The patch titled
gfp.h: GFP_THISNODE can go to other nodes if some are unpopulated
has been added to the -mm tree. Its filename is
gfph-gfp_thisnode-can-go-to-other-nodes-if-some-are-unpopulated.patch
*** Remember to use Documentation/SubmitChecklist when testing your code ***
See http://www.zip.com.au/~akpm/linux/patches/stuff/added-to-mm.txt to find
out what to do about this
------------------------------------------------------
Subject: gfp.h: GFP_THISNODE can go to other nodes if some are unpopulated
From: Nishanth Aravamudan <nacc@xxxxxxxxxx>
While testing my sysfs per-node hugepage allocator
(http://marc.info/?l=linux-mm&m=117935849517122&w=2), I found that an
alloc_pages_node(nid, GFP_THISNODE) request would sometimes return a struct
page such that page_to_nid(page) != nid. This was because, on that
particular machine, nodes 0 and 1 are populated and nodes 2 and 3 are not.
When a page is requested get_page_from_freelist() relies on
zonelist->zones[0]->zone_pgdat indicating when THISNODE stops. But,
because, say, node 2 has no memory, the first zone_pgdat in the fallback
list points to a different node. Add a comment indicating that THISNODE
may not return pages on THISNODE if the node is unpopulated.
Am working on testing Lee/Anton's patch to add a node_populated_mask and
use that in the hugepage allocator path. But I think this may be a problem
anywhere THISNODE is used and memory is expected to come from the requested
node and nowhere else.
Reworked the comment based on feedback from Christoph Lameter.
Signed-off-by: Nishanth Aravamudan <nacc@xxxxxxxxxx>
Acked-by: Christoph Lameter <clameter@xxxxxxx>
Cc: David Rientjes <rientjes@xxxxxxxxxx>
Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
---
include/linux/gfp.h | 6 ++++++
1 file changed, 6 insertions(+)
diff -puN include/linux/gfp.h~gfph-gfp_thisnode-can-go-to-other-nodes-if-some-are-unpopulated include/linux/gfp.h
--- a/include/linux/gfp.h~gfph-gfp_thisnode-can-go-to-other-nodes-if-some-are-unpopulated
+++ a/include/linux/gfp.h
@@ -67,6 +67,12 @@ struct vm_area_struct;
__GFP_HIGHMEM)
#ifdef CONFIG_NUMA
+/*
+ * NOTE: GFP_THISNODE allocates from the first available pgdat (== node
+ * structure) from the zonelist of the requested node. The first pgdat
+ * may be the pgdat of another node if the requested node has no memory
+ * on its own.
+ */
#define GFP_THISNODE (__GFP_THISNODE | __GFP_NOWARN | __GFP_NORETRY)
#else
#define GFP_THISNODE ((__force gfp_t)0)
_
Patches currently in -mm which might be from nacc@xxxxxxxxxx are
hugetlb-remove-unnecessary-nid-initialization.patch
gfph-gfp_thisnode-can-go-to-other-nodes-if-some-are-unpopulated.patch
-
To unsubscribe from this list: send the line "unsubscribe mm-commits" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
