Re: [PATCH] futex: move debug_rt_mutex_free_waiter() further down

2017-03-13 Thread Peter Zijlstra 
On Wed, Mar 08, 2017 at 04:29:02PM +0100, Sebastian Andrzej Siewior wrote:

>  kernel/futex.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/kernel/futex.c b/kernel/futex.c
> index 00ec4a01d3f5..73abfe0da4d0 100644
> --- a/kernel/futex.c
> +++ b/kernel/futex.c
> @@ -3046,11 +3046,11 @@ static int futex_wait_requeue_pi(u32 __user *uaddr, 
> unsigned int flags,
>   WARN_ON(!q.pi_state);
>   pi_mutex = _state->pi_mutex;
>   ret = rt_mutex_wait_proxy_lock(pi_mutex, to, _waiter);
> - debug_rt_mutex_free_waiter(_waiter);
>  
>   spin_lock(q.lock_ptr);
>   if (ret && !rt_mutex_cleanup_proxy_lock(pi_mutex, _waiter))
>   ret = 0;
> + debug_rt_mutex_free_waiter(_waiter);
>  
>   /*
>* Fixup the pi_state owner and possibly acquire the lock if we
> 

Thanks, folded.

Re: [PATCH] futex: move debug_rt_mutex_free_waiter() further down

2017-03-13 Thread Peter Zijlstra 
On Wed, Mar 08, 2017 at 04:29:02PM +0100, Sebastian Andrzej Siewior wrote:

>  kernel/futex.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/kernel/futex.c b/kernel/futex.c
> index 00ec4a01d3f5..73abfe0da4d0 100644
> --- a/kernel/futex.c
> +++ b/kernel/futex.c
> @@ -3046,11 +3046,11 @@ static int futex_wait_requeue_pi(u32 __user *uaddr, 
> unsigned int flags,
>   WARN_ON(!q.pi_state);
>   pi_mutex = _state->pi_mutex;
>   ret = rt_mutex_wait_proxy_lock(pi_mutex, to, _waiter);
> - debug_rt_mutex_free_waiter(_waiter);
>  
>   spin_lock(q.lock_ptr);
>   if (ret && !rt_mutex_cleanup_proxy_lock(pi_mutex, _waiter))
>   ret = 0;
> + debug_rt_mutex_free_waiter(_waiter);
>  
>   /*
>* Fixup the pi_state owner and possibly acquire the lock if we
> 

Thanks, folded.

Re: [PATCH] futex: move debug_rt_mutex_free_waiter() further down

2017-03-08 Thread Steven Rostedt 
On Wed, 8 Mar 2017 16:37:32 +0100
Sebastian Andrzej Siewior  wrote:

> On 2017-03-08 16:29:02 [+0100], To Peter Zijlstra wrote:
> > Without this, futex_requeue_pi_signal_restart will trigger
> > 
> > |kernel BUG at locking/rtmutex_common.h:55!
> > |Call Trace:
> > | rt_mutex_cleanup_proxy_lock+0x54/0x90
> > | futex_wait_requeue_pi.constprop.21+0x387/0x4d0
> > | do_futex+0x289/0xbf0
> > |RIP: remove_waiter+0x157/0x170 RSP: c9e0fbe0
> > 
> > with BUG  != pointer once this patch is applied.  
> 
> My wording is wrong. This BUG_ON() statement described here in this
> patch (together with the test case mentioned) will trigger once
> 
>  "[PATCH -v5 12/14] futex,rt_mutex: Restructure rt_mutex_finish_proxy_lock()"
> 
> is applied.
>

Now I read this. Ignore my last email.

-- Steve

Re: [PATCH] futex: move debug_rt_mutex_free_waiter() further down

2017-03-08 Thread Steven Rostedt 
On Wed, 8 Mar 2017 16:37:32 +0100
Sebastian Andrzej Siewior  wrote:

> On 2017-03-08 16:29:02 [+0100], To Peter Zijlstra wrote:
> > Without this, futex_requeue_pi_signal_restart will trigger
> > 
> > |kernel BUG at locking/rtmutex_common.h:55!
> > |Call Trace:
> > | rt_mutex_cleanup_proxy_lock+0x54/0x90
> > | futex_wait_requeue_pi.constprop.21+0x387/0x4d0
> > | do_futex+0x289/0xbf0
> > |RIP: remove_waiter+0x157/0x170 RSP: c9e0fbe0
> > 
> > with BUG  != pointer once this patch is applied.  
> 
> My wording is wrong. This BUG_ON() statement described here in this
> patch (together with the test case mentioned) will trigger once
> 
>  "[PATCH -v5 12/14] futex,rt_mutex: Restructure rt_mutex_finish_proxy_lock()"
> 
> is applied.
>

Now I read this. Ignore my last email.

