[RFC,V2,2/5] qapi: add event helper functions

This file hold some functions that do not need to be generated.

Signed-off-by: Wenchao Xia <xiawenc@linux.vnet.ibm.com>
---
 include/qapi/qmp-event.h |   22 ++++++++++++++++++
 qapi/Makefile.objs       |    1 +
 qapi/qmp-event.c         |   56 ++++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 79 insertions(+), 0 deletions(-)
 create mode 100644 include/qapi/qmp-event.h
 create mode 100644 qapi/qmp-event.c

On Fri,  3 Jan 2014 07:10:31 +0800
Wenchao Xia <xiawenc@linux.vnet.ibm.com> wrote:

> This file hold some functions that do not need to be generated.
> 
> Signed-off-by: Wenchao Xia <xiawenc@linux.vnet.ibm.com>
> ---
>  include/qapi/qmp-event.h |   22 ++++++++++++++++++
>  qapi/Makefile.objs       |    1 +
>  qapi/qmp-event.c         |   56 ++++++++++++++++++++++++++++++++++++++++++++++
>  3 files changed, 79 insertions(+), 0 deletions(-)
>  create mode 100644 include/qapi/qmp-event.h
>  create mode 100644 qapi/qmp-event.c
> 
> diff --git a/include/qapi/qmp-event.h b/include/qapi/qmp-event.h
> new file mode 100644
> index 0000000..2baf093
> --- /dev/null
> +++ b/include/qapi/qmp-event.h
> @@ -0,0 +1,22 @@
> +/*
> + * QMP Event related
> + *
> + * Copyright IBM, Corp. 2014
> + *
> + * Authors:
> + *  Wenchao Xia   <xiawenc@linux.vnet.ibm.com>
> + *
> + * This work is licensed under the terms of the GNU GPLv2+ or later.
> + * See the COPYING.LIB file in the top-level directory.
> + *
> + */
> +
> +#ifndef QMP_EVENT_H
> +#define QMP_EVENT_H
> +
> +#include "qapi/error.h"
> +#include "qapi/qmp/qdict.h"
> +
> +QDict *qmp_event_build_dict(const char *event_name);
> +
> +#endif
> diff --git a/qapi/Makefile.objs b/qapi/Makefile.objs
> index 1f9c973..d14b769 100644
> --- a/qapi/Makefile.objs
> +++ b/qapi/Makefile.objs
> @@ -3,3 +3,4 @@ util-obj-y += qmp-output-visitor.o qmp-registry.o qmp-dispatch.o
>  util-obj-y += string-input-visitor.o string-output-visitor.o
>  
>  util-obj-y += opts-visitor.o
> +util-obj-y += qmp-event.o
> diff --git a/qapi/qmp-event.c b/qapi/qmp-event.c
> new file mode 100644
> index 0000000..dc81ec2
> --- /dev/null
> +++ b/qapi/qmp-event.c
> @@ -0,0 +1,56 @@
> +/*
> + * QMP Event related
> + *
> + * Copyright IBM, Corp. 2014
> + *
> + * Authors:
> + *  Wenchao Xia   <xiawenc@linux.vnet.ibm.com>
> + *
> + * This work is licensed under the terms of the GNU GPLv2+ or later.
> + * See the COPYING.LIB file in the top-level directory.
> + *
> + */
> +
> +#include <inttypes.h>
> +
> +#include "qemu-common.h"
> +#include "qapi/qmp-event.h"
> +#include "qapi/qmp/qstring.h"
> +#include "qapi/qmp/qjson.h"
> +
> +#ifdef _WIN32
> +#include "sysemu/os-win32.h"
> +#endif
> +
> +#ifdef CONFIG_POSIX
> +#include "sysemu/os-posix.h"
> +#endif
> +
> +static void timestamp_put(QDict *qdict)
> +{
> +    int err;
> +    QObject *obj;
> +    qemu_timeval tv;
> +
> +    err = qemu_gettimeofday(&tv);
> +    if (err < 0) {
> +        return;
> +    }

Hmm, I see this has always existed (and I guess I did it myself), but it's
not quite right. Sending an event w/o time info wouldn't be complaint to
the protocol spec. It's a good idea to fix this now. We have three options:

 1. abort()

 2. Skip sending the event altogether

 3. Add a bogus time value (say seconds=0 and microseconds=0)

I don't know what's best, but I guess I'd do item 3. Although I wonder
if zero is any better then no info at all (it's certainly complaint, but
not a valid info). Maybe skip the event then?


