On Thursday 04 May 2006 10:46 pm, Nathan Becker wrote: > Yes, GART_IOMMU is already turned on. Do you want me to send more > detailed debugging messages?
And when you tell the kernel to only use 2 GB, it works again right? Just trying to rule out hardware or hardware-init bugs. Maybe you could change both dma masks to be 31 bits and then force the IOMMU active, so it gives out addresses below 2GB. You should be able to make it so that the EHCI controller never sees addresss with the high bit set, making your 4 GB config act like a 2 GB one (via IOMMU). Which one hopes means "working" ... it'd be good to know a workaround! This isn't neceessarily a variant of that known NForce silicon bug, but you could help confirm that by changing the "3strikes" message so it prints out urb->transfer_dma ... and maybe make qh_completions() dump the dma addresses of the qh and qtd too (iff urb->status becomes -EPROTO). ------------------------------------------------------- Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642 _______________________________________________ linux-usb-devel@lists.sourceforge.net To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
