qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Fabiano Rosas <farosas@suse.de>
To: Bin Guo <guobin@linux.alibaba.com>
Cc: peterx@redhat.com, qemu-devel@nongnu.org
Subject: Re: [PATCH v2 14/24] migration: Use visitors in migrate_params_test_apply
Date: Thu, 23 Oct 2025 11:46:51 -0300	[thread overview]
Message-ID: <87v7k5smjo.fsf@suse.de> (raw)
In-Reply-To: <20251015021603.21933-1-guobin@linux.alibaba.com>

Bin Guo <guobin@linux.alibaba.com> writes:

> Fabiano Rosas wrote on Mon, 30 Jun 2025 16:59:03 -0300:
>
>> Convert the code in migrate_params_test_apply() from an open-coded
>> copy of every migration parameter to a copy using visitors. The
>> current code has conditionals for each parameter's has_* field, which
>> is exactly what the visitors do.
>> 
>> This hides the details of QAPI from the migration code and avoids the
>> need to update migrate_params_test_apply() every time a new migration
>> parameter is added. Both were very confusing and while the visitor
>> code can become a bit involved, there is no need for new contributors
>> to ever touch it.
>> 
>> Change the name of the function to a more direct reference of what it
>> does: merging the user params with the temporary copy.
>> 
>> Move the QAPI_CLONE_MEMBERS into the caller, so QAPI_CLONE can be used
>> and there's no need to allocate memory in the migration
>> code. Similarly, turn 'tmp' into a pointer so the proper qapi_free_
>> routine can be used.
>> 
>> An extra call to migrate_mark_all_params_present() is now needed
>> because the visitors update the has_ field for non-present fields, but
>> we actually want them all set so migrate_params_apply() can copy all
>> of them.
>> 
>> Signed-off-by: Fabiano Rosas <farosas@suse.de>
>> ---
>>  migration/options.c | 157 +++++++++++++++-----------------------------
>>  1 file changed, 54 insertions(+), 103 deletions(-)
>> 
>> diff --git a/migration/options.c b/migration/options.c
>> index 6619b5f21a..695bec5b8f 100644
>> --- a/migration/options.c
>> +++ b/migration/options.c
>> 
>> -static void migrate_params_test_apply(MigrationParameters *params,
>> -                                      MigrationParameters *dest)
>> +static bool migrate_params_merge(MigrationParameters *dst,
>> +
>> +    ...
>> +    /* read in from src */
>> +    v = qobject_output_visitor_new(&ret_out);
>> +    ok = visit_type_MigrationParameters(v, NULL, &src, errp);
>> +    if (!ok) {
>> +        goto out;
>>      }
>> +    visit_complete(v, &ret_out);
>> +    visit_free(v);
>>  
>> -    if (params->has_max_bandwidth) {
>> -        dest->max_bandwidth = params->max_bandwidth;
>> +    /*
>> +     * Write to dst but leave existing fields intact (except for has_*
>> +     * which will be updated according to their presence in src).
>> +     */
>> +    v = qobject_input_visitor_new(ret_out);
>> +    ok = visit_start_struct(v, NULL, NULL, 0, errp);
>> +    if (!ok) {
>> +        goto out;
>>      }
>> -
>> -    if (params->has_avail_switchover_bandwidth) {
>> -        dest->avail_switchover_bandwidth = params->avail_switchover_bandwidth;
>> +    ok = visit_type_MigrationParameters_members(v, dst, errp);
>> +    if (!ok) {
>> +        goto out;
>>      }
>> -
>> -    if (params->has_downtime_limit) {
>> -        dest->downtime_limit = params->downtime_limit;
>> +    ok = visit_check_struct(v, errp);
>> +    visit_end_struct(v, NULL);
>> +    if (!ok) {
>> +        goto out;
>>      }
>>  
>> -    if (params->has_x_checkpoint_delay) {
>> -        dest->x_checkpoint_delay = params->x_checkpoint_delay;
>> -    }
>> +out:
>> +    visit_free(v);
>> +    qobject_unref(ret_out);
>> +    return ok;
>> }
>
> If visit_start_struct is executed successfully, then visit_end_struct
> should be executed. IMHO:
>
>     v = qobject_input_visitor_new(ret_out);
>     ok = visit_start_struct(v, NULL, NULL, 0, errp);
>     if (!ok) {
>         goto out;
>     }
>
>     ok = visit_type_MigrationParameters_members(v, dst, errp);
>     if (!ok) {
>         goto out_end;
>     }
>
>     ok = visit_check_struct(v, errp);
>
> out_end:
>     visit_end_struct(v, NULL);
>
> out:
>     visit_free(v);
>     qobject_unref(ret_out);
>
>     return ok;
> }
>

