v2.6.22.1-rt5

we are pleased to announce the v2.6.22.1-rt5 kernel (collected and 
assembled by Thomas), which can be downloaded from the usual place:
  
  http://redhat.com/~mingo/realtime-preempt/
  
more info about the -rt patchset can be found in the RT wiki:
   
  http://rt.wiki.kernel.org

Changes since -rt4:

 - MM fix: PCP pages locking (Peter Zijlstra)

 - yield() fix: remove stale RCU-unlock (Daniel Walker)

 - dont do check_pgt_cache() in idle (found by Daniel Walker)

 - moved 8 patches, created 6 new patches out of larger patches to 
   make the -rt queue more bisectable (Daniel Walker)

 - cleanup (Daniel Walker)

 - fix the latency tracer under paravirt

to build a 2.6.22.1-rt5 tree, the following patches should be applied:
  
  http://kernel.org/pub/linux/kernel/v2.6/linux-2.6.22.1.tar.bz2
  http://redhat.com/~mingo/realtime-preempt/patch-2.6.22.1-rt5
  
	Ingo

Ingo Molnar wrote:

we are pleased to announce the v2.6.22.1-rt5 kernel (collected and 
assembled by Thomas), which can be downloaded from the usual place:
  
  http://redhat.com/~mingo/realtime-preempt/
  
more info about the -rt patchset can be found in the RT wiki:
   
  http://rt.wiki.kernel.org

Changes since -rt4:

 - MM fix: PCP pages locking (Peter Zijlstra)

 - yield() fix: remove stale RCU-unlock (Daniel Walker)

 - dont do check_pgt_cache() in idle (found by Daniel Walker)

 - moved 8 patches, created 6 new patches out of larger patches to 
   make the -rt queue more bisectable (Daniel Walker)

 - cleanup (Daniel Walker)

 - fix the latency tracer under paravirt

to build a 2.6.22.1-rt5 tree, the following patches should be applied:
  
  http://kernel.org/pub/linux/kernel/v2.6/linux-2.6.22.1.tar.bz2
  http://redhat.com/~mingo/realtime-preempt/patch-2.6.22.1-rt5
  
	Ingo
-

Maybe I was too quick, but `make all` on is failing here:

...
In file included from arch/i386/mm/pgtable.c:16:
include/linux/quicklist.h: In function ‘quicklist_alloc’:
include/linux/quicklist.h:50: warning: unused variable ‘q’
include/linux/quicklist.h: In function ‘__quicklist_free’:
include/linux/quicklist.h:79: error: ‘per_cpu__quicklist’ undeclared
(first use in this function)
include/linux/quicklist.h:79: error: (Each undeclared identifier is
reported only once
include/linux/quicklist.h:79: error: for each function it appears in.)
include/linux/quicklist.h:79: warning: type defaults to ‘int’ in
declaration of ‘type name’
include/linux/quicklist.h:79: error: invalid type argument of ‘unary *’
make[1]: *** [arch/i386/mm/pgtable.o] Error 1
make: *** [arch/i386/mm] Error 2
...

.config is from previous -rt4:
http://www.rncbc.org/datahub/config-2.6.22.1-rt4.0

Bye now.
-- 
rncbc aka Rui Nuno Capela
rncbc@rncbc.org

On Mon, 2007-07-23 at 22:29 +0100, Rui Nuno Capela wrote:

Ingo Molnar wrote:

quoted

we are pleased to announce the v2.6.22.1-rt5 kernel (collected and 
assembled by Thomas), which can be downloaded from the usual place:
  
  http://redhat.com/~mingo/realtime-preempt/
  
more info about the -rt patchset can be found in the RT wiki:
   
  http://rt.wiki.kernel.org

Changes since -rt4:

 - MM fix: PCP pages locking (Peter Zijlstra)

 - yield() fix: remove stale RCU-unlock (Daniel Walker)

 - dont do check_pgt_cache() in idle (found by Daniel Walker)

 - moved 8 patches, created 6 new patches out of larger patches to 
   make the -rt queue more bisectable (Daniel Walker)

