All of lore.kernel.org
 help / color / mirror / Atom feed
From: Fabiano Rosas <farosas@suse.de>
To: Peter Xu <peterx@redhat.com>
Cc: qemu-devel@nongnu.org
Subject: Re: [PATCH 10/18] migration: Validate that all params are set for query
Date: Fri, 04 Sep 2026 12:11:27 -0300	[thread overview]
Message-ID: <87ld9h6ogw.fsf@suse.de> (raw)
In-Reply-To: <apnDfpdyAzo2Vuwc@x1.local>

Peter Xu <peterx@redhat.com> writes:

> On Wed, Sep 02, 2026 at 07:15:38PM -0300, Fabiano Rosas wrote:
>> There are a couple of situations where all fields of a
>> MigrationParameters object need to be marked as present: when cloning
>> an entire object and when creating the transient object in
>> qmp_query_migrate(). The query-migrate-parameters QMP command contract
>> requires that all parameters, except block-bitmap-mapping, are present
>> in the output.
>> 
>> Validate that a given object has all has_* fields set to true.
>> 
>> Signed-off-by: Fabiano Rosas <farosas@suse.de>
>> ---
>>  migration/options.c | 54 +++++++++++++++++++++++++++++++++++++++++++++
>>  1 file changed, 54 insertions(+)
>> 
>> diff --git a/migration/options.c b/migration/options.c
>> index bd7be8f9832..5d17acdd881 100644
>> --- a/migration/options.c
>> +++ b/migration/options.c
>> @@ -12,6 +12,7 @@
>>   */
>>  
>>  #include "qemu/osdep.h"
>> +#include "qemu/cutils.h"
>>  #include "qemu/error-report.h"
>>  #include "qemu/units.h"
>>  #include "exec/target_page.h"
>> @@ -23,8 +24,10 @@
>>  #include "qapi/qmp/qerror.h"
>>  #include "qapi/qobject-input-visitor.h"
>>  #include "qapi/qobject-output-visitor.h"
>> +#include "qobject/qbool.h"
>>  #include "qobject/qdict.h"
>>  #include "qobject/qnull.h"
>> +#include "qobject/qstring.h"
>>  #include "system/runstate.h"
>>  #include "migration/colo.h"
>>  #include "migration/cpr.h"
>> @@ -1149,12 +1152,63 @@ static void migrate_mark_all_params_present(MigrationParameters *p)
>>      }
>>  }
>>  
>> +static bool assert_all_params_present(MigrationParameters *params, Error **errp)
>> +{
>> +    g_autoptr(QDict) d = migrate_params_to_dict(params, errp);
>> +    const QDictEntry *e = NULL;
>> +    int i = 0;
>> +
>> +    if (!d) {
>> +        return false;
>> +    }
>> +
>> +    for (e = qdict_first(d); e; e = qdict_next(d, e), i++) {
>> +        const char *key = qdict_entry_key(e);
>> +        const char *p;
>> +
>> +        if (strstart(key, "tls-", &p)) {
>> +            QString *s = qobject_to(QString, qdict_entry_value(e));
>> +
>> +            if (!s) {
>> +                break;
>> +            }
>> +        } else if (strstart(key, "has-", &p)) {
>
> Does the qdict contain any has- field?  
>

¬¬

I guess that decides the fate of the MigrationParameter enum.

> visit_type_MigrationParameters_members:
>
>     if (visit_optional(v, "announce-initial", &obj->has_announce_initial)) {
>         if (!visit_type_size(v, "announce-initial", &obj->announce_initial, errp)) {
>             return false;
>         }
>     }
>     ...
>
> It seems the has_* fields are only used to identify existance of objects,
> not converted.
>
>> +            if (qdict_haskey(d, p)) {
>> +                QBool *b = qobject_to(QBool, qdict_entry_value(e));
>> +
>> +                If (!b || !qbool_get_bool(b)) {
>> +                    break;
>> +                }
>> +            }
>> +        }
>> +    }
>> +
>> +    if (i && !e) {
>> +        return true;
>> +    }
>> +
>> +    /*
>> +     * Should never happen, but avoid asserting becase this is
>> +     * reachable from QMP.
>
> IIUC as long as this fact shouldn't be changed by any possible user input,
> we could still assert.  But I understand you want to be careful, maybe
> either (1) directly assert, or (2) change the function name,
> s/assert/check/?  I vote (1).
>
> Said that, if the qdict trick didn't work it beats the whole patch.. so
> IMHO we can also leave this sanity check for later too.  Your call.
>
>> +     */
>> +    error_setg(errp, "Missing parameter. Query output will be incomplete.");
>> +    return false;
>> +}
>> +
>>  MigrationParameters *qmp_query_migrate_parameters(Error **errp)
>>  {
>>      MigrationState *s = migrate_get_current();
>>      MigrationParameters *params = QAPI_CLONE(MigrationParameters,
>>                                               &s->parameters);
>>  
>> +    /*
>> +     * Validate all parameters have their has_* field set to true as
>> +     * consequence of the initial migrate_mark_all_params_present().
>> +     */
>> +    if (!assert_all_params_present(params, errp)) {
>> +        return NULL;
>> +    }
>> +
>>      /*
>>       * The block-bitmap-mapping breaks the expected API of
>>       * query-migrate-parameters of having all members present. To keep
>> -- 
>> 2.53.0
>> 


  reply	other threads:[~2026-09-04 15:12 UTC|newest]