New code after Peter's suggestions is:

+/*
+ * Merge @src over @dst by copying deep clones of the present members
+ * from @src to @dst. Non-present on @src are left untouched on @dst.
+ */
+#define QAPI_MERGE(type, dst_, src_)                                    \
+    ({                                                                  \
+        QObject *out_ = NULL;                                           \
+        Visitor *v_;                                                    \
+        /* read in from src */                                          \
+        v_ = qobject_output_visitor_new(&out_);                         \
+        visit_type_ ## type (v_, NULL, &src_, &error_abort);            \
+        visit_complete(v_, &out_);                                      \
+        visit_free(v_);                                                 \
+        /*                                                              \
+         * Write to dst but leave existing fields intact (except for    \
+         * has_* which will be updated according to their presence in   \
+         * src).                                                        \
+         */                                                             \
+        v_ = qobject_input_visitor_new(out_);                           \
+        visit_start_struct(v_, NULL, NULL, 0, &error_abort);            \
+        visit_type_ ## type ## _members(v_, dst_, &error_abort);        \
+        visit_check_struct(v_, &error_abort);                           \
+        visit_end_struct(v_, NULL);                                     \
+        visit_free(v_);                                                 \
+        qobject_unref(out_);                                            \
+    })



  reply	other threads:[~2025-10-23 14:47 UTC|newest]

