On Fri, Sep 21, 2007 at 05:32:08PM -0400, Steven Rostedt wrote:
>
> --
> On Fri, 21 Sep 2007, Paul E. McKenney wrote:
>
> > On Thu, Sep 20, 2007 at 02:34:11PM -0400, Steven Rostedt wrote:
> > > In recent development of the RT kernel, some of our experimental code
> > > corrupted the rcu header.
On Sat, Sep 22, 2007 at 03:04:47AM -0400, Kyle Moffett wrote:
> On Sep 21, 2007, at 17:32:08, Steven Rostedt wrote:
> >On Fri, 21 Sep 2007, Paul E. McKenney wrote:
> >>On Thu, Sep 20, 2007 at 02:34:11PM -0400, Steven Rostedt wrote:
> >>>-#define RCU_HEAD_INIT { .next = NULL, .func = NULL }
>
On Sep 21, 2007, at 17:32:08, Steven Rostedt wrote:
On Fri, 21 Sep 2007, Paul E. McKenney wrote:
On Thu, Sep 20, 2007 at 02:34:11PM -0400, Steven Rostedt wrote:
-#define RCU_HEAD_INIT { .next = NULL, .func = NULL }
+#ifdef CONFIG_RCU_CRC_HEADER_CHECK
+
+#define RCU_CRC_MAGIC 0xC4809168UL
On Sep 21, 2007, at 17:32:08, Steven Rostedt wrote:
On Fri, 21 Sep 2007, Paul E. McKenney wrote:
On Thu, Sep 20, 2007 at 02:34:11PM -0400, Steven Rostedt wrote:
-#define RCU_HEAD_INIT { .next = NULL, .func = NULL }
+#ifdef CONFIG_RCU_CRC_HEADER_CHECK
+
+#define RCU_CRC_MAGIC 0xC4809168UL
On Sat, Sep 22, 2007 at 03:04:47AM -0400, Kyle Moffett wrote:
On Sep 21, 2007, at 17:32:08, Steven Rostedt wrote:
On Fri, 21 Sep 2007, Paul E. McKenney wrote:
On Thu, Sep 20, 2007 at 02:34:11PM -0400, Steven Rostedt wrote:
-#define RCU_HEAD_INIT { .next = NULL, .func = NULL }
+#ifdef
On Fri, Sep 21, 2007 at 05:32:08PM -0400, Steven Rostedt wrote:
--
On Fri, 21 Sep 2007, Paul E. McKenney wrote:
On Thu, Sep 20, 2007 at 02:34:11PM -0400, Steven Rostedt wrote:
In recent development of the RT kernel, some of our experimental code
corrupted the rcu header. But the side
--
On Fri, 21 Sep 2007, Paul E. McKenney wrote:
> On Thu, Sep 20, 2007 at 02:34:11PM -0400, Steven Rostedt wrote:
> > In recent development of the RT kernel, some of our experimental code
> > corrupted the rcu header. But the side effect (crashing) didn't rear its
> > ugly head until way after
On Thu, Sep 20, 2007 at 02:34:11PM -0400, Steven Rostedt wrote:
> In recent development of the RT kernel, some of our experimental code
> corrupted the rcu header. But the side effect (crashing) didn't rear its
> ugly head until way after the fact. Discussing this with Paul, he
> suggested that
--
>
> This has been wondering me some time. Kernel oopses also use [<%p>],
> but what really for are two sort of braces needed?
I believe the notation of [] has always been a
representation of instruction pointer. Seems that's what's used for all
outputs of instruction pointers that I've seen
On Sep 21 2007 14:02, Peter Zijlstra wrote:
>> +if (unlikely(head->crc != rcu_crc_calc(head)) && !once) {
>> +once++;
>> +printk("BUG: RCU check failed!");
>> +if (head->caller)
>> +printk(" (caller=%p)",
>> +
--
On Fri, 21 Sep 2007, Peter Zijlstra wrote:
> On Thu, 20 Sep 2007 14:34:11 -0400 Steven Rostedt <[EMAIL PROTECTED]>
> wrote:
>
>
> > +static inline void rcu_crc_check(struct rcu_head *head)
> > +{
> > + static int once;
> > + if (unlikely(head->crc != rcu_crc_calc(head)) && !once) {
> > +
On Thu, 20 Sep 2007 14:34:11 -0400 Steven Rostedt <[EMAIL PROTECTED]>
wrote:
> +static inline void rcu_crc_check(struct rcu_head *head)
> +{
> + static int once;
> + if (unlikely(head->crc != rcu_crc_calc(head)) && !once) {
> + once++;
> + printk("BUG: RCU check
On Sep 21 2007 14:02, Peter Zijlstra wrote:
+if (unlikely(head-crc != rcu_crc_calc(head)) !once) {
+once++;
+printk(BUG: RCU check failed!);
+if (head-caller)
+printk( (caller=%p),
+ head-caller);
--
On Fri, 21 Sep 2007, Peter Zijlstra wrote:
On Thu, 20 Sep 2007 14:34:11 -0400 Steven Rostedt [EMAIL PROTECTED]
wrote:
+static inline void rcu_crc_check(struct rcu_head *head)
+{
+ static int once;
+ if (unlikely(head-crc != rcu_crc_calc(head)) !once) {
+ once++;
--
This has been wondering me some time. Kernel oopses also use [%p],
but what really for are two sort of braces needed?
I believe the notation of [some-hex-number] has always been a
representation of instruction pointer. Seems that's what's used for all
outputs of instruction pointers that
On Thu, Sep 20, 2007 at 02:34:11PM -0400, Steven Rostedt wrote:
In recent development of the RT kernel, some of our experimental code
corrupted the rcu header. But the side effect (crashing) didn't rear its
ugly head until way after the fact. Discussing this with Paul, he
suggested that RCU
--
On Fri, 21 Sep 2007, Paul E. McKenney wrote:
On Thu, Sep 20, 2007 at 02:34:11PM -0400, Steven Rostedt wrote:
In recent development of the RT kernel, some of our experimental code
corrupted the rcu header. But the side effect (crashing) didn't rear its
ugly head until way after the
In recent development of the RT kernel, some of our experimental code
corrupted the rcu header. But the side effect (crashing) didn't rear its
ugly head until way after the fact. Discussing this with Paul, he
suggested that RCU should have a "self checking" mechanism to detect
these kind of
In recent development of the RT kernel, some of our experimental code
corrupted the rcu header. But the side effect (crashing) didn't rear its
ugly head until way after the fact. Discussing this with Paul, he
suggested that RCU should have a self checking mechanism to detect
these kind of issues.
19 matches
Mail list logo