spin_lock problem in recent kernel

spin_lock problem in recent kernel

[Kaoru Fukui]

Hi!

I had tried SMP kernel using paulus rsync tree and 2_4 bk tree.
Since 2.4.3pre6 it has spin_lock problem in both tree.

I used that patch when pre6,pre7
Then I could boot kernel and it's work well.

2.4.3pre8 is not boot that has the other spin_lock problem.
(around ip_tables loading)

When i used this patch when pre6, pre7.

Please check spin_lock problem of pre8 with SMP kernel.

the machine is G4 dual
Thanks

Kaoru

diff -urN base/arch/ppc/mm/init.c fixed/arch/ppc/mm/init.c
--- base/arch/ppc/mm/init.c	Thu Mar 22 19:29:46 2001
+++ fixed/arch/ppc/mm/init.c	Fri Mar 23 05:52:33 2001
@@ -436,11 +436,11 @@
 	 * Should check if it is a candidate for a BAT mapping
 	 */

-	spin_lock(&init_mm.page_table_lock);
+//	spin_lock(&init_mm.page_table_lock);
 	err = 0;
 	for (i = 0; i < size && err == 0; i += PAGE_SIZE)
 		err = map_page(v+i, p+i, flags);
-	spin_unlock(&init_mm.page_table_lock);
+//	spin_unlock(&init_mm.page_table_lock);
 	if (err) {
 		if (mem_init_done)
 			vfree((void *)v);


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: spin_lock problem in recent kernel

[Tom Gall]

Hi Kaoru,

  I've found much the same as you... however I've also found that the
patch you have below (which is exactly the same as a change I have made
to the kernel I've been debugging) hasn't helped me too much. The
page_table_lock would still get messed up during the boot sequence on
Power3 with SMP on.

  I don't have any other great enlightenment as I'm taking the day off
... be back at it tomorrow!

  Regards,

  Tom

Kaoru Fukui wrote:
>
> Hi!
>
> I had tried SMP kernel using paulus rsync tree and 2_4 bk tree.
> Since 2.4.3pre6 it has spin_lock problem in both tree.
>
> I used that patch when pre6,pre7
> Then I could boot kernel and it's work well.
>
> 2.4.3pre8 is not boot that has the other spin_lock problem.
> (around ip_tables loading)
>
> When i used this patch when pre6, pre7.
>
> Please check spin_lock problem of pre8 with SMP kernel.
>
> the machine is G4 dual
> Thanks
>
> Kaoru
>
> diff -urN base/arch/ppc/mm/init.c fixed/arch/ppc/mm/init.c
> --- base/arch/ppc/mm/init.c     Thu Mar 22 19:29:46 2001
> +++ fixed/arch/ppc/mm/init.c    Fri Mar 23 05:52:33 2001
> @@ -436,11 +436,11 @@
>          * Should check if it is a candidate for a BAT mapping
>          */
>
> -       spin_lock(&init_mm.page_table_lock);
> +//     spin_lock(&init_mm.page_table_lock);
>         err = 0;
>         for (i = 0; i < size && err == 0; i += PAGE_SIZE)
>                 err = map_page(v+i, p+i, flags);
> -       spin_unlock(&init_mm.page_table_lock);
> +//     spin_unlock(&init_mm.page_table_lock);
>         if (err) {
>                 if (mem_init_done)
>                         vfree((void *)v);
>

--
Regards,

Tom

--------------------------------------------------------------
Linux Guy             "My heart is human, my blood is boiling,
gall@rochcivictheatre.org    my brain IBM" -- Mr Roboto, Styxx

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: spin_lock problem in recent kernel

[Kaoru Fukui]

Hi

This is booing message when kernel-2.4.3pre8-SMP.

