Re: [Qemu-devel] [PATCH 1/8] qemu-option: qemu_opt_set_bool(): fix code duplication

[Markus Armbruster]

Luiz Capitulino <address@hidden> writes:

> On Sat, 21 Jul 2012 10:09:09 +0200
> Markus Armbruster <address@hidden> wrote:
>
>> Luiz Capitulino <address@hidden> writes:
>> 
>> > Call qemu_opt_set() instead of duplicating opt_set().
>> >
>> > Signed-off-by: Luiz Capitulino <address@hidden>
>> > ---
>> >  qemu-option.c | 28 +---------------------------
>> >  1 file changed, 1 insertion(+), 27 deletions(-)
>> >
>> > diff --git a/qemu-option.c b/qemu-option.c
>> > index bb3886c..2cb2835 100644
>> > --- a/qemu-option.c
>> > +++ b/qemu-option.c
>> > @@ -677,33 +677,7 @@ void qemu_opt_set_err(QemuOpts *opts, const char
>> > *name, const char *value,
>> >  
>> >  int qemu_opt_set_bool(QemuOpts *opts, const char *name, bool val)
>> >  {
>> > -    QemuOpt *opt;
>> > -    const QemuOptDesc *desc = opts->list->desc;
>> > -    int i;
>> > -
>> > -    for (i = 0; desc[i].name != NULL; i++) {
>> > -        if (strcmp(desc[i].name, name) == 0) {
>> > -            break;
>> > -        }
>> > -    }
>> > -    if (desc[i].name == NULL) {
>> > -        if (i == 0) {
>> > -            /* empty list -> allow any */;
>> > -        } else {
>> > -            qerror_report(QERR_INVALID_PARAMETER, name);
>> > -            return -1;
>> > -        }
>> > -    }
>> > -
>> > -    opt = g_malloc0(sizeof(*opt));
>> > -    opt->name = g_strdup(name);
>> > -    opt->opts = opts;
>> > -    QTAILQ_INSERT_TAIL(&opts->head, opt, next);
>> > -    if (desc[i].name != NULL) {
>> > -        opt->desc = desc+i;
>> > -    }
>> > -    opt->value.boolean = !!val;
>> > -    return 0;
>> > +    return qemu_opt_set(opts, name, val ? "on" : "off");
>> >  }
>> >  
>> >  int qemu_opt_foreach(QemuOpts *opts, qemu_opt_loopfunc func, void *opaque,
>> 
>> Does a bit more than just obvious code de-duplication.  Two things in
>> particular:
>> 
>> * Error reporting
>> 
>>   Old: qerror_report(); return -1
>> 
>>   New: opt_set() and qemu_opt_set() cooperate, like this:
>>        opt_set(): error_set(); return;
>>        qemu_opt_set():
>>             if (error_is_set(&local_err)) {
>>                 qerror_report_err(local_err);
>>                 error_free(local_err);
>>                 return -1;
>>             }
>> 
>>   I guess the net effect is the same.  Not sure it's worth mentioning in
>>   the commit message.
>
> The end result of calling qemu_opt_set_bool() is the same. The difference
> between qerror_report() and qerror_report_err() is that the former gets error
> information from the error call, while the latter gets error information
> from the Error object. But they do the same thing.
>
>> * New sets opt->str to either "on" or "off" depending on val, then lets
>>   reconstructs the value with qemu_opt_parse().  Which can't fail then.
>>   I figure the latter part has no further impact.  But what about
>>   setting opt->str?  Is this a bug fix?
>
> I don't remember if opt->str is read after the QemuOpt object is built. If
> it's, then yes, this is a bug fix. Otherwise it just make the final
> QemuOpt object more 'conforming'.

Uses of opt->str, and what happens when it isn't set:

* qemu_opt_get(): returns NULL, which means "not set".  Bug can bite
  when value isn't the default value.

* qemu_opt_parse(): passes NULL to parse_option_bool(), which treats it
  like "on".  Wrong if the value is actually false.  Bug can bite when
  qemu_opts_validate() runs after qemu_opt_set_bool().

* qemu_opt_del(): passes NULL to g_free(), which is just fine.

* qemu_opt_foreach(): passes NULL to the callback, which is unlikely to
  be prepared for it.

* qemu_opts_print(): prints NULL, which crashes on some systems.

* qemu_opts_to_qdict(): passes NULL to qstring_from_str(), which
  crashes.

Right now, the only use of qemu_opt_set_bool() is the one in vl.c.
Can't see how to break it, but I didn't look hard.

I recommend to document the bug fix in the commit message.