Re: [ANNOUNCE] 3.0-rc7-rt0

Re: [ANNOUNCE] 3.0-rc7-rt0

[hermann]

> Dear RT Folks,
> 
> I'm pleased to announce the first drop of the 3.0-rc7 based RT
> patch.

Hi 
Many thanks for all this work. 

I try to build the 3.0-rc7-rt0 but fail with the following message:
 CC [M]  fs/ext3/balloc.o
In file included from include/linux/jbd.h:247:0,
                 from fs/ext3/balloc.c:18:
include/linux/jbd_common.h: In function ‘jbd_lock_bh_state’:
include/linux/jbd_common.h:43:2: error: ‘struct buffer_head’ has no member named ‘b_state_lock’
include/linux/jbd_common.h: In function ‘jbd_trylock_bh_state’:
include/linux/jbd_common.h:52:9: error: ‘struct buffer_head’ has no member named ‘b_state_lock’
include/linux/jbd_common.h: In function ‘jbd_is_locked_bh_state’:
include/linux/jbd_common.h:61:27: error: ‘struct buffer_head’ has no member named ‘b_state_lock’
include/linux/jbd_common.h: In function ‘jbd_unlock_bh_state’:
include/linux/jbd_common.h:70:2: error: ‘struct buffer_head’ has no member named ‘b_state_lock’
include/linux/jbd_common.h: In function ‘jbd_lock_bh_journal_head’:
include/linux/jbd_common.h:79:2: error: ‘struct buffer_head’ has no member named ‘b_journal_head_lock’
include/linux/jbd_common.h: In function ‘jbd_unlock_bh_journal_head’:
include/linux/jbd_common.h:88:2: error: ‘struct buffer_head’ has no member named ‘b_journal_head_lock’
make[3]: *** [fs/ext3/balloc.o] Fehler 1
make[2]: *** [fs/ext3] Fehler 2
make[1]: *** [fs] Fehler 2
make[1]: Leaving directory `/home/brummer/Projekte/Kernel/linux-3.0-rc7'
make: *** [debian/stamp/build/kernel] Fehler 2


my system is:
System:    Host box Kernel 2.6.39-rc7 i686 (32 bit gcc 4.5.2) 
           Desktop Xfce 4.8.2 (Gtk 2.24.4) Distro Debian GNU/Linux wheezy/sid
Machine:   System Hewlett-Packard product hp workstation version x2100/2600
           Mobo Hewlett-Packard model HP WMTA System Board version A03 Bios Phoenix version JG.W1.04US date 07/26/2002
CPU:       Single core Intel Pentium 4 CPU (-UP-) cache 512 KB flags (sse sse2) bmips 5183.59 clocked at 2591.796 MHz 
Graphics:  Card: ATI Radeon R200 QH [Radeon 8500] bus-ID: 01:00.0 X.Org 1.10.1 drivers ati,radeon Resolution 1024x768@75.0hz 
           GLX Renderer Mesa X11 GLX Version 2.1 Mesa 7.10.2 Direct Rendering Yes
Audio:     Card Creative Labs SB Live! EMU10k1 driver EMU10K1_Audigy port 5040 bus-ID: 02:0a.0 Sound: ALSA v: 1.0.24
Network:   Card Intel 82801BA/BAM/CA/CAM Ethernet Controller driver e100 v: 3.5.24-k2-NAPI port 5000 bus-ID: 02:08.0
           IF: eth0 state: up speed: 100 Mbps duplex: full mac: 00:30:6e:26:c4:6d
Drives:    HDD Total Size: 58.1GB (57.4% used) 1: /dev/sda IC35L060AVV207 41.2GB 
           2: /dev/sdb IBM 16.9GB 
           Optical: /dev/sr0 model: TOSHIBA DVD-ROM SD-M1612 rev: 1004 dev-links: cdrom,dvd,scd0
           Features: speed: 48x multisession: yes audio: yes dvd: yes rw: none state: running
Partition: ID:/ size: 38G used: 27G (76%) fs: ext3 dev: /dev/sda2 label: N/A uuid: 1c8cbe6a-e15f-4a86-9b2b-2c75628ed183 
           ID:/media/sdb1 size: 15G used: 4.6G (33%) fs: ext4 dev: /dev/sdb1 label: N/A uuid: b6d0dca6-d88a-4096-bd54-c20dc98db436 
           ID:swap-1 size: 0.74GB used: 0.00GB (0%) fs: swap dev: /dev/sda1 label: N/A uuid: N/A 
Unmounted: ID: /dev/sdb5 size: 0.75G label: N/A uuid: 6b246475-8d0a-4414-9657-4586b3a57ea1 
Sensors:   Error: You do not have the sensors app installed.
Info:      Processes 113 Uptime 13:07 Memory 449.0/2026.4MB Runlevel 5 Client Shell

regards
hermann

--
To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [ANNOUNCE] 3.0-rc7-rt0

[Thomas Gleixner]

On Thu, 21 Jul 2011, hermann wrote:
> I try to build the 3.0-rc7-rt0 but fail with the following message:
>  CC [M]  fs/ext3/balloc.o
> In file included from include/linux/jbd.h:247:0,
>                  from fs/ext3/balloc.c:18:

Find below a patch on top of rt0 which fixes this build failure and
some other stuff.

Thanks,

	tglx

-----------

Index: linux-2.6/kernel/timer.c
===================================================================
--- linux-2.6.orig/kernel/timer.c
+++ linux-2.6/kernel/timer.c
@@ -654,6 +654,36 @@ static struct tvec_base *lock_timer_base
 	}
 }
 
+#ifndef CONFIG_PREEMPT_RT_FULL
+static inline struct tvec_base *switch_timer_base(struct timer_list *timer,
+						  struct tvec_base *old,
+						  struct tvec_base *new)
+{
+	/* See the comment in lock_timer_base() */
+	timer_set_base(timer, NULL);
+	spin_unlock(&old->lock);
+	spin_lock(&new->lock);
+	timer_set_base(timer, new);
+	return new;
+}
+#else
+static inline struct tvec_base *switch_timer_base(struct timer_list *timer,
+						  struct tvec_base *old,
+						  struct tvec_base *new)
+{
+	/*
+	 * We cannot do the above because we might be preempted and
+	 * then the preempter would see NULL and loop forever.
+	 */
+	if (spin_trylock(&new->lock)) {
+		timer_set_base(timer, new);
+		spin_unlock(&old->lock);
+		return new;
+	}
+	return old;
+}
+#endif
+
 static inline int
 __mod_timer(struct timer_list *timer, unsigned long expires,
 						bool pending_only, int pinned)
@@ -699,14 +729,8 @@ __mod_timer(struct timer_list *timer, un
 		 * handler yet has not finished. This also guarantees that
 		 * the timer is serialized wrt itself.
 		 */
-		if (likely(base->running_timer != timer)) {
-			/* See the comment in lock_timer_base() */
-			timer_set_base(timer, NULL);
-			spin_unlock(&base->lock);
-			base = new_base;
-			spin_lock(&base->lock);
-			timer_set_base(timer, base);
-		}
+		if (likely(base->running_timer != timer))
+			base = switch_timer_base(timer, base, new_base);
 	}
 
 	timer->expires = expires;
Index: linux-2.6/mm/page_alloc.c
===================================================================
--- linux-2.6.orig/mm/page_alloc.c
+++ linux-2.6/mm/page_alloc.c
@@ -1943,8 +1943,8 @@ __alloc_pages_direct_compact(gfp_t gfp_m
 	if (*did_some_progress != COMPACT_SKIPPED) {
 
