/openbmc/linux/include/net/netfilter/ |
H A D | xt_rateest.h | 1c0d32fd Sun Dec 04 11:48:16 CST 2016 Eric Dumazet <edumazet@google.com> net_sched: gen_estimator: complete rewrite of rate estimators
1) Old code was hard to maintain, due to complex lock chains. (We probably will be able to remove some kfree_rcu() in callers)
2) Using a single timer to update all estimators does not scale.
3) Code was buggy on 32bit kernel (WRITE_ONCE() on 64bit quantity is not supposed to work well)
In this rewrite :
- I removed the RB tree that had to be scanned in gen_estimator_active(). qdisc dumps should be much faster.
- Each estimator has its own timer.
- Estimations are maintained in net_rate_estimator structure, instead of dirtying the qdisc. Minor, but part of the simplification.
- Reading the estimator uses RCU and a seqcount to provide proper support for 32bit kernels.
- We reduce memory need when estimators are not used, since we store a pointer, instead of the bytes/packets counters.
- xt_rateest_mt() no longer has to grab a spinlock. (In the future, xt_rateest_tg() could be switched to per cpu counters)
Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net> 1c0d32fd Sun Dec 04 11:48:16 CST 2016 Eric Dumazet <edumazet@google.com> net_sched: gen_estimator: complete rewrite of rate estimators 1) Old code was hard to maintain, due to complex lock chains. (We probably will be able to remove some kfree_rcu() in callers) 2) Using a single timer to update all estimators does not scale. 3) Code was buggy on 32bit kernel (WRITE_ONCE() on 64bit quantity is not supposed to work well) In this rewrite : - I removed the RB tree that had to be scanned in gen_estimator_active(). qdisc dumps should be much faster. - Each estimator has its own timer. - Estimations are maintained in net_rate_estimator structure, instead of dirtying the qdisc. Minor, but part of the simplification. - Reading the estimator uses RCU and a seqcount to provide proper support for 32bit kernels. - We reduce memory need when estimators are not used, since we store a pointer, instead of the bytes/packets counters. - xt_rateest_mt() no longer has to grab a spinlock. (In the future, xt_rateest_tg() could be switched to per cpu counters) Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
|
/openbmc/linux/net/netfilter/ |
H A D | xt_rateest.c | 1c0d32fd Sun Dec 04 11:48:16 CST 2016 Eric Dumazet <edumazet@google.com> net_sched: gen_estimator: complete rewrite of rate estimators
1) Old code was hard to maintain, due to complex lock chains. (We probably will be able to remove some kfree_rcu() in callers)
2) Using a single timer to update all estimators does not scale.
3) Code was buggy on 32bit kernel (WRITE_ONCE() on 64bit quantity is not supposed to work well)
In this rewrite :
- I removed the RB tree that had to be scanned in gen_estimator_active(). qdisc dumps should be much faster.
- Each estimator has its own timer.
- Estimations are maintained in net_rate_estimator structure, instead of dirtying the qdisc. Minor, but part of the simplification.
- Reading the estimator uses RCU and a seqcount to provide proper support for 32bit kernels.
- We reduce memory need when estimators are not used, since we store a pointer, instead of the bytes/packets counters.
- xt_rateest_mt() no longer has to grab a spinlock. (In the future, xt_rateest_tg() could be switched to per cpu counters)
Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net> 1c0d32fd Sun Dec 04 11:48:16 CST 2016 Eric Dumazet <edumazet@google.com> net_sched: gen_estimator: complete rewrite of rate estimators 1) Old code was hard to maintain, due to complex lock chains. (We probably will be able to remove some kfree_rcu() in callers) 2) Using a single timer to update all estimators does not scale. 3) Code was buggy on 32bit kernel (WRITE_ONCE() on 64bit quantity is not supposed to work well) In this rewrite : - I removed the RB tree that had to be scanned in gen_estimator_active(). qdisc dumps should be much faster. - Each estimator has its own timer. - Estimations are maintained in net_rate_estimator structure, instead of dirtying the qdisc. Minor, but part of the simplification. - Reading the estimator uses RCU and a seqcount to provide proper support for 32bit kernels. - We reduce memory need when estimators are not used, since we store a pointer, instead of the bytes/packets counters. - xt_rateest_mt() no longer has to grab a spinlock. (In the future, xt_rateest_tg() could be switched to per cpu counters) Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
|
H A D | xt_RATEEST.c | 1c0d32fd Sun Dec 04 11:48:16 CST 2016 Eric Dumazet <edumazet@google.com> net_sched: gen_estimator: complete rewrite of rate estimators
1) Old code was hard to maintain, due to complex lock chains. (We probably will be able to remove some kfree_rcu() in callers)
2) Using a single timer to update all estimators does not scale.
