On Jan 1, 2011, at 6:52 PM, Jonathan A. Kollasch wrote: > Hi, > > I noticed the new USB_FRAG_DMA_WORKAROUND option. > I had previously seen this bug on amd64 during heavy umass(4) use on ehci(4). > Maybe we should enable this everywhere by default?
I first saw that on my mpc85xx port but I thought it was a local bug. Then Frank reported something similar on ofppc so I committed it to see if it made his problem go away (which it did). That made me think it is a PowerPC bus_dma bug. But it still confused me since the mpc85xx does not have a cache which is coherent in regards to DMA but ofppc does have a DMA coherent cache. But if you are seeing this on non-PowerPC, I have to wonder if the problem is in the USB code itself since amd64 also has a DMA coherent cache.
