similar to: [nbdkit PATCH v2] main: Add option to disable SR advertisement

Similar to --no-sr, it can be handy for testing a client implementation to have a server that can easily be forced into older behaviors, without having to recompile a one-off hack into a server or dig up an older server binary that lacked a newer feature. To see the patch in action, try things like: $ ./nbdkit -U - -fv --mask-handshake=0 null \ --run 'qemu-nbd --list -k $unixsocket'

When we added support for .extents, we had nbdkit unconditionally support structured replies if the client requests them, and the plugin's .can_extents has no impact on what the server advertises. However, while the plugin API doesn't care whether the client requested SR, there is still a case to be made for allowing a filter to prevent SR, at least for testing purposese (such as

This is similar to the recent --no-sr option - it's a change that is unlikely to ever be used except by someone testing whether a client is compliant to the protocol, but in that niche case, it can be quite handy (it's a lot nicer to be able to purposefully cripple a server from the command line than from a one-off compile, when testing if a client's fallback for a spec-compliant but

This parameter provided a name for the "default export" -- ie. the one and only export returned to the client by NBD_OPT_LIST. But nbdkit traditionally didn't care what export name the client requested. Since 1.16 plugins have been able to serve different content per export name (and return errors for unknown exports), but the -e option didn't reflect that and only created

For mainly historical reasons we tended to use int to store boolean values. However using bool is probably safer in some corner cases (eg. ?v == true? can fail badly if v is an int, but works for bool). bool was added in C99 so let's use it. --- server/internal.h | 16 +++++------ server/connections.c | 28 +++++++++--------- server/crypto.c | 4 +-- server/main.c | 67

Prefer named constants over magic numbers in the 'struct option' list, and expand the list of enums for long-only options so that the call to getopt_long() can switch directly to every option, rather than needing a lengthy if/else chain that grows for every new long option. Patch best viewed with whitespace changes ignored. Signed-off-by: Eric Blake <eblake@redhat.com> ---

Until the next few patches expose and implement new callbacks for filters and plugins, our initial implementation for NBD_CMD_CACHE is to just blindly advertise the feature, and call into .pread with an ignored buffer. Note that for bisection reasons, this patch treats any use of a filter as a forced .can_cache of false to bypass the filter's caching; once all affected filters are patched to

Allow plugins to affect the handling of the new NBD_CMD_FLAG_FAST_ZERO flag, then update affected plugins. In particular, in-memory plugins are always fast; the full plugin is better served by omitting .zero and relying on .pwrite fallback; nbd takes advantage of libnbd extensions proposed in parallel to pass through support; and v2 language bindings expose the choice to their scripts. The

Similar to the existing fua, nozero and noextents filters, add a filter to make it easy to override the basic caching functionality, in part to facilitate timing tests of whether a plugin's cache implementation is worthwhile. A worthwhile test to add to the testsuite would connect the log filter both before and after the nocache filter, to prove how caching requests are altered. However,

nbdkit <= 1.1.28 defaulted to the oldstyle protocol for compatibility with qemu and libguestfs. However qemu >= 2.6 can now work with either protocol and is widely installed. Also newstyle is required for newer features such as export names and TLS. In addition nbd-client dropped support for oldstyle entirely. You can select the oldstyle protocol by adding ‘-o’, and it is still tested.

While we are unlikely to change our decision that -e should not control our response to NBD_OPT_LIST (because we intend to add a new callback .extents_list for that), it turns out that it is a lot easier to write: nbdkit -U - -e foo info --run 'nbdsh -u "$uri" -c "print(h.pread(3, 0))"' than it is to write the equivalent: nbdkit -U - info --run 'nbdsh -u

To allow filters to modify the export name as it passes through the layers this commit makes several changes: The filter .open callback now takes an extra parameter, the export name. This is always non-NULL (for oldstyle it is ""). This string has a short lifetime and filters that need to hang on to it must take a copy. The filter must pass the exportname parameter down to the next

On 9/10/19 6:20 AM, Richard W.M. Jones wrote: > This allows plugins (or filters) to read the export name which was > passed to the server from the client. > > UNFINISHED: > > - Needs tests > --- > TODO | 8 ++++++ > docs/nbdkit-plugin.pod | 39 ++++++++++++++++++++++++++++ > server/connections.c | 1 +

In commit c306fa93ab and neighbors (v1.15.1), a concerted effort went into caching the results of .can_FOO callbacks, with commit messages demonstrating that a plugin with a slow callback should not have that delay magnified multiple times. But nothing was added to the testsuite at the time, and with the sh and eval plugins, we still have two scenarios where we did not take advantage of the

This commit adds a new API which can be used from the connected to state to read back which NBD protocol (eg. oldstyle, newstyle-fixed) we are using. It was helpful to add a new state in newstyle negotiation (%NEWSTYLE.FINISHED) so we can route all successful option negotiations through a single path before moving to the %READY state, allowing us to set h->protocol in one place. ---

The code handling the NBD protocol was located in the same file as the code handling connections, for not really any reason except historical. This is quite a large code movement which splits out the protocol code into four new files: server/protocol-handshake.c initial handshake server/protocol-handshake-newstyle.c " " (newstyle)

CentOS 6 has libguestfs-devel 1.20.11, which predates the support in guestfs_add_drive_opts() for requesting an nbd drive instead of a local file. The guestfs plugin can still be built, so no configure changes are needed; but skip the tests that fail to compile so that 'make check' can at least complete. Signed-off-by: Eric Blake <eblake@redhat.com> --- Even with this patch,

This allows plugins (or filters) to read the export name which was passed to the server from the client. UNFINISHED: - Needs tests --- TODO | 8 ++++++ docs/nbdkit-plugin.pod | 39 ++++++++++++++++++++++++++++ server/connections.c | 1 + server/internal.h | 1 + server/protocol-handshake-newstyle.c | 30

Diff against libnbd’s copy of this file, and change this one until it matches. --- common/protocol/nbd-protocol.h | 76 +++++++++++++++++----------- server/protocol-handshake-newstyle.c | 26 +++++----- server/protocol-handshake-oldstyle.c | 4 +- server/protocol.c | 25 ++++----- tests/test-layers.c | 14 ++--- 5 files changed, 81 insertions(+), 64

We discovered a possible Downgrade Attack in libnbd. Lifecycle --------- Reported: 2019-09-14 Fixed: 2019-09-16 Published: 2019-09-16 There is no CVE number assigned for this issue yet, but the bug is being categorized and processed by Red Hat's security team which may result in a CVE being published later. Description ----------- Libnbd includes the method nbd_set_tls(h,