From: Steven Rostedt <rostedt@goodmis.org>
To: linux-kernel@vger.kernel.org,
linux-rt-users <linux-rt-users@vger.kernel.org>
Cc: Thomas Gleixner <tglx@linutronix.de>,
Carsten Emde <C.Emde@osadl.org>, John Kacur <jkacur@redhat.com>,
Peter Zijlstra <peterz@infradead.org>,
Clark Williams <clark.williams@gmail.com>,
stable-rt@vger.kernel.org
Subject: [PATCH RT 23/25][RFC 3.0.23-rt39-rc1] lglock/rt: Use non-rt for_each_cpu() in -rt code
Date: Tue, 06 Mar 2012 11:16:59 -0500 [thread overview]
Message-ID: <20120306161952.081930351@goodmis.org> (raw)
In-Reply-To: <20120306161636.491172179@goodmis.org>
[-- Attachment #1: 0023-lglock-rt-Use-non-rt-for_each_cpu-in-rt-code.patch --]
[-- Type: text/plain, Size: 3925 bytes --]
From: Steven Rostedt <rostedt@goodmis.org>
Currently the RT version of the lglocks() does a for_each_online_cpu()
in the name##_global_lock_online() functions. Non-rt uses its own
mask for this, and for good reason.
A task may grab a *_global_lock_online(), and in the mean time, one
of the CPUs goes offline. Now when that task does a *_global_unlock_online()
it releases all the locks *except* the one that went offline.
Now if that CPU were to come back on line, its lock is now owned by a
task that never released it when it should have.
This causes all sorts of fun errors. Like owners of a lock no longer
existing, or sleeping on IO, waiting to be woken up by a task that
happens to be blocked on the lock it never released.
Convert the RT versions to use the lglock specific cpumasks. As once
a CPU comes on line, the mask is set, and never cleared even when the
CPU goes offline. The locks for that CPU will still be taken and released.
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Cc: Carsten Emde <C.Emde@osadl.org>
Cc: John Kacur <jkacur@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Clark Williams <clark.williams@gmail.com>
Cc: stable-rt@vger.kernel.org
Link: http://lkml.kernel.org/r/20120301190345.374756214@goodmis.org
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
---
include/linux/lglock.h | 35 ++++++++++++++++++++++++++++++++---
1 files changed, 32 insertions(+), 3 deletions(-)
diff --git a/include/linux/lglock.h b/include/linux/lglock.h
index d8acbcc..9cfab36 100644
--- a/include/linux/lglock.h
+++ b/include/linux/lglock.h
@@ -204,9 +204,31 @@
#else /* !PREEMPT_RT_FULL */
#define DEFINE_LGLOCK(name) \
\
- DEFINE_PER_CPU(struct rt_mutex, name##_lock); \
+ DEFINE_PER_CPU(struct rt_mutex, name##_lock); \
+ DEFINE_SPINLOCK(name##_cpu_lock); \
+ cpumask_t name##_cpus __read_mostly; \
DEFINE_LGLOCK_LOCKDEP(name); \
\
+ static int \
+ name##_lg_cpu_callback(struct notifier_block *nb, \
+ unsigned long action, void *hcpu) \
+ { \
+ switch (action & ~CPU_TASKS_FROZEN) { \
+ case CPU_UP_PREPARE: \
+ spin_lock(&name##_cpu_lock); \
+ cpu_set((unsigned long)hcpu, name##_cpus); \
+ spin_unlock(&name##_cpu_lock); \
+ break; \
+ case CPU_UP_CANCELED: case CPU_DEAD: \
+ spin_lock(&name##_cpu_lock); \
+ cpu_clear((unsigned long)hcpu, name##_cpus); \
+ spin_unlock(&name##_cpu_lock); \
+ } \
+ return NOTIFY_OK; \
+ } \
+ static struct notifier_block name##_lg_cpu_notifier = { \
+ .notifier_call = name##_lg_cpu_callback, \
+ }; \
void name##_lock_init(void) { \
int i; \
LOCKDEP_INIT_MAP(&name##_lock_dep_map, #name, &name##_lock_key, 0); \
@@ -215,6 +237,11 @@
lock = &per_cpu(name##_lock, i); \
rt_mutex_init(lock); \
} \
+ register_hotcpu_notifier(&name##_lg_cpu_notifier); \
+ get_online_cpus(); \
+ for_each_online_cpu(i) \
+ cpu_set(i, name##_cpus); \
+ put_online_cpus(); \
} \
EXPORT_SYMBOL(name##_lock_init); \
\
@@ -255,7 +282,8 @@
void name##_global_lock_online(void) { \
int i; \
rwlock_acquire(&name##_lock_dep_map, 0, 0, _RET_IP_); \
- for_each_online_cpu(i) { \
+ spin_lock(&name##_cpu_lock); \
+ for_each_cpu(i, &name##_cpus) { \
struct rt_mutex *lock; \
lock = &per_cpu(name##_lock, i); \
__rt_spin_lock(lock); \
@@ -266,11 +294,12 @@