 		/* Page migration frees to the PCP lists but we want merging */
-		drain_pages(get_cpu());
-		put_cpu();
+		drain_pages(get_cpu_light());
+		put_cpu_light();
 
 		page = get_page_from_freelist(gfp_mask, nodemask,
 				order, zonelist, high_zoneidx,
Index: linux-2.6/include/linux/buffer_head.h
===================================================================
--- linux-2.6.orig/include/linux/buffer_head.h
+++ linux-2.6/include/linux/buffer_head.h
@@ -74,7 +74,9 @@ struct buffer_head {
 	atomic_t b_count;		/* users using this buffer_head */
 #ifdef CONFIG_PREEMPT_RT_BASE
 	spinlock_t b_uptodate_lock;
-#if defined(CONFIG_JBD) || defined(CONFIG_JBD2)
+#if defined(CONFIG_JBD) || defined(CONFIG_JBD_MODULE) || \
+    defined(CONFIG_JBD2) || defined(CONFIG_JBD2_MODULE)
+	spinlock_t b_state_lock;
 	spinlock_t b_state_lock;
 	spinlock_t b_journal_head_lock;
 #endif

Re: [ANNOUNCE] 3.0-rc7-rt0

[Thomas Gleixner]

On Thu, 21 Jul 2011, Thomas Gleixner wrote:
> On Thu, 21 Jul 2011, hermann wrote:
> > I try to build the 3.0-rc7-rt0 but fail with the following message:
> >  CC [M]  fs/ext3/balloc.o
> > In file included from include/linux/jbd.h:247:0,
> >                  from fs/ext3/balloc.c:18:
> 
> Find below a patch on top of rt0 which fixes this build failure and
> some other stuff.

Bah, that's when you trust other peoples patches. It needs the
following fix as well:

--- linux-2.6.orig/include/linux/buffer_head.h
+++ linux-2.6/include/linux/buffer_head.h
@@ -109,7 +111,8 @@ static inline void buffer_head_init_lock
 {
 #ifdef CONFIG_PREEMPT_RT_BASE
 	spin_lock_init(&bh->b_uptodate_lock);
-#if defined(CONFIG_JBD) || defined(CONFIG_JBD2)
+#if defined(CONFIG_JBD) || defined(CONFIG_JBD_MODULE) || \
+    defined(CONFIG_JBD2) || defined(CONFIG_JBD2_MODULE)
 	spin_lock_init(&bh->b_state_lock);
 	spin_lock_init(&bh->b_journal_head_lock);
 #endif

Re: [ANNOUNCE] 3.0-rc7-rt0

[hermann]

Am Donnerstag, den 21.07.2011, 19:43 +0200 schrieb Thomas Gleixner:
> On Thu, 21 Jul 2011, Thomas Gleixner wrote:
> > On Thu, 21 Jul 2011, hermann wrote:
> > > I try to build the 3.0-rc7-rt0 but fail with the following message:
> > >  CC [M]  fs/ext3/balloc.o
> > > In file included from include/linux/jbd.h:247:0,
> > >                  from fs/ext3/balloc.c:18:
> > 
> > Find below a patch on top of rt0 which fixes this build failure and
> > some other stuff.

Many Thanks for the quick response, when I try to apply the patch's on a
fresh patched 3.0-rc7-rt0, I receive some Hunks :

 patch -p1<../patch-3.0-rc7-rt0.1.patch
patching file kernel/timer.c
Hunk #1 succeeded at 654 with fuzz 1.
Hunk #2 FAILED at 729.
1 out of 2 hunks FAILED -- saving rejects to file kernel/timer.c.rej
patching file mm/page_alloc.c
Hunk #1 FAILED at 1943.
1 out of 1 hunk FAILED -- saving rejects to file mm/page_alloc.c.rej
patching file include/linux/buffer_head.h
Hunk #1 FAILED at 74.
1 out of 1 hunk FAILED -- saving rejects to file
include/linux/buffer_head.h.rej
patching file include/linux/buffer_head.h
Hunk #1 FAILED at 109.
1 out of 1 hunk FAILED -- saving rejects to file
include/linux/buffer_head.h.rej

So I apply them per hand, (well some offset here and there). 
Build is running now, will report back if all went well for me now. 

kind regards
hermann

Re: [ANNOUNCE] 3.0-rt1

[hermann]

Hi

Build fail here with

  CC      arch/x86/kernel/early_printk.o
arch/x86/kernel/early_printk.c:188:17: error: conflicting types for
‘early_vprintk’
include/linux/printk.h:95:13: note: previous declaration of
‘early_vprintk’ was here
make[3]: *** [arch/x86/kernel/early_printk.o] Fehler 1
make[2]: *** [arch/x86/kernel] Fehler 2
make[1]: *** [arch/x86] Fehler 2

it build here when I declare early_vprintk in printk.h as 
asmlinkage void early_vprintk(const char *fmt, va_list args);

System:    Host box Kernel 3.0.0-rc7-rt0 i686 (32 bit gcc 4.5.2) 
           Desktop Xfce 4.8.2 (Gtk 2.24.4) Distro Debian GNU/Linux
wheezy/sid
Machine:   System Hewlett-Packard product hp workstation version
x2100/2600
           Mobo Hewlett-Packard model HP WMTA System Board version A03
Bios Phoenix version JG.W1.04US date 07/26/2002
CPU:       Single core Intel Pentium 4 CPU (-UP-) cache 512 KB flags
(sse sse2) bmips 5184.7 clocked at 2600.00 MHz 

regards
hermann

--
To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [ANNOUNCE] 3.0-rt1

[Thomas Gleixner]

[-- Attachment #1: Type: TEXT/PLAIN, Size: 629 bytes --]

On Sat, 23 Jul 2011, hermann wrote:

> Hi
> 
> Build fail here with
> 
>   CC      arch/x86/kernel/early_printk.o
> arch/x86/kernel/early_printk.c:188:17: error: conflicting types for
> ‘early_vprintk’
> include/linux/printk.h:95:13: note: previous declaration of
> ‘early_vprintk’ was here
> make[3]: *** [arch/x86/kernel/early_printk.o] Fehler 1
> make[2]: *** [arch/x86/kernel] Fehler 2
> make[1]: *** [arch/x86] Fehler 2
> 
> it build here when I declare early_vprintk in printk.h as 
> asmlinkage void early_vprintk(const char *fmt, va_list args);

Grr, right. That's a 32bit thingy. Fixed for -rt2.

Thanks,

	tglx

Re: [ANNOUNCE] 3.0-rc7-rt0

[andi]

Hi!

Actually, I think, the last line added in this patch is wrong...

+	spinlock_t b_state_lock;

leads to duplication of b_state_lock...

