* [Intel-gfx] [RFC RESEND 00/12] locking: Separate lock tracepoints from lockdep/lock_stat (v1.1)
@ 2022-02-08 19:43 ` Namhyung Kim
0 siblings, 0 replies; 22+ messages in thread
From: Namhyung Kim @ 2022-02-08 19:43 UTC (permalink / raw)
To: Peter Zijlstra, Ingo Molnar, Will Deacon, Waiman Long, Boqun Feng
Cc: rcu, Paul E. McKenney, intel-gfx, LKML, Steven Rostedt,
Radoslaw Burny, Byungchul Park, Mathieu Desnoyers, Tejun Heo,
cgroups, Thomas Gleixner, linux-btrfs
Hello,
(Resending with Paul's correct email address, sorry!)
There have been some requests for low-overhead kernel lock contention
monitoring. The kernel has CONFIG_LOCK_STAT to provide such an infra
either via /proc/lock_stat or tracepoints directly.
However it's not light-weight and hard to be used in production. So
I'm trying to separate out the tracepoints and using them as a base to
build a new monitoring system.
* Changes in v1.1
- add some Acked-by's
- add comments on cgroup rstat cpu lock
As the lockdep and lock_stat provide good hooks in the lock functions,
it'd be natural to reuse them. Actually I tried to use lockdep as is
but disables the functionality at runtime (initialize debug_locks = 0,
lock_stat = 0). But it still has unacceptable overhead and the
lockdep data structures also increase memory footprint unnecessarily.
So I'm proposing a separate tracepoint-only configuration and keeping
lockdep_map only with minimal information needed for tracepoints (for
now, it's just name). And then the existing lockdep hooks can be used
for the tracepoints.
The patch 01-06 are preparation for the work. In a few places in the
kernel, they calls lockdep annotation explicitly to deal with
limitations in the lockdep implementation. In my understanding, they
are not needed to analyze lock contention.
To make matters worse, they rely on the compiler optimization (or
macro magic) so that it can get rid of the annotations and their
arguments when lockdep is not configured.
But it's not true any more when lock tracepoints are added and it'd
cause build errors. So I added #ifdef guards for LOCKDEP in the code
to prevent such errors.
In the patch 07 I mechanically changed most of code that depend on
CONFIG_LOCKDEP or CONFIG_DEBUG_LOCK_ALLOC to CONFIG_LOCK_INFO. It
paves the way for the codes to be shared for lockdep and tracepoints.
Mostly, it makes sure that locks are initialized with a proper name,
like in the patch 08 and 09.
I didn't change some places intentionally - for example, timer and
workqueue depend on LOCKDEP explicitly since they use some lockdep
annotations to work around the "held lock freed" warnings. The ocfs2
directly accesses lockdep_map.key so I didn't touch the code for now.
And RCU was because it generates too much overhead thanks to the
rcu_read_lock(). Maybe I need to revisit some of them later.
I added CONFIG_LOCK_TRACEPOINTS in the patch 10 to make it optional.
I found that it adds 2~3% of overhead when I ran `perf bench sched
messaging` even when the tracepoints are disabled. The benchmark
creates a lot of processes and make them communicate by socket pairs
(or pipes). I measured that around 15% of lock acquisition creates
contentions and it's mostly for spin locks (alc->lock and u->lock).
I ran perf record + report with the workload and it showed 50% of the
cpu cycles are in the spin lock slow path. So it's highly affected by
the spin lock slow path. Actually LOCK_CONTENDED() macro transforms
the spin lock code (and others) to use trylock first and then fall
back to real lock function if failed. Thus it'd add more (atomic)
operations and cache line bouncing for the contended cases:
#define LOCK_CONTENDED(_lock, try, lock) \
do { \
if (!try(_lock)) { \
lock_contended(&(_lock)->dep_map, _RET_IP_); \
lock(_lock); \
} \
lock_acquired(&(_lock)->dep_map, _RET_IP_); \
} while (0)
If I modify the macro not to use trylock and to call the real lock
function directly (so the lock_contended tracepoint would be called
always, if enabled), the overhead goes down to 0.x% when the
tracepoints are disabled.
I don't have a good solution as long as we use LOCK_CONTENDED() macro
to separate the contended locking path. Maybe we can make it call
some (generic) slow path lock function directly after failing trylock.
Or move the lockdep annotations into the each lock function bodies and
get rid of the LOCK_CONTENDED() macro entirely. Ideas?
Actually the patch 11 handles the same issue on the mutex code. The
fast version of mutex trylock was attempted only if LOCKDEP is not
enabled and it affects the mutex lock performance in the uncontended
cases too. So I partially reverted the change in the patch 07 to use
the fast functions with lock tracepoints too. Maybe we can use it
with LOCKDEP as well?
The last patch 12 might be controversial and I'd like to move the
lock_acquired annotation into the if(!try) block in the LOCK_CONTEDED
macro so that it can only be called when there's a contention.
Eventually I'm mostly interested in the contended locks only and I
want to reduce the overhead in the fast path. By moving that, it'd be
easy to track contended locks with timing by using two tracepoints.
It'd change lock hold time calculation in lock_stat for the fast path,
but I assume time difference between lock_acquire and lock_acquired
would be small when the lock is not contended. So I think we can use
the timestamp in lock_acquire. If it's not acceptable, we might
consider adding a new tracepoint to track the timing of contended
locks.
This series base on the current tip/locking/core and you get it from
'locking/tracepoint-v1' branch in my tree at:
git://git.kernel.org/pub/scm/linux/kernel/git/namhyung/linux-perf.git
Thanks,
Namhyung
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Byungchul Park <byungchul.park@lge.com>
Cc: rcu@vger.kernel.org
Cc: cgroups@vger.kernel.org
Cc: linux-btrfs@vger.kernel.org
Cc: intel-gfx@lists.freedesktop.org
Namhyung Kim (12):
locking: Pass correct outer wait type info
cgroup: rstat: Make cgroup_rstat_cpu_lock name readable
timer: Protect lockdep functions with #ifdef
workqueue: Protect lockdep functions with #ifdef
drm/i915: Protect lockdep functions with #ifdef
btrfs: change lockdep class size check using ks->names
locking: Introduce CONFIG_LOCK_INFO
locking/mutex: Init name properly w/ CONFIG_LOCK_INFO
locking: Add more static lockdep init macros
locking: Add CONFIG_LOCK_TRACEPOINTS option
locking/mutex: Revive fast functions for LOCK_TRACEPOINTS
locking: Move lock_acquired() from the fast path
drivers/gpu/drm/drm_connector.c | 7 +-
drivers/gpu/drm/i915/i915_sw_fence.h | 2 +-
drivers/gpu/drm/i915/intel_wakeref.c | 3 +
drivers/gpu/drm/i915/selftests/lib_sw_fence.h | 2 +-
.../net/wireless/intel/iwlwifi/iwl-trans.c | 4 +-
.../net/wireless/intel/iwlwifi/iwl-trans.h | 2 +-
drivers/tty/tty_ldsem.c | 2 +-
fs/btrfs/disk-io.c | 4 +-
fs/btrfs/disk-io.h | 2 +-
fs/cifs/connect.c | 2 +-
fs/kernfs/file.c | 2 +-
include/linux/completion.h | 2 +-
include/linux/jbd2.h | 2 +-
include/linux/kernfs.h | 2 +-
include/linux/kthread.h | 2 +-
include/linux/local_lock_internal.h | 18 +-
include/linux/lockdep.h | 170 ++++++++++++++++--
include/linux/lockdep_types.h | 8 +-
include/linux/mmu_notifier.h | 2 +-
include/linux/mutex.h | 12 +-
include/linux/percpu-rwsem.h | 4 +-
include/linux/regmap.h | 4 +-
include/linux/rtmutex.h | 14 +-
include/linux/rwlock_api_smp.h | 4 +-
include/linux/rwlock_rt.h | 4 +-
include/linux/rwlock_types.h | 11 +-
include/linux/rwsem.h | 14 +-
include/linux/seqlock.h | 4 +-
include/linux/spinlock_api_smp.h | 4 +-
include/linux/spinlock_rt.h | 4 +-
include/linux/spinlock_types.h | 4 +-
include/linux/spinlock_types_raw.h | 28 ++-
include/linux/swait.h | 2 +-
include/linux/tty_ldisc.h | 2 +-
include/linux/wait.h | 2 +-
include/linux/ww_mutex.h | 6 +-
include/media/v4l2-ctrls.h | 2 +-
include/net/sock.h | 2 +-
include/trace/events/lock.h | 4 +-
kernel/cgroup/rstat.c | 7 +-
kernel/locking/Makefile | 1 +
kernel/locking/lockdep.c | 40 ++++-
kernel/locking/mutex-debug.c | 2 +-
kernel/locking/mutex.c | 22 ++-
kernel/locking/mutex.h | 7 +
kernel/locking/percpu-rwsem.c | 2 +-
kernel/locking/rtmutex_api.c | 10 +-
kernel/locking/rwsem.c | 4 +-
kernel/locking/spinlock.c | 2 +-
kernel/locking/spinlock_debug.c | 4 +-
kernel/locking/spinlock_rt.c | 8 +-
kernel/locking/ww_rt_mutex.c | 2 +-
kernel/printk/printk.c | 14 +-
kernel/rcu/update.c | 27 +--
kernel/time/timer.c | 8 +-
kernel/workqueue.c | 13 ++
lib/Kconfig.debug | 14 ++
mm/memcontrol.c | 7 +-
mm/mmu_notifier.c | 2 +-
net/core/dev.c | 2 +-
net/sunrpc/svcsock.c | 2 +-
net/sunrpc/xprtsock.c | 2 +-
62 files changed, 391 insertions(+), 180 deletions(-)
base-commit: 1dc01abad6544cb9d884071b626b706e37aa9601
--
2.35.0.263.gb82422642f-goog
^ permalink raw reply [flat|nested] 22+ messages in thread
* [RFC RESEND 00/12] locking: Separate lock tracepoints from lockdep/lock_stat (v1.1)
@ 2022-02-08 19:43 ` Namhyung Kim
0 siblings, 0 replies; 22+ messages in thread
From: Namhyung Kim @ 2022-02-08 19:43 UTC (permalink / raw)
To: Peter Zijlstra, Ingo Molnar, Will Deacon, Waiman Long, Boqun Feng
Cc: LKML, Thomas Gleixner, Steven Rostedt, Byungchul Park,
Paul E. McKenney, Mathieu Desnoyers, Radoslaw Burny, Tejun Heo,
rcu, cgroups, linux-btrfs, intel-gfx
Hello,
(Resending with Paul's correct email address, sorry!)
There have been some requests for low-overhead kernel lock contention
monitoring. The kernel has CONFIG_LOCK_STAT to provide such an infra
either via /proc/lock_stat or tracepoints directly.
However it's not light-weight and hard to be used in production. So
I'm trying to separate out the tracepoints and using them as a base to
build a new monitoring system.
* Changes in v1.1
- add some Acked-by's
- add comments on cgroup rstat cpu lock
As the lockdep and lock_stat provide good hooks in the lock functions,
it'd be natural to reuse them. Actually I tried to use lockdep as is
but disables the functionality at runtime (initialize debug_locks = 0,
lock_stat = 0). But it still has unacceptable overhead and the
lockdep data structures also increase memory footprint unnecessarily.
So I'm proposing a separate tracepoint-only configuration and keeping
lockdep_map only with minimal information needed for tracepoints (for
now, it's just name). And then the existing lockdep hooks can be used
for the tracepoints.
The patch 01-06 are preparation for the work. In a few places in the
kernel, they calls lockdep annotation explicitly to deal with
limitations in the lockdep implementation. In my understanding, they
are not needed to analyze lock contention.
To make matters worse, they rely on the compiler optimization (or
macro magic) so that it can get rid of the annotations and their
arguments when lockdep is not configured.
But it's not true any more when lock tracepoints are added and it'd
cause build errors. So I added #ifdef guards for LOCKDEP in the code
to prevent such errors.
In the patch 07 I mechanically changed most of code that depend on
CONFIG_LOCKDEP or CONFIG_DEBUG_LOCK_ALLOC to CONFIG_LOCK_INFO. It
paves the way for the codes to be shared for lockdep and tracepoints.
Mostly, it makes sure that locks are initialized with a proper name,
like in the patch 08 and 09.
I didn't change some places intentionally - for example, timer and
workqueue depend on LOCKDEP explicitly since they use some lockdep
annotations to work around the "held lock freed" warnings. The ocfs2
directly accesses lockdep_map.key so I didn't touch the code for now.
And RCU was because it generates too much overhead thanks to the
rcu_read_lock(). Maybe I need to revisit some of them later.
I added CONFIG_LOCK_TRACEPOINTS in the patch 10 to make it optional.
I found that it adds 2~3% of overhead when I ran `perf bench sched
messaging` even when the tracepoints are disabled. The benchmark
creates a lot of processes and make them communicate by socket pairs
(or pipes). I measured that around 15% of lock acquisition creates
contentions and it's mostly for spin locks (alc->lock and u->lock).
I ran perf record + report with the workload and it showed 50% of the
cpu cycles are in the spin lock slow path. So it's highly affected by
the spin lock slow path. Actually LOCK_CONTENDED() macro transforms
the spin lock code (and others) to use trylock first and then fall
back to real lock function if failed. Thus it'd add more (atomic)
operations and cache line bouncing for the contended cases:
#define LOCK_CONTENDED(_lock, try, lock) \
do { \
if (!try(_lock)) { \
lock_contended(&(_lock)->dep_map, _RET_IP_); \
lock(_lock); \
} \
lock_acquired(&(_lock)->dep_map, _RET_IP_); \
} while (0)
If I modify the macro not to use trylock and to call the real lock
function directly (so the lock_contended tracepoint would be called
always, if enabled), the overhead goes down to 0.x% when the
tracepoints are disabled.
I don't have a good solution as long as we use LOCK_CONTENDED() macro
to separate the contended locking path. Maybe we can make it call
some (generic) slow path lock function directly after failing trylock.
Or move the lockdep annotations into the each lock function bodies and
get rid of the LOCK_CONTENDED() macro entirely. Ideas?
Actually the patch 11 handles the same issue on the mutex code. The
fast version of mutex trylock was attempted only if LOCKDEP is not
enabled and it affects the mutex lock performance in the uncontended
cases too. So I partially reverted the change in the patch 07 to use
the fast functions with lock tracepoints too. Maybe we can use it
with LOCKDEP as well?
The last patch 12 might be controversial and I'd like to move the
lock_acquired annotation into the if(!try) block in the LOCK_CONTEDED
macro so that it can only be called when there's a contention.
Eventually I'm mostly interested in the contended locks only and I
want to reduce the overhead in the fast path. By moving that, it'd be
easy to track contended locks with timing by using two tracepoints.
It'd change lock hold time calculation in lock_stat for the fast path,
but I assume time difference between lock_acquire and lock_acquired
would be small when the lock is not contended. So I think we can use
the timestamp in lock_acquire. If it's not acceptable, we might
consider adding a new tracepoint to track the timing of contended
locks.
This series base on the current tip/locking/core and you get it from
'locking/tracepoint-v1' branch in my tree at:
git://git.kernel.org/pub/scm/linux/kernel/git/namhyung/linux-perf.git
Thanks,
Namhyung
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Byungchul Park <byungchul.park@lge.com>
Cc: rcu@vger.kernel.org
Cc: cgroups@vger.kernel.org
Cc: linux-btrfs@vger.kernel.org
Cc: intel-gfx@lists.freedesktop.org
Namhyung Kim (12):
locking: Pass correct outer wait type info
cgroup: rstat: Make cgroup_rstat_cpu_lock name readable
timer: Protect lockdep functions with #ifdef
workqueue: Protect lockdep functions with #ifdef
drm/i915: Protect lockdep functions with #ifdef
btrfs: change lockdep class size check using ks->names
locking: Introduce CONFIG_LOCK_INFO
locking/mutex: Init name properly w/ CONFIG_LOCK_INFO
locking: Add more static lockdep init macros
locking: Add CONFIG_LOCK_TRACEPOINTS option
locking/mutex: Revive fast functions for LOCK_TRACEPOINTS
locking: Move lock_acquired() from the fast path
drivers/gpu/drm/drm_connector.c | 7 +-
drivers/gpu/drm/i915/i915_sw_fence.h | 2 +-
drivers/gpu/drm/i915/intel_wakeref.c | 3 +
drivers/gpu/drm/i915/selftests/lib_sw_fence.h | 2 +-
.../net/wireless/intel/iwlwifi/iwl-trans.c | 4 +-
.../net/wireless/intel/iwlwifi/iwl-trans.h | 2 +-
drivers/tty/tty_ldsem.c | 2 +-
fs/btrfs/disk-io.c | 4 +-
fs/btrfs/disk-io.h | 2 +-
fs/cifs/connect.c | 2 +-
fs/kernfs/file.c | 2 +-
include/linux/completion.h | 2 +-
include/linux/jbd2.h | 2 +-
include/linux/kernfs.h | 2 +-
include/linux/kthread.h | 2 +-
include/linux/local_lock_internal.h | 18 +-
include/linux/lockdep.h | 170 ++++++++++++++++--
include/linux/lockdep_types.h | 8 +-
include/linux/mmu_notifier.h | 2 +-
include/linux/mutex.h | 12 +-
include/linux/percpu-rwsem.h | 4 +-
include/linux/regmap.h | 4 +-
include/linux/rtmutex.h | 14 +-
include/linux/rwlock_api_smp.h | 4 +-
include/linux/rwlock_rt.h | 4 +-
include/linux/rwlock_types.h | 11 +-
include/linux/rwsem.h | 14 +-
include/linux/seqlock.h | 4 +-
include/linux/spinlock_api_smp.h | 4 +-
include/linux/spinlock_rt.h | 4 +-
include/linux/spinlock_types.h | 4 +-
include/linux/spinlock_types_raw.h | 28 ++-
include/linux/swait.h | 2 +-
include/linux/tty_ldisc.h | 2 +-
include/linux/wait.h | 2 +-
include/linux/ww_mutex.h | 6 +-
include/media/v4l2-ctrls.h | 2 +-
include/net/sock.h | 2 +-
include/trace/events/lock.h | 4 +-
kernel/cgroup/rstat.c | 7 +-
kernel/locking/Makefile | 1 +
kernel/locking/lockdep.c | 40 ++++-
kernel/locking/mutex-debug.c | 2 +-
kernel/locking/mutex.c | 22 ++-
kernel/locking/mutex.h | 7 +
kernel/locking/percpu-rwsem.c | 2 +-
kernel/locking/rtmutex_api.c | 10 +-
kernel/locking/rwsem.c | 4 +-
kernel/locking/spinlock.c | 2 +-
kernel/locking/spinlock_debug.c | 4 +-
kernel/locking/spinlock_rt.c | 8 +-
kernel/locking/ww_rt_mutex.c | 2 +-
kernel/printk/printk.c | 14 +-
kernel/rcu/update.c | 27 +--
kernel/time/timer.c | 8 +-
kernel/workqueue.c | 13 ++
lib/Kconfig.debug | 14 ++
mm/memcontrol.c | 7 +-
mm/mmu_notifier.c | 2 +-
net/core/dev.c | 2 +-
net/sunrpc/svcsock.c | 2 +-
net/sunrpc/xprtsock.c | 2 +-
62 files changed, 391 insertions(+), 180 deletions(-)
base-commit: 1dc01abad6544cb9d884071b626b706e37aa9601
--
2.35.0.263.gb82422642f-goog
^ permalink raw reply [flat|nested] 22+ messages in thread
* [RFC RESEND 00/12] locking: Separate lock tracepoints from lockdep/lock_stat (v1.1)
@ 2022-02-08 19:43 ` Namhyung Kim
0 siblings, 0 replies; 22+ messages in thread
From: Namhyung Kim @ 2022-02-08 19:43 UTC (permalink / raw)
To: Peter Zijlstra, Ingo Molnar, Will Deacon, Waiman Long, Boqun Feng
Cc: rcu, Paul E. McKenney, intel-gfx, LKML, Steven Rostedt,
Radoslaw Burny, Byungchul Park, Mathieu Desnoyers, Tejun Heo,
cgroups, Thomas Gleixner, linux-btrfs
Hello,
(Resending with Paul's correct email address, sorry!)
