Caitlin Bestler wrote:
On 6/4/05, Christoph Hellwig <[EMAIL PROTECTED]> wrote:
The Quadrics patch informed zero copy devices of new mappings, but does so when the device is not actively doing work. That latter piece of information is not available for an RDMA device. For normal devices the OS has a reasonable expectation on when the device is performing DMA, because it is doing so in response to a request that the OS knows about. For an RDMA device a DMA transfer can be in progress as as a result of an RDMA Read -- something the host processor knows nothing about.
The Quadrics devices can handle changes in page mappings whilst RDMA operations are in progress. If say a page is being evicted due to swap pressure, then the Quadrics NIC is first informed by an IOPROC invalidate call. This will cause any future device accesses to that page to fault and the network operation to stall. The Quadrics NIC will continue the RDMA once a new valid translation has been loaded by the OS. Cheers Addy. _______________________________________________ openib-general mailing list [email protected] http://openib.org/mailman/listinfo/openib-general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
