"Michael S. Tsirkin" <m...@redhat.com> writes: > On Mon, Jan 21, 2013 at 10:22:03PM +1030, Rusty Russell wrote: >> Rusty Russell <ru...@rustcorp.com.au> writes: >> > "Michael S. Tsirkin" <m...@redhat.com> writes: >> >>> + iov->iov[iov->i].iov_base = (__force __user void *)addr; >> >>> + iov->iov[iov->i].iov_len = desc.len; >> >> >> >> The following comment from the previous version still applies: >> >> > This looks like it won't do the right thing if desc.len spans multiple >> >> > ranges. I don't know if this happens in practice but this is something >> >> > vhost supports ATM. >> >> in otgher words, we might need to split a single desc to multiple >> >> iov entries. >> > >> > Ah, separate offsets for consecutive ranges, right. I'd prefer to say >> > "don't do that", but qemu is rarely sane. I'll fix it. >> >> Actually, you make the same assumption for vhost, with your use of >> getuser and putuser for accessing the ring. > > There's no requirement that ring is mapped directly into guest > memory. If a ring is not contigious qemu can allocate > it's own virtuall contigious rings and copy data back and forth.
True, but it's the guest which allocates the ring. If QEMU sets up a guest with a offset-discontiguous mapping, vhost would be unreliable today. >> The complexity and cost of handling this is significant, > > Why is it? Just add a while loop, increment desc.addr > and decrement desc.len. Indirect handling now needs to re-check the boundaries for every descriptor it fetches, and handle the case where a single descriptor overlaps the boundary. The ability to verify the indirect descriptor table all at once makes the code fast and neat. >From your other mail: > It seems it could happen if e.g. one region is ROM and another one is > RAM. This means it won't be used for the ring (which is R/W) but could > be for the data. That case wouldn't be so bad, I think, since it's a simple loop. I'll create some patches and see if it's too ugly to live... Thanks, Rusty. _______________________________________________ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
