Re: [PATCH] tree/tiny rcu: Add debug RCU head option (v2)

[Mathieu Desnoyers <mathieu.desnoyers@efficios.com>]

* Paul E. McKenney (paulmck@linux.vnet.ibm.com) wrote:
> On Fri, Mar 19, 2010 at 12:13:53PM +0800, Lai Jiangshan wrote:
> > CC: Christoph Lameter <cl@linux-foundation.org>
> > CC: Pekka Enberg <penberg@cs.helsinki.fi>
> > 
> > I have written a patch like this before.
> > 
> > but SLUB opposed me, so I did not post it.
> > 
> > @ slub.c
> > static void free_slab(struct kmem_cache *s, struct page *page)
> > {
> > 	if (unlikely(s->flags & SLAB_DESTROY_BY_RCU)) {
> > 		/*
> > 		 * RCU free overloads the RCU head over the LRU
> > 		 */
> > 		struct rcu_head *head = (void *)&page->lru;
> > 
> > 		call_rcu(head, rcu_free_slab);
> > 	} else
> > 		__free_slab(s, page);
> > }
> > 
> > I think it is not your patch's fault, but the above code has
> > already existed.
> > 
> > In your patch, struct rcu_head::debug will share the same memory of
> > the other field of struct page, (especially share the struct page::flags
> > which I encountered)
> > 
> > I tested it very long time before, so I can't give you dmesg.
> 
> Ouch!!!
> 
> If this only causes a problem for SLUB, maybe a workaround is to make
> the config variable depend on !SLUB.  But this would seem to increase
> the urgency of using the debug objects

Hrm.. looking into the port to debugobjects, there are a few things that are not
precisely well set at the moment for it:

- We need to have two distinct active states, with a check that we move between
  the two. This does not seem to be currently supported by debugobjects.
- RCU heads are never explicitely freed. So yeah, that means we will have to
  modify a lot of callers.

