From: Paul Gortmaker <paul.gortmaker@windriver.com>
To: <stable@vger.kernel.org>, <linux-kernel@vger.kernel.org>
Cc: Robert Richter <robert.richter@amd.com>,
Carl Love <carll@us.ibm.com>,
Paul Gortmaker <paul.gortmaker@windriver.com>
Subject: [v2.6.34-stable 47/77] oprofile: Fix locking dependency in sync_start()
Date: Tue, 8 Jan 2013 18:35:26 -0500 [thread overview]
Message-ID: <1357688156-25387-48-git-send-email-paul.gortmaker@windriver.com> (raw)
In-Reply-To: <1357688156-25387-1-git-send-email-paul.gortmaker@windriver.com>
From: Robert Richter <robert.richter@amd.com>
-------------------
This is a commit scheduled for the next v2.6.34 longterm release.
http://git.kernel.org/?p=linux/kernel/git/paulg/longterm-queue-2.6.34.git
If you see a problem with using this for longterm, please comment.
-------------------
commit 130c5ce716c9bfd1c2a2ec840a746eb7ff9ce1e6 upstream.
This fixes the A->B/B->A locking dependency, see the warning below.
The function task_exit_notify() is called with (task_exit_notifier)
.rwsem set and then calls sync_buffer() which locks buffer_mutex. In
sync_start() the buffer_mutex was set to prevent notifier functions to
be started before sync_start() is finished. But when registering the
notifier, (task_exit_notifier).rwsem is locked too, but now in
different order than in sync_buffer(). In theory this causes a locking
dependency, what does not occur in practice since task_exit_notify()
is always called after the notifier is registered which means the lock
is already released.
However, after checking the notifier functions it turned out the
buffer_mutex in sync_start() is unnecessary. This is because
sync_buffer() may be called from the notifiers even if sync_start()
did not finish yet, the buffers are already allocated but empty. No
need to protect this with the mutex.
So we fix this theoretical locking dependency by removing buffer_mutex
in sync_start(). This is similar to the implementation before commit:
750d857 oprofile: fix crash when accessing freed task structs
which introduced the locking dependency.
Lockdep warning:
oprofiled/4447 is trying to acquire lock:
(buffer_mutex){+.+...}, at: [<ffffffffa0000e55>] sync_buffer+0x31/0x3ec [oprofile]
but task is already holding lock:
((task_exit_notifier).rwsem){++++..}, at: [<ffffffff81058026>] __blocking_notifier_call_chain+0x39/0x67
which lock already depends on the new lock.
the existing dependency chain (in reverse order) is:
-> #1 ((task_exit_notifier).rwsem){++++..}:
[<ffffffff8106557f>] lock_acquire+0xf8/0x11e
[<ffffffff81463a2b>] down_write+0x44/0x67
[<ffffffff810581c0>] blocking_notifier_chain_register+0x52/0x8b
[<ffffffff8105a6ac>] profile_event_register+0x2d/0x2f
[<ffffffffa00013c1>] sync_start+0x47/0xc6 [oprofile]
[<ffffffffa00001bb>] oprofile_setup+0x60/0xa5 [oprofile]
[<ffffffffa00014e3>] event_buffer_open+0x59/0x8c [oprofile]
[<ffffffff810cd3b9>] __dentry_open+0x1eb/0x308
[<ffffffff810cd59d>] nameidata_to_filp+0x60/0x67
[<ffffffff810daad6>] do_last+0x5be/0x6b2
[<ffffffff810dbc33>] path_openat+0xc7/0x360
[<ffffffff810dbfc5>] do_filp_open+0x3d/0x8c
[<ffffffff810ccfd2>] do_sys_open+0x110/0x1a9
[<ffffffff810cd09e>] sys_open+0x20/0x22
[<ffffffff8146ad4b>] system_call_fastpath+0x16/0x1b
-> #0 (buffer_mutex){+.+...}:
[<ffffffff81064dfb>] __lock_acquire+0x1085/0x1711
[<ffffffff8106557f>] lock_acquire+0xf8/0x11e
[<ffffffff814634f0>] mutex_lock_nested+0x63/0x309
[<ffffffffa0000e55>] sync_buffer+0x31/0x3ec [oprofile]
[<ffffffffa0001226>] task_exit_notify+0x16/0x1a [oprofile]
[<ffffffff81467b96>] notifier_call_chain+0x37/0x63
[<ffffffff8105803d>] __blocking_notifier_call_chain+0x50/0x67
[<ffffffff81058068>] blocking_notifier_call_chain+0x14/0x16
[<ffffffff8105a718>] profile_task_exit+0x1a/0x1c
[<ffffffff81039e8f>] do_exit+0x2a/0x6fc
[<ffffffff8103a5e4>] do_group_exit+0x83/0xae
[<ffffffff8103a626>] sys_exit_group+0x17/0x1b
[<ffffffff8146ad4b>] system_call_fastpath+0x16/0x1b
other info that might help us debug this:
1 lock held by oprofiled/4447:
