From: Eric Dumazet <eric.dumazet@gmail.com>
To: Wei Gu <wei.gu@ericsson.com>
Cc: netdev <netdev@vger.kernel.org>,
Alexander Duyck <alexander.h.duyck@intel.com>,
Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Subject: RE: Question on "net: allocate skbs on local node"
Date: Thu, 07 Apr 2011 06:58:47 +0200 [thread overview]
Message-ID: <1302152327.2701.50.camel@edumazet-laptop> (raw)
In-Reply-To: <D12839161ADD3A4B8DA63D1A134D084026E48B9BEB@ESGSCCMS0001.eapac.ericsson.se>
Le jeudi 07 avril 2011 à 10:16 +0800, Wei Gu a écrit :
> Hi Eric,
> Testing with ixgbe Linux 2.6.38 driver:
> We have a little better thruput figure with this driver, but it looks
> not scalling at all, I always stressed one CPU core/24.
> And when look the perf report for ksoftirqd/24, the most cost function
> is still "_raw_spin_unlock_irqstore" and the IRQ/s is huge, it's
> somehow conflicts with desgin of NAPI. On linux 2.6.32 while the CPU
> was stressed the IRQ will descreased while the NAPI will running much
> on the polling mode. I don't know why on 2.6.38 the IRQ was keep
> increasing.
CC netdev and Intel guys, since they said it should not happen (TM)
IF you dont use DCA (make sure ioatdma module is not loaded), how comes
alloc_iova() is called at all ?
IF you use DCA, how comes its called, since the same CPU serves a given
interrupt ?
>
> CONFIG_TICK_ONESHOT=y
> CONFIG_NO_HZ=y
>
> PerfTop: 512417 irqs/sec kernel:91.3% exact: 0.0% [1000Hz cpu-clock-msecs], (all, 64 CPUs)
> ------------------------------------------------------------------------------------------------------------------------------------------------------
> - 0.82% ksoftirqd/24 [kernel.kallsyms] [k] _raw_spin_unlock_irqrestore
> \u2592 - _raw_spin_unlock_irqrestore
> \u2592 - 44.27% alloc_iova
> \u2592 intel_alloc_iova
> \u2592 __intel_map_single
> \u2592 intel_map_page
> \u2592 - ixgbe_init_interrupt_scheme
> \u2592 - 59.97% ixgbe_alloc_rx_buffers
> \u2592 ixgbe_clean_rx_irq
> \u2592 0xffffffffa033a5
> \u2592 net_rx_action
> u2592 __do_softirq
> \u2592 + call_softirq
> \u2592 - 40.03% ixgbe_change_mtu
> \u2592 ixgbe_change_mtu
> \u2592 dev_hard_start_xmit
> \u2592 sch_direct_xmit
> \u2592 dev_queue_xmit
> \u2592 vlan_dev_hard_start_xmit
> \u2592 hook_func
> \u2592 nf_iterate
> \u2592 nf_hook_slow
> \u2592 NF_HOOK.clone.1
> \u2592 ip_rcv
> \u2592 __netif_receive_skb
> \u2592 __netif_receive_skb
> \u2592 netif_receive_skb
> \u2592 napi_skb_finish
> \u2592 napi_gro_receive
> \u2592 ixgbe_clean_rx_irq
> \u2592 0xffffffffa033a5
> \u2592 net_rx_action
> \u2592 __do_softirq
> \u2592 + call_softirq
> \u2592 + 35.85% find_iova
> \u2592 + 19.44% add_unmap
>
>
> Thanks
> WeiGu
>
next parent reply other threads:[~2011-04-07 4:58 UTC|newest]
Thread overview: 58+ messages / expand[flat|nested] mbox.gz Atom feed top
[not found] <D12839161ADD3A4B8DA63D1A134D084026E48B9BEB@ESGSCCMS0001.eapac.ericsson.se>
2011-04-07 4:58 ` Eric Dumazet [this message]
2011-04-07 5:16 ` Question on "net: allocate skbs on local node" Eric Dumazet
2011-04-07 6:16 ` Eric Dumazet
2011-04-07 7:22 ` Low performance Intel 10GE NIC (3.2.10) on 2.6.38 Kernel Wei Gu
2011-04-07 8:07 ` Eric Dumazet
2011-04-07 8:39 ` Wei Gu
2011-04-07 9:06 ` Eric Dumazet
2011-04-07 11:15 ` Wei Gu
2011-04-07 11:46 ` Eric Dumazet
2011-04-07 13:41 ` Eric Dumazet
2011-04-07 15:58 ` Alexander Duyck
2011-04-07 16:03 ` Eric Dumazet
2011-04-07 16:20 ` Alexander Duyck
2011-04-07 16:37 ` Eric Dumazet
2011-04-08 8:59 ` Wei Gu
2011-04-08 9:07 ` Eric Dumazet
2011-04-08 9:15 ` Wei Gu
2011-04-08 9:49 ` Eric Dumazet
2011-04-08 9:59 ` Wei Gu
2011-04-08 9:41 ` Wei Gu
2011-04-08 12:19 ` Wei Gu
2011-04-08 12:56 ` Eric Dumazet
2011-04-08 14:10 ` Wei Gu
2011-04-08 14:49 ` Stephen Hemminger
2011-04-09 3:51 ` Wei Gu
2011-04-08 15:07 ` Eric Dumazet
2011-04-09 3:27 ` Wei Gu
2011-04-09 6:36 ` Eric Dumazet
2011-04-10 7:02 ` Wei Gu
2011-04-11 14:50 ` Alexander Duyck
2011-04-11 15:00 ` Wei Gu
2011-04-11 15:14 ` Wei Gu
2011-04-11 15:42 ` Eric Dumazet
2011-04-12 1:22 ` Wei Gu
2011-04-12 4:40 ` Wei Gu
2011-04-12 4:56 ` Eric Dumazet
2011-04-12 5:18 ` Wei Gu
2011-04-14 5:42 ` Wei Gu
2011-04-14 6:07 ` Eric Dumazet
2011-04-14 6:33 ` Eric Dumazet
2011-04-14 6:58 ` Wei Gu
2011-04-14 16:42 ` Alexander Duyck
2011-04-14 16:45 ` Eric Dumazet
2011-04-14 16:56 ` Peter Zijlstra
2011-04-14 16:57 ` Eric Dumazet
2011-04-14 17:49 ` Eric Dumazet
2011-04-14 19:08 ` Alexander Duyck
2011-04-15 2:10 ` Wei Gu
2011-04-15 8:57 ` Peter Zijlstra
2011-04-15 9:14 ` Wei Gu
2011-04-18 21:12 ` Jesse Brandeburg
2011-04-19 4:09 ` Wei Gu
2011-04-21 2:57 ` Wei Gu
2011-04-21 3:25 ` Wei Gu
2011-04-08 16:22 ` Alexander Duyck
2011-04-09 3:36 ` Wei Gu
2011-04-09 4:40 ` Alexander H Duyck
2011-04-09 6:12 ` Wei Gu
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1302152327.2701.50.camel@edumazet-laptop \
--to=eric.dumazet@gmail.com \
--cc=alexander.h.duyck@intel.com \
--cc=jeffrey.t.kirsher@intel.com \
--cc=netdev@vger.kernel.org \
--cc=wei.gu@ericsson.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox