[PATCH v2 18/19] qemu: blockjob: Handle bitmaps after finish of normal block-commit

[Eric Blake]

On 3/11/20 7:56 AM, Peter Krempa wrote:
> Merge the bitmaps into base of the block commit after the job finishes.
> 
> Signed-off-by: Peter Krempa <pkrempa at redhat.com>
> ---
>   src/qemu/qemu_blockjob.c | 53 ++++++++++++++++++++++++++++++++++++++++
>   1 file changed, 53 insertions(+)
> 
> diff --git a/src/qemu/qemu_blockjob.c b/src/qemu/qemu_blockjob.c
> index 63f1cc79c3..ed7959175a 100644
> --- a/src/qemu/qemu_blockjob.c
> +++ b/src/qemu/qemu_blockjob.c
> @@ -1049,6 +1049,56 @@ qemuBlockJobDeleteImages(virQEMUDriverPtr driver,
>       }
>   }
> 
> +
> +/**
> + * qemuBlockJobProcessEventCompletedCommitBitmaps:
> + *
> + * Handles the bitmap changes after commit. This function shall return -1 on
> + * monitor failures.

s/shall return/returns/ sounds less stodgy :)

> + */
> +static int
> +qemuBlockJobProcessEventCompletedCommitBitmaps(virDomainObjPtr vm,
> +                                               qemuBlockJobDataPtr job,
> +                                               qemuDomainAsyncJob asyncJob)
> +{
> +    qemuDomainObjPrivatePtr priv = vm->privateData;
> +    g_autoptr(virHashTable) blockNamedNodeData = NULL;
> +    g_autoptr(virJSONValue) actions = NULL;
> +
> +    if (!virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_BLOCKDEV_REOPEN))
> +        return 0;
> +
> +    if (!(blockNamedNodeData = qemuBlockGetNamedNodeData(vm, asyncJob)))
> +        return -1;
> +
> +    if (qemuBlockBitmapsHandleCommitFinish(job->data.commit.top,
> +                                           job->data.commit.base,
> +                                           blockNamedNodeData,
> +                                           &actions,
> +                                           job->data.commit.disabledBitmapsBase) < 0)
> +        return 0;
> +
> +    if (!actions)
> +        return 0;
> +
> +    if (qemuBlockReopenReadWrite(vm, job->data.commit.base, asyncJob) < 0)
> +        return -1;
> +
> +    if (qemuDomainObjEnterMonitorAsync(priv->driver, vm, asyncJob) < 0)
> +        return -1;
> +
> +    qemuMonitorTransaction(priv->mon, &actions);
> +
> +    if (qemuDomainObjExitMonitor(priv->driver, vm) < 0)
> +        return -1;
> +
> +    if (qemuBlockReopenReadOnly(vm, job->data.commit.base, asyncJob) < 0)
> +        return -1;

Earlier in the series, you ignored failure to restore readonly (leaving 
things read-write isn't ideal, but isn't a show-stopper to correct 
operation).  Any reason why that use was different than this where you 
treat it as fatal?

Reviewed-by: Eric Blake <eblake at redhat.com>

-- 
Eric Blake, Principal Software Engineer
Red Hat, Inc.           +1-919-301-3226
Virtualization:  qemu.org | libvirt.org