Thread overview: 50+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-09-02 22:15 [PATCH 00/18] migration: MigrationParameters changes Fabiano Rosas
2026-09-02 22:15 ` [PATCH 01/18] checkpatch: Fix checking of newlines in error messages Fabiano Rosas
2026-09-03 17:37   ` Peter Xu
2026-09-03 17:46     ` Fabiano Rosas
2026-09-03 18:00       ` Peter Xu
2026-09-03 18:35         ` Fabiano Rosas
2026-09-04  8:56         ` Markus Armbruster
2026-09-04  9:15           ` Peter Maydell
2026-09-04 12:09             ` Peter Xu
2026-09-04 14:00   ` Markus Armbruster
2026-09-02 22:15 ` [PATCH 02/18] migration/options.c: Don't export migrate_tls_opts_free Fabiano Rosas
2026-09-02 22:15 ` [PATCH 03/18] migration: Rename variables in qmp_migrate_set_parameters Fabiano Rosas
2026-09-03 17:44   ` Peter Xu
2026-09-02 22:15 ` [PATCH 04/18] migration: Use QAPI_CLONE_MEMBERS in migrate_params_apply Fabiano Rosas
2026-09-02 22:15 ` [PATCH 05/18] migration: Merge parameter structs instead of assigning one by one Fabiano Rosas
2026-09-03 18:20   ` Peter Xu
2026-09-03 19:03     ` Fabiano Rosas
2026-09-02 22:15 ` [PATCH 06/18] migration: Open code migrate_params_apply Fabiano Rosas
2026-09-02 22:15 ` [PATCH 07/18] migration: Stop freeing s->parameters members individually Fabiano Rosas
2026-09-02 22:15 ` [PATCH 08/18] migration: Use migrate_params_free during finalize Fabiano Rosas
2026-09-02 22:15 ` [PATCH 09/18] tests/qtest/migration: Add a test for HMP Fabiano Rosas
2026-09-03 20:38   ` Peter Xu
2026-09-03 20:42     ` Peter Xu
2026-09-02 22:15 ` [PATCH 10/18] migration: Validate that all params are set for query Fabiano Rosas
2026-09-03 18:59   ` Peter Xu
2026-09-04 15:11     ` Fabiano Rosas [this message]
2026-09-02 22:15 ` [PATCH 11/18] migration: Use keyval input visitor in HMP set command Fabiano Rosas
2026-09-03 19:44   ` Peter Xu
2026-09-03 20:24     ` Dr. David Alan Gilbert
2026-09-04 15:26       ` Peter Xu
2026-09-04  9:45   ` Markus Armbruster
2026-09-02 22:15 ` [PATCH 12/18] migration: Change HMP 'info migrate_parameters' output Fabiano Rosas
2026-09-03 20:25   ` Peter Xu
2026-09-02 22:15 ` [PATCH 13/18] migration: Use output visitor in info command Fabiano Rosas
2026-09-04 12:04   ` Peter Xu
2026-09-04 13:41     ` Fabiano Rosas
2026-09-04 14:54       ` Peter Xu
2026-09-04 15:08         ` Fabiano Rosas
2026-09-02 22:15 ` [PATCH 14/18] migration: Rewrite migrate_set_parameter_completion using QDict Fabiano Rosas
2026-09-03 20:23   ` Peter Xu
2026-09-02 22:15 ` [PATCH 15/18] qapi/migration: Remove MigrationParameter Fabiano Rosas
2026-09-04  9:07   ` Markus Armbruster
2026-09-04 12:24   ` Peter Xu
2026-09-04 13:49     ` Fabiano Rosas
2026-09-02 22:15 ` [PATCH 16/18] migration: Add capabilities into MigrationParameters Fabiano Rosas
2026-09-04  9:53   ` Markus Armbruster
2026-09-04 13:58     ` Fabiano Rosas
2026-09-02 22:15 ` [PATCH 17/18] migration: Remove s->capabilities Fabiano Rosas
2026-09-04 12:15   ` Peter Xu
2026-09-02 22:15 ` [PATCH 18/18] qapi/migration: Deprecate capabilities commands Fabiano Rosas

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=87ld9h6ogw.fsf@suse.de \
    --to=farosas@suse.de \
    --cc=peterx@redhat.com \
    --cc=qemu-devel@nongnu.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.