3) Code was buggy on 32bit kernel (WRITE_ONCE() on 64bit quantity is not supposed to work well)
In this rewrite :
- I removed the RB tree that had to be scanned in gen_estimator_active(). qdisc dumps should be much faster.
- Each estimator has its own timer.
- Estimations are maintained in net_rate_estimator structure, instead of dirtying the qdisc. Minor, but part of the simplification.
- Reading the estimator uses RCU and a seqcount to provide proper support for 32bit kernels.
- We reduce memory need when estimators are not used, since we store a pointer, instead of the bytes/packets counters.
- xt_rateest_mt() no longer has to grab a spinlock. (In the future, xt_rateest_tg() could be switched to per cpu counters)
Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net> 1c0d32fd Sun Dec 04 11:48:16 CST 2016 Eric Dumazet <edumazet@google.com> net_sched: gen_estimator: complete rewrite of rate estimators 1) Old code was hard to maintain, due to complex lock chains. (We probably will be able to remove some kfree_rcu() in callers) 2) Using a single timer to update all estimators does not scale. 3) Code was buggy on 32bit kernel (WRITE_ONCE() on 64bit quantity is not supposed to work well) In this rewrite : - I removed the RB tree that had to be scanned in gen_estimator_active(). qdisc dumps should be much faster. - Each estimator has its own timer. - Estimations are maintained in net_rate_estimator structure, instead of dirtying the qdisc. Minor, but part of the simplification. - Reading the estimator uses RCU and a seqcount to provide proper support for 32bit kernels. - We reduce memory need when estimators are not used, since we store a pointer, instead of the bytes/packets counters. - xt_rateest_mt() no longer has to grab a spinlock. (In the future, xt_rateest_tg() could be switched to per cpu counters) Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
|
/openbmc/linux/include/net/ |
H A D | gen_stats.h | 1c0d32fd Sun Dec 04 11:48:16 CST 2016 Eric Dumazet <edumazet@google.com> net_sched: gen_estimator: complete rewrite of rate estimators
1) Old code was hard to maintain, due to complex lock chains. (We probably will be able to remove some kfree_rcu() in callers)
2) Using a single timer to update all estimators does not scale.
3) Code was buggy on 32bit kernel (WRITE_ONCE() on 64bit quantity is not supposed to work well)
In this rewrite :
- I removed the RB tree that had to be scanned in gen_estimator_active(). qdisc dumps should be much faster.
- Each estimator has its own timer.
- Estimations are maintained in net_rate_estimator structure, instead of dirtying the qdisc. Minor, but part of the simplification.
- Reading the estimator uses RCU and a seqcount to provide proper support for 32bit kernels.
- We reduce memory need when estimators are not used, since we store a pointer, instead of the bytes/packets counters.
- xt_rateest_mt() no longer has to grab a spinlock. (In the future, xt_rateest_tg() could be switched to per cpu counters)
Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net> 1c0d32fd Sun Dec 04 11:48:16 CST 2016 Eric Dumazet <edumazet@google.com> net_sched: gen_estimator: complete rewrite of rate estimators 1) Old code was hard to maintain, due to complex lock chains. (We probably will be able to remove some kfree_rcu() in callers) 2) Using a single timer to update all estimators does not scale. 3) Code was buggy on 32bit kernel (WRITE_ONCE() on 64bit quantity is not supposed to work well) In this rewrite : - I removed the RB tree that had to be scanned in gen_estimator_active(). qdisc dumps should be much faster. - Each estimator has its own timer. - Estimations are maintained in net_rate_estimator structure, instead of dirtying the qdisc. Minor, but part of the simplification. - Reading the estimator uses RCU and a seqcount to provide proper support for 32bit kernels. - We reduce memory need when estimators are not used, since we store a pointer, instead of the bytes/packets counters. - xt_rateest_mt() no longer has to grab a spinlock. (In the future, xt_rateest_tg() could be switched to per cpu counters) Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
|
H A D | act_api.h | 1c0d32fd Sun Dec 04 11:48:16 CST 2016 Eric Dumazet <edumazet@google.com> net_sched: gen_estimator: complete rewrite of rate estimators
1) Old code was hard to maintain, due to complex lock chains. (We probably will be able to remove some kfree_rcu() in callers)
2) Using a single timer to update all estimators does not scale.