<snip>
TCP: Hash tables configured (established 8192 bind 8192)
ip_conntrack(2048 buckets,16384 max)
_spin_lock(c02cc470) CPU# 1 NIP c0033404 holder: cpu 1 pc C002B384
_spin_lock(c02ce740) CPU# 0 NIP c0015890 holder: cpu 1 pc C0015890
_spin_lock(c02cc470) CPU# 1 NIP c0033404 holder: cpu 1 pc C002B384
_spin_lock(c02ce740) CPU# 0 NIP c0015890 holder: cpu 1 pc C0015890
_spin_lock(c02cc470) CPU# 1 NIP c0033404 holder: cpu 1 pc C002B384
_spin_lock(c02ce740) CPU# 0 NIP c0015890 holder: cpu 1 pc C0015890
_spin_lock(c02cc470) CPU# 1 NIP c0033404 holder: cpu 1 pc C002B384
_spin_lock(c02ce740) CPU# 0 NIP c0015890 holder: cpu 1 pc C0015890

continue same messages

Any help
Kaoru




From: Tom Gall <gall@rochcivictheatre.org>
To: Kaoru Fukui <k_fukui@highway.ne.jp>
Subject: Re: spin_lock problem in recent kernel
Date: Wed, 28 Mar 2001 13:02:05 -0600
CC: linuxppc-dev@lists.linuxppc.org

> Hi Kaoru,
>
>   I've found much the same as you... however I've also found that the
> patch you have below (which is exactly the same as a change I have made
> to the kernel I've been debugging) hasn't helped me too much. The
> page_table_lock would still get messed up during the boot sequence on
> Power3 with SMP on.
>
>   I don't have any other great enlightenment as I'm taking the day off
> ... be back at it tomorrow!
>
>   Regards,
>
>   Tom
>
> Kaoru Fukui wrote:
> >
> > Hi!
> >
> > I had tried SMP kernel using paulus rsync tree and 2_4 bk tree.
> > Since 2.4.3pre6 it has spin_lock problem in both tree.
> >
> > I used that patch when pre6,pre7
> > Then I could boot kernel and it's work well.
> >
> > 2.4.3pre8 is not boot that has the other spin_lock problem.
> > (around ip_tables loading)
> >
> > When i used this patch when pre6, pre7.
> >
> > Please check spin_lock problem of pre8 with SMP kernel.
> >
> > the machine is G4 dual
> > Thanks
> >
> > Kaoru
> >
> > diff -urN base/arch/ppc/mm/init.c fixed/arch/ppc/mm/init.c
> > --- base/arch/ppc/mm/init.c     Thu Mar 22 19:29:46 2001
> > +++ fixed/arch/ppc/mm/init.c    Fri Mar 23 05:52:33 2001
> > @@ -436,11 +436,11 @@
> >          * Should check if it is a candidate for a BAT mapping
> >          */
> >
> > -       spin_lock(&init_mm.page_table_lock);
> > +//     spin_lock(&init_mm.page_table_lock);
> >         err = 0;
> >         for (i = 0; i < size && err == 0; i += PAGE_SIZE)
> >                 err = map_page(v+i, p+i, flags);
> > -       spin_unlock(&init_mm.page_table_lock);
> > +//     spin_unlock(&init_mm.page_table_lock);
> >         if (err) {
> >                 if (mem_init_done)
> >                         vfree((void *)v);
> >
>
> --
> Regards,
>
> Tom
>
> --------------------------------------------------------------
> Linux Guy             "My heart is human, my blood is boiling,
> gall@rochcivictheatre.org    my brain IBM" -- Mr Roboto, Styxx
>


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Success kernel-243pre8-SMP(patch)(Re: spin_lock problem in recent k ernel

[Kaoru Fukui]

Many thanks Tom,

I have had a success for kernel-2.4.3.pre8-SMP on Dual G4.
This kernel needs the patch.

Kernel Guys, Please look the patch

I will be able to upload this binary kernel into my directory tomorrow.

