Re: [PATCH v4 10/14] migration/multifd: Enable DSA offloading in multifd sender path.

[Daniel P . Berrangé]

On Thu, Apr 25, 2024 at 02:21:13AM +0000, Hao Xiang wrote:
> Multifd sender path gets an array of pages queued by the migration
> thread. It performs zero page checking on every page in the array.
> The pages are classfied as either a zero page or a normal page. This
> change uses Intel DSA to offload the zero page checking from CPU to
> the DSA accelerator. The sender thread submits a batch of pages to DSA
> hardware and waits for the DSA completion thread to signal for work
> completion.
> 
> Signed-off-by: Hao Xiang <hao.xiang@linux.dev>
> ---
>  migration/multifd-zero-page.c | 99 +++++++++++++++++++++++++++++++++--
>  migration/multifd.c           | 27 +++++++++-
>  migration/multifd.h           |  1 +
>  3 files changed, 120 insertions(+), 7 deletions(-)
> 
> diff --git a/migration/multifd-zero-page.c b/migration/multifd-zero-page.c
> index e1b8370f88..4f426289e4 100644
> --- a/migration/multifd-zero-page.c
> +++ b/migration/multifd-zero-page.c

> diff --git a/migration/multifd.c b/migration/multifd.c
> index cfd3a92f6c..7316643d0a 100644
> --- a/migration/multifd.c
> +++ b/migration/multifd.c
> @@ -818,6 +818,8 @@ void multifd_send_shutdown(void)
>  
>      multifd_send_terminate_threads();
>  
> +    dsa_cleanup();
> +
>      for (i = 0; i < migrate_multifd_channels(); i++) {
>          MultiFDSendParams *p = &multifd_send_state->params[i];
>          Error *local_err = NULL;
> @@ -1155,11 +1157,20 @@ bool multifd_send_setup(void)
>      uint32_t page_count = MULTIFD_PACKET_SIZE / qemu_target_page_size();
>      bool use_packets = multifd_use_packets();
>      uint8_t i;
> +    const char *dsa_parameter = migrate_multifd_dsa_accel();
>  
>      if (!migrate_multifd()) {
>          return true;
>      }
>  
> +    if (dsa_init(dsa_parameter)) {
> +        error_setg(&local_err, "multifd: Sender failed to initialize DSA.");
> +        error_report_err(local_err);
> +        return false;
> +    }

This is an example of why all the dsa functions need to report
a via an "Error **err" parameter. The error reported
here is useless as it lacks any meaningful information of what
went wrong.

The multifd_send_setup method itself needs a "Error **errp"
param so it can pass it back up to be reoprted too.

> +
> +    dsa_start();
> +
>      thread_count = migrate_multifd_channels();
>      multifd_send_state = g_malloc0(sizeof(*multifd_send_state));
>      multifd_send_state->params = g_new0(MultiFDSendParams, thread_count);
> @@ -1393,6 +1404,7 @@ void multifd_recv_cleanup(void)
>              qemu_thread_join(&p->thread);
>          }
>      }
> +    dsa_cleanup();
>      for (i = 0; i < migrate_multifd_channels(); i++) {
>          multifd_recv_cleanup_channel(&multifd_recv_state->params[i]);
>      }
> @@ -1568,6 +1580,9 @@ int multifd_recv_setup(Error **errp)
>      uint32_t page_count = MULTIFD_PACKET_SIZE / qemu_target_page_size();
>      bool use_packets = multifd_use_packets();
>      uint8_t i;
> +    const char *dsa_parameter = migrate_multifd_dsa_accel();
> +    int ret;
> +    Error *local_err = NULL;
>  
>      /*
>       * Return successfully if multiFD recv state is already initialised
> @@ -1577,6 +1592,15 @@ int multifd_recv_setup(Error **errp)
>          return 0;
>      }
>  
> +    ret = dsa_init(dsa_parameter);
> +    if (ret != 0) {
> +        error_setg(&local_err, "multifd: Receiver failed to initialize 
> DSA.");
> +        error_propagate(errp, local_err);
> +        return ret;
> +    }
> +
> +    dsa_start();
> +
>      thread_count = migrate_multifd_channels();
>      multifd_recv_state = g_malloc0(sizeof(*multifd_recv_state));
>      multifd_recv_state->params = g_new0(MultiFDRecvParams, thread_count);
> @@ -1616,13 +1640,12 @@ int multifd_recv_setup(Error **errp)
>  
>      for (i = 0; i < thread_count; i++) {
>          MultiFDRecvParams *p = &multifd_recv_state->params[i];
> -        int ret;
> -
>          ret = multifd_recv_state->ops->recv_setup(p, errp);
>          if (ret) {
>              return ret;
>          }
>      }
> +
>      return 0;
>  }
>  
> diff --git a/migration/multifd.h b/migration/multifd.h
> index 16e27db5e9..b3717fae24 100644
> --- a/migration/multifd.h
> +++ b/migration/multifd.h
> @@ -14,6 +14,7 @@
>  #define QEMU_MIGRATION_MULTIFD_H
>  
>  #include "ram.h"
> +#include "qemu/dsa.h"
>  
>  typedef struct MultiFDRecvData MultiFDRecvData;
>  
> -- 
> 2.30.2
> 
> 

With regards,
Daniel
-- 
|: https://berrange.com      -o-    https://www.flickr.com/photos/dberrange :|
|: https://libvirt.org         -o-            https://fstop138.berrange.com :|
|: https://entangle-photo.org    -o-    https://www.instagram.com/dberrange :|