[ofa-general] [RFC 0/1] libmthca: CQ/DMA race on Altix

[Muli Ben-Yehuda]

On Mon, Jul 16, 2007 at 02:47:28PM -0700, Roland Dreier wrote:

>  - Add a flag that gets passed in the normal "direction" parameter
>    of the DMA mapping APIs, which is ignored on most systems and not
>    set by most drivers.  Adds some churn to the internal
>    implementation on all archs, though.

This will potentially break a bunch of things that assume the only
valid values of direction are NONE, TO_DEVICE, FROM_DEVICE, or BOTH
(e.g., include/linux/dma-mapping.h:valid_dma_direction()).

> or
> 
>  - Add new functions dma_map_single_flushing(),
>    dma_map_sg_flushing() and dma_map_page_flushing() that are
>    defined to be the same as the non-flushing variants except on
>    Altix.  Fairly quick and easy to implement, but arguably makes
>    the DMA API even more bloated with even more functions that are
>    only slightly different.
> 
> Dunno...

Looks like we need to harness the collective power of lkml. Better
hope everyone isn't distracted by the flamewar-du-jour.

Cheers,
Muli