[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [PATCH v4 10/18] migration/rdma: Create the multifd recv channels fo
From: |
Zheng Chuan |
Subject: |
Re: [PATCH v4 10/18] migration/rdma: Create the multifd recv channels for RDMA |
Date: |
Sat, 6 Mar 2021 16:45:18 +0800 |
User-agent: |
Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:68.0) Gecko/20100101 Thunderbird/68.6.0 |
On 2021/2/4 2:59, Dr. David Alan Gilbert wrote:
> * Chuan Zheng (zhengchuan@huawei.com) wrote:
>> We still don't transmit anything through them, and we only build
>> the RDMA connections.
>>
>> Signed-off-by: Zhimin Feng <fengzhimin1@huawei.com>
>> Signed-off-by: Chuan Zheng <zhengchuan@huawei.com>
>> ---
>> migration/rdma.c | 69
>> ++++++++++++++++++++++++++++++++++++++++++++++++++++++--
>> 1 file changed, 67 insertions(+), 2 deletions(-)
>>
>> diff --git a/migration/rdma.c b/migration/rdma.c
>> index 996afb0..ed8a015 100644
>> --- a/migration/rdma.c
>> +++ b/migration/rdma.c
>> @@ -3267,6 +3267,40 @@ static void rdma_cm_poll_handler(void *opaque)
>> }
>> }
>>
>> +static bool qemu_rdma_accept_setup(RDMAContext *rdma)
>> +{
>> + RDMAContext *multifd_rdma = NULL;
>> + int thread_count;
>> + int i;
>> + MultiFDRecvParams *multifd_recv_param;
>> + thread_count = migrate_multifd_channels();
>> + /* create the multifd channels for RDMA */
>> + for (i = 0; i < thread_count; i++) {
>> + if (get_multifd_recv_param(i, &multifd_recv_param) < 0) {
>> + error_report("rdma: error getting multifd_recv_param(%d)", i);
>> + return false;
>> + }
>> +
>> + multifd_rdma = (RDMAContext *) multifd_recv_param->rdma;
>> + if (multifd_rdma->cm_id == NULL) {
>> + break;
>> + } else {
>> + multifd_rdma = NULL;
>> + }
>
> I'm confused by what this if is doing - what are the two cases?
>
Since we share the CM channel and CM id with main thread,
we assign the cmd_id through the callback rdma_accept_incoming_migration() for
the multifd thread if cm_id is NULL.
Once it is assigned, we could go to the normal rdma_cm_poll_handler() set
handler.
>> + }
>> +
>> + if (multifd_rdma) {
>> + qemu_set_fd_handler(rdma->channel->fd,
>> + rdma_accept_incoming_migration,
>> + NULL, (void *)(intptr_t)multifd_rdma);
>> + } else {
>> + qemu_set_fd_handler(rdma->channel->fd, rdma_cm_poll_handler,
>> + NULL, rdma);
>> + }
>> +
>> + return true;
>> +}
>> +
>> static int qemu_rdma_accept(RDMAContext *rdma)
>> {
>> RDMACapabilities cap;
>> @@ -3366,6 +3400,10 @@ static int qemu_rdma_accept(RDMAContext *rdma)
>> qemu_set_fd_handler(rdma->channel->fd,
>> rdma_accept_incoming_migration,
>> NULL,
>> (void *)(intptr_t)rdma->return_path);
>> + } else if (migrate_use_multifd()) {
>> + if (!qemu_rdma_accept_setup(rdma)) {
>> + goto err_rdma_dest_wait;
>> + }
>> } else {
>> qemu_set_fd_handler(rdma->channel->fd, rdma_cm_poll_handler,
>> NULL, rdma);
>> @@ -3976,6 +4014,34 @@ static QEMUFile *qemu_fopen_rdma(RDMAContext *rdma,
>> const char *mode)
>> return rioc->file;
>> }
>>
>> +static void migration_rdma_process_incoming(QEMUFile *f,
>> + RDMAContext *rdma, Error **errp)
>> +{
>> + MigrationIncomingState *mis = migration_incoming_get_current();
>> + QIOChannel *ioc = NULL;
>> + bool start_migration = false;
>> +
>> + if (!migrate_use_multifd()) {
>> + rdma->migration_started_on_destination = 1;
>> + migration_fd_process_incoming(f, errp);
>> + return;
>> + }
>> +
>> + if (!mis->from_src_file) {
>> + mis->from_src_file = f;
>> + qemu_file_set_blocking(f, false);
>> + } else {
>> + ioc = QIO_CHANNEL(getQIOChannel(f));
>> + /* Multiple connections */
>> + assert(migrate_use_multifd());
>
> Are you sure that's never triggerable by something trying to connect
> badly? If it was it would be better to error than abort.
>
This is the similiar action with tcp multifd which is introduced by
a429e7f4887313370,
However we will never get there if migrate_use_multifd is false because of
return at the first judgement of function, we could not do it or just put a
warning.
>> + start_migration = multifd_recv_new_channel(ioc, errp);
>
> And what does 'start_migration' mean here - is that meaning that we have
> a full set of connections?
>
Yes, multifd_recv_new_channel returns true when correctly receiving all
channels.
> Dave
>
>> + }
>> +
>> + if (start_migration) {
>> + migration_incoming_process();
>> + }
>> +}
>> +
>> static void rdma_accept_incoming_migration(void *opaque)
>> {
>> RDMAContext *rdma = opaque;
>> @@ -4004,8 +4070,7 @@ static void rdma_accept_incoming_migration(void
>> *opaque)
>> return;
>> }
>>
>> - rdma->migration_started_on_destination = 1;
>> - migration_fd_process_incoming(f, &local_err);
>> + migration_rdma_process_incoming(f, rdma, &local_err);
>> if (local_err) {
>> error_reportf_err(local_err, "RDMA ERROR:");
>> }
>> --
>> 1.8.3.1
>>
--
Regards.
Chuan
[Prev in Thread] |
Current Thread |
[Next in Thread] |
- Re: [PATCH v4 10/18] migration/rdma: Create the multifd recv channels for RDMA,
Zheng Chuan <=