On Thu, 2015-12-03 at 20:51 +0800, Herbert Xu wrote:

On Mon, Nov 30, 2015 at 06:18:59PM +0800, Herbert Xu wrote:

quoted

OK that's better.  I think I see the problem.  The test in
rhashtable_insert_rehash is racy and if two threads both try
to grow the table one of them may be tricked into doing a rehash
instead.

I'm working on a fix.

While the EBUSY errors are gone for me, I can still see plenty
of ENOMEM errors.  In fact it turns out that the reason is quite
understandable.  When you pound the rhashtable hard so that it
doesn't actually get a chance to grow the table in process context,
then the table will only grow with GFP_ATOMIC allocations.

For me this starts failing regularly at around 2^19 entries, which
requires about 1024 contiguous pages if I'm not mistaken.

Well, it will fail before this point if memory is fragmented.

Anyway, __vmalloc() can be used with GFP_ATOMIC, have you tried this ?

diff --git a/lib/rhashtable.c b/lib/rhashtable.c
index a54ff8949f91..9ef5d74963b2 100644
--- a/lib/rhashtable.c
+++ b/lib/rhashtable.c

@@ -120,8 +120,9 @@ static struct bucket_table *bucket_table_alloc(struct rhashtable *ht,
 	if (size <= (PAGE_SIZE << PAGE_ALLOC_COSTLY_ORDER) ||
 	    gfp != GFP_KERNEL)
 		tbl = kzalloc(size, gfp | __GFP_NOWARN | __GFP_NORETRY);
-	if (tbl == NULL && gfp == GFP_KERNEL)
-		tbl = vzalloc(size);
+	if (tbl == NULL)
+		tbl = __vmalloc(size, gfp | __GFP_HIGHMEM | __GFP_ZERO,
+				PAGE_KERNEL);
 	if (tbl == NULL)
 		return NULL;