3) Code was buggy on 32bit kernel (WRITE_ONCE() on 64bit quantity is not supposed to work well)
In this rewrite :
- I removed the RB tree that had to be scanned in gen_estimator_active(). qdisc dumps should be much faster.
- Each estimator has its own timer.
- Estimations are maintained in net_rate_estimator structure, instead of dirtying the qdisc. Minor, but part of the simplification.
- Reading the estimator uses RCU and a seqcount to provide proper support for 32bit kernels.
- We reduce memory need when estimators are not used, since we store a pointer, instead of the bytes/packets counters.
- xt_rateest_mt() no longer has to grab a spinlock. (In the future, xt_rateest_tg() could be switched to per cpu counters)
Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net> 1c0d32fd Sun Dec 04 11:48:16 CST 2016 Eric Dumazet <edumazet@google.com> net_sched: gen_estimator: complete rewrite of rate estimators 1) Old code was hard to maintain, due to complex lock chains. (We probably will be able to remove some kfree_rcu() in callers) 2) Using a single timer to update all estimators does not scale. 3) Code was buggy on 32bit kernel (WRITE_ONCE() on 64bit quantity is not supposed to work well) In this rewrite : - I removed the RB tree that had to be scanned in gen_estimator_active(). qdisc dumps should be much faster. - Each estimator has its own timer. - Estimations are maintained in net_rate_estimator structure, instead of dirtying the qdisc. Minor, but part of the simplification. - Reading the estimator uses RCU and a seqcount to provide proper support for 32bit kernels. - We reduce memory need when estimators are not used, since we store a pointer, instead of the bytes/packets counters. - xt_rateest_mt() no longer has to grab a spinlock. (In the future, xt_rateest_tg() could be switched to per cpu counters) Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
|
H A D | sch_generic.h | 1c0d32fd Sun Dec 04 11:48:16 CST 2016 Eric Dumazet <edumazet@google.com> net_sched: gen_estimator: complete rewrite of rate estimators
1) Old code was hard to maintain, due to complex lock chains. (We probably will be able to remove some kfree_rcu() in callers)
2) Using a single timer to update all estimators does not scale.
3) Code was buggy on 32bit kernel (WRITE_ONCE() on 64bit quantity is not supposed to work well)
In this rewrite :
- I removed the RB tree that had to be scanned in gen_estimator_active(). qdisc dumps should be much faster.
- Each estimator has its own timer.
- Estimations are maintained in net_rate_estimator structure, instead of dirtying the qdisc. Minor, but part of the simplification.
- Reading the estimator uses RCU and a seqcount to provide proper support for 32bit kernels.
- We reduce memory need when estimators are not used, since we store a pointer, instead of the bytes/packets counters.
- xt_rateest_mt() no longer has to grab a spinlock. (In the future, xt_rateest_tg() could be switched to per cpu counters)
Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net> 1c0d32fd Sun Dec 04 11:48:16 CST 2016 Eric Dumazet <edumazet@google.com> net_sched: gen_estimator: complete rewrite of rate estimators 1) Old code was hard to maintain, due to complex lock chains. (We probably will be able to remove some kfree_rcu() in callers) 2) Using a single timer to update all estimators does not scale. 3) Code was buggy on 32bit kernel (WRITE_ONCE() on 64bit quantity is not supposed to work well) In this rewrite : - I removed the RB tree that had to be scanned in gen_estimator_active(). qdisc dumps should be much faster. - Each estimator has its own timer. - Estimations are maintained in net_rate_estimator structure, instead of dirtying the qdisc. Minor, but part of the simplification. - Reading the estimator uses RCU and a seqcount to provide proper support for 32bit kernels. - We reduce memory need when estimators are not used, since we store a pointer, instead of the bytes/packets counters. - xt_rateest_mt() no longer has to grab a spinlock. (In the future, xt_rateest_tg() could be switched to per cpu counters) Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
|
/openbmc/linux/net/core/ |
H A D | gen_stats.c | 1c0d32fd Sun Dec 04 11:48:16 CST 2016 Eric Dumazet <edumazet@google.com> net_sched: gen_estimator: complete rewrite of rate estimators
1) Old code was hard to maintain, due to complex lock chains. (We probably will be able to remove some kfree_rcu() in callers)
2) Using a single timer to update all estimators does not scale.
