Re: [RFC PATCH] dma: Add barrierless dma mapping/unmapping api

From: Randy Dunlap
Date: Tue Jan 26 2010 - 11:50:06 EST


On Mon, 25 Jan 2010 18:35:46 -0800 adharmap@xxxxxxxxxxxxxx wrote:

> From: Abhijeet Dharmapurikar <adharmap@xxxxxxxxxxx>
>
> Tests have shown that there is a significant performance gain
> when a driver uses mapping/unmapping functions without a barrier
> multiple times and calls dsb() only after the last buffer. This
> api adds support for barrierless dma operations on ARMv6
> and ARMv7.
>
> Signed-off-by: Abhijeet Dharmapurikar <adharmap@xxxxxxxxxxxxxx>
> ---
>
> Please refer to the earlier post here http://lkml.org/lkml/2010/1/4/347.
> These changes only are only for the arm v6 and v7
> architecture and introduce dma_map/unmap_single_nobarrier api.
> The idea is to extend the cpu_cache_fns.dma_.*_range functions to accept
> a third boolean argument that tells it whether to execute the barrier or
> skip it. Note that this applies over Linux 2.6.33-rc5 and needs a few changes
> to apply on the linux-next tree.
>
>
> arch/arm/include/asm/cacheflush.h | 12 ++++----
> arch/arm/include/asm/dma-mapping.h | 57 ++++++++++++++++++++++++++++++++++--
> arch/arm/mm/cache-v6.S | 15 ++++++----
> arch/arm/mm/cache-v7.S | 19 +++++++++++-
> arch/arm/mm/dma-mapping.c | 17 ++++++-----
> drivers/staging/dream/pmem.c | 4 +-
> 6 files changed, 97 insertions(+), 27 deletions(-)


Along with the other comments, the patch is also missing updates
to Documentation/DMA-API.txt ...


---
~Randy
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/