[PATCH] x86_64 - UV fix for memory-only blades

[PATCH] x86_64 - UV fix for memory-only blades

[Jack Steiner]

Fix an endcase in handling UV blades that have memory
but no cpus. The calculation of the highest blade number
failed to handle these blades if they appear last in the
configuration.

Signed-off-by: Jack Steiner <steiner@sgi.com>

---
 arch/x86/kernel/apic/x2apic_uv_x.c |    1 +
 1 file changed, 1 insertion(+)

Index: linux/arch/x86/kernel/apic/x2apic_uv_x.c
===================================================================
--- linux.orig/arch/x86/kernel/apic/x2apic_uv_x.c	2009-05-15 12:00:59.000000000 -0500
+++ linux/arch/x86/kernel/apic/x2apic_uv_x.c	2009-05-15 14:38:16.000000000 -0500
@@ -632,6 +632,7 @@ void __init uv_system_init(void)
 		pnode = (paddr >> m_val) & pnode_mask;
 		blade = boot_pnode_to_blade(pnode);
 		uv_node_to_blade[nid] = blade;
+		max_pnode = max(pnode, max_pnode);
 	}
 
 	map_gru_high(max_pnode);

Re: [PATCH] x86_64 - UV fix for memory-only blades

[Yinghai Lu]

Jack Steiner wrote:
> Fix an endcase in handling UV blades that have memory
> but no cpus. The calculation of the highest blade number
> failed to handle these blades if they appear last in the
> configuration.
> 
> Signed-off-by: Jack Steiner <steiner@sgi.com>
> 
> ---
>  arch/x86/kernel/apic/x2apic_uv_x.c |    1 +
>  1 file changed, 1 insertion(+)
> 
> Index: linux/arch/x86/kernel/apic/x2apic_uv_x.c
> ===================================================================
> --- linux.orig/arch/x86/kernel/apic/x2apic_uv_x.c	2009-05-15 12:00:59.000000000 -0500
> +++ linux/arch/x86/kernel/apic/x2apic_uv_x.c	2009-05-15 14:38:16.000000000 -0500
> @@ -632,6 +632,7 @@ void __init uv_system_init(void)
>  		pnode = (paddr >> m_val) & pnode_mask;
>  		blade = boot_pnode_to_blade(pnode);
>  		uv_node_to_blade[nid] = blade;
> +		max_pnode = max(pnode, max_pnode);
>  	}
>  
>  	map_gru_high(max_pnode);

node_possible_map should already include nodes with memory and without memory already.

maybe you could skip max_pnode calculating in the loop.

YH

Re: [PATCH] x86_64 - UV fix for memory-only blades

[Jack Steiner]

On Tue, May 19, 2009 at 08:13:05AM -0700, Yinghai Lu wrote:
> Jack Steiner wrote:
> > Fix an endcase in handling UV blades that have memory
> > but no cpus. The calculation of the highest blade number
> > failed to handle these blades if they appear last in the
> > configuration.
> > 
> > Signed-off-by: Jack Steiner <steiner@sgi.com>
> > 
> > ---
> >  arch/x86/kernel/apic/x2apic_uv_x.c |    1 +
> >  1 file changed, 1 insertion(+)
> > 
> > Index: linux/arch/x86/kernel/apic/x2apic_uv_x.c
> > ===================================================================
> > --- linux.orig/arch/x86/kernel/apic/x2apic_uv_x.c	2009-05-15 12:00:59.000000000 -0500
> > +++ linux/arch/x86/kernel/apic/x2apic_uv_x.c	2009-05-15 14:38:16.000000000 -0500
> > @@ -632,6 +632,7 @@ void __init uv_system_init(void)
> >  		pnode = (paddr >> m_val) & pnode_mask;
> >  		blade = boot_pnode_to_blade(pnode);
> >  		uv_node_to_blade[nid] = blade;
> > +		max_pnode = max(pnode, max_pnode);
> >  	}
> >  
> >  	map_gru_high(max_pnode);
> 
> node_possible_map should already include nodes with memory and without memory already.
> 
> maybe you could skip max_pnode calculating in the loop.

Terms related to nodes are grossly overloaded.

"pnode" is only loosely related to a node. It is a physical node number that
is a hardware-thing. For example, it is possible to have a 1 node system with a max_pnode
of 128. 

pnodes are used only in the UV infrastructure - they have no generic meaning nor would they apply
to any other platform.


--- jack