Re: linux-next: Tree for June 13 (XEN)

Re: linux-next: Tree for June 13 (XEN)

[Randy Dunlap]

next-20080613 on x86_32 has lots of xen build errors like this:

linux-next-20080613/arch/x86/xen/mmu.c: In function 'drop_mm_ref':
linux-next-20080613/arch/x86/xen/mmu.c:759: error: implicit declaration of function 'xen_smp_call_function_mask'
make[2]: *** [arch/x86/xen/mmu.o] Error 1


---
~Randy
'"Daemon' is an old piece of jargon from the UNIX operating system,
where it referred to a piece of low-level utility software, a
fundamental part of the operating system."

Re: linux-next: Tree for June 13 (XEN)

[Jeremy Fitzhardinge]

Randy Dunlap wrote:
> next-20080613 on x86_32 has lots of xen build errors like this:
>
> linux-next-20080613/arch/x86/xen/mmu.c: In function 'drop_mm_ref':
> linux-next-20080613/arch/x86/xen/mmu.c:759: error: implicit declaration of function 'xen_smp_call_function_mask'
> make[2]: *** [arch/x86/xen/mmu.o] Error 1
>
>   

Ooh, first time I've seen that.  Sounds like Jens' patches are missing 
the appropriate update there (though it's certainly had it in the past).

    J

Re: linux-next: Tree for June 13 (XEN)

[Jens Axboe]

On Fri, Jun 13 2008, Jeremy Fitzhardinge wrote:
> Randy Dunlap wrote:
> >next-20080613 on x86_32 has lots of xen build errors like this:
> >
> >linux-next-20080613/arch/x86/xen/mmu.c: In function 'drop_mm_ref':
> >linux-next-20080613/arch/x86/xen/mmu.c:759: error: implicit declaration of 
> >function 'xen_smp_call_function_mask'
> >make[2]: *** [arch/x86/xen/mmu.o] Error 1
> >
> >  
> 
> Ooh, first time I've seen that.  Sounds like Jens' patches are missing 
> the appropriate update there (though it's certainly had it in the past).

Hmm, will this work or do we need to force xen smp_ops for this one? I
wonder if this is new code and was missed, or what happened in this
case.

diff --git a/arch/x86/xen/mmu.c b/arch/x86/xen/mmu.c
index 3525ef5..8baef77 100644
--- a/arch/x86/xen/mmu.c
+++ b/arch/x86/xen/mmu.c
@@ -569,7 +569,7 @@ static void drop_mm_ref(struct mm_struct *mm)
 	}
 
 	if (!cpus_empty(mask))
-		xen_smp_call_function_mask(mask, drop_other_mm_ref, mm, 1);
+		smp_call_function_mask(mask, drop_other_mm_ref, mm, 1);
 }
 #else
 static void drop_mm_ref(struct mm_struct *mm)

-- 
Jens Axboe

Re: linux-next: Tree for June 13 (XEN)

[Randy Dunlap]

On Sat, 14 Jun 2008 22:31:10 +0200 Jens Axboe wrote:

> On Fri, Jun 13 2008, Jeremy Fitzhardinge wrote:
> > Randy Dunlap wrote:
> > >next-20080613 on x86_32 has lots of xen build errors like this:
> > >
> > >linux-next-20080613/arch/x86/xen/mmu.c: In function 'drop_mm_ref':
> > >linux-next-20080613/arch/x86/xen/mmu.c:759: error: implicit declaration of 
> > >function 'xen_smp_call_function_mask'
> > >make[2]: *** [arch/x86/xen/mmu.o] Error 1
> > >
> > >  
> > 
> > Ooh, first time I've seen that.  Sounds like Jens' patches are missing 
> > the appropriate update there (though it's certainly had it in the past).
> 
> Hmm, will this work or do we need to force xen smp_ops for this one? I
> wonder if this is new code and was missed, or what happened in this
> case.

Builds cleanly.  Thanks.

> diff --git a/arch/x86/xen/mmu.c b/arch/x86/xen/mmu.c
> index 3525ef5..8baef77 100644
> --- a/arch/x86/xen/mmu.c
> +++ b/arch/x86/xen/mmu.c
> @@ -569,7 +569,7 @@ static void drop_mm_ref(struct mm_struct *mm)
>  	}
>  
>  	if (!cpus_empty(mask))
> -		xen_smp_call_function_mask(mask, drop_other_mm_ref, mm, 1);
> +		smp_call_function_mask(mask, drop_other_mm_ref, mm, 1);
>  }
>  #else
>  static void drop_mm_ref(struct mm_struct *mm)


---
~Randy
'"Daemon' is an old piece of jargon from the UNIX operating system,
where it referred to a piece of low-level utility software, a
fundamental part of the operating system."