Kaoru
-------- this patch is needed for SMP kernel-2.4.3.pre8  ---------

diff -urN base/arch/ppc/mm/init.c fixed/arch/ppc/mm/init.c
--- base/arch/ppc/mm/init.c	Thu Mar 29 06:02:35 2001
+++ fixed/arch/ppc/mm/init.c	Fri Mar 30 05:51:16 2001
@@ -437,12 +437,15 @@
 	/*
 	 * Should check if it is a candidate for a BAT mapping
 	 */
-
+#ifndef CONFIG_SMP
 	spin_lock(&init_mm.page_table_lock);
+#endif
 	err = 0;
 	for (i = 0; i < size && err == 0; i += PAGE_SIZE)
 		err = map_page(v+i, p+i, flags);
+#ifndef CONFIG_SMP
 	spin_unlock(&init_mm.page_table_lock);
+#endif
 	if (err) {
 		if (mem_init_done)
 			vfree((void *)v);
diff -urN base/mm/memory.c fixed/mm/memory.c
--- base/mm/memory.c	Thu Mar 29 09:22:35 2001
+++ fixed/mm/memory.c	Fri Mar 30 05:52:17 2001
@@ -1320,9 +1320,13 @@
 		/* "fast" allocation can happen without dropping the lock.. */
 		new = pte_alloc_one_fast(address);
 		if (!new) {
+#ifndef CONFIG_SMP
 			spin_unlock(&mm->page_table_lock);
+#endif
 			new = pte_alloc_one(address);
+#ifndef CONFIG_SMP
 			spin_lock(&mm->page_table_lock);
+#endif
 			if (!new)
 				return NULL;

---------- END -------------

> > This is booing message when kernel-2.4.3pre8-SMP.
> >
> > <snip>
> > TCP: Hash tables configured (established 8192 bind 8192)
> > ip_conntrack(2048 buckets,16384 max)
> > _spin_lock(c02cc470) CPU# 1 NIP c0033404 holder: cpu 1 pc C002B384
> > _spin_lock(c02ce740) CPU# 0 NIP c0015890 holder: cpu 1 pc C0015890

>   Can you look at your system.map for that kernel and tell me what is at
> addresses:

> This is getton  data from system.map

> c00153b0 T schedule_tail
> <C001589>
> c0015a4c T __wake_up
> <snip>
> c002b28c T pte_alloc
> <C002b384>
> c002b3d8 T make_pages_present
> <snip>
> c02cc42c D init_mm
> <c02cc470>
> c02cc4d0 D init_task_union
> <snip>
> c02ce738 d null_pci_ops
> <c02ce740>
> c02ce750 D kernel_flag
>

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: Success kernel-243pre8-SMP(patch)(Re: spin_lock problem in recent kernel

[Takashi Oe]

On Thu, 29 Mar 2001, Tom Gall wrote:

> Hi Kaoru,
>
>   I tried your patch on my 2 way Power3 Model 270 RS/6000 just before I
> left but didn't have enough time to write this note until now.
>
>   Your patch amazingly booted my box quite a bit further... it still
> dies for me but in a new and interesting way, that doesn't appear to be
> related to the page_table_lock. (Token Ring and shared memory
> ...sigh...)
>
>   The part that REALLY REALLY bothers me tho is why does this patch
> work? I *HAVE* CONFIG_SMP on... least I think I do... there's two
> penguins on my screen ... two processors grinding away ... that's
> CONFIG_SMP right??? I'm not losing my mind am I?
>
>   So assuming that CONFIG_SMP is on, this patch shouldn't yeild anything
> different what so ever from a codegen perspective with or without the
> patch for an SMP kernel build. It doesn't make sense.

He is just commenting out those lines.  They are not "#ifdef CONFIG_SMP".


Takashi Oe


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: Success kernel-243pre8-SMP(patch)(Re: spin_lock problem in recent kernel

[Tom Gall]

Hi Kaoru,

  I tried your patch on my 2 way Power3 Model 270 RS/6000 just before I
left but didn't have enough time to write this note until now.

