Re: [Qemu-devel] [PATCH 2/2] qapi: deprecate implicit filters

[Kevin Wolf <kwolf@redhat.com>]

Am 15.08.2019 um 18:07 hat John Snow geschrieben:
> 
> 
> On 8/15/19 6:49 AM, Kevin Wolf wrote:
> > Am 14.08.2019 um 21:27 hat John Snow geschrieben:
> >>
> >>
> >> On 8/14/19 6:07 AM, Vladimir Sementsov-Ogievskiy wrote:
> >>> To get rid of implicit filters related workarounds in future let's
> >>> deprecate them now.
> >>>
> >>> Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
> >>> ---
> >>>  qemu-deprecated.texi      |  7 +++++++
> >>>  qapi/block-core.json      |  6 ++++--
> >>>  include/block/block_int.h | 10 +++++++++-
> >>>  blockdev.c                | 10 ++++++++++
> >>>  4 files changed, 30 insertions(+), 3 deletions(-)
> >>>
> >>> diff --git a/qemu-deprecated.texi b/qemu-deprecated.texi
> >>> index 2753fafd0b..8222440148 100644
> >>> --- a/qemu-deprecated.texi
> >>> +++ b/qemu-deprecated.texi
> >>> @@ -183,6 +183,13 @@ the 'wait' field, which is only applicable to sockets in server mode
> >>>  
> >>>  Use blockdev-mirror and blockdev-backup instead.
> >>>  
> >>> +@subsection implicit filters (since 4.2)
> >>> +
> >>> +Mirror and commit jobs inserts filters, which becomes implicit if user
> >>> +omitted filter-node-name parameter. So omitting it is deprecated, set it
> >>> +always. Note, that drive-mirror don't have this parameter, so it will
> >>> +create implicit filter anyway, but drive-mirror is deprecated itself too.
> >>> +
> >>>  @section Human Monitor Protocol (HMP) commands
> >>>  
> >>>  @subsection The hub_id parameter of 'hostfwd_add' / 'hostfwd_remove' (since 3.1)
> >>> diff --git a/qapi/block-core.json b/qapi/block-core.json
> >>> index 4e35526634..0505ac9d8b 100644
> >>> --- a/qapi/block-core.json
> >>> +++ b/qapi/block-core.json
> >>> @@ -1596,7 +1596,8 @@
> >>>  # @filter-node-name: the node name that should be assigned to the
> >>>  #                    filter driver that the commit job inserts into the graph
> >>>  #                    above @top. If this option is not given, a node name is
> >>> -#                    autogenerated. (Since: 2.9)
> >>> +#                    autogenerated. Omitting this option is deprecated, it will
> >>> +#                    be required in future. (Since: 2.9)
> >>>  #
> >>>  # @auto-finalize: When false, this job will wait in a PENDING state after it has
> >>>  #                 finished its work, waiting for @block-job-finalize before
> >>> @@ -2249,7 +2250,8 @@
> >>>  # @filter-node-name: the node name that should be assigned to the
> >>>  #                    filter driver that the mirror job inserts into the graph
> >>>  #                    above @device. If this option is not given, a node name is
> >>> -#                    autogenerated. (Since: 2.9)
> >>> +#                    autogenerated. Omitting this option is deprecated, it will
> >>> +#                    be required in future. (Since: 2.9)
> >>>  #
> >>>  # @copy-mode: when to copy data to the destination; defaults to 'background'
> >>>  #             (Since: 3.0)
> >>> diff --git a/include/block/block_int.h b/include/block/block_int.h
> >>> index 3aa1e832a8..624da0b4a2 100644
> >>> --- a/include/block/block_int.h
> >>> +++ b/include/block/block_int.h
> >>> @@ -762,7 +762,15 @@ struct BlockDriverState {
> >>>      bool sg;        /* if true, the device is a /dev/sg* */
> >>>      bool probed;    /* if true, format was probed rather than specified */
> >>>      bool force_share; /* if true, always allow all shared permissions */
> >>> -    bool implicit;  /* if true, this filter node was automatically inserted */
> >>> +
> >>> +    /*
> >>> +     * @implicit field is deprecated, don't set it to true for new filters.
> >>> +     * If true, this filter node was automatically inserted and user don't
> >>> +     * know about it and unprepared for any effects of it. So, implicit
> >>> +     * filters are workarounded and skipped in many places of the block
> >>> +     * layer code.
> >>> +     */
> >>> +    bool implicit;
> >>>  
> >>>      BlockDriver *drv; /* NULL means no media */
> >>>      void *opaque;
> >>> diff --git a/blockdev.c b/blockdev.c
> >>> index 36e9368e01..b3cfaccce1 100644
> >>> --- a/blockdev.c
> >>> +++ b/blockdev.c
> >>> @@ -3292,6 +3292,11 @@ void qmp_block_commit(bool has_job_id, const char *job_id, const char *device,
> >>>      BlockdevOnError on_error = BLOCKDEV_ON_ERROR_REPORT;
> >>>      int job_flags = JOB_DEFAULT;
> >>>  
> >>> +    if (!has_filter_node_name) {
> >>> +        warn_report("Omitting filter-node-name parameter is deprecated, it "
> >>> +                    "will be required in future");
> >>> +    }
> >>> +
> >>>      if (!has_speed) {
> >>>          speed = 0;
> >>>      }
> >>> @@ -3990,6 +3995,11 @@ void qmp_blockdev_mirror(bool has_job_id, const char *job_id,
> >>>      Error *local_err = NULL;
> >>>      int ret;
> >>>  
> >>> +    if (!has_filter_node_name) {
> >>> +        warn_report("Omitting filter-node-name parameter is deprecated, it "
> >>> +                    "will be required in future");
> >>> +    }
> >>> +
> >>>      bs = qmp_get_root_bs(device, errp);
> >>>      if (!bs) {
> >>>          return;
> >>>
> >>
> >> This might be OK to do right away, though.
> >>
> >> I asked Markus this not too long ago; do we want to amend the QAPI
> >> schema specification to allow commands to return with "Warning" strings,
> >> or "Deprecated" stings to allow in-band deprecation notices for cases
> >> like these?
> >>
> >> example:
> >>
> >> { "return": {},
> >>   "deprecated": True,
> >>   "warning": "Omitting filter-node-name parameter is deprecated, it will
> >> be required in the future"
> >> }
> >>
> >> There's no "error" key, so this should be recognized as success by
> >> compatible clients, but they'll definitely see the extra information.
> >>
> >> Part of my motivation is to facilitate a more aggressive deprecation of
> >> legacy features by ensuring that we are able to rigorously notify users
> >> through any means that they need to adjust their scripts.
> > 
> > Who would read this, though? In the best case it ends up deep in a
> > libvirt log that nobody will look at because there was no error. In the
> > more common case, the debug level is configured so that QMP traffic
> > isn't even logged.
> > 
> > Kevin
> > 
> 
> I believe you are right, but I also can't shake the feeling that this
> attitude ensures that we'll never find a way to expose this information
> to the end-user. Is this not too defeatist?

