From: Sergei Shtylyov
> Sent: 24 April 2015 19:27
...
> > If you have ethernet hardware that requires tx or rx buffers to be on
> > 4n boundaries you should send it back as 'not fit for purpose'.
>
> The RX buffers can be adjusted with skb_resrerve(), it's only the TX
> buffers that need to be copied...
If the processor can't perform misaligned reads (don't know what is on
your SoC, but I suspect it can't - crossing page boundaries is hard)
then the rx buffer will have to be re-aligned in software.
Even the 'userdata' part will typically end up with an expensive
misaligned buffer copy.
Even on x86 the misaligned transfers are probably measurable.
> I'm afraid we can't. :-)
> However, my colleague has suggested a scheme minimizing the copying:
> only up to 3 first bytes need to be copied to the driver's internal buffers,
> the rest can be sent from an skb itself. That would require substantial
> changes to the driver though...
There might be a restriction on the length of buffer fragments.
You might be able to alternate 14 and 1500+ byte receive buffers.
The frame following a slightly overlong one would be 'wrong'.
David
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