  Your patch amazingly booted my box quite a bit further... it still
dies for me but in a new and interesting way, that doesn't appear to be
related to the page_table_lock. (Token Ring and shared memory
...sigh...)

  The part that REALLY REALLY bothers me tho is why does this patch
work? I *HAVE* CONFIG_SMP on... least I think I do... there's two
penguins on my screen ... two processors grinding away ... that's
CONFIG_SMP right??? I'm not losing my mind am I?

  So assuming that CONFIG_SMP is on, this patch shouldn't yeild anything
different what so ever from a codegen perspective with or without the
patch for an SMP kernel build. It doesn't make sense.

Kaoru Fukui wrote:
>
> Many thanks Tom,
>
> I have had a success for kernel-2.4.3.pre8-SMP on Dual G4.
> This kernel needs the patch.
>
> Kernel Guys, Please look the patch
>
> I will be able to upload this binary kernel into my directory tomorrow.
>
> Kaoru
> -------- this patch is needed for SMP kernel-2.4.3.pre8  ---------
>
> diff -urN base/arch/ppc/mm/init.c fixed/arch/ppc/mm/init.c
> --- base/arch/ppc/mm/init.c     Thu Mar 29 06:02:35 2001
> +++ fixed/arch/ppc/mm/init.c    Fri Mar 30 05:51:16 2001
> @@ -437,12 +437,15 @@
>         /*
>          * Should check if it is a candidate for a BAT mapping
>          */
> -
> +#ifndef CONFIG_SMP
>         spin_lock(&init_mm.page_table_lock);
> +#endif
>         err = 0;
>         for (i = 0; i < size && err == 0; i += PAGE_SIZE)
>                 err = map_page(v+i, p+i, flags);
> +#ifndef CONFIG_SMP
>         spin_unlock(&init_mm.page_table_lock);
> +#endif
>         if (err) {
>                 if (mem_init_done)
>                         vfree((void *)v);
> diff -urN base/mm/memory.c fixed/mm/memory.c
> --- base/mm/memory.c    Thu Mar 29 09:22:35 2001
> +++ fixed/mm/memory.c   Fri Mar 30 05:52:17 2001
> @@ -1320,9 +1320,13 @@
>                 /* "fast" allocation can happen without dropping the lock.. */
>                 new = pte_alloc_one_fast(address);
>                 if (!new) {
> +#ifndef CONFIG_SMP
>                         spin_unlock(&mm->page_table_lock);
> +#endif
>                         new = pte_alloc_one(address);
> +#ifndef CONFIG_SMP
>                         spin_lock(&mm->page_table_lock);
> +#endif
>                         if (!new)
>                                 return NULL;
>
> ---------- END -------------
>
> > > This is booing message when kernel-2.4.3pre8-SMP.
> > >
> > > <snip>
> > > TCP: Hash tables configured (established 8192 bind 8192)
> > > ip_conntrack(2048 buckets,16384 max)
> > > _spin_lock(c02cc470) CPU# 1 NIP c0033404 holder: cpu 1 pc C002B384
> > > _spin_lock(c02ce740) CPU# 0 NIP c0015890 holder: cpu 1 pc C0015890
>
> >   Can you look at your system.map for that kernel and tell me what is at
> > addresses:
>
> > This is getton  data from system.map
>
> > c00153b0 T schedule_tail
> > <C001589>
> > c0015a4c T __wake_up
> > <snip>
> > c002b28c T pte_alloc
> > <C002b384>
> > c002b3d8 T make_pages_present
> > <snip>
> > c02cc42c D init_mm
> > <c02cc470>
> > c02cc4d0 D init_task_union
> > <snip>
> > c02ce738 d null_pci_ops
> > <c02ce740>
> > c02ce750 D kernel_flag
> >
>

--
Regards,

Tom

--------------------------------------------------------------
Linux Guy             "My heart is human, my blood is boiling,
gall@rochcivictheatre.org    my brain IBM" -- Mr Roboto, Styxx