 - cleanup (Daniel Walker)

 - fix the latency tracer under paravirt

to build a 2.6.22.1-rt5 tree, the following patches should be applied:
  
  http://kernel.org/pub/linux/kernel/v2.6/linux-2.6.22.1.tar.bz2
  http://redhat.com/~mingo/realtime-preempt/patch-2.6.22.1-rt5
  
	Ingo
-

Maybe I was too quick, but `make all` on is failing here:

...
In file included from arch/i386/mm/pgtable.c:16:
include/linux/quicklist.h: In function ‘quicklist_alloc’:
include/linux/quicklist.h:50: warning: unused variable ‘q’
include/linux/quicklist.h: In function ‘__quicklist_free’:
include/linux/quicklist.h:79: error: ‘per_cpu__quicklist’ undeclared
(first use in this function)
include/linux/quicklist.h:79: error: (Each undeclared identifier is
reported only once
include/linux/quicklist.h:79: error: for each function it appears in.)
include/linux/quicklist.h:79: warning: type defaults to ‘int’ in
declaration of ‘type name’
include/linux/quicklist.h:79: error: invalid type argument of ‘unary *’
make[1]: *** [arch/i386/mm/pgtable.o] Error 1
make: *** [arch/i386/mm] Error 2

Yeah, the quicklist patch wasn't fully tested .. The delta patch below
is what I had to change to get it working ..

Signed-Off-By: Daniel Walker <redacted>

diff -u linux-2.6/include/linux/quicklist.h linux-2.6.22.1/include/linux/quicklist.h

--- linux-2.6/include/linux/quicklist.h
+++ linux-2.6.22.1/include/linux/quicklist.h

@@ -47,7 +47,6 @@
 
 static inline void *quicklist_alloc(int nr, gfp_t flags, void (*ctor)(void *))
 {
-	struct quicklist *q;
 	void **p = NULL;
 	int cpu;
 

@@ -66,6 +65,7 @@
 static inline void __quicklist_free(int nr, void (*dtor)(void *), void *p,
 	struct page *page)
 {
+	int cpu;
 	struct quicklist *q;
 	int nid = page_to_nid(page);
 

@@ -76,11 +76,11 @@
 		return;
 	}
 
-	q = &get_cpu_var(quicklist)[nr];
+	q = &get_cpu_var_locked(quicklist, &cpu)[nr];
 	*(void **)p = q->page;
 	q->page = p;
 	q->nr_pages++;
-	put_cpu_var(quicklist);
+	put_cpu_var_locked(quicklist, cpu);
 }
 
 static inline void quicklist_free(int nr, void (*dtor)(void *), void *pp)

diff -u linux-2.6/mm/quicklist.c linux-2.6.22.1/mm/quicklist.c

--- linux-2.6/mm/quicklist.c
+++ linux-2.6.22.1/mm/quicklist.c

@@ -80,7 +80,7 @@
 	struct quicklist *ql, *q;
 
 	for_each_online_cpu(cpu) {
-		ql = per_cpu(quicklist, cpu);
+		ql = per_cpu_var_locked(quicklist, cpu);
 		for (q = ql; q < ql + CONFIG_NR_QUICK; q++)
 			count += q->nr_pages;
 	}

On Mon, 2007-07-23 at 14:30 -0700, Daniel Walker wrote:

Yeah, the quicklist patch wasn't fully tested .. The delta patch below
is what I had to change to get it working ..

My bad. I pushed the wrong queue to Ingo.