Thread overview: 79+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-06-30 19:58 [PATCH v2 00/24] migration: Unify capabilities and parameters Fabiano Rosas
2025-06-30 19:58 ` [PATCH v2 01/24] migration: Fix leak of block_bitmap_mapping Fabiano Rosas
2025-07-01  6:12   ` Markus Armbruster
2025-07-03 21:31     ` Peter Xu
2025-07-04  5:09       ` Markus Armbruster
2025-06-30 19:58 ` [PATCH v2 02/24] migration: Add a qdev property for StrOrNull Fabiano Rosas
2025-07-01  6:38   ` Markus Armbruster
2025-07-03 22:32     ` Peter Xu
2025-07-04 12:58       ` Fabiano Rosas
2025-06-30 19:58 ` [PATCH v2 03/24] migration: Normalize tls arguments Fabiano Rosas
2025-07-01  7:46   ` Markus Armbruster
2025-07-01 14:20     ` Fabiano Rosas
2025-07-04 13:12       ` Fabiano Rosas
2025-07-04 15:37         ` Peter Xu
2025-08-20 15:45           ` Fabiano Rosas
2025-10-15  2:31   ` Bin Guo
2025-10-23 14:29     ` Fabiano Rosas
2025-06-30 19:58 ` [PATCH v2 04/24] migration: Remove MigrateSetParameters Fabiano Rosas
2025-07-01  8:00   ` Markus Armbruster
2025-07-03 19:34     ` Fabiano Rosas
2025-07-04  4:25       ` Markus Armbruster
2025-07-04 15:39   ` Peter Xu
2025-06-30 19:58 ` [PATCH v2 05/24] qapi/migration: Don't document MigrationParameter Fabiano Rosas
2025-07-01  8:04   ` Markus Armbruster
2025-07-04 15:40   ` Peter Xu
2025-06-30 19:58 ` [PATCH v2 06/24] migration: Run a post update routine after setting parameters Fabiano Rosas
2025-10-13  6:10   ` Bin Guo
2025-10-23 14:30     ` Fabiano Rosas
2025-06-30 19:58 ` [PATCH v2 07/24] migration: Add a flag to track block-bitmap-mapping input Fabiano Rosas
2025-07-04 15:42   ` Peter Xu
2025-06-30 19:58 ` [PATCH v2 08/24] migration: Remove checks for s->parameters has_* fields Fabiano Rosas
2025-06-30 19:58 ` [PATCH v2 09/24] migration: Do away with usage of QERR_INVALID_PARAMETER_VALUE Fabiano Rosas
2025-07-04 16:04   ` Peter Xu
2025-06-30 19:58 ` [PATCH v2 10/24] migration: Extract code to mark all parameters as present Fabiano Rosas
2025-06-30 19:59 ` [PATCH v2 11/24] migration: Use QAPI_CLONE_MEMBERS in query_migrate_parameters Fabiano Rosas
2025-07-04 16:11   ` Peter Xu
2025-06-30 19:59 ` [PATCH v2 12/24] migration: Use QAPI_CLONE_MEMBERS in migrate_params_test_apply Fabiano Rosas
2025-06-30 19:59 ` [PATCH v2 13/24] migration: Use QAPI_CLONE_MEMBERS in migrate_params_apply Fabiano Rosas
2025-08-13 19:05   ` Peter Xu
2025-08-14 15:04     ` Fabiano Rosas
2025-06-30 19:59 ` [PATCH v2 14/24] migration: Use visitors in migrate_params_test_apply Fabiano Rosas
2025-08-13 20:05   ` Peter Xu
2025-08-14 15:10     ` Fabiano Rosas
2025-08-14 19:40       ` Peter Xu
2025-10-15  2:16   ` Bin Guo
2025-10-23 14:46     ` Fabiano Rosas [this message]
2025-06-30 19:59 ` [PATCH v2 15/24] migration: Cleanup hmp_info_migrate_parameters Fabiano Rosas
2025-08-13 20:40   ` Peter Xu
2025-06-30 19:59 ` [PATCH v2 16/24] migration: Add capabilities into MigrationParameters Fabiano Rosas
2025-07-01  8:25   ` Markus Armbruster
2025-07-04 13:15     ` Fabiano Rosas
2025-07-04 14:04       ` Markus Armbruster
2025-07-04 14:48         ` Fabiano Rosas
2025-07-04 15:04           ` Markus Armbruster
2025-07-04 16:33   ` Peter Xu
2025-06-30 19:59 ` [PATCH v2 17/24] migration: Remove s->capabilities Fabiano Rosas
2025-08-13 20:48   ` Peter Xu
2025-06-30 19:59 ` [PATCH v2 18/24] qapi/migration: Deprecate capabilities commands Fabiano Rosas
2025-07-01  8:30   ` Markus Armbruster
2025-07-01  8:38   ` Jiri Denemark
2025-07-01  9:00     ` Peter Krempa
2025-07-01  9:10     ` Daniel P. Berrangé
2025-08-13 20:50   ` Peter Xu
2025-06-30 19:59 ` [PATCH v2 19/24] migration: Store the initial values used for s->parameters Fabiano Rosas
2025-08-13 21:09   ` Peter Xu
2025-06-30 19:59 ` [PATCH v2 20/24] migration: Allow migrate commands to provide the migration config Fabiano Rosas
2025-07-01  8:35   ` Markus Armbruster
2025-08-13 21:27   ` Peter Xu
2025-08-14 15:13     ` Fabiano Rosas
2025-06-30 19:59 ` [PATCH v2 21/24] tests/qtest/migration: Take reference when passing %p to qtest_qmp Fabiano Rosas
2025-08-13 22:22   ` Peter Xu
2025-08-21 17:20     ` Fabiano Rosas
2025-06-30 19:59 ` [PATCH v2 22/24] tests/qtest/migration: Adapt the capabilities helper to take a config Fabiano Rosas
2025-08-14 14:02   ` Peter Xu
2025-06-30 19:59 ` [PATCH v2 23/24] tests/qtest/migration: Adapt convergence routines to config Fabiano Rosas
2025-06-30 19:59 ` [PATCH v2 24/24] tests/qtest/migration: Pass the migration config to file tests Fabiano Rosas
2025-08-14 14:24   ` Peter Xu
2025-08-14 15:30     ` Fabiano Rosas
2025-08-14 19:45       ` Peter Xu

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=87v7k5smjo.fsf@suse.de \
    --to=farosas@suse.de \
    --cc=guobin@linux.alibaba.com \
    --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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).