regards,
Andi


On 07/21/2011 07:35 PM, Thomas Gleixner wrote:
> On Thu, 21 Jul 2011, hermann wrote:
>    
>> I try to build the 3.0-rc7-rt0 but fail with the following message:
>>   CC [M]  fs/ext3/balloc.o
>> In file included from include/linux/jbd.h:247:0,
>>                   from fs/ext3/balloc.c:18:
>>      
> Find below a patch on top of rt0 which fixes this build failure and
> some other stuff.
>
> Thanks,
>
> 	tglx
>
> -----------
>
> Index: linux-2.6/kernel/timer.c
> ===================================================================
> --- linux-2.6.orig/kernel/timer.c
> +++ linux-2.6/kernel/timer.c
> @@ -654,6 +654,36 @@ static struct tvec_base *lock_timer_base
>   	}
>   }
>
> +#ifndef CONFIG_PREEMPT_RT_FULL
> +static inline struct tvec_base *switch_timer_base(struct timer_list *timer,
> +						  struct tvec_base *old,
> +						  struct tvec_base *new)
> +{
> +	/* See the comment in lock_timer_base() */
> +	timer_set_base(timer, NULL);
> +	spin_unlock(&old->lock);
> +	spin_lock(&new->lock);
> +	timer_set_base(timer, new);
> +	return new;
> +}
> +#else
> +static inline struct tvec_base *switch_timer_base(struct timer_list *timer,
> +						  struct tvec_base *old,
> +						  struct tvec_base *new)
> +{
> +	/*
> +	 * We cannot do the above because we might be preempted and
> +	 * then the preempter would see NULL and loop forever.
> +	 */
> +	if (spin_trylock(&new->lock)) {
> +		timer_set_base(timer, new);
> +		spin_unlock(&old->lock);
> +		return new;
> +	}
> +	return old;
> +}
> +#endif
> +
>   static inline int
>   __mod_timer(struct timer_list *timer, unsigned long expires,
>   						bool pending_only, int pinned)
> @@ -699,14 +729,8 @@ __mod_timer(struct timer_list *timer, un
>   		 * handler yet has not finished. This also guarantees that
>   		 * the timer is serialized wrt itself.
>   		 */
> -		if (likely(base->running_timer != timer)) {
> -			/* See the comment in lock_timer_base() */
> -			timer_set_base(timer, NULL);
> -			spin_unlock(&base->lock);
> -			base = new_base;
> -			spin_lock(&base->lock);
> -			timer_set_base(timer, base);
> -		}
> +		if (likely(base->running_timer != timer))
> +			base = switch_timer_base(timer, base, new_base);
>   	}
>
>   	timer->expires = expires;
> Index: linux-2.6/mm/page_alloc.c
> ===================================================================
> --- linux-2.6.orig/mm/page_alloc.c
> +++ linux-2.6/mm/page_alloc.c
> @@ -1943,8 +1943,8 @@ __alloc_pages_direct_compact(gfp_t gfp_m
>   	if (*did_some_progress != COMPACT_SKIPPED) {
>
>   		/* Page migration frees to the PCP lists but we want merging */
> -		drain_pages(get_cpu());
> -		put_cpu();
> +		drain_pages(get_cpu_light());
> +		put_cpu_light();
>
>   		page = get_page_from_freelist(gfp_mask, nodemask,
>   				order, zonelist, high_zoneidx,
> Index: linux-2.6/include/linux/buffer_head.h
> ===================================================================
> --- linux-2.6.orig/include/linux/buffer_head.h
> +++ linux-2.6/include/linux/buffer_head.h
> @@ -74,7 +74,9 @@ struct buffer_head {
>   	atomic_t b_count;		/* users using this buffer_head */
>   #ifdef CONFIG_PREEMPT_RT_BASE
>   	spinlock_t b_uptodate_lock;
> -#if defined(CONFIG_JBD) || defined(CONFIG_JBD2)
> +#if defined(CONFIG_JBD) || defined(CONFIG_JBD_MODULE) || \
> +    defined(CONFIG_JBD2) || defined(CONFIG_JBD2_MODULE)
> +	spinlock_t b_state_lock;
>   	spinlock_t b_state_lock;
>   	spinlock_t b_journal_head_lock;
>   #endif
> --
> To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>    

Re: [ANNOUNCE] 3.0-rc7-rt0

[hermann]

Am Donnerstag, den 21.07.2011, 20:28 +0200 schrieb andi:
> Hi!
> 
> Actually, I think, the last line added in this patch is wrong...
> 
> +	spinlock_t b_state_lock;
> 
> leads to duplication of b_state_lock...
> 
> regards,
> Andi
> 

Yea, I haven't apply this line here to avoid that.

regards
hermann

> 
> On 07/21/2011 07:35 PM, Thomas Gleixner wrote:
> > On Thu, 21 Jul 2011, hermann wrote:
> >    
> >> I try to build the 3.0-rc7-rt0 but fail with the following message:
> >>   CC [M]  fs/ext3/balloc.o
> >> In file included from include/linux/jbd.h:247:0,
> >>                   from fs/ext3/balloc.c:18:
> >>      
> > Find below a patch on top of rt0 which fixes this build failure and
> > some other stuff.
> >
> > Thanks,
> >
> > 	tglx
> >
> > -----------
> >
> > Index: linux-2.6/kernel/timer.c
> > ===================================================================
> > --- linux-2.6.orig/kernel/timer.c
> > +++ linux-2.6/kernel/timer.c
> > @@ -654,6 +654,36 @@ static struct tvec_base *lock_timer_base
> >   	}
> >   }
> >
> > +#ifndef CONFIG_PREEMPT_RT_FULL
> > +static inline struct tvec_base *switch_timer_base(struct timer_list *timer,
> > +						  struct tvec_base *old,
> > +						  struct tvec_base *new)
> > +{
> > +	/* See the comment in lock_timer_base() */
> > +	timer_set_base(timer, NULL);
> > +	spin_unlock(&old->lock);
> > +	spin_lock(&new->lock);
> > +	timer_set_base(timer, new);
> > +	return new;
> > +}
> > +#else
> > +static inline struct tvec_base *switch_timer_base(struct timer_list *timer,
> > +						  struct tvec_base *old,
> > +						  struct tvec_base *new)
> > +{
> > +	/*
> > +	 * We cannot do the above because we might be preempted and
> > +	 * then the preempter would see NULL and loop forever.
> > +	 */
> > +	if (spin_trylock(&new->lock)) {
> > +		timer_set_base(timer, new);
> > +		spin_unlock(&old->lock);
> > +		return new;
> > +	}
> > +	return old;
> > +}
> > +#endif
> > +
> >   static inline int
> >   __mod_timer(struct timer_list *timer, unsigned long expires,
> >   						bool pending_only, int pinned)
> > @@ -699,14 +729,8 @@ __mod_timer(struct timer_list *timer, un
> >   		 * handler yet has not finished. This also guarantees that
> >   		 * the timer is serialized wrt itself.
> >   		 */
> > -		if (likely(base->running_timer != timer)) {
> > -			/* See the comment in lock_timer_base() */
> > -			timer_set_base(timer, NULL);
> > -			spin_unlock(&base->lock);
> > -			base = new_base;
> > -			spin_lock(&base->lock);
> > -			timer_set_base(timer, base);
> > -		}
> > +		if (likely(base->running_timer != timer))
> > +			base = switch_timer_base(timer, base, new_base);
> >   	}
> >
> >   	timer->expires = expires;
> > Index: linux-2.6/mm/page_alloc.c
> > ===================================================================
> > --- linux-2.6.orig/mm/page_alloc.c
> > +++ linux-2.6/mm/page_alloc.c
> > @@ -1943,8 +1943,8 @@ __alloc_pages_direct_compact(gfp_t gfp_m
> >   	if (*did_some_progress != COMPACT_SKIPPED) {
> >
> >   		/* Page migration frees to the PCP lists but we want merging */
> > -		drain_pages(get_cpu());
> > -		put_cpu();
> > +		drain_pages(get_cpu_light());
> > +		put_cpu_light();
> >
> >   		page = get_page_from_freelist(gfp_mask, nodemask,
> >   				order, zonelist, high_zoneidx,
> > Index: linux-2.6/include/linux/buffer_head.h
> > ===================================================================
> > --- linux-2.6.orig/include/linux/buffer_head.h
> > +++ linux-2.6/include/linux/buffer_head.h
> > @@ -74,7 +74,9 @@ struct buffer_head {
> >   	atomic_t b_count;		/* users using this buffer_head */
> >   #ifdef CONFIG_PREEMPT_RT_BASE
> >   	spinlock_t b_uptodate_lock;
> > -#if defined(CONFIG_JBD) || defined(CONFIG_JBD2)
> > +#if defined(CONFIG_JBD) || defined(CONFIG_JBD_MODULE) || \
> > +    defined(CONFIG_JBD2) || defined(CONFIG_JBD2_MODULE)
> > +	spinlock_t b_state_lock;
> >   	spinlock_t b_state_lock;
> >   	spinlock_t b_journal_head_lock;
> >   #endif
> > --
> > To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in
> > the body of a message to majordomo@vger.kernel.org
> > More majordomo info at  http://vger.kernel.org/majordomo-info.html
> >    
> 

[PATCH] Various fixups from first bug reports

[Uwe Kleine-König]

From: Thomas Gleixner <tglx@linutronix.de>

---
Hello,

this is the patch Thomas sent earlier in this thread with the two fixups
folded in that occured in reply to Thomas' patch.

