Re: [PATCH 2/2] futex: correct futex_requeue futex key ref counting in requeue loop

[Darren Hart <dvhltc@us.ibm.com>]

Darren Hart wrote:
> The requeue loop takes multiple references to key2, but the corresponding
> put loop decrements the refs for key1.  This patch corrects the accounting.
> 
> Build and boot tested on an x86_64 system.
> 
> Signed-off-by: Darren Hart <dvhltc@us.ibm.com>
> Cc: Peter Zijlstra <peterz@infradead.org>
> Cc: Thomas Gleixner <tglx@linutronix.de>
> Cc: Rusty Russell <rusty@au1.ibm.com>
> ---
> 
>  kernel/futex.c |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)
> 
> diff --git a/kernel/futex.c b/kernel/futex.c
> index cf363ce..3b66d91 100644
> --- a/kernel/futex.c
> +++ b/kernel/futex.c
> @@ -919,7 +919,7 @@ out_unlock:
> 
>  	/* drop_futex_key_refs() must be called outside the spinlocks. */
>  	while (--drop_count >= 0)
> -		drop_futex_key_refs(&key1);
> +		drop_futex_key_refs(&key2);
> 
>  	put_futex_key(fshared, &key2);
>  out_put_key1:
> 

Ugh, so I'm having second thoughts about this patch.  I believe what is 
happening here is that the requeue loop requeues each waiter from one 
futex (key1) to another (key2).  It rightly takes a reference to the 
futex at key2 and then decrements the references to key1 by drop_count 
(since the waiters now reference key2, not key1).  The newly taken key2 
references will be dropped in futex_wait() when each waiter is woken up 
and takes the futex.

I apologize for the confusion on this.  Thanks for suggesting I send 
this patch out independently from the rest Peter ;-)

If we can come to a consensus on this, I suggest pulling this patch from 
tip/core/futexes.


-- 
Darren Hart
IBM Linux Technology Center
Real-Time Linux Team