-ENOTENOUGHSLEEP

This is the delta. We'll do a -rt6 tomorrow morning.

	tglx

Index: linux-2.6.22/arch/i386/kernel/hpet.c
===================================================================

--- linux-2.6.22.orig/arch/i386/kernel/hpet.c	2007-07-24 00:02:12.000000000 +0200
+++ linux-2.6.22/arch/i386/kernel/hpet.c	2007-07-24 00:05:05.000000000 +0200

@@ -9,7 +9,6 @@
 #include <linux/pm.h>
 
 #include <asm/fixmap.h>
-#include <asm/i8253.h>
 #include <asm/hpet.h>
 #include <asm/i8253.h>
 #include <asm/io.h>

Index: linux-2.6.22/arch/i386/kernel/process.c
===================================================================

--- linux-2.6.22.orig/arch/i386/kernel/process.c	2007-07-24 00:02:10.000000000 +0200
+++ linux-2.6.22/arch/i386/kernel/process.c	2007-07-24 00:05:08.000000000 +0200

@@ -189,7 +189,6 @@ void cpu_idle(void)
 
 			tick_nohz_stop_sched_tick();
 
-			check_pgt_cache();
 			rmb();
 			idle = pm_idle;
 

Index: linux-2.6.22/include/asm-generic/bug.h
===================================================================

--- linux-2.6.22.orig/include/asm-generic/bug.h	2007-07-24 00:02:11.000000000 +0200
+++ linux-2.6.22/include/asm-generic/bug.h	2007-07-24 00:05:07.000000000 +0200

@@ -92,16 +92,4 @@ struct bug_entry {
 # define WARN_ON_NONRT(condition)	WARN_ON(condition)
 #endif
 
-#ifdef CONFIG_PREEMPT_RT
-# define BUG_ON_RT(c)			BUG_ON(c)
-# define BUG_ON_NONRT(c)		do { } while (0)
-# define WARN_ON_RT(condition)		WARN_ON(condition)
-# define WARN_ON_NONRT(condition)	do { } while (0)
-#else
-# define BUG_ON_RT(c)			do { } while (0)
-# define BUG_ON_NONRT(c)		BUG_ON(c)
-# define WARN_ON_RT(condition)		do { } while (0)
-# define WARN_ON_NONRT(condition)	WARN_ON(condition)
-#endif
-
 #endif

Index: linux-2.6.22/kernel/sched.c
===================================================================

--- linux-2.6.22.orig/kernel/sched.c	2007-07-24 00:02:12.000000000 +0200
+++ linux-2.6.22/kernel/sched.c	2007-07-24 00:05:36.000000000 +0200

@@ -4924,7 +4924,6 @@ asmlinkage long sys_sched_yield(void)
 	 * no need to preempt or enable interrupts:
 	 */
 	spin_unlock_no_resched(&rq->lock);
-	rcu_read_unlock();
 
 	__schedule();
 

Index: linux-2.6.22/kernel/softirq.c
===================================================================

--- linux-2.6.22.orig/kernel/softirq.c	2007-07-24 00:02:11.000000000 +0200
+++ linux-2.6.22/kernel/softirq.c	2007-07-24 00:05:08.000000000 +0200

@@ -102,7 +102,6 @@ static void wakeup_softirqd(int softirq)
 
 	if (unlikely(!tsk))
 		return;
-#if 1
 #if defined(CONFIG_PREEMPT_SOFTIRQS) && defined(CONFIG_PREEMPT_HARDIRQS)
 	/*
 	 * Optimization: if we are in a hardirq thread context, and

@@ -117,7 +116,6 @@ static void wakeup_softirqd(int softirq)
 			(tsk->normal_prio == current->normal_prio))
 		return;
 #endif
-#endif
 	/*
 	 * Wake up the softirq task:
 	 */

Index: linux-2.6.22/include/linux/quicklist.h
===================================================================

--- linux-2.6.22.orig/include/linux/quicklist.h	2007-07-24 00:02:00.000000000 +0200
+++ linux-2.6.22/include/linux/quicklist.h	2007-07-24 00:05:08.000000000 +0200

@@ -18,7 +18,7 @@ struct quicklist {
 	int nr_pages;
 };
 
-DECLARE_PER_CPU(struct quicklist, quicklist)[CONFIG_NR_QUICK];
+DECLARE_PER_CPU_LOCKED(struct quicklist, quicklist)[CONFIG_NR_QUICK];
 
