> On Mon, Aug 04, 2008 at 06:48:22PM +0530, ext Gadiyar, Anand wrote: > > MUSB RTL v1.4 has a hardware issue which results in a DMA controller > > hang when TX and RX DMA channels are simultaneously enabled. This > > affects at least OMAP2430 and OMAP34XX. > > > > Since RX transfers are in Mode 0 and anyway result in one DMA interrupt > > per packet, we can use System DMA to unload the RX fifos. MUSB DMA can > > be used for all TX channels as before. > > > > Tested with full-duplex TX and RX transfers using g_ether. Runs for 24 > > hours without a hang. Without this patch, the hang occurs within minutes. > > I just don't like that amount of ifdefs, as we were discussing out of > the list we should use runtime checks where possible. > > Let's wait on this one since Anand will send another patch set later on > this week. Updated patch coming up. Run-time checks are still not implemented. It would be nice to have, but I fail to see on what basis I can have the runtime check. Suggestions are welcome. I've left in only two #ifdefs - one for the dma callback function, and the other for the call to omap_request_dma(). The other ifdefs are now gone - replaced with runtime checks based on whether a system DMA channel has been acquired. Also fixed a small bug which would make this patch not compile. - Anand -- To unsubscribe from this list: send the line "unsubscribe linux-omap" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
