| /openbmc/linux/drivers/net/ethernet/broadcom/bnx2x/ |
| H A D | bnx2x_cmn.h | diff e52fcb2462ac484e6dd6e68869536609f0216938 Mon Nov 14 00:05:34 CST 2011 Eric Dumazet <eric.dumazet@gmail.com> bnx2x: uses build_skb() in receive path
bnx2x uses following formula to compute its rx_buf_sz :
dev->mtu + 2*L1_CACHE_BYTES + 14 + 8 + 8 + 2
Then core network adds NET_SKB_PAD and SKB_DATA_ALIGN(sizeof(struct
skb_shared_info))
Final allocated size for skb head on x86_64 (L1_CACHE_BYTES = 64,
MTU=1500) : 2112 bytes : SLUB/SLAB round this to 4096 bytes.
Since skb truesize is then bigger than SK_MEM_QUANTUM, we have lot of
false sharing because of mem_reclaim in UDP stack.
One possible way to half truesize is to reduce the need by 64 bytes
(2112 -> 2048 bytes)
Instead of allocating a full cache line at the end of packet for
alignment, we can use the fact that skb_shared_info sits at the end of
skb->head, and we can use this room, if we convert bnx2x to new
build_skb() infrastructure.
skb_shared_info will be initialized after hardware finished its
transfert, so we can eventually overwrite the final padding.
Using build_skb() also reduces cache line misses in the driver, since we
use cache hot skb instead of cold ones. Number of in-flight sk_buff
structures is lower, they are recycled while still hot.
Performance results :
(820.000 pps on a rx UDP monothread benchmark, instead of 720.000 pps)
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
CC: Eilon Greenstein <eilong@broadcom.com>
CC: Ben Hutchings <bhutchings@solarflare.com>
CC: Tom Herbert <therbert@google.com>
CC: Jamal Hadi Salim <hadi@mojatatu.com>
CC: Stephen Hemminger <shemminger@vyatta.com>
CC: Thomas Graf <tgraf@infradead.org>
CC: Herbert Xu <herbert@gondor.apana.org.au>
CC: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Acked-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
| H A D | bnx2x_cmn.c | diff 036d2df9b3167598a9c9f1c13d9039f7e6cb0083 Mon Dec 12 17:40:53 CST 2011 Dmitry Kravkov <dmitry@broadcom.com> bnx2x: properly update skb when mtu > 1500
Since commit e52fcb2462ac484e6dd6e68869536609f0216938 newly allocated
skb for small packets are not updated properly and dropped by stack.
Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
diff e52fcb2462ac484e6dd6e68869536609f0216938 Mon Nov 14 00:05:34 CST 2011 Eric Dumazet <eric.dumazet@gmail.com> bnx2x: uses build_skb() in receive path
bnx2x uses following formula to compute its rx_buf_sz :
dev->mtu + 2*L1_CACHE_BYTES + 14 + 8 + 8 + 2
Then core network adds NET_SKB_PAD and SKB_DATA_ALIGN(sizeof(struct
skb_shared_info))
Final allocated size for skb head on x86_64 (L1_CACHE_BYTES = 64,
MTU=1500) : 2112 bytes : SLUB/SLAB round this to 4096 bytes.
Since skb truesize is then bigger than SK_MEM_QUANTUM, we have lot of
false sharing because of mem_reclaim in UDP stack.
One possible way to half truesize is to reduce the need by 64 bytes
(2112 -> 2048 bytes)
Instead of allocating a full cache line at the end of packet for
alignment, we can use the fact that skb_shared_info sits at the end of
skb->head, and we can use this room, if we convert bnx2x to new
build_skb() infrastructure.
skb_shared_info will be initialized after hardware finished its
transfert, so we can eventually overwrite the final padding.
Using build_skb() also reduces cache line misses in the driver, since we
use cache hot skb instead of cold ones. Number of in-flight sk_buff
structures is lower, they are recycled while still hot.
Performance results :
(820.000 pps on a rx UDP monothread benchmark, instead of 720.000 pps)
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
CC: Eilon Greenstein <eilong@broadcom.com>
CC: Ben Hutchings <bhutchings@solarflare.com>
CC: Tom Herbert <therbert@google.com>
CC: Jamal Hadi Salim <hadi@mojatatu.com>
CC: Stephen Hemminger <shemminger@vyatta.com>
CC: Thomas Graf <tgraf@infradead.org>
CC: Herbert Xu <herbert@gondor.apana.org.au>
CC: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Acked-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
| H A D | bnx2x_ethtool.c | diff e52fcb2462ac484e6dd6e68869536609f0216938 Mon Nov 14 00:05:34 CST 2011 Eric Dumazet <eric.dumazet@gmail.com> bnx2x: uses build_skb() in receive path
bnx2x uses following formula to compute its rx_buf_sz :
dev->mtu + 2*L1_CACHE_BYTES + 14 + 8 + 8 + 2
Then core network adds NET_SKB_PAD and SKB_DATA_ALIGN(sizeof(struct
skb_shared_info))
Final allocated size for skb head on x86_64 (L1_CACHE_BYTES = 64,
MTU=1500) : 2112 bytes : SLUB/SLAB round this to 4096 bytes.