 /*
  * The two key functions quicklist_alloc and quicklist_free are inline so

@@ -30,19 +30,30 @@ DECLARE_PER_CPU(struct quicklist, quickl
  * The fast patch in quicklist_alloc touched only a per cpu cacheline and
  * the first cacheline of the page itself. There is minmal overhead involved.
  */
-static inline void *quicklist_alloc(int nr, gfp_t flags, void (*ctor)(void *))
+static inline void *__quicklist_alloc(int cpu, int nr, gfp_t flags, void (*ctor)(void *))
 {
 	struct quicklist *q;
 	void **p = NULL;
 
-	q =&get_cpu_var(quicklist)[nr];
+	q = &__get_cpu_var_locked(quicklist, cpu)[nr];
 	p = q->page;
 	if (likely(p)) {
 		q->page = p[0];
 		p[0] = NULL;
 		q->nr_pages--;
 	}
-	put_cpu_var(quicklist);
+	return p;
+}
+
+static inline void *quicklist_alloc(int nr, gfp_t flags, void (*ctor)(void *))
+{
+	struct quicklist *q;
+	void **p = NULL;
+	int cpu;
+
+	(void)get_cpu_var_locked(quicklist, &cpu)[nr];
+	p = __quicklist_alloc(cpu, nr, flags, ctor);
+	put_cpu_var_locked(quicklist, cpu);
 	if (likely(p))
 		return p;
 

Index: linux-2.6.22/mm/quicklist.c
===================================================================

--- linux-2.6.22.orig/mm/quicklist.c	2007-07-24 00:02:00.000000000 +0200
+++ linux-2.6.22/mm/quicklist.c	2007-07-24 00:05:08.000000000 +0200

@@ -19,7 +19,7 @@
 #include <linux/module.h>
 #include <linux/quicklist.h>
 
-DEFINE_PER_CPU(struct quicklist, quicklist)[CONFIG_NR_QUICK];
+DEFINE_PER_CPU_LOCKED(struct quicklist, quicklist)[CONFIG_NR_QUICK];
 
 #define FRACTION_OF_NODE_MEM	16
 

@@ -51,8 +51,9 @@ void quicklist_trim(int nr, void (*dtor)
 {
 	long pages_to_free;
 	struct quicklist *q;
+	int cpu;
 
-	q = &get_cpu_var(quicklist)[nr];
+	q = &get_cpu_var_locked(quicklist, &cpu)[nr];
 	if (q->nr_pages > min_pages) {
 		pages_to_free = min_pages_to_free(q, min_pages, max_free);
 

@@ -61,7 +62,7 @@ void quicklist_trim(int nr, void (*dtor)
 			 * We pass a gfp_t of 0 to quicklist_alloc here
 			 * because we will never call into the page allocator.
 			 */
-			void *p = quicklist_alloc(nr, 0, NULL);
+			void *p = __quicklist_alloc(cpu, nr, 0, NULL);
 
 			if (dtor)
 				dtor(p);

@@ -69,7 +70,7 @@ void quicklist_trim(int nr, void (*dtor)
 			pages_to_free--;
 		}
 	}
-	put_cpu_var(quicklist);
+	put_cpu_var_locked(quicklist, cpu);
 }
 
 unsigned long quicklist_total_size(void)

On Tue, 2007-07-24 at 00:05 +0200, Thomas Gleixner wrote:

On Mon, 2007-07-23 at 14:30 -0700, Daniel Walker wrote:

quoted

Yeah, the quicklist patch wasn't fully tested .. The delta patch below
is what I had to change to get it working ..

My bad. I pushed the wrong queue to Ingo.

-ENOTENOUGHSLEEP

This is the delta. We'll do a -rt6 tomorrow morning.

-EWAYNOTENOUGHSLEEP

I sent out the wrong delta. /me goes to sleep and fixes tomorrow.

	tglx

* Rui Nuno Capela [off-list ref] wrote:

Maybe I was too quick, but `make all` on is failing here:

does -rt6 work better?

