Except for RDMA migration, other parts of the RDMA subsystem have been
removed since 9.1.

Due to the lack of unit tests and CI tests for RDMA migration, int the
past developing cycles, a few fatal errors were introduced and broke the
RDMA migration, and these issues[1][2] were not fixed until some time later.

Modern network cards (TCP/IP) can also provide high bandwidth
(similar to RDMA) to handle the large amount of data generated during
migration.

Issue a warning to inform the end users of the RDMA migration status.

[1] https://lore.kernel.org/r/20230920090412.726725-1-lizhijian@fujitsu.com
[2] https://lore.kernel.org/r/CAHEcVy7HXSwn4Ow_Kog+Q+TN6f_kMeiCHevz1qGM-fbxBPp1hQ@mail.gmail.com

CC: Peter Xu <peterx@redhat.com>
CC: Philippe Mathieu-Daudé <philmd@linaro.org>
CC: Fabiano Rosas <farosas@suse.de>
CC: Thomas Huth <thuth@redhat.com>
CC: Daniel P. Berrangé <berrange@redhat.com>
CC: Yu Zhang <yu.zhang@ionos.com>
Signed-off-by: Li Zhijian <lizhijian@fujitsu.com>
---
 docs/about/deprecated.rst | 15 +++++++++++++++
 migration/migration.c     |  1 +
 2 files changed, 16 insertions(+)

diff --git a/docs/about/deprecated.rst b/docs/about/deprecated.rst
index 7b548519b5..fe70a7009e 100644
--- a/docs/about/deprecated.rst
+++ b/docs/about/deprecated.rst
@@ -529,3 +529,18 @@ Compression method fails too much.  Too many races.  We are going to
 remove it if nobody fixes it.  For starters, migration-test
 compression tests are disabled because they fail randomly.  If you need
 compression, use multifd compression methods.
+
+RDMA migration (since 9.1)
+''''''''''''''''''''''''''''''''''
+
+The QEMU project intends to remove the whole RDMA subsystem from the
+code base in a future release without replacement unless somebody steps
+up and improves the situation. So far, except for RDMA migration, other
+parts of the RDMA subsystem have been removed since 9.1.
+
+Due to the lack of unit tests and CI tests for RDMA migration, in the past
+developing cycles, a few fatal errors were introduced and broke the RDMA
+migration, and these issues were not fixed until some time later.
+
+Modern network cards (TCP/IP) can also provide high bandwidth (similar to RDMA)
+to handle the large amount of data generated during migration.
diff --git a/migration/migration.c b/migration/migration.c
index 9fe8fd2afd..807d66bbba 100644
--- a/migration/migration.c
+++ b/migration/migration.c
@@ -563,6 +563,7 @@ bool migrate_uri_parse(const char *uri, MigrationChannel **channel,
             qapi_free_InetSocketAddress(isock);
             return false;
         }
+        warn_report("RDMA migration is deprecated and will be removed in a future release");
         addr->transport = MIGRATION_ADDRESS_TYPE_RDMA;
     } else if (strstart(uri, "tcp:", NULL) ||
                 strstart(uri, "unix:", NULL) ||
-- 
2.41.0

On 1/4/24 05:59, Li Zhijian wrote:
> Except for RDMA migration, other parts of the RDMA subsystem have been
> removed since 9.1.
> 
> Due to the lack of unit tests and CI tests for RDMA migration, int the
> past developing cycles, a few fatal errors were introduced and broke the
> RDMA migration, and these issues[1][2] were not fixed until some time later.
> 
> Modern network cards (TCP/IP) can also provide high bandwidth
> (similar to RDMA) to handle the large amount of data generated during
> migration.
> 
> Issue a warning to inform the end users of the RDMA migration status.
> 
> [1] https://lore.kernel.org/r/20230920090412.726725-1-lizhijian@fujitsu.com
> [2] https://lore.kernel.org/r/CAHEcVy7HXSwn4Ow_Kog+Q+TN6f_kMeiCHevz1qGM-fbxBPp1hQ@mail.gmail.com
> 
> CC: Peter Xu <peterx@redhat.com>
> CC: Philippe Mathieu-Daudé <philmd@linaro.org>
> CC: Fabiano Rosas <farosas@suse.de>
> CC: Thomas Huth <thuth@redhat.com>
> CC: Daniel P. Berrangé <berrange@redhat.com>
> CC: Yu Zhang <yu.zhang@ionos.com>
> Signed-off-by: Li Zhijian <lizhijian@fujitsu.com>
> ---
>   docs/about/deprecated.rst | 15 +++++++++++++++
>   migration/migration.c     |  1 +
>   2 files changed, 16 insertions(+)

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>

On Mon, Apr 01, 2024 at 11:59:47AM +0800, Li Zhijian wrote:
> Except for RDMA migration, other parts of the RDMA subsystem have been
> removed since 9.1.
> 
> Due to the lack of unit tests and CI tests for RDMA migration, int the
> past developing cycles, a few fatal errors were introduced and broke the
> RDMA migration, and these issues[1][2] were not fixed until some time later.
> 
> Modern network cards (TCP/IP) can also provide high bandwidth
> (similar to RDMA) to handle the large amount of data generated during
> migration.
> 
> Issue a warning to inform the end users of the RDMA migration status.
> 
> [1] https://lore.kernel.org/r/20230920090412.726725-1-lizhijian@fujitsu.com
> [2] https://lore.kernel.org/r/CAHEcVy7HXSwn4Ow_Kog+Q+TN6f_kMeiCHevz1qGM-fbxBPp1hQ@mail.gmail.com
> 
> CC: Peter Xu <peterx@redhat.com>
> CC: Philippe Mathieu-Daudé <philmd@linaro.org>
> CC: Fabiano Rosas <farosas@suse.de>
> CC: Thomas Huth <thuth@redhat.com>
> CC: Daniel P. Berrangé <berrange@redhat.com>
> CC: Yu Zhang <yu.zhang@ionos.com>
> Signed-off-by: Li Zhijian <lizhijian@fujitsu.com>

Thanks, Zhijian and everyone.

Acked-by: Peter Xu <peterx@redhat.com>

I'll keep this around for more days for a better exposure, and this will be
included in the 1st 9.1 pull if no objections.

-- 
Peter Xu