Since skb truesize is then bigger than SK_MEM_QUANTUM, we have lot of
false sharing because of mem_reclaim in UDP stack.
One possible way to half truesize is to reduce the need by 64 bytes
(2112 -> 2048 bytes)
Instead of allocating a full cache line at the end of packet for
alignment, we can use the fact that skb_shared_info sits at the end of
skb->head, and we can use this room, if we convert bnx2x to new
build_skb() infrastructure.
skb_shared_info will be initialized after hardware finished its
transfert, so we can eventually overwrite the final padding.
Using build_skb() also reduces cache line misses in the driver, since we
use cache hot skb instead of cold ones. Number of in-flight sk_buff
structures is lower, they are recycled while still hot.
Performance results :
(820.000 pps on a rx UDP monothread benchmark, instead of 720.000 pps)
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
CC: Eilon Greenstein <eilong@broadcom.com>
CC: Ben Hutchings <bhutchings@solarflare.com>
CC: Tom Herbert <therbert@google.com>
CC: Jamal Hadi Salim <hadi@mojatatu.com>
CC: Stephen Hemminger <shemminger@vyatta.com>
CC: Thomas Graf <tgraf@infradead.org>
CC: Herbert Xu <herbert@gondor.apana.org.au>
CC: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Acked-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
| H A D | bnx2x.h | diff e52fcb2462ac484e6dd6e68869536609f0216938 Mon Nov 14 00:05:34 CST 2011 Eric Dumazet <eric.dumazet@gmail.com> bnx2x: uses build_skb() in receive path
bnx2x uses following formula to compute its rx_buf_sz :
dev->mtu + 2*L1_CACHE_BYTES + 14 + 8 + 8 + 2
Then core network adds NET_SKB_PAD and SKB_DATA_ALIGN(sizeof(struct
skb_shared_info))
Final allocated size for skb head on x86_64 (L1_CACHE_BYTES = 64,
MTU=1500) : 2112 bytes : SLUB/SLAB round this to 4096 bytes.
Since skb truesize is then bigger than SK_MEM_QUANTUM, we have lot of
false sharing because of mem_reclaim in UDP stack.
One possible way to half truesize is to reduce the need by 64 bytes
(2112 -> 2048 bytes)
Instead of allocating a full cache line at the end of packet for
alignment, we can use the fact that skb_shared_info sits at the end of
skb->head, and we can use this room, if we convert bnx2x to new
build_skb() infrastructure.
skb_shared_info will be initialized after hardware finished its
transfert, so we can eventually overwrite the final padding.
Using build_skb() also reduces cache line misses in the driver, since we
use cache hot skb instead of cold ones. Number of in-flight sk_buff
structures is lower, they are recycled while still hot.
Performance results :
(820.000 pps on a rx UDP monothread benchmark, instead of 720.000 pps)
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
CC: Eilon Greenstein <eilong@broadcom.com>
CC: Ben Hutchings <bhutchings@solarflare.com>
CC: Tom Herbert <therbert@google.com>
CC: Jamal Hadi Salim <hadi@mojatatu.com>
CC: Stephen Hemminger <shemminger@vyatta.com>
CC: Thomas Graf <tgraf@infradead.org>
CC: Herbert Xu <herbert@gondor.apana.org.au>
CC: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Acked-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
| H A D | bnx2x_main.c | diff e52fcb2462ac484e6dd6e68869536609f0216938 Mon Nov 14 00:05:34 CST 2011 Eric Dumazet <eric.dumazet@gmail.com> bnx2x: uses build_skb() in receive path
bnx2x uses following formula to compute its rx_buf_sz :
dev->mtu + 2*L1_CACHE_BYTES + 14 + 8 + 8 + 2
Then core network adds NET_SKB_PAD and SKB_DATA_ALIGN(sizeof(struct
skb_shared_info))
Final allocated size for skb head on x86_64 (L1_CACHE_BYTES = 64,
MTU=1500) : 2112 bytes : SLUB/SLAB round this to 4096 bytes.
Since skb truesize is then bigger than SK_MEM_QUANTUM, we have lot of
false sharing because of mem_reclaim in UDP stack.
One possible way to half truesize is to reduce the need by 64 bytes
(2112 -> 2048 bytes)
Instead of allocating a full cache line at the end of packet for
alignment, we can use the fact that skb_shared_info sits at the end of
skb->head, and we can use this room, if we convert bnx2x to new
build_skb() infrastructure.
skb_shared_info will be initialized after hardware finished its
transfert, so we can eventually overwrite the final padding.
Using build_skb() also reduces cache line misses in the driver, since we
use cache hot skb instead of cold ones. Number of in-flight sk_buff
structures is lower, they are recycled while still hot.
Performance results :
(820.000 pps on a rx UDP monothread benchmark, instead of 720.000 pps)
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
CC: Eilon Greenstein <eilong@broadcom.com>
CC: Ben Hutchings <bhutchings@solarflare.com>
CC: Tom Herbert <therbert@google.com>
CC: Jamal Hadi Salim <hadi@mojatatu.com>
CC: Stephen Hemminger <shemminger@vyatta.com>
CC: Thomas Graf <tgraf@infradead.org>
CC: Herbert Xu <herbert@gondor.apana.org.au>
CC: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Acked-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
|