer
2021-04-16  9:23 ` [PATCH v4 30/37] pthread: Introduce __pthread_early_init Florian Weimer
2021-04-16  9:23 ` [PATCH v4 31/37] nptl: Move internal symbol __mutex_aconf into libc Florian Weimer
2021-04-16  9:24 ` [PATCH v4 32/37] nptl: pthread_mutex_lock, pthread_mutex_unock single-threaded optimization Florian Weimer
2021-04-16  9:24 ` Florian Weimer [this message]
2021-04-16  9:24 ` [PATCH v4 34/37] nptl: Move core mutex functions into libc Florian Weimer
2021-04-16  9:24 ` [PATCH v4 35/37] nptl: Move core condition variable " Florian Weimer
2021-04-16  9:24 ` [PATCH v4 36/37] nptl: Move setxid broadcast implementation " Florian Weimer
2021-04-16  9:24 ` [PATCH v4 37/37] nptl: Remove remnants of the libc/libpthread forwarder interface Florian Weimer
2021-04-21 11:47 ` [PATCH v4 00/37] libpthread: Remove NPTL forwarders Florian Weimer
2021-04-21 13:42   ` H.J. Lu
2021-04-22  7:35   ` Szabolcs Nagy
2021-04-22  7:58     ` Florian Weimer
2021-04-22  8:56       ` Szabolcs Nagy

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=24b83fba24f43c3a200addb13b6e6c6aa2f28ff3.1618564630.git.fweimer@redhat.com \
    --to=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).