Re: [Libguestfs] [libnbd PATCH v4 01/25] block_status: Add some sanity checking of server lengths

On Wed, Aug 02, 2023 at 08:50:21PM -0500, Eric Blake wrote: > Previously, we had not been doing any validation of server extent > responses, which means a client query at an offset near the end of the > export can result in a buggy server sending a response longer than the > export length and potentially confusing the client. The NBD spec also > says that an extent length should be non-zero so that a successful > block status call makes progress. It is easy enough to track that the > server has not overflowed the export size, and that we ensure an error > on no progress even when the buggy server claims success. Since the > spec says a client should be prepared for a block status result to be > truncated, the client should not care whether the truncation happened > at the server or at libnbd after validating the server's response. > > In the process, this patch reorganizes some of the code so that early > exits are obvious, leading for less indentation in the success path. > > Adding this sanity checking now makes it easier for future patches to > do orthogonal support for a server's 32- or 64-bit reply, vs. a > client's 32- or 64-bit API call. Once 64-bit replies are in play, we > will additionally have to worry about a 64-bit reply that overflows a > 32-bit API callback without exceeding the exportsize. Similarly, > since nbd_get_size() already caps export size at 63 bits (based on > off_t limitations), we have guaranteed that a 64-bit API callback will > never see an extent length that could appear negative in a 64-bit > signed type (at least OCaml benefits from that guarantee, since its > only native 64-bit integer type is signed). > > Signed-off-by: Eric Blake <eblake(a)redhat.com&gt; > --- Acked-by: Richard W.M. Jones <rjones(a)redhat.com&gt;