On 22/09/2023 23:42, Peter Xu wrote:
> On Wed, Sep 20, 2023 at 05:04:11PM +0800, Li Zhijian wrote:
>> From: Li Zhijian <lizhij...@cn.fujitsu.com>
>>
>> Destination will fail with:
>> qemu-system-x86_64: rdma: Too many requests in this message
>> (3638950032).Bailing.
>>
>> migrate with RDMA is different from tcp. RDMA has its own control
>> message, and all traffic between RDMA_CONTROL_REGISTER_REQUEST and
>> RDMA_CONTROL_REGISTER_FINISHED should not be disturbed.
>>
>> find_dirty_block() will be called during RDMA_CONTROL_REGISTER_REQUEST
>> and RDMA_CONTROL_REGISTER_FINISHED, it will send a extra traffic to
>> destination and cause migration to fail.
>>
>> Since there's no existing subroutine to indicate whether it's migrated
>> by RDMA or not, and RDMA is not compatible with multifd, we use
>> migrate_multifd() here.
>>
>> Fixes: 294e5a4034 ("multifd: Only flush once each full round of memory")
>> Signed-off-by: Li Zhijian <lizhij...@cn.fujitsu.com>
>> ---
>> migration/ram.c | 3 ++-
>> 1 file changed, 2 insertions(+), 1 deletion(-)
>>
>> diff --git a/migration/ram.c b/migration/ram.c
>> index 9040d66e61..89ae28e21a 100644
>> --- a/migration/ram.c
>> +++ b/migration/ram.c
>> @@ -1399,7 +1399,8 @@ static int find_dirty_block(RAMState *rs,
>> PageSearchStatus *pss)
>> pss->page = 0;
>> pss->block = QLIST_NEXT_RCU(pss->block, next);
>> if (!pss->block) {
>> - if (!migrate_multifd_flush_after_each_section()) {
>> + if (migrate_multifd() &&
>> + !migrate_multifd_flush_after_each_section()) {
>> QEMUFile *f = rs->pss[RAM_CHANNEL_PRECOPY].pss_channel;
>> int ret = multifd_send_sync_main(f);
>> if (ret < 0) {
>> --
>> 2.31.1
>>
>
> Maybe better to put that check at the entry of
> migrate_multifd_flush_after_each_section()?
>
It sounds good to me:
diff --git a/migration/options.c b/migration/options.c
index 1d1e1321b0a..327bcf2fbe4 100644
--- a/migration/options.c
+++ b/migration/options.c
@@ -368,7 +368,7 @@ bool migrate_multifd_flush_after_each_section(void)
{
MigrationState *s = migrate_get_current();
- return s->multifd_flush_after_each_section;
+ return !migrate_multifd() || s->multifd_flush_after_each_section;
}
bool migrate_postcopy(void)
That changes make migrate_multifd_flush_after_each_section() always true when
multifd is disabled.
Thanks
> I also hope that some day there's no multifd function called in generic
> migration code paths..
>
