Re: [PATCH v8 0/5] skb_array: array based FIFO for skbs

2016-06-15 Thread David Miller
From: "Michael S. Tsirkin" Date: Mon, 13 Jun 2016 23:54:26 +0300 > This is in response to the proposal by Jason to make tun > rx packet queue lockless using a circular buffer. > My testing seems to show that at least for the common usecase > in networking, which isn't lockless,

Re: [PATCH v8 0/5] skb_array: array based FIFO for skbs

2016-06-15 Thread David Miller
From: "Michael S. Tsirkin" Date: Mon, 13 Jun 2016 23:54:26 +0300 > This is in response to the proposal by Jason to make tun > rx packet queue lockless using a circular buffer. > My testing seems to show that at least for the common usecase > in networking, which isn't lockless, circular buffer >

Re: [PATCH v8 0/5] skb_array: array based FIFO for skbs

2016-06-14 Thread Jesper Dangaard Brouer
On Mon, 13 Jun 2016 23:54:26 +0300 "Michael S. Tsirkin" wrote: > > Jesper, Jason, I know that both of you tested this, > please post Tested-by tags for whatever was tested. Yes, I tested this via "external" kernel modules, not your framework. I have verified that this V8

Re: [PATCH v8 0/5] skb_array: array based FIFO for skbs

2016-06-14 Thread Jesper Dangaard Brouer
On Mon, 13 Jun 2016 23:54:26 +0300 "Michael S. Tsirkin" wrote: > > Jesper, Jason, I know that both of you tested this, > please post Tested-by tags for whatever was tested. Yes, I tested this via "external" kernel modules, not your framework. I have verified that this V8 patchset is

[PATCH v8 0/5] skb_array: array based FIFO for skbs

2016-06-13 Thread Michael S. Tsirkin
This is in response to the proposal by Jason to make tun rx packet queue lockless using a circular buffer. My testing seems to show that at least for the common usecase in networking, which isn't lockless, circular buffer with indices does not perform that well, because each index access causes a

[PATCH v8 0/5] skb_array: array based FIFO for skbs

2016-06-13 Thread Michael S. Tsirkin
This is in response to the proposal by Jason to make tun rx packet queue lockless using a circular buffer. My testing seems to show that at least for the common usecase in networking, which isn't lockless, circular buffer with indices does not perform that well, because each index access causes a