> +
> +    obj = qobject_from_jsonf("{ 'seconds': %" PRId64 ", "
> +                                "'microseconds': %" PRId64 " }",
> +                                (int64_t) tv.tv_sec, (int64_t) tv.tv_usec);
> +    qdict_put_obj(qdict, "timestamp", obj);
> +}
> +
> +/*
> + * Build a QDict, then fill event name and time stamp, caller should free the
> + * QDict after usage.
> + */
> +QDict *qmp_event_build_dict(const char *event_name)
> +{
> +    QDict *dict = qdict_new();
> +    qdict_put(dict, "event", qstring_from_str(event_name));
> +    timestamp_put(dict);
> +    return dict;
> +}

于 2014/1/7 6:23, Luiz Capitulino 写道:
> On Fri,  3 Jan 2014 07:10:31 +0800
> Wenchao Xia <xiawenc@linux.vnet.ibm.com> wrote:
>
>> This file hold some functions that do not need to be generated.
>>
>> Signed-off-by: Wenchao Xia <xiawenc@linux.vnet.ibm.com>
>> ---
>>   include/qapi/qmp-event.h |   22 ++++++++++++++++++
>>   qapi/Makefile.objs       |    1 +
>>   qapi/qmp-event.c         |   56 ++++++++++++++++++++++++++++++++++++++++++++++
>>   3 files changed, 79 insertions(+), 0 deletions(-)
>>   create mode 100644 include/qapi/qmp-event.h
>>   create mode 100644 qapi/qmp-event.c
>>
>> diff --git a/include/qapi/qmp-event.h b/include/qapi/qmp-event.h
>> new file mode 100644
>> index 0000000..2baf093
>> --- /dev/null
>> +++ b/include/qapi/qmp-event.h
>> @@ -0,0 +1,22 @@
>> +/*
>> + * QMP Event related
>> + *
>> + * Copyright IBM, Corp. 2014
>> + *
>> + * Authors:
>> + *  Wenchao Xia   <xiawenc@linux.vnet.ibm.com>
>> + *
>> + * This work is licensed under the terms of the GNU GPLv2+ or later.
>> + * See the COPYING.LIB file in the top-level directory.
>> + *
>> + */
>> +
>> +#ifndef QMP_EVENT_H
>> +#define QMP_EVENT_H
>> +
>> +#include "qapi/error.h"
>> +#include "qapi/qmp/qdict.h"
>> +
>> +QDict *qmp_event_build_dict(const char *event_name);
>> +
>> +#endif
>> diff --git a/qapi/Makefile.objs b/qapi/Makefile.objs
>> index 1f9c973..d14b769 100644
>> --- a/qapi/Makefile.objs
>> +++ b/qapi/Makefile.objs
>> @@ -3,3 +3,4 @@ util-obj-y += qmp-output-visitor.o qmp-registry.o qmp-dispatch.o
>>   util-obj-y += string-input-visitor.o string-output-visitor.o
>>
>>   util-obj-y += opts-visitor.o
>> +util-obj-y += qmp-event.o
>> diff --git a/qapi/qmp-event.c b/qapi/qmp-event.c
>> new file mode 100644
>> index 0000000..dc81ec2
>> --- /dev/null
>> +++ b/qapi/qmp-event.c
>> @@ -0,0 +1,56 @@
>> +/*
>> + * QMP Event related
>> + *
>> + * Copyright IBM, Corp. 2014
>> + *
>> + * Authors:
>> + *  Wenchao Xia   <xiawenc@linux.vnet.ibm.com>
>> + *
>> + * This work is licensed under the terms of the GNU GPLv2+ or later.
>> + * See the COPYING.LIB file in the top-level directory.
>> + *
>> + */
>> +
>> +#include <inttypes.h>
>> +
>> +#include "qemu-common.h"
>> +#include "qapi/qmp-event.h"
>> +#include "qapi/qmp/qstring.h"
>> +#include "qapi/qmp/qjson.h"
>> +
>> +#ifdef _WIN32
>> +#include "sysemu/os-win32.h"
>> +#endif
>> +
>> +#ifdef CONFIG_POSIX
>> +#include "sysemu/os-posix.h"
>> +#endif
>> +
>> +static void timestamp_put(QDict *qdict)
>> +{
>> +    int err;
>> +    QObject *obj;
>> +    qemu_timeval tv;
>> +
>> +    err = qemu_gettimeofday(&tv);
>> +    if (err < 0) {
>> +        return;
>> +    }
>
> Hmm, I see this has always existed (and I guess I did it myself), but it's
> not quite right. Sending an event w/o time info wouldn't be complaint to
> the protocol spec. It's a good idea to fix this now. We have three options:
>
>   1. abort()
>
>   2. Skip sending the event altogether
>
>   3. Add a bogus time value (say seconds=0 and microseconds=0)
>
> I don't know what's best, but I guess I'd do item 3. Although I wonder
> if zero is any better then no info at all (it's certainly complaint, but
> not a valid info). Maybe skip the event then?
>

