Re: [PATCH v2 07/11] block: add x-blockdev-amend qmp command
On Fri, 2019-11-08 at 11:36 +0100, Max Reitz wrote:
> On 08.11.19 10:26, Maxim Levitsky wrote:
> > 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
> > > > ---
> > > > 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
>
> [...]
>
> > > > +static int coroutine_fn blockdev_amend_run(Job *job, Error **errp)
> > > > +{
> > > > +BlockdevAmendJob *s = container_of(job, BlockdevAmendJob, common);
> > > > +int ret;
> > > > +
> > > > +job_progress_set_remaining(>common, 1);
> > > > +ret = s->bs->drv->bdrv_co_amend(s->bs, s->opts, s->force, errp);
> > > > +job_progress_update(>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).
>
> True, and we could always add it later.
>
> I suppose I was mostly wondering because bdrv_amend_options already has
> all of that infrastructure and I was assuming that qcow2's bdrv_co_amend
> implementation would make some use of the existing function. Well, it
> doesn’t, so *shrug*
>
> [...]
>
> > > > +/*
> > > > + * 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 from the blockdev-create code
> > (which I did note in the copyright of the file)
>
> Well, you noted that it’s heavily based on it, not that it’s just C
'heavily based' is a polite way to say that file is copied and then changed
to fit new purpose, isn't it :-)
>
> So I suppose the comment is just wrong here?
Yes, I absolutely missed this part since I don't know the block layer well
enough.
Thanks for explaining this on IRC, that blockdev_create job is special
in a sense that block device state doesn't exit yet for it (I also was already
bitten
by this in luks block driver), thus the job runs in main AIO context.
No need to do so in amend, and thus I'll remove that wrongly copy'ed
comment.
Thanks a lot!!
Best regards,
Maxim Levitsky
>
> Max
>
Re: [PATCH v2 07/11] block: add x-blockdev-amend qmp command
On 08.11.19 10:26, Maxim Levitsky wrote:
> 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
>>> ---
>>> 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
[...]
>>> +static int coroutine_fn blockdev_amend_run(Job *job, Error **errp)
>>> +{
>>> +BlockdevAmendJob *s = container_of(job, BlockdevAmendJob, common);
>>> +int ret;
>>> +
>>> +job_progress_set_remaining(>common, 1);
>>> +ret = s->bs->drv->bdrv_co_amend(s->bs, s->opts, s->force, errp);
>>> +job_progress_update(>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).
True, and we could always add it later.
I suppose I was mostly wondering because bdrv_amend_options already has
all of that infrastructure and I was assuming that qcow2's bdrv_co_amend
implementation would make some use of the existing function. Well, it
doesn’t, so *shrug*
[...]
>>> +/*
>>> + * 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 from the blockdev-create code
> (which I did note in the copyright of the file)
Well, you noted that it’s heavily based on it, not that it’s just C
So I suppose the comment is just wrong here?
Max
signature.asc
Description: OpenPGP digital signature
Re: [PATCH v2 07/11] block: add x-blockdev-amend qmp command
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
> > ---
> > 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 00..9bd28e08e7
> > --- /dev/null
> > +++ b/block/amend.c
> > @@ -0,0 +1,116 @@
> > +/*
> > + * Block layer code related to image options amend
> > + *
> > + * Copyright (c) 2018 Kevin Wolf
> > + * Copyright (c) 2019 Maxim Levitsky
> > + *
> > + * 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(>common, 1);
> > +ret = s->bs->drv->bdrv_co_amend(s->bs, s->opts, s->force, errp);
> > +job_progress_update(>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);
> > +
Re: [PATCH v2 07/11] block: add x-blockdev-amend qmp command
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
> > ---
> > 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 implment
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 00..9bd28e08e7
> > --- /dev/null
> > +++ b/block/amend.c
> > @@ -0,0 +1,116 @@
> > +/*
> > + * Block layer code related to image options amend
> > + *
> > + * Copyright (c) 2018 Kevin Wolf
> > + * Copyright (c) 2019 Maxim Levitsky
> > + *
> > + * 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(>common, 1);
> > +ret = s->bs->drv->bdrv_co_amend(s->bs, s->opts, s->force, errp);
> > +job_progress_update(>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);
> > +
Re: [PATCH v2 07/11] block: add x-blockdev-amend qmp command
On 13.09.19 00:30, Maxim Levitsky wrote:
> Signed-off-by: Maxim Levitsky
> ---
> 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.
> 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 00..9bd28e08e7
> --- /dev/null
> +++ b/block/amend.c
> @@ -0,0 +1,116 @@
> +/*
> + * Block layer code related to image options amend
> + *
> + * Copyright (c) 2018 Kevin Wolf
> + * Copyright (c) 2019 Maxim Levitsky
> + *
> + * 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(>common, 1);
> +ret = s->bs->drv->bdrv_co_amend(s->bs, s->opts, s->force, errp);
> +job_progress_update(>common, 1);
It would be nice if the amend job could make use of the progress
reporting that we have in place for amend.
> +
> +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