3) Code was buggy on 32bit kernel (WRITE_ONCE() on 64bit quantity is not supposed to work well)
In this rewrite :
- I removed the RB tree that had to be scanned in gen_estimator_active(). qdisc dumps should be much faster.
- Each estimator has its own timer.
- Estimations are maintained in net_rate_estimator structure, instead of dirtying the qdisc. Minor, but part of the simplification.
- Reading the estimator uses RCU and a seqcount to provide proper support for 32bit kernels.
- We reduce memory need when estimators are not used, since we store a pointer, instead of the bytes/packets counters.
- xt_rateest_mt() no longer has to grab a spinlock. (In the future, xt_rateest_tg() could be switched to per cpu counters)
Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net> 1c0d32fd Sun Dec 04 11:48:16 CST 2016 Eric Dumazet <edumazet@google.com> net_sched: gen_estimator: complete rewrite of rate estimators 1) Old code was hard to maintain, due to complex lock chains. (We probably will be able to remove some kfree_rcu() in callers) 2) Using a single timer to update all estimators does not scale. 3) Code was buggy on 32bit kernel (WRITE_ONCE() on 64bit quantity is not supposed to work well) In this rewrite : - I removed the RB tree that had to be scanned in gen_estimator_active(). qdisc dumps should be much faster. - Each estimator has its own timer. - Estimations are maintained in net_rate_estimator structure, instead of dirtying the qdisc. Minor, but part of the simplification. - Reading the estimator uses RCU and a seqcount to provide proper support for 32bit kernels. - We reduce memory need when estimators are not used, since we store a pointer, instead of the bytes/packets counters. - xt_rateest_mt() no longer has to grab a spinlock. (In the future, xt_rateest_tg() could be switched to per cpu counters) Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
|
H A D | gen_estimator.c | 1c0d32fd Sun Dec 04 11:48:16 CST 2016 Eric Dumazet <edumazet@google.com> net_sched: gen_estimator: complete rewrite of rate estimators
1) Old code was hard to maintain, due to complex lock chains. (We probably will be able to remove some kfree_rcu() in callers)
2) Using a single timer to update all estimators does not scale.
3) Code was buggy on 32bit kernel (WRITE_ONCE() on 64bit quantity is not supposed to work well)
In this rewrite :
- I removed the RB tree that had to be scanned in gen_estimator_active(). qdisc dumps should be much faster.
- Each estimator has its own timer.
- Estimations are maintained in net_rate_estimator structure, instead of dirtying the qdisc. Minor, but part of the simplification.
- Reading the estimator uses RCU and a seqcount to provide proper support for 32bit kernels.
- We reduce memory need when estimators are not used, since we store a pointer, instead of the bytes/packets counters.
- xt_rateest_mt() no longer has to grab a spinlock. (In the future, xt_rateest_tg() could be switched to per cpu counters)
Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net> 1c0d32fd Sun Dec 04 11:48:16 CST 2016 Eric Dumazet <edumazet@google.com> net_sched: gen_estimator: complete rewrite of rate estimators 1) Old code was hard to maintain, due to complex lock chains. (We probably will be able to remove some kfree_rcu() in callers) 2) Using a single timer to update all estimators does not scale. 3) Code was buggy on 32bit kernel (WRITE_ONCE() on 64bit quantity is not supposed to work well) In this rewrite : - I removed the RB tree that had to be scanned in gen_estimator_active(). qdisc dumps should be much faster. - Each estimator has its own timer. - Estimations are maintained in net_rate_estimator structure, instead of dirtying the qdisc. Minor, but part of the simplification. - Reading the estimator uses RCU and a seqcount to provide proper support for 32bit kernels. - We reduce memory need when estimators are not used, since we store a pointer, instead of the bytes/packets counters. - xt_rateest_mt() no longer has to grab a spinlock. (In the future, xt_rateest_tg() could be switched to per cpu counters) Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
|
/openbmc/linux/net/sched/ |
H A D | sch_drr.c | 1c0d32fd Sun Dec 04 11:48:16 CST 2016 Eric Dumazet <edumazet@google.com> net_sched: gen_estimator: complete rewrite of rate estimators
1) Old code was hard to maintain, due to complex lock chains. (We probably will be able to remove some kfree_rcu() in callers)
2) Using a single timer to update all estimators does not scale.
