[PATCH 12/15] powerpc: Do not use op_model_cell under LPAR

[PATCH 12/15] powerpc: Do not use op_model_cell under LPAR

[Ishizaki Kou]

op_model_cell supports native Cell processor. Under LPAR environment,
it cannot work.

Signed-off-by: Kou Ishizaki <kou.ishizaki.co.jp>
---

Index: linux-powerpc-git/arch/powerpc/oprofile/common.c
diff -u linux-powerpc-git/arch/powerpc/oprofile/common.c:1.1.1.1 linux-powerpc-git/arch/powerpc/oprofile/common.c:1.2
--- linux-powerpc-git/arch/powerpc/oprofile/common.c:1.1.1.1	Wed Dec  6 08:25:43 2006
+++ linux-powerpc-git/arch/powerpc/oprofile/common.c	Wed Dec  6 08:43:15 2006
@@ -149,6 +149,8 @@
 #ifdef CONFIG_PPC64
 #ifdef CONFIG_PPC_CELL_NATIVE
 		case PPC_OPROFILE_CELL:
+			if (firmware_has_feature(FW_FEATURE_LPAR))
+				return -ENODEV;
 			model = &op_model_cell;
 			break;
 #endif

Re: [PATCH 12/15] powerpc: Do not use op_model_cell under LPAR

[Michael Ellerman]

[-- Attachment #1: Type: text/plain, Size: 1136 bytes --]

On Tue, 2006-12-12 at 13:02 +0900, Ishizaki Kou wrote:
> op_model_cell supports native Cell processor. Under LPAR environment,
> it cannot work.
> 
> Signed-off-by: Kou Ishizaki <kou.ishizaki.co.jp>
> ---
> 
> Index: linux-powerpc-git/arch/powerpc/oprofile/common.c
> diff -u linux-powerpc-git/arch/powerpc/oprofile/common.c:1.1.1.1 linux-powerpc-git/arch/powerpc/oprofile/common.c:1.2
> --- linux-powerpc-git/arch/powerpc/oprofile/common.c:1.1.1.1	Wed Dec  6 08:25:43 2006
> +++ linux-powerpc-git/arch/powerpc/oprofile/common.c	Wed Dec  6 08:43:15 2006
> @@ -149,6 +149,8 @@
>  #ifdef CONFIG_PPC64
>  #ifdef CONFIG_PPC_CELL_NATIVE
>  		case PPC_OPROFILE_CELL:
> +			if (firmware_has_feature(FW_FEATURE_LPAR))
> +				return -ENODEV;
>  			model = &op_model_cell;
>  			break;
>  #endif

Is FW_FEATURE_LPAR is the right switch to use here, I'm not sure.

cheers

-- 
Michael Ellerman
OzLabs, IBM Australia Development Lab

wwweb: http://michael.ellerman.id.au
phone: +61 2 6212 1183 (tie line 70 21183)

We do not inherit the earth from our ancestors,
we borrow it from our children. - S.M.A.R.T Person

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [PATCH 12/15] powerpc: Do not use op_model_cell under LPAR

[Benjamin Herrenschmidt]

On Tue, 2006-12-12 at 15:13 +1100, Michael Ellerman wrote:
> On Tue, 2006-12-12 at 13:02 +0900, Ishizaki Kou wrote:
> > op_model_cell supports native Cell processor. Under LPAR environment,
> > it cannot work.
> > 
> > Signed-off-by: Kou Ishizaki <kou.ishizaki.co.jp>
> > ---
> > 
> > Index: linux-powerpc-git/arch/powerpc/oprofile/common.c
> > diff -u linux-powerpc-git/arch/powerpc/oprofile/common.c:1.1.1.1 linux-powerpc-git/arch/powerpc/oprofile/common.c:1.2
> > --- linux-powerpc-git/arch/powerpc/oprofile/common.c:1.1.1.1	Wed Dec  6 08:25:43 2006
> > +++ linux-powerpc-git/arch/powerpc/oprofile/common.c	Wed Dec  6 08:43:15 2006
> > @@ -149,6 +149,8 @@
> >  #ifdef CONFIG_PPC64
> >  #ifdef CONFIG_PPC_CELL_NATIVE
> >  		case PPC_OPROFILE_CELL:
> > +			if (firmware_has_feature(FW_FEATURE_LPAR))
> > +				return -ENODEV;
> >  			model = &op_model_cell;
> >  			break;
> >  #endif
> 
> Is FW_FEATURE_LPAR is the right switch to use here, I'm not sure.

For now, it probably is as none of the LPAR env. for cell work with that
op_model.

Ben.

Re: [PATCH 12/15] powerpc: Do not use op_model_cell under LPAR

[Michael Ellerman]

[-- Attachment #1: Type: text/plain, Size: 1636 bytes --]

On Tue, 2006-12-12 at 15:53 +1100, Benjamin Herrenschmidt wrote:
> On Tue, 2006-12-12 at 15:13 +1100, Michael Ellerman wrote:
> > On Tue, 2006-12-12 at 13:02 +0900, Ishizaki Kou wrote:
> > > op_model_cell supports native Cell processor. Under LPAR environment,
> > > it cannot work.
> > > 
> > > Signed-off-by: Kou Ishizaki <kou.ishizaki.co.jp>
> > > ---
> > > 
> > > Index: linux-powerpc-git/arch/powerpc/oprofile/common.c
> > > diff -u linux-powerpc-git/arch/powerpc/oprofile/common.c:1.1.1.1 linux-powerpc-git/arch/powerpc/oprofile/common.c:1.2
> > > --- linux-powerpc-git/arch/powerpc/oprofile/common.c:1.1.1.1	Wed Dec  6 08:25:43 2006
> > > +++ linux-powerpc-git/arch/powerpc/oprofile/common.c	Wed Dec  6 08:43:15 2006
> > > @@ -149,6 +149,8 @@
> > >  #ifdef CONFIG_PPC64
> > >  #ifdef CONFIG_PPC_CELL_NATIVE
> > >  		case PPC_OPROFILE_CELL:
> > > +			if (firmware_has_feature(FW_FEATURE_LPAR))
> > > +				return -ENODEV;
> > >  			model = &op_model_cell;
> > >  			break;
> > >  #endif
> > 
> > Is FW_FEATURE_LPAR is the right switch to use here, I'm not sure.
> 
> For now, it probably is as none of the LPAR env. for cell work with that
> op_model.