I think the discussed approach that seemed most likely to me to succeed
was adding a command line option that makes QEMU just crash if you use a
deprecated feature, and enable that in libvirt test cases (or possibly
even any non-release builds, though maybe it's a bit harsh there).

> I think deprecation notices in the QMP stream has two benefits:
> 
> 1) Any direct usages via qmp-shell or manual JSON connection are likely
> to see this message in development or testing. I feel the usage of QEMU
> directly is more likely to increase with time as other stacks seek to
> work around libvirt.
> 
> [Whether or not they should is another question, but I believe the
> current reality to be that people are trying to.]

I don't know about other people, but as a human user, I don't care about
deprecation notices. As long as something works, I use it, and once I
get an error message back, I'll use something else.

If I manually enter drive_mirror and get a warning back, that doesn't
tell me that libvirt still does the same thing and needs to be fixed. It
just tells me that in the future I might need to change the commands
that I use manually.

I guess this would still prevent adding new libvirt features that build
on deprecated QEMU features because some manual testing will be involved
there. But was this ever a problem?

> 2) Programmatic deprecation notices can't be presented to a user at all
> if we don't send them; at least this way it becomes libvirt's problem
> over what to do with them. Perhaps even just in testing and regression
> suites libvirt can assert that it sees no deprecation warnings (or
> whitelist certain ones it knows about.)
> 
> In the case of libvirt, it's not even necessarily about making sure the
> end user sees it, because it isn't even necessarily the user's fault --
> it's libvirt's. This is a sure-fire programmatic way to communicate
> compatibility changes to libvirt.

If libvirt uses this to make test cases fail, it could work.

Kevin