3) Code was buggy on 32bit kernel (WRITE_ONCE() on 64bit quantity is not supposed to work well)
In this rewrite :
- I removed the RB tree that had to be scanned in gen_estimator_active(). qdisc dumps should be much faster.
- Each estimator has its own timer.
- Estimations are maintained in net_rate_estimator structure, instead of dirtying the qdisc. Minor, but part of the simplification.
- Reading the estimator uses RCU and a seqcount to provide proper support for 32bit kernels.
- We reduce memory need when estimators are not used, since we store a pointer, instead of the bytes/packets counters.
- xt_rateest_mt() no longer has to grab a spinlock. (In the future, xt_rateest_tg() could be switched to per cpu counters)
Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net> 1c0d32fd Sun Dec 04 11:48:16 CST 2016 Eric Dumazet <edumazet@google.com> net_sched: gen_estimator: complete rewrite of rate estimators 1) Old code was hard to maintain, due to complex lock chains. (We probably will be able to remove some kfree_rcu() in callers) 2) Using a single timer to update all estimators does not scale. 3) Code was buggy on 32bit kernel (WRITE_ONCE() on 64bit quantity is not supposed to work well) In this rewrite : - I removed the RB tree that had to be scanned in gen_estimator_active(). qdisc dumps should be much faster. - Each estimator has its own timer. - Estimations are maintained in net_rate_estimator structure, instead of dirtying the qdisc. Minor, but part of the simplification. - Reading the estimator uses RCU and a seqcount to provide proper support for 32bit kernels. - We reduce memory need when estimators are not used, since we store a pointer, instead of the bytes/packets counters. - xt_rateest_mt() no longer has to grab a spinlock. (In the future, xt_rateest_tg() could be switched to per cpu counters) Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
|
H A D | sch_qfq.c | 1c0d32fd Sun Dec 04 11:48:16 CST 2016 Eric Dumazet <edumazet@google.com> net_sched: gen_estimator: complete rewrite of rate estimators
1) Old code was hard to maintain, due to complex lock chains. (We probably will be able to remove some kfree_rcu() in callers)
2) Using a single timer to update all estimators does not scale.
3) Code was buggy on 32bit kernel (WRITE_ONCE() on 64bit quantity is not supposed to work well)
In this rewrite :
- I removed the RB tree that had to be scanned in gen_estimator_active(). qdisc dumps should be much faster.
- Each estimator has its own timer.
- Estimations are maintained in net_rate_estimator structure, instead of dirtying the qdisc. Minor, but part of the simplification.
- Reading the estimator uses RCU and a seqcount to provide proper support for 32bit kernels.
- We reduce memory need when estimators are not used, since we store a pointer, instead of the bytes/packets counters.
- xt_rateest_mt() no longer has to grab a spinlock. (In the future, xt_rateest_tg() could be switched to per cpu counters)
Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net> 1c0d32fd Sun Dec 04 11:48:16 CST 2016 Eric Dumazet <edumazet@google.com> net_sched: gen_estimator: complete rewrite of rate estimators 1) Old code was hard to maintain, due to complex lock chains. (We probably will be able to remove some kfree_rcu() in callers) 2) Using a single timer to update all estimators does not scale. 3) Code was buggy on 32bit kernel (WRITE_ONCE() on 64bit quantity is not supposed to work well) In this rewrite : - I removed the RB tree that had to be scanned in gen_estimator_active(). qdisc dumps should be much faster. - Each estimator has its own timer. - Estimations are maintained in net_rate_estimator structure, instead of dirtying the qdisc. Minor, but part of the simplification. - Reading the estimator uses RCU and a seqcount to provide proper support for 32bit kernels. - We reduce memory need when estimators are not used, since we store a pointer, instead of the bytes/packets counters. - xt_rateest_mt() no longer has to grab a spinlock. (In the future, xt_rateest_tg() could be switched to per cpu counters) Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
|
H A D | sch_hfsc.c | 1c0d32fd Sun Dec 04 11:48:16 CST 2016 Eric Dumazet <edumazet@google.com> net_sched: gen_estimator: complete rewrite of rate estimators
1) Old code was hard to maintain, due to complex lock chains. (We probably will be able to remove some kfree_rcu() in callers)
2) Using a single timer to update all estimators does not scale.
