Re: [PATCH] mlx4_en: fix transmit of packages when blue frame is enabled

On Mon, Oct 10, 2011 at 11:01:24AM +0200, Benjamin Herrenschmidt wrote:

The case where things get a bit more nasty is when you try to use MMIO
for low latency small-data type transfers instead of DMA, in which case
you do want the ability for the chipset to write-combine and control the
barriers more precisely.

However, this is hard and Linux doesn't provide very good accessors to
do so, thus you need to be extra careful (see my example about wmb()

In the case of the iomap "copy" operations, my problem is that they
don't properly advertise their lack of ordering since normal iomap does
have full ordering.

I believe they should provide ordering with a barrier before & a barrier
after, eventually with _relaxed variants or _raw variants for those who
"know what they are doing".

Until then I think we need to have the logic working right on ppc and
measure if blue flame buys us any improvement in ppc. If that's not
the case (e.g because write combining is not working), then maybe we
should avoid using blueflame in ppc.
Could any of the guys from IBM check this and give us feedback?

Maybe it's time for us to revive those discussions about providing a
good set of relaxed MMIO accessors with explicit barriers :-)

Cheers,
Ben.