-- Steve

Re: [PATCH] futex: move debug_rt_mutex_free_waiter() further down

2017-03-08 Thread Steven Rostedt 
On Wed, 8 Mar 2017 16:29:02 +0100
Sebastian Andrzej Siewior  wrote:

> Without this, futex_requeue_pi_signal_restart will trigger
> 
> |kernel BUG at locking/rtmutex_common.h:55!
> |Call Trace:
> | rt_mutex_cleanup_proxy_lock+0x54/0x90
> | futex_wait_requeue_pi.constprop.21+0x387/0x4d0
> | do_futex+0x289/0xbf0
> |RIP: remove_waiter+0x157/0x170 RSP: c9e0fbe0
> 
> with BUG  != pointer once this patch is applied.

This sentence makes no sense. It's a no-sensetence ;-)

-- Steve

> 
> Signed-off-by: Sebastian Andrzej Siewior 
> ---
>  kernel/futex.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/kernel/futex.c b/kernel/futex.c
> index 00ec4a01d3f5..73abfe0da4d0 100644
> --- a/kernel/futex.c
> +++ b/kernel/futex.c
> @@ -3046,11 +3046,11 @@ static int futex_wait_requeue_pi(u32 __user *uaddr, 
> unsigned int flags,
>   WARN_ON(!q.pi_state);
>   pi_mutex = _state->pi_mutex;
>   ret = rt_mutex_wait_proxy_lock(pi_mutex, to, _waiter);
> - debug_rt_mutex_free_waiter(_waiter);
>  
>   spin_lock(q.lock_ptr);
>   if (ret && !rt_mutex_cleanup_proxy_lock(pi_mutex, _waiter))
>   ret = 0;
> + debug_rt_mutex_free_waiter(_waiter);
>  
>   /*
>* Fixup the pi_state owner and possibly acquire the lock if we

Re: [PATCH] futex: move debug_rt_mutex_free_waiter() further down

2017-03-08 Thread Steven Rostedt 
On Wed, 8 Mar 2017 16:29:02 +0100
Sebastian Andrzej Siewior  wrote:

> Without this, futex_requeue_pi_signal_restart will trigger
> 
> |kernel BUG at locking/rtmutex_common.h:55!
> |Call Trace:
> | rt_mutex_cleanup_proxy_lock+0x54/0x90
> | futex_wait_requeue_pi.constprop.21+0x387/0x4d0
> | do_futex+0x289/0xbf0
> |RIP: remove_waiter+0x157/0x170 RSP: c9e0fbe0
> 
> with BUG  != pointer once this patch is applied.

This sentence makes no sense. It's a no-sensetence ;-)

-- Steve

> 
> Signed-off-by: Sebastian Andrzej Siewior 
> ---
>  kernel/futex.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/kernel/futex.c b/kernel/futex.c
> index 00ec4a01d3f5..73abfe0da4d0 100644
> --- a/kernel/futex.c
> +++ b/kernel/futex.c
> @@ -3046,11 +3046,11 @@ static int futex_wait_requeue_pi(u32 __user *uaddr, 
> unsigned int flags,
>   WARN_ON(!q.pi_state);
>   pi_mutex = _state->pi_mutex;
>   ret = rt_mutex_wait_proxy_lock(pi_mutex, to, _waiter);
> - debug_rt_mutex_free_waiter(_waiter);
>  
>   spin_lock(q.lock_ptr);
>   if (ret && !rt_mutex_cleanup_proxy_lock(pi_mutex, _waiter))
>   ret = 0;
> + debug_rt_mutex_free_waiter(_waiter);
>  
>   /*
>* Fixup the pi_state owner and possibly acquire the lock if we

Re: [PATCH] futex: move debug_rt_mutex_free_waiter() further down

2017-03-08 Thread Sebastian Andrzej Siewior 
On 2017-03-08 16:29:02 [+0100], To Peter Zijlstra wrote:
> Without this, futex_requeue_pi_signal_restart will trigger
> 
> |kernel BUG at locking/rtmutex_common.h:55!
> |Call Trace:
> | rt_mutex_cleanup_proxy_lock+0x54/0x90
> | futex_wait_requeue_pi.constprop.21+0x387/0x4d0
> | do_futex+0x289/0xbf0
> |RIP: remove_waiter+0x157/0x170 RSP: c9e0fbe0
> 
> with BUG  != pointer once this patch is applied.