Best regards
Uwe

 include/linux/buffer_head.h |    6 ++++--
 kernel/timer.c              |   40 ++++++++++++++++++++++++++++++++--------
 mm/page_alloc.c             |    4 ++--
 3 files changed, 38 insertions(+), 12 deletions(-)

diff --git a/include/linux/buffer_head.h b/include/linux/buffer_head.h
index 7283b77..dbaeaec 100644
--- a/include/linux/buffer_head.h
+++ b/include/linux/buffer_head.h
@@ -74,7 +74,8 @@ struct buffer_head {
 	atomic_t b_count;		/* users using this buffer_head */
 #ifdef CONFIG_PREEMPT_RT_BASE
 	spinlock_t b_uptodate_lock;
-#if defined(CONFIG_JBD) || defined(CONFIG_JBD2)
+#if defined(CONFIG_JBD) || defined(CONFIG_JBD_MODULE) || \
+    defined(CONFIG_JBD2) || defined(CONFIG_JBD2_MODULE)
 	spinlock_t b_state_lock;
 	spinlock_t b_journal_head_lock;
 #endif
@@ -109,7 +110,8 @@ static inline void buffer_head_init_locks(struct buffer_head *bh)
 {
 #ifdef CONFIG_PREEMPT_RT_BASE
 	spin_lock_init(&bh->b_uptodate_lock);
-#if defined(CONFIG_JBD) || defined(CONFIG_JBD2)
+#if defined(CONFIG_JBD) || defined(CONFIG_JBD_MODULE) || \
+    defined(CONFIG_JBD2) || defined(CONFIG_JBD2_MODULE)
 	spin_lock_init(&bh->b_state_lock);
 	spin_lock_init(&bh->b_journal_head_lock);
 #endif
diff --git a/kernel/timer.c b/kernel/timer.c
index 20b548f..937799f 100644
--- a/kernel/timer.c
+++ b/kernel/timer.c
@@ -654,6 +654,36 @@ static struct tvec_base *lock_timer_base(struct timer_list *timer,
 	}
 }
 
+#ifndef CONFIG_PREEMPT_RT_FULL
+static inline struct tvec_base *switch_timer_base(struct timer_list *timer,
+						  struct tvec_base *old,
+						  struct tvec_base *new)
+{
+	/* See the comment in lock_timer_base() */
+	timer_set_base(timer, NULL);
+	spin_unlock(&old->lock);
+	spin_lock(&new->lock);
+	timer_set_base(timer, new);
+	return new;
+}
+#else
+static inline struct tvec_base *switch_timer_base(struct timer_list *timer,
+						  struct tvec_base *old,
+						  struct tvec_base *new)
+{
+	/*
+	 * We cannot do the above because we might be preempted and
+	 * then the preempter would see NULL and loop forever.
+	 */
+	if (spin_trylock(&new->lock)) {
+		timer_set_base(timer, new);
+		spin_unlock(&old->lock);
+		return new;
+	}
+	return old;
+}
+#endif
+
 static inline int
 __mod_timer(struct timer_list *timer, unsigned long expires,
 						bool pending_only, int pinned)
@@ -699,14 +729,8 @@ __mod_timer(struct timer_list *timer, unsigned long expires,
 		 * handler yet has not finished. This also guarantees that
 		 * the timer is serialized wrt itself.
 		 */
-		if (likely(base->running_timer != timer)) {
-			/* See the comment in lock_timer_base() */
-			timer_set_base(timer, NULL);
-			spin_unlock(&base->lock);
-			base = new_base;
-			spin_lock(&base->lock);
-			timer_set_base(timer, base);
-		}
+		if (likely(base->running_timer != timer))
+			base = switch_timer_base(timer, base, new_base);
 	}
 
 	timer->expires = expires;
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 57d70b8..33fd80f 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -1943,8 +1943,8 @@ __alloc_pages_direct_compact(gfp_t gfp_mask, unsigned int order,
 	if (*did_some_progress != COMPACT_SKIPPED) {
 
 		/* Page migration frees to the PCP lists but we want merging */
-		drain_pages(get_cpu());
-		put_cpu();
+		drain_pages(get_cpu_light());
+		put_cpu_light();
 
 		page = get_page_from_freelist(gfp_mask, nodemask,
 				order, zonelist, high_zoneidx,
-- 
1.7.5.4

Re: [PATCH] Various fixups from first bug reports

[hermann]

Am Donnerstag, den 21.07.2011, 21:57 +0200 schrieb Uwe Kleine-König:
> From: Thomas Gleixner <tglx@linutronix.de>
> 
> ---
> Hello,
> 
> this is the patch Thomas sent earlier in this thread with the two fixups
> folded in that occured in reply to Thomas' patch.
> 
> Best regards
> Uwe
> 

Hi

well, I run into 2 ops, witch, I guess not related to the rt patch.

first one was with CONFIG_SENSORS_LTC4245, module build stage 2 failed
with undefined reference to of_find_property

second was with CONFIG_LGUEST_GUEST enabled, module build stage 2 failed
with Documentation/lguest not found

Anyhow, I'm now writing from 
 System:    Host box Kernel 3.0.0-rc7-rt0 i686 (32 bit gcc 4.5.2) 
           Desktop Xfce 4.8.2 (Gtk 2.24.4) Distro Debian GNU/Linux
wheezy/sid

many thanks for that, great work, 

regards
hermann


--
To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH] Various fixups from first bug reports

[Uwe Kleine-König]

Hello,

(These two issues occured on 3.0-rc7-rt0, I think they apply to
mainline, too.)

On Fri, Jul 22, 2011 at 07:43:34AM +0200, hermann wrote:
> well, I run into 2 ops, witch, I guess not related to the rt patch.
> 
> first one was with CONFIG_SENSORS_LTC4245, module build stage 2 failed
> with undefined reference to of_find_property
of_find_property isn't exported to modules. Maybe of_get_property should
be used? Should of_find_property be static then? Alternatively export
of_find_property, too. Grant?

> second was with CONFIG_LGUEST_GUEST enabled, module build stage 2 failed
> with Documentation/lguest not found
Hmm, there is Documentation/virtual/lguest. Didn't find where
Documentation/lguest is used from a quick look. Rusty?

Best regards
Uwe

-- 
Pengutronix e.K.                           | Uwe Kleine-König            |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
--
To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH] Various fixups from first bug reports

[hermann]

Am Freitag, den 22.07.2011, 08:57 +0200 schrieb Uwe Kleine-König:
> > second was with CONFIG_LGUEST_GUEST enabled, module build stage 2
> failed
> > with Documentation/lguest not found
> Hmm, there is Documentation/virtual/lguest. Didn't find where
> Documentation/lguest is used from a quick look. Rusty?
> 
> Best regards
> Uwe 

Sorry, I'm far from being a expert, all I can say is that disable 
CONFIG_LGUEST and
CONFIG_LGUEST_GUEST also (auto)disable
CONFIG_PARAVIRT_SPINLOCKS
and then it works, at least here.

regards
hermann

--
To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH] Various fixups from first bug reports

[Trulan Martin]

hermann <brummer- <at> web.de> writes:

> 
> Am Freitag, den 22.07.2011, 08:57 +0200 schrieb Uwe Kleine-König:
> > > second was with CONFIG_LGUEST_GUEST enabled, module build stage 2
> > failed
> > > with Documentation/lguest not found
> > Hmm, there is Documentation/virtual/lguest. Didn't find where
> > Documentation/lguest is used from a quick look. Rusty?
> > 
> > Best regards
> > Uwe 
> 
> Sorry, I'm far from being a expert, all I can say is that disable 
> CONFIG_LGUEST and
> CONFIG_LGUEST_GUEST also (auto)disable
> CONFIG_PARAVIRT_SPINLOCKS
> and then it works, at least here.
> 
> regards
> hermann
> 
> --

CONFIG_LGUEST, etc., had to be disabled for me on 3.0-rc7 as well, I don't think
this is an RT issue.  I'm not sure about CONFIG_PARAVIRT_SPINLOCKS.

