linux-kernel - [PATCH RFC tip/core/rcu 0/15] Remove to-be-unneeded smp_read_barrier

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Date:   Mon, 9 Oct 2017 17:19:51 -0700
From:   "Paul E. McKenney" <paulmck@...ux.vnet.ibm.com>
To:     linux-kernel@...r.kernel.org
Cc:     dhowells@...hat.com, mark.rutland@....com,
        linux-arch@...r.kernel.org, viro@...iv.linux.org.uk, tj@...nel.org,
        cl@...ux.com, davem@...emloft.net, peterz@...radead.org,
        mingo@...hat.com, rostedt@...dmis.org, akpm@...ux-foundation.org,
        corbet@....net, james.l.morris@...cle.com,
        torvalds@...ux-foundation.org
Subject: [PATCH RFC tip/core/rcu 0/15] Remove to-be-unneeded
 smp_read_barrier_depends()

Hello!

Will Deacon has proposed adding smp_read_barrier_depends() to READ_ONCE(),
which would mean that quite a few instances of smp_read_barrier_depends()
would become redundant.  This series depends on Will's change and removes
those smp_read_barrier_depends(), while fixing a bug or two along the
way.  Some of these bugs are subtle, hence posting this sooner rather
than later.

This patch series depends on an in-progress series from Mark Rutland that
changes all calls to ACCESS_ONCE() to either READ_ONCE() or WRITE_ONCE(),
depending.  This patch series also depends on an upcoming patch from Will
Deacon that adds smp_read_barrier_depends() to Alpha's relaxed/release
RMW atomic operations.

As Will pointed out, once all of these patch series are in place, the
core kernel will no longer need any Alpha-specific code, with the sole
exception of READ_ONCE() and of course smp_read_barrier_depends() itself.

Pretty cool, huh?  ;-)

							Thanx, Paul

------------------------------------------------------------------------

 Documentation/memory-barriers.txt         |   11 ++++++-----
 arch/mn10300/kernel/mn10300-serial.c      |    7 +++++--
 drivers/net/ethernet/qlogic/qed/qed_spq.c |    4 +---
 fs/dcache.c                               |   10 +++-------
 include/linux/assoc_array_priv.h          |    5 +++--
 include/linux/percpu-refcount.h           |    6 +++---
 include/linux/rcupdate.h                  |   23 +++++++++++------------
 include/linux/rtnetlink.h                 |    3 +--
 include/linux/seqlock.h                   |    3 +--
 kernel/events/uprobes.c                   |   12 ++++++------
 kernel/locking/qspinlock.c                |   12 +++++-------
 kernel/tracepoint.c                       |    9 ++++-----
 lib/assoc_array.c                         |   20 ++------------------
 lib/percpu-refcount.c                     |    8 ++++----
 mm/ksm.c                                  |    9 +--------
 net/ipv4/netfilter/arp_tables.c           |    7 +------
 net/ipv4/netfilter/ip_tables.c            |    7 +------
 net/ipv6/netfilter/ip6_tables.c           |    7 +------
 security/keys/keyring.c                   |    7 -------
 19 files changed, 59 insertions(+), 111 deletions(-)