My wording is wrong. This BUG_ON() statement described here in this
patch (together with the test case mentioned) will trigger once

 "[PATCH -v5 12/14] futex,rt_mutex: Restructure rt_mutex_finish_proxy_lock()"

is applied.

Sebastian

Re: [PATCH] futex: move debug_rt_mutex_free_waiter() further down

2017-03-08 Thread Sebastian Andrzej Siewior 
On 2017-03-08 16:29:02 [+0100], To Peter Zijlstra wrote:
> Without this, futex_requeue_pi_signal_restart will trigger
> 
> |kernel BUG at locking/rtmutex_common.h:55!
> |Call Trace:
> | rt_mutex_cleanup_proxy_lock+0x54/0x90
> | futex_wait_requeue_pi.constprop.21+0x387/0x4d0
> | do_futex+0x289/0xbf0
> |RIP: remove_waiter+0x157/0x170 RSP: c9e0fbe0
> 
> with BUG  != pointer once this patch is applied.

My wording is wrong. This BUG_ON() statement described here in this
patch (together with the test case mentioned) will trigger once

 "[PATCH -v5 12/14] futex,rt_mutex: Restructure rt_mutex_finish_proxy_lock()"

is applied.

Sebastian

[PATCH] futex: move debug_rt_mutex_free_waiter() further down

2017-03-08 Thread Sebastian Andrzej Siewior 
Without this, futex_requeue_pi_signal_restart will trigger

|kernel BUG at locking/rtmutex_common.h:55!
|Call Trace:
| rt_mutex_cleanup_proxy_lock+0x54/0x90
| futex_wait_requeue_pi.constprop.21+0x387/0x4d0
| do_futex+0x289/0xbf0
|RIP: remove_waiter+0x157/0x170 RSP: c9e0fbe0

with BUG  != pointer once this patch is applied.

Signed-off-by: Sebastian Andrzej Siewior 
---
 kernel/futex.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/kernel/futex.c b/kernel/futex.c
index 00ec4a01d3f5..73abfe0da4d0 100644
--- a/kernel/futex.c
+++ b/kernel/futex.c
@@ -3046,11 +3046,11 @@ static int futex_wait_requeue_pi(u32 __user *uaddr, 
unsigned int flags,
WARN_ON(!q.pi_state);
pi_mutex = _state->pi_mutex;
ret = rt_mutex_wait_proxy_lock(pi_mutex, to, _waiter);
-   debug_rt_mutex_free_waiter(_waiter);
 
spin_lock(q.lock_ptr);
if (ret && !rt_mutex_cleanup_proxy_lock(pi_mutex, _waiter))
ret = 0;
+   debug_rt_mutex_free_waiter(_waiter);
 
/*
 * Fixup the pi_state owner and possibly acquire the lock if we
-- 
2.11.0

[PATCH] futex: move debug_rt_mutex_free_waiter() further down

2017-03-08 Thread Sebastian Andrzej Siewior 
Without this, futex_requeue_pi_signal_restart will trigger

|kernel BUG at locking/rtmutex_common.h:55!
|Call Trace:
| rt_mutex_cleanup_proxy_lock+0x54/0x90
| futex_wait_requeue_pi.constprop.21+0x387/0x4d0
| do_futex+0x289/0xbf0
|RIP: remove_waiter+0x157/0x170 RSP: c9e0fbe0

with BUG  != pointer once this patch is applied.

Signed-off-by: Sebastian Andrzej Siewior 
---
 kernel/futex.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/kernel/futex.c b/kernel/futex.c
index 00ec4a01d3f5..73abfe0da4d0 100644
--- a/kernel/futex.c
+++ b/kernel/futex.c
@@ -3046,11 +3046,11 @@ static int futex_wait_requeue_pi(u32 __user *uaddr, 
unsigned int flags,
WARN_ON(!q.pi_state);
pi_mutex = _state->pi_mutex;
ret = rt_mutex_wait_proxy_lock(pi_mutex, to, _waiter);
-   debug_rt_mutex_free_waiter(_waiter);
 
spin_lock(q.lock_ptr);
if (ret && !rt_mutex_cleanup_proxy_lock(pi_mutex, _waiter))
ret = 0;
+   debug_rt_mutex_free_waiter(_waiter);
 
/*
 * Fixup the pi_state owner and possibly acquire the lock if we
-- 
2.11.0