Re: [PATCH] Fix e1000e with Intel 82572EI that has no hardware timestamp support

[Alexander Duyck]

On 08/06/2014 08:21 AM, Alexander Duyck wrote:
> On 08/06/2014 08:09 AM, Henrique de Moraes Holschuh wrote:
>> On Wed, Aug 6, 2014, at 11:30, Sergei Shtylyov wrote:
>>> On 8/6/2014 6:27 PM, Koehrer Mathias (ETAS/ESW5) wrote:
>>>
>>>> With the Intel 82527EI (driver: e1000e) there is an issue when running
>>>> the ptpd2 program, that leads to a kernel oops.
>>>> The reason is here that in e1000_xmit_frame() a work queue will be
>>>> scheduled that has not been initialized in this case.
>>>> The work queue "tx_hwstamp_work" will only be initialized if
>>>> adapter->flags & FLAG_HAS_HW_TIMESTAMP set.
>>>> This check is missing in e1000_xmit_frame().
>>>
>>>> The following patch adds the missing check.
>>>
>>>     OK, but this time you forgot your sign-off. :-)
>>
>> Also, shouldn't the added test be inside the unlikely() ?
>>
> 
> That shouldn't be necessary, but it might be better to place the new
> check after the unlikely.  So you do the unlikely check first, then the
> adapter check to see if we support Tx timestamping.  That way we can
> avoid any unnecessary checks in the adapter structure.
> 
> Thanks,
> 
> Alex
> 


Actually you might want to even take it one step further and break out
the test as follows (note this is only pseudo-code:
  unlikely(skb->tx_flags & HW_TSTAMP) &&
  (adapter->flags & HAS_HW_TSTAMP) &&
  !adapter->tx_hwtstamp_skb

There isn't much point in wrapping the entire statement in unlikely, by
wrapping just the first test the rest of the conditional checks can be
pushed into the branch which will likely require a jump to some remote
block of code and should streamline the hot-path.

Thanks,

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