3) Code was buggy on 32bit kernel (WRITE_ONCE() on 64bit quantity is not supposed to work well)
In this rewrite :
- I removed the RB tree that had to be scanned in gen_estimator_active(). qdisc dumps should be much faster.
- Each estimator has its own timer.
- Estimations are maintained in net_rate_estimator structure, instead of dirtying the qdisc. Minor, but part of the simplification.
- Reading the estimator uses RCU and a seqcount to provide proper support for 32bit kernels.
- We reduce memory need when estimators are not used, since we store a pointer, instead of the bytes/packets counters.
- xt_rateest_mt() no longer has to grab a spinlock. (In the future, xt_rateest_tg() could be switched to per cpu counters)
Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net> 1c0d32fd Sun Dec 04 11:48:16 CST 2016 Eric Dumazet <edumazet@google.com> net_sched: gen_estimator: complete rewrite of rate estimators 1) Old code was hard to maintain, due to complex lock chains. (We probably will be able to remove some kfree_rcu() in callers) 2) Using a single timer to update all estimators does not scale. 3) Code was buggy on 32bit kernel (WRITE_ONCE() on 64bit quantity is not supposed to work well) In this rewrite : - I removed the RB tree that had to be scanned in gen_estimator_active(). qdisc dumps should be much faster. - Each estimator has its own timer. - Estimations are maintained in net_rate_estimator structure, instead of dirtying the qdisc. Minor, but part of the simplification. - Reading the estimator uses RCU and a seqcount to provide proper support for 32bit kernels. - We reduce memory need when estimators are not used, since we store a pointer, instead of the bytes/packets counters. - xt_rateest_mt() no longer has to grab a spinlock. (In the future, xt_rateest_tg() could be switched to per cpu counters) Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
|
H A D | act_police.c | 1c0d32fd Sun Dec 04 11:48:16 CST 2016 Eric Dumazet <edumazet@google.com> net_sched: gen_estimator: complete rewrite of rate estimators
1) Old code was hard to maintain, due to complex lock chains. (We probably will be able to remove some kfree_rcu() in callers)
2) Using a single timer to update all estimators does not scale.
3) Code was buggy on 32bit kernel (WRITE_ONCE() on 64bit quantity is not supposed to work well)
In this rewrite :
- I removed the RB tree that had to be scanned in gen_estimator_active(). qdisc dumps should be much faster.
- Each estimator has its own timer.
- Estimations are maintained in net_rate_estimator structure, instead of dirtying the qdisc. Minor, but part of the simplification.
- Reading the estimator uses RCU and a seqcount to provide proper support for 32bit kernels.
- We reduce memory need when estimators are not used, since we store a pointer, instead of the bytes/packets counters.
- xt_rateest_mt() no longer has to grab a spinlock. (In the future, xt_rateest_tg() could be switched to per cpu counters)
Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net> 1c0d32fd Sun Dec 04 11:48:16 CST 2016 Eric Dumazet <edumazet@google.com> net_sched: gen_estimator: complete rewrite of rate estimators 1) Old code was hard to maintain, due to complex lock chains. (We probably will be able to remove some kfree_rcu() in callers) 2) Using a single timer to update all estimators does not scale. 3) Code was buggy on 32bit kernel (WRITE_ONCE() on 64bit quantity is not supposed to work well) In this rewrite : - I removed the RB tree that had to be scanned in gen_estimator_active(). qdisc dumps should be much faster. - Each estimator has its own timer. - Estimations are maintained in net_rate_estimator structure, instead of dirtying the qdisc. Minor, but part of the simplification. - Reading the estimator uses RCU and a seqcount to provide proper support for 32bit kernels. - We reduce memory need when estimators are not used, since we store a pointer, instead of the bytes/packets counters. - xt_rateest_mt() no longer has to grab a spinlock. (In the future, xt_rateest_tg() could be switched to per cpu counters) Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
|
H A D | sch_htb.c | 1c0d32fd Sun Dec 04 11:48:16 CST 2016 Eric Dumazet <edumazet@google.com> net_sched: gen_estimator: complete rewrite of rate estimators
1) Old code was hard to maintain, due to complex lock chains. (We probably will be able to remove some kfree_rcu() in callers)
2) Using a single timer to update all estimators does not scale.