Re: linux-next: Tree for June 13 (XEN)

[Jeremy Fitzhardinge]

Jens Axboe wrote:
> On Fri, Jun 13 2008, Jeremy Fitzhardinge wrote:
>   
>> Randy Dunlap wrote:
>>     
>>> next-20080613 on x86_32 has lots of xen build errors like this:
>>>
>>> linux-next-20080613/arch/x86/xen/mmu.c: In function 'drop_mm_ref':
>>> linux-next-20080613/arch/x86/xen/mmu.c:759: error: implicit declaration of 
>>> function 'xen_smp_call_function_mask'
>>> make[2]: *** [arch/x86/xen/mmu.o] Error 1
>>>
>>>  
>>>       
>> Ooh, first time I've seen that.  Sounds like Jens' patches are missing 
>> the appropriate update there (though it's certainly had it in the past).
>>     
>
> Hmm, will this work or do we need to force xen smp_ops for this one? I
> wonder if this is new code and was missed, or what happened in this
> case.
>   

Yes, using smp_call_function_mask is perfectly OK.  The old code was 
just a micro-optimisation.  I'm pretty sure this chunk was in one of 
your patchsets (or perhaps I sent it to you at some point).

> diff --git a/arch/x86/xen/mmu.c b/arch/x86/xen/mmu.c
> index 3525ef5..8baef77 100644
> --- a/arch/x86/xen/mmu.c
> +++ b/arch/x86/xen/mmu.c
> @@ -569,7 +569,7 @@ static void drop_mm_ref(struct mm_struct *mm)
>  	}
>  
>  	if (!cpus_empty(mask))
> -		xen_smp_call_function_mask(mask, drop_other_mm_ref, mm, 1);
> +		smp_call_function_mask(mask, drop_other_mm_ref, mm, 1);
>  }
>  #else
>  static void drop_mm_ref(struct mm_struct *mm)
>
>   

Acked-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>

    J

Re: linux-next: Tree for June 13 (XEN)

[Jens Axboe]

On Sun, Jun 15 2008, Jeremy Fitzhardinge wrote:
> Jens Axboe wrote:
> >On Fri, Jun 13 2008, Jeremy Fitzhardinge wrote:
> >  
> >>Randy Dunlap wrote:
> >>    
> >>>next-20080613 on x86_32 has lots of xen build errors like this:
> >>>
> >>>linux-next-20080613/arch/x86/xen/mmu.c: In function 'drop_mm_ref':
> >>>linux-next-20080613/arch/x86/xen/mmu.c:759: error: implicit declaration 
> >>>of function 'xen_smp_call_function_mask'
> >>>make[2]: *** [arch/x86/xen/mmu.o] Error 1
> >>>
> >>> 
> >>>      
> >>Ooh, first time I've seen that.  Sounds like Jens' patches are missing 
> >>the appropriate update there (though it's certainly had it in the past).
> >>    
> >
> >Hmm, will this work or do we need to force xen smp_ops for this one? I
> >wonder if this is new code and was missed, or what happened in this
> >case.
> >  
> 
> Yes, using smp_call_function_mask is perfectly OK.  The old code was 
> just a micro-optimisation.  I'm pretty sure this chunk was in one of 
> your patchsets (or perhaps I sent it to you at some point).

Hmm yes, not sure myself to be honest, but I think you are right.

> >diff --git a/arch/x86/xen/mmu.c b/arch/x86/xen/mmu.c
> >index 3525ef5..8baef77 100644
> >--- a/arch/x86/xen/mmu.c
> >+++ b/arch/x86/xen/mmu.c
> >@@ -569,7 +569,7 @@ static void drop_mm_ref(struct mm_struct *mm)
> > 	}
> > 
> > 	if (!cpus_empty(mask))
> >-		xen_smp_call_function_mask(mask, drop_other_mm_ref, mm, 1);
> >+		smp_call_function_mask(mask, drop_other_mm_ref, mm, 1);
> > }
> > #else
> > static void drop_mm_ref(struct mm_struct *mm)
> >
> >  
> 
> Acked-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>

Thanks, I just folded it in with the existing patch to avoid breakage.
That one doesn't have an ack from you though, so if you have done a full
review of the x86 bits, I'd appreciate an ack on those from you :-)

-- 
Jens Axboe

Re: linux-next: Tree for June 13 (XEN)

[Jeremy Fitzhardinge]

Jens Axboe wrote:
> Thanks, I just folded it in with the existing patch to avoid breakage.
> That one doesn't have an ack from you though, so if you have done a full
> review of the x86 bits, I'd appreciate an ack on those from you :-)
>   

I've been testing it without problems, at least under Xen.

Acked-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>

    J