Message ID | 1460002330-20965-1-git-send-email-eblake@redhat.com |
---|---|
State | New |
Headers | show |
On 7 Apr 2016, at 05:12, Eric Blake <eblake@redhat.com> wrote: > Commit 730c5830 included a reordering of the fields of > NBD_REP_SERVER, under the guise of putting the variable-sized > data last. However, this makes life harder for sharing > code - the regular NBD_REP_SERVER used by NBD_OPT_LIST > already has _two_ variable-sized fields: name, and an optional > string of additional information, and the wording of the > surrounding text implied that NBD_OPT_GO was just re-purposing > the tail end as binary data instead of a string, not > reorganizing all the fields. > > Furthermore, sticking the same fields last in both NBD_OPT_GO > and NBD_OPT_EXPORT_NAME means that any other future handshake > extension that sends a few more bytes of data behind the > transmission flags can do so in both places without reordering > the first half of NBD_REP_SERVER (such an extension is already > envisioned, where the client and server agree to have the server > advertise minimum, preferred, and maximum block sizes). > > Signed-off-by: Eric Blake <eblake@redhat.com> Reviewed-by: Alex Bligh <alex@alex.org.uk> Thanks for this. I was going to clear this up myself today having introduced the issue. Alex > --- > > I'm trying to implement NBD_OPT_GO in qemu, and found the original > order easier to support than the swapped order. > > doc/proto.md | 4 ++-- > 1 file changed, 2 insertions(+), 2 deletions(-) > > diff --git a/doc/proto.md b/doc/proto.md > index 7b36be4..0775360 100644 > --- a/doc/proto.md > +++ b/doc/proto.md > @@ -744,13 +744,13 @@ Therefore these commands share common documentation. > in place of the default UTF-8 free-form string). The option reply length > MUST be *length of name* + 14, and the option data has the following layout: > > - - 64 bits, size of the export in bytes (unsigned) > - - 16 bits, transmission flags. > - 32 bits, length of name (unsigned) > - Name of the export. This name MAY be different from the one > given in the `NBD_OPT_INFO` or `NBD_OPT_GO` option in case the > server has multiple alternate names for a single export, or a > default export was specified. > + - 64 bits, size of the export in bytes (unsigned) > + - 16 bits, transmission flags. > > The server MUST NOT fail an NBD_OPT_GO sent with the same parameters > as a previous NBD_OPT_INFO which returned successfully (i.e. with > -- > 2.5.5 > > > ------------------------------------------------------------------------------ > _______________________________________________ > Nbd-general mailing list > Nbd-general@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/nbd-general >
diff --git a/doc/proto.md b/doc/proto.md index 7b36be4..0775360 100644 --- a/doc/proto.md +++ b/doc/proto.md @@ -744,13 +744,13 @@ Therefore these commands share common documentation. in place of the default UTF-8 free-form string). The option reply length MUST be *length of name* + 14, and the option data has the following layout: - - 64 bits, size of the export in bytes (unsigned) - - 16 bits, transmission flags. - 32 bits, length of name (unsigned) - Name of the export. This name MAY be different from the one given in the `NBD_OPT_INFO` or `NBD_OPT_GO` option in case the server has multiple alternate names for a single export, or a default export was specified. + - 64 bits, size of the export in bytes (unsigned) + - 16 bits, transmission flags. The server MUST NOT fail an NBD_OPT_GO sent with the same parameters as a previous NBD_OPT_INFO which returned successfully (i.e. with
Commit 730c5830 included a reordering of the fields of NBD_REP_SERVER, under the guise of putting the variable-sized data last. However, this makes life harder for sharing code - the regular NBD_REP_SERVER used by NBD_OPT_LIST already has _two_ variable-sized fields: name, and an optional string of additional information, and the wording of the surrounding text implied that NBD_OPT_GO was just re-purposing the tail end as binary data instead of a string, not reorganizing all the fields. Furthermore, sticking the same fields last in both NBD_OPT_GO and NBD_OPT_EXPORT_NAME means that any other future handshake extension that sends a few more bytes of data behind the transmission flags can do so in both places without reordering the first half of NBD_REP_SERVER (such an extension is already envisioned, where the client and server agree to have the server advertise minimum, preferred, and maximum block sizes). Signed-off-by: Eric Blake <eblake@redhat.com> --- I'm trying to implement NBD_OPT_GO in qemu, and found the original order easier to support than the swapped order. doc/proto.md | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)