But, I am seeing this call trace, related to the b43-lpphy wireless driver
loading.  Everything appears to keep functioning.

[   16.616613] ------------[ cut here ]------------
[   16.616640] WARNING: at net/mac80211/rx.c:2856 ieee80211_rx+0x23/0x6db
[mac80211]()
[   16.616642] Hardware name: Latitude E5500                  
[   16.616644] Modules linked in: af_packet uinput binfmt_misc parport_pc ppdev
lp parport bridge stp bnep rfcomm vboxnetadp vboxnetflt vboxdrv ext3 jbd ipv6
cpufreq_powersave cpufreq_stats cpufreq_ondemand cpufreq_userspace
cpufreq_conservative speedstep_lib fuse loop snd_aloop acpi_cpufreq freq_table
mperf snd_hda_codec_hdmi snd_hda_codec_idt btusb snd_hda_intel snd_hda_codec
bluetooth snd_pcm_oss snd_mixer_oss snd_pcm i915 drm_kms_helper drm
snd_seq_dummy snd_seq_oss snd_seq_midi i2c_algo_bit snd_rawmidi
snd_seq_midi_event snd_seq snd_timer arc4 joydev b43 video snd_seq_device
dell_laptop battery processor ac button dell_wmi sparse_keymap pcspkr wmi dcdbas
yenta_socket pcmcia_rsrc psmouse i2c_i801 i2c_core serio_raw evdev rng_core
mac80211 snd intel_agp intel_gtt soundcore tpm_tis tpm tpm_bios snd_page_alloc
cfg80211 rfkill agpgart ext4 mbcache jbd2 crc16 dm_mirror dm_region_hash dm_log
dm_mod usbhid hid sg sr_mod cdrom sd_mod uhci_hcd ahci ehci_hcd libahci ssb
sdhci_pci usbcore tg3 firewire_ohci libata sdhci pcmcia firewire_core mmc_core
thermal pcmcia_core thermal_sys libphy crc_itu_t [last unloaded: scsi_wait_scan]
[   16.616710] Pid: 2770, comm: irq/17-b43 Not tainted 3.0.0-rc7-rt0-1 #1
[   16.616712] Call Trace:
[   16.616718]  [<c102c807>] ? warn_slowpath_common+0x68/0x79
[   16.616725]  [<f876261a>] ? ieee80211_rx+0x23/0x6db [mac80211]
[   16.616728]  [<c102c825>] ? warn_slowpath_null+0xd/0x10
[   16.616735]  [<f876261a>] ? ieee80211_rx+0x23/0x6db [mac80211]
[   16.616738]  [<c1020e2a>] ? task_rq_lock+0x36/0x64
[   16.616740]  [<c1020d13>] ? set_cpus_allowed_rt+0x11/0xb2
[   16.616743]  [<c1027033>] ? migrate_disable+0xd8/0xf3
[   16.616750]  [<f89e2512>] ? b43_rx+0x40f/0x436 [b43]
[   16.616757]  [<f89e6393>] ? b43_dma_rx+0x211/0x284 [b43]
[   16.616763]  [<f89cfdb3>] ? b43_do_interrupt_thread+0x41c/0x4dd [b43]
[   16.616767]  [<c1015841>] ? __io_apic_modify_irq+0x39/0x57
[   16.616771]  [<c10691aa>] ? unmask_irq+0x11/0x1a
[   16.616776]  [<f89cfe89>] ? b43_interrupt_thread_handler+0x15/0x27 [b43]
[   16.616779]  [<c10680dc>] ? irq_thread_fn+0x10/0x23
[   16.616781]  [<c106801a>] ? irq_thread+0xa0/0x152
[   16.616783]  [<c10680cc>] ? irq_thread+0x152/0x152
[   16.616786]  [<c1067f7a>] ? irq_finalize_oneshot+0x9d/0x9d
[   16.616788]  [<c103fcb4>] ? kthread+0x63/0x68
[   16.616791]  [<c103fc51>] ? __init_kthread_worker+0x36/0x36
[   16.616794]  [<c1248936>] ? kernel_thread_helper+0x6/0xd
[   16.616795] ---[ end trace 0000000000000002 ]---

Thanks,
Trulan



--
To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH] Various fixups from first bug reports

[Rusty Russell]

On Fri, 22 Jul 2011 08:57:11 +0200, Uwe Kleine-König <u.kleine-koenig@pengutronix.de> wrote:
> Hello,
> 
> (These two issues occured on 3.0-rc7-rt0, I think they apply to
> mainline, too.)
> 
> On Fri, Jul 22, 2011 at 07:43:34AM +0200, hermann wrote:
> > well, I run into 2 ops, witch, I guess not related to the rt patch.
> > 
> > first one was with CONFIG_SENSORS_LTC4245, module build stage 2 failed
> > with undefined reference to of_find_property
> of_find_property isn't exported to modules. Maybe of_get_property should
> be used? Should of_find_property be static then? Alternatively export
> of_find_property, too. Grant?
> 
> > second was with CONFIG_LGUEST_GUEST enabled, module build stage 2 failed
> > with Documentation/lguest not found
> Hmm, there is Documentation/virtual/lguest. Didn't find where
> Documentation/lguest is used from a quick look. Rusty?

No, it's not compiled from the kernel build, at least in mainline...

If it is in the -r tree, you need to
s,Documentation/lguest,Documentation/virtual/lguest, as it has moved.

Cheers,
Rusty.
--
To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH] Various fixups from first bug reports

[Uwe Kleine-König]

Hello,

On Fri, Jul 22, 2011 at 08:34:59PM +0930, Rusty Russell wrote:
> On Fri, 22 Jul 2011 08:57:11 +0200, Uwe Kleine-König <u.kleine-koenig@pengutronix.de> wrote:
> > > second was with CONFIG_LGUEST_GUEST enabled, module build stage 2 failed
> > > with Documentation/lguest not found
> > Hmm, there is Documentation/virtual/lguest. Didn't find where
> > Documentation/lguest is used from a quick look. Rusty?
> 
> No, it's not compiled from the kernel build, at least in mainline...
> 
> If it is in the -r tree, you need to
> s,Documentation/lguest,Documentation/virtual/lguest, as it has moved.
I don't understand that and didn't try to reproduce. 
hermann, maybe you can follow up with your failing .config and try to
reproduce on v3.0 (without rt)?