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: Success kernel-243pre8-SMP(patch)(Re: spin_lock problem in recent kernel

[Tony Mantler]

At 8:18 PM -0600 3/29/2001, Tom Gall wrote:
>  So assuming that CONFIG_SMP is on, this patch shouldn't yeild anything
>different what so ever from a codegen perspective with or without the
>patch for an SMP kernel build. It doesn't make sense.
>
>Kaoru Fukui wrote:
[...]
>> +#ifndef CONFIG_SMP
>>         spin_lock(&init_mm.page_table_lock);
>> +#endif

That's #if*n*def, not #ifdef. :)

And speaking of #ifndef, why are these spinlocks here in the first place
and/or why do they now need to be removed? Just suddenly removing locks
because of a new deadlock condition gets my spidey senses tingling.


Cheers - Tony 'Nicoya' Mantler :)


--
Tony "Nicoya" Mantler - Renaissance Nerd Extraordinaire - nicoya@apia.dhs.org
Winnipeg, Manitoba, Canada           --           http://nicoya.feline.pp.se/


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: Success kernel-243pre8-SMP(patch)(Re: spin_lock problem in recent kernel

[Tom Gall]

Takashi Oe wrote:
> He is just commenting out those lines.  They are not "#ifdef CONFIG_SMP".

Sigh ... some days it just doesn't pay to get out of bed... I missed the
'n'....

*BONK *

Thanks!

> Takashi Oe

--
Regards,

Tom

--------------------------------------------------------------
Linux Guy             "My heart is human, my blood is boiling,
gall@rochcivictheatre.org    my brain IBM" -- Mr Roboto, Styxx

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: Success kernel-243pre8-SMP(patch)(Re: spin_lock problem inrecent kernel

[Tom Gall]

Tony Mantler wrote:

> That's #if*n*def, not #ifdef. :)
>
> And speaking of #ifndef, why are these spinlocks here in the first place
> and/or why do they now need to be removed? Just suddenly removing locks
> because of a new deadlock condition gets my spidey senses tingling.

It all stinks to high heaven, that's for sure. Removing the
page_table_lock from being locked in ioremap, saves us from stopping
cold during the first ioremap (at least on Power3). And as far as the
second, I'm in complete agreement...  warning warning danger will
robinson!

--
Regards,

Tom

--------------------------------------------------------------
Linux Guy             "My heart is human, my blood is boiling,
gall@rochcivictheatre.org    my brain IBM" -- Mr Roboto, Styxx

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: Success kernel-243pre8-SMP(patch)(Re: spin_lock problem in recent k ernel

[Geert Uytterhoeven]

On Fri, 30 Mar 2001, Kaoru Fukui wrote:
> +#ifndef CONFIG_SMP
>  	spin_lock(&init_mm.page_table_lock);
> +#endif

AFAIK spin_lock() is void when not compiling for SMP, so this is equal to just
removing the call.

Gr{oetje,eeting}s,

						Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
							    -- Linus Torvalds


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: Success kernel-243pre8-SMP(patch)(Re: spin_lock problem inrecent kernel

[Olaf Hering]

On Thu, Mar 29, Tom Gall wrote:

>
> Tony Mantler wrote:
>
> > That's #if*n*def, not #ifdef. :)
> >
> > And speaking of #ifndef, why are these spinlocks here in the first place
> > and/or why do they now need to be removed? Just suddenly removing locks
> > because of a new deadlock condition gets my spidey senses tingling.
>
> It all stinks to high heaven, that's for sure. Removing the
> page_table_lock from being locked in ioremap, saves us from stopping
> cold during the first ioremap (at least on Power3). And as far as the
> second, I'm in complete agreement...  warning warning danger will
> robinson!

That patch solved the boot problem on my 43p 260. Last progress output
was MMU:exit.
It boots and runs fine now.


Gruss Olaf

--
 $ man clone

BUGS
       Main feature not yet implemented...

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/