On Mon, Feb 10, 2014 at 12:27:45PM -0500, Luiz Capitulino wrote: > From: Luiz capitulino <lcapitulino@xxxxxxxxxx> > > If an allocation from the node specified by the nid argument fails, > memblock_virt_alloc_internal() automatically tries to allocate memory > from other nodes. > > This is fine is the caller don't care which node is going to allocate > the memory. However, there are cases where the caller wants memory to > be allocated from the specified node only. If that's not possible, then > memblock_virt_alloc_internal() should just fail. > > This commit adds a new flags argument to memblock_virt_alloc_internal() > where the caller can control this behavior. > > Signed-off-by: Luiz capitulino <lcapitulino@xxxxxxxxxx> > --- > mm/memblock.c | 10 +++++++--- > 1 file changed, 7 insertions(+), 3 deletions(-) > > diff --git a/mm/memblock.c b/mm/memblock.c > index 39a31e7..b0c7b2e 100644 > --- a/mm/memblock.c > +++ b/mm/memblock.c > @@ -1028,6 +1028,8 @@ phys_addr_t __init memblock_alloc_try_nid(phys_addr_t size, phys_addr_t align, i > return memblock_alloc_base(size, align, MEMBLOCK_ALLOC_ACCESSIBLE); > } > > +#define ALLOC_SPECIFIED_NODE_ONLY 0x1 > + It's not a perfect fit but you could use gfp_t and GFP_THISNODE. The meaning of the flag is recognised and while you are not using it with a page allocator, we already use GFP flags with the slab allocator without confusion. -- Mel Gorman SUSE Labs -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@xxxxxxxxx. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>