[v1] Fix multifd with big number of channels

[Qemu-devel] [PATCH 5/6] migration: add some multifd traces

Posted by Juan Quintela 6 years, 5 months ago

Signed-off-by: Juan Quintela <quintela@redhat.com>
---
 migration/ram.c        | 3 +++
 migration/trace-events | 4 ++++
 2 files changed, 7 insertions(+)

diff --git a/migration/ram.c b/migration/ram.c
index f1aec95f83..25a211c3fb 100644
--- a/migration/ram.c
+++ b/migration/ram.c
@@ -1173,6 +1173,7 @@ static void *multifd_send_thread(void *opaque)
 
 out:
     if (local_err) {
+        trace_multifd_send_error(p->id);
         multifd_send_terminate_threads(local_err);
     }
 
@@ -1203,6 +1204,7 @@ static void multifd_new_send_channel_async(QIOTask *task, gpointer opaque)
     QIOChannel *sioc = QIO_CHANNEL(qio_task_get_source(task));
     Error *local_err = NULL;
 
+    trace_multifd_new_send_channel_async(p->id);
     if (qio_task_propagate_error(task, &local_err)) {
         migrate_set_error(migrate_get_current(), local_err);
         multifd_save_cleanup();
@@ -1496,6 +1498,7 @@ bool multifd_recv_new_channel(QIOChannel *ioc, Error **errp)
                                 atomic_read(&multifd_recv_state->count));
         return false;
     }
+    trace_multifd_recv_new_channel(id);
 
     p = &multifd_recv_state->params[id];
     if (p->c != NULL) {
diff --git a/migration/trace-events b/migration/trace-events
index 9fbef614ab..5d85f8bf83 100644
--- a/migration/trace-events
+++ b/migration/trace-events
@@ -81,7 +81,9 @@ migration_bitmap_sync_start(void) ""
 migration_bitmap_sync_end(uint64_t dirty_pages) "dirty_pages %" PRIu64
 migration_bitmap_clear_dirty(char *str, uint64_t start, uint64_t size, unsigned long page) "rb %s start 0x%"PRIx64" size 0x%"PRIx64" page 0x%lx"
 migration_throttle(void) ""
+multifd_new_send_channel_async(uint8_t id) "channel %d"
 multifd_recv(uint8_t id, uint64_t packet_num, uint32_t used, uint32_t flags, uint32_t next_packet_size) "channel %d packet_num %" PRIu64 " pages %d flags 0x%x next packet size %d"
+multifd_recv_new_channel(uint8_t id) "channel %d"
 multifd_recv_sync_main(long packet_num) "packet num %ld"
 multifd_recv_sync_main_signal(uint8_t id) "channel %d"
 multifd_recv_sync_main_wait(uint8_t id) "channel %d"
@@ -89,7 +91,9 @@ multifd_recv_terminate_threads(bool error) "error %d"
 multifd_recv_thread_can_start(uint8_t id) "channel %d"
 multifd_recv_thread_end(uint8_t id, uint64_t packets, uint64_t pages) "channel %d packets %" PRIu64 " pages %" PRIu64
 multifd_recv_thread_start(uint8_t id) "%d"
+multifd_save_setup_wait(uint8_t id) "%d"
 multifd_send(uint8_t id, uint64_t packet_num, uint32_t used, uint32_t flags, uint32_t next_packet_size) "channel %d packet_num %" PRIu64 " pages %d flags 0x%x next packet size %d"
+multifd_send_error(uint8_t id) "channel %d"
 multifd_send_sync_main(long packet_num) "packet num %ld"
 multifd_send_sync_main_signal(uint8_t id) "channel %d"
 multifd_send_sync_main_wait(uint8_t id) "channel %d"
-- 
2.21.0

Re: [Qemu-devel] [PATCH 5/6] migration: add some multifd traces

Posted by Dr. David Alan Gilbert 6 years, 5 months ago

* Juan Quintela (quintela@redhat.com) wrote:
> Signed-off-by: Juan Quintela <quintela@redhat.com>


Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>

Queued
> ---
>  migration/ram.c        | 3 +++
>  migration/trace-events | 4 ++++
>  2 files changed, 7 insertions(+)
> 
> diff --git a/migration/ram.c b/migration/ram.c
> index f1aec95f83..25a211c3fb 100644
> --- a/migration/ram.c
> +++ b/migration/ram.c
> @@ -1173,6 +1173,7 @@ static void *multifd_send_thread(void *opaque)
>  
>  out:
>      if (local_err) {
> +        trace_multifd_send_error(p->id);
>          multifd_send_terminate_threads(local_err);
>      }
>  
> @@ -1203,6 +1204,7 @@ static void multifd_new_send_channel_async(QIOTask *task, gpointer opaque)
>      QIOChannel *sioc = QIO_CHANNEL(qio_task_get_source(task));
>      Error *local_err = NULL;
>  
> +    trace_multifd_new_send_channel_async(p->id);
>      if (qio_task_propagate_error(task, &local_err)) {
>          migrate_set_error(migrate_get_current(), local_err);
>          multifd_save_cleanup();
> @@ -1496,6 +1498,7 @@ bool multifd_recv_new_channel(QIOChannel *ioc, Error **errp)
>                                  atomic_read(&multifd_recv_state->count));
>          return false;
>      }
> +    trace_multifd_recv_new_channel(id);
>  
>      p = &multifd_recv_state->params[id];
>      if (p->c != NULL) {
> diff --git a/migration/trace-events b/migration/trace-events
> index 9fbef614ab..5d85f8bf83 100644
> --- a/migration/trace-events
> +++ b/migration/trace-events
> @@ -81,7 +81,9 @@ migration_bitmap_sync_start(void) ""
>  migration_bitmap_sync_end(uint64_t dirty_pages) "dirty_pages %" PRIu64
>  migration_bitmap_clear_dirty(char *str, uint64_t start, uint64_t size, unsigned long page) "rb %s start 0x%"PRIx64" size 0x%"PRIx64" page 0x%lx"
>  migration_throttle(void) ""
> +multifd_new_send_channel_async(uint8_t id) "channel %d"
>  multifd_recv(uint8_t id, uint64_t packet_num, uint32_t used, uint32_t flags, uint32_t next_packet_size) "channel %d packet_num %" PRIu64 " pages %d flags 0x%x next packet size %d"
> +multifd_recv_new_channel(uint8_t id) "channel %d"
>  multifd_recv_sync_main(long packet_num) "packet num %ld"
>  multifd_recv_sync_main_signal(uint8_t id) "channel %d"
>  multifd_recv_sync_main_wait(uint8_t id) "channel %d"
> @@ -89,7 +91,9 @@ multifd_recv_terminate_threads(bool error) "error %d"
>  multifd_recv_thread_can_start(uint8_t id) "channel %d"
>  multifd_recv_thread_end(uint8_t id, uint64_t packets, uint64_t pages) "channel %d packets %" PRIu64 " pages %" PRIu64
>  multifd_recv_thread_start(uint8_t id) "%d"
> +multifd_save_setup_wait(uint8_t id) "%d"
>  multifd_send(uint8_t id, uint64_t packet_num, uint32_t used, uint32_t flags, uint32_t next_packet_size) "channel %d packet_num %" PRIu64 " pages %d flags 0x%x next packet size %d"
> +multifd_send_error(uint8_t id) "channel %d"
>  multifd_send_sync_main(long packet_num) "packet num %ld"
>  multifd_send_sync_main_signal(uint8_t id) "channel %d"
>  multifd_send_sync_main_wait(uint8_t id) "channel %d"
> -- 
> 2.21.0
> 
--
Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK

Re: [Qemu-devel] [PATCH 5/6] migration: add some multifd traces

Posted by Philippe Mathieu-Daudé 6 years, 5 months ago

On 8/14/19 4:02 AM, Juan Quintela wrote:
> Signed-off-by: Juan Quintela <quintela@redhat.com>
> ---
>  migration/ram.c        | 3 +++
>  migration/trace-events | 4 ++++
>  2 files changed, 7 insertions(+)
> 
> diff --git a/migration/ram.c b/migration/ram.c
> index f1aec95f83..25a211c3fb 100644
> --- a/migration/ram.c
> +++ b/migration/ram.c
> @@ -1173,6 +1173,7 @@ static void *multifd_send_thread(void *opaque)
>  
>  out:
>      if (local_err) {
> +        trace_multifd_send_error(p->id);
>          multifd_send_terminate_threads(local_err);
>      }
>  
> @@ -1203,6 +1204,7 @@ static void multifd_new_send_channel_async(QIOTask *task, gpointer opaque)
>      QIOChannel *sioc = QIO_CHANNEL(qio_task_get_source(task));
>      Error *local_err = NULL;
>  
> +    trace_multifd_new_send_channel_async(p->id);
>      if (qio_task_propagate_error(task, &local_err)) {
>          migrate_set_error(migrate_get_current(), local_err);
>          multifd_save_cleanup();
> @@ -1496,6 +1498,7 @@ bool multifd_recv_new_channel(QIOChannel *ioc, Error **errp)
>                                  atomic_read(&multifd_recv_state->count));
>          return false;
>      }
> +    trace_multifd_recv_new_channel(id);
>  
>      p = &multifd_recv_state->params[id];
>      if (p->c != NULL) {
> diff --git a/migration/trace-events b/migration/trace-events
> index 9fbef614ab..5d85f8bf83 100644
> --- a/migration/trace-events
> +++ b/migration/trace-events
> @@ -81,7 +81,9 @@ migration_bitmap_sync_start(void) ""
>  migration_bitmap_sync_end(uint64_t dirty_pages) "dirty_pages %" PRIu64
>  migration_bitmap_clear_dirty(char *str, uint64_t start, uint64_t size, unsigned long page) "rb %s start 0x%"PRIx64" size 0x%"PRIx64" page 0x%lx"
>  migration_throttle(void) ""
> +multifd_new_send_channel_async(uint8_t id) "channel %d"
>  multifd_recv(uint8_t id, uint64_t packet_num, uint32_t used, uint32_t flags, uint32_t next_packet_size) "channel %d packet_num %" PRIu64 " pages %d flags 0x%x next packet size %d"
> +multifd_recv_new_channel(uint8_t id) "channel %d"
>  multifd_recv_sync_main(long packet_num) "packet num %ld"
>  multifd_recv_sync_main_signal(uint8_t id) "channel %d"
>  multifd_recv_sync_main_wait(uint8_t id) "channel %d"
> @@ -89,7 +91,9 @@ multifd_recv_terminate_threads(bool error) "error %d"
>  multifd_recv_thread_can_start(uint8_t id) "channel %d"
>  multifd_recv_thread_end(uint8_t id, uint64_t packets, uint64_t pages) "channel %d packets %" PRIu64 " pages %" PRIu64
>  multifd_recv_thread_start(uint8_t id) "%d"
> +multifd_save_setup_wait(uint8_t id) "%d"

This one is not used.

For the rest:
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>

>  multifd_send(uint8_t id, uint64_t packet_num, uint32_t used, uint32_t flags, uint32_t next_packet_size) "channel %d packet_num %" PRIu64 " pages %d flags 0x%x next packet size %d"
> +multifd_send_error(uint8_t id) "channel %d"
>  multifd_send_sync_main(long packet_num) "packet num %ld"
>  multifd_send_sync_main_signal(uint8_t id) "channel %d"
>  multifd_send_sync_main_wait(uint8_t id) "channel %d"
>