Sure, there's probably twenty different things we could use at the
moment, I'm just wondering out loud if LPAR is the best thing to reduce
the chance we have to change it in the future.

cheers

-- 
Michael Ellerman
OzLabs, IBM Australia Development Lab

wwweb: http://michael.ellerman.id.au
phone: +61 2 6212 1183 (tie line 70 21183)

We do not inherit the earth from our ancestors,
we borrow it from our children. - S.M.A.R.T Person

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [PATCH 12/15] powerpc: Do not use op_model_cell under LPAR

[Ishizaki Kou]

> On Tue, 2006-12-12 at 15:53 +1100, Benjamin Herrenschmidt wrote:
> > On Tue, 2006-12-12 at 15:13 +1100, Michael Ellerman wrote:
> > > On Tue, 2006-12-12 at 13:02 +0900, Ishizaki Kou wrote:
> > > > op_model_cell supports native Cell processor. Under LPAR environment,
> > > > it cannot work.
> > > > 
> > > > Signed-off-by: Kou Ishizaki <kou.ishizaki.co.jp>
> > > > ---
> > > > 
> > > > Index: linux-powerpc-git/arch/powerpc/oprofile/common.c
> > > > diff -u linux-powerpc-git/arch/powerpc/oprofile/common.c:1.1.1.1 linux-powerpc-git/arch/powerpc/oprofile/common.c:1.2
> > > > --- linux-powerpc-git/arch/powerpc/oprofile/common.c:1.1.1.1	 Wed Dec  6 08:25:43 2006
> > > > +++ linux-powerpc-git/arch/powerpc/oprofile/common.c		 Wed Dec  6 08:43:15 2006
> > > > @@ -149,6 +149,8 @@
> > > >  #ifdef CONFIG_PPC64
> > > >  #ifdef CONFIG_PPC_CELL_NATIVE
> > > >			case PPC_OPROFILE_CELL:
> > > > +			if (firmware_has_feature(FW_FEATURE_LPAR))
> > > > +					return -ENODEV;
> > > >							model = &op_model_cell;
> > > >								break;
> > > >  #endif
> > > 
> > > Is FW_FEATURE_LPAR is the right switch to use here, I'm not sure.
> > 
> > For now, it probably is as none of the LPAR env. for cell work with that
> > op_model.
> 
> Sure, there's probably twenty different things we could use at the
> moment, I'm just wondering out loud if LPAR is the best thing to reduce
> the chance we have to change it in the future.

We will implement Cell LPAR support in the future. The code doesn't
exist now, and it tries to use oprofile_timer by returning -ENODEV
here.
				   
Thank you,
Kou Ishizaki
Toshiba

Re: [PATCH 12/15] powerpc: Do not use op_model_cell under LPAR

[Arnd Bergmann]

On Tuesday 12 December 2006 05:13, Michael Ellerman wrote:
>   On Tue, 2006-12-12 at 13:02 +0900, Ishizaki Kou wrote:
> > op_model_cell supports native Cell processor. Under LPAR environment,
> > it cannot work.
> > 
> > Signed-off-by: Kou Ishizaki <kou.ishizaki.co.jp>

Acked-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>

> 
> Is FW_FEATURE_LPAR is the right switch to use here, I'm not sure.

Yes, I think it is. The code accesses I/O memory areas that can not
work under any hypervisor, so FW_FEATURE_LPAR makes perfect sense
to me.

	Arnd <><

Re: [PATCH 12/15] powerpc: Do not use op_model_cell under LPAR

[Michael Ellerman]

[-- Attachment #1: Type: text/plain, Size: 1001 bytes --]

On Tue, 2006-12-12 at 20:55 +0100, Arnd Bergmann wrote:
> On Tuesday 12 December 2006 05:13, Michael Ellerman wrote:
> >   On Tue, 2006-12-12 at 13:02 +0900, Ishizaki Kou wrote:
> > > op_model_cell supports native Cell processor. Under LPAR environment,
> > > it cannot work.
> > > 
> > > Signed-off-by: Kou Ishizaki <kou.ishizaki.co.jp>
> 
> Acked-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>
> 
> > 
> > Is FW_FEATURE_LPAR is the right switch to use here, I'm not sure.
> 
> Yes, I think it is. The code accesses I/O memory areas that can not
> work under any hypervisor, so FW_FEATURE_LPAR makes perfect sense
> to me.

Yeah it's good enough.

But to be pedantic, P (LPAR) implies Q (HV), but !P does not imply !Q :)

cheers

-- 
Michael Ellerman
OzLabs, IBM Australia Development Lab

wwweb: http://michael.ellerman.id.au
phone: +61 2 6212 1183 (tie line 70 21183)

We do not inherit the earth from our ancestors,
we borrow it from our children. - S.M.A.R.T Person

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]