3) Code was buggy on 32bit kernel (WRITE_ONCE() on 64bit quantity is not supposed to work well)
In this rewrite :
- I removed the RB tree that had to be scanned in gen_estimator_active(). qdisc dumps should be much faster.
- Each estimator has its own timer.
- Estimations are maintained in net_rate_estimator structure, instead of dirtying the qdisc. Minor, but part of the simplification.
- Reading the estimator uses RCU and a seqcount to provide proper support for 32bit kernels.
- We reduce memory need when estimators are not used, since we store a pointer, instead of the bytes/packets counters.
- xt_rateest_mt() no longer has to grab a spinlock. (In the future, xt_rateest_tg() could be switched to per cpu counters)
Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net> 1c0d32fd Sun Dec 04 11:48:16 CST 2016 Eric Dumazet <edumazet@google.com> net_sched: gen_estimator: complete rewrite of rate estimators 1) Old code was hard to maintain, due to complex lock chains. (We probably will be able to remove some kfree_rcu() in callers) 2) Using a single timer to update all estimators does not scale. 3) Code was buggy on 32bit kernel (WRITE_ONCE() on 64bit quantity is not supposed to work well) In this rewrite : - I removed the RB tree that had to be scanned in gen_estimator_active(). qdisc dumps should be much faster. - Each estimator has its own timer. - Estimations are maintained in net_rate_estimator structure, instead of dirtying the qdisc. Minor, but part of the simplification. - Reading the estimator uses RCU and a seqcount to provide proper support for 32bit kernels. - We reduce memory need when estimators are not used, since we store a pointer, instead of the bytes/packets counters. - xt_rateest_mt() no longer has to grab a spinlock. (In the future, xt_rateest_tg() could be switched to per cpu counters) Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
|
H A D | act_api.c | 1c0d32fd Sun Dec 04 11:48:16 CST 2016 Eric Dumazet <edumazet@google.com> net_sched: gen_estimator: complete rewrite of rate estimators
1) Old code was hard to maintain, due to complex lock chains. (We probably will be able to remove some kfree_rcu() in callers)
2) Using a single timer to update all estimators does not scale.
3) Code was buggy on 32bit kernel (WRITE_ONCE() on 64bit quantity is not supposed to work well)
In this rewrite :
- I removed the RB tree that had to be scanned in gen_estimator_active(). qdisc dumps should be much faster.
- Each estimator has its own timer.
- Estimations are maintained in net_rate_estimator structure, instead of dirtying the qdisc. Minor, but part of the simplification.
- Reading the estimator uses RCU and a seqcount to provide proper support for 32bit kernels.
- We reduce memory need when estimators are not used, since we store a pointer, instead of the bytes/packets counters.
- xt_rateest_mt() no longer has to grab a spinlock. (In the future, xt_rateest_tg() could be switched to per cpu counters)
Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net> 1c0d32fd Sun Dec 04 11:48:16 CST 2016 Eric Dumazet <edumazet@google.com> net_sched: gen_estimator: complete rewrite of rate estimators 1) Old code was hard to maintain, due to complex lock chains. (We probably will be able to remove some kfree_rcu() in callers) 2) Using a single timer to update all estimators does not scale. 3) Code was buggy on 32bit kernel (WRITE_ONCE() on 64bit quantity is not supposed to work well) In this rewrite : - I removed the RB tree that had to be scanned in gen_estimator_active(). qdisc dumps should be much faster. - Each estimator has its own timer. - Estimations are maintained in net_rate_estimator structure, instead of dirtying the qdisc. Minor, but part of the simplification. - Reading the estimator uses RCU and a seqcount to provide proper support for 32bit kernels. - We reduce memory need when estimators are not used, since we store a pointer, instead of the bytes/packets counters. - xt_rateest_mt() no longer has to grab a spinlock. (In the future, xt_rateest_tg() could be switched to per cpu counters) Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
|
H A D | sch_api.c | 1c0d32fd Sun Dec 04 11:48:16 CST 2016 Eric Dumazet <edumazet@google.com> net_sched: gen_estimator: complete rewrite of rate estimators
1) Old code was hard to maintain, due to complex lock chains. (We probably will be able to remove some kfree_rcu() in callers)
2) Using a single timer to update all estimators does not scale.