void name##_global_unlock_online(void) { \
int i; \
rwlock_release(&name##_lock_dep_map, 1, _RET_IP_); \
- for_each_online_cpu(i) { \
+ for_each_cpu(i, &name##_cpus) { \
struct rt_mutex *lock; \
lock = &per_cpu(name##_lock, i); \
__rt_spin_unlock(lock); \
} \
+ spin_unlock(&name##_cpu_lock); \
} \
EXPORT_SYMBOL(name##_global_unlock_online); \
\
--
1.7.8.3
next prev parent reply other threads:[~2012-03-06 16:59 UTC|newest]
Thread overview: 29+ messages / expand[flat|nested] mbox.gz Atom feed top
2012-03-06 16:16 [PATCH RT 00/25][RFC 3.0.23-rt39-rc1] [ANNOUNCE] 3.0.23-rt39-rc1 release cycle Steven Rostedt
2012-03-06 16:16 ` [PATCH RT 01/25][RFC 3.0.23-rt39-rc1] revert-convert-xtime_lock-to-raw_seqlock Steven Rostedt
2012-03-06 16:16 ` [PATCH RT 02/25][RFC 3.0.23-rt39-rc1] revert-seqlock-create-raw_seqlock Steven Rostedt
2012-03-06 16:16 ` [PATCH RT 03/25][RFC 3.0.23-rt39-rc1] revert-seqlock-use-seqcount Steven Rostedt
2012-03-06 16:16 ` [PATCH RT 04/25][RFC 3.0.23-rt39-rc1] revert-seqlock-remove-unused-functions Steven Rostedt
2012-03-06 22:39 ` John Kacur
2012-03-06 22:42 ` Thomas Gleixner
2012-03-06 16:16 ` [PATCH RT 05/25][RFC 3.0.23-rt39-rc1] x86-64-remove-vsyscall-number-3 Steven Rostedt
2012-03-06 16:16 ` [PATCH RT 06/25][RFC 3.0.23-rt39-rc1] x86-64-emulate-legacy-vsyscalls Steven Rostedt
2012-03-06 16:16 ` [PATCH RT 07/25][RFC 3.0.23-rt39-rc1] x86: vdso: Remove bogus locking in update_vsyscall_tz() Steven Rostedt
2012-03-06 16:16 ` [PATCH RT 08/25][RFC 3.0.23-rt39-rc1] x86: vdso: Use seqcount instead of seqlock Steven Rostedt
2012-03-06 16:16 ` [PATCH RT 09/25][RFC 3.0.23-rt39-rc1] ia64: vsyscall: " Steven Rostedt
2012-03-06 16:16 ` [PATCH RT 10/25][RFC 3.0.23-rt39-rc1] seqlock: Remove unused functions Steven Rostedt
2012-03-06 16:16 ` [PATCH RT 11/25][RFC 3.0.23-rt39-rc1] seqlock: Use seqcount Steven Rostedt
2012-03-06 16:16 ` [PATCH RT 12/25][RFC 3.0.23-rt39-rc1] seqlock: Provide seq_spin_* functions Steven Rostedt
2012-03-06 16:16 ` [PATCH RT 13/25][RFC 3.0.23-rt39-rc1] fs: fs_struct use seqlock Steven Rostedt
2012-03-06 16:16 ` [PATCH RT 14/25][RFC 3.0.23-rt39-rc1] fs: dentry " Steven Rostedt
2012-03-06 16:16 ` [PATCH RT 15/25][RFC 3.0.23-rt39-rc1] timekeeping: Split xtime_lock Steven Rostedt
2012-03-06 16:16 ` [PATCH RT 16/25][RFC 3.0.23-rt39-rc1] seqlock: Prevent rt starvation Steven Rostedt
2012-03-06 16:16 ` [PATCH RT 17/25][RFC 3.0.23-rt39-rc1] fs: Protect open coded isize seqcount Steven Rostedt
2012-03-06 16:16 ` [PATCH RT 18/25][RFC 3.0.23-rt39-rc1] net: u64_stat: Protect seqcount Steven Rostedt
2012-03-06 16:16 ` [PATCH RT 19/25][RFC 3.0.23-rt39-rc1] timer: Fix hotplug for -rt Steven Rostedt
2012-03-06 16:16 ` [PATCH RT 20/25][RFC 3.0.23-rt39-rc1] futex/rt: Fix possible lockup when taking pi_lock in proxy handler Steven Rostedt
2012-03-06 16:16 ` [PATCH RT 21/25][RFC 3.0.23-rt39-rc1] ring-buffer/rt: Check for irqs disabled before grabbing reader lock Steven Rostedt
2012-03-06 16:16 ` [PATCH RT 22/25][RFC 3.0.23-rt39-rc1] sched/rt: Fix wait_task_interactive() to test rt_spin_lock state Steven Rostedt
2012-03-06 16:16 ` Steven Rostedt [this message]
2012-03-06 16:17 ` [PATCH RT 24/25][RFC 3.0.23-rt39-rc1] cpu: Make hotplug.lock a "sleeping" spinlock on RT Steven Rostedt
2012-03-06 16:17 ` [PATCH RT 25/25][RFC 3.0.23-rt39-rc1] Linux 3.0.23-rt39-rc1 Steven Rostedt
2012-03-06 22:20 ` [PATCH RT 00/25][RFC 3.0.23-rt39-rc1] [ANNOUNCE] 3.0.23-rt39-rc1 release cycle Steven Rostedt
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=20120306161952.081930351@goodmis.org \
--to=rostedt@goodmis.org \
--cc=C.Emde@osadl.org \
--cc=clark.williams@gmail.com \
--cc=jkacur@redhat.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-rt-users@vger.kernel.org \
--cc=peterz@infradead.org \
--cc=stable-rt@vger.kernel.org \
--cc=tglx@linutronix.de \
/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
Powered by JetHome