There have been some requests for low-overhead kernel lock contention
monitoring. The kernel has CONFIG_LOCK_STAT to provide such an infra
either via /proc/lock_stat or tracepoints directly.
However it's not light-weight and hard to be used in production. So
I'm trying to separate out the tracepoints and using them as a base to
build a new monitoring system.
* Changes in v1.1
- add some Acked-by's
- add comments on cgroup rstat cpu lock
As the lockdep and lock_stat provide good hooks in the lock functions,
it'd be natural to reuse them. Actually I tried to use lockdep as is
but disables the functionality at runtime (initialize debug_locks = 0,
lock_stat = 0). But it still has unacceptable overhead and the
lockdep data structures also increase memory footprint unnecessarily.
So I'm proposing a separate tracepoint-only configuration and keeping
lockdep_map only with minimal information needed for tracepoints (for
now, it's just name). And then the existing lockdep hooks can be used
for the tracepoints.
The patch 01-06 are preparation for the work. In a few places in the
kernel, they calls lockdep annotation explicitly to deal with
limitations in the lockdep implementation. In my understanding, they
are not needed to analyze lock contention.
To make matters worse, they rely on the compiler optimization (or
macro magic) so that it can get rid of the annotations and their
arguments when lockdep is not configured.
But it's not true any more when lock tracepoints are added and it'd
cause build errors. So I added #ifdef guards for LOCKDEP in the code
to prevent such errors.
In the patch 07 I mechanically changed most of code that depend on
CONFIG_LOCKDEP or CONFIG_DEBUG_LOCK_ALLOC to CONFIG_LOCK_INFO. It
paves the way for the codes to be shared for lockdep and tracepoints.
Mostly, it makes sure that locks are initialized with a proper name,
like in the patch 08 and 09.
I didn't change some places intentionally - for example, timer and
workqueue depend on LOCKDEP explicitly since they use some lockdep
annotations to work around the "held lock freed" warnings. The ocfs2
directly accesses lockdep_map.key so I didn't touch the code for now.
And RCU was because it generates too much overhead thanks to the
rcu_read_lock(). Maybe I need to revisit some of them later.
I added CONFIG_LOCK_TRACEPOINTS in the patch 10 to make it optional.
I found that it adds 2~3% of overhead when I ran `perf bench sched
messaging` even when the tracepoints are disabled. The benchmark
creates a lot of processes and make them communicate by socket pairs
(or pipes). I measured that around 15% of lock acquisition creates
contentions and it's mostly for spin locks (alc->lock and u->lock).
I ran perf record + report with the workload and it showed 50% of the
cpu cycles are in the spin lock slow path. So it's highly affected by
the spin lock slow path. Actually LOCK_CONTENDED() macro transforms
the spin lock code (and others) to use trylock first and then fall
back to real lock function if failed. Thus it'd add more (atomic)
operations and cache line bouncing for the contended cases:
#define LOCK_CONTENDED(_lock, try, lock) \
do { \
if (!try(_lock)) { \
lock_contended(&(_lock)->dep_map, _RET_IP_); \
lock(_lock); \
} \
lock_acquired(&(_lock)->dep_map, _RET_IP_); \
} while (0)
If I modify the macro not to use trylock and to call the real lock
function directly (so the lock_contended tracepoint would be called
always, if enabled), the overhead goes down to 0.x% when the
tracepoints are disabled.
I don't have a good solution as long as we use LOCK_CONTENDED() macro
to separate the contended locking path. Maybe we can make it call
some (generic) slow path lock function directly after failing trylock.
Or move the lockdep annotations into the each lock function bodies and
get rid of the LOCK_CONTENDED() macro entirely. Ideas?
Actually the patch 11 handles the same issue on the mutex code. The
fast version of mutex trylock was attempted only if LOCKDEP is not
enabled and it affects the mutex lock performance in the uncontended
cases too. So I partially reverted the change in the patch 07 to use
the fast functions with lock tracepoints too. Maybe we can use it
with LOCKDEP as well?
The last patch 12 might be controversial and I'd like to move the
lock_acquired annotation into the if(!try) block in the LOCK_CONTEDED
macro so that it can only be called when there's a contention.
Eventually I'm mostly interested in the contended locks only and I
want to reduce the overhead in the fast path. By moving that, it'd be
easy to track contended locks with timing by using two tracepoints.
It'd change lock hold time calculation in lock_stat for the fast path,
but I assume time difference between lock_acquire and lock_acquired
would be small when the lock is not contended. So I think we can use
the timestamp in lock_acquire. If it's not acceptable, we might
consider adding a new tracepoint to track the timing of contended
locks.
This series base on the current tip/locking/core and you get it from
'locking/tracepoint-v1' branch in my tree at:
git://git.kernel.org/pub/scm/linux/kernel/git/namhyung/linux-perf.git
Thanks,
Namhyung
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Byungchul Park <byungchul.park@lge.com>
Cc: rcu@vger.kernel.org
Cc: cgroups@vger.kernel.org
Cc: linux-btrfs@vger.kernel.org
Cc: intel-gfx@lists.freedesktop.org
Namhyung Kim (12):
locking: Pass correct outer wait type info
cgroup: rstat: Make cgroup_rstat_cpu_lock name readable
timer: Protect lockdep functions with #ifdef
workqueue: Protect lockdep functions with #ifdef
drm/i915: Protect lockdep functions with #ifdef
btrfs: change lockdep class size check using ks->names
locking: Introduce CONFIG_LOCK_INFO
locking/mutex: Init name properly w/ CONFIG_LOCK_INFO
locking: Add more static lockdep init macros
locking: Add CONFIG_LOCK_TRACEPOINTS option
locking/mutex: Revive fast functions for LOCK_TRACEPOINTS
locking: Move lock_acquired() from the fast path
drivers/gpu/drm/drm_connector.c | 7 +-
drivers/gpu/drm/i915/i915_sw_fence.h | 2 +-
drivers/gpu/drm/i915/intel_wakeref.c | 3 +
drivers/gpu/drm/i915/selftests/lib_sw_fence.h | 2 +-
.../net/wireless/intel/iwlwifi/iwl-trans.c | 4 +-
.../net/wireless/intel/iwlwifi/iwl-trans.h | 2 +-
drivers/tty/tty_ldsem.c | 2 +-
fs/btrfs/disk-io.c | 4 +-
fs/btrfs/disk-io.h | 2 +-
fs/cifs/connect.c | 2 +-
fs/kernfs/file.c | 2 +-
include/linux/completion.h | 2 +-
include/linux/jbd2.h | 2 +-
include/linux/kernfs.h | 2 +-
include/linux/kthread.h | 2 +-
include/linux/local_lock_internal.h | 18 +-
include/linux/lockdep.h | 170 ++++++++++++++++--
include/linux/lockdep_types.h | 8 +-
include/linux/mmu_notifier.h | 2 +-
include/linux/mutex.h | 12 +-
include/linux/percpu-rwsem.h | 4 +-
include/linux/regmap.h | 4 +-
include/linux/rtmutex.h | 14 +-
include/linux/rwlock_api_smp.h | 4 +-
include/linux/rwlock_rt.h | 4 +-
include/linux/rwlock_types.h | 11 +-
include/linux/rwsem.h | 14 +-
include/linux/seqlock.h | 4 +-
include/linux/spinlock_api_smp.h | 4 +-
include/linux/spinlock_rt.h | 4 +-
include/linux/spinlock_types.h | 4 +-
include/linux/spinlock_types_raw.h | 28 ++-
include/linux/swait.h | 2 +-
include/linux/tty_ldisc.h | 2 +-
include/linux/wait.h | 2 +-
include/linux/ww_mutex.h | 6 +-
include/media/v4l2-ctrls.h | 2 +-
include/net/sock.h | 2 +-
include/trace/events/lock.h | 4 +-
kernel/cgroup/rstat.c | 7 +-
kernel/locking/Makefile | 1 +
kernel/locking/lockdep.c | 40 ++++-
kernel/locking/mutex-debug.c | 2 +-
kernel/locking/mutex.c | 22 ++-
kernel/locking/mutex.h | 7 +
kernel/locking/percpu-rwsem.c | 2 +-
kernel/locking/rtmutex_api.c | 10 +-
kernel/locking/rwsem.c | 4 +-
kernel/locking/spinlock.c | 2 +-
kernel/locking/spinlock_debug.c | 4 +-
kernel/locking/spinlock_rt.c | 8 +-
kernel/locking/ww_rt_mutex.c | 2 +-
kernel/printk/printk.c | 14 +-
kernel/rcu/update.c | 27 +--
kernel/time/timer.c | 8 +-
kernel/workqueue.c | 13 ++
lib/Kconfig.debug | 14 ++
mm/memcontrol.c | 7 +-
mm/mmu_notifier.c | 2 +-
net/core/dev.c | 2 +-
net/sunrpc/svcsock.c | 2 +-
net/sunrpc/xprtsock.c | 2 +-
62 files changed, 391 insertions(+), 180 deletions(-)
base-commit: 1dc01abad6544cb9d884071b626b706e37aa9601
--
2.35.0.263.gb82422642f-goog
^ permalink raw reply [flat|nested] 22+ messages in thread
* [PATCH 01/12] locking: Pass correct outer wait type info
2022-02-08 19:43 ` Namhyung Kim
(?)
(?)
@ 2022-02-08 19:43 ` Namhyung Kim
-1 siblings, 0 replies; 22+ messages in thread
From: Namhyung Kim @ 2022-02-08 19:43 UTC (permalink / raw)
To: Peter Zijlstra, Ingo Molnar, Will Deacon, Waiman Long, Boqun Feng
Cc: LKML, Thomas Gleixner, Steven Rostedt, Byungchul Park,
Paul E. McKenney, Mathieu Desnoyers, Radoslaw Burny
In lockdep_init_map_wait(), it didn't pass the given outer argument
to lockdep_init_map_type and use LD_WAIT_INV unconditionally. It
seems like a copy-and-paste bug, let's fix it.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
---
include/linux/lockdep.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/include/linux/lockdep.h b/include/linux/lockdep.h
index 467b94257105..0cc2b338a006 100644
--- a/include/linux/lockdep.h
+++ b/include/linux/lockdep.h
@@ -192,7 +192,7 @@ static inline void
lockdep_init_map_waits(struct lockdep_map *lock, const char *name,
struct lock_class_key *key, int subclass, u8 inner, u8 outer)
{
- lockdep_init_map_type(lock, name, key, subclass, inner, LD_WAIT_INV, LD_LOCK_NORMAL);
+ lockdep_init_map_type(lock, name, key, subclass, inner, outer, LD_LOCK_NORMAL);
}
static inline void
--
2.35.0.263.gb82422642f-goog
^ permalink raw reply related [flat|nested] 22+ messages in thread
* [PATCH 02/12] cgroup: rstat: Make cgroup_rstat_cpu_lock name readable
@ 2022-02-08 19:43 ` Namhyung Kim
0 siblings, 0 replies; 22+ messages in thread
From: Namhyung Kim @ 2022-02-08 19:43 UTC (permalink / raw)
To: Peter Zijlstra, Ingo Molnar, Will Deacon, Waiman Long, Boqun Feng
Cc: LKML, Thomas Gleixner, Steven Rostedt, Byungchul Park,
Paul E. McKenney, Mathieu Desnoyers, Radoslaw Burny, Zefan Li,
Johannes Weiner, cgroups, Tejun Heo
The raw_spin_lock_init() uses the argument to name its lockdep map.
But passing per_cpu_ptr() macro directly makes it a very very long
name as it expanded like below:
({ do { const void *__vpp_verify = (typeof((&cgroup_rstat_cpu_lock) ...
Let's fix it by passing a local variable instead. With this change,
the name now looks like:
cgrp_rstat_cpu_lock
Cc: Zefan Li <lizefan.x@bytedance.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: cgroups@vger.kernel.org
Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
---
kernel/cgroup/rstat.c | 12 ++++++++++--
1 file changed, 10 insertions(+), 2 deletions(-)
diff --git a/kernel/cgroup/rstat.c b/kernel/cgroup/rstat.c
index 9d331ba44870..371f54a59c38 100644
--- a/kernel/cgroup/rstat.c
+++ b/kernel/cgroup/rstat.c
@@ -286,9 +286,17 @@ void cgroup_rstat_exit(struct cgroup *cgrp)
void __init cgroup_rstat_boot(void)
{
int cpu;
+ raw_spinlock_t *cgrp_rstat_cpu_lock;
- for_each_possible_cpu(cpu)
- raw_spin_lock_init(per_cpu_ptr(&cgroup_rstat_cpu_lock, cpu));
+ for_each_possible_cpu(cpu) {
+ /*
+ * raw_spin_lock_init() uses the argument to name the lock in
+ * lockdep results. Passing per_cpu_ptr() directly created a
+ * cryptic name as the macro was expanded before stringified.
+ */
+ cgrp_rstat_cpu_lock = per_cpu_ptr(&cgroup_rstat_cpu_lock, cpu);
+ raw_spin_lock_init(cgrp_rstat_cpu_lock);
+ }
}
/*
--
2.35.0.263.gb82422642f-goog
^ permalink raw reply related [flat|nested] 22+ messages in thread
* [PATCH 02/12] cgroup: rstat: Make cgroup_rstat_cpu_lock name readable
@ 2022-02-08 19:43 ` Namhyung Kim
0 siblings, 0 replies; 22+ messages in thread
From: Namhyung Kim @ 2022-02-08 19:43 UTC (permalink / raw)
To: Peter Zijlstra, Ingo Molnar, Will Deacon, Waiman Long, Boqun Feng
Cc: LKML, Thomas Gleixner, Steven Rostedt, Byungchul Park,
Paul E. McKenney, Mathieu Desnoyers, Radoslaw Burny, Zefan Li,
Johannes Weiner, cgroups-u79uwXL29TY76Z2rM5mHXA, Tejun Heo
The raw_spin_lock_init() uses the argument to name its lockdep map.
But passing per_cpu_ptr() macro directly makes it a very very long
name as it expanded like below:
({ do { const void *__vpp_verify = (typeof((&cgroup_rstat_cpu_lock) ...
Let's fix it by passing a local variable instead. With this change,
the name now looks like:
cgrp_rstat_cpu_lock
Cc: Zefan Li <lizefan.x-EC8Uxl6Npydl57MIdRCFDg@public.gmane.org>
Cc: Johannes Weiner <hannes-druUgvl0LCNAfugRpC6u6w@public.gmane.org>
Cc: cgroups-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
Acked-by: Tejun Heo <tj-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
Signed-off-by: Namhyung Kim <namhyung-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
---
kernel/cgroup/rstat.c | 12 ++++++++++--
1 file changed, 10 insertions(+), 2 deletions(-)
diff --git a/kernel/cgroup/rstat.c b/kernel/cgroup/rstat.c
index 9d331ba44870..371f54a59c38 100644
--- a/kernel/cgroup/rstat.c
+++ b/kernel/cgroup/rstat.c
@@ -286,9 +286,17 @@ void cgroup_rstat_exit(struct cgroup *cgrp)
void __init cgroup_rstat_boot(void)
{
int cpu;
+ raw_spinlock_t *cgrp_rstat_cpu_lock;
- for_each_possible_cpu(cpu)
- raw_spin_lock_init(per_cpu_ptr(&cgroup_rstat_cpu_lock, cpu));
+ for_each_possible_cpu(cpu) {
+ /*
+ * raw_spin_lock_init() uses the argument to name the lock in
+ * lockdep results. Passing per_cpu_ptr() directly created a
+ * cryptic name as the macro was expanded before stringified.
+ */
+ cgrp_rstat_cpu_lock = per_cpu_ptr(&cgroup_rstat_cpu_lock, cpu);
+ raw_spin_lock_init(cgrp_rstat_cpu_lock);
+ }
}
/*
--
2.35.0.263.gb82422642f-goog
^ permalink raw reply related [flat|nested] 22+ messages in thread
* [PATCH 03/12] timer: Protect lockdep functions with #ifdef
2022-02-08 19:43 ` Namhyung Kim
` (3 preceding siblings ...)
(?)
@ 2022-02-08 19:43 ` Namhyung Kim
2022-04-10 22:46 ` Thomas Gleixner
-1 siblings, 1 reply; 22+ messages in thread
From: Namhyung Kim @ 2022-02-08 19:43 UTC (permalink / raw)
To: Peter Zijlstra, Ingo Molnar, Will Deacon, Waiman Long, Boqun Feng
Cc: LKML, Thomas Gleixner, Steven Rostedt, Byungchul Park,
Paul E. McKenney, Mathieu Desnoyers, Radoslaw Burny
With upcoming lock tracepoints config, it'd define some of lockdep
functions without enabling CONFIG_LOCKDEP actually. The existing code
assumes those functions will be removed by the preprocessor but it's
not the case anymore. Let's protect the code with #ifdef's explicitly.