3) Code was buggy on 32bit kernel (WRITE_ONCE() on 64bit quantity is not supposed to work well)
In this rewrite :
- I removed the RB tree that had to be scanned in gen_estimator_active(). qdisc dumps should be much faster.
- Each estimator has its own timer.
- Estimations are maintained in net_rate_estimator structure, instead of dirtying the qdisc. Minor, but part of the simplification.
- Reading the estimator uses RCU and a seqcount to provide proper support for 32bit kernels.
- We reduce memory need when estimators are not used, since we store a pointer, instead of the bytes/packets counters.
- xt_rateest_mt() no longer has to grab a spinlock. (In the future, xt_rateest_tg() could be switched to per cpu counters)
Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net> 1c0d32fd Sun Dec 04 11:48:16 CST 2016 Eric Dumazet <edumazet@google.com> net_sched: gen_estimator: complete rewrite of rate estimators 1) Old code was hard to maintain, due to complex lock chains. (We probably will be able to remove some kfree_rcu() in callers) 2) Using a single timer to update all estimators does not scale. 3) Code was buggy on 32bit kernel (WRITE_ONCE() on 64bit quantity is not supposed to work well) In this rewrite : - I removed the RB tree that had to be scanned in gen_estimator_active(). qdisc dumps should be much faster. - Each estimator has its own timer. - Estimations are maintained in net_rate_estimator structure, instead of dirtying the qdisc. Minor, but part of the simplification. - Reading the estimator uses RCU and a seqcount to provide proper support for 32bit kernels. - We reduce memory need when estimators are not used, since we store a pointer, instead of the bytes/packets counters. - xt_rateest_mt() no longer has to grab a spinlock. (In the future, xt_rateest_tg() could be switched to per cpu counters) Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
|
H A D | sch_generic.c | 1c0d32fd Sun Dec 04 11:48:16 CST 2016 Eric Dumazet <edumazet@google.com> net_sched: gen_estimator: complete rewrite of rate estimators
1) Old code was hard to maintain, due to complex lock chains. (We probably will be able to remove some kfree_rcu() in callers)
2) Using a single timer to update all estimators does not scale.
3) Code was buggy on 32bit kernel (WRITE_ONCE() on 64bit quantity is not supposed to work well)
In this rewrite :
- I removed the RB tree that had to be scanned in gen_estimator_active(). qdisc dumps should be much faster.
- Each estimator has its own timer.
- Estimations are maintained in net_rate_estimator structure, instead of dirtying the qdisc. Minor, but part of the simplification.
- Reading the estimator uses RCU and a seqcount to provide proper support for 32bit kernels.
- We reduce memory need when estimators are not used, since we store a pointer, instead of the bytes/packets counters.
- xt_rateest_mt() no longer has to grab a spinlock. (In the future, xt_rateest_tg() could be switched to per cpu counters)
Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net> 1c0d32fd Sun Dec 04 11:48:16 CST 2016 Eric Dumazet <edumazet@google.com> net_sched: gen_estimator: complete rewrite of rate estimators 1) Old code was hard to maintain, due to complex lock chains. (We probably will be able to remove some kfree_rcu() in callers) 2) Using a single timer to update all estimators does not scale. 3) Code was buggy on 32bit kernel (WRITE_ONCE() on 64bit quantity is not supposed to work well) In this rewrite : - I removed the RB tree that had to be scanned in gen_estimator_active(). qdisc dumps should be much faster. - Each estimator has its own timer. - Estimations are maintained in net_rate_estimator structure, instead of dirtying the qdisc. Minor, but part of the simplification. - Reading the estimator uses RCU and a seqcount to provide proper support for 32bit kernels. - We reduce memory need when estimators are not used, since we store a pointer, instead of the bytes/packets counters. - xt_rateest_mt() no longer has to grab a spinlock. (In the future, xt_rateest_tg() could be switched to per cpu counters) Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
|