Best regards
Uwe

-- 
Pengutronix e.K.                           | Uwe Kleine-König            |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
--
To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH] Various fixups from first bug reports

[hermann]

[-- Attachment #1: Type: text/plain, Size: 841 bytes --]

Am Freitag, den 22.07.2011, 21:20 +0200 schrieb Uwe Kleine-König:
> I don't understand that and didn't try to reproduce. 
> hermann, maybe you can follow up with your failing .config and try to
> reproduce on v3.0 (without rt)?
> 
> Best regards
> Uwe 

well, it also didn't build from the 3.0 mainline with that config. After
modules build stage 2, it comes to :
. . . . .
IHEX    firmware/yam/9600.bin
make[1]: Leaving directory `/home/brummer/Projekte/Kernel/linux-3.0'
/usr/bin/make    ARCH=i386 \
			     -C Documentation/lguest
make: *** Documentation/lguest: Datei oder Verzeichnis nicht gefunden.
Schluss.
make: *** [debian/stamp/build/kernel] Fehler 2

[Datei oder Verzeichnis nicht gefunden == file or directory not found
Schlusss == exit
Fehler 2 == error 2]

attached is the config. 

regards
hermann

[-- Attachment #2: .config.bz2 --]
[-- Type: application/x-bzip, Size: 28228 bytes --]

Re: [PATCH] Various fixups from first bug reports

[Robin Gareus]

On 07/23/2011 05:17 AM, hermann wrote:
> Am Freitag, den 22.07.2011, 21:20 +0200 schrieb Uwe Kleine-König:
>> I don't understand that and didn't try to reproduce. 
>> hermann, maybe you can follow up with your failing .config and try to
>> reproduce on v3.0 (without rt)?
>>
>> Best regards
>> Uwe 
> 
> well, it also didn't build from the 3.0 mainline with that config. After
> modules build stage 2, it comes to :
> . . . . .
> IHEX    firmware/yam/9600.bin
> make[1]: Leaving directory `/home/brummer/Projekte/Kernel/linux-3.0'
> /usr/bin/make    ARCH=i386 \
> 			     -C Documentation/lguest
> make: *** Documentation/lguest: Datei oder Verzeichnis nicht gefunden.
> Schluss.
> make: *** [debian/stamp/build/kernel] Fehler 2
> 
Hello Herman,

Are you using make-kpkg?  You need to change the path to lguest
in /usr/share/kernel-package/ruleset/targets/doc.mk and two or three
other files there as Rusty outlined.

HTH,
robin
--
To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH] Various fixups from first bug reports

[hermann]

Am Samstag, den 23.07.2011, 11:57 +0200 schrieb Robin Gareus:
> Hello Herman,
> 
> Are you using make-kpkg?  You need to change the path to lguest
> in /usr/share/kernel-package/ruleset/targets/doc.mk and two or three
> other files there as Rusty outlined.
> 
> HTH,
> robin 

Hi Robin

Yes, using make-kpkg. I changed line 61 from 

rm -f $(DOCDIR)/Documentation/lguest/lguest
$(DOCDIR)/Documentation/lguest/lguest.lds

to 

rm -f $(DOCDIR)/Documentation/virtual/lguest/lguest
$(DOCDIR)/Documentation/virtual/lguest/lguest.lds

but it didn't help, the result is still the same.
/usr/bin/make    ARCH=i386 \
			     -C Documentation/lguest
make: *** Documentation/lguest: Datei oder Verzeichnis nicht gefunden.
Schluss.

Could you wrote a bit more precise what needs to do, please ?

regards
hermann

Re: [PATCH] Various fixups from first bug reports

[Robin Gareus]

On 07/23/2011 04:19 PM, hermann wrote:
> Am Samstag, den 23.07.2011, 11:57 +0200 schrieb Robin Gareus:
>> Hello Herman,
>>
>> Are you using make-kpkg?  You need to change the path to lguest
>> in /usr/share/kernel-package/ruleset/targets/doc.mk and two or three
>> other files there as Rusty outlined.
>>
>> HTH,
>> robin 
> 
> Hi Robin
> 
> Yes, using make-kpkg. I changed line 61 from 
> [..]
> but it didn't help, the result is still the same.
> /usr/bin/make    ARCH=i386 \
> 			     -C Documentation/lguest

That's the one in /usr/share/kernel-package/ruleset/targets/image.mk

I don't remember all locations, there were around a dozen, also
including some $CONFIG_* checks. It was pretty straight forward and -
alas - I did not keep a diff.

`grep -rli lguest /usr/share/kernel-package/` should get you going.

If you're really stuck I can `apt-get --reinstall` and create a diff.
Let me know. If it is Debian, be aware that there are also various other
scripts in need of patching (e.g. /etc/init.d/udev supported_kernel()
does not match the 3.X series, yet),..


> Schluss.
ende, aus :)

robin

Re: [PATCH] Various fixups from first bug reports

[hermann]

Am Samstag, den 23.07.2011, 16:33 +0200 schrieb Robin Gareus:
> On 07/23/2011 04:19 PM, hermann wrote:
> > Am Samstag, den 23.07.2011, 11:57 +0200 schrieb Robin Gareus:
> >> Hello Herman,
> >>
> >> Are you using make-kpkg?  You need to change the path to lguest
> >> in /usr/share/kernel-package/ruleset/targets/doc.mk and two or three
> >> other files there as Rusty outlined.
> >>
> >> HTH,
> >> robin 
> > 
> > Hi Robin
> > 
> > Yes, using make-kpkg. I changed line 61 from 
> > [..]
> > but it didn't help, the result is still the same.
> > /usr/bin/make    ARCH=i386 \
> > 			     -C Documentation/lguest
> 
> That's the one in /usr/share/kernel-package/ruleset/targets/image.mk
> 
> I don't remember all locations, there were around a dozen, also
> including some $CONFIG_* checks. It was pretty straight forward and -
> alas - I did not keep a diff.
> 
> `grep -rli lguest /usr/share/kernel-package/` should get you going.
> 
> If you're really stuck I can `apt-get --reinstall` and create a diff.
> Let me know. If it is Debian, be aware that there are also various other
> scripts in need of patching (e.g. /etc/init.d/udev supported_kernel()
> does not match the 3.X series, yet),..
> 
> 
> > Schluss.
> ende, aus :)
> 
> robin

Thanks robin, get it working now. Have changed the locations in 
/usr/share/kernel-package/ruleset/targets/common.mk
/usr/share/kernel-package/ruleset/targets/doc.mk
/usr/share/kernel-package/ruleset/targets/image.mk

It would be cool if you could remember witch other scripts need to be
patched to support the 3.X series.

I mean I run already the Kernel 3.0.0-rc7-rt0 i686 here on debian/sid
without hazel, but be interested to set up all in a matching manner.  

regards
hermann

Re: [PATCH] Various fixups from first bug reports

[Uwe Kleine-König]

Hello,

On Sat, Jul 23, 2011 at 04:19:13PM +0200, hermann wrote:
> Am Samstag, den 23.07.2011, 11:57 +0200 schrieb Robin Gareus:
> Yes, using make-kpkg. I changed line 61 from 
there is an alternative,

	make deb-pkg

in the vanilla kernel. Depending on why you use make-kpkg it might be
worth a try.

Best regards
Uwe

-- 
Pengutronix e.K.                           | Uwe Kleine-König            |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
--
To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[ANNOUNCE] 3.0-rt1

[Thomas Gleixner]

Dear RT Folks,

I'm pleased to announce the 3.0-rt1 release.

Changes versus 3.0-rc7-rt0:

  * Update to Linus final 3.0 release

  * RTC bugfixes (scheduled for mainline/stable)

  * Long standing (rt only) timer_list bug (see
    timers-avoid-the-base-null-otptimization-on-rt.patch in the split
    out quilt queue)

  * Minor non exciting fixes all over the place

Known issues:

  * Some weird "console=..." commandline + config dependent
    interactions which have been not yet investigated down to their
    root cause. Result in a boot hang. YMMV

Patch against 3.0 can be found here:

  http://www.kernel.org/pub/linux/kernel/projects/rt/patch-3.0-rt1.patch.bz2

The split quilt queue is available at:

  http://www.kernel.org/pub/linux/kernel/projects/rt/patches-3.0-rt1.tar.bz2

Thanks,

	tglx

Re: [ANNOUNCE] 3.0-rt1

[Rolando Martins]

Thanks for the great work!

Is it possible to have RT_GROUP_SCHED and PREEMPT_RT_FULL?