	Ingo

On Tue, 2007-07-24 at 09:39 +0200, Ingo Molnar wrote:

* Rui Nuno Capela [off-list ref] wrote:

quoted

Maybe I was too quick, but `make all` on is failing here:

does -rt6 work better?

Hmmm, -rt6 seems to be gone... was about to download it and it
dissapeared. 

-- Fernando

On Tue, 2007-07-24 at 12:34 -0700, Fernando Lopez-Lezcano wrote:

On Tue, 2007-07-24 at 09:39 +0200, Ingo Molnar wrote:

quoted

* Rui Nuno Capela [off-list ref] wrote:

quoted

Maybe I was too quick, but `make all` on is failing here:

does -rt6 work better?

Hmmm, -rt6 seems to be gone... was about to download it and it
dissapeared. 

Never mind, I see it migrated back to the main page. 
-- Fernando

* Fernando Lopez-Lezcano [off-list ref] wrote:

On Tue, 2007-07-24 at 12:34 -0700, Fernando Lopez-Lezcano wrote:

quoted

On Tue, 2007-07-24 at 09:39 +0200, Ingo Molnar wrote:

quoted

* Rui Nuno Capela [off-list ref] wrote:

quoted

Maybe I was too quick, but `make all` on is failing here:

does -rt6 work better?

Hmmm, -rt6 seems to be gone... was about to download it and it
dissapeared. 

Never mind, I see it migrated back to the main page. 

apparently you caught that 3 seconds window where the .23-rc1-rt1 
release script moved old patches into the older/ directory :-)

	Ingo

On Tue, 2007-07-24 at 22:05 +0200, Ingo Molnar wrote:

* Fernando Lopez-Lezcano [off-list ref] wrote:

quoted

On Tue, 2007-07-24 at 12:34 -0700, Fernando Lopez-Lezcano wrote:

quoted

On Tue, 2007-07-24 at 09:39 +0200, Ingo Molnar wrote:

quoted

* Rui Nuno Capela [off-list ref] wrote:

quoted

Maybe I was too quick, but `make all` on is failing here:

does -rt6 work better?

Hmmm, -rt6 seems to be gone... was about to download it and it
dissapeared. 

Never mind, I see it migrated back to the main page. 

apparently you caught that 3 seconds window where the .23-rc1-rt1 
release script moved old patches into the older/ directory :-)

Yup, good timing... :-) Hard to do again...
(BTW, will you keep 2.6.22.x patches going for a while?)
-- Fernando

* Fernando Lopez-Lezcano [off-list ref] wrote:

quoted

apparently you caught that 3 seconds window where the .23-rc1-rt1 
release script moved old patches into the older/ directory :-)

Yup, good timing... :-) Hard to do again...
(BTW, will you keep 2.6.22.x patches going for a while?)

yeah, that's the plan: to keep .22-rt updated until .23 is released. 
(Thomas agrees with that approach too)

	Ingo

On Tue, 2007-07-24 at 22:34 +0200, Ingo Molnar wrote:

* Fernando Lopez-Lezcano [off-list ref] wrote:

quoted

quoted

apparently you caught that 3 seconds window where the .23-rc1-rt1 
release script moved old patches into the older/ directory :-)

Yup, good timing... :-) Hard to do again...
(BTW, will you keep 2.6.22.x patches going for a while?)

yeah, that's the plan: to keep .22-rt updated until .23 is released. 
(Thomas agrees with that approach too)

Thank you thank you to all involved! That's very good news...
-- Fernando

Hi,

On Tue, 2007-07-24 at 22:34 +0200, Ingo Molnar wrote:

quoted

* Fernando Lopez-Lezcano [off-list ref] wrote:

[...] 

quoted

(BTW, will you keep 2.6.22.x patches going for a while?)

yeah, that's the plan: to keep .22-rt updated until .23 is released. 
(Thomas agrees with that approach too)

When .23 will be released could you explain changes for every following
-rt release in more detailed fashion, please? For whom is interested to
keep .22 updated also when .23 is released.

Thanks.

Ciao,
Alessio