Re: [PATCH v9 06/10] migration: New migrate and migrate-incoming argument 'channels'

[Het Gala]

On 26/07/23 2:05 pm, Daniel P. Berrangé wrote:
On Wed, Jul 26, 2023 at 01:24:48AM +0530, Het Gala wrote:
Sorry, last reply on this patch was accidently replied only to Daniel.
Pasting the reply again so it is received by all the active maintianers
here. Apologies for the error 😅

On 26/07/23 12:07 am, Daniel P. Berrangé wrote:
On Tue, Jul 25, 2023 at 07:34:09PM +0100, Daniel P. Berrangé wrote:
On Fri, Jul 21, 2023 at 02:49:31PM +0000, Het Gala wrote:
MigrateChannelList allows to connect accross multiple interfaces.
Add MigrateChannelList struct as argument to migration QAPIs.

We plan to include multiple channels in future, to connnect
multiple interfaces. Hence, we choose 'MigrateChannelList'
as the new argument over 'MigrateChannel' to make migration
QAPIs future proof.

Suggested-by: Aravind Retnakaran <aravind.retnaka...@nutanix.com>
Signed-off-by: Het Gala <het.g...@nutanix.com>
Acked-by: Markus Armbruster <arm...@redhat.com>
---
   migration/migration-hmp-cmds.c |   6 +-
   migration/migration.c          |  34 ++++++++--
   qapi/migration.json            | 109 ++++++++++++++++++++++++++++++++-
   softmmu/vl.c                   |   2 +-
   4 files changed, 139 insertions(+), 12 deletions(-)

diff --git a/migration/migration-hmp-cmds.c b/migration/migration-hmp-cmds.c
index 9885d7c9f7..49b150f33f 100644
--- a/migration/migration-hmp-cmds.c
+++ b/migration/migration-hmp-cmds.c
@@ -424,7 +424,7 @@ void hmp_migrate_incoming(Monitor *mon, const QDict *qdict)
       Error *err = NULL;
       const char *uri = qdict_get_str(qdict, "uri");
-    qmp_migrate_incoming(uri, &err);
+    qmp_migrate_incoming(uri, false, NULL, &err);
       hmp_handle_error(mon, err);
   }
@@ -705,8 +705,8 @@ void hmp_migrate(Monitor *mon, const QDict *qdict)
       const char *uri = qdict_get_str(qdict, "uri");
       Error *err = NULL;
-    qmp_migrate(uri, !!blk, blk, !!inc, inc,
-                false, false, true, resume, &err);
+    qmp_migrate(uri, false, NULL, !!blk, blk, !!inc, inc,
+                 false, false, true, resume, &err);
       if (hmp_handle_error(mon, err)) {
           return;
       }
diff --git a/migration/migration.c b/migration/migration.c
index f37b388876..bd3a93fc8c 100644
--- a/migration/migration.c
+++ b/migration/migration.c
@@ -466,10 +466,22 @@ static bool migrate_uri_parse(const char *uri,
       return true;
   }
-static void qemu_start_incoming_migration(const char *uri, Error **errp)
+static void qemu_start_incoming_migration(const char *uri, bool has_channels,
+                                          MigrationChannelList *channels,
+                                          Error **errp)
   {
       g_autoptr(MigrationAddress) channel = g_new0(MigrationAddress, 1);
+    /*
+     * Having preliminary checks for uri and channel
+     */
+    if (uri && has_channels) {
+        error_setg(errp, "'uri' and 'channels' arguments are mutually "
+                   "exclusive; exactly one of the two should be present in "
+                   "'migrate-incoming' qmp command ");
+        return;
+    }
This checks is both are present.

Also needs a check if neither are present as that's invalid.
Also it should (temporarily) raise an error if "has_channels" is
set, as while we've added the parameter in QAPI, we've not
implemented it yet. IOW, raise an error now, and remove the
error in a later patch.
Ack. So in total there should be 3 checks right. 1) if 'has_channels' is
set, 2) if 'uri' and 'channels' both are present, 3) if 'uri' and 'channels'
both are absent. Basically right now only uri should allowed and should
atleast be present.
Correct.
Ack.
I think overall only 1) would be enough and should be checked before
'migration_channels_and_uri_compatible()' and if 'has_channels' is set, just
return for now. With this 2) would not be necessary or not come into play in
this patch. 3) will be taken care by
'migration_channels_and_uri_compatible()' itself IMO.
I think all the checks should be in this method, as it is the entrypoint
for execution of the QMP command and thus where parameter validation
should live. Spreading it across methods will make it very easy to open
bugs by refactoring code and not realizing the new codepaths don't do
the right checks.
Okay, I got the concern here. Will add these 3 checks in this patch and 
later remove 'has_channels' check in the patch where we are implementing 
MigrateChannel.
With regards,
Daniel
Regards,
Het Gala