[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <4ADFE426.6040806@gmail.com>
Date: Thu, 22 Oct 2009 06:48:38 +0200
From: Eric Dumazet <eric.dumazet@...il.com>
To: Stephen Hemminger <shemminger@...tta.com>
CC: "David S. Miller" <davem@...emloft.net>,
Linux Netdev List <netdev@...r.kernel.org>
Subject: Re: [PATCH net-next-2.6] rtnetlink: rtnl_setlink() and rtnl_getlink()
changes
Eric Dumazet a écrit :
> Stephen Hemminger a écrit :
>> On Wed, 21 Oct 2009 22:59:31 +0200
>> Eric Dumazet <eric.dumazet@...il.com> wrote:
>>
>>> Stephen, do you think we could change "ip link show dev ethX" to
>>> let it use rtnl_getlink() instead of rtnl_dump_ifinfo() ?
>>>
>> Would work, but not sure what it gains.
>
> It takes about one second to dump the table when we have 25.000 devices
Adding new links takes also lot of time in rtnl_dump_ifinfo(), we could
optimize it using a 256 fanout (using the ifindex hash table instead
of the single list)
But IMHO rtnl_dump_ifinfo() should be used only when needed, not when
querying/adding a particular device.
------------------------------------------------------------------------------
PerfTop: 42745 irqs/sec kernel:88.0% [100000 cycles], (all, 8 CPUs)
------------------------------------------------------------------------------
samples pcnt kernel function
_______ _____ _______________
231146.00 - 52.4% : rtnl_dump_ifinfo
18491.00 - 4.2% : __register_sysctl_paths
17700.00 - 4.0% : mwait_idle
12883.00 - 2.9% : rtnl_fill_ifinfo
12661.00 - 2.9% : schedule
6324.00 - 1.4% : find_busiest_group
5911.00 - 1.3% : _spin_lock_irqsave
4862.00 - 1.1% : dev_get_stats
4726.00 - 1.1% : copy_to_user
4547.00 - 1.0% : __nla_put
4117.00 - 0.9% : sysfs_find_dirent
4090.00 - 0.9% : sysenter_past_esp
3789.00 - 0.9% : fput
3735.00 - 0.8% : __nla_reserve
3699.00 - 0.8% : read_tsc
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Powered by blists - more mailing lists