Thanks,

Mathieu


> 
> 							Thanx, Paul
> 
> > Lai
> > 
> > Mathieu Desnoyers wrote:
> > > Poisoning the rcu_head callback list.
> > > 
> > > Helps finding racy users of call_rcu(), which results in hangs because list
> > > entries are overwritten and/or skipped.
> > > 
> > > A lot of non-initialized rcu list heads will be found with this option enabled.
> > > It is important to do not just blindly initialize them before each call_rcu, but
> > > rather to perform the initialization at the proper location, after the memory
> > > has been allocated, so we can effectively detect incorrect call_rcu users.
> > > 
> > > This patch version does not use "debug objects", although it probably should.
> > > Some day I might find time to look into this transition, but the patch is usable
> > > as is.
> > > 
> > > The weird #ifdef in the networking code comes from an agreement with Eric
> > > Dumazet about how to disable the build check for the networking code.  For those
> > > who wonder about the existence of this build-time check: it generates a build
> > > error when the size of struct dst_entry grows (because it is very performance
> > > sensitive). So the agreement here has been to disable the check when the
> > > DEBUG_RCU_HEAD config option is active.
> > > 
> > > Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
> > > CC: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
> > > CC: mingo@elte.hu
> > > CC: akpm@linux-foundation.org
> > > CC: mingo@elte.hu
> > > CC: laijs@cn.fujitsu.com
> > > CC: dipankar@in.ibm.com
> > > CC: josh@joshtriplett.org
> > > CC: dvhltc@us.ibm.com
> > > CC: niv@us.ibm.com
> > > CC: tglx@linutronix.de
> > > CC: peterz@infradead.org
> > > CC: rostedt@goodmis.org
> > > CC: Valdis.Kletnieks@vt.edu
> > > CC: dhowells@redhat.com
> > > CC: eric.dumazet@gmail.com
> > > CC: Alexey Dobriyan <adobriyan@gmail.com>
> > > CC: Peter Zijlstra <a.p.zijlstra@chello.nl>
> > > ---
> > >  include/linux/rcupdate.h |   13 ++++++++++++-
> > >  include/net/dst.h        |    2 ++
> > >  kernel/rcutiny.c         |    9 +++++++++
> > >  kernel/rcutree.c         |   10 ++++++++++
> > >  lib/Kconfig.debug        |    9 +++++++++
> > >  5 files changed, 42 insertions(+), 1 deletion(-)
> > > 
> > > Index: linux-2.6-lttng/include/linux/rcupdate.h
> > > ===================================================================
> > > --- linux-2.6-lttng.orig/include/linux/rcupdate.h	2010-03-18 20:27:25.000000000 -0400
> > > +++ linux-2.6-lttng/include/linux/rcupdate.h	2010-03-18 20:30:54.000000000 -0400
> > > @@ -49,6 +49,9 @@
> > >  struct rcu_head {
> > >  	struct rcu_head *next;
> > >  	void (*func)(struct rcu_head *head);
> > > +#ifdef CONFIG_DEBUG_RCU_HEAD
> > > +	struct rcu_head *debug;
> > > +#endif
> > >  };
> > >  
> > >  /* Exported common interfaces */
> > > @@ -71,11 +74,19 @@ extern void rcu_init(void);
> > >  #error "Unknown RCU implementation specified to kernel configuration"
> > >  #endif
> > >  
> > > +#ifdef CONFIG_DEBUG_RCU_HEAD
> > > +#define RCU_HEAD_INIT 	{ .next = NULL, .func = NULL, .debug = NULL }
> > > +#define INIT_RCU_HEAD(ptr) do { \
> > > +       (ptr)->next = NULL; (ptr)->func = NULL; (ptr)->debug = NULL; \
> > > +} while (0)
> > > +#else
> > >  #define RCU_HEAD_INIT	{ .next = NULL, .func = NULL }
> > > -#define RCU_HEAD(head) struct rcu_head head = RCU_HEAD_INIT
> > >  #define INIT_RCU_HEAD(ptr) do { \
> > >         (ptr)->next = NULL; (ptr)->func = NULL; \
> > >  } while (0)
> > > +#endif
> > > +
> > > +#define RCU_HEAD(head) struct rcu_head head = RCU_HEAD_INIT
> > >  
> > >  #ifdef CONFIG_DEBUG_LOCK_ALLOC
> > >  extern struct lockdep_map rcu_lock_map;
> > > Index: linux-2.6-lttng/kernel/rcutree.c
> > > ===================================================================
> > > --- linux-2.6-lttng.orig/kernel/rcutree.c	2010-03-18 20:27:25.000000000 -0400
> > > +++ linux-2.6-lttng/kernel/rcutree.c	2010-03-18 20:31:16.000000000 -0400
> > > @@ -39,6 +39,7 @@
> > >  #include <asm/atomic.h>
> > >  #include <linux/bitops.h>
> > >  #include <linux/module.h>
> > > +#include <linux/poison.h>
> > >  #include <linux/completion.h>
> > >  #include <linux/moduleparam.h>
> > >  #include <linux/percpu.h>
> > > @@ -1060,6 +1061,10 @@ static void rcu_do_batch(struct rcu_stat
> > >  		next = list->next;
> > >  		prefetch(next);
> > >  		trace_rcu_tree_callback(list);
> > > +#ifdef CONFIG_DEBUG_RCU_HEAD
> > > +		WARN_ON_ONCE(list->debug != LIST_POISON1);
> > > +		list->debug = NULL;
> > > +#endif
> > >  		list->func(list);
> > >  		list = next;
> > >  		if (++count >= rdp->blimit)
> > > @@ -1350,6 +1355,11 @@ __call_rcu(struct rcu_head *head, void (
> > >  	unsigned long flags;
> > >  	struct rcu_data *rdp;
> > >  
> > > +#ifdef CONFIG_DEBUG_RCU_HEAD
> > > +	WARN_ON_ONCE(head->debug);
> > > +	head->debug = LIST_POISON1;
> > > +#endif
> > > +
> > >  	head->func = func;
> > >  	head->next = NULL;
> > >  
> > > Index: linux-2.6-lttng/lib/Kconfig.debug
> > > ===================================================================
> > > --- linux-2.6-lttng.orig/lib/Kconfig.debug	2010-03-18 20:27:25.000000000 -0400
> > > +++ linux-2.6-lttng/lib/Kconfig.debug	2010-03-18 20:27:38.000000000 -0400
> > > @@ -661,6 +661,15 @@ config DEBUG_LIST
> > >  
> > >  	  If unsure, say N.
> > >  
> > > +config DEBUG_RCU_HEAD
> > > +	bool "Debug RCU callbacks"
> > > +	depends on DEBUG_KERNEL
> > > +	depends on TREE_RCU
> > > +	help
> > > +	  Enable this to turn on debugging of RCU list heads (call_rcu() usage).
> > > +	  Seems to find problems more quickly with stress-tests in single-cpu
> > > +	  mode.
> > > +
> > >  config DEBUG_SG
> > >  	bool "Debug SG table operations"
> > >  	depends on DEBUG_KERNEL
> > > Index: linux-2.6-lttng/include/net/dst.h
> > > ===================================================================
> > > --- linux-2.6-lttng.orig/include/net/dst.h	2010-03-18 20:27:25.000000000 -0400
> > > +++ linux-2.6-lttng/include/net/dst.h	2010-03-18 20:35:02.000000000 -0400
> > > @@ -159,7 +159,9 @@ static inline void dst_hold(struct dst_e
> > >  	 * If your kernel compilation stops here, please check
> > >  	 * __pad_to_align_refcnt declaration in struct dst_entry
> > >  	 */
> > > +#ifndef CONFIG_DEBUG_RCU_HEAD
> > >  	BUILD_BUG_ON(offsetof(struct dst_entry, __refcnt) & 63);
> > > +#endif
> > >  	atomic_inc(&dst->__refcnt);
> > >  }
> > >  
> > > Index: linux-2.6-lttng/kernel/rcutiny.c
> > > ===================================================================
> > > --- linux-2.6-lttng.orig/kernel/rcutiny.c	2010-03-18 20:35:14.000000000 -0400
> > > +++ linux-2.6-lttng/kernel/rcutiny.c	2010-03-18 20:39:12.000000000 -0400
> > > @@ -35,6 +35,7 @@
> > >  #include <linux/init.h>
> > >  #include <linux/time.h>
> > >  #include <linux/cpu.h>
> > > +#include <linux/poison.h>
> > >  
> > >  /* Global control variables for rcupdate callback mechanism. */
> > >  struct rcu_ctrlblk {
> > > @@ -163,6 +164,10 @@ static void __rcu_process_callbacks(stru
> > >  	while (list) {
> > >  		next = list->next;
> > >  		prefetch(next);
> > > +#ifdef CONFIG_DEBUG_RCU_HEAD
> > > +		WARN_ON_ONCE(list->debug != LIST_POISON1);
> > > +		list->debug = NULL;
> > > +#endif
> > >  		list->func(list);
> > >  		list = next;
> > >  	}
> > > @@ -210,6 +215,10 @@ static void __call_rcu(struct rcu_head *
> > >  {
> > >  	unsigned long flags;
> > >  
> > > +#ifdef CONFIG_DEBUG_RCU_HEAD
> > > +	WARN_ON_ONCE(head->debug);
> > > +	head->debug = LIST_POISON1;
> > > +#endif
> > >  	head->func = func;
> > >  	head->next = NULL;
> > >  
> > > 
> > 

-- 
Mathieu Desnoyers
Operating System Efficiency R&D Consultant
EfficiOS Inc.
http://www.efficios.com