Re: [Qemu-devel] [PATCH 5/6] migration: add some multifd traces

Posted by Philippe Mathieu-Daudé 6 years, 5 months ago

On 8/19/19 1:10 PM, Philippe Mathieu-Daudé wrote:
> On 8/14/19 4:02 AM, Juan Quintela wrote:
>> Signed-off-by: Juan Quintela <quintela@redhat.com>
>> ---
>>  migration/ram.c        | 3 +++
>>  migration/trace-events | 4 ++++
>>  2 files changed, 7 insertions(+)
>>
>> diff --git a/migration/ram.c b/migration/ram.c
>> index f1aec95f83..25a211c3fb 100644
>> --- a/migration/ram.c
>> +++ b/migration/ram.c
>> @@ -1173,6 +1173,7 @@ static void *multifd_send_thread(void *opaque)
>>  
>>  out:
>>      if (local_err) {
>> +        trace_multifd_send_error(p->id);
>>          multifd_send_terminate_threads(local_err);
>>      }
>>  
>> @@ -1203,6 +1204,7 @@ static void multifd_new_send_channel_async(QIOTask *task, gpointer opaque)
>>      QIOChannel *sioc = QIO_CHANNEL(qio_task_get_source(task));
>>      Error *local_err = NULL;
>>  
>> +    trace_multifd_new_send_channel_async(p->id);
>>      if (qio_task_propagate_error(task, &local_err)) {
>>          migrate_set_error(migrate_get_current(), local_err);
>>          multifd_save_cleanup();
>> @@ -1496,6 +1498,7 @@ bool multifd_recv_new_channel(QIOChannel *ioc, Error **errp)
>>                                  atomic_read(&multifd_recv_state->count));
>>          return false;
>>      }
>> +    trace_multifd_recv_new_channel(id);
>>  
>>      p = &multifd_recv_state->params[id];
>>      if (p->c != NULL) {
>> diff --git a/migration/trace-events b/migration/trace-events
>> index 9fbef614ab..5d85f8bf83 100644
>> --- a/migration/trace-events
>> +++ b/migration/trace-events
>> @@ -81,7 +81,9 @@ migration_bitmap_sync_start(void) ""
>>  migration_bitmap_sync_end(uint64_t dirty_pages) "dirty_pages %" PRIu64
>>  migration_bitmap_clear_dirty(char *str, uint64_t start, uint64_t size, unsigned long page) "rb %s start 0x%"PRIx64" size 0x%"PRIx64" page 0x%lx"
>>  migration_throttle(void) ""
>> +multifd_new_send_channel_async(uint8_t id) "channel %d"
>>  multifd_recv(uint8_t id, uint64_t packet_num, uint32_t used, uint32_t flags, uint32_t next_packet_size) "channel %d packet_num %" PRIu64 " pages %d flags 0x%x next packet size %d"
>> +multifd_recv_new_channel(uint8_t id) "channel %d"
>>  multifd_recv_sync_main(long packet_num) "packet num %ld"
>>  multifd_recv_sync_main_signal(uint8_t id) "channel %d"
>>  multifd_recv_sync_main_wait(uint8_t id) "channel %d"
>> @@ -89,7 +91,9 @@ multifd_recv_terminate_threads(bool error) "error %d"
>>  multifd_recv_thread_can_start(uint8_t id) "channel %d"
>>  multifd_recv_thread_end(uint8_t id, uint64_t packets, uint64_t pages) "channel %d packets %" PRIu64 " pages %" PRIu64
>>  multifd_recv_thread_start(uint8_t id) "%d"
>> +multifd_save_setup_wait(uint8_t id) "%d"
> 
> This one is not used.
> 
> For the rest:
> Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>

Oh I forgot, uint8_t format is "%u", we don't care so far but it could
be useful to enable the -Wformat-signedness cflag one day.

>>  multifd_send(uint8_t id, uint64_t packet_num, uint32_t used, uint32_t flags, uint32_t next_packet_size) "channel %d packet_num %" PRIu64 " pages %d flags 0x%x next packet size %d"
>> +multifd_send_error(uint8_t id) "channel %d"
>>  multifd_send_sync_main(long packet_num) "packet num %ld"
>>  multifd_send_sync_main_signal(uint8_t id) "channel %d"
>>  multifd_send_sync_main_wait(uint8_t id) "channel %d"
>>

[Qemu-devel] [PATCH 1/6] migration: Add traces for multifd terminate threads
[Qemu-devel] [PATCH 2/6] migration: Make global sem_sync semaphore by channel
[Qemu-devel] [PATCH 3/6] migration: Make sure that all multifd channels have been created
[Qemu-devel] [PATCH 4/6] migration: Make multifd threads wait until all have been created
[Qemu-devel] [PATCH 5/6] migration: add some multifd traces
[Qemu-devel] [PATCH 6/6] RFH: We lost "connect" events