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
---
kernel/time/timer.c | 8 +++++++-
1 file changed, 7 insertions(+), 1 deletion(-)
diff --git a/kernel/time/timer.c b/kernel/time/timer.c
index 85f1021ad459..4af95dbf6435 100644
--- a/kernel/time/timer.c
+++ b/kernel/time/timer.c
@@ -794,7 +794,10 @@ static void do_init_timer(struct timer_list *timer,
if (WARN_ON_ONCE(flags & ~TIMER_INIT_FLAGS))
flags &= TIMER_INIT_FLAGS;
timer->flags = flags | raw_smp_processor_id();
+
+#ifdef CONFIG_LOCKDEP
lockdep_init_map(&timer->lockdep_map, name, key, 0);
+#endif
}
/**
@@ -1409,19 +1412,22 @@ static void call_timer_fn(struct timer_list *timer,
struct lockdep_map lockdep_map;
lockdep_copy_map(&lockdep_map, &timer->lockdep_map);
-#endif
+
/*
* Couple the lock chain with the lock chain at
* del_timer_sync() by acquiring the lock_map around the fn()
* call here and in del_timer_sync().
*/
lock_map_acquire(&lockdep_map);
+#endif
trace_timer_expire_entry(timer, baseclk);
fn(timer);
trace_timer_expire_exit(timer);
+#ifdef CONFIG_LOCKDEP
lock_map_release(&lockdep_map);
+#endif
if (count != preempt_count()) {
WARN_ONCE(1, "timer: %pS preempt leak: %08x -> %08x\n",
--
2.35.0.263.gb82422642f-goog
^ permalink raw reply related [flat|nested] 22+ messages in thread
* [PATCH 04/12] workqueue: Protect lockdep functions with #ifdef
2022-02-08 19:43 ` Namhyung Kim
` (4 preceding siblings ...)
(?)
@ 2022-02-08 19:43 ` Namhyung Kim
-1 siblings, 0 replies; 22+ messages in thread
From: Namhyung Kim @ 2022-02-08 19:43 UTC (permalink / raw)
To: Peter Zijlstra, Ingo Molnar, Will Deacon, Waiman Long, Boqun Feng
Cc: LKML, Thomas Gleixner, Steven Rostedt, Byungchul Park,
Paul E. McKenney, Mathieu Desnoyers, Radoslaw Burny, Tejun Heo,
Lai Jiangshan
With upcoming lock tracepoints config, it'd define some of lockdep
functions without enabling CONFIG_LOCKDEP actually. The existing code
assumes those functions will be removed by the preprocessor but it's
not the case anymore. Let's protect the code with #ifdef's explicitly.
Cc: Tejun Heo <tj@kernel.org>
Cc: Lai Jiangshan <jiangshanlai@gmail.com>
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
---
kernel/workqueue.c | 13 +++++++++++++
1 file changed, 13 insertions(+)
diff --git a/kernel/workqueue.c b/kernel/workqueue.c
index 33f1106b4f99..405e27385f74 100644
--- a/kernel/workqueue.c
+++ b/kernel/workqueue.c
@@ -2279,8 +2279,11 @@ __acquires(&pool->lock)
raw_spin_unlock_irq(&pool->lock);
+#ifdef CONFIG_LOCKDEP
lock_map_acquire(&pwq->wq->lockdep_map);
lock_map_acquire(&lockdep_map);
+#endif
+
/*
* Strictly speaking we should mark the invariant state without holding
* any locks, that is, before these two lock_map_acquire()'s.
@@ -2310,8 +2313,11 @@ __acquires(&pool->lock)
* point will only record its address.
*/
trace_workqueue_execute_end(work, worker->current_func);
+
+#ifdef CONFIG_LOCKDEP
lock_map_release(&lockdep_map);
lock_map_release(&pwq->wq->lockdep_map);
+#endif
if (unlikely(in_atomic() || lockdep_depth(current) > 0)) {
pr_err("BUG: workqueue leaked lock or atomic: %s/0x%08x/%d\n"
@@ -2824,8 +2830,10 @@ void flush_workqueue(struct workqueue_struct *wq)
if (WARN_ON(!wq_online))
return;
+#ifdef CONFIG_LOCKDEP
lock_map_acquire(&wq->lockdep_map);
lock_map_release(&wq->lockdep_map);
+#endif
mutex_lock(&wq->mutex);
@@ -3052,6 +3060,7 @@ static bool start_flush_work(struct work_struct *work, struct wq_barrier *barr,
insert_wq_barrier(pwq, barr, work, worker);
raw_spin_unlock_irq(&pool->lock);
+#ifdef CONFIG_LOCKDEP
/*
* Force a lock recursion deadlock when using flush_work() inside a
* single-threaded or rescuer equipped workqueue.
@@ -3066,6 +3075,8 @@ static bool start_flush_work(struct work_struct *work, struct wq_barrier *barr,
lock_map_acquire(&pwq->wq->lockdep_map);
lock_map_release(&pwq->wq->lockdep_map);
}
+#endif
+
rcu_read_unlock();
return true;
already_gone:
@@ -3084,10 +3095,12 @@ static bool __flush_work(struct work_struct *work, bool from_cancel)
if (WARN_ON(!work->func))
return false;
+#ifdef CONFIG_LOCKDEP
if (!from_cancel) {
lock_map_acquire(&work->lockdep_map);
lock_map_release(&work->lockdep_map);
}
+#endif
if (start_flush_work(work, &barr, from_cancel)) {
wait_for_completion(&barr.done);
--
2.35.0.263.gb82422642f-goog
^ permalink raw reply related [flat|nested] 22+ messages in thread
* [Intel-gfx] [PATCH 05/12] drm/i915: Protect lockdep functions with #ifdef
2022-02-08 19:43 ` Namhyung Kim
@ 2022-02-08 19:43 ` Namhyung Kim
-1 siblings, 0 replies; 22+ messages in thread
From: Namhyung Kim @ 2022-02-08 19:43 UTC (permalink / raw)
To: Peter Zijlstra, Ingo Molnar, Will Deacon, Waiman Long, Boqun Feng
Cc: Paul E. McKenney, intel-gfx, LKML, Steven Rostedt,
Radoslaw Burny, Byungchul Park, Mathieu Desnoyers,
Thomas Gleixner
With upcoming lock tracepoints config, it'd define some of lockdep
functions without enabling CONFIG_LOCKDEP actually. The existing code
assumes those functions will be removed by the preprocessor but it's
not the case anymore. Let's protect the code with #ifdef's explicitly.
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: intel-gfx@lists.freedesktop.org
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
---
drivers/gpu/drm/i915/intel_wakeref.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/drivers/gpu/drm/i915/intel_wakeref.c b/drivers/gpu/drm/i915/intel_wakeref.c
index dfd87d082218..6e4b8d036283 100644
--- a/drivers/gpu/drm/i915/intel_wakeref.c
+++ b/drivers/gpu/drm/i915/intel_wakeref.c
@@ -106,8 +106,11 @@ void __intel_wakeref_init(struct intel_wakeref *wf,
wf->wakeref = 0;
INIT_DELAYED_WORK(&wf->work, __intel_wakeref_put_work);
+
+#ifdef CONFIG_LOCKDEP
lockdep_init_map(&wf->work.work.lockdep_map,
"wakeref.work", &key->work, 0);
+#endif
}
int intel_wakeref_wait_for_idle(struct intel_wakeref *wf)
--
2.35.0.263.gb82422642f-goog
^ permalink raw reply related [flat|nested] 22+ messages in thread
* [PATCH 05/12] drm/i915: Protect lockdep functions with #ifdef
@ 2022-02-08 19:43 ` Namhyung Kim
0 siblings, 0 replies; 22+ messages in thread
From: Namhyung Kim @ 2022-02-08 19:43 UTC (permalink / raw)
To: Peter Zijlstra, Ingo Molnar, Will Deacon, Waiman Long, Boqun Feng
Cc: LKML, Thomas Gleixner, Steven Rostedt, Byungchul Park,
Paul E. McKenney, Mathieu Desnoyers, Radoslaw Burny, Jani Nikula,
Joonas Lahtinen, Rodrigo Vivi, Tvrtko Ursulin, intel-gfx
With upcoming lock tracepoints config, it'd define some of lockdep
functions without enabling CONFIG_LOCKDEP actually. The existing code
assumes those functions will be removed by the preprocessor but it's
not the case anymore. Let's protect the code with #ifdef's explicitly.
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: intel-gfx@lists.freedesktop.org
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
---
drivers/gpu/drm/i915/intel_wakeref.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/drivers/gpu/drm/i915/intel_wakeref.c b/drivers/gpu/drm/i915/intel_wakeref.c
index dfd87d082218..6e4b8d036283 100644
--- a/drivers/gpu/drm/i915/intel_wakeref.c
+++ b/drivers/gpu/drm/i915/intel_wakeref.c
@@ -106,8 +106,11 @@ void __intel_wakeref_init(struct intel_wakeref *wf,
wf->wakeref = 0;
INIT_DELAYED_WORK(&wf->work, __intel_wakeref_put_work);
+
+#ifdef CONFIG_LOCKDEP
lockdep_init_map(&wf->work.work.lockdep_map,
"wakeref.work", &key->work, 0);
+#endif
}
int intel_wakeref_wait_for_idle(struct intel_wakeref *wf)
--
2.35.0.263.gb82422642f-goog
^ permalink raw reply related [flat|nested] 22+ messages in thread
* [PATCH 06/12] btrfs: change lockdep class size check using ks->names
2022-02-08 19:43 ` Namhyung Kim
` (6 preceding siblings ...)
(?)
@ 2022-02-08 19:43 ` Namhyung Kim
-1 siblings, 0 replies; 22+ messages in thread
From: Namhyung Kim @ 2022-02-08 19:43 UTC (permalink / raw)
To: Peter Zijlstra, Ingo Molnar, Will Deacon, Waiman Long, Boqun Feng
Cc: LKML, Thomas Gleixner, Steven Rostedt, Byungchul Park,
Paul E. McKenney, Mathieu Desnoyers, Radoslaw Burny, Chris Mason,
Josef Bacik, linux-btrfs, David Sterba
With upcoming lock tracepoints config, it'd allow some lockdep
annotation code without enabling CONFIG_LOCKDEP actually. In that
config, size of the struct lock_class_key would be 0.
But it'd cause divide-by-zero in btrfs_set_buffer_lockdep_class() due
to ARRAY_SIZE macro. Let's change it to use ks->names[] instead. It
should have the same size as with ks->keys[].
Cc: Chris Mason <clm@fb.com>
Cc: Josef Bacik <josef@toxicpanda.com>
Cc: linux-btrfs@vger.kernel.org
Acked-by: David Sterba <dsterba@suse.com>
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
---
fs/btrfs/disk-io.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index 87a5addbedf6..be41e35bee92 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -190,7 +190,7 @@ void btrfs_set_buffer_lockdep_class(u64 objectid, struct extent_buffer *eb,
{
struct btrfs_lockdep_keyset *ks;
- BUG_ON(level >= ARRAY_SIZE(ks->keys));
+ BUG_ON(level >= ARRAY_SIZE(ks->names));
/* find the matching keyset, id 0 is the default entry */
for (ks = btrfs_lockdep_keysets; ks->id; ks++)
--
2.35.0.263.gb82422642f-goog
^ permalink raw reply related [flat|nested] 22+ messages in thread
* [PATCH 07/12] locking: Introduce CONFIG_LOCK_INFO
2022-02-08 19:43 ` Namhyung Kim
` (7 preceding siblings ...)
(?)
@ 2022-02-08 19:43 ` Namhyung Kim
-1 siblings, 0 replies; 22+ messages in thread
From: Namhyung Kim @ 2022-02-08 19:43 UTC (permalink / raw)
To: Peter Zijlstra, Ingo Molnar, Will Deacon, Waiman Long, Boqun Feng
Cc: LKML, Thomas Gleixner, Steven Rostedt, Byungchul Park,
Paul E. McKenney, Mathieu Desnoyers, Radoslaw Burny
This is a preparatory work to have tracepoints separated from the
lockdep. I'd like to keep minimal information of locks (name, for
now) to be used by the tracepoints in the lockdep_map structure.
To make the work easier, CONFIG_LOCK_INFO was added to indicate that
it needs to save the lock info. And convert existing code using lock
information to depend on it rather than CONFIG_LOCKDEP and/or
CONFIG_DEBUG_LOCK_ALLOC directly. Future users of the lock
information should select it too.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
---
drivers/gpu/drm/drm_connector.c | 2 +-
drivers/gpu/drm/i915/i915_sw_fence.h | 2 +-
drivers/gpu/drm/i915/selftests/lib_sw_fence.h | 2 +-
drivers/net/wireless/intel/iwlwifi/iwl-trans.c | 4 ++--
drivers/net/wireless/intel/iwlwifi/iwl-trans.h | 2 +-
drivers/tty/tty_ldsem.c | 2 +-
fs/btrfs/disk-io.c | 2 +-
fs/btrfs/disk-io.h | 2 +-
fs/cifs/connect.c | 2 +-
fs/kernfs/file.c | 2 +-
include/linux/completion.h | 2 +-
include/linux/jbd2.h | 2 +-
include/linux/kernfs.h | 2 +-
include/linux/kthread.h | 2 +-
include/linux/local_lock_internal.h | 8 ++++----
include/linux/mmu_notifier.h | 2 +-
include/linux/mutex.h | 8 ++++----
include/linux/percpu-rwsem.h | 4 ++--
include/linux/regmap.h | 4 ++--
include/linux/rtmutex.h | 6 +++---
include/linux/rwlock_api_smp.h | 4 ++--
include/linux/rwlock_rt.h | 4 ++--
include/linux/rwlock_types.h | 6 +++---
include/linux/rwsem.h | 6 +++---
include/linux/seqlock.h | 4 ++--
include/linux/spinlock_api_smp.h | 4 ++--
include/linux/spinlock_rt.h | 4 ++--
include/linux/spinlock_types.h | 4 ++--
include/linux/spinlock_types_raw.h | 4 ++--
include/linux/swait.h | 2 +-
include/linux/tty_ldisc.h | 2 +-
include/linux/wait.h | 2 +-
include/linux/ww_mutex.h | 6 +++---
include/media/v4l2-ctrls.h | 2 +-
include/net/sock.h | 2 +-
kernel/locking/mutex-debug.c | 2 +-
kernel/locking/mutex.c | 16 ++++++++--------
kernel/locking/percpu-rwsem.c | 2 +-
kernel/locking/rtmutex_api.c | 10 +++++-----
kernel/locking/rwsem.c | 4 ++--
kernel/locking/spinlock.c | 2 +-
kernel/locking/spinlock_debug.c | 4 ++--
kernel/locking/spinlock_rt.c | 8 ++++----
kernel/locking/ww_rt_mutex.c | 2 +-
kernel/printk/printk.c | 4 ++--
lib/Kconfig.debug | 5 +++++
mm/memcontrol.c | 2 +-
mm/mmu_notifier.c | 2 +-
net/core/dev.c | 2 +-
net/sunrpc/svcsock.c | 2 +-
net/sunrpc/xprtsock.c | 2 +-
51 files changed, 96 insertions(+), 91 deletions(-)
diff --git a/drivers/gpu/drm/drm_connector.c b/drivers/gpu/drm/drm_connector.c
index a50c82bc2b2f..94931b32a491 100644
--- a/drivers/gpu/drm/drm_connector.c
+++ b/drivers/gpu/drm/drm_connector.c
@@ -676,7 +676,7 @@ const char *drm_get_connector_force_name(enum drm_connector_force force)
}
}
-#ifdef CONFIG_LOCKDEP
+#ifdef CONFIG_LOCK_INFO
static struct lockdep_map connector_list_iter_dep_map = {
.name = "drm_connector_list_iter"
};
diff --git a/drivers/gpu/drm/i915/i915_sw_fence.h b/drivers/gpu/drm/i915/i915_sw_fence.h
index a7c603bc1b01..8c05d161a069 100644
--- a/drivers/gpu/drm/i915/i915_sw_fence.h
+++ b/drivers/gpu/drm/i915/i915_sw_fence.h
@@ -43,7 +43,7 @@ void __i915_sw_fence_init(struct i915_sw_fence *fence,
i915_sw_fence_notify_t fn,
const char *name,
struct lock_class_key *key);
-#ifdef CONFIG_LOCKDEP
+#ifdef CONFIG_LOCK_INFO
#define i915_sw_fence_init(fence, fn) \
do { \
static struct lock_class_key __key; \
diff --git a/drivers/gpu/drm/i915/selftests/lib_sw_fence.h b/drivers/gpu/drm/i915/selftests/lib_sw_fence.h
index e54d6bc23dc3..ad7de5187830 100644
--- a/drivers/gpu/drm/i915/selftests/lib_sw_fence.h
+++ b/drivers/gpu/drm/i915/selftests/lib_sw_fence.h
@@ -12,7 +12,7 @@
#include "../i915_sw_fence.h"
-#ifdef CONFIG_LOCKDEP
+#ifdef CONFIG_LOCK_INFO
#define onstack_fence_init(fence) \
do { \
static struct lock_class_key __key; \
diff --git a/drivers/net/wireless/intel/iwlwifi/iwl-trans.c b/drivers/net/wireless/intel/iwlwifi/iwl-trans.c
index 9236f9106826..c0e7038a939f 100644
--- a/drivers/net/wireless/intel/iwlwifi/iwl-trans.c
+++ b/drivers/net/wireless/intel/iwlwifi/iwl-trans.c
@@ -21,7 +21,7 @@ struct iwl_trans *iwl_trans_alloc(unsigned int priv_size,
const struct iwl_cfg_trans_params *cfg_trans)
{
struct iwl_trans *trans;
-#ifdef CONFIG_LOCKDEP
+#ifdef CONFIG_LOCK_INFO
static struct lock_class_key __key;
#endif
@@ -31,7 +31,7 @@ struct iwl_trans *iwl_trans_alloc(unsigned int priv_size,
trans->trans_cfg = cfg_trans;
-#ifdef CONFIG_LOCKDEP
+#ifdef CONFIG_LOCK_INFO
lockdep_init_map(&trans->sync_cmd_lockdep_map, "sync_cmd_lockdep_map",
&__key, 0);
#endif
diff --git a/drivers/net/wireless/intel/iwlwifi/iwl-trans.h b/drivers/net/wireless/intel/iwlwifi/iwl-trans.h
index 1bcaa3598785..47ef1e852d85 100644
--- a/drivers/net/wireless/intel/iwlwifi/iwl-trans.h
+++ b/drivers/net/wireless/intel/iwlwifi/iwl-trans.h
@@ -1011,7 +1011,7 @@ struct iwl_trans {
struct dentry *dbgfs_dir;
-#ifdef CONFIG_LOCKDEP
+#ifdef CONFIG_LOCK_INFO
struct lockdep_map sync_cmd_lockdep_map;
#endif
diff --git a/drivers/tty/tty_ldsem.c b/drivers/tty/tty_ldsem.c
index 3be428c16260..87c44f3f0c27 100644
--- a/drivers/tty/tty_ldsem.c
+++ b/drivers/tty/tty_ldsem.c
@@ -57,7 +57,7 @@ struct ldsem_waiter {
void __init_ldsem(struct ld_semaphore *sem, const char *name,
struct lock_class_key *key)
{
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
/*
* Make sure we are not reinitializing a held semaphore:
*/
diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index be41e35bee92..2a3a257ced49 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -144,7 +144,7 @@ struct async_submit_bio {
* same as our lockdep setup here. If BTRFS_MAX_LEVEL changes, this code
* needs update as well.
