On 08/31/22 16:39, Eric Blake wrote:
> As promised in the previous patch, also test the new
> nbd_opt_set_meta_context() API in Python, OCaml, and Golang.
>
> The fact that Go slams all unit tests into the same namespace is
> somewhat annoying; it required munging test 240 now that 250 wants to
>
On 08/31/22 16:39, Eric Blake wrote:
> Upstream NBD clarified (see NBD commit 13a4e33a8) that since
> NBD_OPT_LIST_META_CONTEXT is stateless on the server side, it is
> acceptable (but not mandatory) for servers to accept it without the
> client having pre-negotiated structured replies. We aren't
On Thu, Sep 01, 2022 at 04:01:39PM +0800, Ming Lei wrote:
> Maybe you should use one nbd disk, which has the closest code path
> with nbdublk.
Good idea - I have started a heavy test using nbd.ko as the
backing. Let's see what happens after 12 hours or more.
Rich.
--
Richard Jones,
On 08/31/22 21:51, Eric Blake wrote:
> Similar to nbd_supports_tls(), it is nice to know from a feature probe
> whether we are likely to have VSOCK support before even trying more
> expensive APIs like nbd_connect_uri("nbd+vsock://...").
>
> ---
>
>
On Wed, Aug 31, 2022 at 12:06:37PM +0100, Richard W.M. Jones wrote:
> On Wed, Aug 31, 2022 at 06:33:25PM +0800, Ming Lei wrote:
> > On Wed, Aug 31, 2022 at 11:22:04AM +0100, Richard W.M. Jones wrote:
> > >
> > > After running those tests for an hour or two I got a nasty kernel
> > > memory
On Wed, Aug 31, 2022 at 11:22:04AM +0100, Richard W.M. Jones wrote:
>
> After running those tests for an hour or two I got a nasty kernel
> memory corruption issue in the test VM, see attached.
Another one from running overnight.
Rich.
--
Richard Jones, Virtualization Group, Red Hat
On Thu, Sep 01, 2022 at 11:26:10AM +0800, Ming Lei wrote:
> On Wed, Aug 31, 2022 at 04:06:16PM +0100, Richard W.M. Jones wrote:
> > I have also added the changes to Fedora:
> >
> > https://koji.fedoraproject.org/koji/taskinfo?taskID=91468555
> >
> > Please consider becoming a Fedora packager so
On 08/31/22 16:39, Eric Blake wrote:
> Instead of open-coding a linked list, we can utilize our vector.h
> helpers. No semantic change intended.
> ---
> lib/internal.h | 17 +
> generator/states-newstyle-opt-meta-context.c | 36
>
On Wed, Aug 31, 2022 at 09:39:19AM -0500, Eric Blake wrote:
> Upstream NBD clarified (see NBD commit 13a4e33a8) that since
> NBD_OPT_LIST_META_CONTEXT is stateless on the server side, it is
> acceptable (but not mandatory) for servers to accept it without the
> client having pre-negotiated
On Thu, Sep 01, 2022 at 11:21:38AM +0200, Laszlo Ersek wrote:
> On 08/31/22 16:39, Eric Blake wrote:
> > Upstream NBD clarified (see NBD commit 13a4e33a8) that since
> > NBD_OPT_LIST_META_CONTEXT is stateless on the server side, it is
> > acceptable (but not mandatory) for servers to accept it
On Thu, Sep 01, 2022 at 05:17:14PM +0100, Richard W.M. Jones wrote:
> >
> > [*]It's easier to skip on server failure than to try and write an
> > nbdkit patch to add yet another --config feature probe just to depend
> > on new-enough nbdkit to gracefully probe in advance if the server
> > should
On Wed, Aug 31, 2022 at 09:39:20AM -0500, Eric Blake wrote:
> Add a new control knob nbd_set_request_meta_context(), modeled after
> the existing nbd_set_request_structured_replies(), to make it possible
> to skip the NBD_OPT_SET_META_CONTEXT half of the two-command sequence
> currently performed
On Thu, Sep 01, 2022 at 08:07:33AM -0500, Eric Blake wrote:
> On Thu, Sep 01, 2022 at 08:11:17AM +0200, Laszlo Ersek wrote:
> > On 08/31/22 21:51, Eric Blake wrote:
> > > Similar to nbd_supports_tls(), it is nice to know from a feature probe
> > > whether we are likely to have VSOCK support before
On Thu, Sep 01, 2022 at 12:14:40PM +0100, Richard W.M. Jones wrote:
> On Thu, Sep 01, 2022 at 04:01:39PM +0800, Ming Lei wrote:
> > Maybe you should use one nbd disk, which has the closest code path
> > with nbdublk.
>
> Good idea - I have started a heavy test using nbd.ko as the
> backing.
