Mathias Krause 0b6b098efc padata: make the sequence counter an atomic_t
Using a spinlock to atomically increase a counter sounds wrong -- we've
atomic_t for this!

Also move 'seq_nr' to a different cache line than 'lock' to reduce cache
line trashing. This has the nice side effect of decreasing the size of
struct parallel_data from 192 to 128 bytes for a x86-64 build, e.g.
occupying only two instead of three cache lines.

Those changes results in a 5% performance increase on an IPsec test run
using pcrypt.

Btw. the seq_lock spinlock was never explicitly initialized -- one more
reason to get rid of it.

Signed-off-by: Mathias Krause <mathias.krause@secunet.com>
Acked-by: Steffen Klassert <steffen.klassert@secunet.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2013-10-30 12:02:58 +08:00
..
2013-08-26 14:09:03 -04:00
2013-09-06 13:21:16 -07:00
2013-09-04 16:26:56 -07:00
2013-08-26 18:40:56 -04:00
2013-08-27 01:44:40 +02:00
2013-07-10 23:41:18 +01:00
2013-07-23 16:01:28 -07:00
2013-07-15 11:25:00 +09:30
2013-08-26 20:15:23 +09:00
2013-08-22 22:13:54 -07:00
2013-08-02 11:34:30 +08:00
2013-08-23 10:22:29 +02:00
2013-07-18 13:05:23 -07:00
2013-08-22 20:30:15 -07:00
2013-08-09 10:49:00 +02:00
2013-07-26 16:19:48 -07:00
2013-07-16 22:00:14 -07:00
2013-07-09 10:33:30 -07:00
2013-08-12 17:54:05 +02:00
2013-08-12 15:27:01 +00:00
2013-08-12 15:27:01 +00:00
2013-09-06 13:30:06 -07:00
2013-09-03 10:07:40 -07:00
2013-07-09 12:09:43 -07:00
2013-08-13 17:57:48 -07:00
2013-07-24 15:12:53 -07:00
2013-07-10 18:11:34 -07:00