Groups > linux.kernel > #1233609 > unrolled thread

[GIT PULL rcu/urgent] Fix two more 4.3 regressions

Started by	"Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
First post	2015-09-27 06:20 +0200
Last post	2015-09-28 08:10 +0200
Articles	2 — 2 participants

Back to article view | Back to linux.kernel

  [GIT PULL rcu/urgent] Fix two more 4.3 regressions "Paul E. McKenney" <paulmck@linux.vnet.ibm.com> - 2015-09-27 06:20 +0200
    Re: [GIT PULL rcu/urgent] Fix two more 4.3 regressions Ingo Molnar <mingo@kernel.org> - 2015-09-28 08:10 +0200

#1233609 — [GIT PULL rcu/urgent] Fix two more 4.3 regressions

From	"Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Date	2015-09-27 06:20 +0200
Subject	[GIT PULL rcu/urgent] Fix two more 4.3 regressions
Message-ID	<qdbBn-4ra-1@gated-at.bofh.it>

Hello, Ingo,

This series contains two small fixes for regressions introduced during
the past merge window:

1.	Recent versions of gcc are fatally confused by use of a const
	int variable for array bounds in inline functions, one of which
	was introduced by 114b7fd4b (rcu: Create rcu_sync infrastructure).
	For more information on the gcc bug, see:

		https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67055#c13

	This workaround for that gcc bug, courtesy of Oleg Nesterov,
	simply substitutes the constant expression for the two uses of
	the const int variable.

2.	Commit 2cd6ffafec06 (rcu: Extend expedited funnel locking
	to rcu_data structure) extended contention-reducing
	funnel locking from the rcu_node structures to the rcu_data
	structures, but failed to create a separate lock class for the
	rcu_sched_data structures, which is needed due to the fact that
	synchronize_rcu_expedited() invokes synchronize_sched_expedited().
	In the absence of this separate lock class, lockdep can give
	false-positive splats.	This commit therefore creates the
	required lock class so that lockdep can see that there is in
	fact no deadlock.

Both fixes have been exposed to -next and 0day Test Robot testing.

These changes are available in the git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/paulmck/linux-rcu.git for-mingo

for you to fetch changes up to 19a5ecde086a6a5287978b12ae948fa691b197b7:

  rcu: Suppress lockdep false positive for rcp->exp_funnel_mutex (2015-09-20 21:01:22 -0700)

----------------------------------------------------------------
Oleg Nesterov (1):
      rcu: Change _wait_rcu_gp() to work around GCC bug 67055

Paul E. McKenney (1):
      rcu: Suppress lockdep false positive for rcp->exp_funnel_mutex

 include/linux/rcupdate.h | 11 +++++------
 kernel/rcu/tree.c        |  5 +++++
 2 files changed, 10 insertions(+), 6 deletions(-)

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [next] | [standalone]

#1233870

From	Ingo Molnar <mingo@kernel.org>
Date	2015-09-28 08:10 +0200
Message-ID	<qdzNn-5Ey-13@gated-at.bofh.it>
In reply to	#1233609

* Paul E. McKenney <paulmck@linux.vnet.ibm.com> wrote:

> Hello, Ingo,
> 
> This series contains two small fixes for regressions introduced during
> the past merge window:
> 
> 1.	Recent versions of gcc are fatally confused by use of a const
> 	int variable for array bounds in inline functions, one of which
> 	was introduced by 114b7fd4b (rcu: Create rcu_sync infrastructure).
> 	For more information on the gcc bug, see:
> 
> 		https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67055#c13
> 
> 	This workaround for that gcc bug, courtesy of Oleg Nesterov,
> 	simply substitutes the constant expression for the two uses of
> 	the const int variable.
> 
> 2.	Commit 2cd6ffafec06 (rcu: Extend expedited funnel locking
> 	to rcu_data structure) extended contention-reducing
> 	funnel locking from the rcu_node structures to the rcu_data
> 	structures, but failed to create a separate lock class for the
> 	rcu_sched_data structures, which is needed due to the fact that
> 	synchronize_rcu_expedited() invokes synchronize_sched_expedited().
> 	In the absence of this separate lock class, lockdep can give
> 	false-positive splats.	This commit therefore creates the
> 	required lock class so that lockdep can see that there is in
> 	fact no deadlock.
> 
> Both fixes have been exposed to -next and 0day Test Robot testing.
> 
> These changes are available in the git repository at:
> 
>   git://git.kernel.org/pub/scm/linux/kernel/git/paulmck/linux-rcu.git for-mingo
> 
> for you to fetch changes up to 19a5ecde086a6a5287978b12ae948fa691b197b7:
> 
>   rcu: Suppress lockdep false positive for rcp->exp_funnel_mutex (2015-09-20 21:01:22 -0700)
> 
> ----------------------------------------------------------------
> Oleg Nesterov (1):
>       rcu: Change _wait_rcu_gp() to work around GCC bug 67055
> 
> Paul E. McKenney (1):
>       rcu: Suppress lockdep false positive for rcp->exp_funnel_mutex
> 
>  include/linux/rcupdate.h | 11 +++++------
>  kernel/rcu/tree.c        |  5 +++++
>  2 files changed, 10 insertions(+), 6 deletions(-)

Pulled, thanks a lot Paul!

	Ingo
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [prev] | [standalone]

csiph-web

[GIT PULL rcu/urgent] Fix two more 4.3 regressions

Contents

#1233609 — [GIT PULL rcu/urgent] Fix two more 4.3 regressions

#1233870