#0: ((task_exit_notifier).rwsem){++++..}, at: [<ffffffff81058026>] __blocking_notifier_call_chain+0x39/0x67
stack backtrace:
Pid: 4447, comm: oprofiled Not tainted 2.6.39-00007-gcf4d8d4 #10
Call Trace:
[<ffffffff81063193>] print_circular_bug+0xae/0xbc
[<ffffffff81064dfb>] __lock_acquire+0x1085/0x1711
[<ffffffffa0000e55>] ? sync_buffer+0x31/0x3ec [oprofile]
[<ffffffff8106557f>] lock_acquire+0xf8/0x11e
[<ffffffffa0000e55>] ? sync_buffer+0x31/0x3ec [oprofile]
[<ffffffff81062627>] ? mark_lock+0x42f/0x552
[<ffffffffa0000e55>] ? sync_buffer+0x31/0x3ec [oprofile]
[<ffffffff814634f0>] mutex_lock_nested+0x63/0x309
[<ffffffffa0000e55>] ? sync_buffer+0x31/0x3ec [oprofile]
[<ffffffffa0000e55>] sync_buffer+0x31/0x3ec [oprofile]
[<ffffffff81058026>] ? __blocking_notifier_call_chain+0x39/0x67
[<ffffffff81058026>] ? __blocking_notifier_call_chain+0x39/0x67
[<ffffffffa0001226>] task_exit_notify+0x16/0x1a [oprofile]
[<ffffffff81467b96>] notifier_call_chain+0x37/0x63
[<ffffffff8105803d>] __blocking_notifier_call_chain+0x50/0x67
[<ffffffff81058068>] blocking_notifier_call_chain+0x14/0x16
[<ffffffff8105a718>] profile_task_exit+0x1a/0x1c
[<ffffffff81039e8f>] do_exit+0x2a/0x6fc
[<ffffffff81465031>] ? retint_swapgs+0xe/0x13
[<ffffffff8103a5e4>] do_group_exit+0x83/0xae
[<ffffffff8103a626>] sys_exit_group+0x17/0x1b
[<ffffffff8146ad4b>] system_call_fastpath+0x16/0x1b
Reported-by: Marcin Slusarz <marcin.slusarz@gmail.com>
Cc: Carl Love <carll@us.ibm.com>
Signed-off-by: Robert Richter <robert.richter@amd.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
---
drivers/oprofile/buffer_sync.c | 8 ++------
1 file changed, 2 insertions(+), 6 deletions(-)
diff --git a/drivers/oprofile/buffer_sync.c b/drivers/oprofile/buffer_sync.c
index e353ebf..5830bdc 100644
--- a/drivers/oprofile/buffer_sync.c
+++ b/drivers/oprofile/buffer_sync.c
@@ -155,8 +155,6 @@ int sync_start(void)
if (!zalloc_cpumask_var(&marked_cpus, GFP_KERNEL))
return -ENOMEM;
- mutex_lock(&buffer_mutex);
-
err = task_handoff_register(&task_free_nb);
if (err)
goto out1;
@@ -173,7 +171,6 @@ int sync_start(void)
start_cpu_work();
out:
- mutex_unlock(&buffer_mutex);
return err;
out4:
profile_event_unregister(PROFILE_MUNMAP, &munmap_nb);
@@ -190,14 +187,13 @@ out1:
void sync_stop(void)
{
- /* flush buffers */
- mutex_lock(&buffer_mutex);
end_cpu_work();
unregister_module_notifier(&module_load_nb);
profile_event_unregister(PROFILE_MUNMAP, &munmap_nb);
profile_event_unregister(PROFILE_TASK_EXIT, &task_exit_nb);
task_handoff_unregister(&task_free_nb);
- mutex_unlock(&buffer_mutex);
+ barrier(); /* do all of the above first */
+
flush_scheduled_work();
free_all_tasks();
--
1.7.12.1
next prev parent reply other threads:[~2013-01-08 23:35 UTC|newest]
Thread overview: 82+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-01-08 23:34 [v2.6.34-stable 00/77] v2.6.34.14 longterm review Paul Gortmaker
2013-01-08 23:34 ` [v2.6.34-stable 01/77] net: sock: validate data_len before allocating skb in sock_alloc_send_pskb() Paul Gortmaker
2013-01-08 23:34 ` [v2.6.34-stable 02/77] time: Improve sanity checking of timekeeping inputs Paul Gortmaker
2013-01-08 23:34 ` [v2.6.34-stable 03/77] time: Avoid making adjustments if we haven't accumulated anything Paul Gortmaker
2013-01-08 23:34 ` [v2.6.34-stable 04/77] time: Move ktime_t overflow checking into timespec_valid_strict Paul Gortmaker
2013-01-08 23:34 ` [v2.6.34-stable 05/77] ALSA: hda_intel: ALSA HD Audio patch for Intel Patsburg DeviceIDs Paul Gortmaker
2013-01-08 23:34 ` [v2.6.34-stable 06/77] ALSA: hda: add Vortex86MX PCI ids Paul Gortmaker
2013-01-08 23:34 ` [v2.6.34-stable 07/77] ALSA: hda - Add support for VMware controller Paul Gortmaker
2013-01-08 23:34 ` [v2.6.34-stable 08/77] ALSA: hda - Reduce pci id list for Intel with class id Paul Gortmaker
2013-01-08 23:34 ` [v2.6.34-stable 09/77] ALSA: hda - ALSA HD Audio patch for Intel Panther Point DeviceIDs Paul Gortmaker
2013-01-08 23:34 ` [v2.6.34-stable 10/77] ALSA: hda: Use position_fix=1 for Acer Aspire 5538 to enable capture on internal mic Paul Gortmaker
