Re: wl12xx: third submission

[Kalle Valo <kalle.valo@iki.fi>]

Johannes Berg <johannes@sipsolutions.net> writes:

> On Thu, 2009-04-23 at 20:49 +0300, Kalle Valo wrote:
>
>> No problem, I can update wl12xx as soon as you finalise the interface.
>> Just let me know.
>
> Well, I just posted it earlier as [PATCH] :)

Thanks, I'll start working on it tomorrow.

>> > how about IPv4 alignment? Could you read the packet header first,
>> > memmove that and then read the rest, i.e. do two wl12xx_spi_mem_read
>> > calls? Might or might not be more efficient...
>> 
>> SPI access add latency, it would be faster to read as much as possible
>> in one transaction.
>
> Question is how it measures against the CPU doing memmove() on the
> entire frame when necessary.

I don't have any numbers, but for example omap2_mcspi is scheduling
processes etc. so I would say that memmove() is cheaper. Does anyone
have more knowledge about this?

>> Is there anything which in your opinion prevents inclusion to
>> wireless-testing?
>
> Give that you and Bob are working on it, I don't think so. Except the
> API issue, which means it won't compile when John merges my patch.

The compilation fix is easy to solve, I'll take care of it.

Thanks for your comments.

-- 
Kalle Valo