Sander reports following splat after netfilter nat bysrc table got
converted to rhashtable:
swapper/0: page allocation failure: order:3,
mode:0x2084020(GFP_ATOMIC|__GFP_COMP)
CPU: 0 PID: 0 Comm: swapper/0 Not tainted 4.8.0-rc1 [..]
[<ffffffff811633ed>] warn_alloc_failed+0xdd/0x140
[<ffffffff811638b1>] __alloc_pages_nodemask+0x3e1/0xcf0
[<ffffffff811a72ed>] alloc_pages_current+0x8d/0x110
[<ffffffff8117cb7f>] kmalloc_order+0x1f/0x70
[<ffffffff811aec19>] __kmalloc+0x129/0x140
[<ffffffff8146d561>] bucket_table_alloc+0xc1/0x1d0
[<ffffffff8146da1d>] rhashtable_insert_rehash+0x5d/0xe0
[<ffffffff819fcfff>] nf_nat_setup_info+0x2ef/0x400
The failure happens when allocating the spinlock array.
Even with GFP_KERNEL its unlikely for such a large allocation
to succeed.
Thomas Graf pointed me at inet_ehash_locks_alloc(), so in addition
to adding NOWARN for atomic allocations this also makes the bucket-array
sizing more conservative.
In commit 095dc8e0c3686 ("tcp: fix/cleanup inet_ehash_locks_alloc()"),
Eric Dumazet says: "Budget 2 cache lines per cpu worth of 'spinlocks'".
IOW, consider size needed by a single spinlock when determining
number of locks per cpu.
Currently, rhashtable just allocates 128 locks per cpu which gives
factor of 4 more than what inet hashtable uses with same number of
cpus.
For LOCKDEP, we now allocate a lot less locks than before (1 per cpu on
my test box) so we no longer need to pretend we only have two cpus.
Some sizes (64 byte L1 cache, 4 byte per spinlock, numbers in bytes):
cpus: 1 2 4 8 16 32 64
old: 1k 1k 4k 8k 16k 16k 16k
new: 128 256 512 1k 2k 4k 8k
With 72-byte spinlock (LOCKDEP):
cpus : 1 2 4 8 16 32 64
old: 9k 18k 18k 18k 18k 18k 18k
new: 72 144 288 575 ~1k ~2.3k ~4k
Reported-by: Sander Eikelenboom <li...@eikelenboom.it>
Suggested-by: Thomas Graf <tg...@suug.ch>
Signed-off-by: Florian Westphal <f...@strlen.de>
---
Alernatively we could lower BUCKET_LOCKS_PER_CPU to 32
and keep the CONFIG_PROVE_LOCKING ifdef around.
Any preference?
Thanks!
lib/rhashtable.c | 14 +++++++-------
1 file changed, 7 insertions(+), 7 deletions(-)
diff --git a/lib/rhashtable.c b/lib/rhashtable.c
index 5d845ff..92cf5a9 100644
--- a/lib/rhashtable.c
+++ b/lib/rhashtable.c
@@ -30,7 +30,8 @@
#define HASH_DEFAULT_SIZE 64UL
#define HASH_MIN_SIZE 4U
-#define BUCKET_LOCKS_PER_CPU 128UL
+#define BUCKET_LOCKS_PER_CPU max_t(unsigned int, \
+ 2 * L1_CACHE_BYTES / sizeof(spinlock_t),
1)
static u32 head_hashfn(struct rhashtable *ht,
const struct bucket_table *tbl,
@@ -63,14 +64,10 @@ EXPORT_SYMBOL_GPL(lockdep_rht_bucket_is_held);
static int alloc_bucket_locks(struct rhashtable *ht, struct bucket_table *tbl,
gfp_t gfp)
{
- unsigned int i, size;
-#if defined(CONFIG_PROVE_LOCKING)
- unsigned int nr_pcpus = 2;
-#else
unsigned int nr_pcpus = num_possible_cpus();
-#endif
+ unsigned int i, size;
- nr_pcpus = min_t(unsigned int, nr_pcpus, 32UL);
+ nr_pcpus = min_t(unsigned int, nr_pcpus, 64UL);
size = roundup_pow_of_two(nr_pcpus * ht->p.locks_mul);
/* Never allocate more than 0.5 locks per bucket */
@@ -83,6 +80,9 @@ static int alloc_bucket_locks(struct rhashtable *ht, struct
bucket_table *tbl,
tbl->locks = vmalloc(size * sizeof(spinlock_t));
else
#endif
+ if (gfp != GFP_KERNEL)
+ gfp |= __GFP_NOWARN | __GFP_NORETRY;
+
tbl->locks = kmalloc_array(size, sizeof(spinlock_t),
gfp);
if (!tbl->locks)
--
2.7.3