On Thu, Sep 01, 2022 at 11:21:38AM +0200, Laszlo Ersek wrote:
> On 08/31/22 16:39, Eric Blake wrote:
> > Upstream NBD clarified (see NBD commit 13a4e33a8) that since
> > NBD_OPT_LIST_META_CONTEXT is stateless on the server side, it is
> > acceptable (but not mandatory) for servers to accept it
On Wed, Aug 31, 2022 at 09:39:23AM -0500, Eric Blake wrote:
> Since NBD_OPT_LIST_META_CONTEXTS is stateless from the server's point
> of view, there is no reason to make it clear state on the client's
> side. Since we do not otherwise modify h->meta_contexts while
> processing
On Thu, Sep 01, 2022 at 08:11:17AM +0200, Laszlo Ersek wrote:
> On 08/31/22 21:51, Eric Blake wrote:
> > Similar to nbd_supports_tls(), it is nice to know from a feature probe
> > whether we are likely to have VSOCK support before even trying more
> > expensive APIs like
On Thu, Sep 01, 2022 at 11:04:09AM +0200, Laszlo Ersek wrote:
...
> >
> > Rather than open-code a cleanup loop on all error paths, I instead fix
> > the problem by adding a meta_valid witness that is only set to true in
> > select places. The obvious place is when handling NBD_REP_ACK; but it
On 09/01/22 11:21, Laszlo Ersek wrote:
> On 08/31/22 16:39, Eric Blake wrote:
>> Upstream NBD clarified (see NBD commit 13a4e33a8) that since
>> NBD_OPT_LIST_META_CONTEXT is stateless on the server side, it is
>> acceptable (but not mandatory) for servers to accept it without the
>> client having
On 08/31/22 16:39, Eric Blake wrote:
> If a server replies to NBD_OPT_SET_META_CONTEXT first with
> NBD_REP_META_CONTEXT, then with NBD_REP_ERR_*, we have already
> modified h->meta_contexts, but fail to clean it up before moving on to
> OPT_GO in the OPT_META_CONTEXT.CHECK_REPLY's default:
On Thu, Sep 01, 2022 at 05:29:21PM +0100, Richard W.M. Jones wrote:
> On Wed, Aug 31, 2022 at 09:39:20AM -0500, Eric Blake wrote:
> > Add a new control knob nbd_set_request_meta_context(), modeled after
> > the existing nbd_set_request_structured_replies(), to make it possible
> > to skip the
On Thu, Sep 01, 2022 at 09:42:25PM +0100, Richard W.M. Jones wrote:
> On Thu, Sep 01, 2022 at 02:55:45PM -0500, Eric Blake wrote:
> > On Thu, Sep 01, 2022 at 05:17:14PM +0100, Richard W.M. Jones wrote:
> > > >
> > > > [*]It's easier to skip on server failure than to try and write an
> > > >
On Thu, Sep 01, 2022 at 05:33:43PM +0100, Richard W.M. Jones wrote:
> > How about the following wording:
> >
> > Returns true if libnbd was compiled with support for the C
> > family of sockets, or false if not. Note that on the Linux
> > operating system, this returns true if there is
On Thu, Sep 01, 2022 at 03:38:34PM -0500, Eric Blake wrote:
> On Thu, Sep 01, 2022 at 05:29:21PM +0100, Richard W.M. Jones wrote:
> > On Wed, Aug 31, 2022 at 09:39:20AM -0500, Eric Blake wrote:
> > > Add a new control knob nbd_set_request_meta_context(), modeled after
> > > the existing
On Thu, Sep 01, 2022 at 03:46:23PM -0500, Eric Blake wrote:
> On Thu, Sep 01, 2022 at 09:53:11AM +0200, Laszlo Ersek wrote:
> > On 08/31/22 16:39, Eric Blake wrote:
> > > As promised in the previous patch, also test the new
> > > nbd_opt_set_meta_context() API in Python, OCaml, and Golang.
> > >
On Thu, Sep 01, 2022 at 05:25:55PM +0100, Richard W.M. Jones wrote:
> >
> > Can we introduce some xfuncname pattern for these "state machine" C
> > files? The STATE_MACHINE hunk header is totally useless. I'd like
> > "NEWSTYLE.OPT_META_CONTEXT.START". :)
>
> It took me a while to work out what
On Thu, Sep 01, 2022 at 02:55:45PM -0500, Eric Blake wrote:
> On Thu, Sep 01, 2022 at 05:17:14PM +0100, Richard W.M. Jones wrote:
> > >
> > > [*]It's easier to skip on server failure than to try and write an
> > > nbdkit patch to add yet another --config feature probe just to depend
> > > on
On Thu, Sep 01, 2022 at 09:53:11AM +0200, Laszlo Ersek wrote:
> On 08/31/22 16:39, Eric Blake wrote:
> > As promised in the previous patch, also test the new
> > nbd_opt_set_meta_context() API in Python, OCaml, and Golang.
> >
> > The fact that Go slams all unit tests into the same namespace is
>
28 matches
Mail list logo