   I think user want a way, to know error happens. If it is
skepted, then we should report it in stderr or a special message in
monitor. In my opinion, sending an event would be the easist way.
We can set time to zero and doc "timestamp == 0 means failure in
getting host time", and this avoid impacting existing user who always
seeking key "timestamp".

>
>> +
>> +    obj = qobject_from_jsonf("{ 'seconds': %" PRId64 ", "
>> +                                "'microseconds': %" PRId64 " }",
>> +                                (int64_t) tv.tv_sec, (int64_t) tv.tv_usec);
>> +    qdict_put_obj(qdict, "timestamp", obj);
>> +}
>> +
>> +/*
>> + * Build a QDict, then fill event name and time stamp, caller should free the
>> + * QDict after usage.
>> + */
>> +QDict *qmp_event_build_dict(const char *event_name)
>> +{
>> +    QDict *dict = qdict_new();
>> +    qdict_put(dict, "event", qstring_from_str(event_name));
>> +    timestamp_put(dict);
>> +    return dict;
>> +}
>

On 01/06/2014 03:23 PM, Luiz Capitulino wrote:
> On Fri,  3 Jan 2014 07:10:31 +0800
> Wenchao Xia <xiawenc@linux.vnet.ibm.com> wrote:
> 
>> This file hold some functions that do not need to be generated.
>>
>> Signed-off-by: Wenchao Xia <xiawenc@linux.vnet.ibm.com>
>> ---

>> +static void timestamp_put(QDict *qdict)
>> +{
>> +    int err;
>> +    QObject *obj;
>> +    qemu_timeval tv;
>> +
>> +    err = qemu_gettimeofday(&tv);
>> +    if (err < 0) {
>> +        return;
>> +    }
> 
> Hmm, I see this has always existed (and I guess I did it myself), but it's
> not quite right. Sending an event w/o time info wouldn't be complaint to
> the protocol spec. It's a good idea to fix this now. We have three options:
> 
>  1. abort()
> 
>  2. Skip sending the event altogether
> 
>  3. Add a bogus time value (say seconds=0 and microseconds=0)
> 
> I don't know what's best, but I guess I'd do item 3. Although I wonder
> if zero is any better then no info at all (it's certainly complaint, but
> not a valid info). Maybe skip the event then?

I have a patch pending[1] for libvirt (missed libvirt 1.2.2, but will be
in 1.2.3) that exposes raw qemu events through libvirt-qemu.so
(basically, as a debugging aid, similar to libvirt's 'virsh
qemu-monitor-command', it will add 'virsh qemu-monitor-event').  In my
implementation, I documented that seconds=-1 implies a missing timestamp
(at which point microseconds is ignored, but can be set to 0).  So I'm
in favor of option 3, but with -1 rather than 0 for seconds.

[1]https://www.redhat.com/archives/libvir-list/2014-February/msg00000.html