Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754178AbdDJPeD (ORCPT ); Mon, 10 Apr 2017 11:34:03 -0400 Received: from mx2.suse.de ([195.135.220.15]:37784 "EHLO mx2.suse.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754151AbdDJPeB (ORCPT ); Mon, 10 Apr 2017 11:34:01 -0400 X-Amavis-Alert: BAD HEADER SECTION, Duplicate header field: "References" From: Jiri Slaby To: stable@vger.kernel.org Cc: linux-kernel@vger.kernel.org, Eric Dumazet , "David S . Miller" , Jiri Slaby Subject: [PATCH 3.12 029/142] net: net_enable_timestamp() can be called from irq contexts Date: Mon, 10 Apr 2017 17:31:50 +0200 Message-Id: <084d8e5416a7ee2c520317e2d687c2499fb8bdbb.1491838390.git.jslaby@suse.cz> X-Mailer: git-send-email 2.12.2 In-Reply-To: References: In-Reply-To: References: Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 2850 Lines: 98 From: Eric Dumazet 3.12-stable review patch. If anyone has any objections, please let me know. =============== [ Upstream commit 13baa00ad01bb3a9f893e3a08cbc2d072fc0c15d ] It is now very clear that silly TCP listeners might play with enabling/disabling timestamping while new children are added to their accept queue. Meaning net_enable_timestamp() can be called from BH context while current state of the static key is not enabled. Lets play safe and allow all contexts. The work queue is scheduled only under the problematic cases, which are the static key enable/disable transition, to not slow down critical paths. This extends and improves what we did in commit 5fa8bbda38c6 ("net: use a work queue to defer net_disable_timestamp() work") Fixes: b90e5794c5bd ("net: dont call jump_label_dec from irq context") Signed-off-by: Eric Dumazet Reported-by: Dmitry Vyukov Signed-off-by: David S. Miller Signed-off-by: Jiri Slaby --- net/core/dev.c | 35 +++++++++++++++++++++++++++++++---- 1 file changed, 31 insertions(+), 4 deletions(-) diff --git a/net/core/dev.c b/net/core/dev.c index 349ee899b3f0..a8574b4264cb 100644 --- a/net/core/dev.c +++ b/net/core/dev.c @@ -1595,27 +1595,54 @@ EXPORT_SYMBOL(call_netdevice_notifiers); static struct static_key netstamp_needed __read_mostly; #ifdef HAVE_JUMP_LABEL static atomic_t netstamp_needed_deferred; +static atomic_t netstamp_wanted; static void netstamp_clear(struct work_struct *work) { int deferred = atomic_xchg(&netstamp_needed_deferred, 0); + int wanted; - while (deferred--) - static_key_slow_dec(&netstamp_needed); + wanted = atomic_add_return(deferred, &netstamp_wanted); + if (wanted > 0) + static_key_enable(&netstamp_needed); + else + static_key_disable(&netstamp_needed); } static DECLARE_WORK(netstamp_work, netstamp_clear); #endif void net_enable_timestamp(void) { +#ifdef HAVE_JUMP_LABEL + int wanted; + + while (1) { + wanted = atomic_read(&netstamp_wanted); + if (wanted <= 0) + break; + if (atomic_cmpxchg(&netstamp_wanted, wanted, wanted + 1) == wanted) + return; + } + atomic_inc(&netstamp_needed_deferred); + schedule_work(&netstamp_work); +#else static_key_slow_inc(&netstamp_needed); +#endif } EXPORT_SYMBOL(net_enable_timestamp); void net_disable_timestamp(void) { #ifdef HAVE_JUMP_LABEL - /* net_disable_timestamp() can be called from non process context */ - atomic_inc(&netstamp_needed_deferred); + int wanted; + + while (1) { + wanted = atomic_read(&netstamp_wanted); + if (wanted <= 1) + break; + if (atomic_cmpxchg(&netstamp_wanted, wanted, wanted - 1) == wanted) + return; + } + atomic_dec(&netstamp_needed_deferred); schedule_work(&netstamp_work); #else static_key_slow_dec(&netstamp_needed); -- 2.12.2