From: Eric Dumazet
Date: Mon, 13 Jun 2016 20:21:49 -0700
> qdiscs/classes are changed under RTNL protection and often
> while blocking BH and root qdisc spinlock.
>
> When lots of skbs need to be dropped, we free
> them under these locks causing TX/RX freezes,
> and more generally latency spikes
On 16-06-13 11:21 PM, Eric Dumazet wrote:
qdiscs/classes are changed under RTNL protection and often
while blocking BH and root qdisc spinlock.
When lots of skbs need to be dropped, we free
them under these locks causing TX/RX freezes,
and more generally latency spikes.
I saw spikes of 50+ ms o
On Tue, 2016-06-14 at 19:13 -0700, Cong Wang wrote:
> No objection from me. It looks like a good optimization
> before we can improve the qdisc root spinlock.
>
> Just one nit: You probably want to keep rtnl_kfree_skbs()
> within qdisc layer unless you have any plan to use it
> in other places.
On Mon, Jun 13, 2016 at 8:21 PM, Eric Dumazet wrote:
> qdiscs/classes are changed under RTNL protection and often
> while blocking BH and root qdisc spinlock.
>
> When lots of skbs need to be dropped, we free
> them under these locks causing TX/RX freezes,
> and more generally latency spikes.
>
>
qdiscs/classes are changed under RTNL protection and often
while blocking BH and root qdisc spinlock.
When lots of skbs need to be dropped, we free
them under these locks causing TX/RX freezes,
and more generally latency spikes.
I saw spikes of 50+ ms on quite fast hardware...
This patch series