2013-01-08 23:34 ` [v2.6.34-stable 11/77] cifs: fix cifs stable patch cifs-fix-oplock-break-handling-try-2.patch Paul Gortmaker
2013-01-08 23:34 ` [v2.6.34-stable 12/77] gro: reset vlan_tci on reuse Paul Gortmaker
2013-01-08 23:34 ` [v2.6.34-stable 13/77] md: Fix handling for devices from 2TB to 4TB in 0.90 metadata Paul Gortmaker
2013-01-08 23:34 ` [v2.6.34-stable 14/77] md: Don't truncate size at 4TB for RAID0 and Linear Paul Gortmaker
2013-01-08 23:34 ` [v2.6.34-stable 15/77] genalloc: stop crashing the system when destroying a pool Paul Gortmaker
2013-01-08 23:34 ` [v2.6.34-stable 16/77] inotify: stop kernel memory leak on file creation failure Paul Gortmaker
2013-01-08 23:34 ` [v2.6.34-stable 17/77] xfs: validate acl count Paul Gortmaker
2013-01-08 23:34 ` [v2.6.34-stable 18/77] xfs: fix acl count validation in xfs_acl_from_disk() Paul Gortmaker
2013-01-08 23:34 ` [v2.6.34-stable 19/77] x86, ioapic: initialize nr_ioapic_registers early in mp_register_ioapic() Paul Gortmaker
2013-01-08 23:34 ` [v2.6.34-stable 20/77] i2c-algo-bit: Generate correct i2c address sequence for 10-bit target Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 21/77] eCryptfs: Extend array bounds for all filename chars Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 22/77] PCI hotplug: shpchp: don't blindly claim non-AMD 0x7450 device IDs Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 23/77] ARM: 7161/1: errata: no automatic store buffer drain Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 24/77] ALSA: lx6464es - fix device communication via command bus Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 25/77] SUNRPC: Ensure we return EAGAIN in xs_nospace if congestion is cleared Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 26/77] timekeeping: add arch_offset hook to ktime_get functions Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 27/77] p54spi: Add missing spin_lock_init Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 28/77] p54spi: Fix workqueue deadlock Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 29/77] nl80211: fix MAC address validation Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 30/77] staging: usbip: bugfix for deadlock Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 31/77] staging: comedi: fix oops for USB DAQ devices Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 32/77] Staging: comedi: fix signal handling in read and write Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 33/77] USB: whci-hcd: fix endian conversion in qset_clear() Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 34/77] usb: ftdi_sio: add PID for Propox ISPcable III Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 35/77] usb: option: add SIMCom SIM5218 Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 36/77] USB: usb-storage: unusual_devs entry for Kingston DT 101 G2 Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 37/77] Silencing 'killing requests for dead queue' Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 38/77] sched, x86: Avoid unnecessary overflow in sched_clock Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 39/77] x86/mpparse: Account for bus types other than ISA and PCI Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 40/77] oprofile, x86: Fix crash when unloading module (nmi timer mode) Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 41/77] genirq: Fix race condition when stopping the irq thread Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 42/77] tick-broadcast: Stop active broadcast device when replacing it Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 43/77] ALSA: sis7019 - give slow codecs more time to reset Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 44/77] ALSA: hda/realtek - Fix Oops in alc_mux_select() Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 45/77] ARM: davinci: dm646x evm: wrong register used in setup_vpif_input_channel_mode Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 46/77] oprofile: Free potentially owned tasks in case of errors Paul Gortmaker