*/
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
# if BTRFS_MAX_LEVEL != 8
# error
# endif
diff --git a/fs/btrfs/disk-io.h b/fs/btrfs/disk-io.h
index 5e8bef4b7563..31ff5e95cb92 100644
--- a/fs/btrfs/disk-io.h
+++ b/fs/btrfs/disk-io.h
@@ -148,7 +148,7 @@ int btrfs_init_root_free_objectid(struct btrfs_root *root);
int __init btrfs_end_io_wq_init(void);
void __cold btrfs_end_io_wq_exit(void);
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
void btrfs_set_buffer_lockdep_class(u64 objectid,
struct extent_buffer *eb, int level);
#else
diff --git a/fs/cifs/connect.c b/fs/cifs/connect.c
index 11a22a30ee14..b117027019b2 100644
--- a/fs/cifs/connect.c
+++ b/fs/cifs/connect.c
@@ -2573,7 +2573,7 @@ cifs_match_super(struct super_block *sb, void *data)
return rc;
}
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
static struct lock_class_key cifs_key[2];
static struct lock_class_key cifs_slock_key[2];
diff --git a/fs/kernfs/file.c b/fs/kernfs/file.c
index 9414a7a60a9f..5cb58ec61ba8 100644
--- a/fs/kernfs/file.c
+++ b/fs/kernfs/file.c
@@ -994,7 +994,7 @@ struct kernfs_node *__kernfs_create_file(struct kernfs_node *parent,
kn->ns = ns;
kn->priv = priv;
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
if (key) {
lockdep_init_map(&kn->dep_map, "kn->active", key, 0);
kn->flags |= KERNFS_LOCKDEP;
diff --git a/include/linux/completion.h b/include/linux/completion.h
index 51d9ab079629..b6c408e62291 100644
--- a/include/linux/completion.h
+++ b/include/linux/completion.h
@@ -64,7 +64,7 @@ static inline void complete_release(struct completion *x) {}
* This macro declares and initializes a completion structure on the kernel
* stack.
*/
-#ifdef CONFIG_LOCKDEP
+#ifdef CONFIG_LOCK_INFO
# define DECLARE_COMPLETION_ONSTACK(work) \
struct completion work = COMPLETION_INITIALIZER_ONSTACK(work)
# define DECLARE_COMPLETION_ONSTACK_MAP(work, map) \
diff --git a/include/linux/jbd2.h b/include/linux/jbd2.h
index fd933c45281a..990a87ff8ab0 100644
--- a/include/linux/jbd2.h
+++ b/include/linux/jbd2.h
@@ -1275,7 +1275,7 @@ struct journal_s
*/
__u32 j_csum_seed;
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
/**
* @j_trans_commit_map:
*
diff --git a/include/linux/kernfs.h b/include/linux/kernfs.h
index 861c4f0f8a29..48c5c02395cf 100644
--- a/include/linux/kernfs.h
+++ b/include/linux/kernfs.h
@@ -131,7 +131,7 @@ struct kernfs_elem_attr {
struct kernfs_node {
atomic_t count;
atomic_t active;
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
struct lockdep_map dep_map;
#endif
/*
diff --git a/include/linux/kthread.h b/include/linux/kthread.h
index 3df4ea04716f..d0d5ca007c7a 100644
--- a/include/linux/kthread.h
+++ b/include/linux/kthread.h
@@ -172,7 +172,7 @@ struct kthread_delayed_work {
* kthread_worker.lock needs its own lockdep class key when defined on
* stack with lockdep enabled. Use the following macros in such cases.
*/
-#ifdef CONFIG_LOCKDEP
+#ifdef CONFIG_LOCK_INFO
# define KTHREAD_WORKER_INIT_ONSTACK(worker) \
({ kthread_init_worker(&worker); worker; })
# define DEFINE_KTHREAD_WORKER_ONSTACK(worker) \
diff --git a/include/linux/local_lock_internal.h b/include/linux/local_lock_internal.h
index 975e33b793a7..56f03f588aa7 100644
--- a/include/linux/local_lock_internal.h
+++ b/include/linux/local_lock_internal.h
@@ -9,13 +9,13 @@
#ifndef CONFIG_PREEMPT_RT
typedef struct {
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
struct lockdep_map dep_map;
struct task_struct *owner;
#endif
} local_lock_t;
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
# define LOCAL_LOCK_DEBUG_INIT(lockname) \
.dep_map = { \
.name = #lockname, \
@@ -42,12 +42,12 @@ static inline void local_lock_debug_init(local_lock_t *l)
{
l->owner = NULL;
}
-#else /* CONFIG_DEBUG_LOCK_ALLOC */
+#else /* CONFIG_LOCK_INFO */
# define LOCAL_LOCK_DEBUG_INIT(lockname)
static inline void local_lock_acquire(local_lock_t *l) { }
static inline void local_lock_release(local_lock_t *l) { }
static inline void local_lock_debug_init(local_lock_t *l) { }
-#endif /* !CONFIG_DEBUG_LOCK_ALLOC */
+#endif /* !CONFIG_LOCK_INFO */
#define INIT_LOCAL_LOCK(lockname) { LOCAL_LOCK_DEBUG_INIT(lockname) }
diff --git a/include/linux/mmu_notifier.h b/include/linux/mmu_notifier.h
index 45fc2c81e370..ff88b1f5b173 100644
--- a/include/linux/mmu_notifier.h
+++ b/include/linux/mmu_notifier.h
@@ -264,7 +264,7 @@ struct mmu_interval_notifier {
#ifdef CONFIG_MMU_NOTIFIER
-#ifdef CONFIG_LOCKDEP
+#ifdef CONFIG_LOCK_INFO
extern struct lockdep_map __mmu_notifier_invalidate_range_start_map;
#endif
diff --git a/include/linux/mutex.h b/include/linux/mutex.h
index 8f226d460f51..369c1abbf3d0 100644
--- a/include/linux/mutex.h
+++ b/include/linux/mutex.h
@@ -20,7 +20,7 @@
#include <linux/osq_lock.h>
#include <linux/debug_locks.h>
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
# define __DEP_MAP_MUTEX_INITIALIZER(lockname) \
, .dep_map = { \
.name = #lockname, \
@@ -70,7 +70,7 @@ struct mutex {
#ifdef CONFIG_DEBUG_MUTEXES
void *magic;
#endif
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
struct lockdep_map dep_map;
#endif
};
@@ -134,7 +134,7 @@ extern bool mutex_is_locked(struct mutex *lock);
struct mutex {
struct rt_mutex_base rtmutex;
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
struct lockdep_map dep_map;
#endif
};
@@ -174,7 +174,7 @@ do { \
* See kernel/locking/mutex.c for detailed documentation of these APIs.
* Also see Documentation/locking/mutex-design.rst.
*/
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
extern void mutex_lock_nested(struct mutex *lock, unsigned int subclass);
extern void _mutex_lock_nest_lock(struct mutex *lock, struct lockdep_map *nest_lock);
diff --git a/include/linux/percpu-rwsem.h b/include/linux/percpu-rwsem.h
index 5fda40f97fe9..9d2427579d9a 100644
--- a/include/linux/percpu-rwsem.h
+++ b/include/linux/percpu-rwsem.h
@@ -15,12 +15,12 @@ struct percpu_rw_semaphore {
struct rcuwait writer;
wait_queue_head_t waiters;
atomic_t block;
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
struct lockdep_map dep_map;
#endif
};
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
#define __PERCPU_RWSEM_DEP_MAP_INIT(lockname) .dep_map = { .name = #lockname },
#else
#define __PERCPU_RWSEM_DEP_MAP_INIT(lockname)
diff --git a/include/linux/regmap.h b/include/linux/regmap.h
index 22652e5fbc38..174493a0512a 100644
--- a/include/linux/regmap.h
+++ b/include/linux/regmap.h
@@ -666,12 +666,12 @@ struct regmap *__devm_regmap_init_spi_avmm(struct spi_device *spi,
const char *lock_name);
/*
* Wrapper for regmap_init macros to include a unique lockdep key and name
- * for each call. No-op if CONFIG_LOCKDEP is not set.
+ * for each call. No-op if CONFIG_LOCK_INFO is not set.
*
* @fn: Real function to call (in the form __[*_]regmap_init[_*])
* @name: Config variable name (#config in the calling macro)
**/
-#ifdef CONFIG_LOCKDEP
+#ifdef CONFIG_LOCK_INFO
#define __regmap_lockdep_wrapper(fn, name, ...) \
( \
({ \
diff --git a/include/linux/rtmutex.h b/include/linux/rtmutex.h
index 7d049883a08a..8eafdd6dcf35 100644
--- a/include/linux/rtmutex.h
+++ b/include/linux/rtmutex.h
@@ -56,7 +56,7 @@ extern void rt_mutex_base_init(struct rt_mutex_base *rtb);
*/
struct rt_mutex {
struct rt_mutex_base rtmutex;
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
struct lockdep_map dep_map;
#endif
};
@@ -76,7 +76,7 @@ do { \
__rt_mutex_init(mutex, __func__, &__key); \
} while (0)
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
#define __DEP_MAP_RT_MUTEX_INITIALIZER(mutexname) \
.dep_map = { \
.name = #mutexname, \
@@ -97,7 +97,7 @@ do { \
extern void __rt_mutex_init(struct rt_mutex *lock, const char *name, struct lock_class_key *key);
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
extern void rt_mutex_lock_nested(struct rt_mutex *lock, unsigned int subclass);
extern void _rt_mutex_lock_nest_lock(struct rt_mutex *lock, struct lockdep_map *nest_lock);
#define rt_mutex_lock(lock) rt_mutex_lock_nested(lock, 0)
diff --git a/include/linux/rwlock_api_smp.h b/include/linux/rwlock_api_smp.h
index dceb0a59b692..7fb42c921669 100644
--- a/include/linux/rwlock_api_smp.h
+++ b/include/linux/rwlock_api_smp.h
@@ -142,7 +142,7 @@ static inline int __raw_write_trylock(rwlock_t *lock)
* even on CONFIG_PREEMPT, because lockdep assumes that interrupts are
* not re-enabled during lock-acquire (which the preempt-spin-ops do):
*/
-#if !defined(CONFIG_GENERIC_LOCKBREAK) || defined(CONFIG_DEBUG_LOCK_ALLOC)
+#if !defined(CONFIG_GENERIC_LOCKBREAK) || defined(CONFIG_LOCK_INFO)
static inline void __raw_read_lock(rwlock_t *lock)
{
@@ -217,7 +217,7 @@ static inline void __raw_write_lock_nested(rwlock_t *lock, int subclass)
LOCK_CONTENDED(lock, do_raw_write_trylock, do_raw_write_lock);
}
-#endif /* !CONFIG_GENERIC_LOCKBREAK || CONFIG_DEBUG_LOCK_ALLOC */
+#endif /* !CONFIG_GENERIC_LOCKBREAK || CONFIG_LOCK_INFO */
static inline void __raw_write_unlock(rwlock_t *lock)
{
diff --git a/include/linux/rwlock_rt.h b/include/linux/rwlock_rt.h
index 8544ff05e594..140c418c51b5 100644
--- a/include/linux/rwlock_rt.h
+++ b/include/linux/rwlock_rt.h
@@ -6,7 +6,7 @@
#error Do not #include directly. Use <linux/spinlock.h>.
#endif
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
extern void __rt_rwlock_init(rwlock_t *rwlock, const char *name,
struct lock_class_key *key);
#else
@@ -84,7 +84,7 @@ static __always_inline void write_lock(rwlock_t *rwlock)
rt_write_lock(rwlock);
}
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
static __always_inline void write_lock_nested(rwlock_t *rwlock, int subclass)
{
rt_write_lock_nested(rwlock, subclass);
diff --git a/include/linux/rwlock_types.h b/include/linux/rwlock_types.h
index 1948442e7750..3e621bfd7cd0 100644
--- a/include/linux/rwlock_types.h
+++ b/include/linux/rwlock_types.h
@@ -5,7 +5,7 @@
# error "Do not include directly, include spinlock_types.h"
#endif
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
# define RW_DEP_MAP_INIT(lockname) \
.dep_map = { \
.name = #lockname, \
@@ -28,7 +28,7 @@ typedef struct {
unsigned int magic, owner_cpu;
void *owner;
#endif
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
struct lockdep_map dep_map;
#endif
} rwlock_t;
@@ -57,7 +57,7 @@ typedef struct {
typedef struct {
struct rwbase_rt rwbase;
atomic_t readers;
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
struct lockdep_map dep_map;
#endif
} rwlock_t;
diff --git a/include/linux/rwsem.h b/include/linux/rwsem.h
index f9348769e558..c488485861f5 100644
--- a/include/linux/rwsem.h
+++ b/include/linux/rwsem.h
@@ -16,7 +16,7 @@
#include <linux/atomic.h>
#include <linux/err.h>
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
# define __RWSEM_DEP_MAP_INIT(lockname) \
.dep_map = { \
.name = #lockname, \
@@ -60,7 +60,7 @@ struct rw_semaphore {
#ifdef CONFIG_DEBUG_RWSEMS
void *magic;
#endif
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
struct lockdep_map dep_map;
#endif
};
@@ -127,7 +127,7 @@ static inline int rwsem_is_contended(struct rw_semaphore *sem)
struct rw_semaphore {
struct rwbase_rt rwbase;
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
struct lockdep_map dep_map;
#endif
};
diff --git a/include/linux/seqlock.h b/include/linux/seqlock.h
index 37ded6b8fee6..c673f807965e 100644
--- a/include/linux/seqlock.h
+++ b/include/linux/seqlock.h
@@ -64,7 +64,7 @@
*/
typedef struct seqcount {
unsigned sequence;
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
struct lockdep_map dep_map;
#endif
} seqcount_t;
@@ -79,7 +79,7 @@ static inline void __seqcount_init(seqcount_t *s, const char *name,
s->sequence = 0;
}
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
# define SEQCOUNT_DEP_MAP_INIT(lockname) \
.dep_map = { .name = #lockname }
diff --git a/include/linux/spinlock_api_smp.h b/include/linux/spinlock_api_smp.h
index 51fa0dab68c4..94e5ddbcc2d1 100644
--- a/include/linux/spinlock_api_smp.h
+++ b/include/linux/spinlock_api_smp.h
@@ -99,7 +99,7 @@ static inline int __raw_spin_trylock(raw_spinlock_t *lock)
* even on CONFIG_PREEMPTION, because lockdep assumes that interrupts are
* not re-enabled during lock-acquire (which the preempt-spin-ops do):
*/
-#if !defined(CONFIG_GENERIC_LOCKBREAK) || defined(CONFIG_DEBUG_LOCK_ALLOC)
+#if !defined(CONFIG_GENERIC_LOCKBREAK) || defined(CONFIG_LOCK_INFO)
static inline unsigned long __raw_spin_lock_irqsave(raw_spinlock_t *lock)
{
@@ -134,7 +134,7 @@ static inline void __raw_spin_lock(raw_spinlock_t *lock)
LOCK_CONTENDED(lock, do_raw_spin_trylock, do_raw_spin_lock);
}
-#endif /* !CONFIG_GENERIC_LOCKBREAK || CONFIG_DEBUG_LOCK_ALLOC */
+#endif /* !CONFIG_GENERIC_LOCKBREAK || CONFIG_LOCK_INFO */
static inline void __raw_spin_unlock(raw_spinlock_t *lock)
{
diff --git a/include/linux/spinlock_rt.h b/include/linux/spinlock_rt.h
index 835aedaf68ac..2605668e0fdd 100644
--- a/include/linux/spinlock_rt.h
+++ b/include/linux/spinlock_rt.h
@@ -6,7 +6,7 @@
#error Do not include directly. Use spinlock.h
#endif
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
extern void __rt_spin_lock_init(spinlock_t *lock, const char *name,
struct lock_class_key *key, bool percpu);
#else
@@ -45,7 +45,7 @@ static __always_inline void spin_lock(spinlock_t *lock)
rt_spin_lock(lock);
}
-#ifdef CONFIG_LOCKDEP
+#ifdef CONFIG_LOCK_INFO
# define __spin_lock_nested(lock, subclass) \
rt_spin_lock_nested(lock, subclass)
diff --git a/include/linux/spinlock_types.h b/include/linux/spinlock_types.h
index 2dfa35ffec76..fb9e778a0ee5 100644
--- a/include/linux/spinlock_types.h
+++ b/include/linux/spinlock_types.h
@@ -18,7 +18,7 @@ typedef struct spinlock {
union {
struct raw_spinlock rlock;
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
# define LOCK_PADSIZE (offsetof(struct raw_spinlock, dep_map))
struct {
u8 __padding[LOCK_PADSIZE];
@@ -49,7 +49,7 @@ typedef struct spinlock {
typedef struct spinlock {
struct rt_mutex_base lock;
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
struct lockdep_map dep_map;
#endif
} spinlock_t;
diff --git a/include/linux/spinlock_types_raw.h b/include/linux/spinlock_types_raw.h
index 91cb36b65a17..564092a30cc4 100644
--- a/include/linux/spinlock_types_raw.h
+++ b/include/linux/spinlock_types_raw.h
@@ -17,7 +17,7 @@ typedef struct raw_spinlock {
unsigned int magic, owner_cpu;
void *owner;
#endif
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
struct lockdep_map dep_map;
#endif
} raw_spinlock_t;
@@ -26,7 +26,7 @@ typedef struct raw_spinlock {
#define SPINLOCK_OWNER_INIT ((void *)-1L)
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
# define RAW_SPIN_DEP_MAP_INIT(lockname) \
.dep_map = { \
.name = #lockname, \
diff --git a/include/linux/swait.h b/include/linux/swait.h
index 6a8c22b8c2a5..643c9fe68d63 100644
--- a/include/linux/swait.h
+++ b/include/linux/swait.h
@@ -75,7 +75,7 @@ extern void __init_swait_queue_head(struct swait_queue_head *q, const char *name
__init_swait_queue_head((q), #q, &__key); \
} while (0)
-#ifdef CONFIG_LOCKDEP
+#ifdef CONFIG_LOCK_INFO
# define __SWAIT_QUEUE_HEAD_INIT_ONSTACK(name) \
({ init_swait_queue_head(&name); name; })
# define DECLARE_SWAIT_QUEUE_HEAD_ONSTACK(name) \
diff --git a/include/linux/tty_ldisc.h b/include/linux/tty_ldisc.h
index e85002b56752..5af6fb3649ab 100644
--- a/include/linux/tty_ldisc.h
+++ b/include/linux/tty_ldisc.h
@@ -20,7 +20,7 @@ struct ld_semaphore {
unsigned int wait_readers;
struct list_head read_wait;
struct list_head write_wait;
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
struct lockdep_map dep_map;
#endif
};
diff --git a/include/linux/wait.h b/include/linux/wait.h
index 851e07da2583..aa811a05b070 100644
--- a/include/linux/wait.h
+++ b/include/linux/wait.h
@@ -70,7 +70,7 @@ extern void __init_waitqueue_head(struct wait_queue_head *wq_head, const char *n
__init_waitqueue_head((wq_head), #wq_head, &__key); \
} while (0)
-#ifdef CONFIG_LOCKDEP
+#ifdef CONFIG_LOCK_INFO
# define __WAIT_QUEUE_HEAD_INIT_ONSTACK(name) \
({ init_waitqueue_head(&name); name; })
# define DECLARE_WAIT_QUEUE_HEAD_ONSTACK(name) \
diff --git a/include/linux/ww_mutex.h b/include/linux/ww_mutex.h
index bb763085479a..12a6ad18176d 100644
--- a/include/linux/ww_mutex.h
+++ b/include/linux/ww_mutex.h
@@ -63,7 +63,7 @@ struct ww_acquire_ctx {
struct ww_class *ww_class;
void *contending_lock;
#endif
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
struct lockdep_map dep_map;
#endif
#ifdef CONFIG_DEBUG_WW_MUTEX_SLOWPATH
@@ -142,7 +142,7 @@ static inline void ww_acquire_init(struct ww_acquire_ctx *ctx,
ctx->done_acquire = 0;
ctx->contending_lock = NULL;
#endif
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
debug_check_no_locks_freed((void *)ctx, sizeof(*ctx));
lockdep_init_map(&ctx->dep_map, ww_class->acquire_name,
&ww_class->acquire_key, 0);
@@ -184,7 +184,7 @@ static inline void ww_acquire_done(struct ww_acquire_ctx *ctx)
*/
static inline void ww_acquire_fini(struct ww_acquire_ctx *ctx)
{
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
mutex_release(&ctx->dep_map, _THIS_IP_);
#endif
#ifdef DEBUG_WW_MUTEXES
diff --git a/include/media/v4l2-ctrls.h b/include/media/v4l2-ctrls.h
index b3ce438f1329..f2c30a4fe203 100644
--- a/include/media/v4l2-ctrls.h
+++ b/include/media/v4l2-ctrls.h
@@ -489,7 +489,7 @@ int v4l2_ctrl_handler_init_class(struct v4l2_ctrl_handler *hdl,
unsigned int nr_of_controls_hint,
struct lock_class_key *key, const char *name);
-#ifdef CONFIG_LOCKDEP
+#ifdef CONFIG_LOCK_INFO
/**
* v4l2_ctrl_handler_init - helper function to create a static struct
diff --git a/include/net/sock.h b/include/net/sock.h
index ff9b508d9c5f..e88c7de283ed 100644
--- a/include/net/sock.h
+++ b/include/net/sock.h
@@ -103,7 +103,7 @@ typedef struct {
* the slock as a lock variant (in addition to
* the slock itself):
*/
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
struct lockdep_map dep_map;
#endif
} socket_lock_t;
diff --git a/kernel/locking/mutex-debug.c b/kernel/locking/mutex-debug.c
index bc8abb8549d2..33fc3c06b714 100644
--- a/kernel/locking/mutex-debug.c
+++ b/kernel/locking/mutex-debug.c
@@ -79,7 +79,7 @@ void debug_mutex_unlock(struct mutex *lock)
void debug_mutex_init(struct mutex *lock, const char *name,
struct lock_class_key *key)
{
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
/*
* Make sure we are not reinitializing a held lock:
*/
diff --git a/kernel/locking/mutex.c b/kernel/locking/mutex.c
index 5e3585950ec8..8733b96ce20a 100644
--- a/kernel/locking/mutex.c
+++ b/kernel/locking/mutex.c
@@ -149,7 +149,7 @@ static inline bool __mutex_trylock(struct mutex *lock)
return !__mutex_trylock_common(lock, false);
}
-#ifndef CONFIG_DEBUG_LOCK_ALLOC
+#ifndef CONFIG_LOCK_INFO
/*
* Lockdep annotations are contained to the slow paths for simplicity.
