Re: [patch] nodemask: make NODEMASK_ALLOC more general

[David Rientjes <rientjes@google.com>]

On Fri, 2 Oct 2009, Christoph Lameter wrote:

> > NODEMASK_ALLOC(x, m) assumes x is a type of struct, which is unnecessary.
> > It's perfectly reasonable to use this macro to allocate a nodemask_t,
> > which is anonymous, either dynamically or on the stack depending on
> > NODES_SHIFT.
> 
> There is currently only one user of NODEMASK_ALLOC which is
> NODEMASK_SCRATCH.
> 

That changes with Lee's patchset for mempolicy hugepage allocations and 
freeing, he'll be using it in generic hugetlb code.

> Can we generalize the functionality here? The macro is basically choosing
> between a slab allocation or a stack allocation depending on the
> configured system size.
> 
> NUMA_COND__ALLOC(<type>, <min numa nodes for not using stack>,
> <variablename>)
> 
> or so?
> 

I assume we could, although it would be slightly messy because we'd be 
coding a stack allocation in a macro when comparing the passed value 
against CONFIG_NODES_SHIFT.

> Its likely that one way want to allocate other structures on the stack
> that may get too big if large systems need to be supported.
> 

I don't think we currently have any examples of that other than 
nodemask_t.  We allocate arrays of length MAX_NUMNODES quite often for 
things like node_to_cpumask_map, struct bootnode, etc, but no longer on 
the stack even in NUMA emulation.  I'd be interested to see any 
non-nodemask use cases.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>