[GIT RFC PULL] RCU changes for 2.6.35

[GIT RFC PULL] RCU changes for 2.6.35

[Paul E. McKenney]

Hello, Ingo,

Here is the first installment of RCU changes for 2.6.35.  These commits
are based on top of the rcu/urgent set sent yesterday.  If you would rather
rcu/urgent be merged into tip/core/rcu and then this set based on top of
that, or some other arrangement, please let me know.

git://git.kernel.org/pub/scm/linux/kernel/git/paulmck/linux-2.6-rcu.git rcu/next

There are two more sets of commits queued:

o	Mathieu's debugobjects changes.  These need to be reworked in
	order to coordinate with other maintainer trees, and also to
	get the relevant Acked-by:s.  This will eventually apply in
	three stages: (1) add new RCU APIs, (2) apply these APIs into
	various trees, and (3) remove old RCU APIs.

o	Arnd's "sparse" checks.  These require a quadruple-action approach:
	(1) add a new RCU API member rcu_dereference_index_check(),
	(2) a couple of commits must be accepted into maintainer trees
	(networking, vfs, and mce), (3) a pair of RCU-infrastructure
	patches, and finally (4) a number of patches to apply __rcu
	attributes to various trees.

							Thanx, Paul

------------------>
 Lai Jiangshan (4)
	 rcu: optionally leave lockdep enabled after RCU lockdep splat
	 rcu: make dead code really dead
	 rcu: move some code from macro to function
	 rcu: ignore offline CPUs in last non-dyntick-idle CPU check

 Paul E. McKenney (16)
	 rcu: substitute set_need_resched for sending resched IPIs
	 rcu: Fix bogus CONFIG_PROVE_LOCKING in comments to reflect reality.
	 rcu: fix now-bogus rcu_scheduler_active comments.
	 rcu: shrink rcutiny by making synchronize_rcu_bh() be inline
	 rcu: rename rcutiny rcu_ctrlblk to rcu_sched_ctrlblk
	 rcu: refactor RCU's context-switch handling
	 rcu: slim down rcutiny by removing rcu_scheduler_active and friends
	 rcu: enable CPU_STALL_VERBOSE by default
	 rcu: disable CPU stall warnings upon panic
	 rcu: print boot-time console messages if RCU configs out of ordinary
	 rcu: improve RCU CPU stall-warning messages
	 rcu: permit discontiguous cpu_possible_mask CPU numbering
	 rcu: reduce the number of spurious RCU_SOFTIRQ invocations
	 rcu: improve the RCU CPU-stall warning documentation
	 rcu: make SRCU usable in modules
	 rcu: RCU_FAST_NO_HZ must check RCU dyntick state

Re: [GIT RFC PULL] RCU changes for 2.6.35

[Ingo Molnar]

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

> Hello, Ingo,
> 
> Here is the first installment of RCU changes for 2.6.35.  These commits
> are based on top of the rcu/urgent set sent yesterday.  If you would rather
> rcu/urgent be merged into tip/core/rcu and then this set based on top of
> that, or some other arrangement, please let me know.
> 
> git://git.kernel.org/pub/scm/linux/kernel/git/paulmck/linux-2.6-rcu.git rcu/next

FYI, it doesnt build on x86 allyesconfig:

In file included from kernel/rcutree.c:1961:
kernel/rcutree_plugin.h: In function ?rcu_needs_cpu?:
kernel/rcutree_plugin.h:1066: error: invalid type argument of ?->? (have ?struct rcu_dynticks?)
kernel/rcutree_plugin.h:1067: error: invalid type argument of ?->? (have ?struct rcu_dynticks?)

Thanks,

	Ingo

Re: [GIT RFC PULL] RCU changes for 2.6.35

[Paul E. McKenney]

On Sat, May 08, 2010 at 10:34:21AM +0200, Ingo Molnar wrote:
> 
> * Paul E. McKenney <paulmck@linux.vnet.ibm.com> wrote:
> 
> > Hello, Ingo,
> > 
> > Here is the first installment of RCU changes for 2.6.35.  These commits
> > are based on top of the rcu/urgent set sent yesterday.  If you would rather
> > rcu/urgent be merged into tip/core/rcu and then this set based on top of
> > that, or some other arrangement, please let me know.
> > 
> > git://git.kernel.org/pub/scm/linux/kernel/git/paulmck/linux-2.6-rcu.git rcu/next
> 
> FYI, it doesnt build on x86 allyesconfig:
> 
> In file included from kernel/rcutree.c:1961:
> kernel/rcutree_plugin.h: In function ?rcu_needs_cpu?:
> kernel/rcutree_plugin.h:1066: error: invalid type argument of ?->? (have ?struct rcu_dynticks?)
> kernel/rcutree_plugin.h:1067: error: invalid type argument of ?->? (have ?struct rcu_dynticks?)

<red face>

I forgot to move CONFIG_RCU_FAST_NO_HZ from my development tests to my
regression tests.  I have done so now.  I have also added allyesconfig
to my compile tests, which I should have done long ago.

I am testing the fix, and will rebase the stack onto 2.6.34-rc7 when it
comes out.

							Thanx, Paul

Re: [GIT RFC PULL] RCU changes for 2.6.35

[Ingo Molnar]

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

> On Sat, May 08, 2010 at 10:34:21AM +0200, Ingo Molnar wrote:
> > 
> > * Paul E. McKenney <paulmck@linux.vnet.ibm.com> wrote:
> > 
> > > Hello, Ingo,
> > > 
> > > Here is the first installment of RCU changes for 2.6.35.  These commits
> > > are based on top of the rcu/urgent set sent yesterday.  If you would rather
> > > rcu/urgent be merged into tip/core/rcu and then this set based on top of
> > > that, or some other arrangement, please let me know.
> > > 
> > > git://git.kernel.org/pub/scm/linux/kernel/git/paulmck/linux-2.6-rcu.git rcu/next
> > 
> > FYI, it doesnt build on x86 allyesconfig:
> > 
> > In file included from kernel/rcutree.c:1961:
> > kernel/rcutree_plugin.h: In function ?rcu_needs_cpu?:
> > kernel/rcutree_plugin.h:1066: error: invalid type argument of ?->? (have ?struct rcu_dynticks?)
> > kernel/rcutree_plugin.h:1067: error: invalid type argument of ?->? (have ?struct rcu_dynticks?)
> 
> <red face>
> 
> I forgot to move CONFIG_RCU_FAST_NO_HZ from my development tests to my
> regression tests.  I have done so now.  I have also added allyesconfig
> to my compile tests, which I should have done long ago.
> 
> I am testing the fix, and will rebase the stack onto 2.6.34-rc7 when it
> comes out.

Cool, thanks Paul!

	Ingo