On Tue, 9 Nov 2010, Holger Eitzenberger wrote:
> Hi,
>
> using e1000e driver version 1.2.10 and kernel version 2.6.32.24 I see
> the kernel go BUG() sporadically at the time 'ethtool -p eth0 3' comes
> back.
>
> Network hardware is four times 'Intel Corporation 82583V Gigabit Network
> Connect
2010/11/1 xiaolin :
> In e1000 driver, there is ew32(IMC, ~0) in the function of e1000_intr before
> scheduling adapter->napi.
>
> However, there is no such kind operation in e1000e.
>
> My question is whether NIC hardware irq is disabled during the NAPI/ksoftirqd
> processing?
yes, it is disabl
On 11/10/2010 9:09 AM, Alexander Duyck wrote:
> On 11/10/2010 8:54 AM, Stephen Hemminger wrote:
>> On Wed, 10 Nov 2010 14:33:01 -0200
>> Flavio Leitner wrote:
>>
>>> On Wed, Nov 10, 2010 at 07:58:55AM -0800, Stephen Hemminger wrote:
On Wed, 10 Nov 2010 10:56:38 -0200
Flavio Leitner wr
On 11/10/2010 8:54 AM, Stephen Hemminger wrote:
> On Wed, 10 Nov 2010 14:33:01 -0200
> Flavio Leitner wrote:
>
>> On Wed, Nov 10, 2010 at 07:58:55AM -0800, Stephen Hemminger wrote:
>>> On Wed, 10 Nov 2010 10:56:38 -0200
>>> Flavio Leitner wrote:
>>>
Hello,
The driver is calcul
On Wed, 10 Nov 2010 14:33:01 -0200
Flavio Leitner wrote:
> On Wed, Nov 10, 2010 at 07:58:55AM -0800, Stephen Hemminger wrote:
> > On Wed, 10 Nov 2010 10:56:38 -0200
> > Flavio Leitner wrote:
> >
> > >
> > > Hello,
> > >
> > > The driver is calculating the amount of bytes transmitted assuming
On Wed, Nov 10, 2010 at 07:58:55AM -0800, Stephen Hemminger wrote:
> On Wed, 10 Nov 2010 10:56:38 -0200
> Flavio Leitner wrote:
>
> >
> > Hello,
> >
> > The driver is calculating the amount of bytes transmitted assuming
> > the way TCP builds GSO packets which are non-linear. Therefore,
> > skb
On Wed, 10 Nov 2010 10:56:38 -0200
Flavio Leitner wrote:
>
> Hello,
>
> The driver is calculating the amount of bytes transmitted assuming
> the way TCP builds GSO packets which are non-linear. Therefore,
> skb_headlen() returns the correct size of the header and the
> math is correct.
>
> How
Hello,
The driver is calculating the amount of bytes transmitted assuming
the way TCP builds GSO packets which are non-linear. Therefore,
skb_headlen() returns the correct size of the header and the
math is correct.
However, the skb can be linearized if it is copied using
skb_copy(), which leads