Thanks.
Rolando Martins


On Fri, Jul 22, 2011 at 9:37 PM, Thomas Gleixner <tglx@linutronix.de> wrote:
> Dear RT Folks,
>
> I'm pleased to announce the 3.0-rt1 release.
>
> Changes versus 3.0-rc7-rt0:
>
>  * Update to Linus final 3.0 release
>
>  * RTC bugfixes (scheduled for mainline/stable)
>
>  * Long standing (rt only) timer_list bug (see
>    timers-avoid-the-base-null-otptimization-on-rt.patch in the split
>    out quilt queue)
>
>  * Minor non exciting fixes all over the place
>
> Known issues:
>
>  * Some weird "console=..." commandline + config dependent
>    interactions which have been not yet investigated down to their
>    root cause. Result in a boot hang. YMMV
>
> Patch against 3.0 can be found here:
>
>  http://www.kernel.org/pub/linux/kernel/projects/rt/patch-3.0-rt1.patch.bz2
>
> The split quilt queue is available at:
>
>  http://www.kernel.org/pub/linux/kernel/projects/rt/patches-3.0-rt1.tar.bz2
>
> Thanks,
>
>        tglx
> --
> To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>

Re: [ANNOUNCE] 3.0-rt1

[Thomas Gleixner]

On Fri, 22 Jul 2011, Rolando Martins wrote:

> Thanks for the great work!
> 
> Is it possible to have RT_GROUP_SCHED and PREEMPT_RT_FULL?

Not at the momenent. RT_GROUP_SCHED depends on !PREEMPT_RT_FULL
because it leads to yet undebugged hangs and failures.

Thanks,

	tglx

Re: [ANNOUNCE] 3.0-rt1

[Rolando Martins]

Hi Thomas,

Do you see it being fixed in a near future?

Thanks,
Rolando

On Sat, Jul 23, 2011 at 8:31 AM, Thomas Gleixner <tglx@linutronix.de> wrote:
> On Fri, 22 Jul 2011, Rolando Martins wrote:
>
>> Thanks for the great work!
>>
>> Is it possible to have RT_GROUP_SCHED and PREEMPT_RT_FULL?
>
> Not at the momenent. RT_GROUP_SCHED depends on !PREEMPT_RT_FULL
> because it leads to yet undebugged hangs and failures.
>
> Thanks,
>
>        tglx
>
--
To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [ANNOUNCE] 3.0-rt1

[Thomas Gleixner]

On Sat, 23 Jul 2011, Rolando Martins wrote:
> Hi Thomas,
> 
> Do you see it being fixed in a near future?

Delegating that question to scheduler wizards and folks you created
that mess in the first place.

Re: [ANNOUNCE] 3.0-rt1

[Peter Zijlstra]

On Sat, 2011-07-23 at 12:20 +0200, Thomas Gleixner wrote:
> On Sat, 23 Jul 2011, Rolando Martins wrote:
> > Hi Thomas,
> > 
> > Do you see it being fixed in a near future?
> 
> Delegating that question to scheduler wizards and folks you created
> that mess in the first place.

Uhm what?

Re: [ANNOUNCE] 3.0-rt1

[Thomas Gleixner]

On Sat, 23 Jul 2011, Peter Zijlstra wrote:

> On Sat, 2011-07-23 at 12:20 +0200, Thomas Gleixner wrote:
> > On Sat, 23 Jul 2011, Rolando Martins wrote:
> > > Hi Thomas,
> > > 
> > > Do you see it being fixed in a near future?
> > 
> > Delegating that question to scheduler wizards and folks you created
> > that mess in the first place.
> 
> Uhm what?

RT_GROUP_SCHED & RT 

Re: [ANNOUNCE] 3.0-rt1

[Ed Tomlinson]

On Friday 22 July 2011 16:37:24 Thomas Gleixner wrote:
> Dear RT Folks,
> 
> I'm pleased to announce the 3.0-rt1 release.

Thomas,

It does not build here with CONFIG_RCU_TORTURE_TEST=m

Without that it build here on amd64.

Thanks,
Ed

Re: [ANNOUNCE] 3.0-rt1

[Thomas Gleixner]

On Fri, 22 Jul 2011, Ed Tomlinson wrote:

> On Friday 22 July 2011 16:37:24 Thomas Gleixner wrote:
> > Dear RT Folks,
> > 
> > I'm pleased to announce the 3.0-rt1 release.
> 
> Thomas,
> 
> It does not build here with CONFIG_RCU_TORTURE_TEST=m

Fix queued for -rt2.

Thanks,

	tglx

Re: [ANNOUNCE] 3.0-rt1

[Maarten Lankhorst]

On 07/22/2011 10:37 PM, Thomas Gleixner wrote:
> Dear RT Folks,
>
> I'm pleased to announce the 3.0-rt1 release.
>
> Changes versus 3.0-rc7-rt0:
>
>   * Update to Linus final 3.0 release
>
>   * RTC bugfixes (scheduled for mainline/stable)
>
>   * Long standing (rt only) timer_list bug (see
>     timers-avoid-the-base-null-otptimization-on-rt.patch in the split
>     out quilt queue)
>
>   * Minor non exciting fixes all over the place
>
> Known issues:
>
>   * Some weird "console=..." commandline + config dependent
>     interactions which have been not yet investigated down to their
>     root cause. Result in a boot hang. YMMV
Sweet, works here :)

Only the nvidia driver fails to build because of rtmutex and migrate enable/disable being GPL only, you might want to reconsider the GPL only status of them.

~Maarten

Re: [ANNOUNCE] 3.0-rt1

[Peter W. Morreale]

On Sat, 2011-07-23 at 03:05 +0200, Maarten Lankhorst wrote:
> On 07/22/2011 10:37 PM, Thomas Gleixner wrote:
> > Dear RT Folks,
> >
> > I'm pleased to announce the 3.0-rt1 release.
> >
> > Changes versus 3.0-rc7-rt0:
> >
> >   * Update to Linus final 3.0 release
> >
> >   * RTC bugfixes (scheduled for mainline/stable)
> >
> >   * Long standing (rt only) timer_list bug (see
> >     timers-avoid-the-base-null-otptimization-on-rt.patch in the split
> >     out quilt queue)
> >
> >   * Minor non exciting fixes all over the place
> >
> > Known issues:
> >
> >   * Some weird "console=..." commandline + config dependent
> >     interactions which have been not yet investigated down to their
> >     root cause. Result in a boot hang. YMMV
> Sweet, works here :)
> 
> Only the nvidia driver fails to build because of rtmutex and migrate enable/disable being GPL only, you might want to reconsider the GPL only status of them.
> 
> ~Maarten

Or, alternatively, please ask Nvidia to GPL their driver.

Best
-PWM


> --
> To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [ANNOUNCE] 3.0-rt1

[Mike Galbraith]

On Sat, 2011-07-23 at 07:34 -0600, Peter W. Morreale wrote:
> On Sat, 2011-07-23 at 03:05 +0200, Maarten Lankhorst wrote:

> > Only the nvidia driver fails to build because of rtmutex and migrate enable/disable being GPL only, you might want to reconsider the GPL only status of them.
> > 
> > ~Maarten
> 
> Or, alternatively, please ask Nvidia to GPL their driver.

I know a couple folks for whom this will eventually become a problem.

One is doing (very) high performance aircraft simulation, where
rendering performance highly important.  AFAIK there is currently no
viable alternative to "the blob"

	-Mike

Re: [ANNOUNCE] 3.0-rt1

[Mike Galbraith]

On Wed, 2011-08-10 at 07:14 +0200, Mike Galbraith wrote:
> On Sat, 2011-07-23 at 07:34 -0600, Peter W. Morreale wrote:
> > On Sat, 2011-07-23 at 03:05 +0200, Maarten Lankhorst wrote:
> 
> > > Only the nvidia driver fails to build because of rtmutex and migrate enable/disable being GPL only, you might want to reconsider the GPL only status of them.
> > > 
> > > ~Maarten
> > 
> > Or, alternatively, please ask Nvidia to GPL their driver.
> 
> I know a couple folks for whom this will eventually become a problem.

(maybe one of them will buy nvidia, and GPL the driver.. right:)

Re: [ANNOUNCE] 3.0-rt1

[Peter W. Morreale]

On Wed, 2011-08-10 at 07:14 +0200, Mike Galbraith wrote:
> On Sat, 2011-07-23 at 07:34 -0600, Peter W. Morreale wrote:
> > On Sat, 2011-07-23 at 03:05 +0200, Maarten Lankhorst wrote:
> 
> > > Only the nvidia driver fails to build because of rtmutex and migrate enable/disable being GPL only, you might want to reconsider the GPL only status of them.
> > > 
> > > ~Maarten
> > 
> > Or, alternatively, please ask Nvidia to GPL their driver.
> 
> I know a couple folks for whom this will eventually become a problem.
> 
> One is doing (very) high performance aircraft simulation, where
> rendering performance highly important.  AFAIK there is currently no
> viable alternative to "the blob"
> 
> 	-Mike
> 

Great!  Perhaps they can talk to Nvidia as well.  

Best,
-PWM

Re: [ANNOUNCE] 3.0-rt1

[Maarten Lankhorst]

Hi Dietrich,

On 07/23/2011 08:57 PM, Sven-Thorsten Dietrich wrote:
> On Jul 22, 2011, at 6:05 PM, Maarten Lankhorst wrote:
>
>> On 07/22/2011 10:37 PM, Thomas Gleixner wrote:
>>> Dear RT Folks,
>>>
>>> I'm pleased to announce the 3.0-rt1 release.
>>>
>>> Changes versus 3.0-rc7-rt0:
>>>
>>>  * Update to Linus final 3.0 release
>>>
>>>  * RTC bugfixes (scheduled for mainline/stable)
>>>
>>>  * Long standing (rt only) timer_list bug (see
>>>    timers-avoid-the-base-null-otptimization-on-rt.patch in the split
>>>    out quilt queue)
>>>
>>>  * Minor non exciting fixes all over the place
>>>
>>> Known issues:
>>>
>>>  * Some weird "console=..." commandline + config dependent
>>>    interactions which have been not yet investigated down to their
>>>    root cause. Result in a boot hang. YMMV
>> Sweet, works here :)
>>
>> Only the nvidia driver fails to build because of rtmutex and migrate enable/disable being GPL only, you might want to reconsider the GPL only status of them.
> Can you reconsider using nVidia ?
You see, I've been living two lives. In one life, I'm Maarten B Lankhorst. I work on wine, and since a week or so I've been working on nouveau. And I... help fix kernel bugs that I'm affected by. My other life is in computers, where I go by the hacker alias ROFLGAMER, participating actively in computer simulated environments and am playing graphic intensive games using wine. Unfortunately when I do that I just want things to work, nouveau keeps getting better all the time, but wine's d3d is a lot more tested with the nvidia blob. ;)

~Maarten

Re: [ANNOUNCE] 3.0-rt1

[Mike Galbraith]

On Fri, 2011-07-22 at 22:37 +0200, Thomas Gleixner wrote:
> Dear RT Folks,
> 
> I'm pleased to announce the 3.0-rt1 release.

drivers-dca-convert-dcalock-to-raw.patch needs incremental update.

---
 drivers/dca/dca-core.c |    6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

Index: linux-3.0-rt/drivers/dca/dca-core.c
===================================================================
--- linux-3.0-rt.orig/drivers/dca/dca-core.c
+++ linux-3.0-rt/drivers/dca/dca-core.c
@@ -101,10 +101,10 @@ static void unregister_dca_providers(voi
 
 	INIT_LIST_HEAD(&unregistered_providers);
 
-	spin_lock_irqsave(&dca_lock, flags);
+	raw_spin_lock_irqsave(&dca_lock, flags);
 
 	if (list_empty(&dca_domains)) {
-		spin_unlock_irqrestore(&dca_lock, flags);
+		raw_spin_unlock_irqrestore(&dca_lock, flags);
 		return;
 	}
 
@@ -116,7 +116,7 @@ static void unregister_dca_providers(voi
 
 	dca_free_domain(domain);
 
-	spin_unlock_irqrestore(&dca_lock, flags);
+	raw_spin_unlock_irqrestore(&dca_lock, flags);
 
 	list_for_each_entry_safe(dca, _dca, &unregistered_providers, node) {
 		dca_sysfs_remove_provider(dca);

Re: [ANNOUNCE] 3.0-rt1

[Thomas Gleixner]

On Sat, 23 Jul 2011, Mike Galbraith wrote:

> On Fri, 2011-07-22 at 22:37 +0200, Thomas Gleixner wrote:
> > Dear RT Folks,
> > 
> > I'm pleased to announce the 3.0-rt1 release.
> 
> drivers-dca-convert-dcalock-to-raw.patch needs incremental update.

Indeed. Thanks for spotting.

Re: [ANNOUNCE] 3.0-rt1

[Juergen Beisert]

Hi Thomas,

on ARM (in this case a Samsung S3C2440 based platform) I get:

kernel/built-in.o: In function `vprintk':
/home/jb/[...]/linux-3.0/kernel/printk.c:862: undefined reference to `early_vprintk'
kernel/built-in.o: In function `register_console':
/home/jb/[...]linux-3.0/kernel/printk.c:1580: undefined reference to `early_console_initialized'
make[1]: *** [vmlinux] Error 1

Did I forget something to enable in my kernel config?

jbe

-- 
Pengutronix e.K.                              | Juergen Beisert             |
Linux Solutions for Science and Industry      | http://www.pengutronix.de/  |

Re: [ANNOUNCE] 3.0-rt1

[Robin Gareus]

On 07/23/2011 04:29 PM, Juergen Beisert wrote:
> Hi Thomas,
> 
> on ARM (in this case a Samsung S3C2440 based platform) I get:
> 
> kernel/built-in.o: In function `vprintk':
> /home/jb/[...]/linux-3.0/kernel/printk.c:862: undefined reference to `early_vprintk'
> kernel/built-in.o: In function `register_console':
> /home/jb/[...]linux-3.0/kernel/printk.c:1580: undefined reference to `early_console_initialized'
> make[1]: *** [vmlinux] Error 1
> 
> Did I forget something to enable in my kernel config?

I've seen the same error when trying to work around vprintk issue on
i386 with CONFIG_EXPERT=y  CONFIG_EARLY_PRINTK=n

instead of fixing 'printk.c' I've changed 'include/linux/printk.h:95'
 -void early_vprintk(const char *fmt, va_list args);
 +asmlinkage void early_vprintk(const char *fmt, va_list args);

and compiled with CONFIG_EARLY_PRINTK=y. Not sure if that is an option
on ARM.

Herman has reported the latter issue earlier today.

ciao,
robin

Re: [ANNOUNCE] 3.0-rt1

[Thomas Gleixner]

On Sat, 23 Jul 2011, Juergen Beisert wrote:

> Hi Thomas,
> 
> on ARM (in this case a Samsung S3C2440 based platform) I get:
> 
> kernel/built-in.o: In function `vprintk':
> /home/jb/[...]/linux-3.0/kernel/printk.c:862: undefined reference to `early_vprintk'
> kernel/built-in.o: In function `register_console':
> /home/jb/[...]linux-3.0/kernel/printk.c:1580: undefined reference to `early_console_initialized'
> make[1]: *** [vmlinux] Error 1
> 
> Did I forget something to enable in my kernel config?

No, that's my fault. I merged a late minute debugfeature yesterday and
forgot to enable early_printk on ARM. Will be fixed in -rt2 which i'm
about to release.

Thanks,

	tglx