2013-01-08 23:35 ` Paul Gortmaker [this message]
2013-01-08 23:35 ` [v2.6.34-stable 48/77] percpu: fix first chunk match in per_cpu_ptr_to_phys() Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 49/77] percpu: fix chunk range calculation Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 50/77] xfrm: Fix key lengths for rfc3686(ctr(aes)) Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 51/77] linux/log2.h: Fix rounddown_pow_of_two(1) Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 52/77] jbd/jbd2: validate sb->s_first in journal_get_superblock() Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 53/77] Make TASKSTATS require root access Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 54/77] hfs: fix hfs_find_init() sb->ext_tree NULL ptr oops Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 55/77] export __get_user_pages_fast() function Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 56/77] oprofile, x86: Fix nmi-unsafe callgraph support Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 57/77] ext4: avoid hangs in ext4_da_should_update_i_disksize() Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 58/77] USB: cdc-acm: add IDs for Motorola H24 HSPA USB module Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 59/77] udf: Fortify loading of sparing table Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 60/77] udf: Avoid run away loop when partition table length is corrupted Paul Gortmaker
2013-01-10 14:43 ` Ben Hutchings
2013-01-10 17:03 ` Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 61/77] sctp: malloc enough room for asconf-ack chunk Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 62/77] sctp: Fix list corruption resulting from freeing an association on a list Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 63/77] sctp: ABORT if receive, reassmbly, or reodering queue is not empty while closing socket Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 64/77] sctp: Enforce retransmission limit during shutdown Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 65/77] SCTP: fix race between sctp_bind_addr_free() and sctp_bind_addr_conflict() Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 66/77] KVM: x86: Prevent starting PIT timers in the absence of irqchip support Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 67/77] perf_events: Fix races in group composition Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 68/77] perf: Fix tear-down of inherited group events Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 69/77] sched: fix divide by zero at {thread_group,task}_times Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 70/77] mutex: Place lock in contended state after fastpath_lock failure Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 71/77] crypto: ghash - Avoid null pointer dereference if no key is set Paul Gortmaker
2013-01-09 2:56 ` Nick Bowler
2013-01-09 14:56 ` Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 72/77] net: Fix ip link add netns oops Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 73/77] sched/rt: Fix task stack corruption under __ARCH_WANT_INTERRUPTS_ON_CTXSW Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 74/77] rwsem: Remove redundant asmregparm annotation Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 75/77] um: Use RWSEM_GENERIC_SPINLOCK on x86 Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 76/77] x86: Get rid of asmregparm Paul Gortmaker
2013-01-08 23:35 ` [v2.6.34-stable 77/77] x86: Don't use the EFI reboot method by default Paul Gortmaker
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=1357688156-25387-48-git-send-email-paul.gortmaker@windriver.com \
--to=paul.gortmaker@windriver.com \
--cc=carll@us.ibm.com \
--cc=linux-kernel@vger.kernel.org \
--cc=robert.richter@amd.com \
--cc=stable@vger.kernel.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 NNTP newsgroup(s).