* There is nothing that would stop spreading the lockdep annotations outwards
@@ -245,7 +245,7 @@ static void __mutex_handoff(struct mutex *lock, struct task_struct *task)
}
}
-#ifndef CONFIG_DEBUG_LOCK_ALLOC
+#ifndef CONFIG_LOCK_INFO
/*
* We split the mutex lock/unlock logic into separate fastpath and
* slowpath functions, to reduce the register pressure on the fastpath.
@@ -533,7 +533,7 @@ static noinline void __sched __mutex_unlock_slowpath(struct mutex *lock, unsigne
*/
void __sched mutex_unlock(struct mutex *lock)
{
-#ifndef CONFIG_DEBUG_LOCK_ALLOC
+#ifndef CONFIG_LOCK_INFO
if (__mutex_unlock_fast(lock))
return;
#endif
@@ -591,7 +591,7 @@ __mutex_lock_common(struct mutex *lock, unsigned int state, unsigned int subclas
if (ww_ctx->acquired == 0)
ww_ctx->wounded = 0;
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
nest_lock = &ww_ctx->dep_map;
#endif
}
@@ -778,7 +778,7 @@ int ww_mutex_trylock(struct ww_mutex *ww, struct ww_acquire_ctx *ww_ctx)
}
EXPORT_SYMBOL(ww_mutex_trylock);
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
void __sched
mutex_lock_nested(struct mutex *lock, unsigned int subclass)
{
@@ -937,7 +937,7 @@ static noinline void __sched __mutex_unlock_slowpath(struct mutex *lock, unsigne
wake_up_q(&wake_q);
}
-#ifndef CONFIG_DEBUG_LOCK_ALLOC
+#ifndef CONFIG_LOCK_INFO
/*
* Here come the less common (and hence less performance-critical) APIs:
* mutex_lock_interruptible() and mutex_trylock().
@@ -1078,7 +1078,7 @@ int __sched mutex_trylock(struct mutex *lock)
}
EXPORT_SYMBOL(mutex_trylock);
-#ifndef CONFIG_DEBUG_LOCK_ALLOC
+#ifndef CONFIG_LOCK_INFO
int __sched
ww_mutex_lock(struct ww_mutex *lock, struct ww_acquire_ctx *ctx)
{
@@ -1109,7 +1109,7 @@ ww_mutex_lock_interruptible(struct ww_mutex *lock, struct ww_acquire_ctx *ctx)
}
EXPORT_SYMBOL(ww_mutex_lock_interruptible);
-#endif /* !CONFIG_DEBUG_LOCK_ALLOC */
+#endif /* !CONFIG_LOCK_INFO */
#endif /* !CONFIG_PREEMPT_RT */
/**
diff --git a/kernel/locking/percpu-rwsem.c b/kernel/locking/percpu-rwsem.c
index 70a32a576f3f..98ff434a5f95 100644
--- a/kernel/locking/percpu-rwsem.c
+++ b/kernel/locking/percpu-rwsem.c
@@ -20,7 +20,7 @@ int __percpu_init_rwsem(struct percpu_rw_semaphore *sem,
rcuwait_init(&sem->writer);
init_waitqueue_head(&sem->waiters);
atomic_set(&sem->block, 0);
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
debug_check_no_locks_freed((void *)sem, sizeof(*sem));
lockdep_init_map(&sem->dep_map, name, key, 0);
#endif
diff --git a/kernel/locking/rtmutex_api.c b/kernel/locking/rtmutex_api.c
index 900220941caa..ce08dabf4f93 100644
--- a/kernel/locking/rtmutex_api.c
+++ b/kernel/locking/rtmutex_api.c
@@ -40,7 +40,7 @@ void rt_mutex_base_init(struct rt_mutex_base *rtb)
}
EXPORT_SYMBOL(rt_mutex_base_init);
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
/**
* rt_mutex_lock_nested - lock a rt_mutex
*
@@ -59,7 +59,7 @@ void __sched _rt_mutex_lock_nest_lock(struct rt_mutex *lock, struct lockdep_map
}
EXPORT_SYMBOL_GPL(_rt_mutex_lock_nest_lock);
-#else /* !CONFIG_DEBUG_LOCK_ALLOC */
+#else /* !CONFIG_LOCK_INFO */
/**
* rt_mutex_lock - lock a rt_mutex
@@ -517,7 +517,7 @@ static __always_inline int __mutex_lock_common(struct mutex *lock,
return ret;
}
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
void __sched mutex_lock_nested(struct mutex *lock, unsigned int subclass)
{
__mutex_lock_common(lock, TASK_UNINTERRUPTIBLE, subclass, NULL, _RET_IP_);
@@ -557,7 +557,7 @@ void __sched mutex_lock_io_nested(struct mutex *lock, unsigned int subclass)
}
EXPORT_SYMBOL_GPL(mutex_lock_io_nested);
-#else /* CONFIG_DEBUG_LOCK_ALLOC */
+#else /* CONFIG_LOCK_INFO */
void __sched mutex_lock(struct mutex *lock)
{
@@ -585,7 +585,7 @@ void __sched mutex_lock_io(struct mutex *lock)
io_schedule_finish(token);
}
EXPORT_SYMBOL(mutex_lock_io);
-#endif /* !CONFIG_DEBUG_LOCK_ALLOC */
+#endif /* !CONFIG_LOCK_INFO */
int __sched mutex_trylock(struct mutex *lock)
{
diff --git a/kernel/locking/rwsem.c b/kernel/locking/rwsem.c
index 69aba4abe104..8da694940165 100644
--- a/kernel/locking/rwsem.c
+++ b/kernel/locking/rwsem.c
@@ -304,7 +304,7 @@ rwsem_owner_flags(struct rw_semaphore *sem, unsigned long *pflags)
void __init_rwsem(struct rw_semaphore *sem, const char *name,
struct lock_class_key *key)
{
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
/*
* Make sure we are not reinitializing a held semaphore:
*/
@@ -1378,7 +1378,7 @@ void __init_rwsem(struct rw_semaphore *sem, const char *name,
{
init_rwbase_rt(&(sem)->rwbase);
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
debug_check_no_locks_freed((void *)sem, sizeof(*sem));
lockdep_init_map_wait(&sem->dep_map, name, key, 0, LD_WAIT_SLEEP);
#endif
diff --git a/kernel/locking/spinlock.c b/kernel/locking/spinlock.c
index 7f49baaa4979..e814ca0b76c3 100644
--- a/kernel/locking/spinlock.c
+++ b/kernel/locking/spinlock.c
@@ -34,7 +34,7 @@ EXPORT_PER_CPU_SYMBOL(__mmiowb_state);
* even on CONFIG_PREEMPT, because lockdep assumes that interrupts are
* not re-enabled during lock-acquire (which the preempt-spin-ops do):
*/
-#if !defined(CONFIG_GENERIC_LOCKBREAK) || defined(CONFIG_DEBUG_LOCK_ALLOC)
+#if !defined(CONFIG_GENERIC_LOCKBREAK) || defined(CONFIG_LOCK_INFO)
/*
* The __lock_function inlines are taken from
* spinlock : include/linux/spinlock_api_smp.h
diff --git a/kernel/locking/spinlock_debug.c b/kernel/locking/spinlock_debug.c
index 14235671a1a7..011f66515693 100644
--- a/kernel/locking/spinlock_debug.c
+++ b/kernel/locking/spinlock_debug.c
@@ -16,7 +16,7 @@
void __raw_spin_lock_init(raw_spinlock_t *lock, const char *name,
struct lock_class_key *key, short inner)
{
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
/*
* Make sure we are not reinitializing a held lock:
*/
@@ -35,7 +35,7 @@ EXPORT_SYMBOL(__raw_spin_lock_init);
void __rwlock_init(rwlock_t *lock, const char *name,
struct lock_class_key *key)
{
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
/*
* Make sure we are not reinitializing a held lock:
*/
diff --git a/kernel/locking/spinlock_rt.c b/kernel/locking/spinlock_rt.c
index 48a19ed8486d..22cd3eb36c98 100644
--- a/kernel/locking/spinlock_rt.c
+++ b/kernel/locking/spinlock_rt.c
@@ -56,7 +56,7 @@ void __sched rt_spin_lock(spinlock_t *lock)
}
EXPORT_SYMBOL(rt_spin_lock);
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
void __sched rt_spin_lock_nested(spinlock_t *lock, int subclass)
{
spin_acquire(&lock->dep_map, subclass, 0, _RET_IP_);
@@ -129,7 +129,7 @@ int __sched rt_spin_trylock_bh(spinlock_t *lock)
}
EXPORT_SYMBOL(rt_spin_trylock_bh);
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
void __rt_spin_lock_init(spinlock_t *lock, const char *name,
struct lock_class_key *key, bool percpu)
{
@@ -239,7 +239,7 @@ void __sched rt_write_lock(rwlock_t *rwlock)
}
EXPORT_SYMBOL(rt_write_lock);
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
void __sched rt_write_lock_nested(rwlock_t *rwlock, int subclass)
{
rtlock_might_resched();
@@ -269,7 +269,7 @@ void __sched rt_write_unlock(rwlock_t *rwlock)
}
EXPORT_SYMBOL(rt_write_unlock);
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
void __rt_rwlock_init(rwlock_t *rwlock, const char *name,
struct lock_class_key *key)
{
diff --git a/kernel/locking/ww_rt_mutex.c b/kernel/locking/ww_rt_mutex.c
index d1473c624105..aecb2e4e5f07 100644
--- a/kernel/locking/ww_rt_mutex.c
+++ b/kernel/locking/ww_rt_mutex.c
@@ -56,7 +56,7 @@ __ww_rt_mutex_lock(struct ww_mutex *lock, struct ww_acquire_ctx *ww_ctx,
if (ww_ctx->acquired == 0)
ww_ctx->wounded = 0;
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
nest_lock = &ww_ctx->dep_map;
#endif
}
diff --git a/kernel/printk/printk.c b/kernel/printk/printk.c
index 82abfaf3c2aa..e45664e0ca30 100644
--- a/kernel/printk/printk.c
+++ b/kernel/printk/printk.c
@@ -93,7 +93,7 @@ EXPORT_SYMBOL_GPL(console_drivers);
*/
int __read_mostly suppress_printk;
-#ifdef CONFIG_LOCKDEP
+#ifdef CONFIG_LOCK_INFO
static struct lockdep_map console_lock_dep_map = {
.name = "console_lock"
};
@@ -1752,7 +1752,7 @@ SYSCALL_DEFINE3(syslog, int, type, char __user *, buf, int, len)
* They allow to pass console_lock to another printk() call using a busy wait.
*/
-#ifdef CONFIG_LOCKDEP
+#ifdef CONFIG_LOCK_INFO
static struct lockdep_map console_owner_dep_map = {
.name = "console_owner"
};
diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug
index 14b89aa37c5c..5f64ffe23c35 100644
--- a/lib/Kconfig.debug
+++ b/lib/Kconfig.debug
@@ -1394,6 +1394,7 @@ config LOCKDEP
select STACKTRACE
select KALLSYMS
select KALLSYMS_ALL
+ select LOCK_INFO
config LOCKDEP_SMALL
bool
@@ -1447,6 +1448,10 @@ config DEBUG_LOCKDEP
additional runtime checks to debug itself, at the price
of more runtime overhead.
+config LOCK_INFO
+ bool
+ default n
+
config DEBUG_ATOMIC_SLEEP
bool "Sleep inside atomic section checking"
select PREEMPT_COUNT
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 09d342c7cbd0..d4ecfdd5eb8f 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -1671,7 +1671,7 @@ static int mem_cgroup_soft_reclaim(struct mem_cgroup *root_memcg,
return total;
}
-#ifdef CONFIG_LOCKDEP
+#ifdef CONFIG_LOCK_INFO
static struct lockdep_map memcg_oom_lock_dep_map = {
.name = "memcg_oom_lock",
};
diff --git a/mm/mmu_notifier.c b/mm/mmu_notifier.c
index 459d195d2ff6..26da67834cba 100644
--- a/mm/mmu_notifier.c
+++ b/mm/mmu_notifier.c
@@ -22,7 +22,7 @@
/* global SRCU for all MMs */
DEFINE_STATIC_SRCU(srcu);
-#ifdef CONFIG_LOCKDEP
+#ifdef CONFIG_LOCK_INFO
struct lockdep_map __mmu_notifier_invalidate_range_start_map = {
.name = "mmu_notifier_invalidate_range_start"
};
diff --git a/net/core/dev.c b/net/core/dev.c
index 1baab07820f6..7548a6c606ca 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -406,7 +406,7 @@ static RAW_NOTIFIER_HEAD(netdev_chain);
DEFINE_PER_CPU_ALIGNED(struct softnet_data, softnet_data);
EXPORT_PER_CPU_SYMBOL(softnet_data);
-#ifdef CONFIG_LOCKDEP
+#ifdef CONFIG_LOCK_INFO
/*
* register_netdevice() inits txq->_xmit_lock and sets lockdep class
* according to dev->type
diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index 478f857cdaed..14d87c2d0df1 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -74,7 +74,7 @@ static void svc_sock_free(struct svc_xprt *);
static struct svc_xprt *svc_create_socket(struct svc_serv *, int,
struct net *, struct sockaddr *,
int, int);
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
static struct lock_class_key svc_key[2];
static struct lock_class_key svc_slock_key[2];
diff --git a/net/sunrpc/xprtsock.c b/net/sunrpc/xprtsock.c
index d8ee06a9650a..cd66a6608ba2 100644
--- a/net/sunrpc/xprtsock.c
+++ b/net/sunrpc/xprtsock.c
@@ -1719,7 +1719,7 @@ static void xs_local_set_port(struct rpc_xprt *xprt, unsigned short port)
{
}
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
+#ifdef CONFIG_LOCK_INFO
static struct lock_class_key xs_key[3];
static struct lock_class_key xs_slock_key[3];
--
2.35.0.263.gb82422642f-goog
^ permalink raw reply related [flat|nested] 22+ messages in thread
* [PATCH 08/12] locking/mutex: Init name properly w/ CONFIG_LOCK_INFO
2022-02-08 19:43 ` Namhyung Kim
` (8 preceding siblings ...)
