[PATCH] Re allocate the RX DMA buffer to avoid skb conflict Handle all RX packets in the DMA ring buffer

[YanBo]

On Tue, May 28, 2013 at 3:07 PM, YanBo <dreamfly281 at gmail.com> wrote:
> On Tue, May 28, 2013 at 2:55 PM, Eugene Krasnikov <k.eugene.e at gmail.com> wrote:
>>> In my testing, the irq_tx_complete90 will not be triggered after TX
>>> fire, but the TX status has to be
>>> clear in order to continue TX, then it just moved to RX handle to as a
>>> workaround to fix this issue.
>>>
>>> That is indeed need be move to TX handler when we get the irq_tx_complete works
>>
>> irq_tx_complete is not generated because it's disabled completely
>> right now. That is also one of the reasons why we a leaking memory as
>> Pontus stated before.
>> Do not think it's a good workaround because there can be a situation
>> when there is no RX packets. Let's better implement a proper TX
>> handling first.
>>
>>> but will not include in this patch. what your opinion?
>>
>> Let's split it in couple of patches:
>> 1) Implement tx_comp status
>> 2) Fix dmamap/unmap
>> 3) Process more then one RX packet per interrupt
>> 4) Refill RX skb
>> 5) Clear channel status register
>>
> 2)&4)should be together cause the unmap is used for refill the rx skb
>
> 1)&5) can be together cause clear the TX status  is one of part of tx_comp
>
> 3) can be separately theoretically, but consider this has been combine
> with 2)4) and reviewed for several times,
> I just don't want waste time to separate them again and then combine
> them, it has not too much sense now although I think you are
> right about  writing the patch as small as possible to include just a
> simple function,
>
> I'll resend a patch include 2)&)4  and 3)  for review and submit
> another patch for 1)&5) later
>
Has send the patch named for 2)&)4  and 3), please help to review, thanks

BR /Yanbo