lkml.org 
[lkml]   [2004]   [Sep]   [20]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
SubjectRe: hotplug e1000 failed after 32 times
From
Date
On Mon, 2004-09-20 at 12:15, Chris Leech wrote:
> On Sun, 19 Sep 2004 22:55:51 -0400, Jeff Garzik <jgarzik@pobox.com> wrote:
> > Li Shaohua wrote:
> > > I'm not familiar with the NIC driver, but this problem really is
> > > annoying. The gurus, please consider a solution. It's not an uncommon
> > > case. I believe it's common in a big system with hotplug support. I can
> > > understand why the driver doesn't support more than 32 a card, but one
> > > card with 32 times hotplug failed is a little ugly.
> >
> > There should be no problem at all with the driver supporting 32 NICs...
> > in fact if it cannot support at least 99 NICs, I would consider that a
> > bug.
>
> I'd like to hear more about what the failure condition actually is,
> see any system messages logged.
>
> The 32 NIC limitation is for configuration via the module parameters,
> but e1000 itself doesn't impose any other limitations on the number of
> NICs. I know of testing situations where 360 e1000 ports were in use
> on a large system. I've run hotplug tests, powering slots on and off
> while passing traffic in a fail-over teaming configuration, for weeks
> without problem. I'll run a hotplug test myself as soon as I get a
> chance, but I don't see any reason why it should be failing.
>
> As for the module parameter limitations, yes only 32 ports can be
> configured that way and the counter doesn't roll back with
> hot-removes. I think we can live with those limitations. Indexed
> parameters make no sense in a hotplug environment, use ethtool.
>
> The module parameters in e1000 work for a large number or users who
> can live with those limitations. That's why they're still in the
> driver. As long as they work, I think they should stay. But I'm not
> in favor of adding more code, or making the existing code more
> complex, to try and remove those limitations. That's why everything
> can be configured through ethtool.
Below is the dmesg I got.
>ACPI: PCI interrupt 0000:11:01.0[A] -> GSI 120 (level, low) -> IRQ 58
>e1000: eth1: e1000_probe: Intel(R) PRO/1000 Network Connection
>e1000: eth1: e1000_check_options: Warning: no configuration for board
#36
>e1000: eth1: e1000_check_options: Using defaults for all values
but the default value isn't correct, such as the 'speed' is 65535.
I think you are right, I can change the parameter using ethtool. This
way I don't need the driver changes. Thanks.

Thanks,
Shaohua


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 14:06    [W:0.055 / U:0.332 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site