(?)
@ 2022-02-08 19:43 ` Namhyung Kim
-1 siblings, 0 replies; 22+ messages in thread
From: Namhyung Kim @ 2022-02-08 19:43 UTC (permalink / raw)
To: Peter Zijlstra, Ingo Molnar, Will Deacon, Waiman Long, Boqun Feng
Cc: LKML, Thomas Gleixner, Steven Rostedt, Byungchul Park,
Paul E. McKenney, Mathieu Desnoyers, Radoslaw Burny
The mutex.dep_map.name was set in debug_mutex_init() which is defined
only if CONFIG_DEBUG_MUTEXES is on. Let's fix it to work with
CONFIG_LOCK_INFO as well.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
---
kernel/locking/mutex.h | 7 +++++++
1 file changed, 7 insertions(+)
diff --git a/kernel/locking/mutex.h b/kernel/locking/mutex.h
index 0b2a79c4013b..b7b584ecdfef 100644
--- a/kernel/locking/mutex.h
+++ b/kernel/locking/mutex.h
@@ -41,5 +41,12 @@ extern void debug_mutex_init(struct mutex *lock, const char *name,
# define debug_mutex_add_waiter(lock, waiter, ti) do { } while (0)
# define debug_mutex_remove_waiter(lock, waiter, ti) do { } while (0)
# define debug_mutex_unlock(lock) do { } while (0)
+
+#ifdef CONFIG_LOCK_INFO
+# define debug_mutex_init(lock, name, key) \
+ lockdep_set_class_and_name(lock, key, name)
+#else
# define debug_mutex_init(lock, name, key) do { } while (0)
+#endif /* !CONFIG_LOCK_INFO */
+
#endif /* !CONFIG_DEBUG_MUTEXES */
--
2.35.0.263.gb82422642f-goog
^ permalink raw reply related [flat|nested] 22+ messages in thread
* [PATCH 09/12] locking: Add more static lockdep init macros
2022-02-08 19:43 ` Namhyung Kim
` (9 preceding siblings ...)
(?)
@ 2022-02-08 19:43 ` Namhyung Kim
-1 siblings, 0 replies; 22+ messages in thread
From: Namhyung Kim @ 2022-02-08 19:43 UTC (permalink / raw)
To: Peter Zijlstra, Ingo Molnar, Will Deacon, Waiman Long, Boqun Feng
Cc: LKML, Thomas Gleixner, Steven Rostedt, Byungchul Park,
Paul E. McKenney, Mathieu Desnoyers, Radoslaw Burny
Add STATIC_LOCKDEP_MAP_INIT_{WAIT,TYPE} macros and use it for various
lock init codes. This helps having different implementations of
CONFIG_LOCK_INFO like lockdep and tracepoints.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
---
drivers/gpu/drm/drm_connector.c | 5 ++---
include/linux/local_lock_internal.h | 10 ++++------
include/linux/lockdep.h | 23 +++++++++++++++++------
include/linux/mutex.h | 6 ++----
include/linux/rtmutex.h | 8 +++-----
include/linux/rwlock_types.h | 5 +----
include/linux/rwsem.h | 8 +++-----
include/linux/spinlock_types_raw.h | 24 ++++++++----------------
kernel/printk/printk.c | 10 ++++------
kernel/rcu/update.c | 27 +++++++++------------------
mm/memcontrol.c | 5 ++---
11 files changed, 55 insertions(+), 76 deletions(-)
diff --git a/drivers/gpu/drm/drm_connector.c b/drivers/gpu/drm/drm_connector.c
index 94931b32a491..7f470de2ef2b 100644
--- a/drivers/gpu/drm/drm_connector.c
+++ b/drivers/gpu/drm/drm_connector.c
@@ -677,9 +677,8 @@ const char *drm_get_connector_force_name(enum drm_connector_force force)
}
#ifdef CONFIG_LOCK_INFO
-static struct lockdep_map connector_list_iter_dep_map = {
- .name = "drm_connector_list_iter"
-};
+static struct lockdep_map connector_list_iter_dep_map =
+ STATIC_LOCKDEP_MAP_INIT("drm_connector_list_iter", NULL);
#endif
/**
diff --git a/include/linux/local_lock_internal.h b/include/linux/local_lock_internal.h
index 56f03f588aa7..dd280fcfadec 100644
--- a/include/linux/local_lock_internal.h
+++ b/include/linux/local_lock_internal.h
@@ -16,12 +16,10 @@ typedef struct {
} local_lock_t;
#ifdef CONFIG_LOCK_INFO
-# define LOCAL_LOCK_DEBUG_INIT(lockname) \
- .dep_map = { \
- .name = #lockname, \
- .wait_type_inner = LD_WAIT_CONFIG, \
- .lock_type = LD_LOCK_PERCPU, \
- }, \
+# define LOCAL_LOCK_DEBUG_INIT(lockname) \
+ .dep_map = STATIC_LOCKDEP_MAP_INIT_TYPE(#lockname, NULL, \
+ LD_WAIT_CONFIG, LD_WAIT_INV, \
+ LD_LOCK_PERCPU), \
.owner = NULL,
static inline void local_lock_acquire(local_lock_t *l)
diff --git a/include/linux/lockdep.h b/include/linux/lockdep.h
index 0cc2b338a006..38cbef7601c7 100644
--- a/include/linux/lockdep.h
+++ b/include/linux/lockdep.h
@@ -237,6 +237,19 @@ static inline void lockdep_init_map(struct lockdep_map *lock, const char *name,
#define lockdep_set_novalidate_class(lock) \
lockdep_set_class_and_name(lock, &__lockdep_no_validate__, #lock)
+/*
+ * To initialize a lockdep_map statically use this macro.
+ * Note that _name must not be NULL.
+ */
+#define STATIC_LOCKDEP_MAP_INIT(_name, _key) \
+ { .name = (_name), .key = (void *)(_key), }
+
+#define STATIC_LOCKDEP_MAP_INIT_WAIT(_name, _key, _inner) \
+ { .name = (_name), .key = (void *)(_key), .wait_type_inner = (_inner), }
+
+#define STATIC_LOCKDEP_MAP_INIT_TYPE(_name, _key, _inner, _outer, _type) \
+ { .name = (_name), .key = (void *)(_key), .wait_type_inner = (_inner), \
+ .wait_type_outer = (_outer), .lock_type = (_type), }
/*
* Compare locking classes
*/
@@ -377,6 +390,10 @@ static inline void lockdep_set_selftest_task(struct task_struct *task)
#define lockdep_set_novalidate_class(lock) do { } while (0)
+#define STATIC_LOCKDEP_MAP_INIT(_name, _key) { }
+#define STATIC_LOCKDEP_MAP_INIT_WAIT(_name, _key, _inner) { }
+#define STATIC_LOCKDEP_MAP_INIT_TYPE(_name, _key, _inner, _outer, _type) { }
+
/*
* We don't define lockdep_match_class() and lockdep_match_key() for !LOCKDEP
* case since the result is not well defined and the caller should rather
@@ -432,12 +449,6 @@ enum xhlock_context_t {
};
#define lockdep_init_map_crosslock(m, n, k, s) do {} while (0)
-/*
- * To initialize a lockdep_map statically use this macro.
- * Note that _name must not be NULL.
- */
-#define STATIC_LOCKDEP_MAP_INIT(_name, _key) \
- { .name = (_name), .key = (void *)(_key), }
static inline void lockdep_invariant_state(bool force) {}
static inline void lockdep_free_task(struct task_struct *task) {}
diff --git a/include/linux/mutex.h b/include/linux/mutex.h
index 369c1abbf3d0..b2d018250a41 100644
--- a/include/linux/mutex.h
+++ b/include/linux/mutex.h
@@ -22,10 +22,8 @@
#ifdef CONFIG_LOCK_INFO
# define __DEP_MAP_MUTEX_INITIALIZER(lockname) \
- , .dep_map = { \
- .name = #lockname, \
- .wait_type_inner = LD_WAIT_SLEEP, \
- }
+ , .dep_map = STATIC_LOCKDEP_MAP_INIT_WAIT(#lockname, \
+ NULL, LD_WAIT_SLEEP)
#else
# define __DEP_MAP_MUTEX_INITIALIZER(lockname)
#endif
diff --git a/include/linux/rtmutex.h b/include/linux/rtmutex.h
index 8eafdd6dcf35..887ffcd5fc09 100644
--- a/include/linux/rtmutex.h
+++ b/include/linux/rtmutex.h
@@ -77,11 +77,9 @@ do { \
} while (0)
#ifdef CONFIG_LOCK_INFO
-#define __DEP_MAP_RT_MUTEX_INITIALIZER(mutexname) \
- .dep_map = { \
- .name = #mutexname, \
- .wait_type_inner = LD_WAIT_SLEEP, \
- }
+#define __DEP_MAP_RT_MUTEX_INITIALIZER(mutexname) \
+ .dep_map = STATIC_LOCKDEP_MAP_INIT_WAIT(#mutexname, \
+ NULL, LD_WAIT_SLEEP)
#else
#define __DEP_MAP_RT_MUTEX_INITIALIZER(mutexname)
#endif
diff --git a/include/linux/rwlock_types.h b/include/linux/rwlock_types.h
index 3e621bfd7cd0..438d8639a229 100644
--- a/include/linux/rwlock_types.h
+++ b/include/linux/rwlock_types.h
@@ -7,10 +7,7 @@
#ifdef CONFIG_LOCK_INFO
# define RW_DEP_MAP_INIT(lockname) \
- .dep_map = { \
- .name = #lockname, \
- .wait_type_inner = LD_WAIT_CONFIG, \
- }
+ .dep_map = STATIC_LOCKDEP_MAP_INIT_WAIT(#lockname, NULL, LD_WAIT_CONFIG)
#else
# define RW_DEP_MAP_INIT(lockname)
#endif
diff --git a/include/linux/rwsem.h b/include/linux/rwsem.h
index c488485861f5..39126e6d97a1 100644
--- a/include/linux/rwsem.h
+++ b/include/linux/rwsem.h
@@ -17,11 +17,9 @@
#include <linux/err.h>
#ifdef CONFIG_LOCK_INFO
-# define __RWSEM_DEP_MAP_INIT(lockname) \
- .dep_map = { \
- .name = #lockname, \
- .wait_type_inner = LD_WAIT_SLEEP, \
- },
+# define __RWSEM_DEP_MAP_INIT(lockname) \
+ .dep_map = STATIC_LOCKDEP_MAP_INIT_WAIT(#lockname, \
+ NULL, LD_WAIT_SLEEP),
#else
# define __RWSEM_DEP_MAP_INIT(lockname)
#endif
diff --git a/include/linux/spinlock_types_raw.h b/include/linux/spinlock_types_raw.h
index 564092a30cc4..006250640e76 100644
--- a/include/linux/spinlock_types_raw.h
+++ b/include/linux/spinlock_types_raw.h
@@ -27,23 +27,15 @@ typedef struct raw_spinlock {
#define SPINLOCK_OWNER_INIT ((void *)-1L)
#ifdef CONFIG_LOCK_INFO
-# define RAW_SPIN_DEP_MAP_INIT(lockname) \
- .dep_map = { \
- .name = #lockname, \
- .wait_type_inner = LD_WAIT_SPIN, \
- }
-# define SPIN_DEP_MAP_INIT(lockname) \
- .dep_map = { \
- .name = #lockname, \
- .wait_type_inner = LD_WAIT_CONFIG, \
- }
+# define RAW_SPIN_DEP_MAP_INIT(lockname) \
+ .dep_map = STATIC_LOCKDEP_MAP_INIT_WAIT(#lockname, NULL, LD_WAIT_SPIN)
-# define LOCAL_SPIN_DEP_MAP_INIT(lockname) \
- .dep_map = { \
- .name = #lockname, \
- .wait_type_inner = LD_WAIT_CONFIG, \
- .lock_type = LD_LOCK_PERCPU, \
- }
+# define SPIN_DEP_MAP_INIT(lockname) \
+ .dep_map = STATIC_LOCKDEP_MAP_INIT_WAIT(#lockname, NULL, LD_WAIT_CONFIG)
+
+# define LOCAL_SPIN_DEP_MAP_INIT(lockname) \
+ .dep_map = STATIC_LOCKDEP_MAP_INIT_TYPE(#lockname, NULL, LD_WAIT_CONFIG,\
+ LD_WAIT_INV, LD_LOCK_PERCPU)
#else
# define RAW_SPIN_DEP_MAP_INIT(lockname)
# define SPIN_DEP_MAP_INIT(lockname)
diff --git a/kernel/printk/printk.c b/kernel/printk/printk.c
index e45664e0ca30..7889df01a378 100644
--- a/kernel/printk/printk.c
+++ b/kernel/printk/printk.c
@@ -94,9 +94,8 @@ EXPORT_SYMBOL_GPL(console_drivers);
int __read_mostly suppress_printk;
#ifdef CONFIG_LOCK_INFO
-static struct lockdep_map console_lock_dep_map = {
- .name = "console_lock"
-};
+static struct lockdep_map console_lock_dep_map =
+ STATIC_LOCKDEP_MAP_INIT("console_lock", NULL);
#endif
enum devkmsg_log_bits {
@@ -1753,9 +1752,8 @@ SYSCALL_DEFINE3(syslog, int, type, char __user *, buf, int, len)
*/
#ifdef CONFIG_LOCK_INFO
-static struct lockdep_map console_owner_dep_map = {
- .name = "console_owner"
-};
+static struct lockdep_map console_owner_dep_map =
+ STATIC_LOCKDEP_MAP_INIT("console_owner", NULL);
#endif
static DEFINE_RAW_SPINLOCK(console_owner_lock);
diff --git a/kernel/rcu/update.c b/kernel/rcu/update.c
index 156892c22bb5..8202ab6ddb4c 100644
--- a/kernel/rcu/update.c
+++ b/kernel/rcu/update.c
@@ -243,30 +243,21 @@ core_initcall(rcu_set_runtime_mode);
#ifdef CONFIG_DEBUG_LOCK_ALLOC
static struct lock_class_key rcu_lock_key;
-struct lockdep_map rcu_lock_map = {
- .name = "rcu_read_lock",
- .key = &rcu_lock_key,
- .wait_type_outer = LD_WAIT_FREE,
- .wait_type_inner = LD_WAIT_CONFIG, /* PREEMPT_RT implies PREEMPT_RCU */
-};
+struct lockdep_map rcu_lock_map = /* PREEMPT_RT implies PREEMPT_RCU */
+ STATIC_LOCKDEP_MAP_INIT_TYPE("rcu_read_lock", &rcu_lock_key,
+ LD_WAIT_CONFIG, LD_WAIT_FREE, 0);
EXPORT_SYMBOL_GPL(rcu_lock_map);
static struct lock_class_key rcu_bh_lock_key;
-struct lockdep_map rcu_bh_lock_map = {
- .name = "rcu_read_lock_bh",
- .key = &rcu_bh_lock_key,
- .wait_type_outer = LD_WAIT_FREE,
- .wait_type_inner = LD_WAIT_CONFIG, /* PREEMPT_RT makes BH preemptible. */
-};
+struct lockdep_map rcu_bh_lock_map = /* PREEMPT_RT makes BH preemptable. */
+ STATIC_LOCKDEP_MAP_INIT_TYPE("rcu_read_lock_bh", &rcu_bh_lock_key,
+ LD_WAIT_CONFIG, LD_WAIT_FREE, 0);
EXPORT_SYMBOL_GPL(rcu_bh_lock_map);
static struct lock_class_key rcu_sched_lock_key;
-struct lockdep_map rcu_sched_lock_map = {
- .name = "rcu_read_lock_sched",
- .key = &rcu_sched_lock_key,
- .wait_type_outer = LD_WAIT_FREE,
- .wait_type_inner = LD_WAIT_SPIN,
-};
+struct lockdep_map rcu_sched_lock_map =
+ STATIC_LOCKDEP_MAP_INIT_TYPE("rcu_read_lock_sched", &rcu_sched_lock_key,
+ LD_WAIT_SPIN, LD_WAIT_FREE, 0);
EXPORT_SYMBOL_GPL(rcu_sched_lock_map);
// Tell lockdep when RCU callbacks are being invoked.
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index d4ecfdd5eb8f..a561a6c66b2f 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -1672,9 +1672,8 @@ static int mem_cgroup_soft_reclaim(struct mem_cgroup *root_memcg,
}
#ifdef CONFIG_LOCK_INFO
-static struct lockdep_map memcg_oom_lock_dep_map = {
- .name = "memcg_oom_lock",
-};
+static struct lockdep_map memcg_oom_lock_dep_map =
+ STATIC_LOCKDEP_MAP_INIT("memcg_oom_lock", NULL);
#endif
static DEFINE_SPINLOCK(memcg_oom_lock);
--
2.35.0.263.gb82422642f-goog
^ permalink raw reply related [flat|nested] 22+ messages in thread
* [PATCH 10/12] locking: Add CONFIG_LOCK_TRACEPOINTS option
2022-02-08 19:43 ` Namhyung Kim
` (10 preceding siblings ...)
(?)
@ 2022-02-08 19:43 ` Namhyung Kim
-1 siblings, 0 replies; 22+ messages in thread
From: Namhyung Kim @ 2022-02-08 19:43 UTC (permalink / raw)
To: Peter Zijlstra, Ingo Molnar, Will Deacon, Waiman Long, Boqun Feng
Cc: LKML, Thomas Gleixner, Steven Rostedt, Byungchul Park,
Paul E. McKenney, Mathieu Desnoyers, Radoslaw Burny
This option enables the following tracepoints for lock analysis. This
can be used to trace lock/unlock pairs and their contentions without
enabling LOCKDEP and LOCK_STAT.
* lock_acquire
* lock_release
* lock_contended
* lock_acquired
It selects CONFIG_LOCK_INFO to have lock names properly, and make the
lockdep functions inline to check if the corresponding tracepoint is
enabled before calling the function. Then it could avoid unnecessary
overhead setting up the arguments in the tracepoint when disabled.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
---
include/linux/lockdep.h | 141 +++++++++++++++++++++++++++++++++-
include/linux/lockdep_types.h | 8 +-
include/trace/events/lock.h | 4 +-
kernel/locking/Makefile | 1 +
kernel/locking/lockdep.c | 40 +++++++++-
lib/Kconfig.debug | 9 +++
6 files changed, 195 insertions(+), 8 deletions(-)
diff --git a/include/linux/lockdep.h b/include/linux/lockdep.h
index 38cbef7601c7..4e728d2957db 100644
--- a/include/linux/lockdep.h
+++ b/include/linux/lockdep.h
@@ -14,6 +14,8 @@
#include <linux/smp.h>
#include <asm/percpu.h>
+#include <linux/tracepoint-defs.h>
+
struct task_struct;
/* for sysctl */
@@ -367,12 +369,104 @@ static inline void lockdep_set_selftest_task(struct task_struct *task)
{
}
+#ifdef CONFIG_LOCK_TRACEPOINTS
+DECLARE_TRACEPOINT(lock_acquire);
+DECLARE_TRACEPOINT(lock_release);
+
+extern void __lock_acquire(struct lockdep_map *lock, unsigned int subclass,
+ int trylock, int read, int check,
+ struct lockdep_map *nest_lock, unsigned long ip);
+extern void __lock_release(struct lockdep_map *lock, unsigned long ip);
+
+static inline void lock_acquire(struct lockdep_map *lock, unsigned int subclass,
+ int trylock, int read, int check,
+ struct lockdep_map *nest_lock, unsigned long ip)
+{
+ if (tracepoint_enabled(lock_acquire))
+ __lock_acquire(lock, subclass, trylock, read, check, nest_lock, ip);
+}
+
+static inline void lock_release(struct lockdep_map *lock, unsigned long ip)
+{
+ if (tracepoint_enabled(lock_release))
+ __lock_release(lock, ip);
+}
+#else /* !CONFIG_LOCK_TRACEPOINTS */
# define lock_acquire(l, s, t, r, c, n, i) do { } while (0)
# define lock_release(l, i) do { } while (0)
+#endif /* CONFIG_LOCK_TRACEPOINTS */
+
# define lock_downgrade(l, i) do { } while (0)
# define lock_set_class(l, n, k, s, i) do { } while (0)
# define lock_set_subclass(l, s, i) do { } while (0)
# define lockdep_init() do { } while (0)
+
+#ifdef CONFIG_LOCK_INFO
+
+static inline void
+lockdep_init_map_type(struct lockdep_map *lock, const char *name,
+ struct lock_class_key *key, int subclass,
+ u8 inner, u8 outer, u8 type)
+{
+ if (!name)
+ name = "NULL";
+
+ lock->name = name;
+}
+
+static inline void
+lockdep_init_map_waits(struct lockdep_map *lock, const char *name,
+ struct lock_class_key *key, int subclass,
+ u8 inner, u8 outer)
+{
+ lockdep_init_map_type(lock, name, key, subclass, inner, outer, 0);
+}
+
+static inline void
+lockdep_init_map_wait(struct lockdep_map *lock, const char *name,
+ struct lock_class_key *key, int subclass, u8 inner)
+{
+ lockdep_init_map_waits(lock, name, key, subclass, inner, 0);
+}
+
+static inline void
+lockdep_init_map(struct lockdep_map *lock, const char *name,
+ struct lock_class_key *key, int subclass)
+{
+ lockdep_init_map_wait(lock, name, key, subclass, 0);
+}
+
+/* Reinitialize a lock name - other info will be ignore. */
+# define lockdep_set_class(lock, key) \
+ lockdep_init_map(&(lock)->dep_map, #key, key, 0)
+
+# define lockdep_set_class_and_name(lock, key, name) \
+ lockdep_init_map(&(lock)->dep_map, name, key, 0)
+
+# define lockdep_set_class_and_subclass(lock, key, sub) \
+ lockdep_init_map(&(lock)->dep_map, #key, key, sub)
+
+# define lockdep_set_subclass(lock, sub) \
+ lockdep_init_map(&(lock)->dep_map, #lock, NULL, sub)
+
+# define lockdep_set_novalidate_class(lock) \
+ lockdep_set_class_and_name(lock, NULL, #lock)
+
+/*
+ * To initialize a lockdep_map statically use this macro.
+ * Note that _name must not be NULL.
+ */
+# define STATIC_LOCKDEP_MAP_INIT(_name, _key) \
+ { .name = (_name), }
+
+# define STATIC_LOCKDEP_MAP_INIT_WAIT(_name, _key, _inner) \
+ { .name = (_name), }
+
+# define STATIC_LOCKDEP_MAP_INIT_TYPE(_name, _key, _inner, _outer, _type) \
+ { .name = (_name), }
+
+#else /* !CONFIG_LOCK_INFO */
+
# define lockdep_init_map_type(lock, name, key, sub, inner, outer, type) \
do { (void)(name); (void)(key); } while (0)
# define lockdep_init_map_waits(lock, name, key, sub, inner, outer) \
@@ -394,6 +488,8 @@ static inline void lockdep_set_selftest_task(struct task_struct *task)
#define STATIC_LOCKDEP_MAP_INIT_WAIT(_name, _key, _inner) { }
#define STATIC_LOCKDEP_MAP_INIT_TYPE(_name, _key, _inner, _outer, _type) { }
+#endif /* CONFIG_LOCK_INFO */
+
/*
* We don't define lockdep_match_class() and lockdep_match_key() for !LOCKDEP
* case since the result is not well defined and the caller should rather
@@ -479,7 +575,48 @@ do { \
____err; \
})
-#else /* CONFIG_LOCK_STAT */
+#elif defined(CONFIG_LOCK_TRACEPOINTS)
+
+DECLARE_TRACEPOINT(lock_contended);
+DECLARE_TRACEPOINT(lock_acquired);
+
+extern void __lock_contended(struct lockdep_map *lock, unsigned long ip);
+extern void __lock_acquired(struct lockdep_map *lock, unsigned long ip);
+
+static inline void lock_contended(struct lockdep_map *lock, unsigned long ip)
+{
+ if (tracepoint_enabled(lock_contended))
+ __lock_contended(lock, ip);
+}
+
+static inline void lock_acquired(struct lockdep_map *lock, unsigned long ip)
+{
+ if (tracepoint_enabled(lock_acquired))
+ __lock_acquired(lock, ip);
+}
+
+#define LOCK_CONTENDED(_lock, try, lock) \
+do { \
+ if (!try(_lock)) { \
+ lock_contended(&(_lock)->dep_map, _RET_IP_); \
+ lock(_lock); \
+ } \
+ lock_acquired(&(_lock)->dep_map, _RET_IP_); \
+} while (0)
+
+#define LOCK_CONTENDED_RETURN(_lock, try, lock) \
+({ \
+ int ____err = 0; \
+ if (!try(_lock)) { \
+ lock_contended(&(_lock)->dep_map, _RET_IP_); \
+ ____err = lock(_lock); \
+ } \
+ if (!____err) \
+ lock_acquired(&(_lock)->dep_map, _RET_IP_); \
+ ____err; \
+})
+
+#else /* !CONFIG_LOCK_STAT && !CONFIG_LOCK_TRACEPOINTS */
#define lock_contended(lockdep_map, ip) do {} while (0)
#define lock_acquired(lockdep_map, ip) do {} while (0)
@@ -490,7 +627,7 @@ do { \
#define LOCK_CONTENDED_RETURN(_lock, try, lock) \
lock(_lock)
-#endif /* CONFIG_LOCK_STAT */
+#endif /* CONFIG_LOCK_STAT || CONFIG_LOCK_TRACEPOINTS */
#ifdef CONFIG_PROVE_LOCKING
extern void print_irqtrace_events(struct task_struct *curr);
diff --git a/include/linux/lockdep_types.h b/include/linux/lockdep_types.h
index d22430840b53..340138c9023e 100644
--- a/include/linux/lockdep_types.h
+++ b/include/linux/lockdep_types.h
@@ -197,9 +197,13 @@ struct pin_cookie { unsigned int val; };
struct lock_class_key { };
/*
- * The lockdep_map takes no space if lockdep is disabled:
+ * The lockdep_map takes no space if lockdep and lock_info is disabled:
*/
-struct lockdep_map { };
+struct lockdep_map {
+#ifdef CONFIG_LOCK_INFO
+ const char *name;
+#endif
+};
struct pin_cookie { };
diff --git a/include/trace/events/lock.h b/include/trace/events/lock.h
index d7512129a324..74667841ae02 100644
--- a/include/trace/events/lock.h
+++ b/include/trace/events/lock.h
@@ -8,7 +8,7 @@
#include <linux/lockdep.h>
#include <linux/tracepoint.h>
-#ifdef CONFIG_LOCKDEP
+#if defined(CONFIG_LOCKDEP) || defined(CONFIG_LOCK_TRACEPOINTS)
TRACE_EVENT(lock_acquire,
@@ -62,7 +62,7 @@ DEFINE_EVENT(lock, lock_release,
TP_ARGS(lock, ip)
);
-#ifdef CONFIG_LOCK_STAT
+#if defined(CONFIG_LOCK_STAT) || defined(CONFIG_LOCK_TRACEPOINTS)
DEFINE_EVENT(lock, lock_contended,
diff --git a/kernel/locking/Makefile b/kernel/locking/Makefile
index d51cabf28f38..d772a37470a6 100644
--- a/kernel/locking/Makefile
+++ b/kernel/locking/Makefile
@@ -17,6 +17,7 @@ endif
obj-$(CONFIG_DEBUG_IRQFLAGS) += irqflag-debug.o
obj-$(CONFIG_DEBUG_MUTEXES) += mutex-debug.o
obj-$(CONFIG_LOCKDEP) += lockdep.o
+obj-$(CONFIG_LOCK_TRACEPOINTS) += lockdep.o
ifeq ($(CONFIG_PROC_FS),y)
obj-$(CONFIG_LOCKDEP) += lockdep_proc.o
endif
diff --git a/kernel/locking/lockdep.c b/kernel/locking/lockdep.c
index 2e6892ec3756..f234e981d1c7 100644
--- a/kernel/locking/lockdep.c
+++ b/kernel/locking/lockdep.c
@@ -58,11 +58,13 @@
#include <asm/sections.h>
-#include "lockdep_internals.h"
-
#define CREATE_TRACE_POINTS
#include <trace/events/lock.h>
+#ifdef CONFIG_LOCKDEP
+
+#include "lockdep_internals.h"
+
#ifdef CONFIG_PROVE_LOCKING
int prove_locking = 1;
module_param(prove_locking, int, 0644);
@@ -6566,3 +6568,37 @@ void lockdep_rcu_suspicious(const char *file, const int line, const char *s)
dump_stack();
}
EXPORT_SYMBOL_GPL(lockdep_rcu_suspicious);
+
+#elif defined(CONFIG_LOCK_TRACEPOINTS)
+
+void __lock_acquire(struct lockdep_map *lock, unsigned int subclass,
+ int trylock, int read, int check,
+ struct lockdep_map *nest_lock, unsigned long ip)
+{
+ trace_lock_acquire(lock, subclass, trylock, read, check, nest_lock, ip);
+}
+EXPORT_SYMBOL_GPL(__lock_acquire);
+EXPORT_TRACEPOINT_SYMBOL_GPL(lock_acquire);
+
+void __lock_release(struct lockdep_map *lock, unsigned long ip)
+{
+ trace_lock_release(lock, ip);
+}
+EXPORT_SYMBOL_GPL(__lock_release);
+EXPORT_TRACEPOINT_SYMBOL_GPL(lock_release);
+
+void __lock_contended(struct lockdep_map *lock, unsigned long ip)
+{
+ trace_lock_contended(lock, ip);
+}
+EXPORT_SYMBOL_GPL(__lock_contended);
+EXPORT_TRACEPOINT_SYMBOL_GPL(lock_contended);
+
+void __lock_acquired(struct lockdep_map *lock, unsigned long ip)
+{
+ trace_lock_acquired(lock, ip);
+}
+EXPORT_SYMBOL_GPL(__lock_acquired);
+EXPORT_TRACEPOINT_SYMBOL_GPL(lock_acquired);
+
+#endif /* CONFIG_LOCKDEP */
diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug
index 5f64ffe23c35..749de6c1ba50 100644
--- a/lib/Kconfig.debug
+++ b/lib/Kconfig.debug
@@ -1452,6 +1452,15 @@ config LOCK_INFO
bool
default n
+config LOCK_TRACEPOINTS
+ bool "Enable lock tracepoints"
+ depends on !LOCKDEP
+ select LOCK_INFO
+ help
+ This enables all lock events required by "perf lock", subcommand of
+ perf, without enabling LOCKDEP and LOCK_STAT. If you want to use
+ "perf lock", you also need to turn on CONFIG_EVENT_TRACING.
+
config DEBUG_ATOMIC_SLEEP
bool "Sleep inside atomic section checking"
select PREEMPT_COUNT
--
2.35.0.263.gb82422642f-goog
^ permalink raw reply related [flat|nested] 22+ messages in thread
* [PATCH 11/12] locking/mutex: Revive fast functions for CONFIG_LOCK_TRACEPOINTS
2022-02-08 19:43 ` Namhyung Kim
` (11 preceding siblings ...)
(?)
@ 2022-02-08 19:43 ` Namhyung Kim
-1 siblings, 0 replies; 22+ messages in thread
From: Namhyung Kim @ 2022-02-08 19:43 UTC (permalink / raw)
To: Peter Zijlstra, Ingo Molnar, Will Deacon, Waiman Long, Boqun Feng
Cc: LKML, Thomas Gleixner, Steven Rostedt, Byungchul Park,
Paul E. McKenney, Mathieu Desnoyers, Radoslaw Burny
The CONFIG_LOCK_TRACEPOINTS used the same path as CONFIG_DEBUG_ALLOC
or CONFIG_LOCKDEP but it caused performance impact on mutex as it
removes _fast variants at the beginning.
I'm not entirely sure why it removed the fast versions when lockdep is
on. It seems easy to add the required annotation when the fast
version is succeeded as far as the tracpoints are concerned.
It reduces around 2% of elapsed time when I ran a micro-benchmark. It
was `perf bench sched messaging -p -l 1000000` and it had a pretty low
lock contention rate (under 1% of total lock acquisition). So the
improvement should come from the fast path.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
---
include/linux/mutex.h | 2 +-
kernel/locking/mutex.c | 30 ++++++++++++++++++++----------
2 files changed, 21 insertions(+), 11 deletions(-)
diff --git a/include/linux/mutex.h b/include/linux/mutex.h
index b2d018250a41..49bde305453e 100644
--- a/include/linux/mutex.h
+++ b/include/linux/mutex.h
@@ -172,7 +172,7 @@ do { \
* See kernel/locking/mutex.c for detailed documentation of these APIs.
* Also see Documentation/locking/mutex-design.rst.
*/
-#ifdef CONFIG_LOCK_INFO
+#ifdef CONFIG_DEBUG_LOCK_ALLOC
extern void mutex_lock_nested(struct mutex *lock, unsigned int subclass);
extern void _mutex_lock_nest_lock(struct mutex *lock, struct lockdep_map *nest_lock);
diff --git a/kernel/locking/mutex.c b/kernel/locking/mutex.c
index 8733b96ce20a..f8bc4ae312a0 100644
--- a/kernel/locking/mutex.c
+++ b/kernel/locking/mutex.c
@@ -149,7 +149,7 @@ static inline bool __mutex_trylock(struct mutex *lock)
return !__mutex_trylock_common(lock, false);
}
-#ifndef CONFIG_LOCK_INFO
+#ifndef CONFIG_DEBUG_LOCK_ALLOC
/*
* Lockdep annotations are contained to the slow paths for simplicity.
* There is nothing that would stop spreading the lockdep annotations outwards
@@ -245,7 +245,7 @@ static void __mutex_handoff(struct mutex *lock, struct task_struct *task)
}
}
-#ifndef CONFIG_LOCK_INFO
+#ifndef CONFIG_DEBUG_LOCK_ALLOC
/*
* We split the mutex lock/unlock logic into separate fastpath and
* slowpath functions, to reduce the register pressure on the fastpath.
@@ -281,6 +281,8 @@ void __sched mutex_lock(struct mutex *lock)
if (!__mutex_trylock_fast(lock))
__mutex_lock_slowpath(lock);
+ else
+ mutex_acquire(&lock->dep_map, 0, 0, _RET_IP_);
}
EXPORT_SYMBOL(mutex_lock);
#endif
@@ -533,9 +535,11 @@ static noinline void __sched __mutex_unlock_slowpath(struct mutex *lock, unsigne
*/
void __sched mutex_unlock(struct mutex *lock)
{
-#ifndef CONFIG_LOCK_INFO
- if (__mutex_unlock_fast(lock))
+#ifndef CONFIG_DEBUG_LOCK_ALLOC
+ if (__mutex_unlock_fast(lock)) {
+ mutex_release(&lock->dep_map, _RET_IP_);
return;
+ }
#endif
__mutex_unlock_slowpath(lock, _RET_IP_);
}
@@ -591,7 +595,7 @@ __mutex_lock_common(struct mutex *lock, unsigned int state, unsigned int subclas
if (ww_ctx->acquired == 0)
ww_ctx->wounded = 0;
-#ifdef CONFIG_LOCK_INFO
+#ifdef CONFIG_DEBUG_LOCK_ALLOC
nest_lock = &ww_ctx->dep_map;
#endif
}
@@ -778,7 +782,7 @@ int ww_mutex_trylock(struct ww_mutex *ww, struct ww_acquire_ctx *ww_ctx)
}
EXPORT_SYMBOL(ww_mutex_trylock);
-#ifdef CONFIG_LOCK_INFO
+#ifdef CONFIG_DEBUG_LOCK_ALLOC
void __sched
mutex_lock_nested(struct mutex *lock, unsigned int subclass)
{
@@ -937,7 +941,7 @@ static noinline void __sched __mutex_unlock_slowpath(struct mutex *lock, unsigne
wake_up_q(&wake_q);
}
-#ifndef CONFIG_LOCK_INFO
+#ifndef CONFIG_DEBUG_LOCK_ALLOC
/*
* Here come the less common (and hence less performance-critical) APIs:
* mutex_lock_interruptible() and mutex_trylock().
