Re: [PATCH v2 07/11] block: add x-blockdev-amend qmp command

[Maxim Levitsky]

On Fri, 2019-10-04 at 20:53 +0200, Max Reitz wrote:
> On 13.09.19 00:30, Maxim Levitsky wrote:
> > Signed-off-by: Maxim Levitsky <address@hidden>
> > ---
> >  block/Makefile.objs       |   2 +-
> >  block/amend.c             | 116 ++++++++++++++++++++++++++++++++++++++
> >  include/block/block_int.h |  23 ++++++--
> >  qapi/block-core.json      |  26 +++++++++
> >  qapi/job.json             |   4 +-
> >  5 files changed, 163 insertions(+), 8 deletions(-)
> >  create mode 100644 block/amend.c
> 
> I think I’d move this (and everything to belongs to it) to a different
> series.
I already kind of do this, patches prior to this one fully implement
the existing amend code path, while this and patches after this
one implement the qmp x-blockdev-amend code path.

i don't mind sending this as two separate patch series, now that
first refactoring patch series is committed upsteam.


> 
> > diff --git a/block/Makefile.objs b/block/Makefile.objs
> > index 35f3bca4d9..10d0308792 100644
> > --- a/block/Makefile.objs
> > +++ b/block/Makefile.objs
> > @@ -18,7 +18,7 @@ block-obj-y += block-backend.o snapshot.o qapi.o
> >  block-obj-$(CONFIG_WIN32) += file-win32.o win32-aio.o
> >  block-obj-$(CONFIG_POSIX) += file-posix.o
> >  block-obj-$(CONFIG_LINUX_AIO) += linux-aio.o
> > -block-obj-y += null.o mirror.o commit.o io.o create.o
> > +block-obj-y += null.o mirror.o commit.o io.o create.o amend.o
> >  block-obj-y += throttle-groups.o
> >  block-obj-$(CONFIG_LINUX) += nvme.o
> >  
> > diff --git a/block/amend.c b/block/amend.c
> > new file mode 100644
> > index 0000000000..9bd28e08e7
> > --- /dev/null
> > +++ b/block/amend.c
> > @@ -0,0 +1,116 @@
> > +/*
> > + * Block layer code related to image options amend
> > + *
> > + * Copyright (c) 2018 Kevin Wolf <address@hidden>
> > + * Copyright (c) 2019 Maxim Levitsky <address@hidden>
> > + *
> > + * Heavily based on create.c
> > + *
> > + * Permission is hereby granted, free of charge, to any person obtaining a 
> > copy
> > + * of this software and associated documentation files (the "Software"), 
> > to deal
> > + * in the Software without restriction, including without limitation the 
> > rights
> > + * to use, copy, modify, merge, publish, distribute, sublicense, and/or 
> > sell
> > + * copies of the Software, and to permit persons to whom the Software is
> > + * furnished to do so, subject to the following conditions:
> > + *
> > + * The above copyright notice and this permission notice shall be included 
> > in
> > + * all copies or substantial portions of the Software.
> > + *
> > + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS 
> > OR
> > + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
> > + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
> > + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR 
> > OTHER
> > + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING 
> > FROM,
> > + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS 
> > IN
> > + * THE SOFTWARE.
> > + */
> > +
> > +#include "qemu/osdep.h"
> > +#include "block/block_int.h"
> > +#include "qemu/job.h"
> > +#include "qemu/main-loop.h"
> > +#include "qapi/qapi-commands-block-core.h"
> > +#include "qapi/qapi-visit-block-core.h"
> > +#include "qapi/clone-visitor.h"
> > +#include "qapi/error.h"
> > +
> > +typedef struct BlockdevAmendJob {
> > +    Job common;
> > +    BlockdevCreateOptions *opts;
> > +    BlockDriverState *bs;
> > +    bool force;
> > +} BlockdevAmendJob;
> > +
> > +static int coroutine_fn blockdev_amend_run(Job *job, Error **errp)
> > +{
> > +    BlockdevAmendJob *s = container_of(job, BlockdevAmendJob, common);
> > +    int ret;
> > +
> > +    job_progress_set_remaining(&s->common, 1);
> > +    ret = s->bs->drv->bdrv_co_amend(s->bs, s->opts, s->force, errp);
> > +    job_progress_update(&s->common, 1);
> 
> It would be nice if the amend job could make use of the progress
> reporting that we have in place for amend.

I also thought about it, but is it worth it?

I looked through the status reporting of the qcow2 amend
code (which doesn't really allowed to be run through
qmp blockdev-amend, due to complexity of changing 
the qcow2 format on the fly).



> 
> > +
> > +    qapi_free_BlockdevCreateOptions(s->opts);
> > +
> > +    return ret;
> > +}
> > +
> > +static const JobDriver blockdev_amend_job_driver = {
> > +    .instance_size = sizeof(BlockdevAmendJob),
> > +    .job_type      = JOB_TYPE_AMEND,
> > +    .run           = blockdev_amend_run,
> > +};
> > +
> > +void qmp_x_blockdev_amend(const char *job_id,
> > +                        const char *node_name,
> > +                        BlockdevCreateOptions *options,
> > +                        bool has_force,
> > +                        bool force,
> > +                        Error **errp)
> > +{
> > +    BlockdevAmendJob *s;
> > +    const char *fmt = BlockdevDriver_str(options->driver);
> > +    BlockDriver *drv = bdrv_find_format(fmt);
> > +    BlockDriverState *bs = bdrv_find_node(node_name);
> > +
> > +    /*
> > +     * If the driver is in the schema, we know that it exists. But it may 
> > not
> > +     * be whitelisted.
> > +     */
> > +    assert(drv);
> > +    if (bdrv_uses_whitelist() && !bdrv_is_whitelisted(drv, false)) {
> > +        error_setg(errp, "Driver is not whitelisted");
> > +        return;
> > +    }
> > +
> > +    if (bs->drv != drv) {
> > +        error_setg(errp,
> > +                   "x-blockdev-amend doesn't support changing the block 
> > driver");
> > +        return;
> > +
> > +    }
> > +
> > +    /* Error out if the driver doesn't support .bdrv_co_amend */
> > +    if (!drv->bdrv_co_amend) {
> > +        error_setg(errp, "Driver does not support x-blockdev-amend");
> > +        return;
> > +    }
> > +
> > +    /*
> > +     * Create the block job
> > +     * TODO Running in the main context. Block drivers need to error out 
> > or add
> > +     * locking when they use a BDS in a different AioContext.
> 
> Why shouldn’t the job just run in the node’s context?

This is shameless copy&pasta from the blockdev-create code
(which I did note in the copyright of the file)


Best regards,
        Maxim Levitsky

> 
> Max
>