From: Kristoffer Glembo <kristoffer@xxxxxxxxxxx> Date: Fri, 27 Nov 2009 17:43:35 +0100 > I'm adding LEON support for the DMA-API into ioport.c and at the > same time wanting to clean things up somewhat. > > One thing that bugs me a little bit is this one: > > #define mmu_inval_dma_area(p, l)/* Anton pulled it out for 2.4.0-xx */ > > It seems this was never implemented. I'm not sure which machines needs > to invalidate the cache after DMA but LEON needs it in some > configurations and as far as I understand the microsparc IIep needs it > as well. I'm not so familiar with where this chip is used but the > JavaStation (Krups) uses it and has the Sun HappyMeal ethernet card > doing DMA from the PCI bus. > > Can anyone explain why this was never implemented and instead removed? > Is it not needed even on the IIep ? I don't know offhand, sorry. But I seem to remember that Microsparc-IIep is coherent all the way to the L1 caches with DMA so it wouldn't need flushes after DMA at all. -- To unsubscribe from this list: send the line "unsubscribe sparclinux" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html