@@ -964,8 +968,10 @@ int __sched mutex_lock_interruptible(struct mutex *lock)
{
might_sleep();
- if (__mutex_trylock_fast(lock))
+ if (__mutex_trylock_fast(lock)) {
+ mutex_acquire(&lock->dep_map, 0, 0, _RET_IP_);
return 0;
+ }
return __mutex_lock_interruptible_slowpath(lock);
}
@@ -988,8 +994,10 @@ int __sched mutex_lock_killable(struct mutex *lock)
{
might_sleep();
- if (__mutex_trylock_fast(lock))
+ if (__mutex_trylock_fast(lock)) {
+ mutex_acquire(&lock->dep_map, 0, 0, _RET_IP_);
return 0;
+ }
return __mutex_lock_killable_slowpath(lock);
}
@@ -1078,7 +1086,7 @@ int __sched mutex_trylock(struct mutex *lock)
}
EXPORT_SYMBOL(mutex_trylock);
-#ifndef CONFIG_LOCK_INFO
+#ifndef CONFIG_DEBUG_LOCK_ALLOC
int __sched
ww_mutex_lock(struct ww_mutex *lock, struct ww_acquire_ctx *ctx)
{
@@ -1087,6 +1095,7 @@ ww_mutex_lock(struct ww_mutex *lock, struct ww_acquire_ctx *ctx)
if (__mutex_trylock_fast(&lock->base)) {
if (ctx)
ww_mutex_set_context_fastpath(lock, ctx);
+ mutex_acquire(&lock->base.dep_map, 0, 0, _RET_IP_);
return 0;
}
@@ -1102,6 +1111,7 @@ ww_mutex_lock_interruptible(struct ww_mutex *lock, struct ww_acquire_ctx *ctx)
if (__mutex_trylock_fast(&lock->base)) {
if (ctx)
ww_mutex_set_context_fastpath(lock, ctx);
+ mutex_acquire(&lock->base.dep_map, 0, 0, _RET_IP_);
return 0;
}
--
2.35.0.263.gb82422642f-goog
^ permalink raw reply related [flat|nested] 22+ messages in thread
* [PATCH 12/12] locking: Move lock_acquired() from the fast path
2022-02-08 19:43 ` Namhyung Kim
` (12 preceding siblings ...)
(?)
@ 2022-02-08 19:43 ` Namhyung Kim
-1 siblings, 0 replies; 22+ messages in thread
From: Namhyung Kim @ 2022-02-08 19:43 UTC (permalink / raw)
To: Peter Zijlstra, Ingo Molnar, Will Deacon, Waiman Long, Boqun Feng
Cc: LKML, Thomas Gleixner, Steven Rostedt, Byungchul Park,
Paul E. McKenney, Mathieu Desnoyers, Radoslaw Burny
The lock_acquired() function is used by CONFIG_LOCK_STAT to track wait
time for contended locks. So it's meaningful only if the given lock
is in the slow path (contended). Let's move the call into the if
block so that we can skip it in the fast path. This also move the
tracepoint to be called only after lock_contended().
It might affect bounce_acquired stats rarely (if it's on a different
cpu than when you call lock_acquire) but I'm not sure it's possible in
uncontended cases. Otherwise, this should have no functional changes
in the LOCKDEP and LOCK_STAT.
Userspace tools that use the tracepoint might see the difference, but
I think most of them can handle the missing lock_acquired() in
non-contended case properly as it's the case when using a trylock
function to grab a lock. At least it seems ok for the perf
tools ('perf lock' command specifically).
Add similar change in the __mutex_lock_common() so that it can call
lock_acquired() only after lock_contended().
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
---
Documentation/locking/lockstat.rst | 4 ++--
include/linux/lockdep.h | 12 ++++++------
kernel/locking/mutex.c | 4 +---
3 files changed, 9 insertions(+), 11 deletions(-)
diff --git a/Documentation/locking/lockstat.rst b/Documentation/locking/lockstat.rst
index 536eab8dbd99..3638ad1113c2 100644
--- a/Documentation/locking/lockstat.rst
+++ b/Documentation/locking/lockstat.rst
@@ -28,11 +28,11 @@ The graph below shows the relation between the lock functions and the various
| __contended
| |
| <wait>
+ | |
+ | __acquired
| _______/
|/
|
- __acquired
- |
.
<hold>
.
diff --git a/include/linux/lockdep.h b/include/linux/lockdep.h
index 4e728d2957db..63b75ad2e17c 100644
--- a/include/linux/lockdep.h
+++ b/include/linux/lockdep.h
@@ -559,8 +559,8 @@ do { \
if (!try(_lock)) { \
lock_contended(&(_lock)->dep_map, _RET_IP_); \
lock(_lock); \
+ lock_acquired(&(_lock)->dep_map, _RET_IP_); \
} \
- lock_acquired(&(_lock)->dep_map, _RET_IP_); \
} while (0)
#define LOCK_CONTENDED_RETURN(_lock, try, lock) \
@@ -569,9 +569,9 @@ do { \
if (!try(_lock)) { \
lock_contended(&(_lock)->dep_map, _RET_IP_); \
____err = lock(_lock); \
+ if (!____err) \
+ lock_acquired(&(_lock)->dep_map, _RET_IP_); \
} \
- if (!____err) \
- lock_acquired(&(_lock)->dep_map, _RET_IP_); \
____err; \
})
@@ -600,8 +600,8 @@ do { \
if (!try(_lock)) { \
lock_contended(&(_lock)->dep_map, _RET_IP_); \
lock(_lock); \
+ lock_acquired(&(_lock)->dep_map, _RET_IP_); \
} \
- lock_acquired(&(_lock)->dep_map, _RET_IP_); \
} while (0)
#define LOCK_CONTENDED_RETURN(_lock, try, lock) \
@@ -610,9 +610,9 @@ do { \
if (!try(_lock)) { \
lock_contended(&(_lock)->dep_map, _RET_IP_); \
____err = lock(_lock); \
+ if (!____err) \
+ lock_acquired(&(_lock)->dep_map, _RET_IP_); \
} \
- if (!____err) \
- lock_acquired(&(_lock)->dep_map, _RET_IP_); \
____err; \
})
diff --git a/kernel/locking/mutex.c b/kernel/locking/mutex.c
index f8bc4ae312a0..e67b5a16440b 100644
--- a/kernel/locking/mutex.c
+++ b/kernel/locking/mutex.c
@@ -605,8 +605,6 @@ __mutex_lock_common(struct mutex *lock, unsigned int state, unsigned int subclas
if (__mutex_trylock(lock) ||
mutex_optimistic_spin(lock, ww_ctx, NULL)) {
- /* got the lock, yay! */
- lock_acquired(&lock->dep_map, ip);
if (ww_ctx)
ww_mutex_set_context_fastpath(ww, ww_ctx);
preempt_enable();
@@ -708,10 +706,10 @@ __mutex_lock_common(struct mutex *lock, unsigned int state, unsigned int subclas
debug_mutex_free_waiter(&waiter);
-skip_wait:
/* got the lock - cleanup and rejoice! */
lock_acquired(&lock->dep_map, ip);
+skip_wait:
if (ww_ctx)
ww_mutex_lock_acquired(ww, ww_ctx);
--
2.35.0.263.gb82422642f-goog
^ permalink raw reply related [flat|nested] 22+ messages in thread
* Re: [PATCH 02/12] cgroup: rstat: Make cgroup_rstat_cpu_lock name readable
@ 2022-02-09 20:20 ` Waiman Long
0 siblings, 0 replies; 22+ messages in thread
From: Waiman Long @ 2022-02-09 20:20 UTC (permalink / raw)
To: Namhyung Kim, Peter Zijlstra, Ingo Molnar, Will Deacon, Boqun Feng
Cc: LKML, Thomas Gleixner, Steven Rostedt, Byungchul Park,
Paul E. McKenney, Mathieu Desnoyers, Radoslaw Burny, Zefan Li,
Johannes Weiner, cgroups, Tejun Heo
On 2/8/22 14:43, Namhyung Kim wrote:
> The raw_spin_lock_init() uses the argument to name its lockdep map.
> But passing per_cpu_ptr() macro directly makes it a very very long
> name as it expanded like below:
>
> ({ do { const void *__vpp_verify = (typeof((&cgroup_rstat_cpu_lock) ...
>
> Let's fix it by passing a local variable instead. With this change,
> the name now looks like:
>
> cgrp_rstat_cpu_lock
>
> Cc: Zefan Li <lizefan.x@bytedance.com>
> Cc: Johannes Weiner <hannes@cmpxchg.org>
> Cc: cgroups@vger.kernel.org
> Acked-by: Tejun Heo <tj@kernel.org>
> Signed-off-by: Namhyung Kim <namhyung@kernel.org>
> ---
> kernel/cgroup/rstat.c | 12 ++++++++++--
> 1 file changed, 10 insertions(+), 2 deletions(-)
>
> diff --git a/kernel/cgroup/rstat.c b/kernel/cgroup/rstat.c
> index 9d331ba44870..371f54a59c38 100644
> --- a/kernel/cgroup/rstat.c
> +++ b/kernel/cgroup/rstat.c
> @@ -286,9 +286,17 @@ void cgroup_rstat_exit(struct cgroup *cgrp)
> void __init cgroup_rstat_boot(void)
> {
> int cpu;
> + raw_spinlock_t *cgrp_rstat_cpu_lock;
>
> - for_each_possible_cpu(cpu)
> - raw_spin_lock_init(per_cpu_ptr(&cgroup_rstat_cpu_lock, cpu));
> + for_each_possible_cpu(cpu) {
> + /*
> + * raw_spin_lock_init() uses the argument to name the lock in
> + * lockdep results. Passing per_cpu_ptr() directly created a
> + * cryptic name as the macro was expanded before stringified.
> + */
> + cgrp_rstat_cpu_lock = per_cpu_ptr(&cgroup_rstat_cpu_lock, cpu);
> + raw_spin_lock_init(cgrp_rstat_cpu_lock);
> + }
> }
>
> /*
That is a standalone fix. Maybe Tejun can take it directly into the
cgroup tree.
Acked-by: Waiman Long <longman@redhat.com>
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [PATCH 02/12] cgroup: rstat: Make cgroup_rstat_cpu_lock name readable
@ 2022-02-09 20:20 ` Waiman Long
0 siblings, 0 replies; 22+ messages in thread
From: Waiman Long @ 2022-02-09 20:20 UTC (permalink / raw)
To: Namhyung Kim, Peter Zijlstra, Ingo Molnar, Will Deacon, Boqun Feng
Cc: LKML, Thomas Gleixner, Steven Rostedt, Byungchul Park,
Paul E. McKenney, Mathieu Desnoyers, Radoslaw Burny, Zefan Li,
Johannes Weiner, cgroups-u79uwXL29TY76Z2rM5mHXA, Tejun Heo
On 2/8/22 14:43, Namhyung Kim wrote:
> The raw_spin_lock_init() uses the argument to name its lockdep map.
> But passing per_cpu_ptr() macro directly makes it a very very long
> name as it expanded like below:
>
> ({ do { const void *__vpp_verify = (typeof((&cgroup_rstat_cpu_lock) ...
>
> Let's fix it by passing a local variable instead. With this change,
> the name now looks like:
>
> cgrp_rstat_cpu_lock
>
> Cc: Zefan Li <lizefan.x-EC8Uxl6Npydl57MIdRCFDg@public.gmane.org>
> Cc: Johannes Weiner <hannes-druUgvl0LCNAfugRpC6u6w@public.gmane.org>
> Cc: cgroups-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
> Acked-by: Tejun Heo <tj-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
> Signed-off-by: Namhyung Kim <namhyung-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
> ---
> kernel/cgroup/rstat.c | 12 ++++++++++--
> 1 file changed, 10 insertions(+), 2 deletions(-)
>
> diff --git a/kernel/cgroup/rstat.c b/kernel/cgroup/rstat.c
> index 9d331ba44870..371f54a59c38 100644
> --- a/kernel/cgroup/rstat.c
> +++ b/kernel/cgroup/rstat.c
> @@ -286,9 +286,17 @@ void cgroup_rstat_exit(struct cgroup *cgrp)
> void __init cgroup_rstat_boot(void)
> {
> int cpu;
> + raw_spinlock_t *cgrp_rstat_cpu_lock;
>
> - for_each_possible_cpu(cpu)
> - raw_spin_lock_init(per_cpu_ptr(&cgroup_rstat_cpu_lock, cpu));
> + for_each_possible_cpu(cpu) {
> + /*
> + * raw_spin_lock_init() uses the argument to name the lock in
> + * lockdep results. Passing per_cpu_ptr() directly created a
> + * cryptic name as the macro was expanded before stringified.
> + */
> + cgrp_rstat_cpu_lock = per_cpu_ptr(&cgroup_rstat_cpu_lock, cpu);
> + raw_spin_lock_init(cgrp_rstat_cpu_lock);
> + }
> }
>
> /*
That is a standalone fix. Maybe Tejun can take it directly into the
cgroup tree.
Acked-by: Waiman Long <longman-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [PATCH 02/12] cgroup: rstat: Make cgroup_rstat_cpu_lock name readable
@ 2022-02-09 23:29 ` Tejun Heo
0 siblings, 0 replies; 22+ messages in thread
From: Tejun Heo @ 2022-02-09 23:29 UTC (permalink / raw)
To: Waiman Long
Cc: Namhyung Kim, Peter Zijlstra, Ingo Molnar, Will Deacon,
Boqun Feng, LKML, Thomas Gleixner, Steven Rostedt,
Byungchul Park, Paul E. McKenney, Mathieu Desnoyers,
Radoslaw Burny, Zefan Li, Johannes Weiner, cgroups
On Wed, Feb 09, 2022 at 03:20:42PM -0500, Waiman Long wrote:
> That is a standalone fix. Maybe Tejun can take it directly into the cgroup
> tree.
>
> Acked-by: Waiman Long <longman@redhat.com>
Yeah, I sure can but the change is rather immaterial from cgroup POV and it
might be easier to route it with the rest of the patchset to avoid cross
tree pulls. I'd be happy either way. Please let me know once things settle
and the patchset is ready for merging.
Thanks.
--
tejun
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [PATCH 02/12] cgroup: rstat: Make cgroup_rstat_cpu_lock name readable
@ 2022-02-09 23:29 ` Tejun Heo
0 siblings, 0 replies; 22+ messages in thread
From: Tejun Heo @ 2022-02-09 23:29 UTC (permalink / raw)
To: Waiman Long
Cc: Namhyung Kim, Peter Zijlstra, Ingo Molnar, Will Deacon,
Boqun Feng, LKML, Thomas Gleixner, Steven Rostedt,
Byungchul Park, Paul E. McKenney, Mathieu Desnoyers,
Radoslaw Burny, Zefan Li, Johannes Weiner,
cgroups-u79uwXL29TY76Z2rM5mHXA
On Wed, Feb 09, 2022 at 03:20:42PM -0500, Waiman Long wrote:
> That is a standalone fix. Maybe Tejun can take it directly into the cgroup
> tree.
>
> Acked-by: Waiman Long <longman-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
Yeah, I sure can but the change is rather immaterial from cgroup POV and it
might be easier to route it with the rest of the patchset to avoid cross
tree pulls. I'd be happy either way. Please let me know once things settle
and the patchset is ready for merging.
Thanks.
--
tejun
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [PATCH 03/12] timer: Protect lockdep functions with #ifdef
2022-02-08 19:43 ` [PATCH 03/12] timer: Protect lockdep functions with #ifdef Namhyung Kim
@ 2022-04-10 22:46 ` Thomas Gleixner
0 siblings, 0 replies; 22+ messages in thread
From: Thomas Gleixner @ 2022-04-10 22:46 UTC (permalink / raw)
To: Namhyung Kim, Peter Zijlstra, Ingo Molnar, Will Deacon,
Waiman Long, Boqun Feng
Cc: LKML, Steven Rostedt, Byungchul Park, Paul E. McKenney,
Mathieu Desnoyers, Radoslaw Burny
On Tue, Feb 08 2022 at 11:43, Namhyung Kim wrote:
> With upcoming lock tracepoints config, it'd define some of lockdep
> functions without enabling CONFIG_LOCKDEP actually. The existing code
> assumes those functions will be removed by the preprocessor but it's
> not the case anymore. Let's protect the code with #ifdef's
> explicitly.
This is lame. It's certainly possible to come up with replacement macros
which are compiled out when LOCKDEP is off for these use cases which are
not involved in lock tracepoint magic. Then just change these places
over to the new macros and keep the code free from this #ifdef ugliness.
Thanks,
tglx
^ permalink raw reply [flat|nested] 22+ messages in thread
end of thread, other threads:[~2022-04-10 22:46 UTC | newest]
Thread overview: 22+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-02-08 19:43 [Intel-gfx] [RFC RESEND 00/12] locking: Separate lock tracepoints from lockdep/lock_stat (v1.1) Namhyung Kim
2022-02-08 19:43 ` Namhyung Kim
2022-02-08 19:43 ` Namhyung Kim
2022-02-08 19:43 ` [PATCH 01/12] locking: Pass correct outer wait type info Namhyung Kim
2022-02-08 19:43 ` [PATCH 02/12] cgroup: rstat: Make cgroup_rstat_cpu_lock name readable Namhyung Kim
2022-02-08 19:43 ` Namhyung Kim
2022-02-09 20:20 ` Waiman Long
2022-02-09 20:20 ` Waiman Long
2022-02-09 23:29 ` Tejun Heo
2022-02-09 23:29 ` Tejun Heo
2022-02-08 19:43 ` [PATCH 03/12] timer: Protect lockdep functions with #ifdef Namhyung Kim
2022-04-10 22:46 ` Thomas Gleixner
2022-02-08 19:43 ` [PATCH 04/12] workqueue: " Namhyung Kim
2022-02-08 19:43 ` [Intel-gfx] [PATCH 05/12] drm/i915: " Namhyung Kim
2022-02-08 19:43 ` Namhyung Kim
2022-02-08 19:43 ` [PATCH 06/12] btrfs: change lockdep class size check using ks->names Namhyung Kim
2022-02-08 19:43 ` [PATCH 07/12] locking: Introduce CONFIG_LOCK_INFO Namhyung Kim
2022-02-08 19:43 ` [PATCH 08/12] locking/mutex: Init name properly w/ CONFIG_LOCK_INFO Namhyung Kim
2022-02-08 19:43 ` [PATCH 09/12] locking: Add more static lockdep init macros Namhyung Kim
2022-02-08 19:43 ` [PATCH 10/12] locking: Add CONFIG_LOCK_TRACEPOINTS option Namhyung Kim
2022-02-08 19:43 ` [PATCH 11/12] locking/mutex: Revive fast functions for CONFIG_LOCK_TRACEPOINTS Namhyung Kim
2022-02-08 19:43 ` [PATCH 12/12] locking: Move lock_acquired() from the fast path Namhyung Kim
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.